Genome size and β-diversity
The ranges of genome size for the species that occurred at each study site increased from west to east in our transect, with the maximum genome size for each study site also increasing from west to east (Fig. 1). Precipitation, soil nitrogen, and soil phosphorus (Fig S2) were all relatively higher at the eastern than that from the western end, which is consistent with our hypothesis that species with larger genomes would be more frequent where resource availability was greater.
Pairwise species turnover among communities (βturnover) showed a positive correlation with increasing environmental dissimilarity (Fig. 2a), in line with many previous findings (Robroeket al. 2017; Mori et al. 2018). Plant species were then divided into two subsets based on whether their genome was greater or less than 2.5 pg (the median genome size for angiosperms globally (Leitch & Leitch 2013)). Consistent with our hypothesis, environmental variables explained more variation in βturnover for the large-GS than that for the small-GS species (Fig. 1b, d-h; Fig. 2a), while geographic distance explained more variation in βturnover for the small-GS than the large-GS species (Fig. 2b-c). For each specific environmental predictor, altitude, soil N, soil P, and MAP [marginally] explained more variation in βturnover for the large-GS species.
These results indicate that the relative importance (or effects) of environmental heterogeneity and geographic distance (which can indicate, in part, dispersal limitation) on plant β-diversity depend greatly on genome size. To check the sensitivity of our approach for separating species into large- and small-GS groups based on a single threshold (2.5 pg/1C), we repeated the GDM analysis after dividing the species based on the median value of our studied species (1.59 pg/1C). We found similar results, thus we report the results based on 2.5-threshold here in the main text and that of the 1.59-threshold in Fig S5. In addition, for variation partitioning, we also assessed the sensitivity of our results to four different thresholds in genome size [Fig. 3; median and mean values for the present study (1.59 and 3.19 pg) or for global angiosperms (2.5 and 5.90 pg) (Leitch & Leitch 2013)]. We observed qualitatively similar results, with the environment explaining the most variation for the large-GS species and geographic distance explaining the most variation for the small-GS species (Fig. 3a-c). The thresholds differed mainly in how variation was partitioned for species classified to the large genome size category, with environment explaining more variation and geographic distance explaining less variation as the threshold increased, indicating that the species with the largest genomes were most strongly impacted by environmental filtering (Fig. 3).
We found spatial distances explained less variation in β-diversity for the large-GS species. In an attempt to link genome size and dispersal capacity among plant species, we also included the primary dispersal mode for the dominant species that we observed (Dataset S1). This includes genera that contain multiple species with large genomes, such as Allium and Agropyron , both of which tend to have an animal dispersal mode in seed dispersal. Previous studies have shown that genome size is positively correlated with seed mass (Beaulieuet al. 2007). Thus, lower dispersal limitation for the large-GS species might be due to their production of larger seeds and having a better chance of attracting dispersers. While confirmation requires further work, the associations that we have uncovered make the study of genome size a potentially powerful tool for understanding dispersal patterns in driving plant community assembly.