Genetic diversity and differentiation
For many cultivated crops, the consequences of artificial selection, founder effects and genetic bottlenecks result in considerable decline in genetic diversity [33–35], which may reduce their capacity for long-term survival and evolution in dynamic environments. Due to the comparatively brief domestication period and the exchange of genes with wild populations during extensive cultivation, the genetic diversity of medicinal plants typically declined much milder than that of crops. In our fieldwork, we investigated 13 sites of wild O. japonicus, 7 CMD cultivation bases in the villages along the Fujiang River in Sichuan Province, 3 ZMD cultivation bases, 2 of which were situated in traditional genuine producing area Cixi and one of which located in new producing area Santai. The present study indicated that cultivated O. japonicus exhibited a striking decrease in genetic diversity (hT = 0.521, hS = 0.030, VT= 0.842, VS= 0.067) compared to wild populations (hT =0.849, hS =0.341, VT= 0.859, VS= 0.204) after a few hundred years of domestication history. Out of 19 cpDNA haplotypes in the species, only 4 haplotypes were detected in cultivated populations. All these domestication result of decreased genetic diversity is similar to that of Scrophularia ningpoensis [36] and Angelica dahurica [8]. These dramatic changes in the genetic diversity of O. japonicus during such a short domestication history are most likely caused by founder effect and the clonal mode of reproduction. Firstly, the founder effect is that when a population is established and developed by a small number of individuals, the genetic information carried by these few individuals does not completely reflect the genetic information of their source population, resulting in a low genetic diversity of the resulting new population. The loss of genetic diversity for medical plants like Allium mongolicum [37], Leonurus cardiaca [38] and Lycium barbarum [39] are more or less related to the founder effect. In this study, due to the founder effect, only 2 haplotypes of cultivated O. japonicus was gathered from haplotypes of wild populations, meaning that the founders contained only a few samples of the genetic diversity of wild populations. Secondly, while the domestication history of O. japonicus (600 years) was shorter than other medical plants (2000 years), the genetic diversity of cultivate O. Japonicus dramatically decreased. It is probably attributed to the fact that wild O. Japonicus may expand via fruits, whereas the cultivated O. Japonicus were influenced by asexual reproduction and artificial selection. In addition, previous researches demonstrated that genetic bottlenecks caused by domestication in annual and perennial fruit crops. It concluded that perennial crops preserved a higher percentage of the genetic variation present in their wild progenitors than annual crops [40]. And juvenile phase length is a principal difference between domesticated annual and domesticated perennial fruit crops. In this study, the cultivation time before harvest was 2 or 3 years for ZMD and 1 year for CMD. ZMD (Hd = 0.06900; π = 0.00049) showed a higher genetic diversity than CMD (Hd = 0.06900; π = 0.00049). And gene flow of ZMD-wild (Nm=1.490) is higher than that of CMD-wild (Nm=0.830). Thus, it can be inferred that the relatively lengthy juvenile phases coupled with ongoing cultivar–wild gene flow due to the change of habitat may contribute to milder genetic bottlenecks in perennial ZMD than annual CMD, resulting in relatively high genetic diversity in ZMD.
In addition to reducing the genetic diversity of medical plants, domestication also affects the structural patterns of genetic variation [41]. For instance, notable genetic divergence was found between wild and cultivated Angelica dahurica and between its two cultivars (Fct = 0.148, P = 0.000) [8]. Our result of AMOVA analysis indicated that 21.8% of the total genetic variation occurred between cultivated and wild O. japonicus, while 98.37% of the total genetic variation occurred between CMD and ZMD (Table 3). Further analysis revealed that the genetic differentiation between ZMD and wild O. japonicus (FST = 0.72934) was lower than that between CMD and wild O. japonicus (FST = 0.82566). The results of mismatch distribution and neutrality test indicated that both ZMD and CMD had undergone a recent population expansion. Frequent extinctions and recolonizations of local populations can also be an important source of gene flow (Slatkin 1987). It has been reported that ZMD production areas have experienced significant changes in recent years [42]. Due to urbanization and regional economic reasons, most farmers in the traditional production area of ZMD (Hangzhou) no longer plant ZMD. Sanmen County, located in Taizhou City, Zhejiang Province, has gradually became a new producing region of ZMD. As a result, it is probable that one of the factors contributing to the higher gene flow between ZMD and wild populations (Nm = 1.490) than between CMD and wild populations (Nm = 0.830) is the shrinkage of traditional production regions and the creation of new production areas. In a group of completely isolated populations, genetic drift tends to fix different alleles in different local populations. Genetic drift is the unpredictable change in gene frequency due to finite population size. Gene flow between any populations will prevent complete fixation, but gene flow must exceed a certain level (Nm > 1) to prevent substantial genetic differentiation due to genetic drift [43]. This also clarified why the gene flow between ZMD and wild populations (Nm = 1.490) is able to withstand the genetic differentiation caused by genetic drift while the gene flow between CMD and wild populations (Nm = 0.830) is unable.
As for the genetic differentiation between CMD and ZMD, on the one hand, we believe that it is the result of artificial selection and the isolation by distance. Our surveys conducted in cultivated regions indicated that seedlings of CMD are prostrate CMKY-2, whereas the majority of seedlings of ZMD are erect O. japonicus and a small number are prostrate O. japonicus [44]. This combination of artificial selection and clonal propagation of O. japonicus rapidly fixes the alleles carried by selected individuals, which means that one or two haplotypes are fixed in each cultivar population, leading to large genetic differences between ZMD and CMD. Moreover, the herbal textual researches and the description of the local planters exhibited there was no record about the introduction of one cultivar of O. japonicus into another cultivar's growing region, which may have prevented gene flow and then leading to genetic differentiation between CMD and ZMD. Finally, population differentiation is also influenced by geographical distance. Isolation by distance is a model of population structure in which genetic differences between populations increase with geographic scale [45]. In other words, populations that are adjacent to each other are more similar than that are far apart, because they are linked by larger gene exchange, unless some impassable isolation breaks this mechanism and reduces communication between some populations and their neighbors. In our result, a greater NST (0.920) than GST (0.753) indicated the presence of phylogeographical structure of O. japonicus. Combined with the result of mantel test, it can be inferred that the restricted gene flow with isolated by distance contributed to the population structure of O. japonicus [46]. Geographical isolation can contribute to the lack of gene flow among populations, and then leading to the genetic differentiation between two cultivars.
Origin of cultivated O. japonicus
Designing strategies for the protection of genetic resources and the breeding of new varieties can be assisted by identifying the original site of domestication and evaluating the evolution history of wild-domesticated species [47]. Before Tang Dynasty (1200 years ago), it was believed that O. japonicus came from the wild and mainly distributed in Henan, Jiangsu, Zhejiang and Anhui provinces. On cultivation history, ZMD can trace back to Song dynasty (600 year ago) while CMD can trace back to Ming Dynasty (500 years ago) [20]. As the first study exploring the origin of O. japonicus, we explored the ancestral distribution of O. japonicus and its historical changes based on 4 chloroplast DNA data. ML phylogenetic tree (Fig. 3) showed that compared with the outgroup, ZMD and CMD grouped as a cluster. Further studies found ZMD and some individuals in the wild population BZ located in Sichuan grouped as a cluster, while CMD and the wild population NC located in Sichuan grouped as another cluster. These results suggested that the cultivated O. japonicus is more closely linked to wild populations in Sichuan than the wild O. japonicus in Zhejiang, which is similar to study on the genetic diversity of cultivated O. japonicus based on ISSR markers [29]. Therefore, the wild populations NC and BZ located in the Sichuan province may have played an important role in the origin of cultivated O. japonicus.
The TCS network claimed that 2 haplotypes (H1 and H2) found in CMD and 2 haplotypes (H2 and H4) found in ZMD were derived from H6 and H13 respectively. The 2 haplotypes (H6 and H13) both found only in the wild population BZ and TQ located in Sichuan province. All of the aforementioned haplotypes were derived from H13, which lied at the center of the haplotype network and only found in the wild population TQ located in Sichuan. It may be the ancestor of the two cultivars in this study. Furthermore, as the most frequent and widely distributed haplotype, H1 was found not only in CMD but also in the wild population NC, while H3, the one occurred in most ZMD also found in the wild population BZ, showing that the wild populations in Sichuan made a great contribution to the origin of O. japonicus.
Result of historical biogeography and the model of species distribution indicated that the ancestor of O. japonicus was located in Sichuan, and then experienced multiple vicariance-dispersal between Sichuan and Zhejiang due to emergence suitable climatic conditions after LGM. This may explain why haplotypes H5 appeared in different populations located in Sichuan, Zhejiang, Hubei, Hunan, Guizhou and other provinces. In addition, simulation results of species distribution also suggested that Sichuan province should be one of the glacial refugia for O. japonicus.
In summary, it can be inferred that O. japonicus was widely distributed in China during LIG period. With the passage of time, the cold climate caused O. japonicus to migrate east of Hengduan Mountains and south of Qinling Mountains during the LGM period. The present distribution was formed after several vicariance-dispersal events on the basis. More importantly, wild populations similar to the extant TQ located in Sichuan Province may be the ancestor of cultivated O. japonicus.
Sustainable cultivation of O. japonicus
Wild populations have been considered as a critical resource for breeding efforts in medical plants because their genetic information are used infrequently and generally exhibit no reproductively isolation from cultivated populations [48]. Previous study comparing cultivated and wild populations of O. japonicus indicated that cultivated O. japonicus has much lower genetic diversity than that of wild populations (Fig. 1, Table 1). The large-scale cultivation of O. japonicus has affected the genetic diversity of wild populations. For example, wild population NC showed a single, identical haplotypes (H1) to CMD. It may result in reduction in plasticity of cultivated O. japonicus to respond to changes in climate, pathogen populations, agricultural practices, or quality requirements. Hence, it is necessary to improve the genetic diversity of cultivated O. japonicus through sexual reproduction by using wild resources.
Accumulation evidence indicated that a proper assessment of genetic resources of medicinal plant is useful information for the development of conservation plans to protect genetic diversity. For example, protection of medicinal plants like Angelica dahurica [8], Aspidopterys obcordata var. obcordate [49] and Vitex rotundifolia [11] were mostly based on the conservation of genetic diversity. Our study exhibited wild populations DZ, HC and BZ had more cpDNA haplotypes and higher genetic diversity (Hd = 0.83300, π = 0.00568; Hd = 0.80300, π = 0.01341; Hd = 0.70500, π = 0.00488) than other populations (Fig. 1, Table 2). In addition, populations located in glacial refugia should be considered as priority areas for conservation because they not only preserve the genetic differentiation of species during the period of climate upheaval, but also serve as the starting point for post-glacial re-dispersal of species which can continue to maintain the continuity of genetic diversity [50–52]. In our study the ENM results suggested that east of Hengduan Mountains and south of Qinling Mountains was the glacial refugia of O. japonicus during LIG, while RASP analysis indicated that the ancestor of O. japonicus was located in Sichuan. These results demonstrated that Sichuan should be the starting point for post-glacial and re-dispersal of O. japonicus. In summary, populations DZ, HC and BZ that showed high genetic diversity as well as populations TQ and BZ that located in Sichuan with high genetic diversity should be served as key objects for biodiversity protection of O. japonicus and preferentially selected as the gene pool for improving the genotypes of cultivated O. japonicus.