Designing monitoring programs that can infer species absence: how much effort is enough?
Date:
Summary
A key issue in the design of monitoring studies is how to balance the trade-off between the number of monitored locations and the effort invested at each site. A single report is typically enough to be sure the site is occupied, but what if multiple visits lead to no detections? Knowing where species are absent may be as important as knowing where they are present. As such several sets of guidelines have been proposed to recommend the number of successive non-detections required in order to have confidence that a site is truly unoccupied.
We discuss two types of prior information (species detectability and expected prevalence) that can be used to design occupancy studies with the optimum trade-off, and compare two of the main frameworks that are used to design occupancy studies. We use data from 12 reptile species to demonstrate differences between these 2 frameworks, and illustrate the sensitivity of each framework to the quality of prior information available.
Although a-priori expectations of species prevalence influence the likelihood of detecting a species, we argue that in most cases it is not necessary to consider this when designing monitoring. In many cases the differences between the frameworks do not differ substantially, and for hard to detect species, both methods suggest that an unfeasible amount of effort be expended at each site to confirm its true state. More problematically, for rare species, considering species prevalence at the design stage would lead to a monitoring program which erroneously classifies an unknown number of occupied sites as unoccupied. We conclude that in most cases, more simplistic guidelines are to be preferred- not only do they not advocate giving up on rare species, but also the uncertainties surrounding the likelihood of detecting a species can be easily understood and communicated to managers and decision-makers.