Missing Data in Sea Turtle Population Monitoring: A Bayesian Statistical Framework Accounting for Incomplete Sampling

Abstract

Monitoring how populations respond to sustained conservation measures is essential to detect changes in their population status and determine the effectiveness of any interventions. In the case of sea turtles, their populations are difficult to assess because of their complicated life histories. Ground-derived clutch counts are most often used as an index of population size for sea turtles; however, data are often incomplete with varying sampling intensity within and among sites and seasons. To address these issues, we: (1) develop a Bayesian statistical modelling framework that can be used to account for sampling uncertainties in a robust probabilistic manner within a given site and season; and (2) apply this to a previously unpublished long-term sea turtle dataset (n = 17 years) collated for the Republic of the Congo, which hosts two sympatrically nesting species of sea turtle (leatherback turtle [Dermochelys coriacea] and olive ridley turtle [Lepidochelys olivacea]). The results of this analysis suggest that leatherback turtle nesting levels dropped initially and then settled into quasi-cyclical levels of interannual variability, with an average of 573 (mean, 95% prediction interval: 554–626) clutches laid annually between 2012 and 2017. In contrast, nesting abundance for olive ridley turtles has increased more recently, with an average of 1,087 (mean, 95% prediction interval: 1,057–1,153) clutches laid annually between 2012 and 2017. These findings highlight the regional and global importance of this rookery with the Republic of the Congo, hosting the second largest documented populations of olive ridley and the third largest for leatherback turtles in Central Africa; and the fourth largest non-arribada olive ridley rookery globally. Furthermore, whilst the results show that Congo’s single marine and coastal national park provides protection for over half of sea turtle clutches laid in the country, there is scope for further protection along the coast. Although large parts of the African coastline remain to be adequately monitored, the modelling approach used here will be invaluable to inform future status assessments for sea turtles given that most datasets are temporally and spatially fragmented.

From terra incognita to hotspot: the largest South Pacific green turtle nesting population in the forgotten reefs of New Caledonia

“…The observers separated the tracks into four categories: from the previous night, < 4 days old, < 1 month old and > 1 month old. However, based on current knowledge, tracks older than 10 days are generally no longer visible (Omeyer et al, 2022). We therefore recoded these data to take this into account.…”

“…Quantifying the nesting of marine turtles has become the preferred tool for evaluating their status (Delcroix et al, 2014; Laloë et al, 2020; Omeyer et al, 2022). This approach considers the fact that some nights are not monitored during a season (missing intra-season data) whilst providing information at the scale of the nesting season (phenology) and between nesting seasons (trend).…”

Computation of the Distribution of the Sum of Independent Negative Binomial Random Variables

“…The sum of independent NB variables is of special interest in different contexts, such as the study of animal distribution [4,5], fecal egg counts in infected goats [6], the number of emergency medical calls [7], empirical distribution of the duration of wet periods in days [8] or insurance risk [9]. When the sum of several independent NB counts is available, determining the distribution of ∑ X i with X i ∼ NB(r i ; p i ) is a problem.…”

“…Our solution for computation of the convolution method, presented here, is novel and has proven to be robust for extensive testing. A naïve tolerance condition has been used by one of the authors of this note (MG) (recursion stops when the change is lower than the tolerance limit) as in [4,5], but the other author (JB) found that outputs can be strongly biased in some conditions. It has been the beginning of a collaboration between the two authors to understand and solve the origin of this bias.…”

Conservation‐relevant reproductive parameters from long‐term tagging of two globally important sea turtle nesting aggregations in Central Africa

“…Nest monitoring has been ongoing for nearly 2 decades, with particular focus on Conkouati‐Douli National Park, which until recently was the only coastal and marine park in the country, created in 1999. The Park covers approximately 60 km, corresponding to 35% of the country's coastline, and hosts more than 50% of all D. coriacea and L. olivacea nesting in the Republic of the Congo (Omeyer et al, 2022). As of 2022, Conkouati‐Douli has been extended and two further MPAs have been created in Mvassa and Loango Bay to protect important sea turtle foraging and developmental habitat (Metcalfe et al, 2020).…”