Landform has long been considered one of the primary controls on forest community structure; however, it is still unclear how the strength of such species-topography couplings varies in response to recurring disturbance events. We evaluated this question in the context of repeated prescribed fire applied to a forest landscape in eastern Kentucky, USA. The landscape encompassed different areas of varying treatment frequencies: no fire, less frequent fires (two times, 6-year interval), and frequent fires (four times, 1 or 2-year interval) over 8 years. For each of 32 plots (10 m × 40 m), species-landform data were collected seven times between 2002 and 2010. Results of canonical correspondence analysis showed generally decreasing importance of terrain attributes to the overall tree species composition, during and after the period of fire disturbance. Before fire, species composition showed high fractal dimensions and low Moran's I, implying that the complex terrain of the area provided local, site-specific topographic controls on community structure. As fire was repeatedly applied, fractal dimensions decreased and Moran's I increased, indicating that plots possessed increasingly similar vegetation characteristics regardless of site-specific terrain conditions; that is, local topography no longer acted as the primary driver of species composition because such a short-range spatial control became overwhelmed by a longer-range variation dictated by fire. Following a period of fire disturbance, forest modelers are recommended to avoid species distribution modeling heavily based on topographic parameters and to explicitly take into account potentially increasing spatial autocorrelation in species composition.
|Number of pages
|Published - Jul 2014
Bibliographical noteFunding Information:
Acknowledgments This study (#14-09-006) is connected with a project of the Kentucky Agricultural Experiment Station. The research was partially funded by the USDA-USDI Joint Fire Science Program (01-3-3-14, 04-2-1-06) through cooperative research agreements with the Southern Research Station, McIntire-Stennis Funds administered by the Kentucky Agricultural Experiment Station, and a Challenge Cost-Share Agreement with the Daniel Boone National Forest. We thank our USDA Forest Service collaborators, including E.J. Bunzendahl, Jeff Lewis, and David Loftis, among others. This study could not have been conducted without the field and laboratory support of many people, especially Gretchen Sovkoplas, Autumn Foushee, Jessi Lyons, Heather Alexander, Elizabeth Loucks, Beth Blankenship, Stephanie Green, and many others. Jonathan Phillips and Soohyun Jung provided constructive comments on an earlier version of this paper.
- Canonical correspondence analysis
- Fractal dimension
- Spatial autocorrelation
- Species distribution modeling
ASJC Scopus subject areas
- Ecology, Evolution, Behavior and Systematics