Genes under the selection and driven by the environment
Despite being closely related, CN and NW differ in morphology and ecology (Figure S10). To explore the genetic mechanism of the differentiation, we identified 1168 outlier SNPs from the 143,318 filtered SNPs according to a 0.05 threshold for the q-value between the CN and NW lineages based on the Bayesian method applied in BAYESCAN (Figure 4A). The outlier SNPs were in 487 genes that were identified under positive selection (Table S8). The FST values of outlier SNPs were significantly higher than those of others, which suggests that the divergence of the lineages from their ancestral population was possibly caused by strong divergent selection. GO enrichment analyses of positively selected genes were significantly enriched in ABC-2 type transporter family protein (GO:0016887, p= 0.0151), and we then integrated the candidate genes with organ development, reproductive isolation, biotic and abiotic stress responses. Among these genes , several (e.g., PAD2 (Parisy et al., 2007), DRB3 (Mehdi, Reza, Hassan, Shima, & Gholamreza, 2022) and EDM2 (Eulgem et al., 2007)) are involved in plant immunity, whereas some are associated with the stress of drought (e.g.,MAPKKK21 (L. Wu, Chang, Wang, Wu, & Wang, 2021) and SCL1(Manohara Reddy et al., 2018)) and cold (e.g., CER3 (Rahman et al., 2021) and LOS1 (Bielsa, Ávila-Alonso, Fernández i Martí, Grimplet, & Rubio-Cabetas, 2021)). Moreover, two genes (GRF2(Beltramino et al., 2018) and PHT4;2 (Irigoyen, Karlsson, Kuruvilla, Spetea, & Versaw, 2011)) participate in the regulation of leaf size, and several candidate genes are associated with flowering time (e.g., FT (Tyagi et al., 2018), FLK (Lim et al., 2004), MYB30 (L. Zhu et al., 2020) and FY (Henderson, Liu, Drea, Simpson, & Dean, 2005)), flower organ size(KNU (Bollier et al., 2018) and CKX5 (Bartrina, Otto, Strnad, Werner, & Schmülling, 2011)), flower color (MYB113 (Jiao, Zhao, Wu, Song, & Li, 2020)) and pollen development (WLIM1 (Yang et al., 2022) and EXO70A2 (Marković et al., 2020)). The rapid divergence of candidate genes with reproductive function might drive prezygotic isolation across lineages.
Additionally, the Mantel test revealed a significant effect of IBD on the genome-wide SNPs of the A. viridiflora complex (Figure 4B). Because genetic divergence can result from selection driven by heterogeneous environments via geographic distance, redundancy analysis (RDA) was then implemented to examine the bioclimatic factors related to the 143,318 SNPs. The contribution of seven environmental variables in RDA space is shown in Figure S11. The strongest predictor was the mean diurnal range (Bio2), followed by isothermality (Bio3), mean temperature of the wettest quarter (Bio8), temperature seasonality (Bio4), precipitation seasonality (Bio15), annual mean temperatures (Bio1) and precipitation of the driest quarter (Bio17). BAYENV2 identified 179 SNPs significantly associated with the 7 examined environmental variables (Figure 4C). The outlier SNPs were in 83 genes and significantly enriched in “PEBP (phosphatidylethanolamine-binding protein) family protein” (GO:0003712, 2 genes, p = 0.034), which was previously known to be involved in various physiological processes, such as seasonal growth (Khosa, Bellinazzo, Kamenetsky Goldstein, Macknight, & Immink, 2021), seed germination (B. Zhang, Li, Li, & Yu, 2020) and floral transition (Y. Zhu, Klasfeld, & Wagner, 2021). In addition, seven genes were adaptively differentiated under temperature and water stress, indicating that the crosstalk between the two modules observed might be important in the process of local adaptation (Table S9).