Introduction

Habitats vary in their temporal and spatial stability and, as Southwood (1962, 1977) recognised, this has important consequences for the evolution of migration in insects. The spatiotemporal structure of the habitat, he argued, would act as a ‘template’ selecting individuals of migratory phenotype appropriate to the prevailing conditions. Based on this assumption, Southwood predicted that insects occupying habitats whose ‘favourableness’ varied in time and/or space would have relatively high migratory potential, whereas those in more permanent, continuous habitats would have relatively low migratory potential, due to costs generally assumed to be associated with the ability to migrate (Rankin & Burchsted, 1992). Between-species comparisons within a number of insect orders generally support this prediction: migratory species tend to be associated with ‘temporary’ habitats, non-migratory with ‘permanent’ ones (Southwood, 1962; Johnson, 1969; Roff, 1990a; Denno et al., 1991).

However, comparative studies that fail to take account of phylogenetic and other constraints on evolution can generate misleading results (Harvey & Pagel, 1991; but see Denno et al., 1991 for an excellent use of the comparative method). An alternative approach to the problem is to make within-species comparisons of populations in environments differing in their spatiotemporal stabilities, the assumption being that the observed levels of migratory potential will reflect the prevailing levels of habitat heterogeneity. However, there is also a problem associated with this method – that the predicted response to habitat heterogeneity may be obscured by the mechanisms regulating migratory potential.