Toona ciliata Roem. is a tall deciduous or semi-evergreen tree, and a precious timber species belonging to genus Toona (Meliaceae). T. ciliata was listed as a wild endangered species under secondary national protection (Yu, 1999). The natural distribution area of T. ciliata is limited to East Asia, South Asia and Australia. In China, T. ciliata is mainly distributed in South China, Central China, East China and Southwest China, and the natural population has sporadic distribution characteristics (Long et al., 2011; Li et al., 2016; Wang et al., 2018). In these areas, it is known as “Chinese mahogany” (Long et al., 2011; Wang et al., 2018) due to its high timber quality, so it has been overexploited and its natural regeneration is slow, leading to the reduction of its natural distribution (Wen et al., 2012; Chen et al., 2014).

So far, studies on T. ciliata at home and abroad have mainly focused on plant physiology and biochemistry, ecology, plant introduction and breeding, and related topics (Malairajan et al., 2007; Duan et al., 2015; Wang et al., 2016a; Wang et al., 2016b; Wang et al., 2016c; Huang et al., 2017; Wang et al., 2018). Genetic studies on T. ciliata have been reported with respect to phenotypic variation (Cai et al., 2018; Wang et al., 2018; Wang et al., 2019) and molecular markers (SRAP) (Li et al., 2016; Zhan et al., 2016). The SSR (Simple Sequence Repeat) molecular marker is a neutral molecular marker randomly distributed in the genome (Zietkievicz et al., 1994). Due to its advantages of co-dominant inheritance, wide distribution, good stability and repeatability (Powell et al., 1996), SSR technology has become an ideal genetic marker technology. It has been widely used in the genetic studies of some endangered and rare tree species, such as Eucommia ulmoides (Miao et al., 2017), Taxus wallichiana var. mairei (Yi et al., 2013), Parashorea chinensis (Zhang and Li, 2011), Pteroceltis tatarinowii (Fan et al., 2018) and Phoebe bournei (Liu et al., 2019). However, the application of SSR markers in the study of genetic diversity in genus Toona is seldom reported. Liu et al. (2013) compared the genetic diversity of central and peripheral populations of Toona ciliata var. pubescens by using SSR markers. Zhan et al. (2016) established an optimal reaction system of SSR-PCR and screened the highly polymorphic primers fitted for the SSR analysis of T. ciliata, which laid an academic foundation for genetic research on natural populations of T. ciliata.

Studies of the genetic characteristics of T. ciliata should consider the selection of provenances in different geographical distribution areas and the use of different molecular marker technologies in the investigation and evaluation of T. ciliata resources. We studied 192 germplasm samples from 24 natural T. ciliata populations from five provinces using SSR markers. Through the analyses of genetic diversity, population genetic differentiation, clustering and correlations between geographic distance and genetic distance, we aimed to reveal the genetic diversity level of T. ciliate, and the cause of that diversity, to provide a theoretical basis for the protection and development of T. ciliata germplasm resources.

In May 2016, experimental samples were collected based on the investigation of 14 naturally-distributed populations of T. ciliata in the provinces of Hubei, Hunan, Jiangxi, Guizhou and Guangxi Zhuang Autonomous Region (Table 1). Eight samples were collected for each population and the distances between sampled plants were all greater than 40 m. Small healthy leaflets on adult plants without pests and diseases were collected for a total of 192 testing samples. All leaflets were rapidly dehydrated with a large amount of color-changing silica gel and then kept at low temperature for subsequent DNA extraction. See Table 1 for specific sampling data for each population.

2.2.1 DNA extraction

Genomic DNA was extracted from leaflets of T. ciliata using the CTAB method. Purity and quality of extracted DNA were measured with 1.0% agarose gel electrophoresis (AGE), and the DNA concentration was measured with a UV spectrophotometer.

2.2.2 SSR-PCR amplification

Based on the published literatures (Liu et al., 2009; Liu et al., 2013; Liu et al., 2016; Zhan et al., 2016), 29 pairs of primers with good polymorphism were selected. The SSR- PCR (polymerase chain reaction) was performed on the Bio-Rad ptc-200 PCR apparatus (Bio-Rad Laboratories, USA). The reaction system was amplified in 19 μl PCR molecular marker solutions, the specific components of which (after being optimized) were: 12.1 μl dd H₂O, 2 μl template DNA, 0.1 μl Taq enzyme, 2.0 μl 10×PCR buffer, 1.8 μl MgCl₂ (25 mmol L^-1) and 1.0 μl primer mixture (10 μmol).

The PCR thermal cycling was: pre-denaturation at 94 ℃ for 5 min, denaturation at 94 ℃ for 45 s, annealing at 55 ℃ for 45 s, with a total of 30 cycles, and extension at 72 ℃ for 45 s. Then, the sequences amplified by PCR were extended at 72 ℃ for 10 min and maintained at 4 ℃ for 5 min. At the end of PCR, the amplified solutions were stored in the refrigerator at 4 ℃ for future use. A total of seven pairs of primers with stable amplification and good repeatability were selected for SSR analysis using the T. ciliata samples. Primer information is shown in Table 2.

The amplified banding patterns were recorded as 0 or 1. In the positions with the same mobility rate, each position with a band was denoted as 1, while positions without a band were denoted as 0. DataFormater software (Fan et al., 2016) was used to transform the data to meet the input requirements of the different analysis software programs. POPGENE 1.32 was employed to obtain the required genetic parameters, including observed number of alleles (N_a), effective number of alleles (N_e), observed heterozygosity (H_o), expected heterozygosity (H_e), Shannon information index (I), polymorphism information content (PIC), population inbreeding coefficient (F_is), genetic differentiation coefficient (F_st), the number of migrants per generation (N_m) and Nei’s genetic distance (Nei et al., 1983). NTSYS-pc 2.10s was used to draw the UPGMA-based dendrogram. TFPGA software was used to conduct the Mantel test for geographical distance and Nei’s genetic distance (Mantel, 1967). SPSS 22 and Excel 2013 were applied for data processing.

From 29 pairs of SSR primers, seven pairs of markers were obtained which had stable amplification, effective polymorphic information content and uniform genome-wide distribution (Table 2). A total of 17 alleles (N_a) was detected, yielding a mean value of 2.7143 alleles for each marker, with the variation ranging from 2 to 4, as can be seen in Table 3. The number of detected effective alleles (N_e) was 15.8214, for an average of 2.2602 alleles for each marker. The percentage of polymorphic bands (PPB) ranged from 49.92%-93.70%, and the mean was 78.82%, indicating that there was little difference between N_a and N_e, and that the detected loci were evenly distributed in each population. The average polymorphism information content (PIC) of each marker was 0.5182 and the arrangement of loci by PIC values was: S11 (0.7473) > T07 (0.5789) > S5 (0.5340) > T05 (0.5211) > S422 (0.5122) > S22 (0.4429) >T02 (0.2909). Shannon indexes (I) varied from 0.1447 to 1.2094, with an average of 0.7321. The observed heterozygosity (H_o) spanned 0.0000 to 0.5965, with an average value of 0.1055. The expected heterozygosity (H_e) ranged from 0.4266 to 0.6749, with an average of 0.4956. Nei’s genetic diversity index (H) values were between 0.4228 and 0.6675, with a mean of 0.4909. These results showed that the polymorphism of the seven pairs of SSR primers in the populations of T. ciliata was lower but effective, so these primers could be applied to provide a good analysis of the genetic diversity of T. ciliata.

The analyses of the genetic diversity parameters of the 24 populations showed that the number of alleles (N_a) varied between 1.0000 and 2.4286, with an average of 1.2629. The number of effective alleles (N_e) ranged from 1.0000 to 2.2286, with an average of 1.2081. The polymorphic in- formation content (PIC) was 0‒100.00%, and the mean was 19.05%. The observed heterozygosity (H_o) was between 0.0000 and 0.2857, with an average of 0.1136. The expected heterozygosity (H_e) spanned 0.0000 to 0.6190, with a mean of 0.1493, indicating that the diversity level of all populations was lower. Nei’s genetic diversity index (H) was between 0.0000 and 0.5159, with an average of 0.1044. Only P16 had a higher genetic diversity than that of the species level (H = 0.4909) while the other populations with H > 0.1000 were ranked in value as: P16 > P6 > P13 > P15 > P10 > P22 > P8. Shannon information indexes (I) were in the range of 0.0000‒0.8015, with a mean of 0.1546, indicating a low level of genetic diversity (Table 4).

The coefficient of inbreeding (F_is) reveals the deletion or excess of heterozygous genotypes in the total group of samples (Table 5). Among the seven loci, there were excess hybrid genes in five of them, and deleted hybrid genes in two loci (S11 and T02); overall, the mean heterozygosity of the populations was higher, indicating an inbreeding phenomenon of T. ciliata. This inbreeding phenomenon might be related to the characteristics of a small population or a high degree of geographic or environmental isolation (Wang et al., 2016a).

The genetic differentiation index (F_st) is an important indicator of inter-population genetic differentiation. The mean value of F_st was 0.7727, indicating a high degree of genetic differentiation among the populations. The F_st of T02 (0.2374) was the lowest, but even it reached a high level of genetic differentiation, while the F_st of S5 (0.9148) and S22 (0.9148) were both at the highest value. Gene flow (if N_m > 1) can play a homogenizing role, that is, it can effectively inhibit the differentiation between populations. But when N_m ˂ 1, genetic differentiation between populations definitely occurs (Wright, 1951). The mean N_m of the 24 T. ciliata populations was 0.0735, showing a low level of genetic exchanges between the populations, which inevitably resulted in the higher F_st.

Table 6 shows that the Nei’s genetic distances of the 24 T. ciliata populations were between 0.000 and 2.635, with an average of 0.548. The genetic distance between P9 and P18 was the longest, while the genetic distance between P1 and P3 was the shortest. The geographical distances among the 24 populations ranged from 20.202 to 1154.471 km. The geographical distance between P1 and P23 was the longest, while that between P14 and P24 was the shortest, with an average of 491.180 km. By using the UPGMA method, the genetic consistency among the populations of T. ciliata was clustered (Fig. 1). The populations of Guizhou Province and Guangxi Zhuang Autonomous Region were grouped together; while the populations of Hunan Province were clustered in another group; and the populations of Hubei were in a third group. The results showed that the 24 populations of T. ciliata were clearly clustered according to geographical distances.

The Mantel test (Mantel, 1967) was performed on the normalized logarithmic geographic distances between the different populations and Nei’s genetic distances (Fig. 2). The results revealed an extremely significant correlation between Nei’s genetic distance and geographic distance for the group of 24 T. ciliata populations (r=0.631, P=0.009 ˂0.05).

At the species level, Nei’s genetic diversity index (H = 0.4909) was consistent with (but slightly lower than) Shannon diversity index, indicating that the genetic diversity of populations was at a lower level. The level and distribution pattern of genetic diversity of a given plant species are the results of geographical distribution, breeding system, human interferences and many other factors (Wu et al., 2019), among which natural environmental differences can cause isolation between populations. Widely distributed in China, and adapted to complex and diverse environments, T. ciliata has derived rich genetic diversity. In this study, the latitudes of sampled T. ciliata populations spanned 24°02'12"- 32°01'36"N, and the difference between the flowering phases of the northernmost and southernmost distribution areas is greater than 30 d. Therefore, T. ciliata trees from the northern populations and the southern populations could not pollinate each other, resulting in reproductive isolation. Limited pollen and seed diffusion might contribute to a lower level of effective gene flow, and can easily cause a high proportion of self-pollination in a species (Liu et al., 2009). T. ciliata is mainly distributed in mountainous areas (at altitudes of 300-2260 m), and its sporadic distribution and small population together with overcutting, all add to the declines of its habitats and natural resources (Long et al., 2011), which contribute to the high degree of habitat isolation and obstruct the gene flow, resulting in the low genetic diversity of the T. ciliata populations. Except for P16, the genetic diversity levels of the other 23 populations were lower than that of the species level (H = 0.4909), which was higher than that found by Li et al. (2016) for the whole distribution area of T. ciliata in China (H = 0.3770). However, the average genetic diversity level of the populations in this study (H = 0.1044) was lower than that found by Li et al. (2016) (H = 0.1805), which might be related to the differences in sampling sites.

T. ciliata is a highly heliophilous species. So, if the plants in the forest cannot reach the canopy, then their competitiveness is insufficient, and the small-and medium-sized plants under the canopy often die (Wang et al., 2019). Therefore, the natural habitats of T. ciliata are along streams, rivers or narrow forest margins with optimal luminous conditions. When the suitable habitats shrink, the number of trees declines sharply, and smaller populations become more typical (Wang et al., 2016a). Meanwhile, species with diffusive incompetence are more susceptible to the influence of edge locations, compared with species which have long- distance diffusion competence (Lesica and Allendorf, 1995; Peakal and Smouse, 2012). A strategy of scattered survival would make it difficult for T. ciliata trees to maintain extensive gene exchanges, even within the populations, which may lead to low genetic diversity at the population level.

The high degree of genetic differentiation indicates that the homologous probability of two gametes being randomly selected from any non-cohabitation populations is low, and, therefore, the homogeneity of genetic composition of the population is also low. The genetic differentiation coefficient (F_st = 0.7727) was higher than that found for T. ciliata var. pubescens (a T. ciliata variety) in central (0.1520) and peripheral populations (0.3045) (Liu et al., 2009), showing that at the species level, the genetic variation (77.27%) within T. ciliata populations was higher than the genetic variation between groups (22.73%), and that the genetic differentiation within populations was the main factor causing the genetic variation of T. ciliata. This result was consistent with Li’s finding that 79.26% of the genetic differentiation existed between populations through AMOVA analysis (Li et al., 2016).

Genetic differentiation is influenced by gene flow, natural selection and mutations (Schaal et al., 1998). Gene flow (N_m) is the flow of genes between populations and an important factor affecting the genetic differentiation of populations. A greater level of gene flow between populations causes more homogeneous populations (Slatkin, 1981). However, as long as the gene flow is in pleiotropy, it can prevent the genetic differentiation caused by genetic drift between populations when the inter-population migration per generation is N_m ≥ 1 (Hamrick et al., 1995). In this study, we found that gene exchange between populations was low (N_m = 0.0735 ˂ 1), and thus increasing the genetic differentiation between populations. Since gene flow mainly comes from seed flow or pollen flow, geographical isolation caused by mountains or rivers could block it (Nagel et al., 2015; Yi et al., 2018).

Nei’s genetic distance was significantly related to the geographical distance of the T. ciliata populations. First of all, there were six populations in one group, including P4 of Guangxi Zhuang Autonomous Region as well as P1, P3 P7, P9, and P10 from Guizhou Province. In addition, P2, P5 and P8 from Hunan Province were grouped together, while all 14 populations in Hubei were grouped together. Such a clustering reflected the differences in the geographical distribution areas of T. ciliata (both in the north and in the south), and, as a result, reproduction between the populations was almost completely isolated and the gene flow was greatly blocked.

As an important source of genetic diversity, wildlife may possess valuable genetic resources which can serve as the basis for resource utilization. Therefore, systematic research and scientific protection of wildlife should be emphasized (Wu et al., 2019). For protecting germplasm resources of T. ciliata, it is necessary to select and breed superior populations according to the higher genetic diversity of the species resources and their genetically-differentiated characteristics. P16, for example, has the highest genetic diversity (H = 0.5159), which is higher than that of the species level (H = 0.4909). As we observed in this investigation, the natural resources of T. ciliata in Xuan’en region were very rich. This might give rise to a higher probability of gene exchanges within the populations, which might help to effectively reduce the genetic differentiation.

We can generally conclude that the larger distribution area of T. ciliata results in the lower genetic diversity of the species, but the higher genetic diversity at the population level as a whole. The differences in the geographical distribution areas of T. ciliata can add to reproductive isolation. Furthermore, the geographical and environmental characteristics within smaller areas in each group coupled with the resource pressure from human activities have led to the unique clustering pattern. For example, terrain blockage, human interference, and frequent rainfall in the flowering period could bring the reduction of gene exchanges within the populations, resulting in lower genetic diversity within populations. Meanwhile, natural selection and genetic mutations may increase the genetic differentiation.

The key element of germplasm breeding of T. ciliata lies in the selection of families and plants with high genetic diversities within the different populations (Yang et al., 2017; Wu et al., 2019). The populations with the lowest genetic diversities, P14, P17 and P21, for example, might harbor higher potentials, so pursuing on-site protection together with ex-situ protection strategies is recommended (Yi et al., 2018). The selection of parents in crossbreeding, genetic relationships between individuals separated by geographical distances and the parent populations (or individuals) should all be taken into account (Yang et al., 2017), as well as the ecological and timber objectives, so as to maximize the preservation and utilization of the genetic diversity of T. ciliata.