Differential Effects of Elevation and Microtopography on Gopher Tortoise Burrow Distributions in Southern Florida
In southern Florida, Gopher Tortoises (Gopherus polyphemus) occupy mesic flatwoods and Florida scrub communities where habitat and climatic conditions differ from other portions of the species' range. Both of these habitats appear suboptimal for tortoises due to saturated soils in mesic flatwoods and low forage abundance in scrub. Nonetheless, these habitats support large numbers of tortoises in southern Florida, albeit at low intensities. We assessed influences of elevation and microtopography on the spatial distributions of tortoise burrows and examined burrow use patterns within six sites at Avon Park Air Force Range in south-central Florida. The six sites differed in dominant soil types and vegetation communities, allowing comparisons of burrow distributions among mesic flatwoods, Florida scrub, and mixed flatwoods-scrub habitats (two replicate sites each). Point-process modeling identified significant influences of microtopography on burrow intensities that superseded the effects of site-wide elevation trends in five of the six sites. The effects of microtopography were most pronounced in flatwoods, suggesting greater reliance on areas of slightly higher elevation in mesic habitat, presumably in response to saturated soils and frequent flooding. Burrow use patterns during an exceedingly wet year also suggested that tortoises respond behaviorally to unsuitable hydrology by moving frequently among burrows that were flooded with groundwater. Microtopographic variation may be an important predictor of small-scale habitat use for fossorial reptiles, especially in mesic soils, which could be readily explored using increasingly available LiDAR-derived elevation data combined with the modeling approach demonstrated here.
Map of Avon Park Air Force Range in south-central Florida showing locations of study sites where Gopher Tortoise burrows were mapped in 2011 (FW1, SC1), 2015 (FW2, SC2, MX1), and 2016 (MX2).
Elevation surfaces for the FW2 flatwoods site. Elevation data (meters above sea level) were based on a digital elevation model (DEM) (A); each pixel was 0.5 m2. We fit (orthogonal) low-order polynomials to the DEM, where the fitted values represent the polynomial trend surface (B), and the residuals represent microtopography (C). To sharpen the image of the microtopography surface, we used a probability integral transformation (D). For surfaces A–C, Gopher Tortoise burrow locations are shown as white dots and the study area boundary is represented as a black line; for surface D, Gopher Tortoise burrow locations are shown as black dots and the study area boundary is represented as a white line.
Microtopographic elevation surfaces (meters above sea level) sharpened by probability integral transformation for (A) FW1-2011, (B) FW1-2015, (C) SC1-2015, (D) SC2, (E) MX1, and (F) MX2. Black dots are Gopher Tortoise burrow locations and white lines are the study area boundaries. SC1-2011 is not shown because Gopher Tortoise burrow distributions at this site were very similar in 2011 and 2015 (Supplemental Fig. C; see Data Accessibility).
Regression coefficients (β) with 95% confidence intervals for different components of elevation were used (additively) as predictors of log burrow intensity for different study sites/years. For most sites there were weak positive effects of increases in the polynomial trend surfaces (=squares), coupled with stronger positive effects of increases in microtopographic elevations (=circles). For FW1-2011 and FW1-2015, however, there were negative effects of the polynomial trend surfaces, while effects of microtopography were positive. Symbols are color coded depending on habitat type: flatwoods = blue, scrub = red, and mixed = turquoise. Non-overlapping confidence intervals indicate that coefficients for microtopography were significantly larger for flatwoods sites than for scrub or mixed sites. For MX2, only results for microtopography are displayed because a model containing microtopography as the single predictor could not be rejected. Results for SC2 are not displayed because the null model (complete spatial randomness) could not be rejected.
Intensities of used and unused, non-collapsed Gopher Tortoise burrows in flatwoods sites (FW), scrub sites (SC), and sites having a mix of both habitat types (MX). X-axis bar labels reflect the habitat types and years they were surveyed.
Contributor Notes
Associate Editor: J. D. Litzgus.