AMPHIBIAN CONSERVATION IN HUMAN SHAPED ENVIRONMENTS: LANDSCAPE DYNAMICS, HABITAT MODELING AND METAPOPULATION ANALYSES

摘要: Global biodiversity is experiencing a worrying decline. Habitats destruction, associated to their degradation and fragmentation are among the greatest causes. Amphibians particularly interesting because they more threatened decline rapidly than either birds or mammals. In this context, objective of our research improve some methodological approaches offer practical scientific bases for decision making in landscape management amphibian conservation. Our study focuses on fragmented Swiss landscapes. We developed method that uses land-cover data expert knowledge enable spatially explicit assessment 1) temporal changes nature conservation value 2) rehabilitation potential landscape. applied Geographical Information Systems (GIS) based approach Rhone plain we used years 1900 as reference state. The constitutes helpful tool communication, decision-making biological planning. Effective optimal species strategies can only be formulated after relationships between distribution environmental factors have been identified. Concerning amphibians, several exist but generally suffer from two limitations: spatial autocorrelation (i.e. dependency observations) rarely explicitly analyzed, even if it may affect accuracy species-habitat models. showed how measured included logistic models with example agile frog (Rana dalmatina) north-eastern Switzerland. Moran's I autologistic model including measure arrangement response variables). found not considered, then conclusions incorrect. effect occurrence ponds often assumed equal every direction (isotropic). However, barriers inhospitable surfaces reduce movement patterns area around accessible species. This implies ideal circular has reality shape depending surrounding determine habitat variables distribution, considering physical ponds. studied species: common toad (Bufo bufo) temporaria) plain. demonstrated reducing boundaries following barriers, allowed compute predictors which better explained distribution. These results suggested proposed pertinent traditional buffers analysis. stress necessity consider ecological corridors order avoid incorrect inferences. Species usually established single region. It unknown whether identified directly transferred another geographical area. examined landscape-level six by measuring correlations presence 655 five different regions. analyzed using information-theoretic Akaike criterion (AIC). For out species, best predict site occupation probability depends suggest caution needed when predictions done built other geographic also observed connectivity was explicative variables. addition, scale at affected varied pond km Management should conducted taking into account characteristics multiple scales. Finally, landscape, separating patches metapopulation models, considered population viability analyses. explored patch occupancy sensitive Euclidean (shortest) versus landscape-based distance (least-cost). found: theoretical standpoint, inter-patch affects occupancy; standpoints, priority management. illustrated case metapopulations Yellow-bellied Toad successfully doubtlessly extended any functioning defined basis (e.g. pond, nesting place, den …), structured structure during movement. By improving combining approaches, likely provide wildlife managers tools valuable making.