**1. Introduction**

Since at least the beginning of the Anthropocene epoch (beginning between ca. 7000 BP to 1964) [1], human activities have profoundly threatened the survival of other, non-human species on Earth. Human activities include overuse of resources, which are not renewable quickly enough to meet demands, fragmentation of once-continuous habitats to make way for infrastructure, and intentionally or unintentionally transporting species around the globe into new environments where they may threaten the naturally occurring

**Citation:** Wei, X.; Harris, A.; Cui, Y.; Dai, Y.; Hu, H.; Yu, X.; Jiang, R.; Wang, F. Inferring the Potential Geographic Distribution and Reasons for the Endangered Status of the Tree Fern, *Sphaeropteris lepifera*, in Lingnan, China Using a Small Sample Size. *Horticulturae* **2021**, *7*, 496. https:// doi.org/10.3390/horticulturae7110496

Academic Editors: Rosario Paolo Mauro, Carlo Nicoletto and Leo Sabatino

Received: 20 September 2021 Accepted: 10 November 2021 Published: 15 November 2021

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

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

species [2]. These activities have had the outcome of fundamentally altering the local and global patterns of geographic distributions of many organisms [3].

Understanding the natural environmental tolerances and preferences of species is fundamental to developing in-situ and ex-situ conservation strategies that seek to increase the populations of species, especially those that have become rare due to anthropogenic impacts. Unfortunately, the breadth of environmental tolerances and preferences of species cannot always be inferred by examining populations in the field. This is because it is often difficult or impossible to examine all populations or measure all environmental dimensions of a habitat and because species do not always occupy the full extent of the habitat available to them due to various biological constraints and environmental barriers [4,5]. To overcome these challenges, several broad classes of approaches have been utilized to extrapolate from the existing, or realized, distributions, of species to infer their potential distributions [6]. Among these classes are process-based approaches and correlative approaches. Process-based approaches focus on aspects of species biology, such as dispersal capabilities and pollination syndrome (i.e., in the case of plants). In contrast, correlative approaches involve building models based on existing distributional sites and environmental factors at those sites to make inferences about the potential distribution. At present, one of the most common correlative approaches is ecological niche modeling (ENM), especially via maximum entropy in MaxEnt [7]. Process-based and correlative approaches have exceptional potential in combination [8] to inform conservation strategies because the former can predict the natural potential of a species, while the latter can elucidate in what environments and geographic locations human-mediated conservation efforts may protect and encourage that potential. However, unfortunately, process-based and correlative approaches are rarely used together.

In this study, we sought to combine process-based understanding with correlative modeling to infer the potential distribution of *Sphaeropteris lepifera* (J. Sm. ex Hook.) R.M. Tryon, a tree fern that mainly occurs in China [9]. *S. lepifera* is a large species, characterized by an erect trunk (often more than 6 m tall and up to 20 cm in diameter near base) bearing distinct leaf scars [10]. Its leaves are pale green to green, about 3 m long, with scales and spreading hairs on the abaxial sides. The sori are often on the back of the smallest leaf divisions, or pinnules, and lack indusia. *S. lepifera* differs from *S. guangxiensis* Y.F. Gu & Y.H. Yan by having sparse spiny warts on its laminae and pinnae rachides and small scales on the pinnae rachides and abaxially on pinnule rachides. *S. guangxiensis* lacks the scales and warts on these structures [11]. This tree fern is listed as endangered within China as a grade two protected plant [12]. *S. lepifera* is treated in the family Cyatheaceae and is valued medicinally [13], horticulturally [14], and as a scientific model for evolution and biogeography [15,16]. Despite its present conservation status, this fern represents an ancient lineage that originated ca. 200 million years ago, became relatively widespread, and, later, underwent extinctions leading to its present restricted distributional range [17]. Thus, it has survived and multiplied since the Mesozoic time period. Within China, *S. lepifera* is primarily distributed in Taiwan [10], but a few wild populations have been found in coastal areas of mainland China, such as in Guangxi and Yunnan, including on near-shore islands, like Hainan [18–20]. Moreover, the species has also been discovered in Japan (Ryukyu Islands) and the Philippines, where it is likely to be native [21–24]. In Guangdong Province of China, *S. lepifera* was first found on Nan'ao Island of Shantou City by the Second National Key Protected Wild Plant Resources Survey of Guangdong Province in 2018. At present, the Nan'ao Island population, while tiny, is the only population in Guangdong Province showing new plant recruitment. Thus, the population is presently protected in-situ [25].

The causes of a rarity of *S. lepifera* have been considered in several prior studies [26–28]. One set of possible causes are environmental constraints on reproduction; especially on the ability of flagellated, swimming sperm to reach the female reproductive apparatus, the archegonium. In *S. lepifera*, the sperm originates from antheridia, which separated in space on the gamete-producing body from archegonia, thus necessitating at least a thin film of water for fertilization to occur. Aside from water, other constraints on sperm reaching

archegonia may especially arise due to the presence of heavy metal pollutants in the environment. For example, the rotation frequency and displacement speed of the flagella of sperm are negatively impacted by lead (Pb2+), while cadmium (Cd2+) may inhibit the ability of sperm to follow chemical signals that otherwise guide them to receptive archegonia [26,27]. However, heavy metals are likely not the main reason for limitations in reproduction asany species of Cyatheaceae have been shown to produce gametophytes but not sporophytes under experimental conditions where soil nutrients are free of heavy metals [28]. There are 19 species of Cyatheaceae in China with a protected conservation status [29], and only nine species, including *S. lepifera*, have been successfully bred under controlled conditions [28,30].

Protection of a species requires not only a detailed understanding of its unique biology but also accurate models of its environmental tolerances to support both meaningful in-situ and ex situ conservation actions. Recent developments in ENM have led to its application in diverse conservation issues, including prediction of suitable habitat and species ranges [31,32] and how human activities affect the distribution of species [33,34], but applications for the conservation of protected species are still somewhat limited, especially in combination with the biology of the species.

The main objective of our study of *S. lepifera* was to apply process-based understaning with correlative-based modeling to species distributional modeling to (1) determine the main environmental and biological factors affecting the wild distribution of the species, so as to better focus the scope of conservation and (2) assess suitable areas for ex situ conservation. We especially sought to find suitable areas in Guangdong, Hong Kong, and Guanxi Provinces for ex situ conservation because these places had or have wild populations of *S. lepifera*. To accomplish our study, we focused on the population on Nan'ao Island, which is the only one within the province that has been observed to undergo natural recruitment without human intervention. By comprehensively analyzing the environment of *S. lepifera* on Nan'ao Island and its environmental interactions, we expect to determine other similarly suitable areas for the species for reintroduction. We believe that our results provide a comprehensive framework for future endeavors at protection, reintroduction, and sustainable utilization of *S. lepifera*.

### **2. Study Area and Methods**

### *2.1. Field Investigation*

From 2018 to 2020, we carried out field surveys on Nan'ao Island of Guangdong Province. The island is located at 116◦53–117◦19 E and 23◦11–23◦32 N within the subtropical maritime climate zone, having four seasons, including mild winters and relatively cool summers. Within Nan'ao Island, *S. lepifera* occurs on the main island, which has an area of 111.44 km2.

On Nan'ao Island, we investigated plant recruitment of populations of *S. lepifera* as well as diameter, height, crown breadth, and health status of adult individuals of the species. We regarded adults as those individuals with heights greater than 3 m.

To determine the vascular plant composition of the community where *S. lepifera* is located, we also performed a Drude scale analysis, which is a quantitative method of assessing taxonomic composition, abundance, and cover [35,36]. To accomplish this, we established 20 m × 20 m plots centered around adult individuals and surveyed all cooccurring vascular plants within four 10 × 10 m quadrats nested within each plot. Within each quadrat, we sought to assign each species to one of the seven levels of the Drude scale [37]: plants of high sociability (soc; *plantae sociales*), three levels of copious or numerous (cop1-3; *copiose intermixtae*), sparse–sporadic (sp; *sparsae*, *sporadice intermixtae*), solitary (sol; *plantae solitariae*), and low abundance and/or singular individuals (uni; *unicum*). Assignments to these groups were determined based on relative-cover (relative cover = species cover/quadrat area), and the specific classification was as follows: soc (*rc* ≥ 75%); cop3 (75% < *rc* ≤ 50%); cop2 (50% < *rc* ≤ 25%); cop1 (15% < *rc* ≤ 25%); sp (5% < *rc* ≤ 15%); sol (5% < *rc* ≤ 3%); uni (*rc* ≤ 2%), where *rc* refers to relative cover. We performed the Drude

analysis separately for the tree, shrub, and herb layers, which we delimited on the basis of height with shrubs being 3–6 m tall and trees and herbs being greater and lesser in height, respectively.

### *2.2. Reintroduction Experiments*

From July to September 2020, we carried out reintroductions of *S. lepifera* to Nan'ao Island using plants initially grown in the greenhouse. We planted a total of 120 fiddleheads that had been growing in the greenhouse for about six months; 40 at the site of the existing population located in Nan'ao Island, and 80 others divided into two sites selected for translocation experiments. We also conducted preliminary translocation experiments at Maofeng Mountain in Guangzhou, Renhua County in Shaoguan, and Bajia Town in Yangchun, based on preliminary site assessments. At these sites, we planted 20 plants each, and at all sites, including Nan'ao Island, we planted the fiddleheads near a shady slope with a stream. We performed surveys at all sites of reintroduction and translocation in February 2021 to determine fiddlehead survival.

### *2.3. Gametophytic Development*

In order to investigate possible biological reasons for the relative rarity of *S. lepifera*, we conducted a spore breeding experiment. Ferns have a relatively unique life cycle in which both gametophytes and sporophytes can live independently. However, compared with the sporophyte, the structure of the gametophyte of *S. lepifera* is small, fragile, and composed of only one cell layer, whereas the sporophyte is ultimately arborescent. Therefore, gametophytic development of *S. lepifera* might be involved in its rarity or its distributional patterns.

We studied gametophytic development using spores newly germinated under lab conditions. We obtained the spores in September 2018 at our field site on Nan'ao Island. Specifically, we collected 50 mature sporophylls from two individuals of *S. lepifera* and stored these in paper bags in a cool and ventilated place until the spores fell off naturally after about seven days. We separated the spores from the dry vegetative material and stored them at 4 ◦C before further processing.

For germination, we placed the spores on 1/2 concentration MS medium after disinfecting them by applying 4% NaClO solution for 5 min. We grew the spores in culture under 16 h/d of 5000 Lux at 25 ◦C following recommendations for this species in a prior study [38]. Beginning on the third day, we sampled spores for microscopic examination every other day of two weeks. We selected characteristic spores for making into temporary slides at different stages, which we photographed under an OlympusBX43 light microscope.

Following the observations of germination, we transplanted the four-week-old gametophytes to sterilized soil. We divided this cohort of plants into four treatment groups containing 90–100 gametophytes each to determine tolerances of the species to temperature and moisture conditions. For temperature, we grew one group at 18 ◦C and the other at 25 ◦C both with daily watering. For moisture, we watered one group daily and the other once per week, and grew these plants at 25 ◦C. We regarded 25 ◦C and daily watering as "normal" (or control) growth conditions based on field surveys and the available literature. In all other respects, factors between the groups, such as lighting, were maintained as constant. One month later, we counted the number of sporophytes as a measure of successful fertilization.

### *2.4. Ecological Niche Modeling*

We generated ENMs in MaxEnt 3.4.1 to predict potentially suitable distributional areas for *S. lepifera*. For generating ENMs in MaxEnt, we obtained occurrence data from three sources: our field site where a natural population occurs, populations conserved ex situ in natural reserves, and georeferenced specimen records [9] from the Global Biodiversity Information Facility (GBIF) [9]. Using all obtained data records, we performed data thinning with resampling to only one occurrence per 1 km × 1 km area using the R

package, dismo [39] in R 4.0.5 [40,41]. Following data thinning, 409 occurrences remained, including one point from our field site and five ex-situ conservation sites in Guangdong Province (Table 1, Table S1).


**Table 1.** Specimen data from our field site and ex situ reintroduction areas.

We constructed the ENMs using the 19 environmental BioClim variables from World-CLIM 2.0 (Bio1-Bio19) [42–44], as well as slope and aspect derived from the Digital Elevation Model (DEM) downloaded from the same source. We also obtained landcover data as a raster map from the literature [45], and human population data from the Center for International Earth Science Information Network (CIESIN) [46]. We did not directly include elevation, which is widely known to be tightly linked with temperature variables.We obtained or resampled all variables at 30 arc-second resolution and clipped them to the boundaries of the distributional map that we established using the Database of Global Administrative Areas (GADM) [45].

Within MaxEnt, we determined the contribution of each environmental variable to the model as a whole by performing a jackknife test as an extension of model building. At the same time, in order to test the robustness of the model, we repeated the modeling process 12 times through the crossvalidate method [47], and we used "random seed" to randomize each run. We applied the area under the curve (AUC) to determine the predictive ability of each model.
