*1.1. Theoretical Background*

The Sustainable Development Goals (SDGs), the Aichi Biodiversity Targets (ABTs) and the 2 ◦C target of the Paris Climate Agreement all aim to secure human well-being in a sustainable world [1–3]. Even though human wellbeing essentially relies on the preservation of biodiversity to ensure ecosystem

functioning [4,5], current global conservation e fforts fail to halt biodiversity decline, which occurs at an unprecedented rate and will continue to do so due to ongoing as well as new threats [6–8]. This has extensive negative e ffects on economy and society due to ecosystem service losses [9] and clearly demonstrates that the key role of biodiversity for human well-being is not easily recognized by politicians aiming to achieve these global targets. Despite recent e fforts to uncover the synergies and trade-o ffs between the goals and targets of these conventions [10–13], there remains a high risk that nations will cherry-pick a few goals suiting their priorities and fail to tackle those harder to accomplish [14]. Similarly, the ABTs have been criticized for their conflicting interests and lack of indicators, particularly concerning the drivers of biodiversity loss [15,16]. While new indicators are constantly being developed and others improved or upgraded regarding their availability for more countries [17], essential knowledge gaps remain, particularly concerning data from developing countries [18]. Identifying key knowledge gaps is important for assessing biodiversity threats [19] as well as for improving monitoring, managemen<sup>t</sup> and investment strategies [20,21]. Approaching the end of the United Nations Decade on Biodiversity [22], we aim to identify key knowledge gaps concerning biodiversity loss in a particular priority conservation area as an example for many other priority conservation areas to be used by practitioners to streamline funds, resources and e fforts to tackle the ongoing biodiversity crisis while simultaneously secure livelihoods.

Most people's well-being predominantly depends on ecosystem services provided by terrestrial biodiversity. However, there are trade-o ffs between protection of life on land and human wellbeing. Strict area protection for conservation has frequently cut o ff the local poor from essential resources and thus led to famine and increased social inequality threatening their survival [23–25]. We therefore have focused our knowledge gap search on the protection of "Life on land" (SDG 15) and major trade-o ffs of this central goal with SDG 2 (zero hunger—directly connected to food security through agricultural land use), and SDG 10 (reduced inequalities) (Figure 1). The predominant trade-o ff between SDG 2 and SDG 15 is the expansion of agricultural areas to reduce hunger (SDG 2) resulting in competition for land with SDG 15 aiming to protect natural ecosystems like forests and their biodiversity [15]. Progress in achieving SDG 15 has often reduced that of SDG 10 [26–28]. Social and material inequality have been shown to be harmful to the environment and thus to people's health and well-being [29,30].

**Figure 1.** Trade-o ffs between Sustainable Development Goal (SDG) 2, SDG 10 and SDG 15.

#### *1.2. Regional Focus Area: the Dry Deciduous Forests of Western Madagascar*

As countries and regions di ffer in their demographics, geographies and governance [31], and are di fferently a ffected by climate change and biodiversity threats [32], we selected one regional example which promises high conservation payo ff (in terms of biodiversity protection): the dry deciduous forest in western Madagascar. Madagascar is one of many developing countries rich in biodiversity and natural resources, but economically highly disadvantaged [33]. It is one of the poorest countries on Earth, where malnutrition is prevalent and where about 80% of the population lives below the poverty line [33,34]. Despite a growth in GDP, poverty has been increasing [35] and over half of the population faces food insecurity [36]. Madagascar's population has increased by 400% over the last five decades to about 23.5 million [37,38]. This massive population growth, demanding an increasing share of land for agriculture, can be considered the major indirect driver of dry forest loss in western Madagascar [39]. However, political turmoil, insecurity and corruption are also major issues in Madagascar. Environmental crimes encompass illegal land clearance for large-scale agricultural expansion, illegal timber and wildlife trade, and mining [40]. Political crises occur on an almost decadal basis, preventing transparency and accountability of governmental actions [33,41,42]. Currently, Madagascar is listed at fifth position in the global ranking of increased risk of notable changes to the Global Peace Index [43], dropped severely in the Rule of Law Index [40], and only ranks 155th out of 180 nations in the International Corruption Perception Index [44]. Moreover, Madagascar ranks in the bottom 10 of 51 African countries assessed for their performance in reaching the SDGs [45]. Due to successful international awareness-raising of the uniqueness of and threats to Madagascar's biodiversity, the country has seen a quadrupling in protected areas since 2003, with about 10% of the country being protected for conservation [46,47]. However, like in many tropical countries, conservation has focused on rainforests in the east of Madagascar, while the diverse dry forests in the west and south have rather been neglected [39,48,49]. But even where protected areas had been declared, these largely failed to prevent forest loss and degradation [33,50,51] due to a lack of law enforcement and high levels of corruption [33,40,42,47]. At least 13 protected areas are considered mere "paper parks" as they are totally devoid of managemen<sup>t</sup> [47]. Generally, many knowledge gaps remain in Madagascar for achieving fair and equitable biodiversity conservation [52]. We chose the highly threatened dry deciduous forest of western Madagascar because forest losses predominantly caused by slash-and-burn agriculture and illegal logging are still dramatic there despite some national and international conservation attempts. Almost three quarters of the population live from subsistence farming practicing slash-and-burn agriculture [46,53] due to lack of alternatives but also because this farming practice is part of the people's cultural identity [54,55]. Slash-and-burn agriculture is conducted in two stages: during the dry season (June to September), woody undergrowth is cut and stacked around trees; at the beginning of the growth season (October) these piles of undergrowth are ignited, resulting in the destruction of all vegetation except for a few dead blackened tree trunks [56]. This form of agriculture requires little labor and—because of the ashes—requires no addition of nutrients for two to five years [57]. Afterwards, the land needs to be left fallow for several years (at least 20 years within the dry forest—see [58]). A reduced or lacking fallow period results in severe nutrient loss and too frequent burning favors the establishment of introduced and invasive species, preventing native species regeneration [58,59]. A growing need for agricultural land of the rapidly increasing population hinders sustainable managemen<sup>t</sup> that would allow for su fficient regeneration time of soils in agricultural fields. Thus, the rate of turning primary forests into agricultural fields by slash-and-burn practices to generate fertile farmland is increasing. About 40% of the forest have been lost since 1970 [39]. Fragmentation has been immense, so that few areas of primary forest larger than 800 ha remain – too small to contain viable populations of many species like larger lemurs [60]. Fragmentation also hinders animals to disperse and migrate to cope with climate change conditions [61,62]. The largest remaining area of dry deciduous forest occurs in central-western Madagascar [63]: Menabe Central ranks among the hottest biodiversity hotspots in the world [60], particularly due to its exceptionally high rate of endemism and intense anthropogenic threats [64]. While the total number of species is lower in the dry than in the humid forests, species richness is exceptionally high by global comparison with other dry forests [39]. Madagascar's dry forests harbor several locally endemic vertebrate species such as Madame Berthe's mouse lemur (*Microcebus berthae*), the giant jumping rat (*Hypogeomys antimena*), the narrow-striped mongoose (*Mungotictis decemlineata*), and plenty of other endangered species [65–68]. *M. berthae*, the smallest primate in the world, is particularly vulnerable to anthropogenic disturbances and restricted to core habitats of this biome [66,69,70]. Without immediate protection of its habitat, this species will likely be extinct by 2050 [63], just as many other vertebrate species of lemurs, rodents or tenrecs have already been driven to extinction [39]. This is of concern due to the roles these

species play for ecosystem functioning, its regeneration abilities and hence long-term persistence which for example depends on the seed dispersing role of lemurs [71]. Western Madagascar's dry deciduous forests are also considered of high conservation importance by a range of other approaches identifying priority conservation areas, for example the "Global 200" ecoregions [72] due to many endemic and endangered species and the Key Biodiversity Areas with a high biological value and intense anthropogenic pressure [73,74].

With respect to the people inhabiting this priority conservation area, social inequality is a huge issue hampering sustainable resource utilization as well as human well-being. Despite ambitious intentions, limited understanding of social–economic and ecological contexts prevented the realization of an e ffective protected area network [47]. For example, in spite of plans to involve communities in decision-making processes under the Durban Vision aiming to triple the amount of protected areas [47], conservation decisions have mostly been top-down enforced, sharply restricting or banning the use of local resources on which the local population relied [48,75]. This led to increased hunger, poverty and inequality of the already poor population as well as widespread criticism of the prioritization of the survival of lemurs over the survival of people [ibid.]. Due to funding, personnel and time restrictions of this project, we were unable to involve local community members in the collation and ranking of knowledge gaps in this study and therefore had to rely on predominantly academic experts for our knowledge gap ranking. To increase the chance of locals' needs to be considered in the ranking of knowledge gaps, we asked our participants whether they had work experience with local people.

A range of attempts have been made to identify knowledge gaps or important research questions, e.g., in the reports of the Intergovernmental Platform for Biodiversity and Ecosystem Services (IPBES) or by the yearly scan for the most important 100 ecological research questions at the time (see [21] at the global scale, and [76] at the national (UK) scale). However, acting usually occurs at the regional or local level and thus considering the specific local or regional conditions and needs of people at this level is important [77]. By choosing a regional example (i.e., the western dry deciduous forest of Madagascar) and conducting a consultation exercise using questionnaires with professionals from various fields (including subsistence farming and forest biodiversity, human development, governance, etc.), we followed an inclusive approach to "overcome the limitations of a consultation exercise of global aspirations" (Oldekop et al. 2016).

#### *1.3. Aims of Our Study*

We aimed to provide practitioners, politicians and funding bodies with a top 10 list of knowledge gaps concerning the slash-and-burn problematic in western Madagascar which need to be filled most urgently. This top 10 list is intended to be used as a guideline for decision making in research e fforts and funding distribution. We undertook the following steps to pinpoint these key knowledge gaps:


In this paper, we will present and discuss the expert-identified top 10 priority knowledge gaps, link them to the SDGs, ABTs and 2 ◦C target of the Paris Climate Agreement and examine the e ffects of experts' backgrounds on the ranking to check whether a particular type of characteristic (e.g., profession, age, gender) influenced the ranking scores.

#### **2. Materials and Methods**

#### *2.1. Knowledge Gap Identification at the Global Scale and Sorting into Categories*

In a first literature-based step, knowledge gaps related to the trade-o ffs between SDG 15 (life on land) and SDG 2 (zero hunger) as well as SDG 10 (reduced inequality) were identified at a global scale to spot knowledge gaps that are not directly addressed in the regional literature dealing with the slash-and-burn problematic in western Madagascar. We conducted a modified "snowball principle" based literature search using any paper which yielded relevant knowledge gaps to identify further relevant papers, as well as to check for subsequent citing publications [79,80]. Papers studied for knowledge gaps were restricted to those from the last 10 years, as recommended for a snowball literature search [81]. Further knowledge gaps were added by the authors based on their own professional experience.

Knowledge gaps relevant for the slash-and-burn problematic in dry deciduous forests of western Madagascar were listed (n = 71). Most of the literature sourced to select relevant knowledge gaps for this area concerned the core area of Menabe Central. Again, we used the snowball principle for our literature search as well as judgement of relevance by author LS who has extensive work experience in central western Madagascar. We separated these 71 knowledge gaps into ecological (n = 26), socio–economic (n = 24), and political categories (n = 14), because these categories relate to the di fferent spheres of sustainability (i.e., economic, social, environmental and political, see [82] and because it is necessary to identify solutions that have "traction in the social, economic, and political arenas in which conservation action must take place" [83]). In addition, we listed the category "general" (n = 7)—comprising knowledge gaps that apply to any of the other categories (see Questionnaire in the Supplementary Material S1).

#### *2.2. Identification of Key Knowledge Gaps through Expert Ranking*

Knowledge gaps relevant for the slash-and-burn problematic in western Madagascar were listed in a questionnaire (see S1) designed following the guidelines of McLa fferty [84]. To keep the questionnaire short, we combined some knowledge gaps which were closely linked. After discussion of the questionnaires with some experts of our focal area in Madagascar, the questionnaires were refined and then sent to representatives from universities and other research institutions as well as NGOs that have a long professional experience in Madagascar (via Email or LinkedIn). For time and logistic reasons, we could not include local people in the consultation process. However, to make sure that their perspective was represented, we have consulted experts that had been predominately working with rural people. We used snowball sampling [85] to recruit further experts by asking those contacted to spread the questionnaires further via their own personal networks.

The questionnaires consisted of two parts. Part I asked experts about their background (position/ occupation, institution/ organization, nationality, age) and Madagascar work experience, i.e., what type of work related to the slash-and-burn problematic they have done, for how long they have worked in this field, whether they have worked with locals and if so, in which context. We checked these background variables to check whether they qualified as experts and to determine potential bias in ranking scores. For example, people which feel strongly attached to a place, younger people, as well as women are often more environmentally concerned than those with little place attachment, older people, and men, respectively [86]. Age and work experience were divided into categories (age: <25 years, between 25–35 years, 36–45 years, 46–65 years and > 65 years; work experience: <1 year, 1–5 years, 6–10 years, >10 years). Additionally, participants could provide any other background information about themselves they considered relevant to define their expertise. In part II, experts

were introduced to the ranking scheme and then asked to rank the knowledge gaps by importance to pinpoint key knowledge gaps. The ranking categories reached from 0 (no priority) to 3 (extremely high priority). To accommodate for the lack of expert engagemen<sup>t</sup> in compiling the list of knowledge gaps, experts were given the opportunity to sugges<sup>t</sup> additional knowledge gaps which they considered more important than those they ranked as a knowledge gap of category 3. A copy of the full questionnaire, including the cover letter and a last page with options for comments, optional email provisioning to receive information about the project outcome, thanks, and the declaration of consent can be found in the Questionnaire provided in S1.

Questionnaires were created in Adobe Acrobat professional. The ranking categories were fixed to the ranking scheme (0–3), text sections allowed unlimited words and no grammar correction was chosen to avoid false auto-correction. The survey was conducted in February and March 2019. To keep the identity of our experts anonymous, responses of individuals are not presented in a personalized way but only in non-assignable categories (Table S2a,b).

Results of this survey are of mainly descriptive nature. We ranked knowledge gaps by their mean (highest to lowest). Most background variables were unsuitable to use for statistical tests as we did not have enough scores for each category (position/occupation, institution/organization, nationality, age, years of work experience in western Madagascar) or because almost all participants belonged to just one category, i.e., 82% had >10 years of work experience and 96% had work experience with locals. The variables "position/occupation" and "institution/organization" had too many levels to be tested. Other variables with several levels were merged together in the following way: The five age groups were merged into young (>25 and 25–35), medium age (36–45) and old (46 –65, >65) age. The variable 'nationality' was reduced to two levels: Malagasy and other nationalities. Background variables with su fficiently even distribution were tested for their influence in ranking the knowledge gaps using Kruskal–Wallis tests with Holm's adjusted p-values controlling for family-wise Type I errors [87,88]. All analyses were performed in R 3.6.1 [89]. Due to well-known interlinkages of our focal SDGs (2, 10 and 15) to other SDGs, various ABTs and the 2 ◦C target, e.g., [10,12], we linked our top 10 knowledge gaps to other relevant targets of these agreements, thus pointing out which targets of these agreements will be better reached when filling these priority knowledge gaps.
