CNLs and TNLs
We did not find a significant relationship between CMI and TNL counts (p = 0.695, Figure 5D, MSregion = 14.072, MSerror = 86.559, F = 0.1626. Degrees of freedom from Satterthwaite’s method: dfregion = 1, dferror = 10.215), and only a marginal significance with CNL counts (Figure 5C; p=0.0634, MSregion = 651.19, MSerror = 150.22, F = 4.3349. Degrees of freedom from Satterthwaite’s method: dfregion = 1, dferror = 10.22).
Due to having only one plant sampled per prairie, we report the results of a fixed-effect linear regression model for the PacBio dataset. Despite a much lower inferential power (n=9), we still found a positive correlation between CMI and the PacBio R genes (Figure 5B; p=0.019). There was a significant relationship between CMI and PacBio TNL counts (Figure S1B; p=0.018), but not CNLs (Figure S1A; p=0.14).