Optimal design of butterfly occupancy surveys and testing if occupancy converts to abundance for sparse populations

Occupancy has several important advantages over abundance methods and may be the best choice for monitoring sparse populations. Here we use simulations to evaluate competing designs (number of sites vs. number of surveys) for occupancy monitoring, with emphasis on sparse populations of the endangered Karner blue butterfly (Lycaeides melissa samuelis Nabokov). Because conservation planning is usually abundance-based, we also ask whether detection/non-detection data may reliably convert to abundance, hypothesizing that occupancy provides a more dependable shortcut when populations are sparse. Count-index and distance sampling were conducted across 50 habitat patches containing variably sparse Karner blue populations. We used occupancy-detection model estimates as simulation inputs to evaluate primary replication tradeoffs, and used peak counts and population densities to evaluate the occupancy-abundance relationship. Detection probability and therefore optimal design of occupancy monitoring was strongly temperature dependent. Assuming a quality threshold of 0. 075 root-mean square error for the occupancy estimator, the minimum allowable effort was 360 (40 sites × 9 surveys) for spring generation and 200 (20 sites × 10 surveys) for summer generation. A mixture model abundance estimator for repeated detection/non-detection data was biased low for high-density and low-density populations, suggesting that occupancy may not provide a reliable shortcut in abundance-based conservation planning for sparse butterfly populations.