MEPS 634:213-223 (2020)  -  DOI: https://doi.org/10.3354/meps13202

ABSTRACT: Accurately assessing trends in wildlife populations is often an ultimate goal of conservation-oriented monitoring so that conservation measures can be adopted to halt and ultimately reverse decline in population numbers. Confidence and power in detecting trends in wildlife populations are highly dependent on the biology of the target species and measurement error resulting from the survey design. We investigated the impacts of sampling errors on the power and magnitude of trend detection, using long-term surveys of 51 populations of sea turtles comprising 5 different species. We show that errors of ca. 20% in annual abundance estimates have relatively little impact on the number of years required to detect trends or the magnitude of detectable trends (for example, for a 5% change, there was an additional 2 yr for green turtles, 3 yr for leatherback turtles and 4 yr for hawksbill, flatback and loggerhead turtles). An error of ca. 20% is associated with sampling regimes with approximately 5% survey coverage. Given the naturally high inter-annual variability in nesting abundance, we suggest survey effort may be better spread across years rather than within years and may be optimised by coupling several different monitoring techniques (such as count surveys with capture-mark-recapture surveys) to optimise information on count trends and other parameters relating to the functioning of the population.

KEY WORDS: Abundance · Sampling · Turtle · Population size estimation · Trends · Accuracy · Error · Precision