Enhancing models of species-habitat relationships with lidar remote sensing: A case study estimating the probability of burrow abandonment by the threatened gopher tortoise (Gopherus polyphemus)

Background/Question/Methods

Multi-scale information of habitat structure is necessary to derive ecologically relevant models of species-habitat interactions. However, it is logistically difficult to collect fine-grain information on habitat structure at broad spatial extents using only traditional field based methods. Lidar remote sensing is a powerful tool for acquiring data on habitat structure over broad spatial scales and when combined with field methods can be used to build both descriptive and predictive models. The goal of this study was to use a combination of lidar and field methods to determine how the probability that a gopher tortoise abandons its burrow is affected by changes in vegetation structure. Through field surveys in 2011, 1584 tortoise burrows were located and classified as either “abandoned” or “not abandoned” over 350 ha of sandhill habitat in central Florida. Estimates of relative canopy cover in the area surrounding a burrow were derived from discrete-return lidar acquired over the study area and these estimates were used to develop logistic regression models to predict the probability of burrow abandonment at multiple spatial scales. Akaike’s information criterion (AIC) was used to select the logistic regression model with the best fit to the observed data.

Results/Conclusions

The best-fitting model was highly significant (P<0.001) and included an estimate of canopy cover within a 10 m radius of a burrow. This model was used to calculate the mean probability of a gopher tortoise burrow being classified as “abandoned” as a function of the relative percentage of canopy cover. This probability of abandonment increases in a positive, nearly linear manner from a mean of 0.389 (95% confidence interval: 0.336 – 0.441) with 0% relative canopy cover to a mean of 0.775 (95% confidence interval: 0.698 – 0.853) with 100% relative canopy cover. Thus, each percent increase in canopy cover within a 10 m radius of a burrow increases the probability that the burrow is “abandoned” by 1.7%. These results suggest that vegetation change is an important predictor of gopher tortoise burrow abandonment. The marriage of lidar remote sensing and traditional field methods allows us to move beyond simple statements of effect or no effect, and instead measure the functional response of species to changes in the physical vegetation structure of their habitats. Such ecological models are necessary to truly understand species-habitat interactions and design effective conservation strategies for long-term persistence of populations.