Missouri Botanical Garden Open Conference Systems, TDWG 2014 ANNUAL CONFERENCE

Font Size: 
Using GBIF data to test niche vs. neutrality theories at a continental scale, and the value of data cleaning
Tomer Gueta, Avi Bar-Massada, Yohay Carmel

Building: Elmia Congress Centre, Jönköping
Room: Rydbergsalen
Date: 2014-10-28 12:00 PM – 12:15 PM
Last modified: 2014-10-04

Abstract


The processes underlying patterns in the diversity, abundance, and composition of species within a given landscape are under rugged debate known as ‘niche vs. neutrality’. The continuum hypothesis states that rather than being mutually exclusive, niche and neutral theories are located at the two ends of a continuum, and therefore species assembly in communities is an outcome of both niche and neutral processes. The question is what drives the location of communities along this continuum? Modeling studies suggested that species richness is a main determinant, with species-rich communities being closer to the neutral end while species-poor communities are driven more strongly by niche processes. Here, we expand this notion by asking if species richness can affect the relative roles of niche versus neutral processes in driving species geographic ranges. This study uses the massively accumulated species occurrence data available in GBIF to test the continuum hypothesis in the context of species ranges at the continental scale. Specifically, we focused on mammal species in Australia, and asked if species richness affects the relative role of niche processes in generating species distributions.

The distribution of species is driven by environmental conditions as well as biotic interactions. Here, we suggest that by quantifying the relationship between species richness and the role of environmental variables in determining species ranges, we can gain insight about the relative roles of niche versus neutral processes in driving species ranges. We assume that the performance of a species distribution model (SDM) is a proxy for the strength of environmental factors affecting the distribution of a given species. We divided Australia into multiple geographical subsets characterized by varying species richness levels. We obtained over one million records of mammalian occurrence data in Australia from the Australian GBIF node. In each subset, we generated SDMs for each species using Maximum Entropy Modeling (MaxEnt), based on 24 environmental and climatic predictive variables, and analyzed the relationship between species richness of a geographical subset and the performance of SDMs of multiple species. To assess model performance, we quantified the training ‘gain’ value (a measure of goodness of fit) for each model. In each subset, we calculated 12 species richness values, based on 12 different species groupings representing optional guilds, using different biological characteristics (taxon, trophic level and body weight). For species that were present in a sufficient number of subsets, we performed a Spearman rank correlation test between species richness values and the MaxEnt gain values. To evaluate the effect of data cleaning on our results, we conducted the entire analysis twice, before and after data cleaning.

Results suggest that the notion of species-environment affinity is weaker in species-rich areas. The role of environmental factors in determining species distribution decreased with species richness, in line with the prediction of the neutral theory.  The pattern was stable across different geographical subsets and different potential guilds. The pattern became consistently stronger after data cleaning procedures were implemented; this strongly proclaims the value of data cleaning.