HYDROTHERMAL‐VENT ALVINELLID POLYCHAETE DISPERSAL IN THE EASTERN PACIFIC. 2. A METAPOPULATION MODEL BASED ON HABITAT SHIFTS

Marine organisms typically fall into two main categories: those with a high level of population structuring and those with a low one. The first are often found to be poor dispersers, following isolation by distance or stepping‐stone theoretical predictions. The second are commonly associated with high‐dispersal taxa and are best described by the island model. Deep‐sea hydrothermal vent systems represent a good model for studying one‐dimensional metapopulations. Whereas isolation by distance might be expected to be the rule in such a system for species with limited dispersal capabilities, a biological paradox can be observed: an apparent genetic homogeneity in some vent species with short‐scale dispersal potential, in a one‐dimensional fragmented habitat. This can be explained if one key assumption of the existing models is not met: gene flow between populations and genetic drift may not have the time to equilibrate. Geophysical models revealed that hydrothermal convection is intrinsically unstable, inducing processes of coalescence or splitting of venting areas in a chaotic manner. This is likely to generate frequent extinctions and recolonizations. Theoretical genetic predictions derived from extinctions/recolonizations cannot satisfactorily model a situation where habitat shifts are frequent and constantly affect the metapopulation equilibrium. Because neither the island and the stepping‐stone models nor the classical metapopulation models resemble the hydrothermal vent reality, we present here a realistic model developed to provide a compromise between existing conceptual models and what is currently known of the biology and ecology of one of the most peculiar and best‐studied vent species, the polychaete Alvinella pompejana . This model allows us to define the boundaries between which the metapopulation is evolutionary stable in an unstable context. Simulations show different patterns in which metapopulation size and recolonization vary but reach an equilibrium despite chaotic vent extinctions. In contrast, the model also shows that displacing habitat continuously affects the equilibrium between gene flow and drift. As a consequence, the time required to balance these evolutionary forces can never be attained, leading to chaotic fluctuations in F ‐statistics. Those fluctuations are mainly due to stochastic changes of the interpatch distance which affect migration rates. The shifting of active zones of venting can episodically counterbalance differentiation and allow a long‐term genetic homogenization at the ridge scale. These findings lead to a new concept in which the exchanges between populations would mainly depend on the habitat's movements along the ridge axis rather than the organim's dispersal. We therefore propose a new model based on patch‐network displacements in which transient contact zones allow low levels of gene flow throughout the metapopulation.