Optimizing sampling design to infer the number of marine turtles nesting on low and high density sea turtle rookeries using convolution of negative binomial distribution

Reliable monitoring of wildlife populations represents a non-negligible cost, and in a limited-resource world, resources allocated to monitoring are not devoted to actions to solve identified problems.I explore resource efficient survey designs based on a negative binomial distribution including variable survey intervals for marine turtles using track counts as an index of female activity. In the modified procedure, all new tracks between two monitoring patrols are recorded. These data are analyzed by statistical models that take advantage of the statistical properties of the sum of counts.The outputs of models with different lagged monitoring dates (3–10 days) are compared with the outputs of daily surveys using extrapolations from high and low density populations. Results show that the quality of the estimates is similar when total time series analysis is compared with situations in which only a fourth, a seventh, or a tenth of monitoring daily during the season are used.This solution permits the reallocation of funds from monitoring to other conservation activities. Furthermore, the efficient sampling design and the statistical methods allow getting similar information with less effort.