Thinking outside the patch: a multi-species comparison of conceptual models from real-world landscapes

Abstract

Abstract

Context When modeling a species’ distribution, landscapes can alternatively be conceptualized following patch- or gradient-based approaches. However, choosing the most suitable conceptualization is difficult and methods for empirical validation are still lacking.

Objectives To address the conditions under which a given conceptual model is more suitable, taking into account landscape context and species trait dependency effects. Patch- and gradient-based conceptualizations were built based on two structurally different landscapes: variegated and mosaic. We hypothesize that: (H1) gradient-based models better describe variegated landscapes while patch-based models perform better in mosaic landscapes; and (H2) gradient-based models will fit generalist species better while patch-based models will suit specialists better.

Methods We modeled the distribution of eleven bird species in each landscape using each conceptualization. We determined the suitability of each conceptual model to fit statistical models by looking for cross-species responses and deviations from best models.

Results We found no clear support for our hypotheses. Although patch-based models performed better in mosaic landscapes (H1), they also provided useful conceptualizations in variegated landscapes. However, when patches showed high heterogeneity, gradient-based approaches better fit specialist species (H2).

Conclusions The suitability of a given conceptual model depends on the interaction between species habitat specialization, and the intrinsic spatial heterogeneity of the landscape and the ability of each conceptualization to capture it. Gradient-based models provide better information on resource allocation, while patch-based models offer a simplified perspective on landscape attributes. Future research should consider the nature of both species and landscapes in order to avoid bias from inadequate landscape conceptualizations.