Fig. 3

Observed and model-predicted relationship between prevalence and allelic richness. Black circles represent data for each of the 115 clusters. Red shaded regions indicate the 95, 80 and 50% predictive intervals of the best-fitting model, and the red line represents the median prediction. The best fitting model also includes a sample size term, and so these predictions were generated assuming a sample size of 14 (the median sample size in the data)