Forest succession in tropical hardwood hammocks of the Florida Keys 19 years after Hurricane Andrew

Background/Question/Methods

Previous work in Florida Keys tropical hardwood hammocks, demonstrating hurricane-caused mortality of early successional trees and post-hurricane establishment of late-successional seedlings, suggests that hurricanes accelerate succession.  In this study we revisited the same forests 19 years post-hurricane to characterize a long-term successional trajectory. We expected that 1) tree species assemblages 19 years post-hurricane would be later successional than pre-hurricane assemblages, and 2) 19 years post-hurricane, tree and seedling species would be close in successional stage.  Tree and seedling data were collected during 2013 in 27 sites ranging in successional age from 21 to 133 years since last stand-initiating disturbance.  We conducted weighted averaging regression on pre-hurricane tree species from all sites to calculate a successional age optimum and tolerance for each species.  We then used species successional age optima and tolerances in weighted averaging calibration to calculate inferred site ages (ISAs) based on pre-hurricane tree species, tree species 2 and 19 years post-hurricane, and seedling species 19 years post-hurricane.  To quantify a short-term successional trajectory after the hurricane, we compared ISAs based on tree species 2 years post-hurricane to ISAs based on pre-hurricane trees.  To quantify a long-term successional trajectory, we compared ISAs based on tree species 19 years post-hurricane to ISAs based on tree species 2 years post-hurricane.  We also compared ISAs based on 2013 seedling species with ISAs based on 2013 tree species.

Results/Conclusions

Surprisingly, we found no difference in mean inferred stand age between 1992 (pre-hurricane) and 1994 (2 years post-hurricane) trees (paired t-test p = 0.297), or between 1994 and 2013 (19 years post-hurricane) trees (paired t-test p = 0.058).  2013 seedlings were later successional than 2013 trees (mean difference in inferred stand age + 95% CI = 33.9 + 7.9 years, p < 0.0001), but when we compared inferred stand ages based on small (< 10 cm height) 1995 seedlings and inferred stand ages based on large (> 10 cm height) 2013 seedlings, we found no difference (mean difference + 95% CI = 3.37 + 16.0 years, p = 0.602).  Although earlier studies suggested accelerated succession in response to hurricanes in these forests, evidence of an accelerated trajectory was not present in tree assemblages 19 years after Hurricane Andrew.  Instead, our results suggest that seedlings established soon (< 3 years) after the hurricane remain in a seedling bank, and may be candidates for subsequent recruitment into the canopy via canopy gaps.