Fig. S3. Bootstrapping to calculate CV estimates and relate them to release year shows variability across locations
CV was calculated using sampling with replacement from five measurements for each variety that was measured at least five times. Linear regression was performed on CV values for release year. The p-value was determined for comparing that slope to 0 using a t-test. This was repeating 1000 times and the density of p-values is shown. The red vertical lines indicate p = 0.05. In a-c) results are shown across all locations. In a) the p-value distribution is shown. In b) the distribution of R2 values is shown, and in c) the distribution of the slope estimates across the 1000 bootstraps is shown. In d-i) the p-value distributions are shown for the six locations.