Date of Degree


Document Type


Degree Name





Frank Burbrink

Committee Members

Mike Hickerson

S. Elizabeth Alter

Sara Ruane

Brian Smith

Chris Raxworthy

Subject Categories

Biodiversity | Computational Biology | Evolution | Genomics | Population Biology


Madagascar, historical demography, diversification, Machine Learning, comparative phylogeography, herpetology


A long history of isolation coupled with complex topographic and ecological landscapes makes Madagascar ideal for exploring the historical factors that have shaped patterns of population diversity and endemism. Many species-level studies have suggested Late Quaternary climate change may have influenced population dynamics in the tropics, but Madagascar’s ecologically unique biomes or individual species properties may have driven idiosyncratic responses to these shifts. Using community-scale population genetic data I implement a hierarchical approximate Bayesian computation (hABC) approach to evaluate the degree of synchronous population expansion during glacial cycles across herpetofaunal assemblages both within and across discrete biomes and taxonomic groups. I integrate results from Bayesian model averaging to identify whether intrinsic and extrinsic conditions played a role in driving individualistic demographic change. I find that demographic responses are not uniform across groups, with more than 50% of all populations showing signal of recent expansion. Our explanatory models indicate species occupying narrow elevational ranges had a higher probability of expansion, while amphibian assemblages showed higher genetic diversity and greater departures from population neutrality. Expansion events were largely asynchronous, with coexpansion found in less than half of all populations. Humid-restricted taxa are the exception to this trend, with around 69–74% of all populations coexpanding during the start of the Last Glacial Period at around 100 kya, supporting the hypothesis of a more extensive humid forest cover for Madagascar during this time. I show that exploration of trait and habitat-specific demographic processes is crucial to understanding and protecting the exceptional biotic richness and endemism found in Madagascar.

In my second chapter, I use genomic data from population level sampling across the island’s snakes within a comparative statistical phylogeographic framework to investigate patterns of population structure and pulses of simultaneous demographic shifts for Madagascar’s pseudoxyrhophiine snakes to determine if demographic histories been individualistic or synchronous relative to Pleistocene climate change. I identified 21 populations with a high probability of recent expansion and 12 with a history of population bottleneck. I conducted a hierarchical Random Forest analysis to estimate the proportion of lineages experiencing synchronous demographic dynamics and the timing of these events. For expanding populations, I recovered a strong signal of synchronous expansion in the Late Pleistocene after the Last Glacial Maximum. Contracting lineages showed evidence of temporally concordant bottlenecks before the onset of widespread anthropogenic Holocene fire disruption, indicating historical climate was more important in structuring contemporary patterns of diversity. I find little evidence of geographic, ecological or taxonomic signal in population diversity estimates or probability of expansion. These results suggest that areas that promoted demographic stability are not shared among broadly distributed taxa, and instead demographic change in these populations is primarily influenced by localized habitat features and species interactions.

Demographic processes like population expansion often result in events like population divergence that ultimately influence patterns of genetic diversity and richness. In my third chapter, I investigate previously proposed and novel hypotheses that address the historical processes driving these patterns on Madagascar. The many widespread lineages present in the snake fauna of the island span multiple landscape and climatic barriers and represent an unparalleled opportunity to investigate the origins of endemic diversity on the island. Using genome-wide sampling, I test whether these hypotheses explain phylogeographic patterns across four codistributed, endemic snake lineages widely distributed throughout Madagascar. I evaluated the performance of two advanced regression techniques, Gradient Boosting Machine (GBM) and Random Forest regression (RF) to estimate the timing of lineage divergence, migration, and historical demographic parameters of four species groups to highlight spatial and temporal patterns of differentiation. I find that GBM outperforms RF, but both approaches estimate divergence times for the most recent diversification events in the Mid to Late Pleistocene, when overall rates in speciation declined across Madagascar snakes. Divergence was coupled with low migration rates for all species groups examined here. For at least two species groups, I find strong evidence that both climatic stability, historical refugia, and current environmental gradients have jointly contributed to population divergence over time. For dispersal limited organisms with parapatric and sympatric distributions, Madagascar’s complex ecographic and topographic history may have contributed to rapid differentiation following extreme community turnover following glacial cycles in the Quaternary.