Genetic diversity, population structure, and traditional culture of Camellia reticulata

Abstract Camellia reticulata is an arbor tree that has been cultivated in southwestern China by various sociolinguistic groups for esthetic purposes as well as to derive an edible seed oil. This study examined the influence of management, socio‐economic factors, and religion on the genetic diversity patterns of Camellia reticulata utilizing a combination of ethnobotanical and molecular genetic approaches. Semi‐structured interviews and key informant interviews were carried out with local communities in China's Yunnan Province. We collected plant material (n = 190 individuals) from five populations at study sites using single‐dose AFLP markers in order to access the genetic diversity within and between populations. A total of 387 DNA fragments were produced by four AFLP primer sets. All DNA fragments were found to be polymorphic (100%). A relatively high level of genetic diversity was revealed in C. reticulata samples at both the species (H sp = 0.3397, I sp = 0.5236) and population (percentage of polymorphic loci = 85.63%, H pop = 0.2937, I pop = 0.4421) levels. Findings further revealed a relatively high degree of genetic diversity within C. reticulata populations (Analysis of Molecular Variance = 96.31%). The higher genetic diversity within populations than among populations of C. reticulata from different geographies is likely due to the cultural and social influences associated with its long cultivation history for esthetic and culinary purposes by diverse sociolinguistic groups. This study highlights the influence of human management, socio‐economic factors, and other cultural variables on the genetic and morphological diversity of C. reticulata at a regional level. Findings emphasize the important role of traditional culture on the conservation and utilization of plant genetic diversity.

of biodiversity is called for to support the maintenance of ecosystem processes that ultimately sustain human life (Khairo, 2016).
Traditional cultural practices and local ecological knowledge of smallholder farmers and indigenous communities that has accumulated over generations have been widely recognized to contribute to biodiversity conservation (Bohn, Diemont, Gibbs, Stehman, & Vega, 2014;Shen et al., 2012) and may potentially be tapped for addressing the rapid loss of biodiversity around the world. Previous studies have acknowledged the role of traditional knowledge and culture practices of smallholder farmers and indigenous communities for biodiversity conservation at the species, genetic, ecosystem, and landscape levels (Altieri, 2004). Many traditional management practices, customs, and beliefs have been reported to contribute to biodiversity protection including seed exchange systems (Labeyrie, Thomas, Muthamia, & Leclerc, 2016), marriage exchanges (Delêtre, McKey, & Hodkinson, 2011), religious rituals (Mazumdar & Mazumdar, 2012), and dietary traditions (Penafiel, Lachat, Espinel, Van Damme, & Kolsteren, 2011).
Camellia reticulata Lindl. (Theaceae), an evergreen arbor (Liu & Gu, 2011) known as "Yunnan shan-cha" in Chinese, has a long history of being cultivated by various sociolinguistic groups of China's Yunnan Province including the Yi, Bai, Naxi, Hui, Miao, and Lisu for its ornamental, horticultural, and cultural values as well as for the oil derived from its seed (Xin et al., 2015). The Bai ethnic group cultivates C. reticulata in their gardens as a symbol of gentility, and the Yi pay special respect to camellias while communities in Tengchong extract an edible oil from the seeds (Wang & Ruan, 2012). C. reticulata is naturally distributed in Yunnan, southwest Sichuan and Guizhou Provinces of China (Ming, 2000). Morphologically, C. reticulata has attractive large flowers (7-18 cm in width) with bright red or pink petals (Li, Hashimoto, Shimizu, & Sakata, 2007) that have a long blossoming season in winter.
Camellia reticulata is a perennial, outcrossing, and heterogenous double ploidy species (2n = 30) (Ming, 2000). The cultivation of C. reticulata dates back to the Sui and Tang dynasties (~600 AD). This species has been widely described in classical literature including poems and inscriptions (Wang et al., 2011). Camellias are widely cultivated in Buddhist temples and offered to the Buddha (Xin et al., 2015). The British botanist J. Lindley identified and named Camellia reticulata in the Botanical Register in 1827 and introduced this species to Europe (Ming, 2000). By the 1950s, C. reticulata was found in the gardens of western countries and continues to be desired as an ornamental including by the potted flower industry (Hattan et al., 2016).

Managers of C. reticulata have bred this resource with other
Camellia species to produce colorful cultivars with different colors and blossom periods (Zhou et al., 2014). During the process of polyploidization and hybridization, other Camellia species have likely contributed to the genetic diversity of C. reticulata (Hong, 2010;Wang & Ruan, 2012;Yuan, Cornille, Giraud, Cheng, & Hu, 2014) and resulted in over 500 cultivars and hybrids (Chen, Wang, Xia, Xu, & Pei, 2005). However, with rapid environmental and socio-economic change as well as a trend toward monocultures and reduced genetic diversity (Barrett, Travis, & Dasgupta, 2011;Cardinale et al., 2012), it is becoming important to understand the current status of plant genetic diversity of C. reticulata (Wang, Chiang, Roux, Hao, & Ge, 2007). While numerous studies have reported on the importance of traditional management practices for biodiversity conservation, the complex interaction between cultural practices and genetic diversity of C. reticulata has not been studied (Hong, 2010).
In this study, we integrated ethnobotanical and molecular genetic approaches to examine the influence of cultural factors on the conservation of genetic diversity of C. reticulata. Semi-structured and key informant interviews were carried out with managers of C. reticulata in central and western parts of Yunnan Province of China. Samples from five distinct populations of C. reticulata were collected from Yunnan in order to analyze the genetic diversity and population structure using AFLP analysis. It is expected that the genetic diversity of C. reticulata has been influenced by the cultural practices of various sociolinguistic groups who have managed, cultivated, and conserved this tree resource.

| Ethnobotanical survey and plant collections
Research was carried out in the main distribution and production areas of C. reticulata. This includes locations in central and western parts of Yunnan Province in five prefectures including the following: Kunming, Chuxiong, Dali, Tengchong (in Baoshao City), and Lijiang ( Figure 1). Ethnobotanical methods including semi-structured interviews and key informant interviews were conducted to understand the history, culture, use, local names, and number of cultivars of C. reticulata. Informants were randomly selected from local community members in different socio-linguistic groups (including the Yi, Bai, Dai, et al. who have their own unique language and culture) at the study sites who were willing to participate in surveys. A total of 120 informants participated in this study including 38 from Kunming, 20 from Chuxiong, 25 from Dali, 26 from Tengchong, and 11 from Lijiang. In addition, we examined relevant literature regarding beliefs, traditional knowledge, and customs on C. reticulata at the study sites.
A total of 190 individuals from five populations were sampled from the five study sites. Their local name and the morphological characteristics including flower color, flower type, and blooming period were recorded according to ethnobotanical survey as well as plant database research (see Appendix S1).

| AFLP analysis of genetic diversity
Approximately 30-40 individuals of C. reticulata were collected from each population. From each sampled plant, young, healthy leaves were collected, silica-dried, and stored at −20°C for laboratory analysis of genetic diversity. Genomic DNA was extracted from 15 to 30 mg dried-leaf samples following modified CTAB method (Doyle & Doyle, 1990). AFLP was performed according to Vos et al. (1995).
Each diluted DNA sample was digested with EcoRI and MseI and ligated to specific adapters using T4 DNA ligase. EcoRI + A and MseI + C primers were used for preamplification (Table S1). PCR reactions were performed in a GeneAmp PCR system 9700 (ABI, USA). The preamplification program consisted of 25 cycles of 30 s at 94°C, 1 min at 56°C, and 1 min at 72°C. Evaluation of the restriction digest, adapter ligation, and the preamplification occurred on a 1.5% TAE buffered agarose gel, using λ Pst as size reference (Tables S2 and S3) EcoRI-ACC/MseI-CTC (PC3), and EcoRI-ACG/MseI-CTC (PC4). The amplification program consisted of 2 min at 94°C, 30 s at 65°C, 2 min at 72°C, followed by 8 cycles of 1 s at 94°C, 30 s at 64°C, and 2 min at 72°C (the temperature decreased after each cycle with 1°C, from 64°C to 56°C) and finally 23 cycles of 1 s at 94°C, 30 s at 56°C and 2 min at 72°C (Table S4). AFLP fragments were separated on a 3130xl Genetic Analyzer (Applied Biosystems). GeneMapper v4.0 (Applied Biosystems) was used to record the signal peak height, the signal area, and the fragment size. The data were exported to Access, and a scoring table was generated as a starting point for the data analysis. The bands were scored as either present (1) or absent (0) across all loci (see Appendix S2).
The proportion of diversity among populations was estimated using the following equation: (H t −H pop )/H t . An analysis of molecular variance (AMOVA) was performed using Arlequin (Schneider, Roessli, & Excoffier, 2000) and analyzed from both among populations and within populations. The UPGMA dendrogram of populations based on AFLP fingerprints was drawn based on pairwise similarities using software TFPGA (Miller, 1997).
To further elucidate relationships among all populations, a modelbased cluster analysis was performed using the program STRUCTURE v2.3.1 (Raj, Stephens, & Pritchard, 2014). STRUCTURE was run 20 times and was carried out by setting the number of clusters (K) from 1 to 14. The optimum number of clusters (K) was processed and identified by STRUCTURE HARVESTER through comparing log probabilities of data for each value of K (Earl & Vonholdt, 2012;Stephens & Pritchard, 2003). The output of structure analyses was visualized using the software CLUMPP v1.1.2 (Kopelman, Mayzel, Jakobsson, Rosenberg, & Mayrose, 2015) and DISTRUCT v1.1 (Rosenberg, 2016).
Data were input in an Excel spreadsheet in order to carry out statistical analysis and to make figures depicting genetic diversity and cultivar diversity of C. reticulata.

| Ethnobotanical survey Camellia reticulata
Results of ethnobotanical survey have been published by the author (Xin et al., 2015) that highlighted the impact of traditional culture on the conservation of C. reticulata. In this survey, we found that there are 206 ancient trees of C. reticulata in Kunming, with 28% of them well maintained in old Buddhist temples including the Panlong temple and Huating Temple. In Chuxiong, 26 of the 72 old C. reticulata trees are maintained in Buddhist temples. In Lijiang, there are two different cultivars of old C. reticulata trees that have existed for over 500 years and are entitled as "the king of camellia." Furthermore, there is an old C. reticulata in Dali that has grown for about 400 years in the highest altitude of 2,750 m among all Camellia species.

| Genetic diversity of Camellia reticulata
A total of 387 AFLP unambiguous bands (100%) with a gene size ranging from 50 bp to 455 bp were obtained from the 190 plant samples collected for this study with the four EcoRI/MseI primer sets (Table 1).
The observed effective number of alleles (N e ), Nei's Gene Diversity (H), Shannon's Information Index (I), total genetic diversity (H t ), genetic diversity within populations (H s ), and gene differentiation coefficient Table 1. From the data displayed in Table 1,   migrants (N m ) was 15.2925 and indicates that the rate of gene flow is high enough for transferring genetic diversity among populations.

| Genetic differentiation within and among populations
Analysis of molecular variance (AMOVA) on genetic differentiation among and within populations of C. reticulata was conducted and is shown in Table 3. Findings from AMOVA revealed that 96.31% of the total genetic variations was contributed by differences within populations (p < .001), which was notably and significantly higher than that among populations (only 3.96% of total genetic variation was due to difference between populations).

| Genetic relationships and geographic distances
The genetic distances among all five populations (Kunming, KM; Lijiang, LJ; Chuxiong, CX; Dali, DL; Tengchong, TC) of C. reticulata were calculated using TFPGA Software v1.3 and are shown in Table 4.
Variation was found in the genetic distance among populations rang- The correlation analysis between genetic distances and geographic distances among the five populations is presented in Fig. S1.
Findings indicate that there was significant correlation between genetic distances and geographic distances among the five populations of C. reticulata (p = .035, p < .05). However, an R 2 value of 0.037 indicated that about 97% of the variation is not explained by the distance between population pairs, but may be caused by other factors such as hybridization by human cultivation practices.
The UPGMA dendrogram based on Nei's genetic distances that was constructed through POPGENE32 v1.32 (Figure 2) clustered all five populations into different groups. Figure 2 shows that there is no significant correspondence between the genetic structure and geographical origin of C. reticulata individuals. Evidence of the population structure of C. reticulata and the distribution of flower characteristics was strengthened by multiple models implemented in the program STRUCTURE v2.3.1 (Figs S2 and S3).

| Morphological structure of C. reticulata and its correspondence with geographical origin
The optimum number of clusters (K) was processed and identified by STRUCTURE, which was K = 7 ( Figure 3). The populations inferred by STRUCTURE for K = 7 were able to distinguish flower characteristics including flower color and flower type. In Figure 3, the red areas   (Table S5). About 88.60% of flower data of all samples were covered by these 12 flower characteristics as shown in Table 5. Double-pink-early maturity (20%) is the most common flower pattern of C. reticulata, followed by semi-double-pink-early (17.37%) and double red-middle (14.21%) ( Table 5).
Correspondence between the morphological structure and geographical origin is presented in Figure 5. Peach pink is the main flower color for all the five sites. Double flower type is the most type for all sites except for Tengchong with a much higher number of single type. Early and middle blooming period are the two main blooming periods for all sites except for Dali that has fewer varieties with an early blooming period. Regardless of the flower color, flower type, or blooming period, Kunming was found to have the most amount of C. reticulata varieties compared to the other four sites. Therefore, morphological structure seems not to have influence geographical distribution and distances.

| Correspondence between genetic diversity and cultivar diversity of C. reticulata
Informants named a total of 75 cultivars of C. reticulata during our ethnobotanical survey and named these cultivars based on their morphological features, local customs, people's interests, and values. The population of C. reticulata in Kunming, Chuxiong, Dali, and Tengchong all has rich cultivar diversity. However, the populations from Kunming, Dali, and Tengchong have higher genetic diversity than the population of Chuxiong ( Figure 6). The cultivars known as "Dalicha," "Shizitou," "Zipao," "Zaotaohong," and "Mudan" were found to be very common and popular in Yunnan with a genetic diversity of 0.3298, 0.3402, 0.3298, 0.3402, and 0.3402, respectively.
As shown in Figure 5, this study did not have significant correspondence between genetic diversity and cultivar diversity of C. reticulata Except for the Chuxiong population, C. reticulata in the other three populations showed almost the same genetic diversity among different cultivars.

| Genetic diversity and cultivar diversity of Camellia reticulata
Camellia reticulata is an endemic and religious tree restricted to Yunnan Province in south China. Our AFLP survey of that five natural populations of C. reticulata revealed a relatively high level of genetic diversity at the species level (H = 0.3578, I = 0.5369, PPL = 85.63%; Tables 1 and 2). Interestingly, a large proportion of genetic variation resided within populations (96.31%). By contrast, the genetic diversity of C. reticulata was relatively low between populations (3.69%) ( Table 3). Findings showed that geographic distance among five populations was irrelevant to their genetic variations and distances ( Figure 2 and Table 4). Therefore, geography is not the driving influence of the genetic diversity within populations of C. reticulata.
Rather, the observed genetic diversity of C. reticulata is more likely due to the accumulation of different genotypes resulting from artificial hybridization facilitated by human cultivation practices and other cultural practices that serve to not only protect old cultivars but also to enhance new cultivars.
Comparison of the genetic data of C. reticulata with genetic data collected from populations of close congeners, C. taliensis (Zhao, Yang, Yang, Kato, & Luo, 2014) and C. japonica (Lin, Hu, Ni, Li, & Qiu, 2013), found no variation among wild populations. C. taliensis has a long domestication history as a tea tree widely distributed in Yunnan and has a relatively moderate to high level of overall gene diversity (H s = 0.597) (Zhao et al., 2014). While no variation was found among wild populations in the AMOVA, most of the variation was detected within popu-  (Lin et al., 2013). While a relatively high level of genetic differentiation among populations of C. reticulata was revealed by AMOVA (22.5%) by this study, relatively low genetic diversity existed within populations. This genetic trend is likely because of overexploitation, frequent human activities, and insufficient conservation management (Lin et al., 2013;Zhao et al., 2014). Previous studies have supported that a long cultivation history, Buddhist practices, and other traditional cultural factors have contributed to the morphological diversity of Camellia trees (Xin et al., 2015). C. reticulata has a long history of use for worship and offerings in Buddhism in temples and altars in Yunnan Province by Bai, Yi, and other sociolinguistic groups (Xin et al., 2015). Many ancient trees of C. reticulata have been maintained for hundreds of years under the influence of Buddhism in temples and sacred areas in Yunnan (Xin et al., 2015). Other religions and beliefs adhered to the study areas are associated with conservation practices including those linked to nature worship, totemism, and ancestor worship (Long, Zhang, Pei, & Chen, 1999). For example, villages in Chuxiong of Yi Autonomous Prefecture build temples (called "Patron God Monastery") (Yang & Sun, 2001) that consist of many varieties of C. reticulata. In addition, tree worship is one of the most important forms of worship in this area, especially for C. reticulata (Liu, Pei, & Chen, 2000). C. reticulata has played a role in the life of Yi of Chuxiong as a tribute to worship ancestors as well as for their farming practice (Lai, 2016). The Bai of Dali further regard C. reticulata a symbol of ancestral/nature spirits, cultural identity and status and give specific names for each cultivar based on different cultural symbols, meanings, and uses (Xin et al., 2015). For example, In Tengchong, a prized cultivar of C. reticulata named Hong-hua-you-cha is considered distinct and highly valued for its seed oil hailed as "Oriental olive oil" that is widely used and commercialized for its high nutrition and medicinal values (Sahari, Ataii, & Hamedi, 2004). Findings from this study support that such cultural practices support rich genetic diversity and cultivar diversity of C. reticulata resources. Findings from our ethnobotanical surveys documented the conservation of old trees in temples within the study area and support that cultural beliefs and practices serve to protect old cultivars of C. reticulata to a large extent as well as influence their species diversity through natural and human selection and hybridization.

| Population genetic structure and morphological structure
In our study, both Nei's genetic differentiation index among popu- rich cultivar diversity in these four populations ( Figure 6) with no-  Although pollen of C. reticulata is dispersed by birds or insects naturally (Kunitake, Hasegawa, Miyashita, & Higuchi, 2004), the mountains in Yunnan where this species occur are distant from each other and have been recognized to result in genetic differentiation among populations (Ellstrand, 2014). Mountain geography has been likened to island geography in facilitating divergent evolution and fostering biodiversity (Winger & Bates, 2015). Gene flow through pollen migration by insects is most likely not a driving factor in the ecological evolution process of C. reticulata populations, but rather artificial propagation by humans that has been facilitated by cultural exchange and other cultural factors between different geographies.

| CONCLUSION
This study highlights that Camellia reticulata resources have relatively high genetic diversity in Yunnan Province of southwestern China and that cultural factors may be a notable driving influence on fostering this diversity compared to geographic distance. AFLP was validated to be useful for examining the genetic evolution of C. reticulata as well as for elucidating genetic relationships of different C. reticulata cultivars by cluster analysis. This is the first study to provide evidence on the genetic diversity, structure, and differentiation within and among populations of C. reticulata. Traditional cultural practices and beliefs of different sociolinguistic groups in the study area of China have likely served an important role in the conservation and enhancement of C. reticulata diversity. We expect this study will be helpful for supporting biodiversity conservation, efforts for cultivar introduction, and further studies of C. reticulata and related species that are valued by different cultural groups.

ACKNOWLEDGMENTS
We would like to thank Veerle Buysens and those at Institute for Agricultural and Fisheries Research for technical and logistic assistance. We are grateful to the farmers of this study for sharing their knowledge and Camellia samples.