Capture–Recapture Methods and Models: Estimating Population Size. Integrated Population Biology and Modeling, Part B. Elsevier; 2019. pp. 33-83. [DOI: 10.1016/bs.host.2018.09.006]

Finite mixtures in capture-recapture surveys for modeling residency patterns in marine wildlife populations

This work aims to show how prior knowledge about the structure of a heterogeneous animal population can be leveraged to improve the abundance estimation from capture-recapture survey data. We combine the Open Jolly-Seber model with finite mixtures and propose a parsimonious specification tailored to the residency patterns of the common bottlenose dolphin. We employ a Bayesian framework for our inference, discussing the appropriate choice of priors to mitigate label-switching and nonidentifiability issues, commonly associated with finite mixture models. We conduct a series of simulation experiments to illustrate the competitive advantage of our proposal over less specific alternatives. The proposed approach is applied to data collected on the common bottlenose dolphin population inhabiting the Tiber River estuary (Mediterranean Sea). Our results provide novel insights into this population's size and structure, shedding light on some of the ecological processes governing its dynamics.

Avoiding bias in estimates of population size for translocation management

Mark-recapture surveys are commonly used to monitor translocated populations globally. Data gathered are then used to estimate demographic parameters, such as abundance and survival, using Jolly-Seber (JS) models. However, in translocated populations initial population size is known and failure to account for this may bias parameter estimates, which are important for informing conservation decisions during population establishment. Here, we provide methods to account for known initial population size in JS models by incorporating a separate component likelihood for translocated individuals, using a maximum-likelihood estimation, with models that can be fitted using either R or MATLAB. We use simulated data and a case study of a threatened lizard species with low capture probability to demonstrate that unconstrained JS models may overestimate the size of translocated populations, especially in the early stages of post-release monitoring. Our approach corrects this bias; we use our simulations to demonstrate that overestimates of population size between 78% and 130% can occur in the unconstrained JS models when the detection probability is below 0.3 compared to 1%-8.9% for our constrained model. Our case study did not show an overestimate; however accounting for the initial population size greatly reduced error in all parameter estimates and prevented boundary estimates. Adopting the corrected JS model for translocations will help managers to obtain more robust estimates of the population sizes of translocated animals, better informing future management including reinforcement decisions, and ultimately improving translocation success.

[Underreporting and exhaustiveness of tuberculosis surveillance systems in a region of Peru: a capture-recapture analysis]

This study sought to determine the underreporting of tuberculosis (TB) cases in the Cajamarca Region, Peru in 2017 and 2018 through the capture-recapture method and to estimate the TB incidence rate using this method, assess the exhaustiveness of surveillance systems, and describe the epidemiological characteristics of TB in this period. A descriptive, retrospective study was performed in the Cajamarca Region; the surveillance systems were analyzed were the TB Management Information System (SIGTB) and the TB Epidemiological System (SIEPI-TB). Based on the number of cases recorded in each system, the capture-recapture method was applied to obtain an estimation of real cases, validating the data with a log-linear model in the R statistical environment. The largest underreporting rates in the systems were 40.7% in 2017 and 25.6% in 2018; the estimated incidence rates in both years were higher than those reported by the official systems, the most exhaustive surveillance system was SIEPI-TB. The results indicate the existence of worrisome underreporting and the need to monitor the TB surveillance systems.