Estimating abundance of a recovering transboundary brown bear population with capture-recapture models

Estimating the size of small populations of large mammals can be achieved via censuses, or complete counts, of recognizable individuals detected over a time period: minimum detected (population) size (MDS). However, as a population grows larger and its spatial distribution expands, the risk of underestimating population size using MDS rapidly increases because the assumption of perfect detection of all individuals in the population is violated. The need to report uncertainty around population size estimates consequently becomes crucial. We explored these biases using the monitoring framework of the critically endangered Pyrenean brown bear that was close to extinction in the mid-1990s, with only five individuals remaining, but was subsequently bolstered by the introduction of 11 bears from Slovenia. Each year since 1996, the abundance of the population has been assessed using MDS and minimum retained (population) size (MRS), which corresponded to a reassessment of the MDS in the light of the new information collected in subsequent years (e.g., adding bears which were not detected the previous years but detected the current year). We used Pollock's closed robust design (PCRD) capture-recapture models applied to the cross-border non-invasive sampling data from France, Spain and Andorra to provide the first published annual abundance and temporal trend estimates of the Pyrenean brown bear population since 2008. Annual population size increased fivefold between 2008 and 2020, going from 13 to 66 individuals. PCRD estimates were globally close to MRS counts and had reasonably narrow associated 95% Credibility Intervals. Even in cases where sampling effort is large compared to population size, the PCRD estimates of population size can diverge from the MDS counts. We report individual heterogeneity in detection that might stem from intraspecific home range size variation that result in individuals that move the most being most likely to be detected. We also found that cubs had a higher mortality rate than adults and subadults, because of infanticide by males, predation, maternal death, or abandonment. Overall, the PCRD capture-recapture modelling approach provides estimates of abundance and demographic rates of the Pyrenean brown bear population, together with associated uncertainty, while minimizing bias due to inter-individual heterogeneity in detection probabilities. We strongly encourage wildlife ecologists and vital for informing management decision-making and assessing population conservation status.