This plot shows the same model output as in Fig. 2, but with detections defined more conservatively, as samples with at least two technical replicates successfully amplifying. Top: probability of detecting eDNA from EAB using the tree core sampling method at sites in New Jersey, USA over time. Bottom: the number of ash tree core samples needed per tree to ensure 95% confidence in detection of EAB larval presence within a tree. Time is represented as growing degree day accumulation, base 50 °F (10 °C). Black lines represent the posterior medians from a Bayesian general additive model; grey shading indicates 80% and 95% credible intervals. Black circles show the proportion of samples with detections on each sampling date calculated from the raw data. The GAM model was fitted and used default vague priors in accordance with Bürkner (2017).