The clustering of spatially associated species unravels patterns in tropical tree species distributions
Open Access
Abstract
Complex distribution data can be summarized by grouping species with similar or overlapping distributions to unravel spatial patterns and separate trends (e.g., of habitat loss) among spatially unique groups. However, such classifications are often heuristic, lacking the transparency, objectivity, and data‐driven rigor of quantitative methods, which limits their interpretability and utility. Here, we develop and illustrate the clustering of spatially associated species, a methodological framework aimed at statistically classifying species using explicit measures of interspecific spatial association. We investigate several association indices and clustering algorithms and show how these methodological choices drive substantial variations in clustering outcomes and performance. To facilitate robust decision‐making, we provide guidance on choosing methods appropriate to one's study objective(s). As a case study, we apply our framework to modeled tree distributions in Borneo and subsequently evaluate the impact of land‐cover change on separate species groupings. Based on the modeled distribution of 390 tree species prior to anthropogenic land‐cover changes, we identified 11 distinct clusters that unraveled ecologically meaningful patterns in Bornean tree distributions. These clusters then enabled us to quantify trends of habitat loss tied to each of those specific clusters, allowing us to discern particularly vulnerable species clusters and their distributions. This study demonstrates the advantages of adopting quantitatively derived clusters of spatially associated species and elucidates the potential of resultant clusters as a spatially explicit framework for investigating distribution‐related questions in ecology, biogeography, and conservation. By adopting our methodological framework and publicly available codes, practitioners can leverage the ever‐growing abundance of distribution data to better understand complex spatial patterns among species distributions and the disparate effects of global changes on biodiversity.Keywords
All Related Versions
This publication has 152 references indexed in Scilit:
- An Update of Wallace’s Zoogeographic Regions of the WorldScience, 2013
- Likelihood analysis of species occurrence probability from presence‐only data for modelling species distributionsMethods in Ecology and Evolution, 2012
- Soils on exposed Sunda Shelf shaped biogeographic patterns in the equatorial forests of Southeast AsiaProceedings of the National Academy of Sciences, 2011
- Investigating habitat-specific plant species pools under climate changeBasic and Applied Ecology, 2010
- Distorted Views of Biodiversity: Spatial and Temporal Bias in Species Occurrence DataPLOS Biology, 2010
- Selecting pseudo-absence data for presence-only distribution modeling: How far should you stray from what you know?Ecological Modelling, 2009
- Species richness, hotspots, and the scale dependence of range maps in ecology and conservationProceedings of the National Academy of Sciences, 2007
- Silhouettes: A graphical aid to the interpretation and validation of cluster analysisJournal of Computational and Applied Mathematics, 1987
- A mathematical model in graph theory for biological classificationJournal of Theoretical Biology, 1966
- Hierarchical Grouping to Optimize an Objective FunctionJournal of the American Statistical Association, 1963