**Toward Unifying Global Hotspots of Wild and Domesticated Biodiversity**

**Samuel Pironon 1,\*, James S. Borrell 1, Ian Ondo 1, Ruben Douglas 1, Charlotte Phillips 2, Colin K. Khoury 3,4, Michael B. Kantar 5, Nathan Fumia 5, Marybel Soto Gomez 6,7, Juan Viruel 1, Rafael Govaerts 1, Félix Forest <sup>1</sup> and Alexandre Antonelli 1,8**


Received: 17 July 2020; Accepted: 27 August 2020; Published: 31 August 2020

**Abstract:** Global biodiversity hotspots are areas containing high levels of species richness, endemism and threat. Similarly, regions of agriculturally relevant diversity have been identified where many domesticated plants and animals originated, and co-occurred with their wild ancestors and relatives. The agro-biodiversity in these regions has, likewise, often been considered threatened. Biodiversity and agro-biodiversity hotspots partly overlap, but their geographic intricacies have rarely been investigated together. Here we review the history of these two concepts and explore their geographic relationship by analysing global distribution and human use data for all plants, and for major crops and associated wild relatives. We highlight a geographic continuum between agro-biodiversity hotspots that contain high richness in species that are intensively used and well known by humanity (i.e., major crops and most viewed species on Wikipedia) and biodiversity hotspots encompassing species that are less heavily used and documented (i.e., crop wild relatives and species lacking information on Wikipedia). Our contribution highlights the key considerations needed for further developing a unifying concept of agro-biodiversity hotspots that encompasses multiple facets of diversity (including genetic and phylogenetic) and the linkage with overall biodiversity. This integration will ultimately enhance our understanding of the geography of human-plant interactions and help guide the preservation of nature and its contributions to people.

**Keywords:** agro-biodiversity; breeding; centres of origin; conservation; crop wild relatives; domestication; geographic distribution; phylogenetic diversity; useful plants; Vavilov centres

#### **1. Introduction**

Biogeographers and conservation biologists have long been interested in identifying and characterizing geographic regions containing a higher concentration of biodiversity and derived natural resources than surrounding areas, ranging from within- and among-species diversity through to ecosystem services [1–3], at different spatial, temporal and taxonomic scales [4]. Centres (also known as hotspots) and peripheries (coldspots) of plant diversity have been shown to be unevenly distributed and to play a fundamental role in shaping ecosystems and delivering associated benefits to humans and other species [5,6]. Mapping efforts contribute to a better fundamental understanding of both biodiversity (e.g., species extinction, diversification and co-existence) [7], and the interaction between people and nature, including the resulting socio-economic benefits and threats [8–11]. This is a particularly urgent endeavour in the current context of a rapidly growing global human population with increasing consumer expectations, posing a serious threat for both plant diversity and its long-term contributions to people [12]. Factors, such as land use and climate change, pollution, direct exploitation of species, and biological invasions have direct and indirect impacts on plant diversity [13,14], potentially undermining current and unrealised plant-based adaptive solutions [15,16] and traditional knowledge associated with plant uses [17]. Documenting the distribution of plant diversity and its uses are, therefore, critical steps towards developing the transformative changes required to achieve socio-economic sustainability, while preserving life on Earth.

Conservationists, constrained by finite resources, have used the concept of biodiversity hotspots to advocate for the allocation of international efforts and resources in regions of the world containing exceptionally diverse, unique and threatened biodiversity [18]. However, biodiversity hotspots often fail to capture the multi-faceted nature of biodiversity. For example, hotspot designations may consider a narrow range of organisms [19], and miss non-terrestrial habitats [3], phylogenetic and functional diversity (but see [20]). The seminal definition of a biodiversity hotspot, which was based solely on plants from tropical forests, highlights this shortcoming [21]. An additional limitation of biodiversity hotspots, as currently defined, is that they rarely consider anthropogenic interactions as anything other than a threat [22]. However, the sustainable and responsible use of nature, as demonstrated by traditional small-scale livelihoods and indigenous communities around the world, presents opportunities for resolving the current biodiversity crisis and global challenges facing humanity [12,15,23]. Finally, although often considered implicitly [13,18], nature's contributions to people are rarely considered when setting global conservation priorities, except for a few regulating ecosystem services [2,3,24].

Agro-biodiversity is a sub-component of biodiversity that accounts for the variety of life that contributes to food and agriculture [25]. In the broadest sense, it can be broken down into two components: (a) Planned agro-biodiversity, which refers to the diversity within and across species (domesticated or undomesticated) that are used by people, and (b) associated agro-biodiversity, which refers to species that surround and/or enhance planned agro-biodiversity [26]. In order to identify species and forms of high potential for improving and sustaining agriculture, research efforts have generally focused on mapping centres of origin and diversity of major domesticated plant species (mostly food crops) and associated wild relatives [27,28]. While the distribution of plants of highest importance for commodity production and human nutrition is now widely studied [6,29], much uncertainty remains about the large fraction of neglected and underutilized species that contribute to a wide array of provisioning, support, regulation and cultural services [30]. Moreover, although recent international treaties incentivized the preservation of plant genetic resources in the face of major global challenges [30], the distribution of useful plant genetic diversity is mainly studied through proxies (e.g., taxonomy, geographic distribution and environmental data) [28]. Likewise, the drivers and spatial patterns of decline in agro-biodiversity remain poorly understood [15,17].

Here we explore the relationship between hotspots of wild plant diversity and regions containing high levels of agro-biodiversity by (i) reviewing the history of the two concepts, (ii) characterizing the degree to which they overlap in space and the human and biological drivers of these geographic (in-) congruencies, and (iii) considering how to better integrate the multi-faceted aspects of useful plant diversity, including genetic and phylogenetic diversity. Finally, we propose a new general framework and discuss future avenues for obtaining an improved understanding and preservation of global hotspots of useful plant diversity [31,32].

#### **2. The History of Diversity Hotspots**

#### *2.1. Biodiversity Hotspots*

The distribution of life on Earth has long been investigated by naturalists aiming to understand, preserve and exploit the natural world [33–35]. These influential observations were followed by a more systematic documentation of the potential distribution of plant diversity globally based on the compilation of data from regional checklists and pioneering modelling efforts [36–38]. It was only in 1988 that British environmentalist Norman Myers (1934–2019) coined the term "hotspot" to define 10 tropical forest areas considered to be both irreplaceable (i.e., due to high concentrations of endemic plant species) and vulnerable (i.e., due to high rates of deforestation) [21]. Eight hotspots were added subsequently, including four in Mediterranean regions [39]. Later, the definition of a hotspot underwent a major update by considering strict quantitative criteria to designate areas containing at least 0.5% of the world's flora, but less than 30% of its original vegetation—resulting in the addition of seven additional hotspots [40]. A recent major update incorporated new data and took account of both terrestrial vertebrates and plants to delineate a total of 35 global biodiversity hotspots (Figure 1) [18]. Besides their importance for biodiversity, hotspots are home to more than two billion people and encompass some of humanity's highest population growth, as well as poverty rates [22].

**Figure 1.** Global distribution of biodiversity hotspots and Vavilov centres. Biodiversity hotspots were first defined by Myers in 1988 [21], and now comprise 35 regions of high species richness, endemism and threat, as last updated in 2011 by Mittermeier and colleagues [18]. Islands constituting biodiversity hotspots are highlighted by outer hotspot limits. Vavilov first defined centres of origin of cultivated species and wild relatives in 1924; he provided an update in 1935, comprising eight primary and three secondary centres [27].

Since the definition of the hotspot concept by Myers, several other approaches have been explored to define and refine important biodiversity areas. Alongside terrestrial plants and vertebrates, other taxa have been considered to date, including marine mammals [41], phytoplankton [42] and soil invertebrates [43]. Conservationists have also argued for the protection of remaining wilderness, considered to be the most ecologically intact areas of the world, as these have been shown to experience substantially fewer threats, and thus, contribute more to the persistence of biodiversity [44]. Whereas, early work focused on species richness, recent studies have also advocated for the consideration of functional traits and evolutionary history, which may permit a better understanding of nature's contribution to both people and nature itself [45]. Incorporating knowledge across various disciplines

is of particular importance as the multiple facets of biodiversity that they represent often do not overlap in space [20,46,47]. Spatial congruencies between biodiversity and direct measures of ecosystem services, such as carbon and water, are also increasingly being investigated [2,3].

Plants have been central to the definition of biodiversity hotspots since the pioneering work of Myers. Regions of high plant diversity have been investigated in multiple ways in the last decades, mainly by attempting to obtain finer continuous maps of (rare) plant species richness, as opposed to considering categorical hotspots. Barthlott and colleagues produced an influential estimate at the scale of ecoregions [5,48,49]. In contrast, researchers at the Royal Botanic Gardens, Kew gathered information at a semi-administrative scale [50,51], and adapted and expanded the concept of Important Plant Areas to the tropics [52], a programme that remains under development [53]. More recently, vascular plant species richness has been interpolated globally based on ~1000 local estimates [54], extrapolated based on the relationship between the occurrences of ~200,000 species and their environment [3], or investigated through the examination of commonness-rarity patterns [55]. Hotspots of high plant richness and endemism include Mesoamerica, the tropical Andes, the Amazon, Brazil's Atlantic rainforest, Central Africa, the western Ghats, South-East Asia, and many islands (e.g., Madagascar, New Guinea), mountainous regions (e.g., Alps, Caucasus) and Mediterranean areas (e.g., Cape floristic region). Despite their international recognition, most of these regions have become increasingly depleted under ongoing human pressures, whereas richness may increase at their periphery through repeated introductions in gardens and disturbed habitats [56]. These processes are leading to a global loss of diversity and increasing biotic homogenisation [57]. In this context, mapping priority areas and taxa for conservation remain, at least, as crucial and urgent as when hotspots were initially identified more than three decades ago.

#### *2.2. Agro-Biodiversity Hotspots*

Starting some 150 years ago, global botanical, geographic, linguistic, and archaeological evidence were combined to identify the geographic origins of crops, including distinctions among Old versus New World species [58]. These works were largely built on developments in plant systematics (e.g., Linnaeus (1707–1787), Alefeld (1820–1872), de Candolle (1806–1893)), phytogeography (e.g., Willdenow (1765–1812), von Humboldt (1769–1859), Wegener (1880–1930)), and evolution by natural and artificial selection (e.g., Darwin (1809–1882)). A number of these scientists were extensive travellers, whose contributions to their fields were catalysed by their voyages. However, none travelled as much as Nikolaï I. Vavilov (1887–1943). Informed by previous phytogeographic research and the rediscovery of Gregor J. Mendel's primary works in genetics, the Russian agricultural scientist pursued genetic variation in crops and their wild relatives, exploring five continents over several decades. Through his field experiences, Vavilov came to propose a set of independent "centres of origin" of cultivated food plants around the world, based fundamentally on where he saw a maximum concentration of diversity of traditional varieties of a wide range of crops, along with their wild relatives. Vavilov initially proposed three centres of origin of forms (1924), progressing to as many as eight primary centres, and including several sub-centres (Figure 1). These putative centres, which later in his tragically curtailed life he called hearths of origin, included Mesoamerica; parts of the Andes, Chile and Brazil-Paraguay; the Mediterranean; the Near East; Ethiopia; Central Asia; India; China; and Indo-Malaysia [27].

Since Vavilov, the regions of origin and diversity of different crops have been debated, investigated and refined, benefiting from an expanding body of archaeological, linguistic, genetic, and taxonomic information [6,59–66]. "Centres of diversity" came to be the preferred term over "centres of origin", to account for the difficulty in assigning an exact place of origin for most crops, and due to the understanding that high concentrations of crop varieties and related wild species are not in every case located where crops were initially domesticated [62]. "Regions" (also "megacentres" per Zhukovsky and "non-centres" per Harlan) rather than "centres" became preferred to reflect the large size of many of these geographic areas, and again, the difficulty of pinpointing exact locations where crops

were domesticated [6,67]. At the present time, multidisciplinary evidence supports the identification of ca. 24–28 different areas around the world where crop domestication occurred independently, mostly beginning in the early to middle Holocene (approximately 11,700–6000 years ago), and in a few cases more recently [66,68,69]. Not all of the identified areas would be considered by most researchers as a "centre" or "region" of origin or diversity, as only a limited number of crops were domesticated in some of these.

There was an acceleration in the movement of crop plants across the globe between 1500–1700, as they were introduced to colonizing countries, their colonies and other regions with emerging export-oriented production [9,70]. Agricultural development and globalisation have made a number of crop species available to consumers worldwide, but in turn, increased homogeneity in global agriculture [71,72]. Added to the geographic decoupling of agricultural production and consumption [73,74], this homogenisation has deteriorated the connection between crops and their geographic origin [6]. Nevertheless, these areas of origin continue to hold foremost importance with regard to crop genetic diversity as their crops diversified for thousands of years under natural and human selection, including via further introgression with wild relatives [75,76]. During Vavilov's voyages a century ago, it was already apparent that the diversity of crops that people grew and consumed was changing as a result of globalisation. Major efforts commenced, particularly during the 1970s and 1980s, to collect traditional crop landraces and wild relatives for safeguarding in genebanks, and to also support in situ conservation, often in collaboration with subsistence agriculturalists [77]. Such efforts continue today, often linked to seed banking and germplasm collections, based on the recognition of persisting gaps in conservation of agro-biodiversity [28,78].

The history of agro-biodiversity was primarily written by colonial powers and white male explorers, conferring little or no room to traditional knowledge holders [79]. Moving ahead in tackling the challenges of mapping, understanding, protecting and further exploring the potential of crop plants and associated wild relatives, it is crucial that benefits of this work are shared in equitable ways [80]. In particular, access must be ensured in low-income countries to new or neglected crops, especially those that offer climate resilience, nutritious contents and other desirable traits.

#### **3. From Biodiversity to Agro-Biodiversity Hotspots: A Geographic Continuum**

The geographic distribution of biodiversity and agro-diversity hotspots has long been investigated, but rarely together. Very little is known about their spatial intricacies despite recent calls for their integration [31]. Several areas of high plant diversity do not include primary regions of agro-biodiversity (e.g., California, Caribbean, Brazilian Cerrado, South Africa, Madagascar, Pacific islands) and a few agro-biodiversity regions are not recognized as diversity hotspots (e.g., large parts of Eastern Asia and India) (Figure 1). Although human activities are altering this pattern [12,14], global plant species richness generally decreases with increasing latitude [5,54], reflecting past environmental changes, land configuration and the evolutionary histories of species [81,82]. On the other hand, the history of global human migrations, civilisations, economy and cultural preferences have been profoundly intertwined with the distribution and availability of natural resources (species richness, abundance and properties) to shape regions of agro-biodiversity [9]. Here, we present a new set of analyses to explore, illustrate and discuss the spatial congruence between biodiversity and agro-biodiversity hotspots, and its relationship with two important processes—human selection of species and species evolutionary history.

#### *3.1. From Popularity to Anonymity*

Vavilov provided an early spatial representation of the origins of cultivated plants by mapping the distribution of major food species. However, by restricting (justifiably) his focus mainly to selected food plants encountered in his field experiences, Vavilov was unable to produce a comprehensive assessment of the distribution of all plant species selected for use by humans across the world. Identifying which plant species are most selected by humans for a broader range of uses in addition

to food, and characterizing their geographic distribution is, thus, key for defining agro-biodiversity hotspots and for relating their variation to the wider context of biodiversity.

In our currently globalized world and big data era, it is now possible to investigate human preferences for species more extensively. Here we assessed the popularity of most vascular plants on the free online encyclopaedia Wikipedia (www.wikipedia.org), using search data as a proxy for cultural preference, knowledge of plant species, use by humans and domestication intensity. As Wikipedia can be edited by anyone at any given time, it cannot be considered a reliable source of information without critical evaluation. However, our analysis remains independent from the quality of Wikipedia articles as it only examines the interaction between users and the web platform by quantifying numbers of page views. Species names and occupied geographic regions were retrieved for 339,924 species from the World Checklist of Vascular Plants (WCVP) [51]. Wikipedia uses the taxonomic backbone of the WCVP, so no additional name matching was performed. Geographic regions were retrieved at the national or sub-national level (finest level three) of the World Geographical Scheme for Recording Plant Distribution (WGSRPD), which was developed by the International Working Group on Taxonomic Databases for Plant Sciences [83]. We retrieved species popularity on Wikipedia as measured by the number of page views over the period 1 January 2016 to 1 January 2020 using the R package pageviews. This information relates to English Wikipedia pages only as it is the language with the highest number of pages overall and often the source of translations into other languages. We ranked these data according to three categories: (1) The 1000 most popular species; this includes plants used as food, medicine, timber, ornamentals and cosmetics (Supplementary Materials), (2) the remaining species covered by Wikipedia (i.e., those with an available page), (3) the remaining species not documented in Wikipedia (i.e., those without a page).

The global richness distribution of the 1000 most popular species is strikingly similar to the Vavilov primary centres with particularly high richness in subtropical regions of the Northern hemisphere (Figure 2a). There are also expansions towards temperate areas (i.e., Eastern North America, Europe and Central Asia), while the Northern Andes, Eastern Africa and the Indo-Malayan regions have relatively low crop richness at a global scale, but high richness within their respective continents. Richness generally increases towards the tropics for species that are documented in Wikipedia but fall outside of the 1000 most popular category, with particularly high concentrations in the Mediterranean and subtropical regions that are characterized by high plant species endemism but low richness in major crops (e.g., Western North America, South Africa, Australia) [5] (Figure 2b). In contrast, richness in species not documented in Wikipedia tends to follow a latitudinal gradient similar to that observed for total plant diversity and most similar to biodiversity hotspots (Figure 2c) [18,84]. These findings highlight the relationship between the spatial structure of biodiversity and agro-biodiversity, and people's knowledge, perception and use of nature. Indeed, we observe the existence of a geographic continuum between popular plant diversity that may be more intensively used by humanity (based on the existence of a Wikipedia page; 2A) and anonymous plant diversity that may be less heavily used (based on the lack of a Wikipedia page; 2C). One artefactual limitation in our assessment of the distribution of plant popularity is the general over-representation of English-speaking regions (e.g., United States, Canada, Australia) given that our data extraction came from English Wikipedia pages only. This also likely explains relatively low values of popular species richness in regions, such as the Andes or Ethiopia (Figure 2a), although those are still visible at the continental scale.

**Figure 2.** Global distribution of the species richness of plants ranging from popular to anonymous based on English Wikipedia page views. Global distribution of (**a**) the 1000 most popular plant species on Wikipedia; (**b**) 49,019 species documented in Wikipedia, excluding the 1000 most popular ones; (**c**) 280,905 species not documented in Wikipedia. Popularity was measured as the number of views of the Wikipedia webpage of each species. Native distribution data was retrieved from the World Checklist of Vascular Plants at the national or sub-national level of the World Geographical Scheme for Recording Plant Distribution (WGSRPD) [51].

#### *3.2. From Domesticates to Wild Relatives*

Alongside major crops, Vavilov was also interested in documenting, mapping, collecting, using, and preserving wild ancestors and closely related species [27]. By considering less known and undomesticated (or less domesticated) parent and sister species, Vavilov directly connected his definition of regions of origin/diversity with wild plant diversity. Recent studies illustrate this link between biodiversity and agro-biodiversity hotspots; these assessed the distribution of the closest and more distant relatives of major crops and found high species richness in plant diversity hotspots, such as the Brazilian Cerrado and Atlantic Forest or South-East Asia [28,85].

Here we assessed the current distribution of 222 major international crops and 2,731 of their wild relative species using comprehensive lists from the USDA ARS GRIN-Global Taxonomy and geographic data from the World Checklist of Vascular Plants, again retrieved at the national and sub-national level (level three) of the WGSRPD [51]. Crop wild relatives are classified across three gene pools based on both their relatedness (using phylogenetics and systematics) and crossing ability with the crop [86,87]: gene pool one comprises the most closely related (even conspecific) wild species that are generally fully interfertile with the crop; gene pool two includes more distant relatives that may be crossed to the crop with more difficulty; and gene pool three typically contains the most distantly related and least compatible species within the genus to which the crop belongs (sometimes including other genera). Here, we assess changes in geographic patterns across a gradient from cultivated species to their closest wild relatives to their more distant wild relatives, by mapping richness for crops and each associated gene pool separately. When more than one species was identified in a gene pool for a given crop, we merged their distribution to assign the same weight to each crop, thus, avoiding overrepresentation of genera with many wild relatives. Geographic data were not available at the infra-specific level (i.e., sub-species, varieties, forms) for all taxa, and so we performed analyses at the species level.

Geographic patterns in major crop species richness strongly overlap with the Vavilov centres (Figure 3a) and are also similar to the 1000 most popular plant species on Wikipedia (Figure 2a). Given that gene pool one is composed of the closest crop wild relatives, including progenitors and/or wild types of the crop species, the distribution of gene pool one species richness is very similar to that of the crops (Figure 3b). Slight increases and decreases are respectively observed inside and outside Europe, which may be explained by more extensive documentation of European crop wild relatives compared to those that occur in other regions [88]. Gene pools two and three provide a more diffuse representation of the Vavilov centres: species richness decreases in most of the Vavilov centres, but increases in surrounding regions, particularly (but not exclusively) towards the tropics and plant diversity hotspots (Figure 3c,d). Although centres of crop diversity remain visible when mapping secondary and tertiary gene pools, the geographic signal diminishes as we move further away from the most explored and popular branches of the tree of life. In contrast, areas of high biodiversity start to emerge, which is even more striking when species richness mapping does not account for the over-representation of genera with many wild relatives [28,85]. This reinforces the existence of a geographic continuum between agro-biodiversity (i.e., widely used and cultivated crops) and biodiversity (i.e., non-domesticated and less used sister species of crops) related to the strength of the interaction between humans and plants, but also plant evolutionary history.

**Figure 3.** Global distribution of the richness of major crops and their wild relatives. Global distribution of (**a**) 222 major crops, (**b**) 361 wild relative species in gene pool one, (**c**) 1040 wild relative species in gene pool two and (**d**) 2358 wild relative species in gene pool three; see text for gene pool characterisation. Species identities and gene pool classifications were retrieved from the USDA ARS GRIN Global Taxonomy. Distribution data was retrieved from the World Checklist of Vascular Plants at level three of the World Geographical Scheme for Recording Plant Distribution (WGSRPD) [51]. When more than one wild relative species was identified for a crop in a gene pool, we merged their distribution to assign the same weight to each crop and avoid genera with many wild relatives to be overrepresented.

#### **4. Integrating Genetic and Phylogenetic Diversity into Agro-Biodiversity Hotspots**

The designation of hotspots of plant species richness and rarity has provided a way to focus attention on the intrinsic value of biodiversity at a global scale. However, the contributions of plants to livelihoods are rarely considered in this context, despite the predominance of utilitarian arguments in conservation. Going beyond species counts (i.e., taxonomic diversity) to describe and understand structural and/or chemical properties of species and their diversity is of critical importance. The wide range of plant properties cannot currently be quantified across all species but is often related to genetic variation within and among species [89]. Characterizing genetic and phylogenetic diversity and their geographic distribution may, therefore, provide a useful framework for identifying other currently unknown forms of diversity and associated usages, and for preserving plant genetic resources.

#### *4.1. Hotspots of Phylogenetic Diversity*

Some species are the sole remaining representative of ancient lineages, while others are part of recent and rapid radiations (i.e., increases in species richness related to elevated speciation rate) that may comprise hundreds of closely related species. Therefore, conservation biologists now account for the fact that the evolutionary histories of species are not equivalent when setting priorities [90]. To address this inconsistency, phylogenetic diversity (PD) was proposed as an approach based on evolutionary information [91]. The PD of a given area is equal to the sum of all the branches on a phylogenetic tree linking the set of species that occur in this area. Areas containing high PD will,

therefore, reflect higher concentrations in distantly related species. While previous studies have cast doubt on the ability of PD to provide a different answer to species richness for prioritisation [92,93], the decoupling of biodiversity patterns, based on taxon richness and evolutionary history, have since been clearly demonstrated (e.g., Reference [94]). Phylogenetic diversity and associated metrics have been widely used to explore biodiversity patterns among and within biodiversity hotspots [95–97].

One of the most important characteristics of PD is its potential to act as a surrogate for feature diversity (i.e., the diversity of characters or traits of species), which encompasses the qualities of plants that are beneficial for humans. While the relationship between PD and feature diversity remains contentious [45,98], it is nevertheless an attribute that is of particular importance for the identification of areas rich in crop wild relatives and species with unexplored uses for humans. Assuming that PD is a suitable surrogate for feature diversity, maintaining PD would not only help retain the evolutionary potential of species, but also maximise the potential unanticipated benefits that biodiversity may have in the future for humans (i.e., biodiversity option values), particularly in the face of global change [99,100]. In the context of agro-biodiversity hotspots and the identification of new sources of plant properties, PD as a metric of choice has a key role to play (Figure 4a). Ultimately, however, the identification of hotspots of agro-biodiversity would be best served by the integration of various metrics capturing the multiple facets of biodiversity [101], and also by considering the human dimension for better understanding the portion, intensity and modes of use of biodiversity by humanity.

**Figure 4.** A proposed general framework for the further inclusion of genetic information into the mapping of agro-biodiversity hotspots. (**a**) A relationship between species evolutionary history and their physical or chemical properties (i.e., features or phenotype) would use phylogenetic diversity as a proxy for feature diversity, the latter being less readily quantifiable across wide ranges of species, regions and features. Combined with species distribution data, phylogenetic diversity could ultimately identifies hotspots of feature diversity and priority areas for the conservation of species' contribution to people; (**b**) By considering phylogenetic information together with population genomics, cytology, life

history and/or ecological data, an Estimated Breeding Value (EBV) could be computed for crop wild relative species and/or populations. Combined with distribution data, hotspots of EBV could then be mapped to identify areas containing high concentrations in valuable wild gene sources for preservation and crop improvement.

#### *4.2. Hotspots of Breeding Value*

Genetic diversity represents the raw material that humans have relied upon for millennia for the maintenance and improvement of crops. Plant breeding is a long-term process [102] aimed at enhancing traits of interest (e.g., yield, quality, disease tolerance, abiotic stress tolerance) using extant variation [103]. There is tremendous genetic and phenotypic diversity in crop wild relatives distributed across the plant tree of life [69,104]. Identifying the top priority branches (from species to populations) that will generate the largest changes in trait values, while having the closest form to current crops is of great interest to plant breeders and agriculturalists. Populations that exist at the edges of distribution ranges may have great utility for breeding as these often occur in more isolated and extreme ecological conditions and display high levels of genetic and phenotypic differentiation [1,66,78,105]. Identification of potentially useful crop wild relatives is generally based on heuristic approaches (e.g., place X has a similar environment to place Y, so translocation is expected to lead to positive results), or by using large-scale germplasm collections to search for specific traits of interest [106,107]. However, these approaches, as currently applied, may not fully explore available plant genetic diversity (e.g., Reference [108]) and are often not available for crops that are less economically important.

There have been recent efforts to identify the geographic regions where crop wild relative species are concentrated [28]. Characterisation of the distribution of populations that are most compatible with existing crops and exhibit phenotypes of interest has somewhat lagged behind. Nevertheless, there are now concerted efforts to identify and incorporate these taxa into breeding programmes for producing viable new cultivars by using integrative approaches that leverage large amounts of data from phylogenetics, population genomics, cytology, life history, ecological niches, and predictions of future environmental conditions that crops may experience [78,109–111].

Building on historic advancements, it may be possible to incorporate the data currently used in heuristic approaches for crop wild relative identification into a more general framework to help the decision-making within breeding programmes. Such a framework could potentially be modelled from the Estimated Breeding Value (EBV), a common breeding programme metric that could expedite selection of donor species. An EBV is the potential of an individual as a genetic parent, considering the heritability of a given trait under selection [103]. Typically, EBVs are obtained from narrow breeding populations of a single species by multiplying the narrow-sense heritability (calculated either by variance decomposition or parent-offspring regression) by the difference between the parent performance and population mean, which provides an estimate of how the progeny of a specific parent will perform relative to an average parent. We propose that EBVs could also be calculated for crop wild relative species and populations, as long as the phenotype of interest is clearly defined. Applied to species and populations instead of individuals, the EBV could go beyond heritability to additionally incorporate biological factors (e.g., ploidy, mating system), evolutionary factors (e.g., phylogenetic relationship), and ecological factors (e.g., species environmental niche) in a hierarchical way for prioritizing species that may be of greater potential use to plant breeding. Functionally, this proposed use of EBV would produce a ranking of species for an individual breeding programme, based on the desired phenotype by crossing interaction. Ultimately, characterizing this variation in utility across species (and populations) may help identify priority areas for in- and ex-situ conservation related to specific breeding targets (Figure 4b).

#### **5. Towards a Unified Concept of Agro-Biodiversity Hotspot**

Hotspots of both biodiversity and agro-biodiversity have long relied on counting numbers of species (i.e., species richness; Figure 5a) and assessing threats (Figure 5b). While biodiversity scientists have mainly focused on numbers of rare species (including many narrowly distributed taxa) for conservation, agronomists have been interested in diversity within gene pools (i.e., numbers of domesticated species and wild relatives) [27,28,85] and within crops (e.g., numbers of landraces) [112]. Although taxon counts remain extremely useful, new approaches are now proposed to account for the multi-faceted nature of (agro-)biodiversity, such as functional diversity accounting for the diversity in species chemical properties and eco-/agri-system functions [113], phylogenetic diversity as a potential proxy for functional and property diversity (Figure 5c) or for identifying gene sources for breeding programmes (Figure 5d). As highlighted by their seminal definition from N. Myers and recent publications [114], biodiversity hotspots are also deeply related to the distribution of a wide range of threats, many of which are shared with agro-biodiversity (e.g., land use and climate change, pollution, biological invasions, over-harvesting), but less formally included in the geographic assessments of the latter (Figure 5b) [12,16]. Although the different facets and threats of agro-biodiversity are not all expected to overlap geographically, our paper proposes to assess them jointly rather than separately.

**Figure 5.** The conceptual framework for the identification of agro-biodiversity hotspots, including (**a**) plant species richness (applicable to infra-specific levels as well); (**b**) threats; (**c**) species (or infra-specific taxa) evolutionary and features diversity; (**d**) crop wild relatives estimated breeding value; and accounting for (**e**) the geographic continuum between hotspots of wild species diversity and regions containing high concentrations in major crops (i.e., highly domesticated species), and its environmental and human drivers. The map does not provide a new estimate of the distribution of agro-biodiversity hotspots, but rather illustrates a combination of the potential two ends of the domestication spectrum: Major crop species richness (Figure 3a) and species richness undocumented in Wikipedia (Figure 2c). High to low species richness is represented from red to blue.

Primary regions of agro-biodiversity have focused on relatively few important crops selected by researchers, whereas they have mainly explored wild relatives of these domesticated species, which are not always considered for other uses than crop improvement. Given the existence of a wide domestication spectrum, ranging from major global crops (i.e., highly domesticated species) to those harvested in the wild [16], we believe that further work on regions of agro-biodiversity should

expand the focus to better include the long list of plants that provide food and other cultural benefits to humanity (Figure 5e). This will be made effective through the documentation of the tremendous diversity of neglected and under-utilized species of the world (with more than 30,000 useful plant species known to date [115]). Many of these species occur naturally in low-income countries, including already established biodiversity hotspots, which are also often home to large human populations and cultural diversity [22,116]. Understanding the drivers of the distributions of nature's contribution to people across the domestication spectrum (from climate and land use to socio-economic factors; Figure 5e) is also fundamental to define hotspots and design conservation and development efforts to sustain socio-environmental sustainability.

The recognition of biodiversity hotspots and agro-biodiversity (Vavilov) centres have played important roles to raise public awareness, foster research and attract political action to preserve and use natural resources sustainably. Given the urgency and magnitude of the global challenges outlined by the United Nations' Sustainable Development Goals and the recent report on biodiversity loss by the intergovernmental science-policy platform on biodiversity and ecosystem Services (IPBES) [12], it is more important than ever to refine, integrate and disseminate such powerful concepts. Failing to protect hotspots of natural resources, and especially agro-biodiversity, would have damaging consequences on nature and human livelihoods, both at those centres and in their peripheries. Our paper calls for the further development and integration of a range of commonly used and more recently proposed indices, while accounting for the key interaction with biodiversity, into the agro-biodiversity hotspot concept.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/9/9/1128/s1.

**Author Contributions:** Conceptualisation, S.P.; data analysis, J.S.B., I.O., S.P.; data curation, C.P., C.K.K., R.G.; writing—original draft preparation, C.K.K., F.F., M.B.K., N.F., S.P.; writing—review and editing, J.S.B., C.P., C.K.K., M.B.K., M.S.G., J.V., F.F., A.A., S.P.; visualisation, I.O., R.D., N.F., M.B.K., S.P.; supervision, A.A., S.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** S.P., R.D. and I.O. received funding from Norway's International Climate and Forest Initiative (NICFI). J.B. received funding from [BB/S014896/1] GCRF BBSRC Landscape genomic-environment diversity data to model existing and novel agri-systems under climate change to enhance food security in Ethiopia. A.A. received funding from the Swedish Research Council, the Swedish Foundation for Strategic Research, the Knut and Alice Wallenberg Foundation, and the Royal Botanic Gardens, Kew.

**Acknowledgments:** We thank Justin Moat and Amanda Cooper for their help with designing the maps; Nicholas Black, Alan Paton, and Robert Turner for their assistance with data from the World Checklist of Vascular Plants; and four anonymous reviewers for constructive suggestions that helped improve this manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Why Seed Physiology Is Important for Genebanking**

**Katherine J. Whitehouse 1, Fiona R. Hay 2,\* and Charlotte Lusty <sup>3</sup>**


Received: 1 April 2020; Accepted: 20 April 2020; Published: 2 May 2020

**Abstract:** Genebank management is a field in its own right; it is multifaceted, requiring a diverse set of skills and knowledge. Seed physiology is one area that is critical to the successful operation of seed genebanks, requiring understanding of seed quality during development and maturation, seed dormancy and germination, and seed longevity in storage of the target species. Careful management of the workflow between these activities, as seeds move from harvest to storage, and the recording and management of all relevant associated data, is key to ensuring the effective conservation of plant genetic resources. This review will discuss various aspects of seed physiology that genebank managers should be aware of, to ensure appropriate decisions are made about the handling and management of their seed collections.

**Keywords:** agrobiodiversity; genebank; genebank management; plant genetic resources; seed physiology; seed quality management

#### **1. Introduction**

Storing seeds in genebanks is the most effective way of conserving and sharing most of our existing agrobiodiversity (that of orthodox species). It is also a relatively simple activity; seeds are dried and stored at low temperature. However, within that simple statement lies a whole series of operations, which, if not carefully followed and controlled, risk the loss of the agrobiodiversity that genebanks are meant to be conserving. Effective management of the individual operations and of the whole workflow, especially when managing a crop genebank with thousands of accessions, requires some understanding of seeds and how they might be impacted in terms of their quality and viability during the different production and processing steps. In this review, we discuss various aspects of seed physiology that we consider particularly relevant to genebanks, including understanding whether a seed is viable or just dormant and how dormancy might be overcome; the acquisition and loss of the ability to tolerate desiccation during seed development; seed ageing leading to loss of viability; understanding how long seed lots will remain above a threshold viability level based on genebank viability monitoring data, modelling, and comparative longevity experiments; how seed longevity might be affected by various pre- and post-harvest factors; and the statistics of analysing seed germination/viability data. The review is directed primarily towards genebanks that conserve agrobiodiversity; the science of seed banking to conserve species diversity has been reviewed elsewhere [1,2]. Some wild species seed banks invest significant resources into researching the seeds of the species they conserve. In contrast, many agricultural genebanks do not have a mandate to do much research and, in the interests of efficiency in genebank operations, avoid collecting data that could be classified as non-essential to managing their collections. We nonetheless touch upon various areas where genebanks could collect data that would be of interest and use to the network of genebanks and to the scientific community. In

relation to this, we also encourage renewed interaction between genebanks and the seed science and testing community.

#### **2. Understanding Why a Seed Lot Is Showing No or Low Germination**

Air-dry seeds of some orthodox species are predicted to survive for very long periods at sub-zero temperatures, typical of long-term genebank storage conditions (−20 ◦C) [3]. However, many genebanks regenerate [regeneration is used to describe the process of planting seeds to produce more seeds for genebank storage, be that as a result of declining viability, where the process might also be called 'rejuvenation', or to increase the number of seeds available ('multiplication')] their material more frequently than longevity predictions would suggest is necessary (unpublished data). This may be because insufficient seeds are being placed into storage [4], because seeds are being distributed at higher rates than expected or a viability monitoring result indicates low viability (i.e., below the acceptable threshold), or simply for reassurance that the seeds in storage, because they are 'fresh', must be viable. Indeed, some accessions may be regenerated frequently because it is difficult to produce seeds in large quantities in one season (because of, for example, a small plot size, poor plant establishment, or low fecundity) and/or because the initial germination result is poor. There are two fundamental reasons why a seed lot might show no or low germination, either in the initial germination test or in a viability monitoring germination test, including seed dormancy and/or non-optimal germination conditions or low seed viability. Indeed, the most important first consideration is to understand whether or not the result is an accurate reflection of the viability of the seed lot.

#### *2.1. Viability versus Germinability*

Ultimately, it is only when a seed germinates that we can be sure that the genetic information it contains can be expressed or exploited, be that for crop production or for scientific research. Thus, a germination test is considered the ultimate method for monitoring 'viability', even though it requires knowledge of dormancy-breaking treatments and optimum germination conditions. However, there are other ways in which the viability of a seed lot can be assessed. At the end of a germination test, any seeds that have not germinated could be assessed via a 'cut-test' to see if they are dead seeds (soft, mouldy); seeds that have not taken up moisture (hard-seededness, a trait that is particular prevalent in seeds from species in families such as the Fabaceae and Malvaceae); empty seeds; or seeds that still have firm, normal-coloured tissues. Other viability tests would be made on a separate sample of seeds. The International Seed Testing Association has a whole chapter of their *International Rules for Seed Testing* dedicated to the use of the tetrazolium test to rapidly assess seed lot viability, based on the living parts of the viable seed staining red, with optimized procedures for specific species or genera. The method usually involves imbibition of water, seed sectioning, and exposure to a solution of 2,3,5-triphenyl tetrazolium chloride or bromide for a specified period of time [5]. How to interpret the results (staining pattern) is also described [5]. The tetrazolium test is often criticised for being somewhat subjective; therefore, it should only be used instead of a germination test when it is not possible to remove seed dormancy. The species and genera included in the *International Rules for Seed Testing* tetrazolium chapter [5] cover the most important crops and wild species conserved by genebanks. Fluorescein diacetate has also been used as a vital stain for seeds, in particular for orchid seeds, which are very small and would thus be difficult to evaluate using tetrazolium [6,7], but also for larger seeds [8]. The methodology is similar to the tetrazolium test, except viability is assessed by observation under a fluorescence microscope, making this method less attractive than the tetrazolium test, which generally does not require any specialised equipment.

If the viability test result is consistent with the germination result, the two test results together can be considered reliable. Indeed, this is how the tetrazolium viability test is used in some genebanks, particularly those working on wild species, where the dormancy mechanism of different and diverse species may not always be known. When the two results are comparable, then the dormancy-breaking treatment(s) and/or germination test conditions are considered appropriate. If the viability result

is higher than the germination result, further dormancy-breaking treatment(s) and/or germination conditions may be tried, if there are sufficient seeds available. It is also possible that the viability result is less than the germination result, in which case the tetrazolium test conditions may not be optimal. Some genebanks systematically use the tetrazolium test for viability monitoring, perhaps because it can ultimately be more straightforward than lots of different, multi-step germination tests. This is not recommended, except perhaps under special circumstances.

#### *2.2. Seed Dormancy and Germination Behaviour*

A non-dormant seed will readily germinate given appropriate environmental conditions, most importantly, moisture and temperature. The optimum temperature range for germination can vary between species, but is likely to reflect the ambient temperature in the habitat where the species naturally occurs or is typically planted, and at the time of year when the plant is expected to germinate [9,10]. In contrast, dormancy mechanisms can be multifaceted and diverse, although still climate-related [9,10]. Species-specific information on germination protocols and dormancy breaking treatments are widely available, again particularly for crop species and their wild relatives [5,10–13]. It is, however, important to be aware that, in some species, dormancy may be released or induced during storage, which may mean that the germination procedure initially used may have to be adapted following storage [14–16].

Where dormancy is suspected, because a cut test indicates that seeds are probably still viable at the end of a germination test, further germination tests with different dormancy-breaking treatments should be performed. The same literature as described in the previous section will help guide what treatments should be prioritized for testing, for example, if the 'problem' species is closely related (same genus) and comes from a similar environment as a species for which dormancy behaviour is documented. Complex, multi-factorial experiments are unlikely to be conducted in genebank laboratories (Table 1), unless there is a plan to publish the findings, in which case a more structured experimental design that can be appropriately analysed (see Section 4 below) might be better received. Nonetheless, it is helpful to share any such 'new' data on the germination requirements of previously unstudied species—precisely because it is useful information for other genebanks and for end-users. Indeed, most genebanks give germination advice when they distribute seed samples, if the information is available.

#### *2.3. Desiccation Tolerance*

Not all plant species produce seeds that tolerate the drying procedure, which is one of the steps in preparing seeds for genebank storage, and seeds of some other species, even if desiccation-tolerant, do not survive for long periods under low-temperature storage. However, the proportion of species that show this non-orthodox seed storage behaviour is low (8%, [17]), and of course, species that are already effectively managed and conserved in genebanks are clearly orthodox. However, seeds of orthodox species can be desiccation-intolerant if they are harvested too early or, conversely, if germination has commenced, for example, if the seeds are prone to vivipary. Seeds acquire physiological traits, including desiccation tolerance, during development. The relative timing of the acquisition of desiccation tolerance varies between species, but is normally acquired by or around the time that seeds reach mass maturity, when the seeds stop accumulating dry matter, and thus reach their maximum dry weight ([18] and references therein). Because dry weight will be reflected in the fresh weight of the seeds and other physical characteristics of the seed, it is perhaps unlikely that seeds of an orthodox species will be harvested before desiccation tolerance is acquired, unless the species flowers and fruits asynchronously. For species with such indeterminate flowering, it may be difficult to avoid harvesting seeds that have yet to acquire desiccation tolerance, and this may be a reason that a seed lot shows poor germination in the initial viability test. It may be possible to remove the desiccation intolerant seeds in the seed lot, by sorting based on density, size, and/or appearance. Alternatively, slow drying the whole seed lot or pre-sorting the seeds according to likely maturity and then slow drying the least mature seeds may allow maturation events to continue in the least mature seeds [19].

**Table 1.** Examples of how genebank operations might differ if resources are limited and there is a need for high levels of efficiency compared with what could be done if resources are available, which would be more sound and/or beneficial from a scientific perspective.


Vivipary is a characteristic of some species or varieties whose seeds have little or no dormancy as the seeds mature on the plant. If the seeds are exposed to germination-promoting conditions, in particular, high moisture owing to, for example, late rain, then the seeds may start to germinate while still on the plant [20]. Vivipary is common in tomatoes and peppers, as well as some cereals (e.g., temperate rice). Upon germination, orthodox seeds lose the ability to tolerate desiccation. Imbibition of water alters the physical properties of the membrane lipid bilayer, making the cellular membranes susceptible to injury upon subsequent desiccation [21–23]. Furthermore, as metabolism is reinstated upon imbibition, subsequent desiccation can lead to the initiation of further lipid peroxidation (see below) in the dehydrated tissues, which can continue upon re-imbibition [21]. Hence, any seeds showing on-plant germination will have to be discarded; indeed, it may not be possible to identify all the seeds that have started to germinate and thus lost desiccation tolerance, meaning that the entire seed lot may have to be rejected.

#### *2.4. Seed Ageing*

Seed ageing is inevitable, regardless of the conditions under which the seeds are stored, though of course, the aim of genebanking is to slow down the rate of ageing so as to preserve the genetic integrity of the seeds and to make sure the genetic information is available for use. Seeds are hygroscopic, meaning they exchange water with their surroundings until they reach equilibrium. The tendency of water to move into the tissues from the external environment is dependent upon the relative humidity (RH) of the atmosphere and the moisture content of the seed. It also depends on the chemical composition (oil content), size, and seed coat properties. Seeds with a higher oil content have a lower moisture content at a specific RH compared with lower oil content seeds. How a seed interacts with water can be explained by sorption isotherms, the shape of which reflects the availability of water within the seeds to support different chemical reactions [24–27]. After the attainment of mass maturity, that is, the end of seed-filling when seeds have reached maximum dry weight, seeds equilibrate with the ambient environmental conditions, usually by losing moisture, unless conditions are very humid. Nonetheless, they may still be harvested at a moisture content conducive to high rates of ageing

reactions. This is why it is critical that seeds are dried as quickly as possible following harvest, in order to reduce the rate of ageing reactions (unless slow drying may allow the continuation of the acquisition of desiccation tolerance, as discussed above).

The type and rate of ageing reactions depend on seed moisture content, temperature, and oxygen availability [28]. For example, respiration will only occur in the presence of oxygen and at a high enough RH that the cytoplasm allows molecular motion. As a result, seeds can be classified as either undergoing wet or dry ageing, determined by the viscosity of the cytoplasm (low viscosity at RH ≥ca.75% (wet ageing) and highly viscous/glassy state at RH <ca.75% (dry ageing)). The difference in the types of reactions that occur during these two states differentially impacts hormonal regulation and signalling pathways involved in longevity [29]. The major contributors to seed deterioration, resulting in loss of membrane integrity, reduced energy metabolism, protein carbonylation, impairment in RNA and protein synthesis, and RNA degradation, are lipid peroxidation and free radical accumulation [30]. DNA damage inhibits effective transcription and replication [31], key processes that are activated during the imbibition stage of germination [32,33]. In this stage, repair mechanisms are activated, in order to restore a functional physiological state [32,33]. However, if damage has accumulated above a certain threshold, seeds will begin to show a delay in germination, as it takes longer to repair all the damage. This explains why seed lot vigour declines and the proportion of abnormal seedlings increases, before complete loss of viability [20], although the temporal pattern from the origin of deterioration to death, either within an individual seed or a seed lot, is still not fully understood [29]. It is important to note however, that if vigour and/or abnormal germination was used to indicate the extent to which a seed lot has already aged, there may be an expectation that accessions be regenerated at an earlier stage of the ageing process. This could lead to more frequent regeneration, which would not be encouraged owing to the inherent costs and risks. Thus, better understanding of the vigour of accessions and how it changes during storage is again perhaps something that is more desirable from a scientific perspective and/or of interest to end-users, rather than being something that genebanks should aspire to assess routinely (Table 1).

#### **3. Understanding How Long a Seed Lot Will Continue to Show Good Germination and the Factors That Influence That Period**

Thresholds for the acceptable percentage viability of a seed lot have been defined in various standards [34,35]. Understanding how long a seed lot is likely to maintain viability above the threshold value will improve the efficiency of genebank operations. For example, in some cases, monitoring intervals could be extended beyond 5 or 10 years, which are the 'default' intervals for species for which evidence of seed longevity is lacking (but which are nonetheless 'expected' to be short- or long-lived, respectively, in long-term genebank storage, at −18 ◦C, in hermetically closed containers [34]). This will save use of seeds, resources, and particularly staff time. Conversely, expectations regarding seed longevity may also highlight where there might be problems owing to biology, perhaps, or where interventions might be appropriate in the seed harvesting–processing–storage chain.

#### *3.1. Historical Viability Monitoring Data*

A number of genebanks have now published historical viability monitoring data for seeds in medium- and/or long-term storage, including the United States Department of Agriculture [36]; the National Agrobiodiversity Center of South Korea [37]; the Centre for Genetic Resources, the Netherlands (CGN [38,39]); the International Rice Research Institute (IRRI [40]); and the International Livestock Research Institute (ILRI [41,42]). Some of these publications highlight the problems of analysing such data, including changes in protocols and/or storage conditions, failure to overcome dormancy, and censoring. Testing of seed lots in genebank storage often ceases once the seed lot is 'replaced' by newly harvested seeds following a cycle of regeneration (literally or metaphorically in that a seed lot may simply cease to be actively managed; Table 1). Thus, the data are censored and most of the data collected have values between 100% and whatever the viability standard might be, typically 75% or 85%, and higher if seed lots are primarily replaced owing to low seed quantity. This makes the data difficult to analyse using methods such as generalized linear modelling (e.g., probit analysis) of germination percentage data, which might otherwise be used (see below).

Nonetheless, these papers in general both confirm the effectiveness of seed genebanks as a means of conserving agrobiodiversity, and perhaps flag accessions that are likely to have shorter seed longevity, for example, accessions of carrot, parsnip, and onion [36] or temperate varieties of rice [40], or accessions that show unexpected behaviour in response to genebank storage, as reported for wheat and barley [39]. In the case of the analysis of the data from the IRRI genebank, it was not deemed necessary to change the monitoring interval, largely because there appeared to be a trend of declining longevity of the seeds being placed into storage, for seeds harvested between 1992 and 2001 (the last year's harvest covered by the analysis was 2002 [40]). On the other hand, CGN concluded that they could delay the first monitoring test to 25 years after harvest (seeds in long-term storage at −20 ◦C [38]). There are also plans to adjust retest intervals, according to genus at the ILRI genebank, to one-third of the time predicted for viability to fall to the viability standard, where those predictions are reliable (Y. Woldemariam and J. Hanson, pers. comm.). The revised intervals will be incorporated into the genebank management software used at ILRI.

From personal experience, downloading, sorting, and analysis of genebank monitoring data is not a straightforward exercise, in part because of the length of time over which the data have been collected and entered (or not) into the database—or different databases—by different staff, with a range of associated errors that accrue over time. In future, this process should not constitute a 'major exercise'; rather, genebank managers should be prepared to 'regularly' analyse data or even have some sort of analysis continuously running in the background, within the genebank management software, though the basis of that analysis (algorithms to handle the data, analytical approach) is not known; machine learning may help in this process [43].

#### *3.2. Seed Longevity Predictions*

Orthodox seeds are not simply defined as those seeds that tolerate desiccation; they are those seeds whose longevity systematically increases in response to a reduction in moisture and temperature [44]. The only longevity model currently available to describe these relations is the Ellis-Roberts viability equations (VEs) [45], which are based on fitting negative cumulative normal distributions to the proportion of viable seeds within samples of seeds stored under different constant moisture and temperature conditions (usually achieved by equilibrating seeds in a particular environment and then sealing them inside air-tight containers, such as aluminium foil bags). The equations have species-specific parameters that quantify the inherent longevity of seeds of a particular species and how longevity changes as moisture content changes. Two further parameters describe the effect of changes in temperature on seed longevity, but the temperature response is thought to be similar across all species [46]. To date, the species-specific parameters have been solved for approximately 70 species, mostly agricultural crops [13]. This means that genebank managers can predict the longevity of seed lots of these species, based on the initial germination test result, for example, using the seed viability constants menu of the Seed Information Database [13]. This is one of the reasons that the initial germination test, ideally made at the time when the seed lot is placed into genebank storage, is so important, not just to confirm the seeds are above the viability threshold, but also to have some indication of how long they might stay above that viability threshold. For example, the predicted length of time it takes for the viability of rice (*Oryza sativa* L.) seeds in long-term storage (6.1% moisture content, −20 ◦C) to fall to 85% is 255 years when the initial viability is 95%, but 103 years if initial viability is 90% (estimated using the Seed Information Database [13]).

Of course, the VEs are not infallible; there is error associated with any prediction, based on the errors associated with the parameter estimates caused by random (and perhaps experimental) errors in the original data [47], though prediction errors have not usually been calculated. Furthermore, the initial viability result is only an estimate of the true viability of a seed lot [48,49]. Of more concern, the species-specific parameters may not be as robust as previously thought [49,50]. The VEs are also critiqued because the cumulative normal distribution has a particular well-defined shape. In fact, deviances from the basic symmetrical sigmoid curve can be accommodated, and in many cases, attributed to a biological response. For example, viability may be more or less maintained above a threshold value that is less than 100% [51] or there may be a dormancy-breaking response at the start of storage, before viability declines [52]. It would also be possible to describe sub-populations within the seed lot that are responding to storage differently, provided there are sufficient data.

Despite the concerns around using the VEs to make predictions of longevity, they are still extremely useful, not least for emphasising the importance of drying; even seemingly small declines in moisture content, over a certain range, result in significant increases in longevity [27]. The low moisture content limit to this range varies between species, largely owing to differences in seed oil content; it also shifts to a higher moisture content as the storage temperature is reduced [53]. Importantly, drying below this moisture content does not result in either further improvement in longevity or decline in longevity [27,53,54]. As some of the original work in this area found that the low moisture content limit was reached when seeds were equilibrated to a rather low humidity (approximately 10% at 20 ◦C for seeds subsequently aged at 65 ◦C) [55], the first official Genebank Standards recommended drying seeds for genebank storage at 10–25 ◦C and 10–15% RH [56]. Many genebanks installed drying rooms set to these conditions to efficiently dry seeds. Some genebanks check the moisture content of seeds before packing. The official seed testing method to determine seed moisture content is to weigh a sample of seeds before and after oven-drying (at 103 or 130 ◦C, depending on species [5]). This is a destructive test and may require far more seeds than a genebank would like to use. A more efficient, non-destructive method is to simply check whether the equilibrium relative humidity (eRH) of the seeds is equal or close to the RH of the environment in which they have been dried, using a suitable hygrometer [57]. If they are in equilibrium, the seeds will not be able to dry further, and are ready to be packed; it is not necessary to know the actual moisture content of the seeds.

The VEs also illustrate why, if different samples of seeds from the same seed lot are stored at the same moisture content (because they are hermetically packed at the same time) at different temperatures, it is not necessary to monitor the viability of both samples from the start. The seeds at the higher temperature will lose viability faster than those at the lower temperature. Thus, for example, it would not be necessary to test the seeds in long-term storage (at the lower temperature) until the viability of the seeds in medium-term storage has declined to the viability standard (Figure 1) [48]. This could save resources if both samples are currently being tested, though of course, it means there would be less viability monitoring data for modelling longevity. Similarly, if samples of seeds are packed for safety duplication (Box 1) in high quality moisture-proof containers at the same time as the samples intended for long-term storage (packed similarly), provided the seeds are stored at the same temperature in the two locations or at a lower temperature in the safety duplicate location than the originating genebank, then it is not necessary to test the seeds that have been sent as the safety duplicate. Moreover, there should not be any real need to include additional samples for retrieval and testing, which would only serve to add extra work load to genebank staff, who may already be challenged to carry out the requisite number of monitoring tests of seeds in the active and base collections.

**Figure 1.** Predicted time for viability to fall from 92% to 85% for seeds of different crops stored in the medium-term store (MTS), the long-term store (LTS), or as a safety duplicate (SD) under LTS conditions in another location. Predications made using the Seed Viability Constants tool of the Seed Information Database [13], based on drying seeds to equilibrium with 15% relative humidity at 15 ◦C. The red stars indicate the timing of viability monitoring tests, at intervals of one-third of the time predicted for viability to fall to 85%.

It is often asked whether moving seeds between temperatures, for example, from the medium-term storage environment to the laboratory, to take a sample for distribution or for viability monitoring, has an effect on the longevity of the seeds. This may happen many times in the 'lifetime' of a genebank seed lot. The evidence, again from applying the VEs, is that there is no effect of moving the seeds per se; the viability is affected only by as much as would be expected from the brief period spent at the higher temperature where the seeds are allowed to equilibrate and are sampled [58].

**Box 1.** Safety duplication and the Svalbard Global Seed Vault.

The Genebank Standards [34] recommend, as a safety measure in case of natural disasters, that a sample of all original seeds collected or seed accessions only held by that genebank should be duplicated in another location. Ideally, this location should be in a different country or even continent i.e., somewhere that is not at risk to the same natural and/or human-caused catastrophes, and preferably one where there is no socio-political uncertainty and environmental risk. This "black-box" collection is not active therefore it is the responsibility of the depositor to ensure the sample size is sufficient (enough to carry out at least three regenerations) and of a high quality (>85% germination). If the storage conditions are the same at both locations then the rate of loss in viability of the black-box collection should equal that of the sample in the original genebank, the viability of which is monitored.

The Svalbard Global Seed Vault (SGSV) on Svalbard, Norway, opened in 2008 and offers the long-term storage of safety duplicates from the world's 1700 national and international genebanks. The vault currently holds more than 1,173,000 samples (www.seedvault.no) and represents the world's largest collection of crop diversity. It is strategically located in a highly secure zone and is the ultimate insurance policy against loss of plant agrobiodiversity.

#### *3.3. Comparative Seed Longevity Studies*

A more robust method for assessing the storage potential of different seed lots would be to carry out a seed storage experiment (SSE) rather than just an initial germination test (Table 1). This involves storing samples of seeds at a relatively high temperature and moisture content (e.g., 45 ◦C after equilibrating seeds at 60% RH and 20 ◦C) and removing a sample at regular intervals for germination testing [59,60]. After probit analysis of the germination data, the seed lots can be ranked according to the time for viability to decrease to a specific viability [61,62] and/or categorized into longevity categories [63]. If the slopes of the survival curves can be constrained to be the same for all of the seed lots (potentially possible if the seeds are of the same crop), then the theoretical initial viability can be inserted into the VEs to make more precise predictions of longevity. Monitoring intervals for the seeds in genebank storage can then be adjusted on a seed lot basis.

#### *3.4. Maturity at Harvest*

Seed longevity continues to increase after the seeds have acquired desiccation tolerance, in the late maturation phase of seed development [64–67]. However, the time when 'final' or 'maximum achievable' longevity is reached and how long it is maintained thereafter will vary between species, varieties, accessions, and seasons. This is because the longevity of maturing seeds still in the field is highly dependent on both the genotype and the environment, and the interaction of the two. This makes it virtually impossible to consistently harvest different accessions when they have peak longevity.

In the late maturation phase of seed development, the seeds no longer have a vascular connection with the mother plant. The moisture content of the seeds, and thus the availability of water to support different types of chemical reactions, will depend on the ambient conditions (humidity and temperature). The drying that occurs in situ is considered important for stimulating the improvement in seed longevity [52,65], which is why it is often recommended to wait for seeds to dry in the field before harvest [18]. This is easy for shatter-resistant crops; for crops and species that readily disperse their seeds after the vascular connection with the mother plant is broken, delaying the harvest may result in loss of seeds. For such species, bagging inflorescences before they enter the late maturation phase may be necessary [54].

If seeds that have dried on the plant subsequently take up moisture before they have been harvested, depending on the moisture content they attain, longevity may in fact continue to improve, either while the seeds have a high moisture content, provided the seeds do not germinate, and/or upon redrying [68]. If an intermittent moisture content is reached, ageing may occur at a rather fast rate until either the seeds dry again or moisture content increases further. If the moisture content increases to a level where respiration can occur, it is possible that the damage accumulated during ageing will be repaired and seed quality restored/maintained [69]. If the seeds have suffered a substantial amount of damage, they may not able to reach the same longevity as was previously attained. It is ultimately the net changes in seed quality (improvement vs. deterioration) that will determine the potential longevity of the seeds when they are harvested. Developmental events that result in improvements in longevity may also continue ex planta if seeds are harvested prematurely and held at conditions similar to what they might naturally encounter in planta [52,60,70], or upon rehydration if seeds were dried too quickly for maximum quality to be attained [71].

As yet, there is no tool to assess how far seeds have progressed through maturation processes before they are harvested, not least because physical changes that occur during maturation vary so widely between species and varieties [72,73]. In temperate and dry climates, the most reliable method is to see whether seeds are in equilibrium with ambient conditions, by placing a sample of seeds in a portable hygrometer and comparing the seed equilibrium relative humidity with the ambient relative humidity.

#### *3.5. Post-Harvest Handling*

The post-harvest environment and seed processing operations also affect seed quality and subsequent longevity [61,74]. If seeds have already dried to equilibrium with ambient conditions and reached a relatively low moisture level (<85% equilibrium relative humidity), seeds should be dried as soon as possible to minimize ageing [75]. In semi-arid and arid climates, drying, if necessary, may be done outside under well-ventilated conditions, but avoiding risks such as insect predation or over-heating in direct sunlight. In such environments, care should also be taken not to allow drying to progress too far, as cracking may occur if the seeds need to be threshed. Otherwise, the seeds should be transferred to a controlled drying environment as soon as possible.

As discussed previously, if seeds are harvested with a high moisture content and are likely to be metabolically active, seeds should be dried under conditions that optimally stimulate late maturation phase metabolism. For example, in rice and soybean, the quality of seeds harvested, when still at a high moisture content, was improved by drying for an initial period at a high temperature (45 ◦C) [50,52,76,77]. It is thought that the loss in moisture is a critical factor controlling the maturation process, by inducing the stress response and other protective mechanisms [78], which significantly increase seed quality. However, drying at such a relatively high temperature is in conflict with the current Genebank Standards that recommend drying mature seeds for long-term storage at a low temperature (5–20 ◦C) and relative humidity (10–25%) [34]. This recommendation was driven by the requirement for a single, simple, and safe procedure for diverse species in all locations worldwide, and assumed that seeds were already comparatively dry (for example, seeds dried using heated air [79], but requiring further drying to a low moisture content for long-term storage). This was determined by combining the seed viability equation with equations describing the effect of environment on seed drying rate and seed temperature in constant-temperature heated-air dryers in contrasting species [80]. Although the "safe" temperature limit for drying seeds varies between species, high temperatures are usually avoided to reduce the risk of seed deterioration, especially when seeds have a high moisture content and during the later stages of drying when evaporative cooling will no longer suppress the temperature within the seeds [79–81]. Despite this, the recommended low temperature and low humidity conditions for post-harvest seed drying are neither species-specific nor dependent upon initial moisture content [34].

#### *3.6. Length of Time Before Storage*

Most genebanks clean seed lots before packing for storage to remove, for example, off-types, immature, diseased, and damaged seeds. This is often a largely manual process or involves the use of fairly basic equipment such as graded sieves or blowing machines. It is important that the seeds do not sit for too long in an uncontrolled environment while cleaning occurs, as this might mean that they take up moisture and rates of ageing increase. Redrying may be necessary before packing, which should also be done in a controlled environment. Most drying and processing environments, even though they are controlled, are maintained at temperatures that are comfortable for genebank staff, but that are thus higher than desirable from the point of view of maintaining the quality of the seeds. Therefore, it is important that the seeds are packed and transferred to proper genebank storage as soon as possible.

#### **4. Statistics of Seed Testing**

Genebanks, as a rule, do not conduct experiments on genebank accessions as part of routine operations (Table 1), but they nonetheless do gather data that may be relevant to analyse, for example, to understand the diversity of the material they conserve, to create core-sets based on the diversity represented, or to identify potential gaps in a collection. Some of these data may be suitable for methods of analysis that are typical within the agricultural sciences. However, viability monitoring and other germination data cannot be treated and analysed in the same way [82]. This is because seed germination is a binary response; a seed will either germinate or not germinate, and a sample of seeds will show a germination result that falls between 0% and 100%.

#### *4.1. Comparing Two Germination*/*Viability Results*

There are a number of reasons why two germination or viability results might be compared, the most obvious being to compare the results of a germination test with that of a viability test (see

above), or the results of a pair of germination tests made using different pre-treatments and/or test conditions. Testing a viability monitoring result against a previous monitoring result or the initial test result would not be common, because genebanks tend to 'accept' a viability result and not worry whether or not it is significantly different from any other value. Indeed, genebanks do not even routinely compare a viability monitoring result against the viability threshold value, even though the germination result is only ever an estimate of the whole seed lot viability (because not all the seeds are tested). One test that could be used for comparing two germination results is the χ<sup>2</sup> test to test whether the probability of success (scored as viable or germinated) is the same for seeds from two different treatments (e.g., stored or not, viability or germination). The algorithm for the χ2-test is likely to be based on maximum likelihood estimation, in which case the test is the same as fitting a single-factor binary logistic regression, that is, a generalized linear model (GLM) with logit link function (the link function transforms the response variable, for example, the proportion of germinating seeds, to the linear 'predictor' variable that is fitted in the analysis) and binomial error distribution based on the number of seeds tested.

#### *4.2. Analysis of a Factorial Germination Experiment*

It may be that a dormancy-germination experiment is multi-factorial, in which case the obvious extension to the χ<sup>2</sup> test is to fit a GLM with different parameters estimating the effect of the different factors included in the experiment. Different link functions may be used to relate the response variable to the predictor [83], but the most common is probably the logit link function, that is, binary logistic regression analysis, for example, [84]. A further extension of this method is to fit a generalized linear mixed model (GLMM) to take into account fixed (factor-related) and random effects (e.g., if seeds are sown across multiple test units such as Petri dishes or rolled paper towels) [82,85]. It is rarely appropriate to use analysis of variance (ANOVA), because the nature of germination data is likely to violate the assumptions of the analysis; that is, that the errors after fitting the model follow the same, normal distribution across treatment groups [82].

#### *4.3. Analysing a Series of Germination Results*

Although not routine in most genebanks, recording the progress of germination of a sample of seeds in a germination test over some days or weeks (or in some cases, even longer) is common in studies related to determining optimum dormancy-breaking/germination requirements and understanding seed vigour (speed of germination). Germination progress data are also collected to assess seed lot behaviour in response to stress (e.g., water stress) or, conversely, to stimulants, for example, [86,87]. By convention, many analyses of such behaviour have neither taken the binomial error distribution of germination behaviour into account nor used independent samples, e.g., [88]. If independent samples are used, for scoring just once and then discarding, again, a GLM or GLMM might be used to analyse the data; if samples are repeatedly scored for germination, time-to-event model fitting would be more appropriate [83]. Germination data for a series of samples from a seed lot in storage can be analysed by fitting a GLM. By convention, this has been by probit analysis (i.e., fitting a GLM with a probit link function) and is the basis of the Ellis-Roberts viability equations (see above).

#### **5. Concluding Remarks**

Genebank managers often come from diverse backgrounds, with different scientific training and experience. On the other hand, to be a genebank manager requires a diverse set of skills and acquaintance with, if not knowledge of, a broad range of topics. As such, it can be difficult to find an ideal person to fill a senior genebank management position, unless there is an applicant who is moving from another genebank or has been an understudy for the role. Consequently, it may be likely, or even inevitable, that some expertise gets overlooked and/or is difficult to find compared with a few decades ago, when for example, plant taxonomy and seed science were perhaps more likely to be covered in botany and plant science degrees. As such, incoming genebank managers may be

on a steep learning curve and be limited by a lack of understanding of some aspect(s) of genebank science. It is key to the management of a genebank that staff understand the reasons behind patterns of germination shown by individual crops and seed lots; are primed to make efforts to increase the efficiency of operations; and are able to communicate this to donors, reviewers, and the public. If not the head of the genebank, then other genebank staff should have training in seed science and technology and/or access to seed physiology experts, and deliberate strategic investments should be made in capacity building in this area.

Seed handling has been a focus of recent quality management audits and external reviews that have been conducted under the CGIAR Genebank Platform [89]. Standard operating procedures (SOPs) have been drafted and audited for key operations at each of the 11 CGIAR genebanks. In addition to document audits, expert reviewers have visited each genebank to see the procedures in practice and to validate individual SOPs. A large number of the reviewers' observations and recommendations concerned seed processing, handling, and multiplication of seed lots and related data management. It is clear that seed quality management practice is a highly dynamic and influential aspect of genebank management, where a lack of oversight at any point in the long history of a permanent genebank can profoundly impact efficiency and long-term conservation. Continual expert input and review, and research and experimentation, as well as innovation and automation on various areas of seed management, clearly play major roles in the sustainability of major seed collections.

In this review, we have discussed various aspects of seed physiology that are particularly relevant for seed genebanks if appropriate decisions about the handling of seed germplasm are to be made. Some of this may seem like common sense to those that are trained in seed science, but in fact, many of the observations relate to questions that do get asked by genebank staff at all management levels. Further, there should be more interaction between the seed science/testing community and genebanks. This would help genebank staff stay abreast of new scientific developments in seed physiology and, in particular, testing methods, and thereby help them to continue to improve the efficiency and effectiveness of genebanks operations. It is also important for seed scientists to be aware of areas on which to focus their research to support the work of the genebanks in conserving and making available plant genetic resources.

**Author Contributions:** F.R.H. and K.J.W. conceived the article; K.J.W., F.R.H., and C.L. wrote the article. All authors have read and agreed to the published version of the manuscript.

**Funding:** The writing of this article was partially funded through the CGIAR Genebank Platform, as part of the Seed Quality Management activity of the Conservation Module of the Platform.

**Acknowledgments:** We thank the editors of this special issue, Andreas Ebert and Jan Engels, for inviting us to write this article and for providing guidance on the content.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

### **Challenges and Prospects for the Conservation of Crop Genetic Resources in Field Genebanks, in In Vitro Collections and**/**or in Liquid Nitrogen**

#### **Bart Panis 1,2,\*,**†**, Manuela Nagel 3,**† **and Ines Van den houwe 1,2,**†


Received: 15 October 2020; Accepted: 18 November 2020; Published: 24 November 2020

**Abstract:** The conservation of crop genetic resources, including their wild relatives, is of utmost importance for the future of mankind. Most crops produce orthodox seeds and can, therefore, be stored in seed genebanks. However, this is not an option for crops and species that produce recalcitrant (non-storable) seeds such as cacao, coffee and avocado, for crops that do not produce seeds at all; therefore, they are inevitably vegetatively propagated such as bananas, or crops that are predominantly clonally propagated as their seeds are not true to type, such as potato, cassava and many fruit trees. Field, in vitro and cryopreserved collections provide an alternative in such cases. In this paper, an overview is given on how to manage and setup a field, in vitro and cryopreserved collections, as well as advantages and associated problems taking into account the practical, financial and safety issues in the long-term. In addition, the need for identification of unique accessions and elimination of duplicates is discussed. The different conservation methods are illustrated with practical examples and experiences from national and international genebanks. Finally, the importance of establishing safe and long-term conservation methods and associated backup possibilities is highlighted in the frame of the global COVID-19 pandemic.

**Keywords:** clonal crops; collection management; cryobiotechnology; cryopreservation; field collections; field maintenance; germplasm storage; in vitro conservation; recalcitrant seeds

#### **1. Introduction**

In the course of crop domestication, many plants have been selected for quantity and/or quality of their seed, while some have been cultivated for their roots, tubers, fruits, stems and leaves. Plant genetic resources for food and agriculture (PGRFA) are of strategic importance to ensure sustainable crop production [1], nutritious food and food security for humans and to enhance economic prosperity of the present and future generations. They comprise the sum of genes, gene combinations or genotypes which serve as a reservoir for direct use in food production systems and for breeding new varieties [2].

Since the beginning of agriculture, selection of plants and seeds for sowing, growing, harvest and storage gave rise to locally adapted varieties, so-called "landraces", that reveal specific variations of morphological and yield characteristics and quality traits. In the mid-19th century, the rediscovery of Gregor Mendel's work and the introduction of breeding schemes resulted in the development of high-yielding and more stress-tolerant varieties leading to higher crop yields. This laid the foundation for the green revolution taking place in the middle of the last century bringing about increased

agricultural production to feed the exponentially growing world population. However, the expansion of industrial mono-cropping with the replacement of landraces by modern breeding varieties has caused the loss of 75% of plant genetic diversity, with more than 90% of crop varieties having disappeared from farmers' fields [3]. In this era of global environmental problems, climate change, and booming population growth, it is of paramount importance that the remaining crop genetic resources are kept available to sustain the agricultural production systems, to feed the world population a healthy diet and to tackle future demanding challenges [4].

Since the 16th century, botanical gardens have collected and preserved a variation of more than 80,000 plant species in about 3400 gardens all over the world [5]. They primarily have an interest in conserving the widest possible plant diversity and crop wild relatives that can be an important source material for breeders. PGRFA are conserved ex situ in specialized repositories, often termed genebanks that have been established since the mid-20th century. In contrast to most botanical gardens, genebanks focus on both intra- and inter-specific crop diversity. There are more than 17,000 national, regional and international institutes and organizations dealing with the conservation and sustainable use of PGRFA [5]. Currently, 711 gene banks and 16 international/regional centers in 90 countries retain more than 5.4 million accessions from over 7051 genera, mainly focusing their conservation efforts on crop species, including landraces and crop wild relatives, breeding materials and cultivars [6].

Most of the major food crops produce orthodox seeds that tolerate intense dehydration and low temperatures, thus seed storage under dry and cool conditions is naturally the most widely adopted method for long-term ex situ conservation at relatively low costs. About 45% of the accessions stored as seeds are cereals, i.e., wheat (*Triticum aestivum*), triticale (*Triticum secale*), rice (*Oryza sativa*), oat (*Avena sativa*), rye (Secale cereale), barley (*Hordeum vulgare*), maize (*Zea mays*) and sorghum (*Sorghum bicolor*), followed by food legumes (15%), forages (9%) and vegetables (7%) [5].

In contrast, a large number of food crops are not storable through seeds and thus need different conservation approaches [7]. This category of plants consists of important species that produce desiccation sensitive, recalcitrant or intermediate seeds, such as coconut (*Cocos nucifera*), cacao (*Theobroma cacao*), avocado (*Persea americana*) and citrus (*Citrus* spp.) and species that are seedless such as edible banana (*Musa* spp.) and garlic (*Allium sativum*). Species including yucca (*Yucca* sp.) and bamboo (*Bambuseae* sp.) that have long life cycles and take years or decades to reproduce also fall into this category. Other species that produce orthodox seeds but require the conservation of particular gene combinations or genotypes, such as root and tuber crops, notably potato (*Solanum tuberosum*), cassava (*Manihot esculenta*), yam (*Dioscorea* spp.), taro (*Colocasia esculenta*) and several fruit and nut trees are also included. These crops are propagated vegetatively, and each genotype needs to be maintained as a clone.

Two main ex situ conservation approaches can be distinguished for these crops: the conservation of plants in field genebanks and the maintenance of propagules in tissue culture, either (i) as active growing cultures in short- and medium-term storage (i.e., in vitro storage), or (ii) in frozen state at ultra-low temperature in liquid nitrogen for long-term storage (cryopreservation). These approaches, the challenges entailed, and prospects offered to secure crop diversity ex situ will be discussed in detail in this chapter and are depicted in Figure 1.

**Figure 1.** Pros and cons of storing crop genetic resources in field genebanks, in vitro and through cryopreservation. Arrows indicate the potential source material and target approach to maintain plant genetic resources for food and agriculture (PGRFA) and can be specific for each plant species.

#### **2. Field Genebanks**

Clonally propagated crops and their wild relatives belong to some 34 plant families, including herbs, shrubs, trees and vines [8]. Agronomically important genotypes that were selected over centuries for their specific properties can only be conserved in a vegetative mode. Consequently, the simplest and most traditional way to establish a collection is to gather specific genotypes from farmers' fields, gardens or in the wild and then grow them in the field genebanks where they continue to grow when maintained appropriately. Even under the highest standards of management, germplasm maintained in the field can deteriorate due to a wide variety of climate conditions, ageing of the plants, diseases and pests, hence the need for timely regeneration. For example, depending on the rootstock and orchard conditions, apple (*Malus* sp.) trees may need to be repropagated periodically after 25–50 years [9].

Some of the largest collections were established at the beginning of the 20th century and are based on the efforts of passionate geneticists and plant explorers such as Nikolai I. Vavilov, Frank N. Meyer [10] and Hans Stubbe among others [11]. In the 1970s, the International Board for Plant Genetic Resources (IBPGR) promoted and sponsored numerous collection missions. To preserve clonal material, 23 field genebanks of nine major crops were established worldwide [12]. As such, approximately 400,000 accessions are currently held in field genebanks of international, regional and national authorities. A recent study commissioned by the Alliance of Bioversity International and the International Center for Tropical Agriculture (CIAT), the International Potato Center (CIP) and the Global Crop Diversity Trust [13] showed that of the 20 institutions surveyed, the vast majority of the clonal plant material is kept in the field. Worldwide, major genebanks maintain potato (98,285 accessions), apple (59,922), cassava (36,529), citrus (36,410), sweet potato (*Ipomoea batatas*, 35,478), coffee (*Co*ff*ea* spp., 30,483) and cacao (23,107). The largest field genebanks are located in the USA (potato, sweet potato and apple), Japan (apple, citrus and sweet potato), Russia (potato and apple) and Brazil (citrus and coffee). In addition, there are numerous smaller collections, such as for grape vine (*Vitis vinifera* L.), garlic

(Figure 2, see Box 1), Jerusalem artichoke (*Helianthus tuberosus* L.), and Andean root crops, which are of high value as luxury foods and condiments or are of regional or religious significance.

#### *2.1. Management of Field Genebanks*

The specific needs of a crop with respect to growth requirements and its multiplication cycle ultimately determine the field conditions and the structural design of the field genebanks. Propagules of some annual crops, such as potato, shallot or yams, are cultivated by up to 10 plants in the field and require good agricultural practices including crop rotation. After harvest, these propagules must be kept under suitable storage conditions until the next growing season [14]. By contrast, woody crops, such as apple, pear, coffee and grapevine are grown at the same locations for many years, often with two plants per accession. Some of these crops require budding or grafting to rootstock resistant to root nematodes, layering or rooting stem cuttings. For example, the US Department of Agriculture—Agricultural Research Service (USDA-ARS) National Plant Germplasm System (NPGS) maintains 5004 apple accessions in the field and 1603 *M.* × *domestica* seed accessions in Geneva, NY, USA. Due to the large number, about 3100 field accessions are only represented by a single tree [15]. By comparison, the JKI Dresden-Pillnitz, but also universities, governmental institutes, communes, non-governmental organizations and private individuals are responsible for the maintenance of fruit collections in Germany. These collections include landraces that partly date back to the 12th century and some of these are at risk. Therefore, the German Fruit Genebank was launched in 2007 and consists of six fruit-specific networks (apple, cherry and plum (both *Prunus* spp.), berries (*Rubus* spp.), strawberry (*Fragaria* spp.), pear). For apple, an expert group selected 743 unique accessions that were duplicated and are represented by two trees each in at least two sites [16]. Although, establishing networks requires additional efforts, the responsibility for maintaining such valuable resources is shared and the partners mutually benefit from the concerted expertise and the security status of the collections.

Depending on the collection strategy and the breeding programs, both annual and perennial crops can vary in proportions of landraces, breeding lines and cultivars and are used as different types of donors for plant breeding. CIP, for example, maintains the largest collection of 4487 potatoes landraces at Huancayo, Peru at a 3200 m elevation. [17]. To increase the productivity and, hence, farmer's incomes in Africa, Asia and Latin America, CIP and partners developed training guides for the positive selection of propagules of landraces that have no visible symptoms of diseases or abiotic stress [18]. By comparison, at IPK (Leibniz Institute of Plant Genetics and Crop Plant Research), in addition to 3300 landraces, also 2900 wild potato accessions are maintained as orthodox seeds. As some potato species are self-incompatible, seeds are propagated as a population in the greenhouse and are screened for resistances, to late blight (*Phytophthora infestans*) on tubers or pale potato cyst nematodes (*Globodera pallida*) [19]. Selected donors are further used to develop introgression lines that can be later used by the breeding industry. In other words, breeding intensity and objectives determine the composition and have a strong impact on the management system of the collections.

**Figure 2.** *Allium* field collection at IPK, Germany.

**Box 1.** Field collection of Allium genetic resources at IPK, Germany.

The IPK maintains one of the largest *Allium* collections worldwide and comprises the Taxonomic Reference Collection of 1300 accessions and 287 species and the *Allium* Crop Collection covering 1400 accessions of 76 species (Figure 2). About 2100 accessions are permanently maintained in the field because of (i) their inability to form seeds, i.e., in garlic and shallots (*Allium cepa* var. *aggregatum*), (ii) the presence of heterozygous seeds, i.e., in onion (*Allium cepa* var. *cepa*), or (iii) the traditional breeding strategy as clonal varieties as in the case of some ornamental *Allium* hybrids [20].

*Allium* accessions, except shallot and onion, are maintained as a permanent crop at the same site and 3 to 6 plants are planted withina2m2 plot. These accessions require permanent management including regular identification, weeding, phytosanitary treatments and seed harvest to avoid the establishment of different accessions or hybrids within the same plots. The soil quality and the continental climate in Central Germany, including mild winters, reduce the risk of pest and diseases and provide optimum growth conditions for a longer term. However, each 5 or 6 years, the Allium gardens are replanted to overcome the problem of soil exhaustion [21].

Special attention must be given to 425 garlic accessions and 82 shallot accessions that are partially replanted every year. In autumn, to avoid infections of nematodes and wireworms, cloves are treated with insecticides and planted in the field. At the end of July of the following year, accessions develop cloves (lateral bulbs) and in some cases bolt and develop bulbils which dry up. These can be used for germplasm distribution. For introduction into tissue culture or cryopreservation, bulbs and bulbils require a further after-ripening period of up to 2 months until the physiological dormancy [22] is broken. During this period, storage of the bulbs and, in the case of garlic also the bulbils, is conducted at between 4 and 10 ◦C [23].

Various strategies have been developed to support the field genetic resources of garlic and shallot. About 30 years ago, in vitro slow-growth maintenance at a temperature between 2 and 10 ◦C was initiated for about 700 accessions. As such, cultures can be kept for up to 12 months without subculturing. However, after many subculture cycles, the accumulation of microbes delimitates the in vitro storage. Therefore, the number of in vitro accessions was reduced to about 25 accessions. Nowadays, the introduction of garlic and shallot accessions into in vitro conditions is only carried out as preparatory step for cryopreservation. Since the PVS3 vitrification approach has been successfully developed for garlic and shallot [24], more than 210 accessions (Table 1) were successfully cryopreserved. To protect further the allelic diversity of garlic, the Research Institute of Crop Production (CRI) in the Czech Republic, the Research Institute of Horticulture (RIH) in Poland and the IPK established a European Core Collection within the frame of "A European Genebank Integrated System" (AEGIS) [25].

#### *2.2. Advantage of Field Collections: Characterization and Evaluation*

The propagation of clonal plants in the field is the conventional method for the preservation of genetic material and traditional knowledge of the farming system [26]. In areas where plants are historically grown, the cultivation can be carried out by local farmers. The sites are usually well-established with pest control and forecasting models used [27]. Their exposure to natural conditions allows a limited selection according to environmental conditions and a competition between plant propagules [8]. A tremendous benefit of holding field collections is that images and voucher specimens can be immediately assessed and made available online [27–29].

Descriptors have been elaborated for a wide range of crop species [27,29,30]. The descriptors and the list of Multi-Crop Passport Descriptors provide an international format that facilitates comparisons among and within collections [14]. The data often follow FAIR—findable, accessible, interoperable and reusable—principles [29] and well-characterized material can be further distributed and evaluated and made available to breeding programs. At IPK in 2019, of 3300 maintained potato cultivars, about 450 accessions were propagated in the field and of these 1027 sub-samples were distributed. A main advantage is that the material is immediately available for evaluation and distribution. For example, at a field-based germplasm collection in New Zealand genomic DNA was extracted and screened for the presence of TG689 and 57R haplotypes linked to the *H1* gene. These haplotypes act as potential predictors for the resistance against a pathotype of potato cyst nematode (*Globodera rostochiensis*), an economically important pest [31]. Similarly, banana germplasm was screened for the resistance against the *Fusarium oxysporum* f. sp. *cubense* tropical race 4 (*Foc TR4*) that has seriously threatened global banana production. From 129 evaluated accessions, 10 were highly resistant to the virus [32]. The readily available data and material can thus have a strong impact on the material selected for breeding.

#### *2.3. Problems Associated with Field Collections*

In field collections, plants are exposed to a natural environment, which can include unfavorable climatic conditions, such as drought, heat or frost, similar to crop production fields. Additionally, pests and diseases can threaten the material, especially the less adapted or susceptible accessions. One of the important apple diseases, fire blight, caused by the bacterium *Erwinia amylovora,* can severely damage or even eradicate susceptible apple accession [33,34]. In 2013, the *Allium* field collection at IPK was challenged with a serious infestation of larval stage click beetles (*Elateridae*), also known as wireworms. Of 2150 accessions, 52 accessions were completely lost and 73 could be recovered by replanting and duplication. Furthermore, depending on the plant species and their life cycles, clonal collections must be rejuvenated periodically. The Tropical Agricultural Research and Higher Education Center (CATIE) in Costa Rica hosts one of the largest collections of *Co*ff*ea arabica* L. and maintains plants that were planted in the 1970s. In the year 2000, a project to rejuvenate the collection was initiated and is still ongoing [14,35]. Compared to reproduction by seeds, vegetatively propagated populations are also susceptible to the accumulation of viruses, bacteria and fungi that might be similar to Muller's ratchet [8]. As a consequence, the maintenance of field collections requires substantial efforts in terms of agricultural measures such as manual selection, evaluation, mechanical weed control, rejuvenation and specific disease control. Therefore, the number of replicates is often limited to between 5 and 10 for cassava, 10 and 12 for sweet potato, 2 and 9 for garlic, 1 and 3 for trees and shrubs and 3 and 20 for bananas [14]. However, spatially separated replicate plants (i.e., trees and vines) can aid in minimizing loss of accessions by pests or mismanagement. However, even provided everything is handled with the utmost care and state-of-the-art technologies, human errors can easily happen and accessions can be mixed up. A genotyping study of 250 accessions of the CIP potato collection showed that only about 80% of the collection was still comparable with voucher mother plants which arrived at CIP about 30 years before [36]. At Arabidopsis stock centers, it was estimated that about 3 to 14% of the materials are misidentified with most errors caused by incorrect labeling [37]. There is also the possibility that genetic lines segregate and spontaneous mutations occur. Especially under environmental stress, whole-genome and whole-methylome sequencing revealed that

the Arabidopsis lineages accumulate 100% more genetic mutations and epigenetic modifications under stress compared to non-stress conditions [38]. This underlines the importance of cautious evaluation, involving trained staff, and the application of complementary preservation methods such as in vitro storage and cryopreservation.

#### **3. In Vitro Collections**

In vitro culture (or tissue culture) of plants is a biotechnological technique in which plant parts are isolated from in vivo plants, disinfected to free the explants from bacteria and fungi and transferred onto well-defined and sterile tissue culture media that provides the plant tissue with the necessary nutrients for growth and multiplication. In a relatively small space, the environment can be controlled precisely and plant growth can be easily observed and manipulated. In vitro approaches are commonly used for large-scale micro-propagation, reproduction purposes including embryo rescue, ploidy manipulations, protoplast fusions and somatic embryogenesis and are appropriate tools for short- and mid-term storage of plant genetic resources. Although the feasibility of using in vitro culture methods for plant genetic resources conservation was already known in the 1970s, it was only in the 1980s that the International Board for Plant Genetic Resources (IBPGR) established a working group of specialists to investigate the critical aspects of in vitro plant conservation [39,40]. Since then, in vitro collections have been setup for many vegetatively propagated crops. Additionally, Genebank Standards for PGRFA maintained in vitro were developed [1], forming the benchmark for establishing standard operating procedures and quality management systems to ensure effective, safe and efficient conservation of these genetic resources. Nowadays, in vitro collections for PGRFA comprise potato (9700 accessions), cassava (8700), sweet potato (6400), yam (3200), banana (2000) and taro (1200) [40–42]. The largest collections are maintained by international organizations such as Bioversity International, the International Center for Tropical Agriculture (CIAT), CIP, the International Institute of Tropical Agriculture (IITA) and in national institutes such as Brazilian Agricultural Research Corporation (EMBRAPA) in Brazil, CRI in Czech Republic, the IPK in Germany and the USDA-NPGS in the USA [13,42].

#### *3.1. Setting Up In Vitro Collections*

At tissue culture facilities, plant tissues are maintained in specific and aseptic growth conditions. To culture explants in vitro, typically tissue culture media are used that contain water, macro and micro nutrients (salts), a gelling agent, plant growth regulators (plant hormones) and sugars. The supplement of sugars as source of energy and building blocks is essential because, unlike in vivo plants, in vitro plants lack the ability to photosynthesize effectively. Explants are kept in a culture room with controlled temperature and light regimes. Since tissue culture material continuously grows and undergoes ageing, plants regularly need to be trimmed, divided in separate propagules and transferred to new culture media. Explant survival and vitality of in vitro plants depend on media composition, temperature and light intensity and further factors such as light spectrum, vessel type and size, number of explants and gas exchange rate [43–45], all of which can affect the growth of the plants.

The goals of applying slow-growth conditions in plant collections is to reduce frequency of subculture events that are labor (and thus cost) intensive and to minimize the risk of loss of accessions due to handling errors and genetic instability induced by the tissue culture environment. Under optimal growth conditions, subculture frequencies range from one to three months, whereas at slow growth conditions, the subculture period can be from one to two years. Three so-called Medium Term Storage (MTS) approaches can be distinguished: physical growth limitation, chemical growth limitation and nutrient limitation [40]; all aimed to reduce the metabolic activity of the plantlets. Physical growth limitation is often achieved by lowering temperature and often combined by applying low light intensities. Cold-tolerant species such as garlic, potato and most *Mentha* species can be kept viable at temperatures from 0 to 5 ◦C and 2 to 4 μmol m−<sup>2</sup> s−<sup>1</sup> for up to 24 months [46], whereas cold sensitive species such as many tropical plants should be stored at relatively high temperatures. Examples are *Musa* [47] (Figure 3, see Box 2) and sweet potato [48] both requiring a temperature of 15 ◦C or higher; for pineapple this is 21 ◦C [49]. Such higher storage temperatures result in shorter sub-culture intervals (six to twelve months). Nutritional growth limitation decreases the supply of carbon and inorganic nutrients whereas chemical growth limitation involves the application of osmotically active agents such as mannitol, sorbitol, polyethylene glycol (PEG) or growth retardants such as abscisic acid (ABA) and hydrazides. In potato, for example, extended periods in MTS can be achieved by different protocols which either apply Murashige and Skoog (MS) medium, sucrose and the growth retardant ancymidol at 6 ◦C resulting in storage periods of 12 months [50]; MS and mannitol at 6 ◦C for storage periods of 16 months [51] or MS, sucrose, mannitol at 6 ◦C for storage periods of 30 months [52]. This shows that often combinations and variations of different slow-growth approaches need to be adapted for different species. An overview of growth limiting protocols is summarized by Chauhan, Singh and Quraishi [53].

#### *3.2. Advantages of In Vitro Collection*

An important benefit of in vitro collections, compared to field collections, is that the plant material is free of most pests and diseases. An exception is in the case of viruses, which can easily be transmitted through tissue culture, often symptomless. As international quarantine restrictions are very stringent for some species, a prerequisite for germplasm to be internationally distributed is that the sample must be healthy and free of harmful pathogens. Several in vitro tissue culture techniques, such as thermotherapy, chemotherapy and meristem tip culture, can be applied to eradicate the viruses [42]. In addition, other eradication methods such as electrotherapy [54] and cryotherapy [55] have been effective in several crops including grape, potato, sweet potato and banana. Cryotherapy using vitrification proved to eliminate effectively Cucumber Mosaic Virus (CMV) and Banana Streak Virus (BSV) from in vitro meristematic tissues of the dessert banana cv. Williams (AAA, Cavendish subgroup) with 30% and 90% of the regenerated plants being CMV- and BSV-free, respectively [56]. Often, combinations of eradication techniques are used to increase their effectiveness.

In contrast to field gene banks, plant materials kept at MTS conditions, also called in vitro "active" collections can be delivered on a year-round basis. For most crops, the availability of samples from a field gene bank is bound to the development stage of the plant and season. In garlic, for example, cloves and bulbils follow a seasonal development and are only available and highly viable between July and February in the northern hemisphere. Moreover, the supply of material from the field can be limited because of the often poor plant propagation rates. For instance, banana has an annual field multiplication rate through suckering as low as five to 20 suckers per year depending on the clone, age of the plant and climatic and culture conditions [57,58]. Furthermore, the international and inter-regional movement of propagules from the field such as suckers of banana, corm pieces or small corms of cocoyam or taro, tubers of oca, ulluco or other plant parts such as wood cuttings of apple or blueberry, involves the risk of transmission of harmful pest and diseases. Strict quarantine measure must thus be taken. For crops with insect- or mite-transmitted viruses, it is therefore useful to maintain virus-free stocks in screenhouses although this implies higher cost for making materials available to users. In general, it is recommended that vegetative material is distributed to requesters as tissue cultures, derived from conventional vegetative propagation material, and indexed free from pathogens. Moreover, tissue culture using multiplication-inducing growth regulators, often cytokinins, allows rapid and mass propagation of plants. Hence, hundreds of propagules can be produced within a few months.

#### *3.3. Problems Associated with In Vitro Collections*

The artificial tissue culture environment can be stressful for plant cells and thus poses a challenge when plant germplasm needs to be maintained for extended periods of time. This can result in so-called somaclonal variations that are changes in the DNA sequences that are not derived from recombination [59]. It can have epigenetic origin which reflects the adaptation process of cells to a different environment [60]. Different factors have an effect on the genetic mutation rate, among them exposure of the chromosomal DNA to different chemicals present in the culture medium and the survival of the variants in a non-selective tissue culture environment. Although the mutagenic activity of plant growth regulators is debated, it is generally accepted that stimulated rapid disorganised growth induces somaclonal variation [59]. This has been shown for thiadiazuron (TDZ) in bananas [61] and the synthetic auxin 2,4-dichlorophenoxyacetic acid (2,4-D) in *Curcuma* aromatic plants [62]. Bairu, et al. [63] observed that, with increasing subculture events, the frequency of somaclonal variation of micropropagated bananas increased, sometimes as high as 72%. However, in potato, random DNA methylation changes occurred only in individual samples at a very low frequency for 3 of 469 markers [64]. Furthermore, no differences in genetic and epigenetic changes have been reported for in vitro propagated pea clones that have been maintained for more than 24 years [65], indicating that somaclonal variation is strongly species-dependent. A general principle is that, the more disorganized the cultured tissues, the higher chance of mutations [59,66]. Therefore, for slow-growth storage, organized tissue systems, such as shoot cultures, are preferred over non organized tissues like callus and suspension cells and the in vitro storage is preferably conducted on hormone-free culture media, as has been used for *Mentha* species [67].

Other constraints in tissue culture collections are the occurrence of cellular ageing and senescence during prolonged cultivation. In eight-year-old peach palm (*Bactris gasipaes*) cultures, processes of plant cell death and senescence were visible through nuclear condensation, cell degradation and the development of large intercellular spaces [68]. The effect of cellular ageing or senescence may appear in parallel with slow growing endophytic microbes that can accumulate over time. Bacteria or fungi are known to colonize almost all healthy plant tissues without necessarily damaging the host or eliciting any defense responses [69]. For example, the in vitro collection of *Allium* species maintained at slow-growth conditions suffers from the presence of endophytes and the storage duration of individual clones is considerably shortened [70]. At the banana collections at the Bioversity Genebank, the tissue cultures are regularly tested for the presence of slow growing (endophytic) microorganisms using a broad spectrum bacterial growth medium. When this method was first applied, the presence of cryptic contaminants in 5% of the stored germplasm was revealed [71]. Although some endopyhtic strains are known to show plant growth promoting behavior such as PsJN in potato [72], the majority is detrimental for in vitro cultures in case they accumulate.

Aside from the challenges that are related to intrinsic processes linked to in vitro culture, external factors can also result in a complete loss of cultures, such as malfunctions in air-conditioning and lighting systems. Moreover, human errors such as accidental microbial contaminations, physical mixing of accession samples, but also documentation errors, e.g., mislabeling, or misidentification, are even more likely to cause serious problems in the operation of an in vitro genebank [73]. Furthermore, mites, thrips and other small arthropods can cause extensive fungal contaminations in tissue cultures and are difficult to eradicate [14]. Therefore, quality management systems including barcode labelling, cleaning management and regular monitoring of the stored materials should be implemented as standard procedures at slow-growth storage facilities [1]. An extra safety measure is duplicating the collection, either in vitro or in cryopreservation, and preferably in another distant location to ensure that the duplicate is properly secured. For example, the *Mentha* collection at IPK consists of two subsets that are maintained in different culture rooms one kept at 2 ◦C and one at 10 ◦C. The CIAT genebank in Colombia sends its duplicates of its cassava in vitro clones off site to CIP, Peru where they are maintained at 23 ◦C and under a controlled photoperiod. For *Musa* at the Bioveristy genebank, maintaining 70% of its in vitro clones in a cryopreserved base collection, a cryopreserved sample is safely duplicated in IRD France.

**Figure 3.** Banana in vitro collection at the International Transit Centre (ITC), Belgium.

**Box 2.** Collection of bananas at the Bioversity Genebank, Belgium.

The international banana collection, also known as the Bioversity International Transit Centre (ITC), Belgium, was founded in 1985. It is the largest repository of *Musa* species in the world holding more than 1600 banana accessions sourced from 38 countries. The genebank, located in a non-banana growing country, does not manage a field collection of its own, but has close links with national and regional field collections around the world, serving as a back-up repository for their accessions. The ITC conserves the widest diversity of cultivated sweet and starchy bananas (75%) belonging to 17 genome groups and 52 subgroups, a representation of the wild genepool with specimens of 34 species (16%), and a range of high yielding and disease-resistant advanced varieties (9%). As bananas are vegetatively propagated with seed fertility limited to the wild forms, ex situ conservation of banana species in the form of in vitro cultures is the most suitable approach when the material should remain available for regeneration, multiplication and international distribution.

Germplasm in the collection is stored as multiple shoot clusters obtained from individual shoot tips. Since the early days, low temperature, combined with light limitation, has been successfully applied as growth-retarding factors to reduce the frequency of subculturing considerably. In the storage room, the temperature is 16 ◦C and light intensity 25 μMol/m2/s (Figure 3). It was demonstrated that under the conditions applied, storage duration is nearly one year on average for the total diversity maintained. However, large differences in transfer interval, ranging between 3 and 22 months, occur among the different genomic groups and even within the same subgroup [47].

To keep the collection alive and in good condition, the active in vitro collection requires close monitoring not only to assess the viability and need for re-culturing of accessions, but also to keep the collection free from contamination [71] which may interfere with the storage and use of the germplasm. In addition, samples of accessions maintained for more than 10 years in vitro are evaluated in the greenhouse from where rejuvenated (renewed) in vitro stocks are established. To keep the levels of somaclonal variation of the stored germplasm as low as possible, older accessions are also systematically grown in the field for an integrity check and in parallel to the field detection of changes in morphology, cytological and molecular (SSR) characterization is performed [74] to confirm the identity of the accession.

Each accession in the active collection has a number of 20 replicate cultures maintained on one single multiplication-inducing growth medium, ensuring safe storage of the clone and direct availability for distribution and use. Over the years, the gene bank has played a crucial role in the international exchange of banana germplasm.

#### **Box 2.** *Cont.*

A system for the safe movement of germplasm is in place with substantial efforts going into the testing and sanitizing of the collection from banana viruses [75]. Each day, three to four clean in vitro samples of the stored banana accessions are distributed to users somewhere in the world to underpin research, breeding and development activities. Depending on the purposes and needs of the germplasm user, the required growth stage of the material may differ: if no tissue culture capacity is available, in vitro rooted plants for direct soil acclimation can be distributed whereas users may need shoot cultures in the multiplication phase for in vitro studies or as virus-free stock for further in vitro propagation. In any case, it is recommended that recipients are provided with detailed handling instructions, the composition of the growth medium and specifications of the growth conditions. Since molecular techniques have become increasingly important in biodiversity studies, scientists and breeders have a growing interest in accessing the DNA of the widest range of genetic diversity present in collections. To meet this changing need, the Bioversity banana genebank established a collection of freeze-dried leaf tissues isolated from greenhouse plants stored at −20 ◦C. Leaf samples of some 850 accessions are readily available to customers, forming a low-cost alternative for DNA-banking and for distribution of living plant material that is time consuming to prepare, requires phytosanitary checks, special packaging, and fast shipping to assure arrival in good conditions.

The first report on successful shoot tip cryopreservation of *Musa* species was published in 1996 [76]. Sucrose precultured meristem clumps belonging to seven different genomic groups were rapidly frozen in liquid nitrogen, from which the shoots were regenerated. However, regeneration rates were low (ranging between 0% and 50%) and variable depending on the cultivar. A new protocol, i.e., droplet vitrification, was therefore needed (see Section 4.1.2) [77] resulting in higher and more cultivar-independent regeneration rates (between 50% and 95%). Since then and thanks to funding provided by the World Bank and the Gatsby Foundation, routine cryopreservation of clean and field characterized banana accessions kept in the in vitro collection was established. Such secure funding sources proved to be important since cryopreservation of *Musa* species is a very labor intensive and costly process, especially because hundreds of tiny meristems need to be excised under the binocular microscope. Using the droplet vitrification protocol, one skilled technical staff member can cryopreserve about 40 to 50 *Musa* accessions per year.

The ITC is a successful example of how complementary conservation approaches can be used to increase the security of collections. With samples of 1175 accessions also maintained in liquid nitrogen (Figure 4), the cryopreserved base collection serves as source for replacing materials in case the accession samples in slow growth are lost due to accidental contamination or genetic variations. This cryopreserved collection is "backed up" (duplicated) in Cryotanks held at the Institut de Recherche pour le Development (IRD), Montpellier, France.

**Figure 4.** Cryopreservation facilities at the ITC.

#### **4. Cryopreserved Collections**

Cryopreservation (or storage of biological material at ultra-low temperatures) is the obvious solution to the above-mentioned limitations, since at these conditions metabolic, physical and chemical alterations are unlikely to occur, even after hundreds of years of storage. Usually, cryogenic storage takes place in liquid nitrogen (−196◦) or its vapor phase (between −140 and −180 ◦C) (Figure 4). The main hurdle associated with cryopreservation is the formation of lethal ice crystals. Complete drying plant tissues, thus preventing the formation of ice crystals is not an option since the presence of water is inevitably linked with life. The only way to avoid ice crystal formation of a watery solution is by making use of the physical phase called "vitrification", i.e., the solidification of a liquid forming an amorphous "or glassy" structure. All cryopreservation procedures developed for biological materials are based on optimizing the chance for vitrification. To attain this, two conditions must be met: (i) application of ultra-rapid cooling rates, limiting the time period that an ice crystal can form before all molecules are immobilized by the ultra-low temperature, and (ii) concentrating the cell solution resulting in relatively more molecules in interference with the organization of water molecules turning into ice crystals.

#### *4.1. Setting-Up Cryopreserved Collections*

Since the first report by Akira Sakai in 1965 [78] on the survival of plant tissues exposed to liquid nitrogen, a wide variety of plant cryopreservation protocols have been established, among them dormant bud cryopreservation, classical (slow) freezing, encapsulation-dehydration, and a range of vitrifications solution-based protocols (for an overview, see [79]). Currently, dormant bud cryopreservation and droplet vitrification are commonly applied (Table 1).


*Plants* **2020**, *9*, 1634

#### 4.1.1. Dormant Bud Cryopreservation

The number of crops or plants species that can be cryopreserved through dormant bud cryopreservation is limited since two requirements must be met: (i) the species produces buds that go into a dormant phase, usually induced in winter by a prolonged period of low temperature and/or photoperiod [88], before being prepared for cryostorage, and (ii) buds recovered from cryopreservation should respond to bud grafting.

A typical protocol for apple consists of (i) collecting dormant field material in mid-winter, (ii) air-dehydrating the twigs at −5 ◦C to 25–30% moisture content, (iii) subsequently applying slow freezing at 1 ◦C h−<sup>1</sup> to <sup>−</sup>30 ◦C and holding this temperature for one hour, before (iv) the twigs are plunged into liquid nitrogen for storage [89]. After rewarming, buds are grafted onto a suitable rootstock. A comparative advantage of this protocol over the other cryopreservation protocols is that, during the whole procedure, no in vitro culture phase is involved; the material for conservation is transferred, after a proper treatment, from the field to the liquid nitrogen tank and, at the time of recovering, back from the tank to the field. This model saves much time [90] and resources and reduces the risks of contamination relative to classic shoot tip-based cryopreservation methods. To date, only apple, pear (*Pyrus* sp.) [91] and sour cherry (*Prunus cerasus*) [92] have been prepared for conservation with this method. An adapted dormant bud cryopreservation protocol where cryopreserved dormant buds are not grafted but directly planted out is developed for *Vaccinium* [87]. Alternatively, buds can be recovered in vitro as in the case of mulberry (*Morus* sp.) [93] and currants (*Ribes* sp.) [94].

#### 4.1.2. Droplet Vitrification

Droplet vitrification can be considered as a "generic" cryopreservation protocol for hydrated tissues, such as in vitro cultures [77,95], as opposed to dry tissues such as seeds and dormant buds. Because of its resulting high post-thaw regeneration rates, relative user-friendliness, and applicability to many plants species, it is now the widest applied protocol for cryopreserved germplasm collections. In short, 1 to 2 mm meristem tips are excised, precultured (or not) on a sucrose medium and treated with two highly concentrated liquids called loading solution (LS) and plant vitrification solution (PVS2 or PVS3) leading to a more concentrated (and thus vitrifiable) cell solution. The dehydrated meristem tips are transferred onto a small strip of aluminum foil and directly plunged in liquid nitrogen for storage. Rapid rewarming takes place in a third liquid, called recovery solution (RS) after which the meristems are transferred onto an in vitro medium for plant recovery. Since its first report in 2005 (See Box 2 [77]), this cryopreservation protocols has now been successfully applied to 111 plant species according the publications reported in Web of Science (Situation 1 July 2020) among which some very important staple foods are found, such as potato [96–98], taro [99], cassava and yam [82], sweet potato [100], and also ornamentals (*Pelargonium* spp. [101], *Nandina domestica* [102]), fruit trees (*Actinidia chinensis* (kiwi fruit)) [103], medicinal plants (*Byrsonima intermedia*) [104] and conifers (*Sequoia sempervirens* (redwood) [105]). For most of these crops/species, cryocollections are being established.

A more recent development in plant cryopreservation is the establishment of two cryo-plate vitrification methods: the vitrification cryo-plate (V Cryo-plate [106]) and dehydration cryo-plate (D Cryo-plate [81,107]). These methods follow the same principles as the droplet vitrification method; small volumes (droplets) cool down more rapidly to the temperature of liquid nitrogen compared to big volumes. The main difference is that, with these recent methods, meristems are enclosed in tiny drops of calcium alginate placed on the aluminum plate before being dehydrated and subsequently plunged in liquid nitrogen. Some advantages of the V Cryo-plate and D Cryo-plate method are (1) simple procedure to store plates in liquid nitrogen, (2) processing large numbers at a time and (3) secured high cooling and heating speed [108]. Post-cryopreservation regeneration rates of these methods are not significantly different from the droplet vitrification method [81], therefore, the choice of method to use is personal preference.

Routine cryopreservation of crop collections began only a few decades ago. Currently, about 18 genebanks have cryopreserved crop collections [13,67,81,98,109]. It is estimated that about 100,000 unique accessions of vegetatively propagated and recalcitrant seed crops potentially need long-term conservation through cryopreservation while currently only about 10,000 accessions are cryopreserved [13].

#### *4.2. Advantages of a Cryopreserved Collection: Safety Backup for Clonally Propagated Crops*

Cryopreservation is a cost- and labor-efficient conservation method that insures genetic stability over time [70], it may also be used for establishing a backup. For orthodox seeds such a safety backup facility already exists—the Svalbard Global Seed Vault (SGSV) in Norway. This facility is built by the Norwegian government and operated by the "Global Crop Diversity Trust" and NordGen, to store safety duplicates of national and international seed collections [110] to protect the diversity from irreversible loss due to natural and human-caused disasters. The SGSV consists of chambers maintained at −18 ◦C, dug into a mountain in the permafrost on Spitzbergen island at the arctic circle. Currently more than one million seed samples are conserved there by many national and international research institutes and genebanks.

Clonal crops can be duplicated in the field or in vitro but the ultimate backup may include a cryopreserved duplication of the conserved accessions. An additional advantage of the cryopreserved backup compared to the global seed vault is that the cryopreservation backup would serve for hundreds of years and does not need to be regenerated after a few decades of storage, depending on the species after 50 to 100 years, as is the case for seeds. Periodic regeneration of cryopreserved tissues is, however, necessary to evaluate successful conservation.

In 2017, a feasibility study concluded that a safety backup facility is currently required to accommodate 5000–10,000 accessions arising from ongoing cryopreservation activities [13] and that this facility should be expanded in a later phase to host all unique clonally propagated crop accessions. Such facility should operate according to the same policies and principles that govern the SGSV.

#### *4.3. Problems Associated with Cryopreserved Collections*

The range of crops represented in "cryobanks" is still rather restricted, and more than 90% of these accessions are composed of a few crops such as potatoes, cassava, apple, bananas and plantains, mulberry, garlic and strawberry. The main reasons why cryopreservation for the long-term conservation of vegetatively propagated crops is not applied more widely and on a larger scale are reviewed in detail by [13]. These challenges are linked to (i) protocol development: for many plant species efficient cryopreservation protocols are not available yet, (ii) problems with implementation of existing cryopreservation protocols such as genotype-specific responses and insufficient supply of healthy plant material, and (iii) challenges related to cryobanking capacities such as insufficient funding, lack of skilled personnel with knowledge on plant genetic resources [111] and lack of equipment/infrastructure.

#### **5. Identification of Unique Accessions and Elimination of Duplicates**

The elimination of duplicates in clonal collections has been an ongoing task since their establishment because (i) core collections need to be set up that contain the widest possible genetic diversity within the smallest number of accessions, and (ii) costs associated with maintaining clonal collections are high. CIP initially gathered more than 15,000 accessions of native potato cultivars from Latin America. Morphological characteristics and electrophoretic bands were used to identify duplicates and to reduce the collections to about 3500 accessions [17]. A reduction of 13% of the duplicates found in white clover resulted in a per accession saving of USD 500 regeneration cost in the field [112]. Additionally, other collections were screened for their redundancies. In eight Dutch apple collections, molecular fingerprinting revealed that 32% of the accessions were duplicates [113], a similar ratio to the duplicate proportion has been found in a barley seed collection [114]. In the US NPGS, about 12.5% of 1910 apple accessions were considered to be identical [115] comparable to the redundancy of a natural population of cacao [116]. Current next-generation sequencing technologies allow an increase

in the number of studied alleles from about 10 SSR markers to several thousand SNPs and elucidate the relationship and ancestries between accessions and species resulting in a better identification of duplicates. The elimination of valuable genotypes has thus to be carefully balanced against the cost of the active preservation.

#### **6. Costs Associated with the Di**ff**erent Conservation Methods**

The aim, costs and safety considerations of a collection are important arguments in defining the necessary conservation approaches. To overcome the major drawbacks of field conservation, in vitro and especially cryopreservation are an option to secure clonal crop collections. The introduction of the genetic resources into in vitro conditions or cryopreservation requires well-equipped laboratories and trained personnel. It also requires the development of crop–specific growth media, the optimization of growth conditions, and the development of cryopreservation protocols. In 2000, the maintenance costs for one cassava accession at CIAT, Colombia, was estimated to be USD 7, USD 25, and USD 43 in the field, in vitro and in cryo conditions, respectively [14]. For garlic, the cost to maintain nine garlic plants (one accession) in the field and replant them annually is EUR 47, about 81% of this amount is due to labor costs. The introduction of one garlic accession into cryopreservation is about EUR 363, of which 52% is labor costs. The type of plant material has a strong impact on the cost of cryopreservation. When bulbils/inflorescences can be used, about EUR 318 is required whereas when cryopreservation is carried out with in vitro plants, the costs can increase to EUR 433 and up to EUR 557. The total depends on the number of in vitro subcultures required to achieve the appropriate number of plants for cryopreservation. Subsequent maintenance of such cryopreserved accessions, however, is rather low since this only demands regular filling up of the cryotanks with liquid nitrogen. Therefore, the break-even point when costs of field conservation are equal to that of cryopreservation is achieved after 8 to 13 years [21]. This estimate is very similar to the calculation made for bananas where the cumulative costs between in vitro and in cryo equate after 15 years. In general, the in vitro initiation of a banana accession at the ITC Leuven, Belgium, costs USD 861. The cost for cryopreserving such an accession is USD 1296. The costs accumulate differentially and are assumed in-perpetuity to reach USD 3686 and USD 1431, respectively. Nowadays, only few in vitro collections, such as the 1200 garlic accessions in Korea [85], about 150 mint accessions at IPK [67] and 150 citrus accessions in NPGS are completely preserved through cryopreservation [117]. However, it has to be considered that the material is only immediately available for distribution, characterization and evaluation when kept either in the field or in vitro.

#### **7. Conclusions**

The objective of a genebank is to conserve crop genetic resources. The advantage of a field genebank is that characterization and evaluation can be performed on mature plants at limited additional cost. Yet, the method has significant restrictions regarding its security, costs, and sustainability [14]. In this respect, in vitro conservation offers advantages over field conservation and is recognized as an invaluable complementary approach to secure the diversity of the collection. With the potential that tissue culture protocols can be developed for practically any plant species, the technique has been successfully applied over the past decades for short- to medium-term conservation of a wide range of crop species [118]. While offering the possibility of disease elimination and rapid clonal propagation of healthy plants, it is a suitable method for exchanging germplasm accommodating national and international phytosanitary requirements and regulations. However, the possible occurrence of genetic instability as a result of enhanced selection pressure under in vitro conditions compared to in vivo may be an obstacle to its use for long-term preservation of plant germplasm [49]. The chance that somatic mutations occur in tissues maintained at the ultra-low storage temperature of liquid nitrogen (−196 ◦C) is very slim since all metabolic processes are suspended [119]. In this regard, cryopreservation is the best option for unlimited and secure conservation of regenerative tissues in the long term. Nevertheless, the latter also requires adequate infrastructure, reliable electricity provision and well-trained staff.

The global COVID-19 pandemic has reminded us that particularly clonal ex situ collections in the field and in vitro that require continuous maintenance are vulnerable and insufficiently secured. "Essential" activities for maintenance of the germplasm in gene banks were temporarily interrupted for concerns out of human safety and health reasons, hence putting valuable genetic resources at risk of loss. While the initial costs of developing a crop- or species-specific cryopreservation protocol [120] and labor input for processing the samples in liquid nitrogen are high, once established, the maintenance of a cryopreserved collection requires very little input of resources including human intervention, making it probably the most cost-effective and secure option for long-term conservation.

**Author Contributions:** All authors contribute equally to this manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** B.P. and I.V.d.h. gratefully acknowledge the Gene Bank CGIAR Research Programme and the CGIAR Research Programme on Roots, Tubers and Bananas (RTB), and the Directorate-general Development Cooperation and Humanitarian, Belgium (DGD) for financial support of the project "Safeguarding vegetatively-propagated crop diversity to nourish people now and in the future". In addition, many thanks to Yu-Chun Liao and Maurizio Lambardi for critical proof-reading the text.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Seed Germination after 30 Years Storage in Permafrost**

#### **Svein Øivind Solberg 1,\*, Guro Brodal 2, Roland von Bothmer 3, Eivind Meen 4, Flemming Yndgaard 5, Christian Andreasen <sup>6</sup> and Åsmund Asdal <sup>5</sup>**


Received: 13 April 2020; Accepted: 29 April 2020; Published: 2 May 2020

**Abstract:** More than 30 years ago, the Nordic Gene Bank established a long-term experiment on seeds stored under permafrost conditions in an abandoned mine corridor in Svalbard, as a tool to monitor storage life under these conditions. The study included seeds from 16 Nordic agricultural and horticultural crops, each represented by two or three cultivars (altogether 38 accessions). All seeds were ultra-dried to 3–5% moisture before being sealed in glass tubes. Germination tests were performed in accordance with the International Seed Testing Association (ISTA) protocols. At the initiation of the experiment, the samples showed good germination with the median value at 92%. The overall picture remained stable over the first twenty to twenty-five years. However, the variation became larger over time and at 30 years, the median value had dropped to 80%. At the lower end, with a high drop in germination, we found rye, wheat, and English ryegrass. At the upper end, we found Kentucky bluegrass and cucumber. The lowest germination was found in samples with the highest initial seed moisture levels. Pre-storage conditions are likely to be of major importance for longevity.

**Keywords:** ex situ conservation; germination; longevity; plant genetic resources; seed storage

#### **1. Introduction**

Most food plants produce seeds that can be stored under low temperature and moisture conditions. Much of our knowledge on seed longevity is based on artificial ageing experiments, where seeds are exposed to suboptimal conditions of elevated temperature and moisture for some weeks, and storage life is predicted based on the seed moisture content, storage temperature, and seed lot characters [1,2]. Such calculations have predicted that high-quality seeds could survive ideal conditions for hundreds of years or more [3,4], which was good news for many gene banks, but might be unrealistic as such studies have rarely been confirmed in long-term storage studies. Long-term seed storage is crucial for ex situ gene bank conservation [5–7]. Gene banks maintain crop diversity and facilitate the utilization of seeds for breeding, research, education, and other purposes [8–11]. Safety back-ups are kept, ideally at a second location, to spread the risks [12,13]. The Global Seed Vault at the Arctic Archipelago of Svalbard was opened in 2008 and is facilitating such back-up collections with a world outreach [14]. However, more than twenty years before this, the Nordic Gene Bank (NGB) started a small seed

storage facility in an abandoned coalmine corridor in Svalbard. Different options had been considered, such as inland ice caves in Greenland or mountain caves in Jotunheimen, Norway, but in the end, a coalmine in the permafrost in Svalbard was chosen due to its good logistics, despite its remote location [15]. The temperature of the Global Seed Vault in Svalbard *was* −18 ◦C compared to the −3.5 ◦C present in the abandoned coalmine corridor. The rock temperature was stable, which meant that it was independent of an external energy supply. As a tool to monitor storage life under these permafrost conditions, a long-term seed storage experiment was initiated. The experiment started in 1986 and included samples for germination monitoring until 2086, thus it was termed "the 100 year experiment". Important crops for Nordic agriculture and horticulture were included. The investigation is still ongoing, and in this paper, we summarize the results after the first 30 years.

#### **2. Results and Discussion**

#### *2.1. Overall Patterns*

The overall picture of seed germination development over storage time across all accessions (species and cultivars) is illustrated by boxplots (Figure 1). A lower and upper percentile defines the box, in which 75% of the observations were found. A marked line denotes the median germination value, and the whiskers and small circles show observations away from the box. At the initiation of the experiment in 1986 (year 0 = y0), most of the seed lots showed an excellent germination ability. The box-range was from 83% to 95%, and the median was at 92% germination. The overall picture was relatively stable over the first twenty to twenty-five years, but the variation increased over time as some seed lots showed a reduced germination. After 30 years, the median value across all the lots was 80%, but with outlier samples below 40%, and some of the lots in the 50–70% range.

**Figure 1.** Boxplots showing the germination percentages across species and cultivars throughout the first thirty years (year 0, year 2.5, year 5, year 7.5, year 10, year 12.5, year 15, year 17.5, year 20, year 25, and year 30).

Figure 2 shows a dendrogram of a cluster analysis of the germination results of the accessions. The dissimilarity values of the fusion level values of the dendrogram indicated the cutting level six clusters to be correct. The largest cluster, cluster 1, contained 19 lots. It contained both normal and lots with seed-borne pathogens (i-prefix lots). Also, cluster 2 contained both normal and seed-borne pathogens, as did cluster 6. In cluster 1 we found Barley 1, Ryegrass 1, Ryegrass 2, Timothy 2, Bluegrass 2, Redclover 2, Rape 1, Rape 2, Onion 1, Onion 2, Carrot 1, Carrot 2, Cauliflower 2, i1 Wheat, i2 Wheat, i Barley, i Meadow\_fescue, i Onion, and i Cabbage. In cluster two, we found Barley 2, Bluegrass 1, Beet 2, Lettuce, Cabbage, Cucumber 1, Cucumber 2, Cauliflower 1, and i-Timothy. In cluster three, we found Wheat 1, Timothy 1, Redclover 1, and i Carrot. In cluster 4, we found Wheat 2. In cluster 5, we found Rye 1 and Rye 2. In cluster 6, we found Beet 1, i Lettuce, and i Beet. Lots with seed-borne pathogens (i-prefix lots) were spread over the clusters, showing that such pathogens do not explain much of the seed longevity. Furthermore, seed lots of the same species were only partly in the same clusters; for example, the two cucumber (*Cucumis sativus* L.) lots and the two rye (*Secale cereale* L.) lots. This also shows that interspecies differences are not of importance for explaining seed longevity.

**Figure 2.** The dendrogram of a cluster analysis of the germination results of the seed lots.

The seeds of all the species remained viable after 30 years in permafrost under the given conditions (dried to 3–5% moisture content and stored in sealed glass ampoules). Crops, but also cultivars, within the same crops showed different results. Similar patterns have also been observed in other long-term experiments, both under ambient [16,17] or −18 ◦C conditions [18–21]. The cold storage of seeds should provide improved longevity compared to ambient storage [22]. Another factor of importance is seed maturation [23,24].

For instance, the potential seed longevity of barley was found to be best if it was harvested a week or so after grain filling was completed [25]. The same has been found in tomatos (*Solanum lycopersicum* L. var. *lycopersicum*) [26]. Furthermore, weather, seed coat damages, diseases, and pests may influence seed storage life [24,27]. In our experiment, the samples were all from newly harvested seeds, but we do not have data on the weather or other pre-harvest conditions. We only assume that the seed lots selected for the experiment were of a high quality.

#### *2.2. Crop-Wise Results*

Our results show good performance for many of the vegetables, while the picture is more varied for cereals and forage species (Figure S1). Table 1 gives the germination result for each seed lot. According to the FAO's gene bank standard [13], "regeneration shall be carried out when the viability drops below 85 percent of the initial viability or when the remaining seed quantity is less than what is required for three sowings of a representative population of the accession." Seed lots with the highest loss in germination, here defined as a loss in 15% or more over the 30 years, were found in both of the rye lots, two of the three English ryegrass (*Lolium perenne* L.) lots (Ryegrass 1, Ryegrass 2), two out of four wheat (*Triticum aestivum* L.) lots (Wheat 1, Wheat 2), one of the three timothy (*Phleum pratense* L.) lots (Timothy 1), one of the three barley (*Hordeum vulgare* L.) lots (Barley 1), and in the one seed lot of meadow fescue (*Schedonorus pratensis* (Huds.) P. Beauv.). The literature shows that especially rye, but also some forage grasses, can be relatively short-lived [28–31]. Intermediate storage performance, with a 5–15% loss in germination over the 30-year period, was found in two of the three lettuce (*Lactuca sativa* L.) lots (i Lettuce, Lettuce 2), two of the three carrot (*Daucus carota* subsp. *sativus* (Hoffm.) Schübl & G. Martens) lots (Carrot 2, i-Carrot), one of the four wheat lots (i1-Wheat), one of the three barley lots (Barley 2), one of the three timothy lots (Timothy 2), one of the two cauliflower (*Brassica oleracea* L. var. *botrytis*) lots (Cauliflower 2), one of the two oilseed rape (*Brassica napus* L.) lots (Rape 1), and one out of the three red clover (*Trifolium pratense* L.) lots (Redclover 1). The most long-lived, with a loss in germination less than 5% after 30 years of storage, were found in all of the three beet (*Beta vulgaris* L.) lots, all of the three onion (*Allium cepa* L.) lots, both of the cucumber lots, both of the Kentucky bluegrass (*Poa pratensis* L.) seed lots, the only cabbage (*Brassica oleracea* L.) seed lot, and in the last of the red clover, cauliflower, barley, wheat, oilseed rape, carrot, and lettuce seed lots. Other studies have also shown that beet seeds as well as cucumber seeds have retained a high germination level over time, but, in contrast to our study, onions are generally found to be short-lived [17–19,29–31]. Our results are a little surprising as we found that onion, and to some extent lettuce, showed no decline in germination over the 30-year period.


**Table 1.** Germination percentages for all the seed lots (values below 70% in bold), and loss in germination (Δ germ) over the first 30 years, calculated by averaging the germination percentage of the three first test occasions (year 0, year 2.5, and year 5) minus the last test occasion (year 30).


**Table 1.** *Cont.*

#### *2.3. Moisture Measurements*

For all samples, the variation between the highest and lowest moisture content over the ten years was between 0.3% and 0.8% (Table 2). The data showed that two lots had a higher moisture content exceeding 5% in the initial test; these were the samples Wheat 2 'Solid', with 6.3% humidity, and Rye 1 'Petkus II', with 5.3% humidity. For wheat, one seed lot showed a low decline, and two seed lots showed a steep decline in germination. Different pre-harvest conditions or genetic factors may explain a steep decline in germination. We saw an effect of drying the seeds to lower than 5% internal moisture content before packing. The two samples with the highest internal humidity (Wheat 2 and Rye 1) were among the ones that showed the most significant drop in germination in our experiment. The current FAO standards [13], which are used by most gene banks, recommend drying for three months at 15 ◦C and 15% RH. According to our experience, this would give a seed moisture content exceeding 5%, thus decreasing the longevity.

**Table 2.** Seed moisture content (in %) in the different lots over the first ten years of the experiment.


#### **3. Conclusions**

The study has so far revealed valuable results concerning the longevity of seeds after 30 years in permafrost. Nine out of the 38 seed lots showed a germination loss exceeding 15%, which is the level recommended by the FAO for carrying out regeneration. Rye and ryegrass in particular showed a rapid decline, while many of the vegetables showed a low decline in germination. The results are relevant for the seeds in the first Nordic back-up collection stored in the abandoned coalmine. A given storage condition is an essential characteristic for comparing results with other experiments. Here, seeds were stored in permafrost with a stable sub-zero temperature. We had no reference material at −18 ◦C conditions. Despite the limited number of samples and the lack of a −18 ◦C control, the observations add knowledge about the longevity of seeds.

#### **4. Materials and Methods**

#### *4.1. Seed Samples and Seed Storage*

In total, 38 seed lots covering 16 crops were included in the study (Table 3). Each species was represented by two to four seed lots, except for cabbage and meadow fescue, which had only one. The seed lot ID was given by crop name and code. A number after the crop name (1 or 2) refers to seed lots where we have no information on seed-borne diseases. A prefix "i" letter applies to seed lots where we know that seed-borne diseases were present.

**Table 3.** Overview of the examined crops and sample information. A prefix number refers to the samples with no information on seed-borne diseases, while a prefix i-letter refers to the samples where we detected seed-borne diseases to be present at the start of the experiment.


The lots were from different cultivars and origins. No cultivation details are available except that the seed lots should be of good quality. Initial germination tests were conducted (y0). Before storage, all the seeds were dried to 3–5% internal moisture content using a Munters (Kista, Sweden) dehumidifier adjusted to 10% RH in a room at 25 ◦C. After that, they were placed in a freezer overnight before they were sealed in glass ampoules. Each seed lot was divided into sub-samples of 2 × 500 seeds for each withdrawal. The sub-samples were boxed and labelled with the date for withdrawal, transported to Longyearbyen, Svalbard, and placed in the abandoned coalmine corridor. The temperature inside the coalmine was measured as −3.5 ◦C (± 0.2 ◦C).

#### *4.2. Germination Studies*

The first set of samples was tested in December 1986 (year 0), with a plan to have the last tests in 2086 (year 100). Since the start of the experiment, 11 boxes have been retrieved. The moisture content of the seed was monitored from year 0 to year 10 (Table 4). The germination tests and the moisture content tests were carried out at the Kimen Seed Laboratory (Ås, Norway). The testing followed the International Seed Testing Association (ISTA) rules [32,33]. The details of the germination conditions for the different species in the study are shown in Table 5. In the case of oilseed rape and onion, the number of days to final count was, however, larger than the number of days in the ISTA protocols. In accordance with the protocols, we germinated 4x100 or 8x50 seeds per sample and each replicate was compared to the mean. If a large variation was detected, the samples were re-tested. Different types of filter paper (Seedburo Equipment Co, Des Plaines, IL, USA) were used as substrates for the germination tests.


**Table 4.** Dates of the retrieval of the samples over the first 30 years, and types of tests conducted.

**Table 5.** Germination conditions for the different crops included in the experiment.


For cereals, in-between paper methods (BP) were used; the seeds were placed on one moist paper with a second paper on top, and thereafter rolled and placed vertically in plastic. Inside the roll, the seeds germinated, and the seedlings developed. For grasses, red clover, and vegetables (except for beet), a Jacobsen apparatus was used [34], which is a plate where circular pieces of wet filter paper and seeds are placed and kept moist by a wick (TP). The seeds of beet were germinated between humidified pleated filter paper. The germination temperature varied from alternating between +20 ◦C for 16 h and +30 ◦C for 8 h to constant ambient temperatures. Lettuce required light for germination, but for

the other species light was not applied. Tests of actual moisture content were made along with seed germination tests in year 0 and the first 10 years (Table 2).

#### *4.3. Data Analyses*

R software was used for statistical examination. One value was missing for year 5 for red clover (cultivar Jokioinen). It was replaced with the average of the four nearest neighbor values: 76.0%. The short descriptor names of the eleven test results are year 0, year 2.5, year 5, year 7.5, year 10, year 12.5, year 15, year 17.5, year 20, year 25, and year 30, where year 0 represents the results of the trial that started in 1986. Summary statistics and boxplots were used to overview the data. The R function 'time series' was used to illustrate the fluctuation of the results over the 11 test occasions over the first 30 years. We calculated the loss in germination (Δ germ) by averaging the germination percentage of the three first test occasions (y0, y2.5, and y5) minus the test occasion at y30. We then categorized the samples into three: the most short-lived group (with Δ germ of 15% or more, which is at a level where regeneration should have been carried out), the intermediate group (Δ germ 5% to 15%), and the most long-lived (Δ germ less than 5%, or with less than 95% germination loss over the 30 years).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/9/5/579/s1, Figure S1: Illustration of germination for each seed lot included in the trial.

**Author Contributions:** Conceptualization, F.Y.; Methodology, G.B., F.Y., and E.M.; Formal analysis, F.Y.; Investigation, S.Ø.S. and F.Y.; Resources, Å.A.; Data curation, F.Y.; Writing—Original draft preparation, S.Ø.S.; Writing—Review and editing, C.A., Å.A., G.B., and R.v.B.; Visualization, F.Y.; Supervision, Å.A. and R.v.B.; Project administration, Å.A.; Funding acquisition, S.Ø.S. and Å.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** We also thank the Nordic Joint Committee for Food and Agricultural Research for network funds to publish the results from the first 30 years of the long-term experiment (Grant number SLU 202100-2817).

**Acknowledgments:** We would like to thank Store Norske Spitsbergen Kulkompani for the seed storage and for organizing transport. A special thanks to the technical staff at the Kimen Seed Laboratory/Norwegian State Seed Testing Station for the exact and laborious germination and moisture tests, and to H. Tangerås who administered, and was responsible for, the laboratory work for most of the 30 years. We would also like to honor the Norwegian Ministry of Agriculture and Food and the Nordic Gene Bank/NordGen staff involved in this long-term seed trial.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analysis, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Progress and Challenges in Ex Situ Conservation of Forage Germplasm: Grasses, Herbaceous Legumes and Fodder Trees**

#### **Jean Hanson 1,\* and Richard H. Ellis <sup>2</sup>**


Received: 19 February 2020; Accepted: 12 March 2020; Published: 2 April 2020

**Abstract:** Forages provide an important livestock feed resource globally, particularly for millions of smallholder farmers, and have important roles in natural resource management and carbon sequestration, reducing soil erosion and mitigating the effects of climate change. Forage germplasm remains the basis for the selection and development of new, higher-yielding and better adaptedgenotypes to meet the increasing demand for livestock feed. Rapid rates of genetic erosion of forage diversity due to land-use change from natural pastures and rangelands to crop production to meet the food security requirements of a growing global population, together with pressures from a changing climate, highlight the necessity for ex situ seed conservation of forage genetic resources to provide germplasm for use by future generations. Whilst many forage species have orthodox seeds, the diverse range of genera and species which provide forage is a challenge in terms of the wide scope of information and understanding on conservation methods that genebank managers require—particularly for tropical forages, many of which are comparatively under-researched. We review the challenges to the conservation of tropical forage species by seed in ex situ genebanks and provide information on optimum methods for their management.

**Keywords:** genebanks; forage germplasm; grasses; legumes; seed storage; conservation; seed longevity; seed germination; monitoring; regeneration

#### **1. Introduction**

The term 'forage' encompasses many different types of plants used for livestock feed, including grasses, herbaceous and tree legumes, as well as other non-leguminous forbs and trees varying greatly in plant habit and adaptation. Forages are an important livestock feed resource globally, particularly for millions of smallholder farmers who rely on natural pastures and grasslands as the basis of their sustainable livestock systems, allowing ruminants to convert feed that cannot be used directly by humans into milk and meat to provide essential nutrients that are required for human health, growth and cognitive development [1]; these animals also provide fibre and skin for clothing and footwear. Forages are effective in maintaining the natural resource base [2]. They stabilise the soil, provide ground cover and windbreaks to prevent or reduce soil erosion, increase soil carbon content by strong rooting and decomposition of leaf litter, and, through symbiotic nitrogen fixation by rhizobia with legumes, they capture nitrogen from the atmosphere to the soil. There has been much research aimed at capturing these benefits, and there are now many alternative ways of introducing sown forages into the farming system [2].

Most forage diversity originates in natural grasslands, one of the largest and most important natural ecosystems [3] covering over 52 million square kilometers globally. Grasslands are home to over a thousand forage species with actual or potential use to support livestock production systems in tropical and sub-tropical areas, although only about 70 species have been commercially developed, mostly through the direct selection of germplasm accessions or by plant breeding in few important species, as feeds in these regions (Table 1). Rapid rates of genetic erosion of forage diversity are occurring in grasslands, mainly due to land-use change converting natural pasture and marginal areas to crop production to meet the food security requirements of a growing global population. This erosion together with pressures from a changing climate, resulting in extreme weather events (such as drought, floods and cyclones), has a negative effect on natural habitats where forage diversity is still found and highlights the necessity for ex situ conservation of forage genetic resources to provide diverse forage germplasm for the use of current and future generations. Most forages in use today are wild species or genotypes selected from wild populations and safeguarding sources of forage germplasm is particularly important because with few forage breeding programmes [4]—and most of those focusing mainly on temperate grassland species, including alfalfa, ryegrasses and clovers—forage germplasm remains the basis for the future selection and development of the new feeds which are particularly needed for tropical and sub-tropical regions.


**Table 1.** Common tropical and sub-tropical forage species and climate zone suitability for cultivation.


**Table 1.** *Cont.*

Zone definitions; Arid zone-100–500 mm rainfall, 0–180 growing days per annum; Semi-arid zone-600–1000 mm rainfall, 0–180 growing days per annum; Sub-humid zone-1000–1500 mm rainfall, 180–270 growing days per annum; Humid zone- >1500 mm rainfall, >270 growing days per annum; Highland zone- >1500 m altitude.

Forages are not as well represented as food crops in ex situ collections with almost 182,000 accessions representing over 1000 species of grasses, legumes and fodder trees maintained in 80 national and international genebanks registered in Genesys (www.genesys-pgr.org) compared to the approximately 7.4 million plant accessions stored in around 1750 genebanks globally [5]. These accessions registered in Genesys were collected from a wide range of sites across tropical and temperate regions (Figure 1) and contain a large amount of diversity with potential use in forage breeding. However, despite the diversity available in ex situ collections, the germplasm from these genebanks remains underutilised in crop and forage breeding programmes. Current genomic technologies that can be used to screen large collections efficiently offer opportunities for increased identification of useful genes and the use of germplasm [5].

#### **2. Management of Forage Germplasm**

Seed genebank protocols have been developed with both crop and wild species in mind, but the focus of most genebanks is very much crop genetic resources conservation with an acknowledgement of the priority to include crop wild relatives in collections. Whilst the knowledge base and dominant thinking was around conservation of crop plants, the same principles have been applied successfully to the ex situ conservation of wild plants [6,7]. The principle difference, and source of problems in implementing standard protocols, is that seeds from wild populations, crop wild relatives and forages tend not to show the high intra-seed lot uniformity of crop seeds [7]. Seed of forage species show considerable variability within and between seed lots and, similarly, within and between accessions of a species that make it difficult to develop inclusive protocols.

In the case of forage species, the information available on seed handling from seed collection through storage to germination, multiplication and regeneration is less than for the major staple food crops, and many of the problems facing forage plant genebank managers are similar to those facing those managing wild species genebanks:


Nonetheless, it is wrong to imply that little is known about seeds of forages: Table 2 summarises examples of selected sources of information on protocols and advice on seed production, dormancy, germination, and survival in genebank storage for various genera of forage grasses and forage legumes. The International Seed Testing Association (ISTA) has included germination test methods for many forage species in the recent rules for seed testing [8]. Detailed information on seed longevity of forages is also included in several publications [9–12]. Even for species or genera in which advice cannot be found, such information sources are useful to indicate how to handle seed of similar species or genera. Nonetheless, some of this information and advice was developed for commercial seed production and may need to be interpreted and adapted for use by genebanks to support the ex situ conservation of forage genetic resources. Some of the particular difficulties in handling forage seed accessions in genebanks stem from the tendency of forage species to show substantial within-accession variation in the timing of flowering, seed development and maturation; they reflect plant diversity in general. As a result, few generalisations are possible when discussing the production, harvesting, and pre-storage processing of seeds of forage species. These diverse species can be determinate or indeterminate in their development; annual, biennial, or perennial; flowers can be self-fertile, require cross-pollination or are apomictic, or a mixture of the three; and flowering may often continue for several weeks [13–19]. Moreover, the spike in most grasses is fragile, with abscission at the rachis or rachilla, and thus, seeds shed to the ground as they mature [20]. Once mature, the seeds also differ in dormancy, germination requirements and in survival during storage [21].


**Table 2.** Examples of information sources (reference numbers shown) on methods of seed production; dormancy (and dormancy-breaking methods), germination (germination testing and promotion for accession monitoring), and seed survival in genebanks (seed survival periods of different genera of forage grasses and forage legumes in genebanks).


**Table 2.** *Cont.*

#### **3. Germplasm Collecting**

One major difference between cultivated and wild forages and the major food crops is the collection location and strategy. Wild forages are found in grasslands and natural pastures, either with small numbers of plants distributed over a large area or as large areas of single species in open grasslands, compared to crop germplasm, which is found as landraces in farmers' fields. Whilst crop landraces have more uniform maturity and little seed shattering, forages tend to vary within accessions in the timing of seed maturity, and in some grasses, seed shattering occurs as seeds ripen or pods dehisce in some forage legumes. This results in some seeds being collected whilst they are immature and unable to survive the drying and storage process, resulting in poor quality or small numbers of seeds being stored [6]. In some species, a post-harvest ripening treatment can be applied, keeping seeds in high humidity to facilitate ripening [22]. With only small numbers of seeds being collected for individual accessions, seed multiplication is usually required before conservation compared to crop landraces, which can often be collected in the required quantity of seeds to go straight for conservation, safety duplication and distribution. In some cases where the species covers large areas, such as in native grasslands, random collection strategies can be applied similar to those used for crops. Some forages must be collected as cuttings because at collection time there are few ripe seeds due to shattering, as reported in the germplasm of *Panicum coloratum* [23], or the species is a poor seeder and only a small percentage of the florets result in caryopses, as in Napier grass (*Pennisetum purpureum*) [24].

#### **4. Multiplication and Regeneration**

Regeneration and multiplication are especially challenging for forages; it can be a slow process for perennial species and accessions represented by few seeds. As explained above, the original sample size of many accessions is small, and most need to be multiplied at the outset to provide sufficient seeds for long term storage. In the absence of information on the amount of outcrossing in many forage species, each accession should be assumed to be fully outcrossing and thus isolated to ensure the maintenance of genetic integrity. In addition, there is a high risk of loss of diversity and changes in genetic integrity due to the small sample size available to use for the regeneration of accessions with few seeds [25]. Accessions of slow-growing and maturing fodder trees often remain in the field for several years before they are capable of flowering to produce enough seeds to meet storage, distribution and safety duplication needs. This results in limited availability of accessions of some species. General guidelines for regeneration procedures for forages have been developed for grasses and forage legumes [26,27].

Protocols for large-scale seed production and multiplication and the basic problems inherent in forage seed multiplication, collection, and regeneration are similar but not identical. Many forages are recognised as crops, but they tend to be selected for vegetative rather than reproductive traits. Thus selection for long periods of vigorous vegetative growth by forage plant breeders to benefit livestock production has had to be tempered by the requirement for the plants to ultimately flower and produce sufficient seeds if new varieties are to be multiplied and disseminated [28]. Moreover, many forages have not been subject to plant breeding, which has focused on *Medicago* species, clovers, ryegrasses, oats, *Brachiaria* grass, Guinea grass and Napier grass. These species have been used in crossing programmes and named varieties are commercially available from seed companies, whilst other forage species are selections from the wild and are naturally adapted to a wide range of contrasting habitats. For range grassland species, for example, to persist requires seeds able to survive in the soil seedbank for considerable periods after shedding, and/or to survive passage through the animal gut. Such characteristics require traits of (considerable) seed dormancy and/or hardseededness, which hinder the promotion of prompt germination when accessions are sown for testing, regeneration, or utilisation. Hence the promotion of seed germination can require considerable intervention, including the use of growth promoters, such as potassium nitrate and gibberellic acid, as well as seed coat removal in grasses and testa scarification in legumes and/or long test periods [21,29].

The major consequence of these inherent traits of forages for genebank regeneration is that as seed harvest time approaches, the seeds are at different stages of development and maturation on any one day, with the likelihood of seed being shed from plants increasing as the harvest is delayed, whilst other seeds ripen. Shedding not only reduces the numbers of seed harvested but may also cause subsequent confusion in the gene bank. Seeds of grass accessions may appear dormant to some, whereas the problem is one of "empty seeds" whereby seed were shed, but the firm seed covering structures may give the impression that the seed remains present, e.g., in *Brachiaria humidicola* [30]. Three different methods are widely used to identify and quantify empty seeds within accessions [29]: dissection (particularly of seeds that fail to germinate when tested); X-radiography; and (rarely used by genebanks to identify empty seeds but routinely used for seed cleaning) the use of a seed blower to separate by density the empty seed fraction. X-radiography can be combined with a subsequent germination test on the same seed sample and so save valuable seeds and provide a permanent image for records [29]. "Empty seeds" are not expected in forage legumes, but X-ray images can identify poorly-developed seeds and also those with insect damage.

The decision as to when to harvest forage seeds is, therefore, a compromise between the quality and uniformity of the seed and maximising their number. Consequently, many genebanks undertake the time- and labour-consuming daily sequential harvesting of ripe seeds: first, to maximise the number of seeds harvested; and second, to reduce selection and change in genetic integrity of the accession which would occur if only early- or late- flowering and maturing plants were selected.

In terms of when to harvest seed, there is no single point in developmental time across all species and all environments where seed quality (subsequent seed storage longevity) is maximal [31]. Nonetheless, harvesting when seeds are at, or almost at, harvest maturity (i.e., their moisture content is close to equilibrium with ambient relative humidity) is a reasonable solution to this dilemma because although maximal seed quality may have been attained prior to this time net seed deterioration is rarely observed on any scale before harvest maturity [31]. This tallies with forage seed production practices where the harvest is when the earliest to develop seeds or pods change colour as they mature visually, to, for example, yellow or brown, or shortly before they would be shed; and similarly in wild species more generally [6,32]. Whilst the lack of uniformity in seed maturation in forages means that the seed crop varies in the timing of harvest maturity, where direct harvesting is not possible, forage seed producers use a variety of seed harvesting approaches to reduce the consequences of variation in developmental timing (Table 3). Combing, bagging, and suction tend to be associated with multiple, sequential harvests of seed from plants where seed maturity date varies considerably—and these approaches are all used by forage plant genebanks.


**Table 3.** Seed harvesting techniques for accessions which show considerable plant-to-plant variation in the timing of seed maturity.

#### **5. Seed Longevity**

Monitoring the viability of seed of accessions in large collections regularly is costly in terms of seeds, labour and other resources, and it is important, therefore, to prioritise monitoring those accessions with brief longevity [34]. Assessing the seed storage longevity of wild species in genebanks is complex and difficult to predict as it is influenced by genotype, production environment, post-harvest handling and processing and storage conditions [7], and until recently, there was limited information on the longevity of many species [6]. Lack of data for evidence-based longevity predictions for forages led to species being classed as good or poor storers [35]. However, recent analysis of over 30 years of germination monitoring data from seeds stored in the medium-term store (seeds stored at circa 8 ◦C with 5% moisture content) in the genebank at ILRI allowed the calculation of seed longevity under these conditions for a range of forage genera and identified many, but not all, forage grass and legume species as having long-lived orthodox seeds [11,12]. Variation was observed in seed longevity among genera and between species of the same genus and indicated that some forage seeds have minimal dormancy and a longevity comparable to seeds of the major food crops, whilst other species showed high levels of hardseededness or dormancy that required dormancy-breaking procedures to be used in the germination test protocols.

Advice on breaking seed dormancy and promoting germination for each of 58 plant families is already published [21]. Typical successful dormancy-breaking procedures suggested therein for grasses include a germination test environment of a constant 16–21 ◦C or alternating temperature of 23/9 ◦C (12 h/12 h) for temperate accessions, or a constant 21–26 ◦C or alternating temperature of 33/19 ◦C (12 h/12 h) for tropical accessions, and/or 10−<sup>3</sup> M potassium nitrate, removal of seed-covering structures, and pre-chilling at 2–6 ◦C for up to 8 weeks. One caution is that although (white) light can promote seed germination, the germination of some grasses can be inhibited by high light intensity: for example, in the temperate grasses *Bromus mollis* and *Bromus sterilis*[36] and the tropical grasses *Echinochloa turnerana*, *Panicum maximum* and *Brachiaria humidicola* [37]. This can be avoided by limiting the period of exposure to light each day, as well as the dose. Typical successful dormancy-breaking/germination promoting procedures suggested for legumes [21] are seed scarification and a germination test environment of a constant 11–16 ◦C or alternating temperature of 23/9 ◦C (12 h/12 h) for temperate accessions, or a constant 21–26 ◦C or alternating temperature of 33/19 ◦C (12 h/12 h) for tropical accessions. One caution is that very dry legume seeds can be sensitive to a rapid uptake of water and so may benefit from initial hydration in a moist atmosphere at 100% relative humidity. A further source of readily-available information on characteristics of seed of diverse species relevant to genebanks is the Seed Information Database (Royal Botanic Gardens Kew, http://data.kew.org/sid/) [38]. This online compilation of various types of seed information from a wide range of sources, including for seed storage behavior [39], is provided to support seed genebank operations globally by the Millennium Seed Bank Project.

Monitoring seed viability in a large collection of many individual accessions using germination tests is labour intensive, time-consuming, and depletes the amount of seeds available for each accession. Hence monitoring should be kept to a minimum, but it is nonetheless essential to determine when germination declines and thus regeneration becomes necessary to avoid the loss of that accession from the genebank. The rational determination of accession monitoring interval is important for efficient, low-cost, but effective genebank management. Analysis of historic genebank data has been applied to provide evidence-based estimates of the different monitoring intervals necessary for diverse forage species [11,12] and indicated that 15-year monitoring intervals are suitable for many long-lived forage seeds with high viability at the time of entering storage. Even in medium-term seed stores, high-quality seed of some forages need only infrequent monitoring (Table 4).

There is a virtuous circle amongst high initial accession viability, excellent storage environment, monitoring period, and regeneration period for the genetic integrity of accessions. The first two support infrequent monitoring and yet more infrequent regeneration, whilst loss in viability during storage, accession depletion through frequent monitoring, and frequent regeneration each put the genetic integrity of accessions at risk [24,40]. Moreover, accession monitoring and regeneration are expensive undertakings. It is essential to retain much larger samples for regeneration than for distribution to users if minor alleles at a low frequency in a genetically heterogeneous accession are to be conserved through cycles of storage and regeneration (infrequent if seed viability is maintained long term). One analysis suggests that, if feasible, large samples of 500 seeds be grown out to regenerate the most original seed source to maintain allelic richness within an accession containing many alleles at low frequency [24].



#### **6. Cost E**ffi**ciency of Managing Wild Species**

The biological differences between crop species on the one hand and forage and wild species on the other are reflected in the costs of management and conservation in genebanks. There is limited data available on the actual costs of conservation other than from the genebanks of the centres of the Consultative Group for International Agricultural Research (CGIAR) [41,42]. Using these data as an example, the cost of conservation and management per sample in CIAT in 2000 was more than double for forages than for beans [41]. Costs also differ between locations even for the conservation of the same species. Studies in 2006 and 2009 calculated the costs of acquisition, characterisation, safety duplication, medium and long term storage, germination and seed health monitoring, regeneration, seed processing, information management, distribution and general management and concluded that the cost of forage germplasm conserved in CIAT was \$315 per accession, whilst in ILRI these estimates ranged from \$125–242 per accession, depending on the type of forage [42]. Such cost comparisons must account for the costs of operating in specific locations worldwide, as well as those incurred by different processes and procedures, and each Centre's prioritisation of species and activities. The reproductive biology of the species, which determines the type of conservation and regeneration procedures, and the quality of management attained in the genebank, contributed most to the variation in the costs of genebank operations [42]. Crops with orthodox seeds had the lowest costs of conservation, whilst vegetatively propagated crops, trees and wild relatives had higher costs.

The size of the collection also contributes to the costs of management. There is a marginal cost to adding an additional accession into a seed genebank with sufficient space to accommodate the seeds, but the benefit to cost ratio is high. The basic questions that genebank managers ask are what is the value of adding additional accessions, which germplasm should genebanks collect and store, and when is the collection large enough and contains the genetic traits demanded by users. The probability of finding accessions of value for users depends on the gene frequency of the trait that is being sought, how many accessions are tested and whether accessions are selected at random or selected based on prior knowledge on where to look for specific genes. Economic principles can be applied to modelling the value of accessions [43]. Sampling more accessions provides a higher probability of finding genes in low frequencies within the accessions but comes at a cost. For rare genes with a probability of 0.01 of presence in accessions, sampling 200 seeds gives an 87% chance of the gene being present in the sample. There is a marginal or incremental probability of finding the gene in the population, and the probability reduces as more accessions are added because if the gene is rare, the marginal value of adding an accession is low because it is difficult to find the gene. If the gene is common, the marginal value of adding accessions is also low because it has probably already been located. The marginal value of adding more accessions is high for intermediate ranges of gene scarcity [43,44]. Economically this supports keeping collections of wild species and forages relatively small, in terms of the number of accessions within a single species, because of their lower probability for use and high costs of conservation resulting from their reproductive biology and conservation challenges.

Cost–benefit analysis could be used to determine if the benefits from using the germplasm in forage development that has been, or could, in future, be realised, justify the conservation costs. However, this would require future germplasm utilisation to be quantified, which is very difficult to do. The value of the germplasm within genebanks lies in the future—when it will be used to transfer useful traits to forage cultivars. The option value of having the germplasm available to respond to as yet unidentified future needs, such as changing feed needs due to climate change or intensifying livestock production systems, and the existence value that society derives from knowing that something exists and will be available for future needs may be more powerful economic drivers of conservation of forage diversity than the cost–benefit analyses that can be done on its actual use [45]. In reality, most genebanks have fixed budgets, and so the management priority is to apply those resources to the best effect.

One example is the relative cost and risk of collecting forages and conserving ex situ in seed or field genebanks or maintaining them in situ in conservation areas within the extensive grasslands where they originated and continue to evolve and adapt. In situ conservation is a good alternative for forage species that are well protected in the wild, provided there is no evidence of a future threat or of historic genetic erosion to these forage genotypes. A good example of this is in sub-Saharan Africa, where many of the indigenous forages are found in well-protected rangelands in national parks [46]. On the other hand, ex situ conservation of forage germplasm is an essential precaution where over-grazing and other poor land management practices occur; it also makes the germplasm more accessible to users.

#### **7. Policy Issues**

The legal framework for crop collections is the multilateral system of the International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGRFA) [47]. This covers the major crops where countries have high interdependence on germplasm for food security, allowing them to be freely exchanged under a standard material transfer agreement (SMTA) to be used for food and agriculture with benefit-sharing provisions based on their commercial use. Germplasm flows through facilitated access under the ITPGRFA from the genebanks of the CGIAR Centres are well documented [48,49] with 4 million samples of germplasm made available under the SMTA from 2007–2016 to national partners and breeders, representing 93% of reported global distributions under the multilateral system [49].

The list of crops covered by the ITPGRFA includes some temperate forages and crop wild relatives that are also used as forages, but not many of the common and important tropical forage species in Table 1 of this review. Expanding collections with new germplasm will need to be done under bilateral agreements on access and benefit-sharing under the terms of the Nagoya Protocol. Reaching bilateral agreements on access and benefit-sharing under the Nagoya Protocol will be complicated and, in some cases, may not be possible in the short term as countries reform their national laws on access [50].

Whilst the economic cost–benefit analysis of conserving large ex situ collections of forages combined with the option to conserve forages in secure conservation areas in national parks indicates that there are other more cost-effective options for the conservation of forage diversity in many cases, the forage germplasm currently held in ex situ collections may well be the more accessible source for research and forage development because it avoids the need to collect and to negotiate conditions of access and benefit-sharing to new germplasm. The latter may take time, and so it is important that the germplasm already collected is properly managed and safeguarded in ex situ genebanks and continues to be made available to users.

#### **8. Conclusions**

Although collections in genebanks are vulnerable to loss of diversity [51] and the conservation of forage in ex situ collections has many challenges as outlined in this review, seed storage remains the most cost-effective and efficient method for their conservation and sustainable use for the immediate future. Ex situ conservation of forage germplasm must be linked to use in forage development to realise the potential benefits from the high costs of conservation. Unlike crops, where genebank accessions are used in breeding programmes to combine with other genotypes, many forage accessions are selected for direct use based on their adaptation, productivity and nutritional quality and released as cultivars due to the limited number of forage breeding programmes [4] and time and cost required to screen large numbers of accessions for useful traits [5]. To use wild species as forages, the challenges faced in seed dormancy, germination and harvesting for seed production of these species will need to be addressed methodically and research carried out to improve ease of cultivation and support adoption.

There are fundamental questions on which germplasm genebanks should collect and store, optimum size of the collection and how much diversity is sufficient to meet user needs remain open for forages. The answer depends on what types of forages are required for future sustainable livestock systems. Climate and environmental issues are shaping the perception of consuming livestock products in western diets, and people are being urged to reduce meat consumption. Yet in the developing world, many children survive on diets deficient in proteins and animal products and require milk and meat in their diets for cognitive development [52]. Livestock production systems are intensifying in the tropics with an increase in crossbred dairy cows with higher feed requirements. Natural pastureland is reducing in area as grassland and forest are converted for cropping, whilst climates are tending to dry and warm, pushing forage production to marginal land or planted as part of intensified crop-livestock production systems. The intensification of dairy systems through the use of higher quality feed in East Africa is predicted to increase milk yields without increasing greenhouse gas emissions, addressing both productivity and environmental issues [53]. This intensification is leading to the production of more crop residues and by-products for feed and, as a result, the need for higher quality protein-rich forages to supplement these lower quality residues. Additionally, in an effort to reduce greenhouse gas emissions, forage producers could reduce the use of inorganic nitrogen fertilisers replacing them with forage legumes, capable of symbiotic nitrogen fixation, and use more perennial forage shrubs and trees to reduce carbon loss from soils from cultivation. Hence, it is likely that the forages of the future will be developed from accessions of forage legumes and fodder tree species from ex situ collections to meet these needs. Given the anticipated continued demand for forage germplasm, challenges in their management will be reduced through research to support their conservation and availability for forage development.

**Author Contributions:** Both authors contributed to this paper's conceptualization; writing—original draft preparation, J.H.; writing—review and editing, J.H. and R.H.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Challenges for Ex Situ Conservation of Wild Bananas: Seeds Collected in Papua New Guinea Have Variable Levels of Desiccation Tolerance**

**Simon Kallow 1,2,3,\*, Kevin Longin 2, Natalia Fanega Sleziak 4, Steven B. Janssens 3,5, Filip Vandelook 3, John Dickie 1, Rony Swennen 2,4,6, Janet Paofa 7, Sebastien Carpentier 2,4 and Bart Panis 2,4**


Received: 30 July 2020; Accepted: 17 September 2020; Published: 21 September 2020

**Abstract:** Ex situ seed conservation of banana crop wild relatives (*Musa* spp. L.), is constrained by critical knowledge gaps in their storage and germination behaviour. Additionally, challenges in collecting seeds from wild populations impact the quality of seed collections. It is, therefore, crucial to evaluate the viability of seeds from such collecting missions in order to improve the value of future seed collections. We evaluate the seed viability of 37 accessions of seven *Musa* species, collected from wild populations in Papua New Guinea, during two collecting missions. Seeds from one mission had already been stored in conventional storage (dried for four months at 15% relative humidity, 20 ◦C and stored for two months at 15% relative humdity, −20 ◦C), so a post-storage test was carried out. Seeds from the second mission were assessed freshly extracted and following desiccation. We used embryo rescue techniques to overcome the barrier of germinating in vivo *Musa* seeds. Seeds from the first mission had low viability (19 ± 27% mean and standard deviation) after storage for two months at 15% relative humidity and −20 ◦C. *Musa balbisiana* Colla seeds had significantly higher post-storage germination than other species (*p* < 0.01). Desiccation reduced germination of the seeds from the second collecting mission, from 84 ± 22% (at 16.7 ± 2.4% moisture content) to 36 ± 30% (at 2.4 ± 0.8% moisture content). There was considerable variation between and (to a lesser extent) within accessions, a proportion of individual seeds of all but one species (*Musa ingens* N.W.Simmonds) survived desiccation and sub-zero temperature storage. We identified that seeds from the basal end of the infructescence were less likely to be viable after storage (*p* < 0.001); and made morphological observations that identify seeds and infructescences with higher viability in relation to their developmental maturity. We highlight the need for research into seed eco-physiology of crop wild relatives in order to improve future collecting missions.

**Keywords:** banana; crop wild relatives; ex situ conservation; desiccation tolerance; *Musa*; Papua New Guinea; plant genetic resources; seed conservation; seed storage behaviour

#### **1. Introduction**

Crop wild relatives (CWRs) possess genetic material useful for improving crops in an increasingly challenging context [1–3]. They comprise a large untapped genepool of alleles potentially useful for breeding [4]. Examples from banana CWRs include improved drought tolerance [5] and resistance to several diseases [6,7]. At the same time, many CWRs are threatened with extinction [8,9] making their conservation imperative for both biodiversity and food security [10,11]. Effective plant conservation employs complementary in situ and ex situ strategies [12,13]. Such an approach is notably encouraged for CWR conservation [14,15]. Accordingly, ambitious efforts to collect and conserve many CWRs ex situ have recently been made [16–20]. However, CWRs, and banana CWRs (*Musa* spp. L.) in particular, are poorly represented in ex situ collections [21].

Banana CWRs are diploid wild species whose fruits contain many dark coloured seeds. Edible bananas, selected to avoid seeds in the fruit pulp, can be diploid, triploid or tetraploid. There are around 80 species in the genus *Musa* [22,23], and over 1000 edible banana cultivars [24,25]. The management of banana germplasm is co-ordinated in a global network of 31 collections containing over 6600 accessions of in vitro or field plants [26]. Only 163 of these accessions are CWRs. Moreover, of these, 122 are of the two most important banana CWRs (*Musa acuminata* Colla and *M. balbisiana* Colla), the other 41 accessions include 33 *Musa* species. Additionally, there are 131 *Musa* seed collections of only seven species stored in seed banks [27]. This means many species are only represented by a single genotype and for many wild banana species, no accessions exist. The diversity of banana CWRs in ex situ conservation is therefore highly constrained and expansion of the inter and intra-specific diversity of the collection is clearly needed.

*Musa* are pioneers or early successional tall herbs of tropical to subtropical rainforests. Native distribution area ranges from Southeast Asia to Pacific regions [22]. Papua New Guinea (PNG), the world's most floristically diverse island [28], is an important centre for both wild banana and cultivar diversity [29–32]. Sixteen wild *Musa* taxa occur in PNG [33]. Several collecting missions have been made in PNG to characterise and collect both cultivar and CWR germplasm [34–38]. These include seed collections, two of which we evaluate here.

Ex situ conservation using seeds can be a highly effective way of conserving the genetic diversity of plant populations [39,40]. This is useful for further conservation activities, phenotyping and breeding. Furthermore, conservation using seed is a relatively cost effective method of ex situ conservation [41]. In order to make high quality seed collections of wild species, understanding of seed development and storage behaviour are crucial [42]. Seed storage behaviour can classically be categorised into three groups. The majority of seeds are easily dried (to 2–5% MC) and stored at sub-zero temperatures, these are *orthodox* seeds [43]. Secondly, *recalcitrant* seeds do not survive drying to below 20–30% moisture content and are sensitive to low temperatures [43]. Finally, seeds that do not fit well into these binary categories are often called *intermediate,* and show partial sensitivity to drying and cold storage in particular [44]. Seeds of recalcitrant and intermediate species should be stored cryogenically, whereas orthodox seeds may be stored conventionally (at −20 ◦C) following desiccation [45].

For wild species, and especially banana CWRs, critical knowledge gaps exist in how best to collect, store and germinate their seeds. For *Musa*, only six species have been assessed for their storage behaviour, results of which are inconclusive [46–52]. Additionally, germination of seeds is notoriously inconsistent and dormancy poorly understood [53–55]. Embryo rescue techniques are therefore commonly used to germinate seeds in breeding programmes [56]. Together, these critical knowledge gaps hamper storage and access to banana genetic material [54].

Substantial challenges associated with collecting seeds from wild species impact the quality of seed collections [57]. Non-uniform seed development across a population, low seed numbers and sub-optimal post-harvest handling may be problematic [57–60]. Post-harvest handling is difficult because it is often not possible to control the temperature and humidity of seeds on collecting missions, e.g., whilst in a vehicle or when moving from place to place. Furthermore, there are significant practical challenges in collecting seed material from populations of wild species, the location of which may be remote, inaccessible and previously unknown. Evaluation of material from actual collecting missions can provide useful concrete evidence of these particular challenges, and lessons can be learnt to improve the quality of collections in the future.

In this study, we make use of seeds from two recent collecting missions to PNG. Seeds from one mission were already stored in Meise Botanic Gardens seed bank (called 'batch 1', and described by Eyland et al. [38]); the others were collected during the course of this investigation (called 'batch 2'). By evaluating seed viability of these collections, we address some of the issues and knowledge gaps described, by answering the following questions: (1) What is the viability of *Musa* seeds stored in Meise Botanic Gardens seed bank (for two months at 15% relative humidity (RH), −20 ◦C)? (2) Do seeds of some *Musa* species have higher viability after storage than others? (3) Do seeds from different parts of the infructescence have higher viability after storage than others? (4) How does desiccation affect seed viability? (5) Is it possible to predict storage behaviour of *Musa* seeds based on their physical properties? (6) Does seed maturity affect viability during dry storage? We use in vitro embryo rescue techniques to quantify viability. This provides the most reliable estimate of viability and removes dormancy constrains that limit germination in *Musa* seeds [61–63].

#### **2. Results**

#### *2.1. Viability Evaluation of Seeds Stored in the Seed Bank*

#### 2.1.1. Overall Viability

The post-storage viability of batch 1 seeds (already stored in Meise Botanic Gardens seed bank) was markedly low with considerable variance between the accessions (Figure 1A). Across all accessions and hands, germination was on average 19 ± 27% (mean and standard deviation used hereafter; empty seeds are excluded in the percentages). In the present study, we use the term *accession* to mean a seed collection from a single individual plant including all the fruits of an infructescence. The term *bunch* refers to an infructescence. A bunch can be subdivided into *hands,* these are groups of fruits from the former clusters of flowers subtended by one bract [64]. Embryos that showed no reaction were 73 ± 29%. Other embryonic reactions, callus formation and embryo darkening without further outgrowth, were minimal (respectively, 0.2 ± 1.2%, and 3 ± 9%). Microbial contamination of the sample was 4 ± 16%. Overall, 24 ± 23% of seeds contained no identifiable embryos.

**Figure 1.** (**A**) Germination responses of embryos rescued from 29 accessions of *Musa* species following drying for 4 months at 15% relative humidity 20 ◦C and storage for 2 months at 15% relative humidity −20 ◦C. 'Hand position' refers to the position in the infructescence of the hand from which seeds were collected, with '1' being closest to the basal end of the bunch (*n* = 23 ± 10 seeds). (**B**) Predicted probability of five embryo rescue outcomes of *Musa acuminata* subsp. *banksii* seeds extracted from different hand positions in the infructescence. Probabilities based on the multinomial logistic regression model of the response of seeds from 50 hands (representing 13 accessions; *n* = 30 seeds for each hand). Shaded areas are 95% standard errors of the estimated regression coefficients.

#### 2.1.2. Effect of Species

*Musa balbisiana* Colla seeds (accessions #2 and #3), showed significantly higher germination than other species after storage (*p* < 0.01), in parallel with less embryos showing no reaction. This is demonstrated by the multinomial logistic regression (MLR) model (Table S1A), whereby the log odds of germination against an increase in no reaction is 0.865, but for all other species, log odds are negative. This is despite one of the three *M. balbisiana* accessions (accession #1) having no viability. One *M. schizocarpa* accession also showed high viability (accession #29, 90%), in contrast to the other three (accessions #26–28).

#### 2.1.3. Effect of Position in the Infructescence

*Musa acuminata* subsp. *banksii* seeds from hands with a higher number, i.e., that were more recently pollinated, had embryos that were significantly more likely to germinate after storage (*p* < 0.001, Figure 1B, Table S1B); again this was concurrent with a reduced likelihood of no reaction.

#### *2.2. E*ff*ect of Desiccation*

#### 2.2.1. Overall Effect of Desiccation

Seeds of batch 2 were tested before and after desiccation. Before desiccation (16.7 ± 2.4% moisture content (MC)), germination was on average 84 ± 22% (Figure 2A). After seven days desiccation (to 2.4 ± 0.8% MC), germination decreased to 36 ± 30%. As embryos dried, their percentage without any sign of germination increased from 3 ± 7% to 55 ± 27%. Again, there was considerable variance between accessions. Accession #37, *M. balbisiana*, was excluded from the analysis as the initial MC was an outlier compared to all the others (35% MC).

**Figure 2.** (**A**) Embryo rescue outcomes of *Musa* seeds (batch 2) before desiccation at 16.7 ± 2.4% moisture content ('Wet'), and after desiccation for seven days in a desiccator to 2.4 ± 0.8% moisture content ('Dry'). Accession and hand numbers are included above each chart. Seeds were germinated using embryo rescue and results recorded 28 days after transfer to the growth medium (*n* = 10). (**B**) Predicted probability of embryo rescue results according to the moisture status of seeds. Plot is on predicted values of the multinomial logistic regression model coefficients in Table S1C, data in Figure 2A. 95% standard errors shown.

#### 2.2.2. Desiccation Tolerance and Species

Viability after desiccation differed by species. The *M. acuminata* subsp. *banksii* and *M. schizocarpa* accessions showed the highest germination after desiccation (55 ± 7%, 54 ± 29%, respectively). *M. maclayi* accessions had the lowest viability after desiccation (3 ± 6%). The MLR model based on drying as a factor showed a significant effect on germination in relation to no reaction (*p* < 0.001, Figure 2B, Table S1C). Additionally, seeds in wet condition were also more likely to darken and be contaminated (*p* < 0.001, *p* < 0.01, respectively). In the model, there is clear interchange of embryos that germinate with those that show no reaction (Figure 2B).

#### *2.3. Prediction of Seed Storage Behaviour*

Predicted seed storage behaviour (using the method of Hong and Ellis [65] and Ellis et al. [66]), identified that accessions straddled both intermediate and orthodox categories (Figure 3). Nine of the 11 accessions were predicted to be intermediate and two orthodox. The accessions predicted to be intermediate exceeded the threshold for weight rather than moisture content (apart from accession #37, *M. balbisiana,* previously identified as an outlier with high MC). There was no correlation between predicted seed storage behaviour and post-desiccation germination. Only seeds from batch 2 were used, as moisture content measurement of seeds before desiccation is required by the model.

**Figure 3.** Predicted storage of behaviour of *Musa* accessions (batch 2) using the diagnostic key of Hong and Ellis [65] and Ellis et al. [66]. Area A includes accessions predicted to have intermediate storage behaviour, accessions in area B are predicted to have orthodox storage behaviour. Accessions are coloured according to the germination percentage of seeds after seven days desiccation to 2.4 ± 0.8% moisture content. Moisture content is calculated on the fresh weight basis ('fwb'). Seeds were germinated using embryo rescues and assessed 28 days after transfer to growth medium.

#### *2.4. Dry storage and Maturity*

#### 2.4.1. Effect on Viability

Two *M. acuminata* subsp. *banksii* accessions were selected from batch 1. Accession #4 was from a less mature and accession #11 from a more mature bunch according to observations in the field. Seeds were tested before and during dry storage (contained within a paper bag, stored in a humidity controlled room at 15%RH, 20 ◦C). Seeds had a mean moisture content of 11.2 ± 2.7% before drying (Figure 4A). After seven days drying, moisture content reduced to 6.6 ± 2.5%. Moisture content remained about the same with further drying time, so that after six months moisture content was 6.5 ± 2.35%. Differences in moisture content between the accessions were not statistically significant.

**Figure 4.** (**A**) Moisture content during dry storage, calculated on fresh weight basis ('fwb'). Seeds from a mature and an immature accession of *Musa acuminata* subsp. *banksii* were used. Seeds were dry stored at 15%RH, 20 ◦C for up to 168 days. (**B**) The effect of dry storage on embryo rescue outcomes. Outcomes recorded 28 days after the transfer of each embryo to growth medium (*n* = 48).

The mature seeds (accession #11) had much higher initial germination rates than the immature (accession #4), 91.3% and 16.3%, respectively (Figure 4B). Embryos from both accessions reduced in germination after seven days drying, to 28.3% for the mature accession and to 9.1% for the immature accession. For the mature seeds, this level of germination remained about the same with further drying time, so that after 6 months of drying, germination was 29.2%. Embryos from the immature accession continued to reduce in germination with further drying time, so that after 6 months dry storage, germination was negligible (2.4%). The proportion of embryos that germinated was notably reduced and exchanged to a correspondingly larger proportion of embryos that displayed no germination reaction, and to a lesser extent, darkening. Contamination also increased with time of drying for the immature seeds.

#### 2.4.2. Effect on Morphology

Observations from magnified images of the selected *M. acuminata* subsp. *banksii* accessions showed apparent under-developed seed coats in the less mature seeds. The different layers of the seed-coat integuments are evident rather than fused, they are also lighter in colour (Figure 5). During drying, these layers were observed to separate. Additionally, there is a noticeably greater effect of desiccation on the structure of the endosperm and shape of the embryo. Less mature seeds display increased airspaces in the endosperm on desiccation. Embryos of the less mature seeds show greater loss of structure during desiccation.

**Figure 5.** Photographs of immature and mature bunches and their seeds of two *Musa acuminata* subsp. *banksii* accessions. Seed images taken before and after 1 month drying in a dry room (15% relative humidity, 20 ◦C). Seed images taken on a Keyence VHX5000 at 150 x magnification.

#### **3. Discussion**

#### *3.1. Key Findings*

This assessment of seed storability of banana CWR seeds from PNG collecting missions illustrates some of the challenges involved in making high quality collections of wild species for ex situ conservation. In particular, this assessment demonstrates some of the difficulties involved in making seed collections of wild species and how critical knowledge gaps impact the value of such collections. Our evaluation shows substantial loss of seed viability during seed banking which can be attributed to variable levels of desiccation tolerance. There was considerable variation between accessions, and some

species (*M. balbisiana* and to a lesser extent *M. schizocarpa* and *M. acuminata* subsp. *banksii*) maintained higher viability during storage compared to others.

#### *3.2. Desiccation Sensitivity*

The low germination rates (19%) of seeds that were stored in the seed bank (batch 1), suggests a problem with maintaining the viability of collections in conventional storage (15%RH, −20 ◦C). However, as this was a viability assessment of seeds already stored, it is not possible to draw specific conclusions as to why viability is low: seeds may have had low initial viability or lost viability during transport, for example. By testing batch 2 seeds both before and after desiccation, it is clear that desiccation sensitivity is a major contributor to loss in viability. On average, these seeds reduced germination from 84% to 36% during rapid desiccation (from 17% to 2.4% moisture content). We therefore surmise that loss of viability is primarily a result of sensitivity to rapid desiccation. Further research is needed to fully understand whether loss in viability is caused by desiccation per se, or whether speed of desiccation is an important contributing factor.

#### *3.3. Seed Storage Behaviour*

Variation in our results, with respect to seed storage classification, is in line with previous studies. For instance, several studies demonstrate desiccation sensitivity where seeds lose viability at 6% MC [46] or, for extracted embryos, to 10–15%MC [50,51,67]. Other studies found that seeds tolerate drying, but do not state to what moisture content [47,48].

High viability of a few accessions stored under very low moisture and sub-zero temperature in the present study, suggests that (at least for *M. balbisiana*) orthodox storage class is likely. For others, our results suggest that intermediate storage classification may be appropriate, as significant proportions of seeds lost viability on desiccation and freezing. *Musa* storage behaviour is, therefore, at the threshold between orthodox and intermediate storage classes, as illustrated by results of the predictive model (Figure 3). It should be noted, however, that for all species (apart from *M. ingens*), a proportion of seeds survived desiccation, or even desiccation and sub-zero storage. Storage class was, therefore, variable within an accession, and orthodox behaviour of at least a small proportion of seeds was possible, if rare. Whilst storage classification is helpful, a continuum of storage behaviour is known to exist [68], even within the same genus or species, depending on when and where seeds were collected [45,69–72]. In the present study, a continuation in desiccation tolerance is evident within the same accession and even from fruits in the same hand.

#### *3.4. Variation between Infructescences*

#### 3.4.1. Species and Climate

Differences in post-storage viability were greater between-infructescences (from different maternal plants), than within-infructescences (from hands of the same plant, Figure 1A). This may be related to differences in seed storage behaviour at the species level, to the maturity level of the whole infructescence or perhaps the microclimate [73]. Viability levels were consequently strongly linked to the fruit-bearing plant.

In batch 1, *M. balbisiana* seeds showed significantly higher post-storage viability than the other species. This species is characterized by a wide, yet often introduced, distribution across the tropics and subtropics. Notably, *M. balbisiana* is not considered a native species to PNG [74,75], but rather has its native distribution in the more seasonal subtropical Northern Indo-Burmese region [76]. By contrast, the other wild banana species studied are native to the Equatorial wet to moist ecoregions of PNG. As such, *M. balbisiana*, has also shown to have high leaf wax content that contributes to drought tolerance [5] and is therefore probably better adapted to seasonal changes in precipitation and temperature than the *Musa* species native to PNG [33,74,75]. This then suggests that within the whole genus, there may well be a range of desiccation tolerance levels possibly according to species

distributions. It should, however, be caveated that our observation is based on only a small number of samples and a wider survey should be carried out for further conclusions. Nonetheless, it is well known that there is a correlation between the bioclimatic distribution of a plant and seed storage behaviour: higher annual precipitation is positively correlated with recalcitrance [77]. Interestingly, differences in precipitation in the native region of the *Musa* species examined here are in fact greater when only the precipitation in the driest quarter of the year is considered, rather than for annual precipitation (see Figure S1). We therefore suggest that the precipitation in the driest quarter might possibly have a stronger correlation to seed storage behaviour for *Musa* than annual precipitation, as the possible impact of a dry season may be masked in seemingly high precipitation regions. We therefore propose that *Musa* seeds collected from species adapted to more pronounced dry seasons may have better desiccation tolerance and therefore better survive storage. However, further research is required in this area.

#### 3.4.2. Seed Maturity

We identified physical properties that were seemingly linked to the level of seed maturity at the time of harvest. Larger fruits with softer pulp texture and seeds with a more powdery endosperm were considered to be more mature. Seeds from the bunch categorised as more mature in the field had greater embryo rescue germination percentages, both before and after dry storage (15%RH, 20 ◦C), compared to the seeds identified from the less mature bunch. In the laboratory, it was observed that the less mature seeds had higher initial moisture content that reduced to a greater extent, and an under-developed seed coat: a light brown inner integument as opposed to dark brown to black, that was less well fused with the outer integument (Figure 5). The small sample size notwithstanding, the importance of seed maturity for desiccation tolerance is consistent with current understanding of the development of desiccation tolerance during late seed maturation [73,78,79]. Desiccation tolerance is acquired at 'mass maturity' after maximum dry weight is achieved and the vascular connection between the maternal plant and seed is terminated [80]. Following this, seed moisture content equilibrates with the environment prior to dispersal. Often this is described as the 'point of natural dispersal' [60]. The difficulty for improving the quality of seed collections is how to translate theory into practice, particularly for seeds that are contained within large pulpy fruits like bananas.

Regarding seeds that were collected from field collections, Simmonds [47] found that, for maximum in vivo germination, *M. balbisiana* seeds should be collected 'mature'. Unfortunately, he did not define what 'mature' meant in this instance. However, he detected a window of six weeks whereby high germination can be achieved (>80%), four weeks before and two weeks after maturity. Additionally, he found that fruit of *M. acuminata* should be collected green or yellow (rather than black or rotten) to achieve high germination. Furthermore, Uma et al. [81] found that at 70% maturity (full maturity being 110 days after (self-)pollination) 'Pisang Jajee' (a *M. acuminata* genotype) embryos were discernible and endosperms had converted from a liquid to semi-solid state; this also coincided with thickening of the integuments. They also found that seeds, in order to germinate, should be at least 90% mature, and immature embryos were more likely to produce calluses.

Collecting mature seeds during collecting missions is much more challenging than from field germplasm collections. Collectors must access bunches before they are consumed and seeds are dispersed by birds and mammals [82–84]. Humans also harvest wild bananas for food, construction and artistry [37,85]. It is therefore important to be able to identify fruits that contain seeds that are mature enough to be desiccation tolerant, without knowing flowering times, whilst they may not have yet attracted frugivores. Based on our results, we suggest that seeds should have powdery endosperms and well-formed integuments with fused layers, without which many seeds will be lost during storage; however, clearer definitions should be developed for collectors.

It may also be possible to improve desiccation tolerance and longevity of seeds by using a treatment that mimics late maturation on the plant, as has been shown for other species [78,86–88]. Indeed, in one study [89], Simmonds found that seeds from 'ripe' and 'over-ripe' bananas that were dried in the fruit at a temperature similar to what may be found on the plant (in an oven at 45 ◦C), germinated better than seeds that were not dried in this manner. Assessing and furthering seed maturity whilst avoiding dispersal is clearly a key factor in improving the quality of future banana seed collecting.

#### *3.5. Variation within Infructescences*

Heterogeneity of maturity within an infructescence has been highlighted as a cause for variable desiccation tolerance within seed accessions of other wild species [57,60]. We observed a small but significant within-infructescence effect, in that seeds from the male bud end of the infructescence (seeds from flowers that were more recently pollenated) were around 15% more likely to germinate (post-storage) than the peduncle basal end (Figure 1B). This, perhaps surprising, effect may be caused be caused by variation in seed-vigour or seed-aging, discussed below.

It is well known for other species, that there are differences in physical and physiological properties in seed-vigour within the same infructescence [90]. For species of temperate regions variation is often correlated with seasonality [91–93]. In tropical species, the effect of climate on seed properties is not well known. However, pollen, seed set and germination success of banana seeds (during breeding programmes) have been found to correlate with climatic conditions [94,95]. Alternatively, seed-vigour, including the ability to tolerate stress, deteriorates according to temperature and moisture [71]. When seeds are kept in the fruit for relatively long periods of time, for example, during collecting missions, seed-aging can occur. This can potentially influence the ability of seeds to withstand the stress of desiccation later on. As seeds from basal hands are produced first, when they are harvested they are already in a more advanced state: fruit may soften quicker and have higher moisture content, and the exocarp may be rotting. This all means that aging is more likely to occur if they are then kept in the fruit during the remainder of the trip and until they are transported to the laboratory (see Figure S2 for a photograph of the fruits of batch 1 after transit to Belgium). This could explain why the older seeds within a bunch may display lower desiccation tolerance. However, as this effect was relatively small compared to the overall maternal effect, it seems that the within-bunch maturity appears at least to have less of an effect than the maturity of the whole bunch.

#### *3.6. Limitations and Assumptions*

#### 3.6.1. Embryo Rescue

We used embryo rescue techniques to estimate viability in the present study. Whilst this is the best current method for estimating *Musa* seed viability, there are limitations and assumptions that should be stated. Firstly, the purpose of a viability measure is to estimate the proportion of seeds that are capable of developing into seedlings or plants [42]. Embryo rescue 'short-cuts' some of the constraints that could limit this process of in vivo germination. For instance, if an embryo germinates in vitro, it does not necessarily mean that it is capable of developing into a seedling or plant. For this to happen the embryo must also push off the seed micropyle cap and develop roots that can access the soil. We accounted for this in our analysis by categorising separately embryos that did not develop fully formed shoots, but rather formed calluses or showed no growth but darkening of the embryo. Secondly, embryo rescue evaluation, in our method, is at 28 days; however, it is possible that germination may be slow and only is evident after this period. According to the literature, 28 days should normally be enough time [61,62,96], but it is an assumption that, at this point in time, the germination process is concluded for all embryos. Finally, the conclusions of this study are based on the assumption that embryos showing no germination reaction are in fact dead. However, it is possible that desiccation does not kill the embryo, but rather causes a deep level of physiological dormancy that is not removed by excision from the rest of the seed. To account for this, we carried out tetrazolium tests on embryos that showed no germination reaction on embryo rescue. These embryos showed no staining. This indicates that embryo rescue produces the maximum measure of viability.

#### 3.6.2. Conservation and Research Material

Whilst the benefit of using seeds from collecting missions allows results to be impactful for future missions, limitations are also introduced by using such material. One of the main limitations we faced was the limited availability of seeds for research. This inevitably constrains the interpretation of results (hence the large amount of deviation) because sample sizes and replicates were small. Seed numbers were limited for two reasons. One, because it is difficult to access seed material in suitable time periods from third parties, despite relevant treaties [97]. Two, because there are conflicting demands for material. There is an expectation and requirement for seeds to be placed into storage 'for conservation'. This may conflict with availability of adequate material for research into how best to store and germinate seeds. Our results highlight the need for seed collecting for research purposes in addition to, and ideally prior to, collecting missions whose primary purpose is conservation. In practice, as here, these two processes often run concurrently.

#### **4. Materials and Methods**

#### *4.1. Study Region*

The study region was between Latitude 2◦ to 8◦ South, and Longitude 141◦ to 151◦ East. Seeds were collected in the Papua New Guinean provinces of Morobe, Madang and Sandaun on the island of New Guinea, and the province of West New Britain on the island of New Britain (Figure 6). These locations are in the tropical and subtropical moist broadleaf forest biomes [98]. Mean annual precipitation and mean annual temperature, at the collecting locations are 2695 ± 562 mm and 24.9 ± 1.9 ◦C, respectively (averages for years 1970–2000) [99].

**Figure 6.** Collecting locations of seeds used in this study (red circles) and relevant province names. Map is shaded according to elevation (meters above sea level, data from: http://srtm.csi.cgiar.org).

#### *4.2. Plant Material*

#### 4.2.1. Accessions

Overall, 37 *Musa* seed accessions were used in this study. Accessions were from a total of seven species: *Musa balbisiana* Colla*, M. acuminata* subsp. *banksii* (F. Muell.) N.W. Simmonds*, M. boman* Argent*, M. ingens* N.W. Simmonds*, M. lolodensis* Cheesman*, M. peekelii* Lauterb*, M. schizocarpa* N.W. Simmonds (Table 1).



#### 4.2.2. Seed Batches

Seeds were collected during two field missions to Papua New Guinea. Batch 1 was collected in May 2019, at the end of the wet season, and included 29 accessions, described by Eyland et al. [38]. Batch 2 was collected in October 2019, at the start of the wet season and contained eight accessions (Table 1).

#### *4.3. Seed Collection, Field Evaluation and Transportation*

Seeds were collected from wild populations that occurred either in primary or secondary forests. At the time of collecting, seed maturity was assessed by dissecting approximately 10 seeds per bunch and examining the embryos and endosperms. Seeds were considered mature when embryos were capitate in shape (mushroom-like) and endosperms were powdery as opposed to wet or milky. Only bunches with seemingly mature seeds were collected (although, some bunches proved to be not

completely mature, see results). Each bunch was photographed on site. Hands were removed from the bunch and numbered according to position, with 1 being at the basal end, i.e., they were produced first. Hands were placed in paper bags, which were then placed in cardboard boxes for storage during the remaining field mission. Accessions were then transported to Belgium for extraction. Transportation was initiated within one week of the end of the two-week collecting mission and took approximately one week to complete by aeroplane. During shipping, temperatures were greater than 0 ◦C and less than 25 ◦C. Fruits were therefore received within four weeks of collecting in the field.

#### *4.4. Seed Processing*

#### 4.4.1. Extraction

Seeds were extracted by peeling the epicarp and crumbling or squashing the endocarp and removing seeds by hand. Excess fruit pulp was removed by washing in running water if necessary. In case fruits were hard, they were soaked in water for 24 h prior to seed extraction. It took a week to extract all the seeds from batch 1, and one day for the seeds of batch 2. Seeds were maintained under ambient laboratory conditions (approximately 60–80% relative humidity, 20 ◦C) for a maximum of seven days whilst all extractions were completed, this also allowed removal of excess water gained during washing. Moisture content of a subset of three accessions of batch 1 seeds and all accessions of batch 2 seeds was measured after extractions were completed (see Section 4.4.2 for method).

#### 4.4.2. Moisture Content Measurement

Moisture content (MC) was calculated on a fresh weight basis (FWB) using the formula:

$$\text{MC}(\%) = \frac{(fresh\,\,weight - dry\,\,weight)}{fresh\,\,weight} \times 100\%$$

Seeds were weighed in plastic boats, dried at 70 ◦C for three days, and re-weighed. The MC of seeds was then calculated. Seeds were dried whole, as seeds coats were previously assessed as water permeable (our own data not shown and see [100]). Our own previous results also showed that embryo moisture content was 2% higher than whole seeds for non-desiccated seeds (at 10%MC) and 3%MC higher after desiccation (to 3%MC). Whole seeds were used here because accurately measuring the moisture content of embryos requires many samples that were not available because of their small size. Three replicates of 10 seeds were used to assess moisture content unless otherwise stated.

#### 4.4.3. Storage

For storage in the seed bank at Meise Botanic Gardens, Belgium, seeds were further dried for four months at 15%RH and 20 ◦C, and then placed in cold storage at 15%RH and −20 ◦C sealed in aluminium envelopes. Seeds were in cold storage for two months prior to viability evaluation. The moisture content of seeds was taken prior to transfer to cold storage.

#### *4.5. Viability Evaluation of Seeds Stored in the Seed Bank*

We used embryo rescue techniques to evaluate viability [61–63]. This is the most effective measure of *Musa* seed viability compared to whole seed germination [61,62,96] and the tetrazolium chloride test [101] (Simon Kallow, pers. obs.).

We evaluated the viability of batch 1 seed accessions that had been stored in the Meise Botanic Gardens cold storage for two months. No pre-storage viability evaluation had been made. The MC of a subset of three accessions was assessed on removal from storage.

For embryo rescue, seeds were sterilised by soaking them in 96% ethanol for 3 min, followed by 20 min in 1% NaOCl (diluted commercial bleach 5%), containing 1 drop of detergent per 100 mL. Seeds were then rinsed three times in sterile water. Embryos were extracted from seeds using a sterile

forceps and scalpel by making an incision in the seed coat next to the micropyle with the scalpel and by manipulating the seed with scalpel and forceps until the testa split open exposing the endosperm and embryo; embryos were then removed by careful manipulation. Embryos were transferred onto autoclaved half MS medium [102] in tubes with the haustorium in contact with medium and the embryonic axis upwards. All procedures were carried out in a laminar flow cabinet. Tubes containing embryos were incubated in the dark at 27 ◦C for 14 days, after which they were put in a growth chamber for an additional 14 days (24 h photo-period, 27 ◦C, 50 μE m−<sup>2</sup> s−<sup>1</sup> illumination provided by 36 W Osram cool-white fluorescent tubes). Six possible observations were recorded after 28 days: empty (no embryo, identified during excision), contamination, no reaction, callus formation, darkening and germination. *Musa* embryos are regarded as non-dormant when cultured in vitro [61–63,96,103], so seeds showing no reaction during this period were considered dead. Embryos that form calluses or that darken are considered alive but unlikely to regenerate into seedlings. An average of 23 ± 10 seeds from an average of 3 hands were tested for 29 accessions. Seed availability for this evaluation was highly constrained.

#### *4.6. E*ff*ect of Desiccation*

Following the results of the viability evaluation of stored seeds (Section 2.1), batch 2 was collected. We assessed the effect of desiccation on the eight accessions included, using embryo rescue (as previously described). This was done before desiccation and then after seven days of enforced desiccation. Seeds were desiccated by placing them on plastic boats suspended over silica gel sealed in a desiccator. The environment in the desiccator was approximately 2.4% RH and 20 ◦C. Ten seeds per accession were used both before and after extraction. Moisture content was measured for each accession before and after desiccation using ten seeds.

#### *4.7. Prediction of Seed Storage Behaviour*

Seed mass and initial moisture content was used to predict seed storage behaviour according to the model of Hong and Ellis [65] and Ellis et al. [66]. For this, seeds with a 1000 seed weight of less than 2500 g and a moisture content of less than 22% are predicted to have orthodox storage behaviour. Seeds with higher mass and moisture content (>4000 g and >40%MC) are predicted to be recalcitrant. Seeds in-between these limits are predicted to be intermediate. Only seeds in batch 2 were used for this prediction as fresh seed moisture content is a requirement. Seeds were weighed within seven days of extraction whilst maintained in ambient conditions (60–80%RH and 20 ◦C) to remove excess moisture gained during extraction. Five replicates of 50 seeds were weighed and the mean of this was used to calculate 1000 seed weight for each accession. Moisture content was measured as described above. Moisture content and mass for each accession was then plotted in a scatter chart.

#### *4.8. Survival during Dry Storage*

#### 4.8.1. Effect of Maturity

Two *M. acuminata* subsp. *banksii* accessions from batch 1 were selected from a seemingly more mature (accession #11) and a less mature bunch (accession #4). Maturity level was identified during collecting. The seemingly mature bunch had darker fruit colour and softer pulp texture, as well as more powdery endosperm compared to the less mature bunch (Figure 5). Subsample seeds were selected across all hands and mixed, so that the seeds used reflected the entire accession. For these selected accessions, embryo rescue was carried out after extraction as described above, and then after 1 week, 1, 3 and 6 months of dry storage in a paper bag stored in a dry room (15%RH, 20 ◦C). Forty eight seeds were used from each accession at each time point. The MC of seeds of each accession at each time point described was measured.

#### 4.8.2. Effect on Morphology

Seeds from the selected *M. acuminata* subsp. *banksii* accessions (#4 and #11) were dissected and photographed at the same time points and conditions described above. A digital microscope (Keyence VHX5000) was used at 150–200 x magnification. Ten seeds were used per accession, condition and time point.

#### *4.9. Statistical Analysis*

Counts from the categorised outcomes of the embryo rescue tests were transformed into lists where each embryo's reaction was a nominal outcome. These data were then used to build multinomial logistic regression (MLR) models to analyse the log-odds of the embryo rescue outcome category using the *nnet* R package [104]. Maximum models were reduced by comparing Akaike information criterion (AIC) and carrying out the likelihood ratio test. Effects plots were produced from the models by predicting and then plotting data using the effects R package [105]. Statistics were carried out in R v 3.6.2 [106].

#### **5. Conclusions**

The aim of this study was to assess the viability of banana seeds collected during two collecting missions in order to inform ex situ conservation of banana CWRs. (1) We found that in general *Musa* seeds collected in PNG and stored in the seed bank had low viability. (2) There was considerable variation between accessions, *Musa balbisiana* seeds had significantly higher post-storage viability than other species. (3) Variation within accessions, according to the position in the infructescence, was significant, with seeds of *M. acuminata* subsp. *banksii* from the basal end having lower viability after storage than from the male bud end. (4) Freshly extracted seeds lost much of their viability during desiccation. (5) Predictions of seed storage behaviour based on physical properties indicate that *Musa* seeds are at the threshold of orthodox and intermediate classification; this is in keeping with our embryo rescue results. (6) *M. acuminata* subsp. *banksii* seeds, identified in the field as more mature, had higher viability before and during dry storage than less mature seeds, but this level was also reduced after dry storage.

This assessment of seed viability demonstrates the importance of advancing understanding of the seed storage behaviour of CWRs. In particular, we show how the ecology and adaption of species and the development of their seeds in time effects the viability of seeds collected for storage.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/9/9/1243/s1, Figure S1: (A) Annual precipitation, and (B) precipitation of the driest quarter (three-month period), across the native distribution of the species evaluated in this study. Data extracted from WorldClim v2.0 based on occurrence records of species (data compiled by A. Mertens). Green dots displays means. Figure S2: Batch 1 fruits during processing after arrival at Meise Botanic Gardens, Belgium. Fruits are separated according to hand position on the bunch. Table S1: Multinomial logistic regression coefficients in log-odds (logits) and standard deviations in parentheses. (A) Embryo rescue outcome of batch 1 post storage. (B) Embryo rescue outcome of post-storage *Musa acuminata* subsp. *banksii* accessions in batch 1 according to hand position (1 being at the basal peduncle end of the infructescence). (C) Embryo rescue outcome of seeds from batch 2 after drying in a desiccator for seven days, a factor with two levels ('Wet' and 'Dry'). Embryo rescue outcome categorised after 28 days. Stars designate significance levels for *p* values \* ≤ 0.05, \*\* ≤ 0.01, \*\*\* ≤ 0.001.

**Author Contributions:** Conceptualization, S.K., K.L., S.B.J., S.C. and B.P.; methodology, S.K., K.L., S.B.J., F.V., J.P., and B.P.; software, S.K.; validation, K.L. and B.P.; formal analysis, S.K.; investigation, S.K., K.L., and N.F.S.; resources, F.V., J.P., S.C. and B.P.; data curation, S.K., K.L., N.F.S., S.J.B., F.V. and B.P.; writing—original draft preparation, S.K.; writing—review and editing, S.K., S.B.J., F.V., J.D., R.S., S.C. and B.P.; visualization, S.K.; supervision, S.J.B., J.D., R.S. and B.P.; project administration, S.K., S.B.J., S.C. and B.P.; funding acquisition, S.B.J., J.D., R.S., S.C. and B.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded as a sub-grant from the University of Queensland from the Bill & Melinda Gates Foundation project 'BBTV mitigation: Community management in Nigeria, and screening wild banana progenitors for resistance' [OPP1130226]. In addition, this study received funding from the Genebank CGIAR Research Program, from Research Foundation-Flanders (FWO) (No. G0 D9318 N) and from CROP TRUST (GS15024). The collection mission in PNG was funded by the Global TRUST foundation project "Crop wild Relatives Evaluation

of drought tolerance in wild bananas from Papua New Guinea, grant no: GS15024. The authors thank all donors who supported this work also through their contributions to the CGIAR Fund (http://www.cgiar.org/funders), and in particular to the CGIAR Research Program Roots, Tubers and Bananas (RTB-CRP).

**Acknowledgments:** We gratefully acknowledge all those who contributed to the making of these seed collections: Joel Pilon, Ian Nabo, Billy Pitalai, Godfried Savi, Philip Daur, Elasanty Yabu, Samson Itau, Joel Lapiu, Tony Kunou, Steven Kambase, Joe Guaf, Nemothy Sinoksor, Julie Sardos, Gabriel Sachter-Smith and David Eyland. Thank you also to Arne Mertens for the data to plot Figure S1 and Richard Ellis for helpful advice.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Review*

### **Born to Eat Wild: An Integrated Conservation Approach to Secure Wild Food Plants for Food Security and Nutrition**

**Teresa Borelli 1,\*, Danny Hunter 1, Bronwen Powell 2, Tiziana Ulian 3, Efisio Mattana 3, Céline Termote 1, Lukas Pawera 4,5, Daniela Beltrame 6, Daniela Penafiel 7,8, Ayfer Tan 9, Mary Taylor <sup>10</sup> and Johannes Engels <sup>1</sup>**


Received: 31 July 2020; Accepted: 24 September 2020; Published: 1 October 2020

**Abstract:** Overlooked in national reports and in conservation programs, wild food plants (WFPs) have been a vital component of food and nutrition security for centuries. Recently, several countries have reported on the widespread and regular consumption of WFPs, particularly by rural and indigenous communities but also in urban contexts. They are reported as critical for livelihood resilience and for providing essential micronutrients to people enduring food shortages or other emergency situations. However, threats derived from changes in land use and climate, overexploitation and urbanization are reducing the availability of these biological resources in the wild and contributing to the loss of traditional knowledge associated with their use. Meanwhile, few policy measures are in place explicitly targeting their conservation and sustainable use. This can be partially attributed to a lack of scientific evidence and awareness among policymakers and relevant stakeholders of the untapped potential of WFPs, accompanied by market and non-market barriers limiting their use. This paper reviews recent efforts being undertaken in several countries to build evidence of the importance of WFPs, while providing examples of cross-sectoral cooperation and multi-stakeholder approaches that are contributing to advance their conservation and sustainable use. An integrated conservation approach is proposed contributing to secure their availability for future generations.

**Keywords:** wild food plants; food security; nutrition data; multi-sectoral collaboration; policy; conservation

#### **1. Introduction**

The practice of consuming wild food plants (WFPs) is as old as human prehistory. Early humans obtained their food by hunting, fishing and gathering these plants, or parts of plants (e.g., stems, roots, flowers, fruits, leaves, buds, and seeds), that were safe for human consumption. It was not until 10,000 years BC that people started settling into more permanent homesteads and domesticating plant species (mostly carbohydrate-rich staples) while maintaining some hunter-gatherer activities and collecting WFPs from the wild [1,2]. This still holds true for some traditional horticultural societies today (e.g., the Machiguenga in South America) [3]. All of the plants we now call domestic crops were once WFPs, altered by human manipulation to achieve domestication by selecting more favorable plant traits. With plant domestication and farming came also the development of weeds; that is, unwanted plant species in cultivated fields, and many of the WFPs consumed today include relatives of today's crops.

Today, the term "wild" is mostly taken to indicate species that grow spontaneously in self-sustaining populations outside cultivated areas, in field margins, forests, woodland, grassland, and wetlands (e.g., paddy fields), independently of human activity [4]. However, the distinction between "wild" and "cultivated" or "domesticated" is not so clear-cut and many WFPs fall somewhere in between these extremes depending on the degree of human intervention and management. For example, they can grow spontaneously in areas that are or have been themselves cultivated [4,5], or, as in the case of the "quelites" greens in Mesoamerica (e.g., the genus *Amaranthus*, *Chenopodium*, *Porophyllum*, *Portulaca*, *Crotalaria,* and *Anoda*), they have become the focus of systematic in situ management practices such as "selective harvesting" and "let standing", with important repercussions on plant communities [6]. Another known management practice is that of "encouraging growing" recorded by Cruz-Garcia [7] in the Peruvian Amazon along the deforestation border. Surveys revealed that, out of 30 wild food plant species identified, 20 are actively managed by local farmers and that most are transplanted from the forest to their agricultural fields for easy access. From these, 57% of the species are classified as weeds, yet are perceived by farmers to play a role in food security, particularly with increasing deforestation and reduced availability of food plants [7].

In this review paper, the term "wild food plants" is extended to all those food plants (herbs and spices included) that are also semi-domesticated, in addition to economically important non-timber forest food products, such as açaí berries and Brazil nuts [8]. As they are often wild relatives of domesticated species, WFPs have potential for domestication and can provide a pool of genetic resources for hybridization and selective breeding [9].

#### **2. The Importance of Wild Food Plants Today**

WFPs continue to play a vital role in the subsistence of many human populations particularly when the availability of food crops is scarce, when household budgets are insufficient to buy enough food or when access to markets is challenging [5,8,10–16]. Wild foods are also integral to traditional food systems and have nutritional and cultural value for many indigenous peoples [4,5,17,18]. Deeply connected to their land, indigenous peoples, who represent 5% of the global population [19], are often the sole custodians of rich and diverse knowledge relating to plant uses and traditional food systems and to local food biodiversity existing within the ecosystems they inhabit [18]. Traditional communities also have better ecological knowledge about local environments and their customary users, making monitoring and regulating of natural resources easier [20].

Although the caloric contribution of WFPs to people's diets is generally low compared to staple foods [21], these species contribute to diet diversification in many geographical settings where otherwise monotonous diets may prevail [22–26]. Wild foods (both plants and non) provided between 1% and 19% of the iron consumed, between 5% and 45% of the calcium and between 0% and 31% of the vitamin A equivalents (RAE) in the diets of women and children in studies from Benin, Tanzania, and the Philippines [21]. These neglected biological resources have, in fact, been shown to contain equally, if not higher amounts, of nutrients than more widely available commercial crops [5,27–29], and, if properly assessed and managed, could be introduced in national food and nutrition security and sovereignty strategies that focus on nutrient adequacy rather than quantity of staples, while being culturally acceptable.

WFPs could also be central to efforts directed at empowering local market actors as well as reducing the distance between consumers and producers and the overreliance on globalized value chains. Although, recent research by Kinnunen et al. [30] highlights the unfeasibility of localizing production for important global staples such as rice, maize and temperate cereals, there is increasing evidence that the local trade of minor crops, traditional varieties, and WFPs has potential to empower communities and increase livelihoods in rural areas, particularly of women and youth [31,32]. Meanwhile, the COVID-19 crisis has revealed the vulnerability of our global food systems to disease-related disruptions and shocks [33–35]. For example, the imposed travel restrictions on people and goods as a result of the lockdowns are causing logistical bottlenecks in food supply chains [36]. Given the national and international trade restrictions, long supply chains are struggling to cope with the rise in food demand for non-perishable food supplies [37], while short supply chains are suffering due to the closing of informal and local open-air markets [38], where the majority of the world's population still obtains fruits, horticultural, and other perishable products [37,39]. At the same time, the pandemic has opened up opportunities for a new food system paradigm that supports local self-sufficiency and domestic agricultural production and sees home and community gardens, traditional agroecosystems, and farmers' markets as essential services [38,40]. With food shortages affecting specialized, high value horticultural crops [41], people are turning to traditional vegetables and WFPs as a sustainable source of food, vitamins and nutrients [42], not to mention for herbal ingredients, traditional medicine formulations or new biopharmaceuticals [38,43,44].

This paper builds mainly on the authors' own efforts being undertaken in several countries to provide evidence of the role of WFPs in supporting nutrition and livelihood security. This paper also provides examples of cross-sectoral cooperation and multi-stakeholder approaches that are contributing to the better conservation and use of WFPs, including by fostering linkages between in and ex situ conservation. In the case of WFPs, "use" includes the various practices and activities involved in (i) domesticating wild species; (ii) the management of wild species and their habitats in and around production systems to promote the delivery of ecosystem services; and (iii) the introduction of wild species into production and consumption systems, for example by creating demand for the species, and regulating their harvesting in the wild. Lastly, details will be provided of a proposed integrated conservation approach that focuses on local interventions based on traditional food systems.

#### *2.1. Diversity (Geographical Use) and Contribution to Diets*

The use of WFPs in many countries is confirmed by national contributions to the recent "State of the World's Biodiversity for Food and Agriculture"—"SOWBFA"—of the Food and Agriculture Organization of the United Nations [45]. Of 91 countries reporting information for compiling the report, 69 nations reported a total of 1955 wild plant species that contribute to food security and nutrition in their respective countries, as well as making diets healthier and more diverse. However, as the examples provided by the authors and recently published papers [46] demonstrate, the number is probably much higher and these species remain largely unreported in national statistics, as does the actual contribution of these biological resources to national economies in many parts of the world [47]. Table S1 in the Supplementary materials lists the wide range of plant families that encompass the edible wild and semi-cultivated plant species researched by the authors and mentioned in the text as contributing to food and nutrition security. The list, as the review carried out by the authors, is by no means exhaustive and could undoubtedly include more.

#### 2.1.1. Africa

As part of the MGU Useful Plants Project (UPP) managed by the Royal Botanic Gardens, Kew, UK, institutional partners working alongside local communities in Botswana, Kenya, Mali, South Africa, and Mexico identified 615 species used for food across the five countries [48]. Information on seed conservation, propagation and traditional uses has been published for 48 of them and is now available on the internet [49]. In Africa, the species included the baobab (*Adansonia digitata)*, the mongongo tree (*Schinziophyton rautanenii*), and the morama bean (*Tylosema esculentum*) [49]. Research undertaken by Bioversity International in the early 1990s has documented 210 African leafy vegetables in Kenya alone [50]. These are wild or semi-domesticated species that are grown mostly for household consumption or traded informally, but which have seen a revival particularly in urban and peri-urban areas [51]. In Western Kenya, between 23 and 42 African leafy vegetables continue to be consumed by local communities depending on the district. Eleven species, including amaranth (*Amaranthus* spp.), spider plant (*Cleome gynandra*), and African nightshades (*Solanum* spp.) were selected for further research as part of the African Leafy Vegetables program from 1996 to 2004 (Bioversity International and EIARD, 2013; Gotor and Irungu, 2010) [51,52] as well as for the Biodiversity for Food and Nutrition Project [28,53]. In addition to filling the nutrient gap, a cost of diet study carried out in Eastern Baringo, Kenya, has shown that wild plant species, especially vegetables, are able to significantly reduce (by 30–70%) the cost of a nutritious meal for women and children aged 6 to 23 months in hypothetically-modeled lowest cost nutritious diets [54].

#### 2.1.2. South America

In Ecuador, one of the top seven mega diverse countries in the world, wild edible fruits and plants collected from a diverse range of habitats play a fundamental role in traditional diets, particularly for the indigenous communities living in forest areas. Studies in the country by Penafiel et al. [55,56], documenting local knowledge on the use of WFPs among the Andean indigenous communities of Guasaganda (Cotopaxi) and the Andean Kichwa mothers of Arosemena Tola (Napo), recorded the culinary use of 49 and 10 WPFs, respectively. Brazil also contains vast amounts of wild food plant diversity [57]. Some of this diversity is of national and regional relevance, e.g., Brazil nut (*Bertholletia excelsa*) and açaí (*Euterpe oleracea*), but most is of local value and its potential nutritional and economic value remains unexplored and unexploited [58]. The "Plants for the Future (PPF) Initiative", a prioritization exercise undertaken by the Ministry of the Environment that set out to explore the wealth of Brazil's plant biodiversity, has identified a considerable number of wild species of nutritional value and market potential. Across the country's five eco-regions, out of 78 native undervalued edible plant species, 49 are found exclusively in the wild (mostly fruits and nuts) [28]. Mostly found in forest areas, the species are managed by family farmers or harvested from the wild by local communities using traditional practices. The link between local communities and nature is such that the Brazilian ministries of Agriculture, Environment, and Social Development have coined the term "sociobiodiversity" to describe these traditionally managed biodiversity-derived goods that are sold in local markets, provide incomes and improve the livelihoods of traditional communities, while protecting biodiversity and the environment.

#### 2.1.3. The Mediterranean

In the Mediterranean, WFPs are still common in traditional cuisine and are widely consumed locally [59,60]. In their compendium of gathered Mediterranean food plants, Rivera et al. [61] identified approximately 2300 different WFPs and fungi taxa in this region alone, of which 1000 are strictly used locally. As part of the Biodiversity for Food and Nutrition (BFN) project, Turkey prioritized 42 wild edible plants for further research [28] out of hundreds of known species [59,62,63], while across Morocco, Nassif and Tanji [64] compiled a list of 246 wild plant species used as food. While many WFPs are only used occasionally or in small regional areas, some are central to Moroccan diets and culinary traditions. Aromatic herbs such as thyme (*Thymus* spp.), mint (*Mentha* spp.), and sage (*Salvia* spp.) are the most widely consumed wild plants; however, they contribute little to the diet in terms of energy (kcal) and nutrients because they are used as condiments [65]. Wild leafy vegetables, on the other hand, are a seasonally important component of Moroccan diets, particularly in rural Morocco where 86% of

households reported consuming wild leafy vegetables (WLVs) on a regular basis [66]. Some of the most commonly consumed WLVs (many of which are also consumed in Turkey and other Mediterranean countries) include mallow (*Malva* spp.), purslane (*Portulaca oleracea)*, goosefoots (*Chenopodium* spp.), docks and sorrels (*Rumex* spp.), fennel (*Foeniculum* sp. cf *F. vulgare*), golden thistle (*Scolymus hispanicus),* and watercress (*Nasturtium o*ffi*cinale)*. They are commonly served as a cooked salad or side dish, eaten in moderate portion sizes (approximately 50 g per meal). Argan oil (*Sideroxylon spinosum*), capers (*Capparis spinosa* and *C. decidua*), acorns (*Quercus* spp.), and the fruits of the strawberry tree (*Arbutus unedo)* as well as jujube (*Ziziphus jujube*), mulberry (*Morus* spp.), and blackberry (*Rubus* spp.) are other commonly consumed plant products in this region [64,65,67].

#### 2.1.4. Asia Pacific

The consumption of WFPs and food trees makes a significant contribution to human health in the Pacific region [68]. In "Food Plants of Papua New Guinea, A Compendium" [69], Bruce French produces a list of food plants, many of which are sourced from the wild, including root crops and staples, legumes, green leafy and other vegetables, nuts, fruits and what are categorized as "minor foods and flavorings". For example, the kernel of wild *karuka* (*Pandanus brosimos*), endemic to Papua New Guinea (PNG), is eaten by about one-third of the rural population [70], particularly by communities living at high altitudes. When the fruit matures, villagers migrate to high altitudes to harvest the fruit and extract the nuts. Nuts have not been recorded in the main highland markets, but it is possible that they are sold in some high-altitude locations [71]. PNG and surrounding region are also one of the few places in the world where communities obtain the majority of their carbohydrate staple from a wild food plant: sago [72,73].

In Niue (Polynesia), the traditional processing of wild arrowroot (*Tacca* spp.) is still an ongoing practice. Starch processed from the root is a local delicacy used to make local puddings and breads [74]. Thaman [75] lists 60 WFPs used in Fiji, noting that these plants play an important role as emergency or famine foods when extreme climatic events disrupt cultivation. Among these are wild marine seaweeds such as sea grapes (*Caulerpa racemosa*), known as nama, and other edible seaweeds that are still widely consumed. "The Guide to the Common Edible and Medicinal Sea Plants of the Pacific Islands" provides an insight into marine WFPs and the benefits that can be gained from their use [76].

In the mosaic tropical landscape of West Sumatra, Indonesia, composed predominantly of rice fields, home gardens, cacao agroforestry, and forests, with the help of local communities, the Food, Agrobiodiversity and Diet (FAD) project has identified 85 WFPs [77]. In this region, WFPs are consumed less than in the past, and the FAD project aimed to raise knowledge and awareness of wild foods by organizing workshops, traditional food competitions, and sharing community materials such as illustrations, posters, and community guidebooks, on food plants for nutrition and health [78].

#### *2.2. Income Generation*

In many parts of the world, WFPs are not only harvested for subsistence [79–81]. Gathered in excess, they are sold in local markets to generate income, thereby contributing to the household economies of gatherers and collectors, usually women, or to bolster the incomes of migrants and unemployed moving from rural to urban areas [82]. For example, in the Chimanimani communities living in the Trans-Frontier Conservation Area in Mozambique, the fruits of *Uapaca kirkiana* and *Strychnos madagascariensis* are sold for a reasonable profit and represent an important source of income outside the maize harvest season (March to May) [83]. In their review paper, Hickey et al. [84] showed that 50% of almost 8000 households sampled in forested areas of 24 developing countries across Asia, Africa, and Latin America derived their income from wild food collection. The study also highlighted that the sale of plant foods contributed 2.3% to total household income across the study sites on three continents, the proportion increasing to 2.8% in Africa and Latin America, particularly in poorer households.

In parts of Turkey, where WFPs are central to traditional cooking, wild edibles are sold unprocessed in local markets and processed (e.g., pickled, canned, or frozen) in district markets or supermarkets via wholesalers and middlemen [85]. In 2012, in the Pacific Island States of Fiji, Samoa, and Tonga the yearly production and revenue from the harvesting and sales of the seaweed *Caulerpa racemosa* was valued at USD 266,492 [86]. However, the true extent of this revenue is not always available. For example, a recent European assessment established the value of collected non-wood forest products, mainly food plants, at € 19.5 billion with value per hectare rising to € 77.8, and ten times above the official European estimates [87]. Many markets for WFPs are informal, and market players may hold back information because of illicit harvesting in local conservation areas [82]. Data about geographic and temporal distribution, production cost, quantity harvested, and price is also often limited. Increased profits can often lead to overexploitation of WFPs and negative outcomes for the entire community [20]. To avert this possibility, participatory research is key to establish sustainable management guidelines and harvest rates, and to monitor the ecological impacts of increased use [83].

#### *2.3. Threats to WFPs*

Despite the realization of the potential use of WFPs in food security and poverty reduction strategies, the SOWBFA, along with other recent global reports [46], warn us that this precious diversity is fast disappearing, particularly in forest habitats [88,89]. Land use changes (e.g., conversion to agriculture, change in agricultural practices and infrastructure development), habitat destruction (resulting from timber harvesting, fuelwood collection, grazing, and forest fires) and overharvesting collectively account for 62% of the threats reported to WFPs in SOWBFA, which mostly grow beyond the limit of protected areas [45,90,91]. The SOWBFA used the Sampled Red List Index for Plants of the International Union for Conservation on Nature (IUCN) [92] to determine that, of a total 822 WFP species considered across 7 different classes, 73% are currently at low risk of extinction (Figure 1), with some classes highly threatened in the wild (e.g., WFPs that are derived from conifers and cycads). However, the IUCN Red List Index for Plants includes global conservation assessments for only one third (31%) of known WFPs. Local assessments for many WFPs that are currently excluded from the IUCN assessment paint a very different story indicating the need to consider community perceptions when ascribing risk class (Table 1). Furthermore, an assessment of the comprehensiveness of conservation of 1587 WFP taxa (including cereals, fruit, and nuts), carried out by the International Center for Tropical Agriculture (CIAT) as part of a larger study to identify conservation gaps for useful plants, shows that only 3.3% of WFPs are sufficiently conserved ex situ, i.e., in gene banks or in other living plant repositories, while 89.1% require urgent off-site conservation measures given the impending threats to their existence [93]. Their continued use in diets, when accompanied by careful sustainable management by the communities consuming them, and protection of WFP habitats, on the other hand, seems to have ensured their momentary conservation in situ, in the natural habitats in which they grow. Of the WFP taxa analyzed 42.1% are sufficiently conserved, 46.7% deserve medium priority and 11.1% require stepping up conservation measures [93]. Nonetheless, Khoury et al. [93] caution against the overreliance on protected areas for the long-term conservation of these species. Rapidly warming temperatures and habitat destruction can alter the species' geographic distribution, driving them across the artificially designated boundaries of many protected areas in pursuit of favourable growing conditions [94].

**Figure 1.** Number of WFPs and fungi on the IUCN Red List of Threatened Species classified by class and risk category Source: IUCN Red List 2017. Adapted from FAO [45].

Given that many WFPs grow in agricultural systems (as weeds, in hedge rows, as wild trees in agroforestry systems, and in small forest patches [5,21]), agricultural change, including intensification, more pesticide use and removal of trees can threaten the existence of these biological resources [12,14,95]. Food production systems that pollute the environment by using large quantities of fertilizers, pesticides, and herbicides, are also major causes of biodiversity loss [45,88,96]. Applying chemical herbicides in rice fields or agroforestry plots, for example, leads to the reduced availability of WFPs in West Sumatra, Indonesia [77]. WFPs that survive aerial spraying with herbicides are contaminated by these harmful substances, making them unfit for human consumption, while pesticides wipe out many of the pollinators needed for plant reproduction.

Overharvesting can also be an important pressure on non-timber forest products, including wild foods [97]. This is the case for Morocco and Turkey. Morocco is the twelfth global exporter of medicinal and aromatic plants, a trade that places extensive harvesting pressure on many of the species traditionally used as herbs [98]. A rapid vulnerability assessment carried out by Lamrani-Alaoui and Hassikou identified six species that grow across wide areas of Morocco's government owned forests (*Thymus satureioides, Lavandula dentata, Origanum compactum, Origanum elongatum, Salvia rosmarinus* and *Fraxinus dimorpha*) as needing urgent conservation, restoration, and sustainable management measures [98].


survivalofnutritiousandlocally

#### *Plants* **2020** , *9*, 1299









Exacerbating these problems in the different geographies are the uncertain effects of climate change, which in many countries is expected to lead to increased variability in seasonality, temperatures and precipitations and increased incidence of hurricanes and wildfires [89]. Climate change is also predicted to severely impact cultivated plants, affecting crop production in specific geographic locations [115], stripping nutrients from staple crops [46,116] and making WFPs all the more important for food and nutrition security. Although generally highly adaptable and often more drought tolerant than cultivated crops, WFPs, as many useful plants, are also not fully resistant to climate change [116]. In the past, many WFPs survived major climatic fluctuations, but thematic studies on the implications of future climate change suggest important impacts on the ability of wild species to survive. This includes WFPs, particularly in tropical regions where economies are already fragile and capacity may be inadequate to protect these species effectively [94,116]. One key impact that could threaten WFP use is the likely shifts in both WFP geographic ranges and phenological changes in ripening times. This could create mismatches with traditional knowledge and practices of the communities that traditionally harvest them [117].

At present, there are very few formal systematic efforts that support and regulate the conservation and sustainable use of WFPs [118]. A sample survey of some of the most recent National Biodiversity Strategies and Action Plans (NBSAPs) submitted to the CBD as part of the reporting requirements of member states (e.g., Chile, Morocco, and Portugal), shows that rarely do these strategies refer specifically to WFPs or, if they do, are very vague in terms of the measures needed to protect them. Actions are mostly limited to ex situ conservation measures [46], while no concrete activities are put forward to support their conservation via sustainable use [45]. Furthermore, appropriate and effective governance mechanisms are seldom in place to safeguard the rights of indigenous people and local communities to sustainably manage and benefit from the use of WFPs (and prevent their over-exploitation by others) [119].

The use of wild species, however, is explicitly recognized as useful for improving food and nutrition security in several international agreements, strategies and action plans: in the 2030 Agenda for Sustainable Development (SDG2, Target 2.5), the International Treaty on Plant Genetic Resources for Food and Agriculture (International Treaty), the Second Global Plan of Action for Plant Genetic Resources for Food and Agriculture (Second GPA), and in the Global Strategy for Plant Conservation of the Convention on Biological Diversity (CBD). The CBD, the main international agreement aimed at conserving biological diversity, accords explicit recognition to sustainable use for the long-term conservation of ecosystems, species and genes, which must continue to be used, but "in a way and at a rate that does not lead to the long-term decline of biological diversity" [120]. Intrinsic in the term "sustainable use" is that it generates benefits (e.g., nutritional, cultural, and financial) for the custodians and users of these wild species. These benefits encourage people to continue conserving these biological resources and the habitats in which they grow or live. However, the real challenge is to ensure this sustainability is maintained given the rising demands on global resources imposed by population growth and economic development, combined with the uncertain effects of climate change mentioned above.

#### **3. Barriers to the Greater Use of WFPs**

The disregard of WFPs for food security and nutrition can be partly attributed to a lack of evidence and awareness among policymakers and other stakeholders of the importance of wild foods to diets, livelihoods, and food security, coupled with a number of market and non-market barriers limiting their untapped potential.

Underpinning the lack of recognition for WFPs is also limited or short-term research and extension funding to support the exploration of non-conventional, traditional and indigenous food resources. Many of these barriers were summarized by Heywood [4] and are still very much valid today:


Like other neglected and underutilized species, additional barriers to the promotion of WFPs in food production and consumption patterns include: limited and fragmented data of the nutritional importance of these species; fragmented data on the quality and nutritional impacts of WFPs on household nutrition [121]; and knowledge gaps on the species' biology and ecology to develop domestication and management strategies [45,46].

Unfavorable and disabling national policies, coupled with the many stakeholders and interests involved, represent an additional obstacle to greater recognition for WFPs. The main policy barriers were identified and summarized by the Strategic Framework for Underutilized Plant Species [122], of which WFPs are part of. These are provided in Table 2 below.

**Table 2.** Barriers that hinder the improvement of national policy frameworks towards supporting WFPs.


Further contributing to the demise of WFPs, is the low recognition of value and perception of these foods as being "women's food" [66] "food for the poor" or "famine foods" to be harvested only when staple crops fail, as well as lack of institutional capacity to mainstream this diversity into national production and consumption patterns [28]. On the other hand, in some regions, such as West Sumatra, communities perceive WFPs positively, but the main barrier to their greater use is their reduced availability caused by land degradation and agriculture intensification [77]. In many places, traditional wild leafy vegetables are disappearing from local diets due to changing dietary patterns and preferences driven by globalization and increasing market integration [123]. Wild leafy vegetables (WLV) and wild food plants in general are undervalued and seen as "un-modern" in Morocco, Turkey [28,66], and many other parts of the world. This lack of value places the role of WFPs in the diet at risk, although it may ease pressures on overharvesting. In Brazil and Kenya, changing dietary patterns and lifestyles has reduced the diversity and availability of wild fruit and vegetables in market settings, which focus instead on a limited number of exotic crops [124]. This has led to people consuming sub-optimal diets that are increasingly unhealthy, unsustainable, and inequitable for many populations [125].

#### *3.1. Contribution to Nutrition and Diets*

The contribution of wild food biodiversity to diets and nutrition is simultaneously limited by a severe lack of food composition data for many neglected and underutilized cultivated and wild foods [126] as well as by a lack of accurate botanical identification for many foods recorded in dietary records or food composition tables [45,127,128]. Nutrient composition data indicates the presence

and quantity of nutrients (e.g., energy, proteins, minerals, and vitamins) as well as the compounds that can impact the bioavailability of nutrients within a food. These data are combined with dietary records of the foods consumed to assess whether individuals or groups are meeting their dietary requirements [129]. Nutrient composition data do not exist for many WFPs, and when they do there may be high variation in nutrient composition for a given species across space and time [130]. The few WFPs that have nutrient composition data and that are included in local food composition tables are often identified by local names. This hinders the use of these data to fill nutrition gaps in other locations where the same species might be present and used but is identified by a different local name. Many data sets lump all wild foods into a single food group (e.g., wild greens). For example, in analyzing data on wild harvests in 24 developing countries across Africa, Asia, and Latin America, Hickey et al. [84] found that only a small percentage (0.9%) of the collected mushrooms were identified by species, the rest was reported nonspecifically as "mushrooms".

In some cases, when WFPs are lost from the diet they may be replaced by similar cultivated species, but in other cases they are not. Anecdotal evidence from Morocco suggests that when people stop or reduce the consumption of WLVs in their diet, these are not replaced with cultivated alternatives, leading to a reduced consumption of any leafy vegetable and fruit and vegetables in general. This is particularly worrying given global recommendations [131] to consume at least 5 servings of fresh fruit and vegetables (including berries, green leafy and cruciferous vegetables, and legumes) per day as a protective measure against cardiovascular diseases and type II diabetes [132–136].

Practical challenges also exist in measuring wild food consumption and contribution to the diet relative to other foods [5,137]. Although in recent years, several investigations have tried to assess the role of wild food biodiversity and the contribution of forests and agroforestry systems to human dietary intakes [13,14], the real dietary contribution of wild food plants, berries, fruit, nuts, and mushrooms harvested within and around people's homesteads and in forested areas remains poorly understood. Geographical variations exist regarding the proportion of WFPs consumed. While in in the global North WFPs mostly have cultural and recreational value [138], in some low-income countries they significantly enrich people's diets [119]. Rowland et al. [13] found that the collection of forest foods represents a regular livelihood strategy for many households and that forest dependent communities living in specific sites in Brazil, Cameroon, and Ethiopia derive as much as 80–96% of wild fruits and vegetables from the forest. In some areas, the nutritional contribution of fruits and vegetables is such to cover 50% and above the minimum dietary recommendation for these food groups [13]. Differences in consumption might also vary by ethnicity. For example, in documenting the traditional food systems of Western Sumatra, Pawera et al. [77] found that different ethnic communities living in the same environment have different knowledge and uses for the same WFPs. Seasonal fluctuations in WFP occurrence and therefore consumption by local communities might also not be adequately captured with a single survey [139]. Other challenges include cultural and language barriers and perceived power imbalances during questionnaire administration that can alter the surveys' accuracy and reliability [137]. There is a huge body of research that only lists the edible species known to community members but neglects to quantify the use of WFPs in local recipes nor is their use standardized in nutrition surveys [121].

#### *3.2. Gathering Grounds, Collection Practices and Use*

An additional knowledge gap is represented by the lack of information on traditional gathering grounds and the sustainability of collection practices. In the SOWBFA, the ecosystem origin reported for WFPs is either from forests (>25%) or unknown (>45%) [45]. An often-overlooked practice is urban and peri-urban foraging for WFPs. In their cross-continental study of urban foraging spanning India, South Africa, Sweden, and the US, Shackleton et al. [140] found that urban foraging is a widespread custom that is practiced independently of wealth and social status and is driven by different motivations varying in time and place. Wooded areas on public land, local lake beds, and other urban habitats harbor nutritionally rich greens and fruits. Even spontaneous vegetation growing in alleyways was

reportedly used by Indian residents for food and culinary use [140]. Aside from direct consumption and small-scale trade, other benefits include "improved physical and psychological health, sense of place, increased ecological knowledge, stronger connections with nature, food, income or cash saving, and a source of pride" [140]. The important cultural ecosystem services offered by these plants are apparent in a study of WFP gathering and consumption trends across Spain [141]. The authors observe that WFPs continue to be used in areas with deep-rooted culinary traditions and in some instances have become gourmet ingredients for chefs. Schulp et al. [142] also suggest that the cultural benefits of wild foods in the European Union might exceed their income and food benefits and observe that wild mushroom and food plant collecting are generally highest in Southern European countries where gastronomic identity is strongest.

#### **4. An Integrated Approach for Conserving and Sustainably Using WFPs**

With the gradual disappearance of WFPs from nature and diets, the question is how to effectively promote their sustainable use and simultaneously conserve them for food security and nutrition. Because they exist on a continuum of human management, from truly wild to semi-domesticated [7], and because the germplasm and other plant material (e.g., tissue, embryos etc.) of some species may not be suitable for ex situ conservation [143], both in situ and ex situ conservation should be combined for optimal results [144–146] (Figure 2). In situ conservation strategies can complement ex situ conservation and allow WFPs to continue to evolve adaptive traits in their natural environments while benefiting those who need them most, particularly in areas where high diversity, rural poverty and malnutrition coexist.

**Figure 2.** Proposed best practices for the long-term co-creation of conservation and sustainable use of WFPs help overcome many of the challenges identified.

Above, we have identified an array of threats to WFPs including: land use changes, deforestation and degradation; agricultural change, intensification and chemical input use; overharvest or unsustainable harvesting; loss of traditional management practices that communities used to promote the production of wild food plants (for example, pruning and burning); and climate change. We also identified a range of barriers that are contributing to the loss of use and value for WFPs, such as, lack of information (diet, nutrition, safety economics, and ecological); lack of harvest, storage and value chain tech and infrastructure; and lack of awareness, education and inclusion in policy and programming. In the subsequent sections of this paper we propose a set of best practice actions that can be taken to support sustainable use and conservation of WFPs. This set of actions laid out in Figure 2 will act to overcome or mitigate against many of the threats and barriers identified.

The proposed set of best practice actions includes: the collection of information (identify the diversity of WFPs that are present in a given environment, information on nutrient composition and contribution to diet, economic importance, and ecological studies to determine sustainable offtake); (ii) prioritize the species with greatest potential to fill nutrition gaps, greatest need in terms of conservation, greatest cultural importance; (iii) protect species that are vulnerable through ex situ conservation; (iv) promote the use and management of WFPs in natural environments (in situ) (including sustainable management and collection guidelines where needed); (v) develop domestication programs where necessary and possible to avoid overexploitation in the wild; (vi) build local capacity to improve storage, processing, value chains, and markets (and all related technology and infrastructure); (vii) integrate WFP into programming and education and other youth outreach so as to raise awareness; (viii) develop and strengthen policies that support the conservation and sustainable use of WFPs; and (ix) and build donor commitment to funding efforts to support sustainable use and conservation of WFPs.

Each community and each WFP are unique, and will require a different set of actions, possibly occurring in a different order. Successful implementation of the set of best practice actions best suited to any given context will require working in a coordinated fashion across disciplines and sectors at the local, regional, and international level, and is largely dependent on the close and active participation of the national and local stakeholders. Due to the limits of time-bound projects (e.g., capacity, resources), it is rare for a single project or intervention to cover all elements or actions needed for a comprehensive and integrated approach. Below we present examples of best practice actions that we believe have successfully helped to further the conservation and sustainable use of WFPs.

#### *4.1. Identify and Prioritize*

The identification of WFPs to include in conservation and sustainable use strategies will almost invariably require close collaboration with indigenous and local communities who are the main users and custodians of this diversity. As opposed to extractive methods, participatory research approaches that integrate traditional and scientific knowledge are the most appropriate to collect information on WFPs while maximizing benefits for the communities involved [147]. Prior to the intervention, the community should be aware and agree on every aspect of the research process so that the methods, the analysis and the purposes of the data collection are clearly understood [148]. Ethnobotanical surveys and free-listing exercises are the most commonly used methods to complement scientific ecogeographic assessments. In the majority of the studies discussed in this paper, focus group discussions conducted with knowledgeable key informants were able to help fill knowledge gaps in WFP availability and use. Useful tools for assessing the potential of WFPs to fill seasonal food insecurity gaps, and low dietary diversity characterized by low fruit and vegetable consumption, are seasonal calendars, such as the one shown in Figure 3 developed by BFN Brazil to investigate flowering and fruiting seasons for wild fruit species. Data can then be transformed into an accessible and understandable tool to assist communities and decision makers adopt healthier diets based on local biodiversity [149].

Market surveys are also useful to understand what WFPs might be available for consumption within a community. A notable example is represented by BFN Turkey, which undertook market surveys and key informant interviews with 2334 local wild plant gatherers, sellers and consumers of WFPs to capture the diversity of WFPs still being used across three ecogeographic regions [28,150]. Documenting the use of wild plants in this participatory way has several benefits that include: (i) facilitating knowledge transmission from elders to younger generations and between community members; (ii) stimulating local innovation without undermining cultural traditions and local governance mechanisms, and (iii) ensuring that the community can use this diversity to address its own questions,

challenges and needs [147]. In Western Kenya, for example, biodiversity surveys and dietary health assessments were followed by a series of participatory workshops that enabled five communities to gain and share knowledge on available wild and cultivated biodiversity, discuss options on ways to use this diversity to improve malnutrition within their communities, rank and prioritize the most suitable species and develop their own community action plans (CAP) towards this end. In collaboration with the ministries of Agriculture and Health, training was then provided to assist with the integration of the chosen species—mostly African leafy vegetables and legumes—into sustainable production systems and diets. One year into CAP implementation, mean dietary diversity scores and the percentage of children meeting minimum dietary diversity had significantly increased in all the households in the sublocation, irrespective of participation in the scheme, indicating the adoption of these best practices by neighboring households. The dietary diversity scores of women from participating households had also significantly increased [151].

**Figure 3.** Research into the flowering and fruiting period of wild fruits and greens within a given geography can be used to develop an adaptable tool for informed decision making at both community and government level. Credit: BFN Brazil.

With all probability, surveys will reveal a long list of species that could be the focus of further research and promotion in food and nutrition strategies. Realistically, limited resources will often require a prioritization exercise that reduces the list to a manageable number. Since the intent is to ensure that WFPs are safeguarded and sustainably consumed, again community participation in the prioritization process is key, for example, to single out species that could be conserved in seed saving facilities, domesticated, or included in breeding programs, or to identify WFPs with the potential to contribute to nutrition, climate-change resilience and other aspects of community well-being. In the earlier example from Turkey, the BFN team developed an ad hoc sustainability index to reduce an initial sample size of 43 species, mostly WFPs, to three target species—foxtail lily (*Eremurus spectabilis*), golden thistle (*Scolymus hispanicus*), and einkorn wheat (*Triticum monococcum*)—which have since been the object of domestication research as well as post-harvest handling and value chain analysis [150].

#### *4.2. The Nutritional Importance of WFPs and Associated Traditional Knowledge*

Understanding the nutrient content and health properties of WFPs (e.g., compositional data) as well as their contribution to diets will also greatly assist in the prioritization process. Compositional data is key to national nutritional planning and for developing locally fitting dietary guidelines. Seldom, however, does nutrition information appear in national food composition tables and

databases, and if data does exist it is either scattered across different sources in institutional databases, in academic journals and grey literature, or is outdated and/or incomplete making data compilation a daunting task [28,150]. An additional hurdle is the standardization of food composition values. Common component names are often expressed inaccurately (e.g., vitamin A: retinol activity equivalents vs. retinol equivalents) or differ in terms of units, denominators, significant figures, maximum decimal places, and conversion factors [152]. Of further importance, is the documentation and protection of traditional knowledge related to consumption and preparation of WFPs, available largely in the recollections of elderly users, i.e., rural, indigenous, and forest-dependent communities, including local farmers, and city migrants. Some information may be available in national floras, in herbaria and in ethnological studies of local human ethnical groups, but additional botanical, culinary, nutritional, cultural research is required to fill this knowledge gap. As explained in Section 4.1, to avoid issues of misuse and abuse of this information, it is important that respondents are always adequately informed about data use, that sources are acknowledged, and that the data is made available in public databases. Biological knowledge on individual species is also frequently lacking but particularly essential for both in situ and ex situ conservation.

One of the most recent and comprehensive attempts to fill the evidence gap in food composition data is provided by the GEF-supported Biodiversity for Food and Nutrition Project (BFN). Led by Brazil, Kenya, Sri Lanka, and Turkey, and implemented by Bioversity International with support from the UN Environment Programme (UNEP) and the Food and Agriculture Organization of the United Nations (FAO), the project has generated food composition data for 185 plant species, many of them wild, particularly in Brazil and Turkey [28,150,153]. Because of the high costs associated with food composition analysis, the four countries carried out literature reviews prior to the project to identify information gaps and narrow down the list of potentially interesting species to a practicable sample size for analysis and to select the species with the greatest potential for conservation, domestication/management, promotion and marketing. Following the literature review and identified data gaps, food composition analysis was carried out for those species and nutrients for which information was missing or incomplete. Examples of the high nutrient content of WFPs was demonstrated as part of the BFN project in Brazil and Turkey [28]. Similar results were obtained in Indonesia by reviewing the country's food composition data [114] in which wild leafy vegetables are reported to contain higher amounts of limiting micronutrients than more commonly consumed greens (Figure 4).

**Figure 4.** Four wild leafy vegetables from West Sumatra are compared to common lettuce (*Lactuca sativa*) in terms of (**a**) iron content (mg) and (**b**) vitamin C content (mg). In the graphs the letters stand for a lettuce; b vegetable fern "pakis" (*Diplazium esculentum*); c nightshade "leunca/ranti" (*Solanum americanum*), d sweet leaf "katuk/nasi-nasi" (*Sauropus androgynus*), and e water mimosa "komen" (*Neptunia prostrata*). Values are expressed per 100g of fresh, raw ingredient. *Source*: Indonesian Food Composition Data [114].

Species selection and prioritization, literature reviews, and generating food composition data is only the first step of a comprehensive and integrated conservation approach.

#### *4.3. Collecting, Storing and Maintaining WFP Diversity*

Once the species have been identified and prioritized, consideration will need to be given to safeguarding the species for future use, either through ex situ or in situ conservation strategies. Particularly for WFPs, ex situ measures are envisioned as a support to their propagation and reintroduction for habitat restoration [154]. In both cases, to be effective, conservation should involve a wide range of stakeholders working together both in the public and private sectors, across the agricultural and environmental domains [145].

In many cases, governments have established national plant genetic resources programs and seed saving facilities (e.g., gene banks) for ex situ conservation.

However, seed and planting material produced by these "formal" facilities are often inaccessible to smallholder farmers due to strict regulations limiting exchange, little involvement of community actors in the governance, and management of these services [146], as well as imbalances in seed availability, access, and quality for smallholders [155]. However, alternatives do exist. The MGU Useful Plants project, for example, worked closely with communities in Botswana, Kenya, Mali, South Africa, and Mexico to select useful indigenous plants for which high-quality seed collections were established. Seed lots were also banked in the five countries as well as being duplicated and tested at the Millennium Seed Bank in Kew [48,154,156]. Research on seed germination helped support plant propagation activities. The propagules were then planted in community gardens while facilities were established or improved at the local level to facilitate conservation of the prioritized species. Training and knowledge on seed conservation in seed conservation, plant propagation, and planting activities were also provided [49]. This form of conserving WFPs, which takes place in situ, either "on farm" or "in the wild" in natural habitats or protected areas, provides greater opportunity for the involvement of local communities. Once hotspots of WFP occurrence are identified, farmers and indigenous communities living within and around those habitats and protected areas should be involved in conservation activities with due recognition given to their roles and rights in managing WFPs. Further guidance on the establishment of sites for active in situ conservation (i.e., where populations are actively monitored and maintained) of WFPs can be found in the "Voluntary Guidelines on the Conservation and Sustainable Use of Crop Wild Relatives and Wild Food Plants" [145].

Midway between these two conservation approaches are community seed banks or gene banks, which are community-maintained facilities that preserve seeds and other planting materials for local use [146]. These are a collective forms of crop conservation that provide farmers with access to seed, planting material, and traditional knowledge that may otherwise be lost [147]. They also foster community engagement and strengthen the understanding of farmers' and community' intellectual property [157]. By documenting and storing biodiversity and associated traditional knowledge, the seed banks also raise awareness of unique biodiversity in a given area. The community-based organizations (CBO) operating in Vihiga County, Western Kenya, have now established their own community seed bank for African indigenous vegetables and legumes [158]. Creating markets for the seeds and planting material can create additional conservation incentives. Such is the case for the communities in Botswana engaged in the MGU UPP who collect the edible seeds of *Tylosema esculentum* (Burch.) A. Schreib (morama bean) for conservation and cultivation, consumption, sale, and processing into numerous marketable food products. Likewise, the Tsetseng community, through their community trust, have become leading innovators in marketable morama products [159].

#### *4.4. Domestication Programmes and Guidelines for Sustainable Collection*

Depending on conservation status and extent of utilization of WFPs, domestication programs may be required to facilitate cultivation of these wild species and thus to ease the pressure on wild populations and rebuild and restore the genetic diversity that has been lost. A successful example is provided by Turkey in its quest to reduce overexploitation of golden thistle (*Scolymus hispanicus* L.). Golden thistle is a flowering plant that is widely consumed across Turkey and is traditionally collected from the wild for its roots and immature leaves that are sold in local markets [62,160]. Selected by the BFN Project as one of the target species for potential commercialization, breeding, and domestication efforts were undertaken by the Aegean Agricultural Research Institute and the University of Anadolu in collaboration with 37 farmers to select, characterize, and evaluate the species [161]. Nurseries established following initial selection of the hardiest plants produced an average yield of 3.9 tons/ha and up to a maximum of 7 tons/ha. A cultivar called "Sari" was registered and seeds distributed to farmers in the ˙ Izmir province. Golden thistle is now cultivated on an area of 100 ha−<sup>1</sup> [161]. To complement seed distribution, guidelines for the sustainable production of golden thistle were also produced to assist farmers in addressing critical aspects such as climate and soil conditions, plant management, harvest, and seed production (Figure 5).

**Figure 5.** From left to right, top to bottom. BFN Turkey work with farmers to test domesticated golden thistle (*Scolymus hispanicus* L.); the sustainable production guidelines; and harvested golden thistle roots ready for sale Credit: BFN Turkey.

#### *4.5. Strengthening Policies in Support of WFP Conservation and Sustainable Use*

Once baseline data has been gathered, guidelines exist to assist countries in preparing a National Plan for the Conservation and Sustainable Use of Wild Food Plants and crop wild relatives, including setting up a monitoring plan for WFP diversity [145]. The scope of the action plan, its application and effectiveness will very much depend on the national context, the existing policy framework and institutional arrangements, the range of stakeholders involved and their interrelationships, as well as on the resources available. Guidelines are also provided for undertaking preparatory work towards this end [145]. Suggestions are made on how to promote wider use of crop wild relatives and WFPs, but few examples are given to show countries what is practical or actionable. One possibility, which has shown great promise in the BFN Project, is to link producers and collectors to institutional or private sector markets enabling them to benefit from the authorized trade of WFPs. Brazil has used its two largest public procurement programs, the Food Procurement Program (PAA) and the National School Feeding Program (PNAE), to stimulate engagement by family farmers and wild plant collectors (known as extractivists) in sustainable agriculture and the management of Brazilian food diversity, including many WFPs. Both PNAE and PAA are in fact obliged to buy a proportion of the food

they distribute from family farmers and pay a 30% bonus for organic or agroecological produce, preferring suppliers from indigenous and traditional communities [162]. PAA also supports activities aimed at the conservation, production, storage, and distribution of local or traditional seed varieties (Beltrame et al., 2020) [28]. Working closely with government actors and using the nutritional data generated as part of the BFN project, BFN Brazil was also able to promote the publication of Ordinance N◦ 284/2018, which officially recognizes the nutritional and sociocultural value of over 100 plant species native to the Amazon, Caatinga, Atlantic Forest, and Cerrado biomes. This has boosted the market value of native biodiversity including WFPs, with ministries now referring to the list in the "Sociobiodiversity Ordinance" to purchase biodiversity and farmers and collectors eager to join the scheme. In order to do so, however, producers must adhere to procurement regulations, and follow training and guidelines for organic production and the sustainable management of these resources in the wild (Figure 6). Similar linkages were fostered in the other three BFN countries, increasing structured demand for African leafy vegetables in Kenya, for WFPs in Turkey and for native fruits such as wood apple (*Limonia acidissima*) in Sri Lanka including via private sector linkages (see next section).

**Figure 6.** Best practices for the sustainable harvesting and management of pequi (*Caryocar brasiliense* Cambess), common to Brazil's Cerrado region. The guidelines are produced by the Ministry of Agriculture, Livestock and Supply (MAPA) in support of producers/extractivists intending to take part in the public procurement schemes.

#### *4.6. Raising Public Awareness of the Importance of WFPs*

Raising public awareness of the important contribution WFPs can make to diets and livelihoods is another effective way to secure research and policy investments targeting their conservation and use and creating a mutually reinforcing virtuous cycle [163]. This is probably the area in which countries invested in protecting WFPs let lose their creativity and excel in finding ingenious, innovative, and culturally acceptable ways of communicating the importance of WFPs to different target groups. Naturally, collaborating and partnering with the broadest range of stakeholders, e.g., farmer groups, NGOs, private sector enterprises, schools, the media, and ministries will ensure that there is clear and cohesive messaging that is able to reach the widest possible audience.

#### 4.6.1. Youth

As future consumers and protectors of biodiversity, youth are an important target audience for WFP messaging. Awareness raising campaigns can take advantage of recurring activities such as biodiversity festivals or food fairs to organize nature walks or competitions for younger participants [28,150] or join forces with relevant ministries (e.g., Environment, Agriculture, and Education) to introduce messaging around WFP conservation and use in curricular activities and courses (Figure 7). For older students, WFPs offer an interesting opportunity for "greening" vocational training, particularly in the food and beverage sector. In Turkey, to raise the profile of WFPs, the BFN project partnered with the Halim Foçali Vocational School organizing a series of lectures and hands-on activities for 16 student chefs, who were trained to recognize and collect local edible species and use them in their cooking classes. Future plans for the institute include the establishment of an herb garden on the school premises where WFPs will be grown and harvested for use in cooking courses. Interest in the program from the National Education Directorate of Foça has led to plans for extending the training to other schools and officially include traditional WFPs in the school curriculum [150]. School gardens are also an effective way of promoting greater interest in biodiversity and can act as important conduit for improving nutrition, well-being and education of schoolchildren and their families [164], as well as acting as conservation networks for tree genetic resources [165] and reviving traditional food systems and culture [164].

**Figure 7.** A nutrition education booklet from Ecuador that includes WFPs as a food group. On the left, the cover depicts the forest as an alternative source of foods, mainly fruits, while on the right, five food groups are shown along with a list of 13 WFPs (mostly aromatic plants) used for preparing hot beverages.

#### 4.6.2. Communities and Households

As previously mentioned, it will be important to ensure that the main users of this diversity the communities that continue collecting and maintaining WFPs are aware of the species' nutritional and sociocultural importance. Seasonal food availability booklets and calendars, such as the one shown in Figure 5, and simple, locally appropriate picture posters (Figure 8) can serve the dual purpose of revitalizing the use of WFPs and imparting basic nutrition information derived from national nutrition guidelines. Translated into local languages, these tools can be used by government extension workers and NGO practitioners to offer an overview of local diet quality and consumption patterns derived from baseline assessments and provide recommendations on how WFPs and other local agrobiodiversity can fill existing nutrient gaps. To avoid issues of overharvesting, it will be important, that the above

information is complemented by easy-to-understand guidelines on the sustainable collection and management of these species, as shown by the informative brochures that accompanied the revival of WFPs in Turkey (Figure 9).

**Figure 8.** Community poster in local language developed in West Sumatra, Indonesia, as part of the "Food, Agrobiodiversity, and Diet project" explaining the health benefits of local food plants that are rich in protein, vitamin A, vitamin C, and iron, and Mandailing children learning about local foods. Credit: Lukas Pawera.

**Figure 9.** Foragers' guide to edible wild plants and illustrations taken from "A children's guide to the collection of wild edibles", produced by BFN Turkey to complement activities aimed at raising the profile of Turkish WFPs. Credit: BFN Turkey.

#### 4.6.3. Policymakers

Policymakers and key change agents who can support the conservation and use of WFPs are to be found within the following sectors: nutrition, health, agriculture, forestry, education, environment, trade, planning, poverty reduction, food security, rural development, economy, and finance at national, regional, and international levels. Whatever their background, for effective decision-making to occur, policymakers need access to timely, independent and reliable information, in a simple and useful form, accompanied by the cost implications of the research, indicating whether it is feasible and affordable [166]. As demonstrated by the endorsement of the "Sociobiodiversity Ordinance" in Brazil, for example, nutrition evidence generated via food composition analysis was critical for expanding the list of sociobiodiversity species to include previously neglected WFPs, and for subsequent policy uptake by national programs dealing with food and nutrition security.

The recognition of WFPs as important elements of healthy diets and rural resilience has thus resulted in increased federal funding (approximately US\$6 Million per year) for public procurement programs to purchase sociobiodiversity products directly from family farmers and provided an indication of the untapped market potential of WFPs in institutional markets [167]. The increased appreciation of the role WFPs play in rural diets is also leading to investigations into the affordability of diets that include WFPs [126]. As mentioned earlier, the study carried out in Eastern Baringo, Kenya, has shown that wild plant species, especially vegetables, are able to significantly reduce (by 30–70%) the cost of nutritious meal for vulnerable groups [54]. The tool, which provides an insight into the affordability of nutritious foods, offers a useful entry point for policy discussion around the types of commodities and delivery channels that are likely to achieve nutritional outcomes particularly for the most vulnerable segments of the population [168].

#### 4.6.4. Broader Audiences

Recent interest in food and gastronomy programs worldwide has acted as the perfect jumping board for WFPs, particularly in developing countries. Many of the approaches adopted by BFN project countries have extensively been described [28,150,153], and broadly involve communities partnering with celebrity chefs, gastronomists, or taking advantage of existing food festivals to organize information and hands-on events onWFP collection, transformation, and cooking (Figure 10). Innovative approaches for reaching out to broader audiences are described in detail by Gee and Lee (2020) who look at emerging youth-led innovations that can be productively applied to the conservation and sustainable use of food biodiversity, including WFPs. The realms of social media and mobile technology are rapidly evolving, and via mobile apps consumers are now able to (i) find local crops in season and plan grocery purchases, (ii) identify plants through a global photo database, (iii) learn about wild edible plants (Wild Edibles and Foraging Flashcard Lite), (iv) and even trace fresh crops back to farms using blockchain technology. On the production side, a growing number of applications, including in developing countries, offer "smart phone farmers" unprecedented access to crop, field, and market information, which could easily be extended to incorporate WFPs. Gee and Lee [169] also explore the benefits of creating conservation networks for biodiversity through international movements such as via "Campesina" and Slow Food, which can connect different actors who are motivated to improve global and community-based food systems using food biodiversity.

**Figure 10.** Front covers of recipe books developed as part of the WFP-focused projects in Brazil, Ecuador and Kenya. Credit: BFN Brazil, IKIAM and BFN Kenya.

#### **5. Conclusions**

While WFPs contribute to the diets and livelihoods of millions of people worldwide at the local level, there is still much that we do not yet fully understand about them and thus their role is not fully appreciated. This makes it a challenge when it comes to decisions and actions that might support more effective national and international conservation, sustainable management, and useful strategies for WFPs. Some of these actions are summarized in Table S2. While there are an increasing number of publications outlining the importance of WFPs, usually at a local level, there is largely a scarcity of data and information at a national level, and conservation assessments are still limited. This fails to convey the full contribution that WFPs make to food security and nutrition and the overall importance of these biological resources to national economies in many parts of the world. Furthermore, while we increasingly learn more about some of the threats which impact WFPs, we still know so little about their biology and ecology as well as the dynamics of their use and how climate change is impacting them now and in the future. The integrated conservation approach described in this paper is intended to guide stakeholders in creating plans and strategies to ensure that WFPs are used sustainably and are conserved for generations to come.

In this review we survey the contribution of WFPs to food security, nutrition, and livelihoods in a variety of geographical settings, many of which have benefited from the availability of donor-funded projects and therefore the dedicated attention of researchers and their organizations. It is by no means a comprehensive review. However, the limited cases and examples it highlights clearly demonstrate that the contribution of WFPs to food security, nutrition, and livelihoods is significant. With increased development attention and research investments, including a more effective enabling policy environment, the role of WFPs could be strengthened in the future.

A greater understanding and appreciation, especially by decision-makers, of the nutritional value of WFPs and their contribution to food security and nutrition could see the enhanced inclusion of WFPs in important national nutrition policy instruments such as dietary guidelines, development plans, or in nutrition education and school curricula. Greater use should also go hand in hand with increased research and investments targeting existing biological and ecological knowledge gaps on WFPs, such as plant demographic studies to calculate sustainable harvest levels in the wild or studies on seed biology and ecology to ensure they are adequately conserved ex situ. If WFPs were provided with greater policy recognition and support, especially through policy incentives and the development of innovative market-based demand approaches (with clear benefits arising to custodians), it would help create longer-term economic viability. This, in turn, could help greatly in better linking the conservation of WFPs and their sustainable traditional management and use, something which is currently missing in most national Plant Genetic Resources conservation strategies and action plans.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/9/10/1299/s1, Table S1: Plant families that include WFPs and semi-cultivated species that are known to contribute to food and nutrition security; Table S2: Summary of actions that can be undertaken across the four pillars by the main stakeholders involved in WFP conservation and use.

**Author Contributions:** Conceptualization, T.B. and D.H.; investigation, T.U., E.M., D.B., C.T., D.P., L.P., B.P., and A.T.; writing—original draft, T.B.; and writing—review and editing, T.B., B.P., T.U., E.M., D.H., D.B., C.T., D.P., L.P., A.T.; M.T., and J.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** Overall support for the Biodiversity for Food and Nutrition (BFN) Project was provided by the Global Environment Facility (GEF Project ID 3808). Co-funding and implementation support were received from the UN Environment Programme; the Food and Agriculture Organization of the United Nations; Bioversity International and the governments of Brazil, Kenya, Sri Lanka, and Turkey. Additional funding was received by the Australian Centre for International Agriculture Research for work in Kenya (HORT2014/100, GP2017/007 and GP2018/101), as well as the CGIAR Research Program on Agriculture for Nutrition and Health (A4NH). For the Food, Agrobiodiversity, and Diet (FAD) Project in West Sumatra, Indonesia, the authors would like to thank the Neys-van Hoogstraten Foundation (Project IN305), and ALFABET mobility for supporting Lukas Pawera under the Erasmus Mundus Action 2 Programme. For work in Kenya, the support of Biovision Foundation Switzerland and A4NH CRP are gratefully acknowledged. The "MGU—Useful Plants Project" was funded by MGU, a kind and generous philanthropist based in Spain. E.M. is supported by the Kew Future Leaders Fellowship—Diversity and Livelihoods, of the Royal Botanic Gardens, Kew, UK.

**Acknowledgments:** The authors would also like to gratefully acknowledge the many colleagues and institutions in participating countries that contributed to the work described in this paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**



#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Conservation of Wild Food Plants and Their Potential for Combatting Food Insecurity in Kenya as Exemplified by the Drylands of Kitui County**

**Fredrick Munyao Mutie 1,2,3, Peninah Cheptoo Rono 1,2,3, Vivian Kathambi 4, Guang-Wan Hu 1,2,\* and Qing-Feng Wang 1,2**


Received: 22 June 2020; Accepted: 8 August 2020; Published: 12 August 2020

**Abstract:** Wild food plants are important resources for people living in dry areas of Kenya. A botanical inventory of vascular plants of Kitui county was compiled from specimens collected during field investigations in Kitui county, at the East African (EA) herbarium and from literature reporting on plants of Kitui county. To obtain an inventory of wild edible plants found in Kitui county, literature reporting on wild edible plants of Kenya were searched and combined with the use reports obtained from field surveys in Kitui county. A total of 199 wild plants found in Kitui county have the potential of being utilized as foods in different ways. Plant species growing either as trees or shrubs (83 species) and herbs (36 species) are the dominant life forms while the best represented plant families are Leguminosae (25 species) and Malvaceae (17 species). Fruits (124 reports) and leaves (56 reports) are the common plant parts collected for food. Fruits (120 species) and vegetables (44 species) are the common wild food types in Kitui county. Further studies on species distribution are necessary to address conservation concerns that may threaten such plants.

**Keywords:** climate change; conservation; crop wild relatives (CWR); drylands; Kitui county; wild food plants

#### **1. Introduction**

Arid and semi-arid areas cover about 80% of the Kenyan land mass and are characterized by a hot and dry climate and soils of poor agricultural potential [1]. The lives of people living in dry areas of Kenya are thus constrained by frequent droughts, which in turn trigger further challenges such as poor grazing resources and poor water quality. This in turn results in poverty and human to human conflicts over scarce resources. These challenges coupled with poor veterinary services make pastoralism an increasingly difficult source of livelihood for dryland communities [2]. Climate manifestation in East Africa has also proved to be difficult where warming is likely to lead to dryness in some areas and a higher precipitation in others [3], meaning that local populations are forced to cope with climatic uncertainties [4,5]. Livelihood diversification is the main way to cope with drought [6]. There is evidence of livelihood diversification in dry areas of Kenya to cope with changing climatic conditions such as the extraction of gums and resins by some communities as a source of income [7,8] while wild plants are also reported as important sources of traditional foods [9–11].

Many foodstuffs consumed in tropical Africa are derived from wild plants [9]. Those plants are utilized in different ways such as fruits, vegetables, cereals, roots, and tubers [11]. Plant derived foods such as fruits and nuts provide many nutritional benefits to the body [12]. Fruits can improve the nutrition of poor people who may suffer from deficiencies in vitamins, minerals, and other macronutrients. Many fruits are also important sources of vitamins A and C which may be lacking in the diet. For example, vitamin C which is found in significant quantities in many fruits is essential for protecting body cells and improves the adsorption of nonheme iron from plant-based foods. As a result of low intake of vitamin A, an estimated 50 million children in Africa are at risk of its deficiency, making it the third greatest health problem in the continent, preceded by malaria and HIV/AIDS [13]. Traditionally, children used to eat wild foods such as fruits and nuts during herding, which served them with nutritional benefits [12]. Although foods from wild plants may not at the present time form a major part of the diet of the local communities in Kenya as exemplified by the life of Dorobo people [14], traditional foods are still culturally accepted and are an integral part of the diet of local inhabitants [11]. For example, among the Dorobo people, plant species such as *Grewia tephrodermis*, *Vangueria madagascariensis*, *Vigna frutescens*, and *Vatovaea pseudolablab* are reported to have been served as staple foods for many years [14]. Furthermore, some traditional local vegetable species such as leaf amaranth are sold in the local markets in Kenya [15]. The local people have the knowledge on preparation and production of traditional foods, which require minimal additional inputs which are affordable to many, including the poor people [15]. Some wild foods also have medicinal properties to the human body and can be processed through various methods such as boiling, fermentation, and sun drying by the local people [11]. In some societies, some traditional foods from wild plants might be considered to be of no or low commercial value hence their collection is mostly meant for local consumption [11]. However, many wild indigenous fruits are sold locally in Kenyan markets such as fruits of *Adansonia digitata* which is also processed by coloring the seed pulp to make a snack. Its products are also in global demand for novel foods, pharmaceuticals, and cosmetics where the European Union, United States, Japan, and South Africa are reportedly potential markets [13].

The exploitation of native flora can be a buffer against periodical famines which are becoming prevalent in tropical areas [9]. About 60% of the Kenyan population face starvation due to lack of physical and economic access to adequate calories [16]. Kenya is endowed with diverse plant species which are estimated to comprise about 6293 indigenous vascular plants [17]. These include an estimated 800 food plants [10] some of which are underutilized food plants [16] such as *Amaranthus* spp. (leaf amaranth), *Solanum americanum* (African nightshade), *Cleome gynandra* (spider plant), *Cucumis dipsaceus* (Hedgehog cucumber), *Commelina forskaolei* (Rat's ear), and *Cucurbita* spp. (pumpkin leaves) which can all be utilized as green leafy vegetables [15]. Such plant species were relied upon in the past as sources of vitamins, minerals, and proteins by rural societies [16]. Despite their importance, ethnobotanical knowledge of traditional wild foods is declining in Kenya [10,11]. Women, children, and herders play important roles during collection of wild edible plants [14,18] and can therefore be considered as important custodians of such knowledge in Kenya. In northern Kenya for example, collection of gums is mostly done by married women in an effort to provide an additional income for their households [8], perhaps adopting the roles of single parenthood especially the widowed. In addition, some wild food plants considered to be of minor significance are gathered by little children and are at times used as diet supplements and emergency foods [9,10]. During collection of some wild edible plants in Kenya such as *Ficus* fruits (figs), *Vangueria* fruits, *Craibia laurentii* nuts, and *Maerua kirkii* nuts, children accompany their mothers to help in gathering while collection of some species such as tubers of *Cyphia glandulosa* is reportedly done by children as they go on with their duties [14]. Collection of wild edible plants is mostly done by poor and illiterate people, where such activities have perhaps been normalized as survival strategies during the dry periods of the year when there are insufficient resources available for human survival [8]. For example, among the hunter-gatherer communities in Kenya, wild foods may comprise the main diet of the day at certain times such as during famines [10].

Introduction of exotic vegetables has diverted the focus on indigenous vegetables in Kenya [16]. Recognizing the value of wild food plants can be useful in conservation of germplasm for the future generations [9] as well as buffering against famine in the changing climatic conditions [18]. The need for the recognition of the nutritional value of traditional foods has resulted in campaigns for them to be incorporated into the rural and urban diets [16]. In spite of this, few studies in Kenya [9,14,19–22] have focused on documentation of wild food plants at local levels. In Kitui county, studies have mostly focused on documentation of medicinal plants [23–25] with little attention given to wild edible plants. Local utilization and acceptance of underutilized vegetable species has been reported in Kitui county where cowpeas are the most popular vegetable species [15]. The overall aim of this study is to highlight the potential of wild edible plants in Kitui county as resources which can be utilized in combatting food insecurity and famine by the rural dryland communities. The study is based on the assumption that the plant species documented as wild food plants elsewhere in Kenya but currently not yet used as such in Kitui county, have the potential to also be adopted and utilized as food plants in this region as well.

#### *1.1. Study Area*

Kitui county is a tract of land located at 0◦10 S and 39◦0 E, between Athi and Tana rivers occupying an area of 30,496.4 km<sup>2</sup> [26,27] (Figure 1). The area is mainly inhabited by Kamba people while Tharaka people are found in the North of Tana River [26]. Kitui Kamba also interact with Oromo and Somali ethnic groups during droughts when the latter two move seeking pastures [2]. The area experiences infrequent rain and lacks permanent waters except in the Athi and Tana rivers, hence water scarcity is a major problem during the dry season. The area also lacks fertile soils hence chronic droughts and famines are major adversities to the people of Kitui [2,26,28]. As a result, the inhabitants rely heavily on forest resources especially in wetter zones near hills [2]. According to the 2009 national census, the population of Kitui county was 1,012,709 with 531,427 females and 481,282 males with a population density of 44 persons per square kilometer [27].

**Figure 1.** Map of Kenya showing the location of Kitui county.

Kitui county is largely a low plateau rising from 300 m above the sea level through various inselbergs reaching to an altitude of about 1638 m above the sea level [28]. The highest altitudes reach about 1800 m above the sea level. The climate of Kitui county varies from arid to semi-arid with a minimum mean annual temperature varying from 14 to 22 ◦C and a maximum mean annual temperature ranging from 26 to 34 ◦C. There are two rainy seasons where the long rains start from March and end in June while the short rains fall from October to December with a mean annual rainfall of 250–1050 mm [27]. Low rainfall amounts are experienced in the extremely hot lowlands while higher rainfall amounts are experienced on the hilltops [27–29]. As a result, the highlands are wetter and highly populated while the dry lowlands are sparsely populated [28]. Considering the amount of rainfall received in the drylands [20] and their elevational ranges above the sea level [30], Kitui county is a typical dryland region.

The vegetation of Kitui county is characterized by low, stunted, dense thorn bushes with thick undergrowth and occasional baobab trees. Much of the area lacks vegetation except on the hills [29] where scrublands and wooded bushlands are found [31] with *Drypetes*, *Combretum*, *Vepris*, and *Croton* as the dominant species [32]. In the dry areas, the dominant vegetation is composed of *Acacia* and *Commiphora* bushlands and woodlands. The vegetation on humid and cooler hills varies and mostly include *Terminalia brownii* and *Acacia polyacantha* while exotic trees such as *Grevillea robusta*, *Cupressus* spp., *Eucalyptus* spp., and *Pinus* spp. are planted on some slopes and mountains [28]. The uncultivated and intact lands are composed of dry bushes [26]. There are several hilltops containing a high diversity of plant and animal species [32]. Such highlands reportedly provide a link between coastal forests and the Kenya highland forests, resulting in the presence of unique species adapted to each individual highland [18,32]. The plant diversity in Kitui county is high and is used for traditional foods, teas, medicines among other uses by the local communities [18,28].

#### *1.2. Food Plants of Kitui County*

The people of Kitui practice mixed farming which involves growing a variety of crops and keeping livestock [18,28] where cattle are kept as a security against famine [33]. In 2009, 80% of the county population was reported to rely on agriculture for economic income [27]. Kitui county is one of the regions which has a high diversity of local foods in Kenya including cultivated food crops. Some vegetable species used by the local communities include African nightshade, cowpeas (*Vigna unguiculata*), *Commelina forskaolii*, leaf amaranth, spider plant, *Cucumis dipsaceus*, and pumpkin leaves [15]. The main food crops cultivated include millet, sorghum (*Sorghum bicolor*), lablab beans (*Lablab purpureus*), pigeon peas (*Cajanus cajan*), cowpeas, maize (*Zea mays*), and green grams (*Vigna* spp.) while mangoes (*Mangifera indica*) are among the cultivated fruits. In some cases, global vegetables such as tomatoes and other green leafy vegetables are also cultivated in addition to the aforementioned traditional vegetable species. Wild fruits eaten include *Adansonia digitata* (baobab), *Grewia villosa*, *Vitex doniana* (Black plum), *Lannea alata*, *Uvaria schefleri*, *Berchemia discolor*, *Azanza garckeana*, *Tamarindus indica* (Indian date)*, Vangueria madagascariensis* (Spanish tamarind), and *Cordia monoica* (Sandpaper saucer-berry) [2,15,18,28]. Goods sold in Kitui local markets during the colonial period include cultivated grain crops, sugarcane, and unspecified vegetables [34].

#### **2. Results and Discussion**

#### *2.1. Diversity of Edible Plants*

A total of 199 plant species in 52 families and 114 genera currently growing in Kitui county have been documented as wild food plants in different parts of Kenya (Table 1). Some of the common wild edible fruits reported during the field work are shown (Figure 2). Leguminosae is the best represented plant family (25 species in 13 genera) followed by Malvaceae (17 species in six genera). Previous studies have reported Leguminosae to be the largest plant family in the flora of various parts of Kitui county [25,32]. It is also the largest plant family in the flora of Kenya [17]. In addition, it has been recorded to comprise most of traditional food plants utilized elsewhere in Kenya [14]. Legumes are important food plants in poor rural African communities where they provide proteins, essential amino acids, macronutrients, minerals, and vitamins. African legumes are also tolerant to drought and are therefore strategic food sources in arid areas especially under the current climatic fluctuations. Despite this, African legumes are poorly studied, and some important economic species are still obtained from the wild [35].


**Table 1.** Number of genera and species of wild edible plants in Kitui county by family.

**Figure 2.** Some of the wild edible fruits encountered during field study in Kitui county: (**A**) *Balanites aegyptiaca* (L.) Delile (Zygophyllaceae); (**B**) *Tamarindus indica* L. (Leguminosae); (**C**) *Berchemia discolor* (Klotzsch) Hemsl. (Rhamnaceae); (**D**) *Vangueria madagascariensis* J.F.Gmel. (Rubiaceae); (**E**) *Cordia sinensis* Lam. (Boraginaceae); (**F**) *Grewia villosa* Willd. (Malvaceae); (**G**) *Commiphora edulis* (Klotzsch) Engl. (Burseraceae); (**H**) *Lannea schweinfurthii* Engl. (Anacardiaceae); (**I**) *Cynanchum hastifolium* K.Schum. (Apocynaceae); (**J**) *Uvaria sche*ffl*eri* Diels (Annonaceae).

#### *2.2. Growth Habit*

Wild edible plant species growing as either shrubs or trees are the dominant life forms (77) recorded followed by herbs (32) and trees (21) (Figure 3). A recent ethnobotanical survey in Kitui county reported shrubs and trees to be the frequent medicinal plants reported by herbalists [25]. The vegetation of Kitui is also characterized by bushlands and woodlands composed of low, stunted thorn bushes and under-growths [28,29]. Such vegetation types are likely to be dominated by shrubs or trees.

**Figure 3.** Growth habits of the wild edible plants in Kitui county.

#### *2.3. Plant Parts Used*

Fruits comprise the majority of plant parts utilized as food (124 reports), followed by leaves (56 reports) while roots and barks are also frequently reported. Other plant parts such as flowers and galls are sparingly reported. A single plant may have different parts collected for food; hence such species are represented by more than one report (Figure 4).

**Figure 4.** Number of species by parts of plants eaten (numbers represent reports per category).

#### *2.4. Food Types Obtained from Wild Edible Plants*

Foods obtained from wild edible plants reported in Kitui county fall into different categories, where the best represented food types are fruits (120 species), vegetables (44 species), and beverages (28 species) (Table 2).


**Table 2.** Classification of food types and the number of species in each type.

Fruits are reported to be eaten raw, cooked, or used in preparation of beverages such as wine and beer. They are also used as food additives such as flavoring agents in foods and soups or as fermenting agents in preparation of local brews. Fruits are among the frequently utilized wild edible plant parts in rural areas of Kenya [9,22]. Consumption of fresh fruits is beneficial to the body since they provide the body with resources such as mineral salts, vitamins A and C, carbohydrates, natural sugars, and water [12]. Some fruits are also consumed as snacks in some rural parts of Kenya [9] while in some regions, some wild fruits are considered to be of little nutritional value and therefore consumed as supplementary and emergency foods [9,28]. Utilization of wild fruits is at times constrained by some fruit plants being widely dispersed in their natural habitats making it difficult to gather enough while other plants produce small fruits which may also be unpalatable [9]. In Kitui county, some wild and cultivated fruits complement each other, where the ripening seasons alternate successively, maintaining a continued supply of fruits to the local communities. The dependence on wild fruits is reported to be higher in drier lowland areas of Kitui where cultivated fruit species are few [28]. Such areas also experience low amounts of rainfall [27–29], making wild fruits an important part of the local diet. In addition, fruits of *Adansonia adansonia*, *Vitex doniana*, *Azanza garckeana*, *Tamarindus indica*, and *Vangueria madagascariensis* are sold at the local markets of Kitui [18]. During the field study at Mutomo subcounty, fruits of *Berchemia discolor* were also reported to be collected and sold at Mutomo market.

Leaves are mostly utilized as green vegetables and as food additives in preparation of tisanes while in some cases, sour leaves are chewed raw. Germinating seeds are also eaten raw or cooked as vegetables [36]. Leafy vegetables are major contributors to local diets of rural populations and are also abundant in local markets. It is likely that they provide similar nutritional composition as cultivated vegetables such as vitamins and minerals and are also of medicinal value to the body [9,12,16]. In Kitui county, deficiencies of vitamin A and zinc are reportedly widespread [15] hence leafy vegetables can play an important role in the diet of the local inhabitants. An advantage of picking wild vegetables is that they provide an opportunity to pick a variety of different plant taxa which in turn offers a diversity in the dietary composition compared to cultivated green vegetables. A single diet of wild vegetables may comprise of different plant taxa thus ensuring maximum nutritional benefits to the body [21]. However, some vegetable plant species bear small leaves, while others are bitter. In addition, nutritional composition and palatability of vegetables vary with season [21]. Combination of such characters mean that wild vegetables require skills and time in their preparation which might result in their avoidance [16]. Similar to cultivated vegetables that are mixed together during times of scarcity, wild vegetables are also mixed to gather enough [9]. Wild vegetables are also mixed with cultivated vegetables to improve the taste [21]. Preparation of wild vegetables may involve boiling to wash them before cooking begins, probably as a way of dealing with bitterness in some vegetable species such as *Solanum americanum* which may contain toxic alkaloids. Bitter tasting or toxic populations of wild vegetables can also be avoided during the time of picking in the wild [9,21]. Leafy vegetables can be obtained from natural habitats such as forests and in disturbed places including farmlands [21]. Cowpeas are the main vegetables in Kitui county while other vegetables are underutilized [15] or used in the absence of cowpeas [28]. For example, *Commelina africana*, a wild vegetable which grows in the farm and in the wild sprouts earlier after the rains, providing an early vegetable before maturation of cowpeas [28]. Some species of wild vegetables such as *Amaranthus graecizans*, *Solanum americanum*, and *Cleome gynandra* occur naturally including in disturbed habitats although their occurrence depends on the right season which coincides with rains [9,21]. Other leafy vegetables such as *Oxygonium sinuatum*, *Commelina africana*, and *C. benghalensis* may occur as invasive weeds in cultivations [9]. Such adaptations to the local environments make indigenous vegetables suitable candidate species for combatting food insecurity by poor people living in dry areas of Kenya [16]. Some wild vegetable plants, which are utilized in other parts of Kenya grow naturally as weeds and may be underutilized by the local communities in Kitui county. Nutritional education and cooking demonstration of underutilized wild leafy vegetables was reported to result to an increase in their utilization in Kitui county [15]. According to Ichikawa et al. [37], major food plants might be shared between different communities while minor food plants may vary from one community to another. Local people are cautious with trying cultivation of vegetables they are not familiar with [15]. To enhance diversification of the ways of obtaining foods by the local communities, it is therefore important to create awareness on the utilization of local food plants not known or less prioritized by the local communities.

Exudates comprise of gums, resins, and wines tapped from plants. Gums and resins are produced by plants throughout the season including during the dry periods of the year. Although most of such exudates are collected from plants in arid and semi-arid areas that are of poor agricultural potential, only small quantities are meant for domestic consumption and much of the material is collected for sale [8]. However, gums and resins are still locally eaten during food scarcity and also have medicinal benefits [38]. Their collection is mostly carried out during the dry season by women and children in poor communities or by opportunists interested in income generation. Harvesting of gums and resins is a viable alternative for strengthening livelihood diversification in the drylands [38] especially during the dry periods of the year when other sources of livelihood such as dependence on livestock resources are constrained by insufficient pasture [8]. Despite the important role played by gums and resins in the lives of the local communities, collectors encounter various challenges such as poor harvesting methods, contamination of the collected materials, and improper post-harvest handling techniques resulting in overall reduction of the quality of the end products and hence low prices in the market. These coupled with the poor markets where the collectors mostly sell to the local shops and further complicated by the presence of local agents and opportunist buyers makes income generation from gums and resins unsustainable. The potential of gums and resins in alleviating poverty in dry areas of Kenya are hence underutilized [8]. Development of better markets for gums and resins would be an important step towards maximizing the benefits of such products to the local communities in dry areas of Kenya including in Kitui county.

Barks are reported to be used in preparation of tisanes or as food additives such as flavorings. Raw roots are eaten as starch foods, cooked as vegetables, or used as food additives. Roots and tubers are important sources of energy since they are rich in starch. Freshly harvested roots also contain a large water content [12]. Roots and stems from some plant species are sweet and succulent hence they are chewed raw to quench thirst. In some instances, roots serve as immediate sources of food especially during grazing when the herders have little or no time to cook. Such methods of utilizing wild plants are important attributes for people to survive in dry areas [20]. Some root tubers are cooked to reduce the poisonous compounds that may be present while others are prepared through drying and pounding before consumption [9,20,36]. For example, the roots of *Thilachium africanum* are poisonous but edible when cooked [39]. Stems of some plant species such as *Albizia amara* are used as food additives, which is boiled in soup and also used as a meat tenderizer. Other plant parts reported are flowers which are eaten raw or picked with leafy parts and prepared together as vegetables, while the internal parts of galls are eaten raw. Wild edible seeds reported include pulses (seeds from legumes), cereals (seeds from grasses), pseudo-cereals (non-grass seeds that serve a similar purpose as cereals), and other seeds which are prepared through boiling, roasting, or eaten raw. Cereals and pseudo-cereals are ground into flour which is made into other dishes such as porridges. Legume seeds are important sources of proteins, iron, niacin, and vitamins hence are used as meat substitutes while other seeds are sources unsaturated fats, vitamins, and minerals such as phosphorus, calcium, and fluorine [12].

Beverages such as beer, wine, and tisanes are also prepared from wild plant parts. Tisanes include both infusions and decoctions taken as beverages, bitter teas, teas with essential oils, stimulant teas, and medicinal teas [40]. Utilization of herbal teas is dated back to the medieval medicine when they were used for therapeutic purposes [41]. Therapeutic classification of herbal teas in Kenya was also done by Ichikawa [14] who referred to them as narcotics and herbal medicines. However, Maundu et al. [36] treated some of them as foods. In this study, infusions and decoctions prepared and taken in place of caffeinated drinks and herbal additives added into caffeinated drinks are categorized as tisanes. They are prepared from the leaves and barks while in some instances, seeds are used. Tisanes are mostly prepared from plant families (Verbanaceae, Lamiaceae, Rutaceae, Burseraceae, and Anacardiaceae) comprising of aromatic, glandular, or resinous and oil producing species [39,42,43] although some species in Leguminosae, Sapindaceae, and Rhamnaceae families are also used. According to Maundu et al. [36], plants with essential oils such as *Ocimum* species are used for flavoring tea. These plants serve both as flavoring agents and as substitutes for caffeinated teas. For example, during a field survey at Mutomo subcounty in Kitui, leaves of *Zanthoxylum chalybeum* were reported by the local residents to add a good flavor to caffeinated teas while the bark of *Acacia nilotica* was said to be used as a substitute for caffeinated teas. Stem bark of *Acacia nilotica* is also reported to be a stimulant [44]. Further studies are needed to determine the role of the reported plant species in preparation of tisanes since such preparations may be categorized as flavorings, teas, coffee substitutes, and as herbal medicines. Some plant species are used in making beverages such as herbal beer and wine. These include the fruits of *Hyphaene compressa* which contain a liquid that is brewed into beer and the fruits of *Cordia sinensis* and *Balanites rotundifolia* which are used in preparation of local brews [36]. Palm exudate liquid, tapped from the vascular bundles of *Phoenix reclinata*, is also drunk as wine [45].

Food additives are also obtained from edible plants. These include spices, herbs, and seasonings which are of small nutritive value hence consumed in small amounts to stimulate appetite by enhancing flavor [12].

An inventory of the wild edible plants in Kitui county is provided in Table 3. Those plant species which at present are already used as wild food plants are marked with an asterisk (\*) and a number sign (#).

**Table 3.** An inventory of wild edible plant species occurring in Kitui county. Information given under 'presence in Kitui county' refers to voucher specimens collected during field work by the authors (designated as SAJIT-Mutie MU), specimens at the East African (EA) herbarium or in publications citing the presence of the species in Kitui county. The plant use information refers to records of use of the plants for anywhere in Kenya, not necessarily in Kitui county, unless the name of the species is accompanied by an Asterix (\*) (indicating the plant was cited during a field survey as edible) or by a number sign (#) indicating the plant use in Kitui county was obtained from literature.



#### **Table 3.** *Cont*.


#### **Table 3.** *Cont*.


**Table 3.** *Cont*.


**Table 3.** *Cont*.


**Table 3.** *Cont*.


**Table 3.** *Cont*.


**Table 3.** *Cont*.


**Table 3.** *Cont*.


#### **Table 3.** *Cont*.


**Table 3.** *Cont*.

#### *2.5. Potential of Crop Wild Relatives (CWR) in Kitui County*

Crop wild relatives (CWR) form an important part of gene pool for the improvement of cultivated crops [55]. The genetic relationship between many of the tropical CWR and the cultivated crops is unknown [56]. In Kenya for example, wild sorghum populations are reportedly widespread in various habitats such as in protected areas, roadsides, and farmlands. Such resources are regarded as weeds in farmers' fields and are facing the risk of genetic contamination through pollen-mediated crop-wild introgression [57]. The negligence of CWR and land races from the notion that they will remain to be readily available in the wild is causing their degradation [58]. Some of the wild plants utilized as wild foods in Kitui county that have cultivated relatives in the area include *Amaranthus* species such as *A. dubius* [15]. The Amaranthaceae family also exhibits the highest diversity of species used as traditional vegetables in Kenya [51], hence such group of plants form an important gene pool for future improvement of cultivated members. The leaves of *Vigna membranacea* (traditional vegetable) are reported to taste similar as cultivated *V. unguiculata*, a species composed of various subspecies and several cultivars in Kenya [36]. *Vigna unguiculata* is also the second most popular grain legume in Kenya after beans, and it is estimated that 85% of the area under its cultivation in Kenya lies in arid and semi-arid areas [51]. In Tharaka for example, an arid area adjacent to Kitui, cowpeas are cultivated by about 80% of the households [59]. Pigeon pea (*Cajanus cajan*) is also an important crop in dry areas although its diversity is limited to only one species [51,59]. *Cajanus cajan* is regarded as an indigenous

plant in Kenya [43], hence wild forms might form an important resource base for improvement of cultivated members especially in dry areas such as Kitui county. Other important cultivated plant species with wild forms in Kitui include *Lagenaria siceraria* and *Citrullus lanatus* [28,36,51]. *Solanum americanum* is also a vegetable species growing in the wild and cultivated in Kitui county [15]. Some of its wild forms are bitter tasting and hence avoided during vegetable collection [9,21]. Such forms might be neglected leading to their possible disappearance in the wild. Although the socio-economic importance of CWR is well known, their conservation has not been systematically addressed and their current extinction levels might result in serious social and economic problems if threats facing them are not adequately addressed [58]. In Kenya, the decline of plant genetic resources is at its peak following the effects of global warming, increased population, and desertification [51]. Conservation efforts of such critically important group of plants is therefore vital if they are to be relied upon in the future [60]. Since it is evident that drylands of Kenya harbor wild plants with a potential to combat food insecurity as exemplified by Kitui county, collection of CWR and other food plants' germplasm and its conservation are important steps towards ensuring maximum benefits from such resources.

#### *2.6. Conservation of Natural Habitats in Kitui County*

Availability of wild food plants depends on the ecology of a given area and the history of its deforestation [9]. In Kenya, there is an ongoing loss of wild food species and the traditional knowledge associated with them especially in areas of high agricultural potential, where much of the original vegetation has been cleared for agriculture and infrastructure [10]. Domestication of some wild vegetables is however reportedly ongoing in some regions where vegetable plants such as *Cleome gynandra*, once introduced continue to self-reseed in subsequent years [9] hence becoming a long-term source of leafy vegetables. Other vegetables species under domestication in Kenya include *Amaranthus* spp., *Solanum americamum*, *Basella alba*, and *Sesamum angustifolium* which may also be spared in the farmland during cultivation of weeds [10]. Many of the food plants occur in natural forests while some are preserved by the local inhabitants in their farmlands. In Kitui county, fruit plants such as *Tamarindus indica* and *Balanites aegyptiaca* are preserved in farmlands for their medicinal uses [25]. Wild fruit trees are also left standing when other plants are being cleared for farmlands or charcoal. Wild food plants of Kitui county are threatened by the local communities who cut them for charcoal, thus also leading to loss of indigenous knowledge associated with them [28]. According to Mutie et al. [25], some medicinal plants in Kitui county such as *Strychnos henningsii* and *Vepris simplicifolia* which are also reported as food plants are decreasing in the wild as a result of human activities. Wild food plants are most important to the communities who reside in dry areas, which are more vulnerable to droughts [9]. Such areas are mostly inhabited by pastoral groups whose major threat to plant diversity is overstocking [10]. Diversity of wild edible plants is also reported to be richer in savanna zones compared to other forests zones [14]. It is also in the drier regions where the vegetation has been conserved to the greatest extent in some regions of Kenya [9]. The hills of Kitui are perceived by the local people to harbor important medicinal and food plant species [18,23]. In addition, high plant diversity and species endemism are reported in the hills of Kitui [32]. Such hills are vital ecosystems for adaptation towards the changing climatic conditions through provision of important ecosystems goods such as wild foods [18]. Mutomo hill plant sanctuary, one of the hills in Kitui county has been recently reported as a potentially important area for conservation of medicinal plants [25]. Conservation of important plants including wild food plants in other hills of Kitui county needs assessment and prioritization through community awareness so as to ease pressure exerted on wild plant populations by the local communities.

#### **3. Materials and Methods**

A botanical inventory of vascular plants of Kitui county was first compiled from data collected during three different botanical surveys in various parts of Kitui county between May 2018 and February 2019 by the Sino-Africa Joint Investigation Team (SAJIT). These include an ethnobotanical survey of medicinal plants carried out in Mutomo subcounty [25], which included citations of wild edible plants by the respondents. Further floral surveys were carried out in Endau hills, Mutitu hills, and Mui basin where the local people cited the wild plants used as food whenever they encountered them. Where possible, specimens at the EA herbarium in Kenya were checked to obtain plant species previously collected from Kitui county. Voucher specimens reported in this study have been deposited at Hubei Institute of Botany (HIB) herbarium in China and at the EA.

The data was supplemented by other data obtained from various literature such as published articles, conference proceedings, botanical survey reports, and the monographs of the Flora of Tropical East Africa reporting on plants of Kitui county (voucher materials are represented by specimen numbers seen in literature or by references citing the presence of the reported plant species in Kitui county). This yielded a plant checklist totaling to 931 vascular plant species, the most comprehensive checklist of the region to date (unpublished results). To obtain an inventory of wild edible plants of Kitui county, literature reporting on wild edible plants of Kenya was searched from various sources and combined with the use reports obtained from field surveys. Data were searched using key words 'plants, flora, edible plants, wild fruits, fruits, livelihood diversification in drylands, vegetable plants, nuts and seeds, useful plants, edible tubers, wild teas.' The key words were combined with 'Kitui' and 'Kenya', each at a time in order to determine the area of data collection. To exhaust the information gathered, if a plant species was found to be edible in Kenya and not yet recorded in Kitui county, another search category was initiated ('Kitui county' plus 'scientific name the plant'). The indigenous plant species and their growth habits were determined based on the local monographs of Kenyan flora [39,43] and the monographs of the Flora of Tropical East Africa [42]. All plant name synonymies were resolved using The Plant List database (http://www.theplantlist.org/). The plant species were then classified into parts utilized for food and into different food categories according to Cook [61]. The data were entered and analyzed in Microsoft Excel 2016.

#### **4. Conclusions**

Although further circumspection is needed before the potential adoption of these plants for food, this study nevertheless presents wild plants as important sources of food for the local communities living in dry areas of Kenya such as Kitui county. Investigation of herbarium materials and further botanical surveys are still necessary to determine the undetected food plants. The local communities have incorporated some conservation measures in their farmlands. There is still a need to sensitize them further on the need of preserving natural habitats and involve them in collecting of germplasm of edible wild plants and their relatives for ex-situ conservation. Disturbed places play an important role in human nutrition through provision of green vegetables, some of which may occur as invasive weeds. Prioritizing on proper harvesting, storage, and marketing of wild foods produced seasonally in large quantities might be an important step in maximizing the nutritional benefits of dryland communities. Lastly, understanding the distribution of ethnobotanical knowledge among individuals and the role of age, gender, and the level of education are important factors in conservation of wild edible plants in dryland areas of Kenya in general.

**Author Contributions:** Investigation and writing: F.M.M.; Investigation: V.K. and P.C.R.; Conceptualization and supervision: G.-W.H. and Q.-F.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research project was supported by the National Natural Science Foundation of China, 31970211 and the Sino-Africa Joint Research Centre, CAS (SAJC201614).

**Acknowledgments:** The authors thank Kitui county government and the Kenya Forest Service for issuing the permits and to the National Museums of Kenya for allowing them to access the East African (EA) herbarium.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Main Challenges and Actions Needed to Improve Conservation and Sustainable Use of Our Crop Wild Relatives**

#### **Johannes M. M. Engels 1,\* and Imke Thormann <sup>2</sup>**


#### Received: 5 July 2020; Accepted: 27 July 2020; Published: 30 July 2020

**Abstract:** Crop wild relatives (CWR, plural CWRs) are those wild species that are regarded as the ancestors of our cultivated crops. It was only at the end of the last century that they were accorded a high priority for their conservation and, thus, for many genebanks, they are a new and somewhat unknown set of plant genetic resources for food and agriculture. After defining and characterizing CWR and their general threat status, providing an assessment of biological peculiarities of CWR with respect to conservation management, illustrating the need for prioritization and addressing the importance of data and information, we made a detailed assessment of specific aspects of CWRs of direct relevance for their conservation and use. This assessment was complemented by an overview of the current status of CWRs conservation and use, including facts and figures on the in situ conservation, on the ex situ conservation in genebanks and botanic gardens, as well as of the advantages of a combination of in situ and ex situ conservation, the so-called complementary conservation approach. In addition, a brief assessment of the situation with respect to the use of CWRs was made. From these assessments we derived the needs for action in order to achieve a more effective and efficient conservation and use, specifically with respect to the documentation of CWRs, their in situ and ex situ, as well as their complementarity conservation, and how synergies between these components can be obtained. The review was concluded with suggestions on how use can be strengthened, as well as the conservation system at large at the local, national, and regional/international level. Finally, based on the foregoing assessments, a number of recommendations were elaborated on how CWRs can be better conserved and used in order to exploit their potential benefits more effectively.

**Keywords:** crop wild relatives; biological features; conservation; use; local; national and global efforts; policy; genetic diversity; gene donors; pre-breeding; breeding; cross-sectoral collaboration

#### **1. Introduction**

Today's cultivated crop plants have undergone more or less drastic changes since their first cultivation and domestication. The first signs of domesticating wild plant (and animal) species date back 10,500 years in Western Asia and domestication has since then been practiced in different parts of the world by different groups of people on new species [1]. The duration and intensity of this domestication process have been very variable from one crop to the other [2]. The one thing that all crops have in common is that they originated from (one or more) wild and naturally occurring species. For a number of crops, the domestication process is well known, based on archaeological finds and (experimental) research. In general, this process started with gathering in particular wild grasses and leguminous species, followed by their cultivation closer to the homestead and gradually undergoing transformation from wild into domesticated taxa [3–5]. In some instances, crops are the result of natural or man-made hybrids between two wild ancestor species (e.g., banana: *Musa acuminata* and

*M. balbisiana*); in other cases, the wild relative is a subspecies of the cultivated crop (e.g., *Vitis vinifera*) or there is no difference between the wild and the domesticated species (e.g., the olive tree, *Olea europea* which has wild, weedy, and cultivated forms, and many forage crops), which are just two different forms of the same species. For other crops, the domestication process is much less known or even completely obscure, including which wild species might have been involved as ancestor(s) of the crop in question (e.g., *Triticum spelta*, spelt). For some crops, the domestication process is still ongoing, especially in local fruit trees [6]. Possibly the most important consequence of the domestication process is that the genetic diversity available in the crop genepool (in the narrow sense) is usually much smaller than that in the related wild species [7,8]. In this paper, we focus on the wild species that are related to our crops, i.e., the crop wild relatives (CWRs). They have in different ways contributed (genetically) to the domestication process and thus can be regarded as the ancestral species or progenitors of our present crops, and they are a valuable resource of genetic diversity and traits for plant breeding.

It has taken several years after the global initiation of systematic collecting and conserving threatened landraces of our crops, somewhere in the 1960/70's, until CWRs were systematically included, both at the national and international level. In 1975, a global collecting program of threatened landraces and CWRs was initiated under the coordination of the International Board for Plant Genetic Resources (IBPGR) and approximately 220,000 samples were collected during more than 1000 collecting missions in more than 130 countries, largely before 1995. The collected materials were sent to and subsequently stored in selected national and regional/international genebanks around the world [9,10]. The inclusion of CWRs in collecting efforts was triggered by the observed genetic erosion, as well as by the apparent need to include more genetic diversity for the advancement of breeding programs of major crops (e.g., potato), triggered by the success of using CWRs in breeding programs, such as the tomato, for specific traits [11]. Due to breeding programs in need for more diversity, the first 'push' for CWR conservation came from the international CGIAR research centers, as well as some (international) breeding companies in the 1970/80's [12].

Only during the past few decades, significant successes of transferring traits from CWRs into cultivated crops have been reported, mostly to overcome biotic stresses, such as pests and diseases, as well as abiotic stresses, such as drought tolerance [8,13]. More recently, adaptability to changing environmental conditions, in particular those caused by climate change, has also become important. Only gradually, CWRs became a priority for the more advanced national plant genetic resources centers for food and agriculture (PGRFA), such as in the USA, UK, Germany, The Netherlands, and Australia. Possibly the biggest 'push' for the conservation of CWRs was the advancement of molecular biology and genetic tools and techniques that greatly facilitate the transfer of traits, genes, and alleles from one species to another, almost independent of how closely they are related to each other.

The above-mentioned developments certainly had an important impact on the increasing (political) conservation priorities accorded to CWRs since the late 1980's/early 1990's. This has been reflected by the inclusion of CWRs in the text of the Convention on Biological Diversity (CBD) [14] and, in 2010, in the AICHI Biodiversity Targets, in particular Target 13, as well as in target 9, of its Global Strategy for Plant Conservation, where CWRs and wild food plants were accorded a high priority for conservation [15]. In almost half of the 18 priority activities of the Second Global Plan of Action (GPA II), adopted in 2011 by the Food and Agricultural Organization of the United Nations (FAO) Member Countries, it makes (again, like in the first GPA agreed upon in 1996) a special reference to CWRs and wild food plants, highlighting the need to strengthen their conservation and sustainable use [16]. More recently, CWRs have been included in the United Nations' Sustainable Development Goals (SDG) [17]. The recent Global Assessment Report on Biodiversity and Ecosystem Services, published in 2019 by the United Nations' Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) [18], mentions CWRs explicitly as species that are important for long-term food security, helping render ecosystems more resilient to stressors including climate change, pests and pathogens, and that lack effective protection. The report highlights the decreasing number

of CWRs and mentions that many hotspots of agrobiodiversity and CWRs are under threat or not formally protected.

In response to this increasing visibility and importance of CWRs in global and international political agendas since the early 1990's, numerous projects, tools, and guidelines have been initiated and developed at local/national, regional, and global levels. Examples for the latter are the voluntary guidelines for the conservation of CWRs and wild food plants at the national level [19] or the interactive toolkit for CWR conservation planning [20].

Besides the more political framework facilitating conservation, technical and managerial considerations are also important in order to effectively include CWR species in routine conservation programs. As treated in the following sections, a number of specific requirements can be identified that determine the ability of genebanks, in particular, to cope more effectively with CWR conservation. Especially, the availability of adequate knowledge and experience to manage this very variable and sometimes extremely difficult category of genetic resources is one of the main hurdles to overcome.

It has been a long and is yet a continuous struggle to get CWRs as a high priority on, in particular, local and national conservation agendas [21,22]. Reasons for this are limited financial resources available to many conservation and use programs; the lack of technological resources to effectively exploit these resources; an increasing debate on access to and availability of PGRFA; the sometimes severe technical challenges, which the conservation of CWRs' can cause to genebanks, due to biological peculiarities of CWRs; as well as the relatively low priorities these resources have for local people. Against this backdrop, the paper investigates the reasons for these constraints, focusing on difficulties, opportunities and synergies that characterize the conservation and use of CWRs. Furthermore, due to the biological peculiarities of CWRs, there is a need for a strong collaboration between actors operating at different levels, especially between local/national and international, as well as between different sectors, such as agriculture and environment.

#### **2. Definition and Classification of CWRs**

A 'simple' and broad definition of a CWR is that all wild species belonging to the same genus (and that coincides in most cases with the same genepool) of a given crop are treated as a crop wild relative [23]. A narrower definition refers to the genepool concept developed by Harlan and de Wet [24]. They used the easiness of crossing a given wild relative with the crop species in question as the basis for their classification. When a CWR species crosses easily with the related crop, the species is defined as a genepool I species (GP1a = cultivated form of the crop and GP1b = wild or weedy form of the crop). Wild relatives from whom genes can be transferred to the crop, but with difficulties using conventional breeding techniques, are included in genepool II. Those wild relatives that cannot be crossed with a given crop and where gene transfer is only possible using sophisticated techniques, such as embryo rescue, somatic fusion or genetic engineering, are defined as genepool III species. Although this classification is very 'utility driven' and from a plant breeding perspective, it makes good practical sense, as crossing barriers are a major limiting factor for the use of CWRs in conventional plant breeding.

However, for the majority of crop complexes, particularly those from tropical areas, too little information on crossability is available to use the genepool concept. Therefore, an alternative concept has been proposed by Maxted et al. [23], based on the existing taxonomic hierarchy to define to which of four recognized taxon groups a given species belongs. Taxon group TG1a corresponds to the crop, CWRs in TG1b correspond to the same species as the crop, CWRs of TG2 are in the same series or section as the crop, TG3 is the same subgenus as the crop, and CWRs of TG4 are those in the same genus. Thus, without detailed information on the reproductive isolation, this concept can be used to establish the degree of relationship between a CWR and a crop [23].

The number of CWR species account for about 21% of the world's flora [19,25], assuming that any species belonging to the same genus as a given crop is a CWR. On that basis, it has been estimated that there are 50,000 to 60,000 CWR and wild food plant species worldwide [19]. For Europe, Kell et al. [26]

found that 17,495 (8624 of them endemic), out of approximately 20,590 species, or 85% of the European flora, comprise crop and CWR species. Maxted et al. [23] argued that a more targeted list of globally important CWR species could be obtained by focusing on the crop genepool GP1b or on taxon groups TG1b and TG2, containing the closest CWR species. By applying this to genera that contain major and minor food crops, as defined by Groombridge and Jenkins [27], that the resulting 77 genera contain 10,739 CWR species that are congeneric to these genera, and of these 221 are very close wild relatives and 471 close wild relatives [25]. Thus, as a working estimate, there would be, globally, around 700 closely related CWR species (i.e., less than 0.26% of the world flora), which are of a high value in terms of their potential use in plant breeding programs and would deserve the highest priority to conserve the genetic diversity of major and minor food crops [21,28].

Vincent et al. [29] used the genepool and taxon group concepts to estimate CWR relatedness for 173 priority crops included in Groombridge and Jenkins [27] and the Annex 1 of the International Treaty for PGRFA. Additional taxa more remotely related to crops were added if they had useful traits for crop improvement. The inventory contains 1667 taxa, belonging to 1392 species in 108 genera and 37 families. It also includes ancillary data, such as their regional and national occurrence, seed storage behavior, and herbaria, housing major collections of CWRs. This inventory is available online as searchable resource, called the Harlan and de Wet inventory, and is actively maintained [30]. This list can be regarded as the most comprehensive one, based on clear criteria. A number of other global priority lists, typically developed in the context of specific projects, are less comprehensive, have less well defined or complex criteria, and have not been used as widely as the list by Vincent et al. [29]. Two African regional checklists [31,32] and several national checklists and inventories have also been developed and are available on the CWR global portal [33].

#### **3. General Threat Status of CWR**

Since the successes of the so-called Green Revolution in the sixties and seventies of the last century, with the breeding of high-yielding varieties of a number of important food crops worldwide, in particular by the CGIAR research centers, a vast replacement of traditional varieties of these crops by the newly bred varieties resulted in a significant loss of genetic diversity and triggered a systematic collecting and conservation of in particular landraces in the newly established genebanks. The Green Revolution also impacted on the agricultural production systems through the promotion of fertilizers and the use of pesticides, leading to a much more intensive agriculture. This development impacted also indirectly on CWRs, especially those that grew in cultivated fields, on field margins and along roadsides. Consequently, they were included in the global collecting efforts coordinated by IBPGR [9]. The authors reported that 25% of the collecting missions were dedicated to CWRs. About 60,000, or 27%, of the 220,000 collected samples were CWRs, mostly forages, including forage shrubs and trees (53.2%), followed by wild cereals (10.4%), wild legumes (9.4%), wild vegetables (7.6%), and wild root and tuber species (7.6%).

As for other wild plant species, the genetic diversity of CWRs continues to be eroded by global threats, such as: changing land use; climate change and natural calamities, becoming possibly the biggest threat through different specific impacts on CWRs; changes in agricultural practices; over-exploitation or excessive use; nitrogen deposition; and invasive species. Other factors include overgrazing and desertification; agricultural subsidies, such as that of biofuel crops, maize, and rubber; the development of aquaculture; reclamation of wasteland; pollution; and others [22].

Specific examples of global threats leading to genetic erosion of CWR species have been presented by [22,34] and [9]. The latter authors noted much fewer publications on genetic erosion of wild plants and CWRs, compared to those on crop species. Jarvis et al. [35] predicted a loss of almost half of the current geographic ranges of CWRs of peanuts in South America, cowpeas in Africa, and wild potatoes in Central and South America. They also projected that between 16% and 22% of these species would go extinct by 2055. Lira et al. [36] concluded from model studies in Mexico that eight of the wild Cucurbitaceae taxa will not survive under accepted climate change models. Erosion of traditional crops and their wild relatives is greatest in cereals, followed by vegetables, fruits, and nuts and food legumes [15]. As part of the GPA II implementation assessment for the period 2012–2014, 32 countries reported to FAO to have conducted more than 5200 PGRFA surveys, covering 1823 species (predominantly wild). Of these, 56.3% were rated as threatened, i.e., they were no longer cultivated or did no longer occur in situ in most of their previous areas of cultivation or occurrence [22].

The most commonly applied means of assessing threats to wild taxa are The IUCN Red List of Threatened Species criteria [37], including for CWRs [38]. Some countries, e.g., Germany, have their own system for assessing endangerment status at national level [39]. IUCN has started to place some focus on CWR threat assessments. Their Plants for People Initiative, for example, included the assessment of high priority CWRs. CWRs are flagged within the IUCN Species Information System. The IUCN Red List of Threatened Species version 2017-2 included 760 CWR assessments [40]. The IUCN Red List status was assessed for 572 CWR species in Europe, and 11.5% of these species were classified as threatened (categories 'vulnerable', 'endangered', or 'critically endangered') and 26 species were reported as 'near threatened' [41]. Bolivia established a red list of CWRs using the IUCN criteria [42]. Maxted et al. [28] reported that the loss of genetic diversity within CWR species is likely to be much greater than the loss of species. Most of the species that are able to survive the threats they are exposed to will suffer some genetic erosion or loss of genetic diversity. The increasing impact of climate change is likely to impose heavy selection pressure on CWR populations. This could easily lead to a loss of genetic diversity and, consequently, species may not be able to adapt as readily and quickly to changing conditions as before. Thus, this vital diversity that is required to underpin food security might not be any more available to breeders [28].

Genetic erosion occurs also in genebanks due to intercrossing with other accessions during regeneration, selection, genetic drift, and shift because of unsuitable growing conditions, loss of viability in storage, or also due to human errors during cultivation. As CWRs are difficult to grow, genebanks might tend to wait as long as possible with regenerating them and, thus, seeds might lose their viability and thus cause genetic erosion [43]. The lack of knowledge about the biology of CWR species, the absence of a good infrastructure for their cultivation, and other factors, such as adequate funding for conservation, might well contribute further to genetic erosion, in particular within accessions [9].

#### **4. An Assessment of Peculiarities of CWRs with Respect to Conservation Management**

#### *4.1. Biological Peculiarities*

CWR species possess characteristics that allow them to survive in nature. Such characteristics are, in many instances, not suitable for cultivation. As CWRs are most valued and valuable as reservoirs of new genetic diversity and traits required by plant breeders, this diversity is evolving in nature while being exposed and adapting to (changing) environmental conditions. Storage in a genebank would not allow such adaptation processes to take place while being conserved. This means that one has to consider where to conserve the CWR, i.e., in their natural habitat (i.e., in situ), in a genebank or botanic garden (i.e., ex situ), and/or a combination of the two. Both the GPA II [16] and the CBD [15] regard in situ conservation as the strategy of choice for CWRs, backed by ex situ.

With respect to in situ conservation, the obvious advantages compared to ex situ conservation are that CWRs can be conserved dynamically, providing for ongoing evolution and for a wider coverage of their genetic diversity. However, a number of preconditions to achieve this are presently not met, including lack of biological information on the species themselves, their taxonomy, distribution, and threat status.

With respect to ex situ conservation, one should realize that crop species have lost most or all of the 'wild' characteristics during the domestication process. Typical examples are shattering, day length sensitivity, variable and non-determined flowering period, fragile ears (in the case of cereals), etc., which CWRs do possess. Thus, their management in an ex situ condition might be very difficult and

requires ample experience. Many wild species have a limited distribution area, compared to most crops, and are an integral part of 'their' natural ecosystem. Their adaptability might be limited and, thus, also their ability to adapt to new environments (i.e., in particular, those of a genebank setting) might be low. Consequently, their optimum ecological conditions should be known when growing them outside their distribution area, in order to produce healthy and vigorous seeds/planting materials for subsequent storage. Furthermore, their biological reproduction 'system' should be known to ensure an effective reproduction, especially in case pollinators are required.

Storage behavior of CWR seeds might be unknown as seed biological aspects are unknown and, thus, require testing to ensure optimum storage conditions; standard viability seed testing methods might not function properly and/or more advanced viability tests might be used; collected seeds might be very variable in quality, i.e., not uniform in their maturation status and, thus, with variable longevity expectations; seeds might have dormancy and/or could possess hard seeds, whereas no treatments are (yet) known; typically only small samples have been collected and, thus, there is in general a need for (immediate) multiplication before storage; possible presence of pest and disease in or on the material (vegetative material, non-orthodox seeds, and/or orthodox seeds) could have implications for outgrowing in the field or greenhouse, for viability testing, storage, and distribution [44].

The lack of knowledge and information on the existence, distribution, and genetic diversity patterns of CWRs make their adequate collecting difficult. This includes the application of the best possible sampling strategy, including the number of plants per population (if there would be such an option to decide), the number of populations for a defined area, or even the entire distribution area of a given CWR, the right timing of the collecting mission, etc. (for details of these and other collecting aspects see [45–47]). This general lack of information is certainly one of the main reasons why CWR genetic diversity is sub-optimally represented in ex situ collections.

Notwithstanding the high importance accorded to in situ conservation of CWR, in particular in protected areas [21], the effectiveness is reported to be more uncertain than in genebanks. At the same time it should be noted that the main rationale for in situ conservation is based on the likelihood that continued exposure to changing selective forces will generate and favor new genetic variation and, thus, there is an increased chance that rare alleles that may be of value to future agriculture are maintained [48].

In addition, considering the rather huge numbers of CWR species reported (50,000–60,000 species), the need to conserve adequate representation of selected populations for each CWR species is creating big challenges for an efficient conservation of CWR diversity [28].

#### *4.2. Managerial Responsibility- and Awareness-Related Issues*

It should be realized when establishing priorities for CWR conservation that their natural distribution does not follow, in most instances, national borders. Consequently, consultations with neighboring countries could facilitate comprehensive and effective conservation of the entire CWR genepool. In addition, information on the spread and possible distribution patterns of the genetic diversity within a given CWR genepool will be very helpful to identify possible sites for in situ conservation and/or to apply the most efficient sampling strategy when collecting.

According to the CBD, the CWR occurrences are under the sovereignty of the countries in which they grow. Therefore, in situ conservation of these species has, necessarily, to include a strong national component and any regional or global in situ conservation approach should be based on and/or aim to integrate or complement such national and local in situ actions. CWR in situ conservation cannot be centralized at national or international level, as is possible with ex situ conservation in genebanks.

According to FAO [21,22], in many countries, CWRs do 'fall between the cracks' of the responsibilities of the environmental and agricultural sectors. This makes it difficult to decide which organizational entity should be the 'logical' institution to assume the conservation responsibility in a given country. Constraints related to this decision are the fact that CWRs are still a not sufficiently known genetic resource, that they have been knowingly or unknowingly included in nature protection

measures without specific management or monitoring activities [28,48], and that they have been maintained by botanic gardens or genebanks without communication with other stakeholders.

Due to the disadvantaged position of CWRs compared to the domesticated genetic resources in most countries, the public awareness on CWRs is, in general, very low; there is no or only a weak political lobby within institutes and countries and, thus, a low priority to apply or provide funding for their conservation. Furthermore, there is a need for training and capacity building; skills such as taxonomy are limited and dwindling, creating dependencies on other organizations and countries. Especially in (remote) rural areas, there is a big need for better awareness and appreciation of CWRs, their diversity, and their role in breeding and adaptation to climate change for sustainable agriculture in order to stand any change of creating sustainable conservation initiatives.

The establishment and operation of in situ conservation sites can present administrative, logistical, and legal problems. For instance, CWR species that occur in 'disturbed' habitats, such as road-sides and field margins, as well as abandoned agricultural areas, will most likely not be 'included' in a protected area [28] and, thus, will require either their 'own' in situ conservation efforts, for instance, as part of an on-farm management scheme, and/or should be included in ex situ conservation. However, in many instances, their existence might not be known to the national PGRFA programs and/or the local authorities or conservation projects and, thus, are not on anybody's radar.

When considering the conservation of CWRs in protected areas, it should be noted that this type of in situ conservation is likely passive, meaning that CWR populations located in protected areas are not being actively managed and monitored, as most of the protected areas that harbor CWR species do not have specific CWR management plans [25]. Active and effective conservation of CWR populations located in protected areas could be achieved by expanding the management plans by including specific actions targeted to CWR [16]. Furthermore, climate change might lead to pronounced range contractions or range shifts for many CWRs. This led Aguirre-Gutiérrez et al. [49] to investigate the impact of climate change on CWRs and to combine this with monitoring programs, as well as collecting of CWRs for backing up in ex situ conditions. They conclude that in situ conservation measures, when ignoring the effects of climate change, will not be effective for many CWR species and that large-scale ex situ conservation actions are needed to safeguard CWRs.

CWRs can create problems for genebanks to manage them in routine operations, in particular, when specific required species information is lacking. For instance, to regenerate or multiply CWR accessions in the field or green or screen house, a genebank manager has to cultivate these wild species and, therefore, has to find answers to manage characteristics, such as a possible low germination rate, the unknown reproductive biology of the species, possibility of small sample sizes, shattering, non-homogenous ripening, etc., in order to meet the agreed standards for genebanks [50,51]. The lack of knowledge, experience, and facilities to adequately manage CWRs in genebanks is widely recognized. Thus, many genebanks will have to seek collaboration with other scientists in the country or with other genebanks that have more expertise in conserving CWRs. One option could be participation in a regional CWR network, through which the coordination of activities with neighboring countries could be achieved, sharing of responsibilities could be obtained, etc. The European Cooperative Program for Plant Genetic Resources (ECPGR) and its virtual European genebank, AEGIS, is an example of such a regional network [52]. At the same time, it should be noted that the conservation of CWRs only ex situ would not be feasible because of the sheer number of species and the need to sample and conserve eco-geographically and genetically diverse populations for each species in a dynamic way [28].

#### *4.3. The Need for Prioritization of CWR Taxa*

Considering the large numbers of species that are classified as CWRs, the usually limited financial resources for conservation, and the fact that many CWR species are not well known and in most cases lack critical information, there is a strong need to set clear priorities for their effective conservation. Possible prioritization criteria for CWRs should address aspects such as:

1. the degree of threat of the species;


These criteria are based on priority-setting criteria that have been used and reported in [53–56]. When countries need to prioritize CWR species they will select a number of these criteria in accordance with their national context. The choice and assigned importance of criteria are therefore likely to vary between countries, while the most commonly included criteria are the economic importance of the related crop, the genetic closeness to the crop, and the threat status of the CWR.

Whereas priority-setting is a 'standard requirement' in conservation, both for in situ as well as for ex situ approaches, there are some specific impediments to the prioritization process of CWRs. Possibly the most important factor is the lack of information/knowledge on the species themselves (see also the following section). Another important constraint is that CWRs are typically not 'directly' used and, thus, not part of a traditional 'food system' (and consequently of a traditional knowledge system) or of an agricultural production system and, thus, their intrinsic value is often not recognized.

#### *4.4. Availability of and Access to Data and Information*

Availability of and access to data and information about CWRs, i.e., their occurrences, distribution, and threat status, their taxonomy, biological characteristics, ecological requirements, habitats, uses and genetic and phenotypic characterization and evaluation, are essential for the planning and implementation of effective conservation and use of CWRs. Existing information is yet mostly scattered, held in different formats (including non-digital) by very disparate entities, many outside the PGR community, and often not readily available. In hardly any data source, CWRs are flagged or tagged as such. Accessing this information is, therefore, resource intensive and time consuming, even more so as comparing datasets is often very difficult due to the variety of standards, formats, and data management models used [26,57–59]. However, quite some progress in proposing descriptors and data collection formats has been made in the past few years, e.g., [26,60–64]. In addition, data are often incomplete and new and/or more data need to be generated or collected. For example, data about occurrences of CWR populations are usually derived from databases of ex situ genebank accessions and herbaria specimen records. These most often do not reflect a comprehensive picture of the species' distribution, can include very old records, and do not include data about the population status of the recorded occurrence. Field surveys and collecting require solid taxonomic knowledge of the local flora, which can be difficult to source. A global database or catalogue that collects into one place data about CWR inventories, occurrences, distribution, and in situ conservation actions currently does not exist.

#### **5. The Current CWR Conservation and Use Status**

#### *5.1. Facts and Figures on CWR Conservation*

#### 5.1.1. In Situ Conservation

Whereas the CBD [14], as well as the GPA [16], recognize the importance of CWR in situ conservation and regard ex situ conservation as a complementary conservation effort, the progress of CWR in situ conservation remains slow and difficult. In the second State of the World (SOW II) report, it is noted that in situ conservation is often envisaged to take place in protected areas or habitats and can be targeted at the species or at the ecosystem in which they occur [21]. However, the report also noted that in situ conservation of wild species of agricultural importance occurs mainly as an

unplanned result of efforts to protect particular habitats or charismatic species. Furthermore, existing in situ protected areas do not always meet the required management standards to maintain CWR populations and their genetic diversity long-term [65]. Whereas the number of protected areas globally has increased considerably and the total area covered by protection expanded from 13 in 1996 to 20.3 million square kilometers in 2020, covering 15.2% of the terrestrial surface [66], it should also be mentioned that, in general, areas with the greatest diversity, for instance within centers of origin and/or diversity of our crops, have received significantly less protection than the global average [21].

Several countries informed as part of the SOW II report [21] the establishment of protected areas for CWRs, e.g., Armenia (CWRs of cereals), Ethiopia (wild populations of *Co*ff*ea arabica*), Mexico (*Zea perennis* and *Z. diploperennis*, CWR species of maize), China (86 in situ conservation sites for CWRs of different crops), Turkey (protected areas for CWRs of cereals and legumes), and Syria (protected areas for CWRs of cereals, legumes, and fruit trees). Hunter and Heywood [55] reported the establishment of a citrus wild relatives' gene sanctuary in northeast India in 1981. A similar genetic reserve for wild relatives, including relatives of lychee, longan, and citrus, was established in Vietnam. They also mentioned that certain wild species of mangoes and other wild relatives are known to occur in biosphere reserves, national parks, and other reserves in India, Indonesia, Singapore, the Philippines, Thailand, and Sri Lanka, but little targeted in situ conservation has been undertaken. In Europe, the first CWR genetic reserves were designated in 2019, when, in Germany, a network of genetic reserves for four wild celery species was established [67–69]. As of February 2020, the network included 15 genetic reserves and more are in the process of being established.

The aforementioned summary assessment of GPA II [22] noted an increased attention to CWRs in the context of in situ conservation and management. Overall, 14.2% of the over 15,000 in situ conservation sites that were listed in 20 country reports had management plans addressing CWRs and wild food plants. A total of 78 activities on in situ conservation and management were implemented with institutional support in 19 countries. A total of 16 countries reported an estimated total of 2141 CWRs, including species from primary and secondary genepools, as well as species previously used for breeding but belonging to the tertiary genepools, and wild food plants, actively conserved in in situ areas. The average per country is amounting to 134 CWR species with a maximum of 840 species in one country. However, the overall developments, with respect to the implementation of the in situ conservation priority activities of GPA II, were limited in scope and the reporting countries rated their achievements with respect to this priority activity as the lowest across all the 18 priority areas that make up the Second GPA [22].

Vincent et al. [65] assessed 167 of the most important food crops for improving food security and income generation and identified 1425 priority CWR species related to these crops. They modeled the distributions of 791 of these priority CWRs as the basis for the identification of 150 sites for in situ conservation. Individual CWR species, in general, were found to be well represented in current protected areas; only 35 (2.5%) of the studied species, related to 28 crops, were distributed exclusively outside of protected areas. If a threshold of 50% or more of the potential genetic diversity of a CWR, based on ecogeographic land characterization diversity [70], occurring within protected areas, is considered adequate for genetic conservation, then 112 of the assessed CWR species are under-conserved, while 91% of CWRs are well represented within existing protected areas. Effectively conserving the top 10 CWR sites inside protected areas and the top 10 sites outside protected areas as defined in the pragmatic scenario, would only require active management of ~2000 km2 globally and would protect 475 CWR species, and 1257 unique CWR/adaptive scenario combinations. Vincent et al. [65] propose to manage these as a global in situ conservation network.

As any other wild species, most of the CWRs might not have any direct economic or nutritional relevance to local communities and, thus, might not be of interest to them. In fact, some of them might even be weedy and constitute a nuisance to local farmers. Therefore, CWRs might not be very attractive for inclusion in a local 'on farm' conservation program [15], and in case their distribution area is not part of a protected area setting, local communities will not be interested in participating in a

conservation activity if no benefits/funding will be provided. Only in cases where the CWR species occur in a protected area (targeted or 'by chance': [15]), their conservation might be easier and more sustainable as long as some sort of a monitoring system does exist.

In some cases, however, CWRs play a known and appreciated role in local and, typically, traditional cropping systems and, thus, will be valued by local farmers or communities. Consequently, conservation approaches might be easier and could directly involve the local people, as long as benefits will be generated through such activities. Examples of such situations include the regular re-domestication of *Dioscorea cayensis* subsp. *rotundata* in Benin; the use of *Dioscorea* spp. in West African countries by facilitating the introgression between wild and domesticated yams, as this is an important improvement strategy; the use of *Ensete ventricosum* in Ethiopia for regular incorporation of 'wild' seedlings into the fields of the cultivated crop; or the selection of wild walnut genotypes for cultivation in Kyrgyzstan [21]. From a crop evolutionary perspective and more related to traditional agricultural production systems, tolerating CWR species which are weeds, especially in field-borders, as pollinators of the cultivated material and, thus, assumingly increasing the genetic diversity of the crop for subsequent selection, is another example. However, also the opposite can be true that CWR play a detrimental role in farmers' field, for instance, as noxious weeds.

#### 5.1.2. Ex Situ Conservation

Traditionally, ex situ conservation is the main approach that countries have taken to conserve CWRs. Genebanks play an important role in the overall conservation of CWR germplasm; in fact, they (should) provide a link between in situ conservation and the users' communities at various levels. This role is essential as they typically are specialized in long-term conservation, distributing or exchanging requested materials, characterizing and evaluating the stored accessions, keeping detailed information records on the individual accessions and, in some instances, conducting pre-breeding activities to facilitate the use.

Genesys, the largest global database on ex situ conserved germplasm accessions, provided data (as of 11.01.2020) for 4,097,112 accessions, of which only 12% are classified as wild material, thus possibly also including some non-CWR species [71]. The European Search Catalogue for Plant Genetic Resources (EURISCO) [72] contains data for 2,023,530 accessions. Among those, 12.15% are reported as wild. According to Ford-Lloyd et al. [34], the 1095 CWR species reported in EURISCO, at the time the research was undertaken, only represented 6% of the 17,495 CWR species found in Europe. This means that 94% of European CWR species are not conserved in ex situ collections.

The SOW II report [21] provides an average percentage of wild species, predominantly CWRs, for each of the 11 major crop groups, varying from 4% (food legumes and fiber crops) to 35% (forages) and 46% (industrial and ornamental plants). The overall mean for the almost 7 million reported accessions of wild plants is 10%, most of them being CWRs.

For a number of reasons, many CWRs are represented by a small number of accessions per species in the collections, both in genebanks and in botanic gardens. As an example, of the 1076 global priority CWR taxa identified in a study about global CWR conservation priorities [73], 'only' 763 or 70.9% are included in genebanks; among those, 257 taxa are represented by less than 10 accessions each. Over 95% of the taxa examined were found to be insufficiently represented in genebank collections with respect to their full range of geographic and ecological variation in their native distribution area. In many instances one would find just few accessions per taxon, e.g., only 5.4% of the CWR taxa in EURISCO are represented by 10 or more accessions, whereas 90.5% of the CWR taxa have less than 5 accessions.

Due to the already mentioned difficulty to collect adequately sized numbers of seeds/plants per population, many of the accessions consist of (too) small quantities of seeds and are genetically poorly sampled [74]. In addition, the stored seed samples have frequently a low(er) viability due to the difficulties to grow them out for regeneration purposes [75]. Another aspect, related to lack of information/knowledge, concerns taxonomic identification of the CWR, including to which crop

genepool they belong. This will directly impact on the priority-setting and possible subsequent conservation, both in situ and ex situ, as well as on their use.

In a study of ex situ holdings of 23 selected genepools of the major crops included in Annex I of the International Treaty, i.e., those materials that countries agreed to form the backbone of the multilateral system of the International Treaty, the authors calculated an average non-weighted percentage of CWR accessions in genebank collections (without the international collections held by the CGIAR genebanks) of the selected genepool worldwide of 9.6%, ranging from 0% for coconuts (there are no CWRs known) to 33% for grass pea (a little bred crop) (Figure 1). The total global holdings considered in the study of the selected genepools (without the collections held by the CGIAR genebanks) were 3,149,371 accessions [76].

**Figure 1.** Percentages of CWR and landrace accessions in genebank collections of 23 selected crop genepools.

When looking at the primary genepool, 242 of the 1667 CWR taxa included in the Harlan and de Wet CWRs inventory were found to be under-represented in ex situ collections and the countries identified as the highest priority for further germplasm collecting are China, Mexico, and Brazil [29]. Khoury et al. [77] used gap analysis to assess the degree of representation of *Cucurbita* CWR taxa in conservation in situ, as well as ex situ in genebanks and botanic gardens. For the *Cucurbita*

genus, including 16 CWR and six cultivated species, the authors established detailed taxon-related ex situ, as well as in situ (i.e., protected areas) conservation priorities and suggested further in situ protected areas that would cover the greatest amount of populations of the largest number of taxa. Khoury et al. [77] concluded that 68.8% of wild *Cucurbita* taxa were assessed as high or medium priority for further collecting for ex situ conservation and 81.3% had a high or medium priority for further protection in situ, including all of the progenitors of the cultivated species. Furthermore, four taxa were listed as having very few accessions and, thus, very limited diversity is available for crop breeding. Khoury et al. [77] suggested that these figures might be considered as 'typical' for the CWRs at large.

Besides their conservation in situ and in genebanks, botanic gardens have also been collecting and storing CWR materials in their collections, as demonstrated by the PlantSearch database, which is an information platform for 1155 botanic gardens that collectively maintain plant, seed, or tissue collections of 589,526 taxa [78]. The database reveals that botanic gardens maintain at least 30% of all known plant species in their own collections, including that more than 41% of species assessed are globally threatened. Many of these wild species are CWRs. Almost one-third (315, or 28.6%) of the 1076 aforementioned global priority CWR taxa are maintained by botanic gardens [79].

A recent major effort of collecting new CWR samples was made by the project "Adapting Agriculture to Climate Change" [80], which focuses on the wild relatives of 29 crops included in Annex 1 of the International Treaty; over 4500 new CWR samples were collected for ex situ storage, evaluated for useful traits, and enhanced or pre-bred for use in crop improvement programs.

#### 5.1.3. Complementary Conservation

As already noted above, both the CBD [15] and GPA II [16] refer to the need to complement in situ conservation efforts with ex situ measures. Genebanks have recognized strengths in facilitating easy and targeted access to specific material (which is problematic for in situ conserved material) and to allow secure and long-term conservation as part of the conservation and use continuum. Especially when environmental change is too rapid for evolutionary change and adaptation, or migration, it can be easily understood how and why ex situ measures would complement or even replace in situ conservation and thus provide for the most effective approach [22,81]. Such a complementary approach requires that in situ and ex situ conservation measures have to be carefully planned and combined, thus securing a holistic combination of the two, which capitalizes on strengths and avoids weaknesses of one or the other. This will require a good understanding of the (seed) biology of the species, their threat status, priorities assigned to the individual CWR species, and other aspects; an assignment of clear responsibilities, including, for instance, to the agricultural and environmental sectors; if applicable, to link conservation and development; adequate and comprehensive information management; facilitation of adequate coordination with other stakeholders and countries; the verification of clear ownership rights over areas where the to-be-conserved CWRs occur; support of public awareness on the importance of CWR conservation; and, where necessary, to ensure the engagement of the broader public.

As an example, Hunter and Changtragoon [82] conclude, on the basis of regional project experiences, that for wild relatives of tropical fruit trees, any conservation strategy should contain elements of both in situ and ex situ conservation and should have a focus on conservation, both inside and outside protected areas. It should also ensure coordination of planning and implementation, institutionalize the practice of wild relative conservation, promote public awareness and understanding, create a suitable policy environment, and highlight the many benefits derived from their sustainable conservation and use. In situ approaches seem feasible for conserving wild relatives of tropical fruit trees, but experiences with targeted species and actions inside and outside protected areas appear to be relatively few. Consequently, wild relatives of tropical fruit trees remain a largely under-conserved natural resource, both ex situ and in situ, and are continuously under threat in their natural habitat from neglect and over-harvesting [82]. Vincent et al. [65] note the generally accepted requirement for complementary conservation, i.e., to also cover in situ conserved materials in genebanks, a process that has started recently. They further see a particular need to develop CWR in situ activities that

enable the conservation of geographically partitioned genetic diversity which retains potential for local environmental-evolutionary adaptation.

#### *5.2. Facts and Figures on CWR Use*

The term 'use' needs to be applied in its widest sense for CWRs. The traditional understanding is the use of genetic diversity in plant breeding by crossing cultivated material, usually advanced varieties with CWRs and through a strong selection to obtain genotypes, with the traits that have been transferred from the CWR species. Furthermore, CWRs are an important target for research on crop evolution and are, indirectly, an important component of research on the origin and spread of agriculture. With the increasing focus of conserving CWR in situ (including on-farm), the 'direct' use of CWRs by local communities and farmers has now also received some more attention. Another dimension of 'using' CWRs is their not well understood and accepted role in and contributions to the evolution of crops and plants at large. Through the overall conservation efforts of the flora (and fauna) in natural habitats and protected areas, of which CWRs are an integral part, they contribute to a healthier environment, healthy ecosystems, and the provision of ecosystem services. However, this latter aspect is not part of the focus of this paper. Furthermore, the appreciation of the economic value of CWRs and their contribution to the global economy is an aspect that would fall under the term 'use'.

In tropical zones, wild fruit harvested from forests contribute significantly to the total income and to sustainable nutritious diets of many rural households, apart from contributing substantially to important ecosystem services [29]. Wild relatives and wild-growing semi-domesticated species of tropical fruit trees also provide services to domesticated fruit trees in terms of resistance to extreme abiotic and biotic stresses through their high levels of genetic diversity [82].

More widely applied is the use of CWRs in pre-breeding and breeding programs and in research, in particular in countries with strong breeding companies, where facilities and technologies, as well as funding, are available to exploit these 'difficult' resources. Today, climate change is causing dramatic changes that are being experienced around the globe, especially global warming and the related increase of severe erratic weather conditions. These changes have a significant impact on agricultural production systems that need to be addressed as well. To allow crops to cope with and/or to adapt to more extreme weather conditions, including heat, drought, flooding, and increased salinity, there is a strong need for more genetic diversity than currently available for most crops from which plant breeders can select specific traits and resistance genes to 'equip' new varieties to cope with these changing conditions. In particular, the use of CWRs, as a known source of traits for introgression into the crops, has proven to offer such solutions, especially to overcome biotic stresses [8]. As CWRs do possess a much wider array of traits and allelic diversity, as well as 'new' genetic variation compared to our modern crops, they are an important asset to be included in the breeding pools of our plant breeders and, thus, to be accorded a high priority in their conservation and research and management activities that facilitate their use by plant breeders, worldwide [73,83,84].

'Historical' examples of CWRs in plant breeding include the use of wild *Aegilops*, *Secale*, *Haynaldia*, and *Agropyron* species in wheat breeding [85], the introduction of resistance to late blight, which is caused by *Phytophthora infestans* and is found in the wild potato *Solanum demissum* [86], as well as other disease resistances and tolerances from different potato CWRs [87]. Resistance against stem rust caused by *Puccinia graminis* subsp. *graminis* derived from the wild wheat *Aegilops tauschii* [88], in another example. In the early 1970's, resistance to the grassy stunt virus was found in wild *Oryza nivara* and now this gene can be found in almost all material bred by the International Rice Research Institute in the Philippines [34]. Maxted and Kell [25] reviewed the use of CWR in crop improvement in 291 papers reporting the identification and transfer of useful traits from 185 CWR taxa into 29 crop species. Wheat and rice accounted for almost 84% of the transfers and 56% of the inter-specific trait transfers related to pest and disease resistances.

The above historical examples demonstrate the past focus on trying to identify traits of interest through phenotypic characterization and evaluation [28]. Whereas the inclusion of genetic diversity from the wild genepool in breeding activities was difficult [21], the advancements in molecular genetics and the related tools allow a much more 'targeted' use of CWRs. Through the possibility of transferring specific parts of the genome, i.e., traits, genes, and/or alleles into the genetic background of improved breeding materials, the hesitation of using CWRs is fading and, thus, their importance for breeding is increasing. According to Ford-Lloyd et al. [34], genomic-based resources, map-based cloning, analysis of quantitative trait loci, gene isolation, and genetic modification are increasingly significant to exploit the potential of CWRs. Genomic databases containing information on genes associated with adaptive characters must increasingly be linked to web-enabled databases of ex situ conserved CWR germplasm, such as EURISCO [72]. Furthermore, predictive characterization, Focused Identification of Germplasm Strategy (FIGS) [28] and eco-geographical filtering method [89] are other promising approaches to facilitate the use of CWRs in breeding.

The number of CWR genomes sequenced has grown significantly over the past decade and in 2016 the number of crop genomes sequenced was 'only' about three times higher than that of sequenced CWR species, which were about 40 [90]. For example, Bertioli et al. [91] sequenced two wild peanut species (*Arachis ipaensis* and *A. duranensis*). Peanut is an important food source for many farmers in the developing world. The CWR genome sequences will provide breeders with new tools for enhancing the crop, and for developing new varieties more resistant to pests, diseases or with improved abiotic tolerance traits. It is hoped that this positive trend of more CWRs to be sequenced continues and thus, allows a better exploitation of the important traits that CWRs harbor, including quantitatively inherited traits.

A study carried out by PwC [92] assigned an indicative value of \$42 billion to the CWRs of 29 major food crops, with a potential to reach a value of \$120 billion in the future. All these 29 crops are included in Annex 1 of the International Treaty on PGRFA. Pimentel et al. [93] reported an estimated value of \$115 billion that CWRs contributed toward increased crop yields per year worldwide. In addition to their economic value, CWRs are also being valued for their not so well-known contributions to ecosystem services [34]. Tyack and Dempewolf [94] have reviewed past economic values of CWRs, including the previously cited studies, and propose an improved conceptual model for understanding the economic value of CWRs under climate change, expanding it from the focus of gross production to including a series of other values and costs.

#### **6. What Needs to Be Done to Conserve and Use CWRs More E**ff**ectively?**

From the information, facts and figures presented above, it is apparent that further concerted assessment and conservation efforts are required in order to keep these valuable resources and the traits therein available and accessible to the users, now and in the future. In this section, we summarize findings and identify actions for efficient conservation and sustainable use of priority CWRs. Important aspects that require attention to underpin the conservation efforts are presented.

#### *6.1. Documentation*

Documentation and availability of CWR data are the basis for the assessments of conservation and threat status, conservation planning, and monitoring, but are yet insufficient to provide more precise assessments and concrete figures about status and trends of CWR diversity. In recent years, tools and descriptors have been developed to support CWR data collection and management (see Supplementary Materials Table S1). The Secretariat of the International Treaty is currently developing a globally agreed descriptor list for CWR data exchange as a further step towards harmonizing CWR data recording and exchange and facilitating the development of national and global CWR databases. Based on these standards and tools, all relevant data at national level required for CWR conservation planning and management should be brought together in an accessible as well as standardized format into national CWR databases or portals. Furthermore, the development of a global CWR data portal,

analogue to Genesys, the global hub for ex situ data, should be considered. National CWR databases could then provide data to this global resource. Such a global portal would allow reaching a better understanding of global CWR distribution and conservation status. It would serve as an important tool for sharing information and supporting more effective planning, conservation, and monitoring at the national and international levels, as well as international collaboration in CWR conservation.

An increased recognition among the actors within the environmental sector responsible for nature protection and protected area management that CWRs constitute a group of very valuable PGRFA, would possibly support flagging and data recording in their respective databases and monitoring activities, and integration of CWR conservation aspects into existing nature protection networks and activities.

#### *6.2. In Situ Conservation*

As each country is responsible for the conservation of the natural resources within its territory, CWR conservation is logically and mainly addressed at national level. To secure these resources effectively and long-term, systematic and coordinated conservation is essential, as well as integrating in situ and ex situ measures. In most occasions, however, CWR in situ conservation has been carried out within the framework of projects, which are limited in time, hardly ever running for more than five years. A more stable organizational and financial basis for CWR conservation at the national level is therefore required in most countries. This can be supported and facilitated by developing a national strategic action plan for CWR conservation.

There is no single method for planning CWR conservation or for developing such a strategic plan, as related factors, such as financial and human resources, availability and quality of baseline data, the range, role and responsibility of relevant stakeholders, or the commitment of national governments, vary between countries. Nevertheless, a series of steps and decisions in the conservation planning process are likely to be common in most situations. These include the development of a CWR checklist, prioritization of CWRs, development of an inventory of the priority CWRs, threat assessments, gap and diversity analyses, and the identification of priority sites and actions for in situ and ex situ conservation [56].

The development of a national CWR organizational plan and an efficient coordination mechanism are important to facilitate coordination and collaboration. These measures require and will greatly benefit from the establishment or provision of a nation-wide information platform that facilitates the routine operations, allows the necessary coordination, and enables adequate reporting. The collaboration between the various important stakeholders at the local, provincial, and national levels is a prerequisite for effective and sustainable conservation operations. At the national level, adequate coordination between, in particular, the ministries of agriculture and environment and their implementation bodies is critically important to facilitate the identification and management of protected areas that target or include CWRs and to allow the participation of key stakeholders in the planning and implementation of projects and activities, including the support of research and awareness creation. Considering the specialized skills and facilities required for efficient and effective conservation of the CWR genepools, close collaboration with neighboring countries, possibly in the context of a regional network, seems to be very important to allow an adequate conservation of the total genetic diversity range of a given CWR species.

#### *6.3. Ex Situ Conservation*

Targeted and adequate collecting of highly threatened and prioritized CWR materials from their natural distribution areas, as well as of populations that are requested for research and use, is a critically important step to avoid genetic erosion and to facilitate use. A close collaboration with local communities and their conservation activities is important, as well as coordination with botanical gardens and other ex situ conservation programs. During collecting, it is important that an adequate number of populations of targeted CWR species is sampled and that the samples are of an adequate size. To ensure effective conservation for each collected CWR species, specific conservation standards need to be used; where necessary, further research might be required. One such research area is on seed biological aspects (see, for instance, [44,51] and/or the application of already developed advanced methods, e.g., on germination testing, using potential markers as volatile compound [95,96], changes in methylation [97,98], or DNA and RNA integrity [99,100]). The morphological and/or molecular characterization as well as further evaluation of conserved samples will be an essential step to facilitate their use, where applicable this should be done in collaboration with neighboring countries. One other example could be the application of cryopreservation of embryos, cells, tissues, or seeds as a long-term conservation method, especially for CWRs that cannot be conserved in the form of orthodox seeds.

A national CWR priority list provides the foundation for targeted collecting of threatened populations and for the development of complementary conservation efforts that reflect the long-term conservation needs, the biology of the species, the needs of users, accessibility to specific materials, and the requirement of exchanging/distributing germplasm. Well planned characterization and evaluation of prioritized accessions will increase our required knowledge and understanding of the genetic diversity aspects of the CWRs and thus enable and facilitate effective conservation as well as the targeted and sustainable use of conserved material.

#### *6.4. Complementary Conservation Approaches*

When planning CWR conservation approaches, a number of considerations will be important to take into account, especially when realizing that in general limited information is available about these resources. Furthermore, different infrastructures and technologies are needed to collect, conserve and monitor the material under conservation. In addition, geographical, technological, scientific as well as political/legal aspects will have to be considered and should complement each other well. As mentioned before, complementary conservation is not a 'method', but rather a conceptual framework that helps with the systematic planning of conservation efforts for a given species and under specific 'local conditions'. An example of such a framework is provided in [101]. So far, little practical experience can be reported. The approach should lead to practical and efficient, long-lasting, and cost-effective conservation activities for a given species. Examples of such pragmatic approaches would be to include populations of CWR species conserved in situ also in ex situ storage as a safety back-up and to facilitate their access for use. In case species cannot be (safely) conserved in situ, for instance, due to financial or administrative constraints or when the species is highly threatened, attempts should be made to conserve the threatened species ex situ in a genebank.

As use might be regarded as the ultimate goal of a conservation effort, it seems obvious to involve the users (primarily breeders) also in a prioritization and conservation planning exercise. Thus, the requirements of possible users of conserved germplasm can be duly reflected in the conservation approach, including specific aspects such as that the conserved materials can be shared easily with users in an appropriate form and quantity.

The very fact that only limited practical experience has been made with complementary conservation, the fact that the best possible combinations will vary from place to place and species to species, means that it will require more research to allow optimal solutions for effective and efficient conservation and sustainable use of individual CWR species to be identified. The development of a generic decision tree and supporting guidelines could be an important contribution to a more comprehensive, effective, and efficient complementary conservation of CWRs, at the various levels.

#### *6.5. Supporting Use*

Concerted efforts that facilitate the use of conserved CWR germplasm, either in in situ or ex situ conditions, are needed to enable a more effective and increased use of the often-unique genetic diversity contained in these threatened resources. Such efforts can be very diverse and include for example better management practices in a genebank or protected area, with respect to the representation of genetic diversity (as populations and/or as pure lines, etc.), ensuring an adequate coverage of the

genetic diversity that exists within a species in the collection, and very importantly increasing the level of characterization and evaluation of individual accessions (both morphological and molecular), providing much more information on the CWRs conserved in genebanks and improving the availability and accessibility of data.

#### *6.6. Strengthening the Conservation System*

In the context of this paper, the national approach is possibly the most relevant one, but with the clear understanding that the 'real action' will have to be undertaken 'on the ground' at the local level and, whenever possible, for both in situ and ex situ approaches. However, when considering the many difficulties to ensure an effective and secured conservation of these species, it is obvious that many of the less well-endowed local genebanks and botanic gardens will require support to implement such conservation activities adequately, in order to contribute to a sustainable and long-term safeguarding of CWR.

#### 6.6.1. National Level

There are a number of steps that need to be addressed at the national level to achieve effective, efficient and long-lasting conservation of CWR. The FAO published voluntary guidelines on the conservation of CWRs and wild food plants that provide an overview of all relevant steps that should be considered while planning and implementing conservation activities [19]. Some of these steps are mentioned in the following list:


A helpful website in preparing and implementing CWR checklists and inventories, as well as conservation strategies, might be the 'CWR Global Portal', established and updated by Bioversity International (now called the Alliance of Bioversity International and CIAT) [104]. It provides access to the Interactive toolkit for CWR conservation planning [56]. Guidelines and tools that can support national CWR documentation, prioritization, conservation planning, and implementation are summarized in Supplementary Materials Table S1.

A close collaboration between the national PGRFA program and those concerned with protected areas in a given country will be indispensable to avoid mistakes, to ensure that the best possible management approaches are being used, and that the existing strengths spread over people and institutions are being combined for successful implementation of in situ conservation. This collaboration can also address concerns that typically only a limited number of CWR species is included in protected areas.

#### 6.6.2. Local Level

The national CWR conservation approach will obviously have to address and include the local level actors' roles and responsibilities. However, often there is very limited published information on specific aspects at the local level that could be included in the planning and implementation processes [55]. A number of obvious aspects can be listed, including the involvement (and active engagement) of all relevant stakeholders in the preparation of management plans for target species. This is a crucial prerequisite when the CWRs are part of a protected area that can no longer be used, for instance, for collecting fresh fruits by the local communities in the neighborhood of such an area. Maxted and Kell [25] included the way to involve local communities in their report as a research question. They also propose an interesting approach in promoting CWR in situ conservation in less formally designated protected areas such as Indigenous and Community Conserved Areas (ICCAs). For the latter, see IUCN [105]. ICCAs are areas where indigenous peoples and local communities have conserved, for millennia, natural environments and species for economic (as well as cultural, spiritual, and aesthetic) reasons, independent of more formal conservation sector interventions. Brooks et al. [106] note that the establishment of genetic or other kinds of reserves for CWRs in areas not yet under protection in times of rapidly rising human population, climate change, and ecosystem instability is a complex goal, which necessitates a carefully researched strategic approach. Sites competing for reserve status would need to be assessed and prioritized for their longer-term sustainability, in terms of the predicted impact of climate change on the site and the economic development plans associated with local communities as well as at the national level [107].

#### 6.6.3. Global Level

Dilemmas with CWRs: Distribution areas of CWR species (at least those of the major food crops) in the tropics/subtropics are, to a large extent, located in countries with limited financial and/or technological resources, limited conservation programs, limited legal frameworks, few breeding program, and which can derive little direct benefits from CWR conservation (especially for local communities). In contrast, interest in these species is largely found in 'the North' where financial and technological resources are ample, knowledge is advanced, and where most of the breeding happens. Access to these species, however, is often limited and thus their use in breeding and research for global benefit difficult. Possibly, the only real solution would be to agree within the framework of the existing global instruments, in particular, the FAO Commission on Genetic Resources for Food and Agriculture and the International Treaty, to accord a high(er) priority to the conservation and sustainable use of these threatened resources, to study them more extensively, and to make the diversity freely available as foreseen by these instruments. A mechanism to enable the badly needed global coordination and facilitation of the frequently complex conservation activities, as well as to provide a platform for identifying and prioritizing research activities on CWRs, would be an important help in effectively and efficiently conserving and sustainably utilizing CWRs.

#### **7. Conclusions**

CWRs have been identified as threatened resources that are understudied, not properly conserved, and that possess a tremendous potential for the breeding of our crops. The latter is particularly important because of climate change, which calls for the urgent development of better adapted crops and varieties for the changing growing conditions in our vulnerable production systems. The protection of the environment is yet another important consideration that can be achieved, or at least important contributions can be made through the increase of crops and varieties that require less harmful inputs and provide still stable and high production levels.

In the above text, we distilled a number of actions that are recommended to be implemented at the various levels, whenever possible, in a timely and collaborative manner. Whereas a number of these recommendations can be implemented by individual countries, others will require agreement and coordination at the global level, where possible, using existing mechanisms and instruments.

#### *7.1. Documentation*


### *7.2. In Situ Conservation*


#### *7.3. Ex Situ Conservation*


#### *7.4. Complementary Conservation and Collaboration*


4. Facilitating/strengthening the collaboration between stakeholders for more effective and efficient conservation, research and use of CWRs as well as to facilitate the transfer of technologies at the local, national, regional, and global levels.

#### *7.5. Conservation System*


**Supplementary Materials:** The following are available online at http://www.mdpi.com/2223-7747/9/8/968/s1, Table S1: Guidelines and tools for CWR conservation, including references [108–112].

**Author Contributions:** Specifically, conceptualization, J.M.M.E.; methodology, J.M.M.E. and I.T.; formal analysis, J.M.M.E. and I.T.; investigation, J.M.M.E. and I.T.; writing-original draft preparation, J.M.M.E.; writing-review and editing, J.M.M.E. and I.T.; visualization, I.T.; supervision, J.M.M.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
