Local Recruitment in the Greater Flamingo: A New Approach Using Capture- Mark-Recapture Data

Abstract

Although the establishment of new individuals in the breeding component of a population is an essential feature of population regulation, only a few attempts have been made to test biological hypotheses about recruitment. Most previous studies rely on ad hoc calculations or are flawed with unwarranted assumptions about survival. We use a recently developed approach, based on capture-mark-recapture, in which analysis of local recruitment is similar to a time-reversed analysis of survival. The basic data set consists of capture histories viewed in reverse order, with initial capture at year of birth, and subsequent observations corresponding to years when the animal has bred. The model considers two essential components, the probability for any breeding individual to reproduce for the first time (␤, the probability of first reproduction) and the probability of recapture (p), both conditional on survival. Contrary to previous attempts at modeling recruitment, the present approach does not assume an age at which breeding propensity stabilizes to a maximum value. The flexibility achieved allows the comparison of recruitment among groups within a population and also allows one to consider the effects of environmental variables, as well as interactions between such effects. Practically, the procedure starts from a global model, based upon the a priori knowledge of the biology of the species, and assesses its fit. Then more parsimonious models are selected using Akaike's Information Criterion and likelihood ratio tests. Finally, maximum likelihood estimates of model parameters are obtained with estimates of precision. We used a modified version of program RELEASE for goodness-of-fit tests, and program SURGE for iterative model fitting and the computation of likelihood ratio tests. We illustrate the method with the study of local recruitment of Greater Flamingos (Phoenicopterus ruber roseus) in the Camargue (southern France) between 1984 and 1994. We found additive effects of age and year to affect recruitment. Breeding propensity increased with age. Recruitment was noticeably higher in the year following an increase in mortality rate due to a particularly severe winter. Longlasting effects of this increased mortality on recruitment were observed in the three following years. There was no evidence for an effect of sex or cohort (year of birth) on recruitment. However, sex, as well as time and age, affected recapture rates. We discuss the various advantages and limitations of the model for the study of local recruitment in long-lived species and mention some potential developments.

Comparing demographic parameters for philopatric and immigrant individuals in a long-lived bird adapted to unstable habitats

“…The specific components of the GOF test pointed out the presence of transient individuals, a consistent result with previous analyses of the same population Tavecchia et al 2007). This lack of fit can be corrected including extra parameters for newly observed birds or transients (Pradel et al 1997), a model noted /*P(t*a)// P(t*a)//*I(t*a)//I(t*a)/pP(a*t)/pI(a*t) (Model 1, Table 2). However, there was evidence of a large residual deviance (v 377 2 = 572.59, P \ 0.01), partly due to the fact that a subset of individuals were systematically observed more than others, a phenomenon known as trap-dependence.…”

“…c i = probability of being present in the population at i -1 for an individual known to be alive at time i, seniority probability hereafter (see Pradel 1996). This is a capture-recapture procedure to estimate recruitment in a population (see also Pradel et al 1997). …”

“…Tavecchia et al (2007) showed that most of these birds did not die but rather dispersed to other colonies. From a methodological point of view, these transients are treated by differentiating the first survival from survival in subsequent years (Pradel et al 1997). Pradel et al (1997) have shown that, at population level, initial survival probability is related to the proportion of transients, s, as:…”

“…We modelled seniority probability, c, by reversing the encounter histories used to estimate survival: seniority probability was the probability that an individual was present before the date of the first resighting (Pradel et al 1997). The complement of c is the probability of recruitment, i.e.…”

“…the probability that an individual was 'new' when first seen. The GOF of the general model c*P(t*a)/cP(t*a)/ c*I (t*a)/cI(t*a)/pP(t*a)/pI(t*a) was taken as the starting model (Pradel et al 1997); hence, we corrected the deviance with the same scale parameter.…”