Analysis of capture-recapture data to study animal population dynamics
2003
Shahrul Anuar Mohd Sah
Estimation of population size from capture-recapture data were calculated using the closed population models inplemented in program CAPTURE which are preferable to Minimum Number Alive (MNA). MNA relies on simple enumeration, accounting for any individuals that miss one trapping session but are subsequently recaptured. It does not provide for any measure of uncertainty by means of a confidence interval, and it is negatively biased under unequal catchability. The closed population models implemented by the program CAPTURE assumed that there are no births, deaths, emigration and immigration during sampling, and they allow for different probabilities of capture: between individuals (Model Mh), between trapping occasions (Model Mt), or according to the trapping history of individuals (Model Mh) or pairwise combinations of models Mh, Mt and Mb. They form the basis of eight different probability models in CAPTURE with different population estimators and confidence limits. Model selection involves two types of tests, the likelihood ratio tests (LRT) with a specific alternative hypothesis (e.g. Model Mo vs. Model Mh) and goodness of fit tests against a general alternative hypothesis (e.g. Model Mh vs. not Model Mh). Tests for the assumption of population closure are also included in the program CAPTURE. The assumption of a closed population model is that there are no births or deaths over the primary trapping session and that every individual has equal catchability. The closure assumption can be tested for all trapping session in our study.
显示更多 [+] 显示较少 [-]