Introduction

Habitat loss and fragmentation due to land use change is considered as the major driver of plant extinction around the world (Hermy, Honnay, Jacquemyn, & Brys, 2014; Tilman et al., 2017). In Java (Indonesia), the habitat loss is characterized by the conversion of forest into small-scale agriculture (Austin, Schwantes, Gu, & Kasibhatla, 2018). Higginbottom, Collar, Symeonakis, and Marsden (2019) reported that around 40% of montane forest areas have loss due to deforestation for the last 28 years. Conservation of plant species through multiple mechanisms is necessary to prevent unprecedented consequences of habitat loss, forest fragmentation, and global climate change (Mantyka-pringle, Martin, & Rhodes, 2012; Pacifici et al., 2015; Volis, 2017). Conservation of plant species in their natural habitats (in situ) is thought to be the preferred method of conservation (Oldfield, 2009); however, conservation outside their natural range (ex situ), including in botanic gardens, arboreta, or seedbanks, can play an unprecedented role in conservation (Potter et al., 2017; D. J. Pritchard, Fa, Oldfield, & Harrop, 2011; H. W. Pritchard et al., 2014).

Currently, there are more than 3,300 botanic gardens around the world (Botanic Gardens Conservation International, 2017a), with a total living plant collections of about 1.3 million individuals, representing more than 526,000 taxa (Botanic Gardens Conservation International, 2017b). Among these living collections, more than 1,300 species are designated critically endangered (CR) or endangered (EN), according to the IUCN Red List (Rivers, Shaw, Beech, & Jones, 2015). The thousands of threatened species conserved in the botanic gardens are irreplaceable resources, not only for conservation but also for research and education. Furthermore, living collections in the botanic gardens are key sources for many elements of conservation programs, including reintroduction, restorations, and population enhancement (Donaldson, 2009; Pennisi, 2010; Sharrock, 2012).

Nevertheless, ex situ conservation in the botanic gardens is facing a huge challenge. The average size of botanic gardens around the world is about 45 ha (Pautasso & Parmentier, 2007), enough to contain only a limited number of living plant collections (Lauterbach, Burkart, & Gemeinholzer, 2012). As a result, the genetic diversity represented in the ex situ collections may not always be adequate for conservation programs (Ensslin, Sandner, & Matthies, 2011). Generally, low number of individuals in a population makes such a population vulnerable to genetic drift and inbreeding depression, possibly causing lower fitness and other deleterious effects that jeopardize future generations (Ellstrand & Elam, 1993; Frankham, 2003). Therefore, limited size of living collections for each endangered plant species in the botanic gardens could have negative effects on conservation. This threat has now become a major concern for many conservation scientists and botanic gardens managers. We were therefore motivated to assess the representation of genetic diversity and intraspecies genetic differentiation of an endemic and threatened tree species from Java (Indonesia) by taking the advantage of a population genetic study, as suggested by Hoban and Schlarbaum (2014).

Vatica bantamensis (Hassk.) Benth. & Hook. ex Miq. was first described by Hasskarl in 1859 under the genus of Anisoptera before being classified under the genus of Vatica in 1862 and accepted under its current name (The International Plant Names Index, 2012; The Plant List, 2013). The genus of Vatica itself was described by Carl Linnaeus in 1771 (The International Plant Names Index, 2012). V. bantamensis is the only species in the genus of Vatica that is endemic to the island of Java, Indonesia (Ashton, 1982), although two other subspecies (V. vanulosa ssp. venulosa and V. javanica ssp. javanica) that are also distributed on the island (Ashton, 1982; Backer & Bakhuizen van den Brink Jr, 1963). V. bantamensis is highly regarded for its timber quality, hence it is used in construction (Soerianegara & Lemmens, 1994). This tree species is endemic only to Mount Payung, Ujung Kulon National Park (UKNP; West Java, Indonesia) and is currently classified as Critically Endangered on the IUCN Red List (Robiansyah, 2018). Mature tree of V. bantamensis can reach up to 30 m height, with diameter of about 40 cm (Ashton, 1982; Wihermanto, Dodo, Kusuma, & Muhiban, 2015). The fruit is composed of hard shell covering one seed called nut, with two shorter lobes or wings which then dispersed by wind (gyration-dispersed; Ashton, 1982; Suzuki & Ashton, 1996). Several species of Vatica (i.e., V. sumatrana and S. sarawakensis) exhibit hermaphrodite reproductive system with mass flowering event every 3 to 4 years (Brearley, Proctor, Nagy, Dalrymple, & Voysey, 2007). V. bantamensis is likely to follow the same pattern.

In this study, we assess the genetic diversity and structure of wild seedlings of V. bantamensis, an endemic and critically endangered dipterocarp from Java (Indonesia). We also estimated genetic differentiation between the wild seedlings and existing ex situ collection from the botanic gardens. We further explored the genetic diversity preserved in the ex situ collection and evaluated the possibility of additional collections from the wild and suggested an effective measure for an improved seedling selection for ex situ conservation strategy.

Materials and Methods

Collecting Sites

The UKNP is located on the westernmost part of the island of Java, Indonesia (Figure 1). Protected since 1921 (Whitten, Soeriaatmadja, & Afiff, 1996), the national park is known as the last remaining habitat of the Javan rhinoceros in the world (Fernando et al., 2006). In the national park, tropical lowland forest is commonly found, consisting mainly of primary and secondary forests. The forests are predominantly composed of palms, bamboos, thorny shrubs, and rattan (Hommel, 1990). The elevation ranges from sea level up to ca. 490 m above sea level (m asl; Whitten et al., 1996).

Figure 1.

(a) Collection sites of Vatica bantamensis on island of Java, Indonesia (map layers were downloaded from  https://www.globalforestwatch.org/,  https://www.naturalearthdata.com/, and  http://www.diva-gis.org/gdata); (b) wild seedlings of V. bantamensis in Ujung Kulon National Park; and (c) living collections in the Bogor Botanic Gardens; inset: leaves of V. bantamensis.

Results

Population Genetic Analysis

A total of 25,949,778 reads were obtained from the 68 samples (online Appendix S1), using the MIG-seq method. After filtering, trimming, and removing short-reads, 18,096,097 reads were retained for SNP calling in stacks software, generating a final catalog of 34,898 loci. From this catalog, a final set of 1,513 SNP loci was successfully called and genotyped. We filtered out loci with minimum minor allele frequency (min_maf) < 5% and removed loci exhibiting high pairwise linkage disequilibrium (threshold r > 0.8) and that deviated from Hardy–Weinberg Equilibrium. A final set of diploid 730 SNP loci were obtained from these procedures. A different parameter setting of minimum minor allele frequency (min_maf = 0.01) gave higher number of SNP loci (1,497). However, further downstream analysis (i.e., DAPC) shows similar result (online Appendix S2).

The proportion of missing data was 19.22% in the wild seedlings and 7.17% in ex situ collection. Nucleotide diversity (π) was higher in the wild seedlings (0.26) than in ex situ collection (0.16). Tajima’s D was lower in the wild seedlings (0.5) than in ex situ collection (1.02). Observed heterozygosity (H_O) values in the wild seedlings and ex situ collection were almost identical (0.73 and 0.76, respectively). Furthermore, a negative inbreeding coefficient (F_IS) was detected in the wild seedlings and ex situ collection (–0.4 and –0.12, respectively), but with relatively high standard error (0.4 and 0.01, respectively; Table 1). On average, HL was lower in the wild seedlings than in ex situ (0.66 and 0.75, respectively; t = 14.48, p < .001; Figure 2). Similarly, IR in the wild is more negative (less related) than in the ex situ (–0.03 and 0.75, respectively; t = 25.21, p < .001). Moreover, the wild seedlings and ex situ collection of V. bantamensis showed strong genetic differentiation (F_ST = 0.32).

Figure 2.

Comparison of the homozygosity-by-loci (HL) and internal relatedness (IR) of the ex situ collection and the wild seedlings of Vatica bantamensis.

Table 1.

Descriptive Statistics of Genetic Diversity of Vatica bantamensis for Each Population, Generated From 730 SNPs.

Genetic Structure of V. bantamensis

The genetic structure of V. bantamensis in the wild seedlings and ex situ collection was assessed using the DAPC method. The Bayesian Information Criterion value indicated that K = 3 was the best representation of the genetic data of V. bantamensis (Figure 3, online Appendix S3). DAPC used two discriminant functions from 21 PCs that accounted for 62.3% of the total variance of the genetic data. The results showed that all individuals from the ex situ collection were grouped into a single cluster, whereas the individuals from the wild seedlings were separated into two distinct clusters. On the other hand, in the STRUCTURE analysis, Evanno’s ΔK indicated an increased peak from K = 3 to K = 4 and started to decrease at K = 5 (Figure 4, online Appendix S3), suggesting three, four, or five clusters were inferred from the data set. AMOVA of the groups inferred from the DAPC supported the separation of all samples into three genetic clusters (ϕ = 0.48, p < .001, Table 2), with the subdivision of the population explained 47.48% of the genetic variation between the populations and 52.52% of the genetic variation within the populations.

Figure 3.

DAPC graph based on SNPs markers depicts three genetic clusters of Vatica bantamensis. Individuals belonging to the ex situ collection are separated from all the individuals from the wild seedlings. Clusters are represented by different color with red color coded for ex situ collection and other colors for wild seedlings.

DA = discriminant analysis; PCA = principal component analysis.

Figure 4.

Genetic structure of Vatica bantamensis inferred using Bayesian clustering implemented in STRUCTURE. Three different clustering results (K = 3, K = 4, and K = 5) are shown. Clusters are represented by different color with red color coded for ex situ collection and other colors for wild seedlings.

Table 2.

Analysis of Molecular Variance for Inferred Genetic Clusters of Vatica bantamensis From DAPC.

Discussion

Conserving the maximum genetic diversity of threatened species in living collections, including botanic gardens, is very challenging (Wee, Surget-groba, & Corlett, 2015), yet very important. Therefore, understanding the current genetic diversity and the genetic structure of known populations is necessary for developing appropriate conservation plans. Several metrics are used to assess genetic diversity in a population. In the current study, we focused on nucleotide diversity (π), Tajima’s D, and the inbreeding coefficient (F_IS), because of their ability to measure overall genetic diversity, reduction in observed heterozygosity in a population (J. Catchen, Bassham, et al., 2013; Holsinger & Weir, 2009), and detection population change based on neutral theory (Nielsen, 2005; Simonsen, Churchill, & Aquadro, 1995; Tajima, 1989a, 1989b).

The nucleotide diversities detected in the wild seedlings and ex situ collection of V. bantamensis were 0.26 and 0.16, respectively. These values were comparable to those reported for Panax spp. (Pan, Wang, Sun, Li, & Gong, 2016) and for Amorphophallus paeoniifolius (Araceae) studied with restriction site-associated DNA sequencing (RAD-seq) in China (Y. Gao et al., 2017). Other studies on tree species, such as Dalbergia cochinchinensis (Fabaceae) in Cambodia (Moritsuka et al., 2017), Populus balsamifera (Salicaceae) in Canada (Olson et al., 2010), Pinus sylvestris (Pinaceae; Wachowiak, Wόjkiewicz, Cavers, & Lewandowski, 2014), and Fagus sylvatica (Fagaceae; Lalagüe et al., 2014) in Europe, reported even lower nucleotide diversity levels than those reported in the current study. The earlier studies used various gene sequences, including chloroplast gene, nuclear gene, and expressed sequence tags to estimate the nucleotide diversity. Various genes or markers used in the studies may result in varying estimates of nucleotide diversity estimation due to variation in evolutionary rates (Nybom, 2004).

In both wild seedlings and ex situ collection of V. bantamensis, our analysis detected positive value for Tajima’s D and negative value for F_IS. A positive Tajima’s D indicates a population differentiation or balancing selection (Charlesworth, 2003). A negative F_IS represents an excess of heterozygosity (J. Catchen, Bassham, et al., 2013) caused by a small reproductive population size, over-dominance, negative assortative mating, or asexual reproduction (Stoeckel et al., 2006). Excess of heterozygosity (negative F_IS) in particular in the wild seedlings might be attributed to a bottleneck event in the past, resulting in a relatively small population size today. V. bantamensis is very rare and narrowly endemic species from the lowland forests in the Mount Payung area, UKNP (Ashton, 1982). Repeated volcanic eruptions of Krakatoa in 1883, 1680, and hundreds of years before (Lamb, 1970) possibly devastated the population of many plant and animal species during these periods, as has been suggested for the Javan rhinoceros (Fernando et al., 2006). These events might have also reduced the size of the V. bantamensis population in UKNP that is close to the volcanic island of Krakatoa, causing a genetic bottleneck. Nevertheless, more thorough historical and demographic analysis is required to test this hypothesis. However, negative F_IS in ex situ collection was more likely due to small reproductive population size.

Our comparison of the total genetic diversity of V. bantamensis between wild seedlings and ex situ collection indicates that the ex situ collection had lower diversity than did the wild one. There have been many cases where ex situ populations exhibit lower levels of genetic diversity than the remnant wild populations (Ensslin et al., 2011; Y. Gao et al., 2017; Lauterbach et al., 2012; Miao et al., 2015; Rucinska & Puchalski, 2011). Nevertheless, there was a study reporting the opposite results, where the ex situ populations included comparable or greater genetic diversity than did the wild populations (J. Gao, He, & Li, 2012; LaBonte, Tonos, Hartel, & Woeste, 2017). Lower genetic diversity in the ex situ collection of V. bantamensis could be attributed to small population size (n = 3) and kinship among individuals.

Kinship was indicated by the higher HL value showed in the ex situ collection (mean HL = 0.75) compared with the wild seedlings (mean HL = 0.66; Figure 2). The HL value has been reported to be associated with the level of correlated paternity (Nora, Aparicio, & Albaladejo, 2016). Hence, a high HL in the ex situ collection of V. bantamensis suggested strong kinship between the three individuals. Correspondingly, parental relatedness analysis using IR index indicated that the ex situ collection is significantly higher than wild seedlings (mean IR = 0.75 compared with –0.03, t = 25.31, p < .001). It shows that ex situ collections have more closely related parent than wild seedlings. Generally, ex situ populations suffer most from the two stages of sampling effect that occurs during the establishment (founder effect) and decay of genetic diversity because of small population size (genetic drift; Yang & Yeh, 1992). In addition, genetically related samples (siblings in the case of the ex situ collection of V. bantamensis) may also have a strong influence on the reduction in genetic diversity, because on average, full siblings share approximately 50% of their genome (Stadele & Vigilant, 2016; Visscher et al., 2006).

Our present study also revealed that the founding collections of V. bantamensis in BBG already suffered from low genetic diversity, as indicated by the higher HL of all the ex situ individuals compared with the wild individuals (Figure 2). This result shows that understanding individual genetic diversity, that is, from using HL index, is important before selecting the wildlings to take part in living collections for ex situ conservation. Selecting for the greatest genetic diversity among all collected wild seedlings from the wild population will benefit the living collection in the future because it minimizes the risk of random genetic drift occurring in a small population.

Our current study also found that there was strong genetic differentiation between the wild seedlings and ex situ collection of V. bantamensis (F_ST = 0.32). The high F_ST value obtained using Weir and Cockerham formula takes into account of unequal sample size; however, a possibility of overestimation is also acknowledged. Genetically distinct populations in the ex situ and wild populations have also been reported in other species, including Silene otites (Caryophyllaceae; Lauterbach et al., 2012), Taxus yunnanensis (Taxaceae; Miao et al., 2015), and Amorphophallus paeoniifolius (Araceae; Y. Gao et al., 2017), indicating distinct genetic differences between cultivated (ex situ) and wild (in situ) populations. The ex situ collection of V. bantamensis has a very small population size and comes from two different generations, the original seedling from the wild population and its progeny. Therefore, a small effective population size (founder effect), interacting with genetic drift (Ellstrand & Elam, 1993; Lauterbach et al., 2012), contributed to the strong genetic differentiation between the ex situ collection and the wild population of V. bantamensis.

We found that, based on DAPC, the genetic structure of V. bantamensis demonstrated clear separation into three genetic clusters (Figure 3). This result was also supported by AMOVA (ϕ = 0.48, p < .001). Based on DAPC analysis, Clusters 1 and 3 consisted of all the individuals from the wild seedlings, whereas Cluster 2 was composed of all the individuals from the ex situ collection. The two genetic clusters (Clusters 1 and 3) were observed in the wild seedlings did not attributed to geographical separation. Although, genetic differentiation between both genetic clusters was relatively low (F_ST = 0.14), the genetic variation may provide genetic pool that is important for future generation (Paaby & Rockman, 2014). Similar pattern on genetic clustering was also inferred based on STRUCTURE analysis with K = 3 (Figure 4). Moreover, group membership in DAPC results was precisely similar to the STRUCTURE analysis for K = 3 (Figure 4, online Appendix S4). On the other hand, small genetic similarity between ex situ individuals and the wild seedlings indicates that the original population source of the ex situ individuals is likely to have been extirpated. In addition, STRUCTURE analysis also detects more genetic clusters (K=4 and K=5) from the data set, indicating possibility of more subdivision within the wild seedlings. However, to avoid possible bias and overrepresentation of clustering (Gilbert, 2016; Puechmaille, 2016), we choose the K = 3 which is congruent with the result of DAPC for the optimal genetic structure.

We suggested that the individuals from genetic Clusters 1 and 3 are potential new sources for collection from the wild to augment diversity in the ex situ collection. These new accessions from the new distinct genetic Clusters (1 and 3) would increase genetic diversity if incorporated into the ex situ collection. Ex situ collections benefits plant conservation by providing propagation materials for population improvement and reintroduction (Cochrane, Crawford, & Monks, 2007; Donaldson, 2009; Pennisi, 2010; Sharrock, 2012). Therefore, the preservation and maintenance of high levels of genetic diversity in ex situ collection are indispensable (Cavender et al., 2015; Hoban, Kallow, & Trivedi, 2018).

Conservation Implications

Conservation action of V. bantamensis is urgently required because of the low population size and the existence of only a single population in the wild, as far as is currently known. Similarly, the number of individuals in the living V. bantamensis collection in BBG is also relatively small. Yet, this is the only available ex situ collection of the species. Low genetic diversity in the ex situ collection indicates that additional collection from the wild is necessary to increase the genetic diversity of the ex situ collection. Therefore, it is important that the two genetic clusters identified from the wild seedlings are also represented in the ex situ collection to increase the genetic diversity.

In addition, we also illustrate in the study how using molecular markers to detect genetic structure in the population of a threatened plant species can help to determine the optimum number of collections for ex situ conservation. We underline the importance of performing such analyses in other living collections in the BBG and other botanic gardens, and to take the approach into account for the future ex situ conservation plans. Finally, our study contributes to the development of more effective and rationalized measure for ex situ conservation strategies in the limited confines of the botanic gardens. In particular, this can be achieved by providing a method to evaluate the genetic representativeness of the ex situ collections and to select the most appropriate candidates for the living collection from the wild to cover most of the genetic variations present in the wild.

Supplemental Material

Supplemental Material

Supplemental Material

Supplemental Material