R Gene Counts Increase with Increasing Climate Moisture Index
We found a strong positive association between R gene count and CMI in both the Illumina and PacBio datasets (Figure 5A,B; Table S3) from the simple linear model. We also found a significant positive correlation between Illumina R gene counts and CMI using the mixed effect model (MSCMI = 9493, MSerror = 1059.6, F = 8.96. Degrees of freedom from Satterthwaite’s method [Satterthwaite 1946]: dfCMI = 1, dferror = 9.73, p = 0.0139, Figure 5A; Table S4), on top of substantial variation among source prairies. Specifically, the model predicted an additional 0.707 kg/m2/month (with a standard error of 0.236 kg/m2/month) R genes for every additional unit of CMI. As average CMI ranged from -57.3 units to -12.9 units, the model predicts that plants from prairies with the highest annual CMI will have 31.4 more R genes than those from prairies with the least CMI. Because most variation in CMI is associated with source region (West, Central, or East), we also modeled source region, rather than CMI as a fixed effect. This analysis revealed a similar, but nominally insignificant association between source region and the number of R genes (MSregion = 8753, MSerror = 4376.5, F = 4.13. Degrees of freedom from Satterthwaite’s method: dfregion = 2, dferror = 8.82, p = 0.0542, see descriptive statistics in the section above and in Table S4).
Our PacBio sequencing recovered more R genes per individual and had relatively fewer R genes annotated as “partial” (relative to “complete”) than the Illumina data (Table S2). Despite much lower power, we still see a significant association between CMI and R gene count in the PacBio data set (MSCMI = 92006, MSerror = 9961, F = 7.193, dfCMI = 1, dferror = 7, p = 0.0315, Figure 5B). Like the case for the Illumina data modeling R gene count from the PacBio data revealed a similarly consistent but nominally insignificant association (MSregion = 53074 MSerror = 10164, F = 4.96. Degrees of freedom dfregion = 2, dferror = 6, p = 0.054, see descriptive statistics in the section above and in Table S4).