DYNAMICS AND VIABILITY OF A METAPOPULATION OF THE ENDANGERED IBERIAN LYNX (LYNX PARDINUS)

The use of metapopulation models in conservation biology is growing ex ponentially, but there is a need for empirical studies that support th eoretical approaches, especially for species with large and long-lived individuals. In this paper we explore the viability and dynamics of a real metapopulation of an endangered mammal by combining field work a nd modeling in order to support conservation decisions and evaluate th eoretical approaches. The Iberian lynx (Lynx pardinus), considered the most vulnerable felid in the world, is restricted to the Iberian Peni nsula in southwestern Europe, The persistence of the species is handic apped by the high fragmentation of its populations. Fewer than 1000 in dividuals are distributed in nine isolated populations, each of them a lso fragmented but with their patches connected by dispersers, in what could be called metapopulations. One of these metapopulations, includ ing similar to 60 individuals, inhabits the Donana National Park (DNP) and its surroundings. Demographic and behavioral data gathered over o ne decade for this lynx population were employed to develop a spatiall y realistic structured model with density-dependent fecundity and migr ation, including demographic and environmental stochasticity. Such a m odel is used to identify the demographic features that determine the d ynamics of this population and to predict its risk of decline under a set of alternative assumptions. A hypothetical lynx metapopulation wit h values of the parameters such as those observed in Donana, but witho ut stochastic events, could sustain itself over time. Results of this deterministic model show how females occupy all the potential breeding territories, while males are below the carrying capacity. The metapop ulation has a source-sink structure, with the sources internal and the sinks external to the national park. Sinks result from reduced surviv al rather than reduced fecundity, as generally assumed. High mortality in sink patches is deterministic, deriving both from within-patch ris ks and from factors related to the landscape matrix among patches. The survival rate of adults with territories in the sources was the most sensitive parameter, leading the dynamics of the metapopulation. When we include demographic stochasticity in the model, the population beco mes extinct 22% of the time within 100 yr, and this value increases to 33.8% when environmental stochasticity is also considered. Most of th e metapopulation extinctions occurred because of the disappearance of males due to sex differences in demographic parameters related to beha vioral aspects (e.g., dispersal rate). Different scenarios were simula ted as modifications affecting either within- or between-patch dynamic s. Changes in the carrying capacity of source and sink patches would h ave very different consequences in terms of metapopulation persistence : one breeding territory increase in the largest source reduces metapo pulation extinction risk from 33.8 to 17.2% in 100 yr, while an increa se of three territories in the largest sink does not modify the extinc tion risk. In this sense, results suggest that the best management str ategy for conservation should be restoring habitat at the source patch es and reducing mortality at the sinks. The results of our models emph asize the need for empirical studies to characterize metapopulations i n nature and distinguish between such terms as source-sink, mainland-i sland, nonequilibrium, or even ''refuge'' metapopulations.