Publication Date


Date of Final Oral Examination (Defense)


Type of Culminating Activity


Degree Title

Master of Science in Biology



Major Advisor

T. Trevor Caughlin, Ph.D.


Jodi Brandt, Ph.D.


Jennifer Forbey, Ph.D.


Boundaries between ecosystems are often biodiversity hotspots with relatively high vulnerability to global change. The boundary between tropical rainforest and savanna ecosystems in the Amazon presents an ecotone that is undergoing a shift in ecosystem structure, as a warming climate promotes the expansion of grassland. How animal communities in the Amazon will respond to changes in ecosystem structure is a crucial unanswered question with implications for the many ecosystem services that animals provide, from a food source for Indigenous people to seed dispersal for vulnerable tree populations. Recent modeling work has forecasted that faunal savannization will occur in the Amazon, as savanna-dwelling animals replace forest specialists. However, empirical data to test these forecasts has remained scarce, due to the need for large-scale data across local and regional forest-savanna gradients. To overcome this difficulty, we quantified associations between terrestrial vertebrates and ecosystem structure using replicated camera traps across a forest-savanna ecotone in central Guyana. To capture continuous gradients in woody biomass across the ecotone, we paired radar-derived measurements of aboveground biomass from Phased Array-type L-band Synthetic Aperture Radar (PALSAR) with animal species presence at camera trap sites, including >54,000 individual photos. We hypothesized that different animal species communities would emerge in sites with different levels of aboveground biomass, representing forest and savanna specialists. We tested this hypothesis with hierarchical Bayesian models for animal species detection and species richness across our study landscapes. Our results did not support the hypothesis that there is a guild of savanna specialists with increased presence in sites with low aboveground biomass. Instead, nearly all (54 out of 56) species showed increased probability of detection in sites with higher aboveground biomass. Consequently, overall species richness was significantly related to aboveground biomass, including a median proportional increase in species richness of 90.0% (CI: 21.57 to 200.0%) for every kiloton of biomass at a site. These results suggest that woody structure plays a critical role in supporting animal species richness at the Amazonian forest-savanna ecotone, including non-forest tree cover such as bush islands, gallery forest, and isolated trees. Ongoing declines in tree cover will likely have detrimental impacts across most groups of animal species. Without landscape conservation strategies to maintain tree cover at the forest-savanna boundary, climate change could have severe consequences for Amazonian animal populations.