Fig. 3.

False discovery rate (FDR) estimates from a mixture of independent and dependent presence-absence vectors. In 3 separate scenarios with null proportions π₀=.25,.50,.75, 2000 presence-absence vectors of m=200 are simulated with occurrence probabilities of p=.5. Each simulation scenario is repeated 20 times and the proposed methods are used to automatically compute p-values and q-values. FDR thresholds are plotted against observed false discovery proportions, where a downward deviation from a theoretically correct diagonal red line indicates a conservative behavior