How to find a metapopulation

Abstract

Where habitat loss and fragmentation is severe, many native species are likely to have reduced levels of dispersal between remnant populations. For those species to avoid regional extinction in fragmented landscapes, they must undergo some kind of metapopulation dynamics so that local extinctions are countered by recolonisation. The importance of spatial dynamics for regional survival means that research into metapopulation dynamics is essential. In this review I explore the approaches taken to examine metapopulation dynamics, highlight the analytical methods used to get the most information out of field data, and discover some of the major research gaps. Statistical models, including Hanski's INCIDENCE FUNCTION model (IFM) are frequently applied to presence-absence data, an approach that is often strengthened using long-term data sets that document extinctions and colonisations. Recent developments are making the IFM more biologically realistic and expanding the range of situations for which the model is relevant. Although accurate predictions using the IFM seem unlikely, it may be useful for ranking management decisions. A key weakness of presence-absence modelling is that the mechanisms underlying spatial dynamics remain inferential, so combining modelling approaches with detailed demographic research is warranted. For species where very large data sets cannot be obtained to facilitate statistical modelling, a demographic approach alone or with stochastic modelling may be the only viable research angle to take. Dispersal is a central process in metapopulation dynamics. Research combining mark-recapture or telemetry methods with model-selection procedures demonstrate that dispersal is frequently oversimplified in conceptual and statistical metapopulation models. Dispersal models like the island model that underlies classic metapopulation theory do not approximate the behaviour of real species in fragmented landscapes. Nevertheless, it remains uncertain if additional biological realism will improve predictions of statistical metapopulation models. Genetic methods can give better estimates of dispersal than direct methods and take less effort, so they should be routinely explored alongside direct ecological methods. Recent development of metacommunity theory (communities connected by dispersal) emphasises a range of mechanisms that complement metapopulation theory. Taking both theories into account will enhance interpretation of field data. The extent of metapopulation dynamics in human modified landscapes remains uncertain, but we have a powerful array of field and analytical approaches for reducing this knowledge gap. The most informative way forward requires that many species are studied in the same fragmented landscape by applying a selection of approaches that reveal complementary aspects of spatial dynamics.