Michael M. Fuller

Spatial Analysis in Community Ecology

The goal of spatial ecology is to understand the processes that govern the geographic patterns of species, communities, and habitats. Species often occur as clusters of individuals of the same species, or recurring groups of species. Sometimes there are clear geographic delineations between species groups (often caused by distinct habitat preferences). But patterns of coexistence frequently vary with spatial scale and for some groups (e.g. forest trees) can be indistinguishable from random sorting.

A broad range of geostatistical tools are available for quantifying spatial patterns. For example, the semivariogram (figure at right) depicts the strength of correlation among sites, as the distance between sites increases. In the figure, tree density is correlated up to a distance of about 70 meters, beyond which density varies randomly. The lines bracketing the points are Monte Carlo envelopes indicating the region of uncorrelated data. This figure was generated using the open-source statistical package, R.

omnidirectional variogram

In addition to my geographic analysis of the invasive Eurasian collared dove in North America, my spatial work also includes:

Example of Spatial Analysis: Tree Species Density

The figures below show the distribution of dominant species in a Costa Rican dry forest (left), and the density of stems within the same forest (right). Here, we used a kernel density smoothing technique to generate contour lines of tree abundance.

3species plot density plot

Using density analysis helped us to visualize the geographic pattern of trees, which gave us a better understanding of how topography influenced species distributions within the commmunity. Previous research indicated that the distribution of species was influenced by random processes, such as dispersal, as well as deterministic processes, such as habitat preferences and assymetric competition. To better understand the relative contribution of such processes, I worked with Brian Enquist to develop a randomization algorithm, which redistributes tree species while preserving the large-scale spatial structure of the forest. The algorithm leaves the 3-dimensional organization of the forest plot intact when generating (simulated) random plots for comparison.

In addition to the above research, I worked with several people affiliated with the ITR research group at the University of Tennessee on spatial modeling approaches. For example, we looked at how geographic variation in local climate can influences the spread of invasive species. Some information on these studies can be found on my projects page.

Contact Information
The Institute for Environmental Modeling