Two decades of fire and water management shape rare plant communities along an elevational gradient in South Florida

Background/Question/Methods

Quantifying the response of plant diversity to large-scale restoration is essential for measuring management success. One of the world’s largest restoration efforts began in 2000 in Everglades National Park through the Comprehensive Everglades Restoration Plan (CERP). CERP, in coordination with ongoing fire management, aims to restore natural hydrological and fire regimes in this dynamic ecosystem. Restored water and fire regimes in the Everglades interact along an elevational gradient between seasonally inundated marl prairie and frequently burned pine rockland in Long Pine Key (LPK). LPK is critical for management success because it holds the largest intact segment of the globally critically imperiled pine rockland ecosystem. To determine the effects of management on plant community richness and composition in LPK, we resampled transects spanning the elevational gradient between marl prairie and pine rockland originally sampled prior to the implementation of current restoration efforts (c. 1997-1999). We measured percent plant cover and species presence/absence and developed two separate generalized linear mixed models to determine the effects of fire frequency, inundation, and elevation on plant species richness and composition across the two time periods. Additionally, we used species random effects to examine how individual species respond to each environmental variable.

Results/Conclusions

We found that plant richness increased at intermediate elevations and with moderate periods of inundation. We also found a strong positive interaction between burn frequency and inundation such that species richness is greatest at high burn frequencies and high levels of inundation. Similarly, for the presence/absence model, the probability of occurrence for any species was greatest at intermediate elevations. There were strong positive interactions between elevation and inundation, and burn frequency and inundation, indicating that the probability of species occurrence depends on the length of inundation at different burn frequencies across the elevational gradient. Sites at intermediate elevations with moderate levels of inundation and sites with high burn frequencies and high levels of inundation showed the highest probability of species occurrence. Species-level random effects show that elevation was the most important predictor for the majority of species, followed by inundation. The results of these two models indicated that most species are found in the middle of the elevational gradient where fire and water are more likely to interact. Additionally, our species-level random effects showed that species endemic to the system increased while invasive species decreased between the two time periods, indicating effective management of these species groups.