Federico Riva, Environmental Geography Dept. of the Institute for Environmental Studies (IVM),Vrije Universiteit Amsterdam, discusses his review article: Incorporating effects of habitat patches into species distribution models
Habitat patches are everywhere – forests surrounded by agriculture, pastures within Alpine forests, or green areas in a city. This patchiness of many ecosystem types is key to if and how species persist in a landscape: some of the most influential theories in the history of ecology use habitat patches as fundamental units, including island biogeography theory and metapopulation theory. From theory, the idea of habitat patches has percolated into environmental management, with concepts such as “minimum patch sizes” for biodiversity conservation. The idea of habitat patches has arguably become one of the most widespread and influential concepts in the environmental sciences.
Since the 1960s, much research has demonstrated that the area, configuration, and diversity of habitat patches affects the distribution of many species. For instance, large patches usually host more species, whether habitat exists fragmented or not can play important roles in moderating species persistence, and landscapes hosting different types of patches also tend to host more species. Therefore, thinking about patches, could help us to better predict the distribution of species for applications in ecology and conservation. Despite these important roles of habitat patches, Species Distribution Models – the most frequently used tool to predict and understand the distribution of species – have been surprisingly disconnected from the science tackling how habitat patches moderate biodiversity patterns. Our review takes on the challenge of formalizing links between the effects of habitat patches (EHPs) and Species Distribution Models (SDMs). Specifically, we identified three major characteristics of habitat patches and of landscapes containing patches – area, configuration, and diversity of habitat – and described how SDMs can be designed to account for these properties
One important aspect that rapidly became clear to the authors, was that incorporating EHPs in SDMs is inextricably tied with issues of spatial scales, particularly the grain (i.e., resolution of raster pixel) at which a specific SDM is designed and fitted. Traditionally, the grain of analysis adopted in most SDMs did not allow for assessing EHPs because environmental covariates were available only at relatively coarse resolutions; nevertheless, the increasing availability of high-resolution remote sensing data is opening unprecedented opportunities to test when EHPs matter in SDMs. However, even among us, we are still having lively discussions about if and when using coarser data might be beneficial, depending on the goal of a SDM application. In other words, it is not clear yet whether smaller resolutions are always better.
We believe that EHPs can be meaningfully incorporated into SDMs, and that this is possible when using the following four steps: (i) choose a proper conceptual representation of the landscapes studied, (ii) select metrics representative of important processes for the species of interest – with particular attention to properly discriminate between characteristics of patches and characteristics of landscapes, (iii) measure such metrics, and (iv) evaluate statistical support for an EHP.
Beyond increasing our ability to accurately predict the distribution of species, prospects for incorporating EHPs in SDMs include important contributions to ongoing conversations in ecology and conservation (e.g., when are small habitat patches useful for biodiversity conservation?). Challenges are related to both theoretical aspects (e.g., when are patches meaningful entities?) and methodological aspects (e.g., how do methods developed in other sub-fields translate to SDMs, such as the “scale of effect” approach?). In our review, we discuss some solutions, e.g., selecting relevant climatic variables as a first step, and testing for different EHPs as a second step to avoid correlation among multi-scale metrics of EHPs. We hope that our review will stimulate the community to come up with approaches to address current limitations.
Ultimately, we believe that there are endless opportunities to incorporate EHPs in SDMs. Efforts in this direction are sorely needed because they will allow advancements in ecological theory and in development of optimal environmental management actions. We hope that our review will facilitate advances at the intersection of several research programs that attempt to understand biodiversity change (e.g., landscape ecology, macroecology, and biogeography), where SDMs have taken a prominent role.
