Patterns in species aggregation across spatial scales and species abundances

Background/Question/Methods

The study of patterns in species distributions has long been a major focus of ecology.  One metric of particular interest is the species-level spatial abundance distribution (SSAD), the distribution of a species’ abundance in quadrats sampled from a large landscape, which characterizes the degree of aggregation in a species’ distribution. Knowledge of the shape of the SSAD across scales is often important for the construction of species-area relationships, the prediction of species’ abundance from presence-absence data, and the derivation of other similar metrics important to conservation.

The negative binomial distribution is a widely used and flexible model for characterizing the shape of the SSAD, as the fitted value of the clustering parameter k for any species at any spatial scale can be used as a measure of the degree of aggregation exhibited by individuals of that species at that scale. Here we examine the best-fit value of k in seven ecosystems (four tropical forests, a temperate forest, a serpentine grassland, and a desert) across a variety of spatial scales, uncover evidence for systematic trends in this parameter, and explore the implications of these trends.

Results/Conclusions

Across all data sets, we find that k increases with species abundance, quadrat area, and mean species abundance per quadrat approximately as a power law with a slope of 0.5. For many species, at many scales, values of k cluster near 1, suggesting a geometric distribution for the SSAD that matches the predictions of the recent Maximum Entropy Theory of Ecology. A random placement model (k -> infinity) is a poor fit for most examined species and scales. 

We suggest that the observed trends in species aggregation across scales may be a candidate for a universal macroecological pattern that should be used to guide the further development of macroecological theory. Although inferring ecological process from pattern alone is problematic, we note that a negative binomial SSAD can arise from a simple quadrat-scale mechanistic model of birth, death, and immigration. This simple model suggests that the observed relationship between k and area can arise naturally when dispersal distances are small relative to quadrat area, while the observed relationship between k and species abundance would require a positive relationship between species abundance and dispersal distance.