Transect-based trapping for area-wide delimitation of insects

Typical delimitation trapping survey designs for area-wide (nonlocalized) insect populations are regularly spaced grids, and alternative shapes have not been evaluated. We hypothesized that transect-based designs could give similar detection rates with significantly shorter servicing distances. We used the TrapGrid model to investigate novel "trap-sect" designs incorporating crossed, spoked, and parallel lines of traps, comparing them to a regular grid, in single survey and multiple-site scenarios. We calculated minimum servicing distances and simulated mean probabilities of detecting a pest population, judging overall performance of trap network designs using both metrics. For single sites, trap-sect designs reduced service distances by 65-89%, and most had similar detection probabilities as the regular grid. Kernel-smoothed intensity plots indicated that the best performing trap-sect designs distributed traps more fully across the area. With multiple sites (3 side by side), results depended on insect dispersal ability. All designs performed similarly in terms of detection for highly mobile insects, suggesting that designs minimizing service distances would be best for such pests. For less mobile pests the best trap-sect designs had 4-6 parallel lines, or 8 spokes, which reduced servicing distances by 33-50%. Comparisons of hypothetical trap-sect arrays to real program trap locations for 2 pests demonstrated that the novel designs reduced both trap numbers and service distances, with little differences in mean nearest trap distance to random pest locations. Trap-sect designs in delimitation surveys could reduce costs and increase program flexibility without harming the ability to detect populations.