Multi-scale response of a North Carolina piedmont forest 15 years after Hurricane Fran

Background/Question/Methods

Wind disturbance can cause patches of major change in forest vegetation structure due to loss in tree basal area and canopy cover.  This can lead to a rapid change in diversity of the forest floor due to increased light penetration and soil disturbance from tip-up mounds.  Ruderal and light-dependent woody species may take advantage of this influx of resources, and can alter forest diversity as secondary succession takes place. In 1990, 19 permanent, 0.1 hectare vegetation plots in forested uplands of the North Carolina Botanical Gardens were established, recording all species as well as DBH for woody species in a nested plot design.  In September of 1996 the eye of Hurricane Fran passed seven kilometer east of these plots causing severe damage. These plots were resampled the following summer in 1997 and again in 2011.  Plots lost an average of 9.1% of their basal area due to hurricane damage, and showed considerable variation ranging from 48.9% loss to 32.9% gain.  Damaged plots were expected to show an immediate increase in species richness with a less evident signal 15 years post-hurricane.  This was also investigated in four 100m² subplots within each plot to account for the patchiness of the disturbance.

Results/Conclusions

One-sample t-tests were used to determine change in richness from the 1990 sampling event in plots that decreased in basal area (damaged) and plots that increased in basal area (undamaged) following the hurricane.  Undamaged plots showed no significant increase in richness (1997: df=6, t=0.09, p=.47; 2011: df=6, t=-1.88, p=0.95).  Damaged plots showed a significant increase in richness in 1997 (df=10, t=6.42, p<0.001), but not in 2011 (df=10, t=0.33, p=0.38).  100m² subplots were analyzed and separately assigned damage classifications based on change in basal area, thus allowing within plot variation.   Based on these subplots, 1997 data showed a significant increase in richness in both damaged (df=29, t=2.36, p<0.05) and undamaged subplots (df=41, t=2.52, p<0.01).  2011 subplots showed no significant increase in either damaged (df=29, t=-2.95, p=.99) or undamaged plots (df=41, t=-1.28, p=0.90).  When subplots were analyzed using their plot level classifications, however, subplot level richness responded similarly to plot level richness (1997 damaged: df=43, t=4.13, p<0.0001); 1997 undamaged: df=27, t=1.06, p=0.15; 2011 damaged: df=43 t=-1.84, p=0.96; 2011 undamaged: df=27, t=-2.13, p=0.98).  This indicates that the richness response is not localized at the subplot level, and is best understood at the 0.1 hectare scale despite the patchiness of the disturbance.