Estimating dispersal in spatiotemporally variable environments using multievent capture-recapture modeling - Archive ouverte HAL Access content directly
Journal Articles Ecology Year : 2018

Estimating dispersal in spatiotemporally variable environments using multievent capture-recapture modeling

(1, 2) , (2) , (1) , (3) , (2)


Dispersal is a key process in ecological and evolutionary dynamics. Spatiotemporal vari- ation in habitat availability and characteristics has been suggested to be one of the main cause involved in dispersal evolution and has a strong influence on metapopulation dynamics. In recent dec- ades, the study of dispersal has led to the development of capture–recapture (CR) models that allow movement between sites to be quantified, while handling imperfect detection. For studies involving numerous recapture sites, Lagrange et al. (2014) proposed a multievent CR model that allows disper- sal to be estimated while omitting site identity by distinguishing between individuals that stay and individuals that move. More recently, Cayuela et al. (2017) extended this model to allow survival and dispersal probabilities to differ for the different types of habitat represented by several sites within a study area. Yet in both of these modeling systems, the state of sites is assumed to be static over time, which is not a realistic assumption in dynamic landscapes. For that purpose, we generalized the multi- event CR model proposed by Cayuela et al. (2017) to allow the estimation of dispersal, survival and recapture probabilities when a site may appear or disappear over time (MODEL 1) or when the char- acteristics of a site fluctuate over space and time (MODEL 2). This paper first presents these two new modeling systems, and then provides an illustration of their efficacy and usefulness by applying them to simulated CR data and data collected on two metapopulations of amphibians. MODEL 1 was tested using CR data recorded on a metapopulation of yellow-bellied toads (Bombina variegata). In this first empirical case, we examined whether the drying-out dynamics of ponds and the past dispersal status of an individual might affect dispersal behavior. Our study revealed that the probability of facul- tative dispersal (i.e., from a pond group that remained available/flooded) fluctuated between years and was higher in individuals that had previously dispersed. MODEL 2 was tested using CR data collected on a metapopulation of great crested newts (Triturus cristatus). In this second empirical example, we investigated whether the density of alpine newts (Ichthyosaura alpestris), a potential competitor, might affect the dispersal and survival of the crested newt. Our study revealed that the departure rate was lower in ponds with a high density of heterospecifics than in ponds with a low density of heterospeci- fics at both inter-annual and intra-annual scales. Moreover, annual survival was slightly higher in ponds with a high density of heterospecifics. Overall, our findings indicate that these multievent CR models provide a highly flexible means of modeling dispersal in dynamic landscapes.
Not file

Dates and versions

hal-01797399 , version 1 (22-05-2018)



Hugo Cayuela, Roger Pradel, Pierre Joly, Eric Bonnaire, Aurélien Besnard. Estimating dispersal in spatiotemporally variable environments using multievent capture-recapture modeling. Ecology, 2018, 99 (5), pp.1150-1163. ⟨10.1002/ecy.2195⟩. ⟨hal-01797399⟩
132 View
0 Download



Gmail Facebook Twitter LinkedIn More