Tue, Aug 03, 2021:On Demand
Background/Question/Methods
Reconstructions of species’ biogeographic histories and attempts to infer the location and number of past refugia as well as migration rates commonly rely on disparate types of data and analyses. These include occurrence data modeled with species distribution models (SDMs), fossil pollen data informed spatio-temporal estimates of taxon relative abundance, and genetic data analyzed with scenario-based models. Generally, these different types of analyses are assumed to provide consistent (if incomplete) results about species’ biogeographic trajectories through time. Using green ash (Fraxinus pennsylvanica) in eastern North America as a case study, we analyze the consistency of inference between the three data types commonly used in historical biogeography. Genetic data were modeled using a spatially-explicit forward-time demographic model coupled with coalescent simulations to generate SNP data. Relative abundance of ash across space and through time was estimated using a Bayesian hierarchical spatio-temporal model, and climatically suitable habitat was modeled using SDMs with multiple algorithms and global climate models. Specifically, we compare the number and locations of inferred refugia during the last glacial maximum (~21,000 years ago) and the pace of range movement (biotic velocity) through time. We also evaluate inferences from integrated analysis of multiple data types (genetics-pollen and genetics-SDMs).
Results/Conclusions The location and number of refugia and biotic velocities through time differed according to the method used in the analysis. Generally, the pollen model predicted one refuge in the Midwest (from Missouri to Minnesota). SDMs predicted a large southern refugium accompanied by 1 to 5 smaller refugia, including the northeast coast. The genetic analysis best supported a multi-refuge model over single-refuge models. The techniques also resulted in different estimates of biotic velocity through time. The genetics-only analysis (which assumes all habitats are equally suitable) indicated velocity was very high soon after the glacial maximum, but then declined to a rate primarily determined by glaciers’ retreat. In contrast, velocities based on pollen and occurrence data suggest accelerations coinciding with known climatic fluctuations associated with the Bølling-Allerød period of dramatic climatic change. Biotic velocities informed by multiple data types are contrasted to those inferred by individual data types, and suggest that integration of multiple lines of evidence may overcome the limitations of each individual data type. Together, our results suggest that coincidence between biogeographic reconstruction methods should be more quantitatively assessed.
Results/Conclusions The location and number of refugia and biotic velocities through time differed according to the method used in the analysis. Generally, the pollen model predicted one refuge in the Midwest (from Missouri to Minnesota). SDMs predicted a large southern refugium accompanied by 1 to 5 smaller refugia, including the northeast coast. The genetic analysis best supported a multi-refuge model over single-refuge models. The techniques also resulted in different estimates of biotic velocity through time. The genetics-only analysis (which assumes all habitats are equally suitable) indicated velocity was very high soon after the glacial maximum, but then declined to a rate primarily determined by glaciers’ retreat. In contrast, velocities based on pollen and occurrence data suggest accelerations coinciding with known climatic fluctuations associated with the Bølling-Allerød period of dramatic climatic change. Biotic velocities informed by multiple data types are contrasted to those inferred by individual data types, and suggest that integration of multiple lines of evidence may overcome the limitations of each individual data type. Together, our results suggest that coincidence between biogeographic reconstruction methods should be more quantitatively assessed.