With an increasing number of habitats becoming fragmented, we are faced with the task of conserving a patchy world. Recent metapopulation theory has focused on the size and isolation of remaining habitat patches. However, the observation that habitat fragmentation often goes hand in hand with habitat deterioration has led some researchers to emphasize habitat quality above the spatial configuration of remaining patches. As a result, managers have had to choose between two seemingly contrasting recommendations: "preserve optimal habitat quality wherever it can be found" as opposed to "preserve a large number of well-connected patches."

In a new paper, Jeremy Thomas et al. [1] reconcile these diverging views. They show that both the configuration and quality of habitat patches affect the distribution of three British grassland butterfly species that differ in their habitat requirements. Thomas et al. first defined the exact preference of each species for subtle details of the habitat. They then looked for the presence of each species in each patch of suitable habitat within its full British range (two species) or within its British stronghold (one species). For all three species, both the isolation of a site, and the habitat quality within it, were important in explaining which patches supported a local population.

The general lesson for practical conservation is clear and important: the informed manager should not make a choice between either maintaining high habitat quality within individual sites or retaining as many interconnected populations as possible, but should strike a balance between the two. But where the balance lies is an open question. For the three butterflies, Thomas et al. found habitat quality to be two or three times better as a predictor of butterfly presence than was isolation, whereas patch area failed to show any detectable effect. Before attempting to generalize this specific finding, we should keep in mind that neither "quality" nor "size" nor "isolation" is easily defined. Hence, a biologically suspect measure might well make one parameter seem less important than another. Thomas et al. measured isolation simply as the distance to the nearest suitable habitat patch. A better measure would reflect the total number of individuals likely to immigrate to this site from existing local populations anywhere in the system, and such a definition might further emphasize the importance of isolation [2]. Second, the relative importance of size, isolation, and quality will depend on the specifics of the system. Size will have a larger effect in a system where patch size is heterogeneous compared with a system where all patches are approximately the same size. Third, the importance of habitat quality will probably vary with geographical location. The effect of habitat quality is generally assumed to be accentuated in populations near the edge of the species range. Nevertheless, habitat quality can usefully be thought of as a third parameter in metapopulation dynamics. Its inclusion can add realism to metapopulation modeling, which is vital for the conservation of our rarest species.