Phenotypic variation and broad-sense heritability
In most individual environments (location and year combinations), plant vigor in the hybrid population exhibited left skewed distributions whereas the distributions in the selfed population were slightly right skewed (Fig. 1 and Fig. 2). Normally distributed patterns or right skewed distributions were showed for plant biomass in the hybrid population. Large phenotypic variation existed for spring green-up in the two populations. Genotype mean values spanned a wide range in all environments (e.g., 1-100% in 2015PKS; Table 1). Population means also varied considerably across environments, ranging from 35 to 81% and 17 to 44% in the hybrid population and the selfed population, respectively (Fig. 1, Fig. 2 and Table 1). For plant vigor, substantial variations were observed among genotypes within each environment, whereas population means were generally similar across environments (6-8 in hybrid population, 3-4 in selfed population; Table 1). For plant biomass, obvious heterosis and selfing depression for the two populations can be found crossing all individual environments (Table 1).
ANOVA indicated that significant effects of genotype, year, and location × year for spring green-up, plant vigor and plant biomass in both populations (Table 2). Significant genotype × location interactions were detected for plant vigor in both populations and plant biomass in the selfed population, but slightly significant genotype × location interactions was showed in the hybrid population. Significant genotype × year interactions were found for spring green-up in both populations and marginally significant (P = 0.0036 in the selfed population) or non-significant (P = 0.0518 in the hybrid population) for plant vigor (Online Resource 1). However, non-significant genotype × year interactions were detected for plant biomass in both populations.
Correlations among spring green-up, plant vigor, and plant biomass in the hybrid and selfed populations are presented in Table 2. Only data collected in 2012 and 2013 were used in correlation analysis as plant biomass data were not available in 2014 and 2015. The correlation between spring green-up and plant biomass was marginally positive (0.09) in the selfed population while it was negative (-0.07) in the hybrid population (Table 2). Plant vigor was positively correlated with plant biomass in both the hybrid population (0.30) and the selfed (0.26). Between spring green-up and plant vigor, significant positive correlation was found in the selfed population (0.31), while non-significant correlation was found between these two traits in the hybrid population (Table 2).
Broad-sense heritabilities of spring green-up and plant vigor were estimated under both single year (i.e., based on single year data across two locations) and joint environments (i.e., based on multiple year data across two locations). For spring green-up, the highest single-year heritabilities were found in 2012, with the value of 0.78 and 0.82 in the hybrid population and selfed population, respectively (Online Resource 2). In subsequent years (2013 to 2015), heritabilities of spring green-up were moderate, ranging from 0.42 to 0.59 in the hybrid population and 0.47 to 0.57 in the selfed population (Online Resource 2). Under joint environments, the heritability of spring green-up was 0.51 in the hybrid population and 0.63 in the selfed population (Table 3). For plant vigor, the highest single-year heritabilities in the two populations were both detected in 2013 with both the value of 0.74 (Online Resource 2). Moreover, the lowest heritabilities of plant vigor in the hybrid population and selfed population were 0.46 in 2015 and 0.45 in 2014, respectively (Online Resource 2). Based on the joint environment analysis, a high heritability (0.80) for plant vigor was observed in the hybrid population, and a moderately high heritability (0.69) was found in the selfed population (Table 3). As the trait plant biomass, single-year heritability ranged from 0.54 to 0.74 in the hybrid and selfed populations (Online Resource 2). For the joint environments, the heritability of plant biomass was 0.63 in the hybrid population and the same in the selfed population (Table 3).
Spring green-up QTLs detection
In the hybrid population, a total of 15 QTLs for spring green-up were detected on seven LGs across individual environments (Table 4). Specifically, four QTLs were discovered in 2012STW, three QTLs in each of 2013PKS, 2014PKS and 2015STW. The QTLs of 2012STW, 2013PKS, 2014PKS, and 2015STW cumulatively explained 29.9, 27.4, 27.5 and 24.4％of the phenotypic variance, respectively. Only one QTL was detected in each of 2012PKS, 2014STW and 2015PKS, explaining 9.6, 11.8 and 9.9％ of the phenotypic variance (Table 4). However, only one QTL was identified on LG 2a in the joint environment analysis (Table 4). The significant QTL between markers sww-532 and nfsg-052 on LG 2a was identified in 2014PKS and 2015PKS as well as the joint environment (Table 4 and Fig. 3). The consistent results in multiple environments indicated that this significant QTL was stably expressed. The effects for the genotypes AD, BC and BD were positive for the QTLs on LGs 2b, 9a and 1a in 2015STW while significant QTLs mapped on LGs 1a, 5a and 6b in 2012STW, 2a in 2014PKS, 2015PKS and the joint environments had negative effects (Table 4).
In the selfed population, four significant QTLs for spring green-up were identified on three LGs based on individual environments and one significant QTL on LG 6b-1 based on the joint environment (Table 5). Two QTLs in 2012STW, one QTL in each of 2012PKS, 2015PKS and 2015STW explained 18.5, 6.3, 7.1 and 7.1％ phenotypic variance, respectively (Table 5). No significant QTL was detected at the two locations in 2013 and 2014 (Table 5). The QTL between PVAAG-3017/3018 and PVGA-1115/1116 on LG 6b-1 was detected in 2012PKS, 2012STW and the joint environment analysis (Table 5 and Fig. 3). The additive effects for spring green-up ranged from -5.76 to 5.96 and only one single QTL located on LG 9a in 2015PKS had the positive effect and other QTLs had negative effects (Table 5).
Plant vigor QTL detection
Six significant QTLs for plant vigor with LOD scores from 3.81 to 6.01 and one QTL with a LOD value of 4.7 were discovered in individual environments and the joint-environment in the hybrid population (Table 4). In the selfed population, four QTLs from individual environments were detected with LOD values from 3.54 to 4.77 and one QTL from the joint-environment analysis with a LOD value of 4.39 (Table 5). The phenotypic variance was explained from 7.7 to 12.6％ in the hybrid population, and from 5.5 to 8.3％in the selfed population (Table 4 and Table 5). Among all the significant QTLs in the hybrid population, the QTL between PVGA-1813/1814 and PVGA-1357/1358 on LG 5a was stably expressed in 2014STW, 2015PKS and the joint environment, which accounted for 34.5％ of the total phenotypic variance, and the additive effects of the three genotypes AD, BC and BD for this QTL ranged from -0.65 to 0.93 (Table 4 and Fig. 3). The QTL between PVCA-815/816 and nfsg-50 were also detected in three environments, 2013PKS, 2014STW and the joint environment, accounting for 18.7％ phenotypic variance in the selfed population, with the additive effects ranging from 0.03 to 0.14 (Table 5 and Fig. 3).
Plant biomass QTL detection
In the hybrid population, interesting findings showed in individual environments and joint-environments, all of the significant QTLs were focused on LG 5a with the LOD values ranging from 3.75 to 5.69 (Fig. 3). Identical QTLs (e.g., the QTL between PVGA-1971/1972 and PVCAG-2197/2198) and adjacent QTLs (e.g., the QTL between PVCAG-2389/2390 and PVCAG-2167/2168 and the QTL between PVCAG-2167/2168 and PVGA-1971/1972) were found between plant biomass and spring green-up or plant biomass and plant vigor. In the selfed population, only one QTL between sww-1622 and sww-2501 was found on LG 2a with the LOD value 4.15 (Fig. 3).
Correspondence between linkage groups and switchgrass genome
High correspondence was found between the nomenclature of genetic linkage groups (a and b) and the subgenome designations (N and K) in the switchgrass genome (Online Resource 3). For both populations, clear correspondences are as follows: LG 1a – Chr 01K, LG 1b – Chr 01N, LG 2a – Chr 02N, LG 2b – Chr 02K, LG 3a – Chr 03K, LG 3b – Chr 03N, LG 4a & 4b – Chr 04N and 04K, LG 5a – Chr 05K, LG 5b – Chr 05N, LG 6a – Chr 06N (the selfed population only), LG 6b – Chr 06K, LG 7a – Chr 07K, LG 8a – Chr 08N, LG 8b – Chr 08K, LG 9a – Chr 09K, LG 9b – Chr 02N. However, LG 6a in the hybrid population largely matched with Chr 07N, indicating the lack of marker coverage on this group. Moreover, LG 7b only contained 3 and 4 markers in the hybrid and selfed population, respectively, and these markers did not correspond to the expected Chr 07N. This is also likely due to the low marker density.