Usually, successful conservation strategies require obtaining the genetic information, of which genetic diversity and population structure are essential parts. Molecular markers become useful tools to study genetic diversity and population structure of natural germplasm resources in non-model plants with no reference genomes (Parida et al., 2009). Compared to the other makers, SSRs possess the advantages of relative abundance, high polymorphism, codominant inheritance, and reproducibility, which have gained considerable importance in plant genetics and breeding (Taheri et al., 2018). In addition, SSRs have also been applied for discovering quantitative trait loci (QTL), linkage map construction between gene and marker, marker assisted selection for desired traits (MAS), and so forth (Kalia et al., 2011).

According to the source, SSRs are classed into genomic SSRs (gSSRs) and expressed sequence tag SSRs (EST-SSRs), the latter of which are located in the genic transcribed regions and are identified by NGS technology (Varshney et al., 2005; Wei et al., 2011). In general, EST-SSRs have been found to be more development-inexpensive, more evolutionarily conserved, and have higher transferability to related species than traditional anonymous gSSRs (Scott et al., 2000; Gadissa et al., 2018). Moreover, they are expected to be less polymorphic than gSSRs because of greater DNA sequence conservation in transcribed regions, but also less prone to null alleles (Postolache et al., 2014). The location of EST-SSRs determines their functional roles. EST-SSRs in CDSs affect the inactivated or activated genes or truncate proteins, in 3′UTR are involved in gene silencing or transcription slippage, and in 5′UTR impact gene transcription and/or translation (Lawson and Zhang, 2006; Gao et al., 2013).

Cephalotaxus oliveri (Cephalotaxaceae) is an endemic conifer species of China. It is a perennial shrub or small tree which is 4 m tall with white stomatal bands between the midvein and marginal bands of abaxial leaves. This plant can be cultivated as an ornamental in gardens, and its wood can be used for the manufacture of farm tools and furniture (Fu et al., 1999). Moreover, a variety of plant alkaloids (such as anticancer alkaloid harringtonine) can be extracted from leaves, branches, seeds, and roots, which have certain curative effects on leukemia and lymphoid sarcoma (Ni et al., 2016; Xiao et al., 2019). The species is distributed in broad-leaved or coniferous forests to the south of Qinling mountains–Huaihe River line and west of Wuyi Mountains at an altitude of 300–1800 m, and the distribution locations mainly include southeastern and northeastern Yunnan, Guizhou, southern and western Sichuan, northwestern Hubei, Hunan, eastern Jiangxi, and northern Guangdong (Fu et al., 1999). However, its natural population sizes decreased significantly in the recent years due to overexploitation, deforestation, and climate change. This species has been listed as vulnerable in the IUCN Red List. Thus, effective strategies are important to ensure the conservation and scientific utilization of C. oliveri. EST-SSRs have been widely developed and applied to numerous coniferous species. Zhang et al. (2015) developed nine EST-SSRs of Larix gmelinii and used them to evaluate the genetic parameters and transferability to three related species. Zeng et al. (2018) developed 11 EST-SSRs of T. grandis, which revealed a moderate level of genetic diversity and two different genetic groups within this species. Li et al. (2021) also investigated the genetic variation and population structure of 20 EST-SSRs. All these studies proved that EST-SSRs are effective molecular markers for studying the genetic diversity of conifers with high transferability. Although, hitherto, few SSRs (Pan et al., 2011; Miao et al., 2012) and ISSRs (Wang et al., 2016) have been developed and applied, there have been no reports on EST-SSR markers for C. oliveri, even for Cephalotaxaceae, using transcriptome data, greatly limiting research on genetic diversity, germplasm preservation, and molecular breeding of this species.

In this study, we used the leaf transcriptome data by Illumina sequencing from C. oliveri, and the objectives were to 1) characterize frequency, distribution, and function of the SSR motifs from transcriptome unigenes; 2) develop and characterize novel EST-SSRs and examine the level of polymorphism; 3) analyze the cross-species transferability of the polymorphic EST-SSRs; and 4) explore the genetic diversity and structure of C. oliveri by polymorphic EST-SSRs.

Currently, transcriptome sequencing through the Illumina platform is the most widely utilized NGS technology for EST-SSR marker development in non-model plants, particularly in conifers with large genomes (Zalapa et al., 2012; Postolache et al., 2014; Vieira et al., 2016). In this study, a total of 36446 unigenes were used to detect SSRs, and finally 4352 unigenes containing 5089 SSRs were identified. The distribution frequency was 11.94%, which was similar to that of C. hainanensis (11.39%) (Qiao et al., 2014) and P. chienii (11.15%) (Xu et al., 2020), higher than that of P. bungeana (9.21%) (Duan et al., 2017), Torreya grandis (2.75%) (Zeng et al., 2018), A. argotaenia (7.68%) (Ruan et al., 2019), and P. koraiensis (6.84%) (Li et al., 2021), lower than that of Lycium barbarum (27.93%) (Chen et al., 2017), Dalbergia odorifera (23.31%) (Liu et al., 2019), and peony (20.38%) (He et al., 2020). The average density of SSRs was 1/11.1 kb, lower than that of C. hainanensis (1/8.08 kb) (Qiao et al., 2014), Glyptostrobus pensilis (1/7.59 kb) (Li et al., 2019), and P. chienii (1/9.18 kb) (Xu et al., 2020), while higher than that in P. dabeshanensis (1/23.08 kb) (Xiang et al., 2015), P. koraiensis (1/17.38 kb) (Du et al., 2017), and L. principis-rupprechtii (1/26.8 kb) (Dong et al., 2018). The differences in frequency and density might be caused by several factors, including dataset size, SSR mining tools, search criteria, and genome structure (Varshney et al., 2005).

A microsatellite locus typically varies in length between 5 and 40 repeats, but longer strings of repeats are possible (Selkoe and Toonen, 2006). In this study, the repeat units of 10, 5, 6, 11, and 12 accounted for 77.42% of the total SSR loci, with the size mainly ranging from 10 to 18 bp. The predominant type was trinucleotide repeats (26.19%), followed by dinucleotide repeats (15.35%) and hexanucleotide repeats (4.30%). This result was similar to that of pervious reports that trinucleotide repeats were the abundant type for other conifers, including C. hainanensis (Qiao et al., 2014), A. argotaenia (Ruan et al., 2019), P. koraiensis (Li et al., 2020), and P. chienii (Xu et al., 2020). Among the trinucleotide repeats, the most abundant motif was AAG/CTT, which was identical to previous findings in Cryptomeria japonica (Ueno et al., 2012), P. halepensis (Pinosio et al., 2014), L. gmelinii (Zhang et al., 2015), T. grandis (Zeng et al., 2018), A. argotaenia (Ruan et al., 2019), and P. chienii (Xu et al., 2020). In addition, this motif was the second abundant motif in C. hainanensis (Qiao et al., 2014) and P. dabeshanensis (Xiang et al., 2015). These results showed that AAG/CTT was conserved in conifers. Of the dinucleotide repeats, AT/AT and AG/CT exhibited high frequency. However, like most conifers, including T. contorta (Majeed et al., 2019) and P. koraiensis (Li et al., 2020), the CG/CG motif was found to be rare in this study. It might be due to methylation of cytosine in CpG sequences, which might potentially inhibit transcription (Gonzalez-Ibeas et al., 2007; Xing et al., 2017).

Some studies have shown that SSRs are much more abundant in the UTRs than in the CDSs of many plants (Li et al., 2004). In the present study, 17.85% of SSRs were found to be located in CDSs in contrast to 82.13% in UTRs. Moreover, trinucleotide repeats were mostly accumulated in the CDSs and other types had a small proportion. This result was consistent with those of the studies in T. contorta (Majeed et al., 2019), Elaeagnus mollis (Liu et al., 2020), and P. chienii (Xu et al., 2020). This might explain that non-trinucleotides negatively selected frameshift mutations in coding regions, contributing to changes in SSR length, and therefore to expression. In contrast, trinucleotides did not generate frameshifts and did not affect gene expression through single-motif length mutations (Metzgar et al., 2000; Taheri et al., 2019).

All SSR-containing unigenes in leaves of C. oliveri were annotated through seven functional databases. 86.72% of these unigenes were annotated in at least one database, which was similar to the annotation result of all unigenes in leaf (He et al., 2021). In the GO classification, most of SSR-containing unigenes were categorized in the cellular process, metabolic process, single-organism process, cell part, cell, binding, and catalytic activity, which are involved in the basic metabolism in the plant cells. Furthermore, KEGG and KOG classification suggested that SSR-containing unigenes were involved in growth and development of C. oliveri with various biological functions.

In this study, among the 238 pairs of randomly designed primers, 127 (53.36%) were of expected size bands. Forty-seven (19.75%) were identified as polymorphic among 14 individuals from seven C. oliveri sampling sites. This polymorphism rate was higher than the results reported for T. grandis (10.38%) (Zeng et al., 2018), P. koraiensis (6.67%) (Li et al., 2020), G. pensilis (13.53%) (Li et al., 2019), and P. chienii (10.67%) (Xu et al., 2020), but lower than the results of P. dabeshanensis (Xiang et al., 2015) (23.17%) and L. principis-rupprechtii (21.67%) (Dong et al., 2018). In conifers with huge genome structure, most EST-SSRs might locate within repetitive DNA, which caused EST-SSRs difficult to be amplified and the polymorphism rate to be low (Wagner et al., 2012).

The transferability rate of the markers corresponds to genomic similarity and can reflect the genetic relationships and extent of sequence conservation between species (Zhang et al., 2014). EST-SSR markers are expected to have a high transferability rate due to conservation of transcribed regions among related species (Kalia et al., 2011). In this study, the transferability of 28 EST-SSRs from C. oliveri to C. fortune was 96.43%, confirming that C. oliveri had a closer relationship with C. fortune than the other two species. This transferability rate was consistent with that of Taxodium 'zhongshansa' (100%) (Cheng et al., 2015), while higher than that of Abies alba (75–81%) (Postolache et al., 2014) and P. chienii (70%) (Xu et al., 2020). Significantly, the transferability rate was more than 70% both for A. argotaenia and P. chienii belonging to Taxaceae, indicating that there may be a high genomic similarity between C. oliveri and Taxaceae species. These results indicated that the markers developed in this study would provide a powerful molecular tool for evolutionary adaptation and genetic relationship analysis in C. fortune, A. argotaenia, and P. chienii.

Genetic diversity is an important indicator in the conservation and management of plant genetic resources (Chen et al., 2018).In this study, genetic parameters were detected in 134 individuals by the 28 EST-SSRs. The mean PIC value was 0.345, indicating a mean level of genetic information, and similar results were observed in C. japonica (0.325) (Ueno et al., 2012), A. argotaenia (0.455) (Ruan et al., 2019), and P. koraiensis (0.404) (Li et al., 2021). Mean A (4.214) was higher than that reported for T. grandis (2.636) (Zeng et al., 2018) and L. principis-rupprechtii (3.850) (Dong et al., 2018), but lower than that for P. koraiensis (6.45) (Li et al., 2020) and P. chienii (6.4) (Xu et al., 2020). The average of Ho and He were 0.279 and 0.382, respectively, which was lower than that for P. dabeshanensis (0.445 and 0.486) (Xiang et al., 2015) and L. principis-rupprechtii (0.487 and 0.490) (Dong et al., 2018). These results revealed that the 134 individuals had a moderate level of genetic diversity, which might be caused by small population size and fragmentation; meanwhile, habitat heterogeneity might restrict gene flow between populations and had an impact on genetic diversity (Wang et al., 2014). However, this study on genetic diversity of C. oliveri showed a lower level than that in the previous study employing gSSRs (Ho = 0.570, He = 0.568) (Miao et al., 2012). These differences might be due to the different types and numbers of genetic markers used in the studies or to the different populations and sample sizes (Tong et al., 2020). Furthermore, the genetic diversity of the population level was lower than that at the species level, which might be due to limited numbers of SSR loci and small population size.

The distribution geographic range, habitat types, and species characteristics may have a significant influence on the population structure, as well as genetic diversity (Jia et al., 2016; Li et al., 2019). A previous report found that 22 populations of C. oliveri were clustered into two groups by constructing a dendrogram, and the YNdws population (Yunnan Province) was clustered into the separated group (Wang et al., 2016). In this study, STRUCTURE and PCoA analyses also revealed a distinct genetic population (site PB) among the seven sampling sites of C. oliveri. The main reason was that site PB was in the southern edge of the distribution area, as the most geographically distant population, which limited the gene flow between site PB and other sites. However, detailed reasons for the population structure need to be further investigated with an increased number of population and materials.

According to the results of genetic diversity and population structure of C. oliveri, we gained the following management implications and recommendations: for the sites (SX, LP) with higher genetic diversity, in situ conservation strategies should be designed to protect the habitats from human disturbance and prevent the loss of genetic diversity. Ex situ conservation programs should be carried out for the sites (EMS, PB) with lower genetic diversity in order to increase the species numbers of these sampling sites and improve their adaptability to the environment. Furthermore, special attention should be given to site PB because of its isolation from the other sampling sites; it is necessary to save the seeds of this site for breeding.

In this study, 5089 EST-SSRs were identified from transcriptome data of C. oliveri, and the distribution, frequency, location, and function of motifs were characterized and evaluated., Twenty-eight EST-SSR markers were developed for C. oliveri with abundant polymorphisms and showed a moderate level of genetic diversity. Genetic structure and PCoA analysis revealed two different genetic groups of natural C. oliveri. In addition, more than 70% of these EST-SSRs could be transferred to other species. These EST-SSRs would enable further genetic investigation in C. oliveri and related species and could be used to ensure effective conservation and breeding applications of C. oliveri in the future.