Introduction

The spatial distribution of individuals is influenced by bottom-up and top-down effects, and by processes both in the local habitat and at larger landscape scales (Strong 1979; Tscharntke and Brandl 2004; Ryall and Fahrig 2006). Patterns of resource distributions are often primary determinants of consumer abundance. For example, herbivore densities vary with the size of host plant patches, with the density of plant individuals and with the presence of other, non-host, plant species in the neighbourhood (Andow 1991; Bender et al. 1998; Bowman et al. 2002; Hambäck and Beckerman 2003). Based on observations of these spatial patterns, Richard Root formulated the resource concentration hypothesis (RCH) more than 35 years ago (Root 1973) and predicted that herbivore densities were higher in areas with high resource concentrations. Subsequent research has revealed a more complex picture, where herbivore densities may be either higher or lower in resource dense areas and where patterns may vary with the spatial scale of the study (Bowers and Matter 1997; Bender et al. 1998; Bowman et al. 2002; Bommarco and Banks 2003; Hambäck and Englund 2005). The limited predictive capacity of RCH can be explained by the incompleteness in the formulation of underlying mechanisms (Bukovinszky et al. 2005; Hambäck and Englund 2005).

Root’s argument was that herbivore densities would be higher in large patches, because immigration rates were higher into and emigration rates were lower from large areas. This argument defines immigration rate as the number of immigrating individuals, not correcting for the fact that a larger number of individuals in large patches will also be distributed across a larger area. A recent paper by Hambäck and Englund (2005) examined RCH mathematically and found a much wider range of predicted density–area relationships. The analysis showed that the main drivers explaining density variation along patch size gradients are (1) the relative role of local versus regional processes and (2) individual search mode. The basic message was straightforward, though hardly surprising, that patterns in large patches are determined by local processes whereas patterns in small patches are determined by regional processes. However, patterns in small patches should also vary with search mode, or specifically with the relative scaling of immigration and emigration rates in relation to patch size.