Comparative phylogeography reveals how a barrier filters and structures taxa in North American warm deserts

Kaiya L. Provost, Edward A. Myers, Brian Tilston SmithbioRxiv 2020 · DOI

TL;DR

This paper investigates how a specific geographic barrier, the Cochise Filter Barrier in North America, influences the genetic divergence of various species. By synthesizing existing data and using simulations, the researchers found that species traits like locomotion and thermoregulation play a key role in how effectively this barrier shapes genetic differences, leading to a complex and varied evolutionary history across the region.

What Problem Does This Paper Address?

Biogeographic barriers are fundamental to understanding how species evolve and distribute across landscapes. However, simply observing a genetic break coinciding with a geographic feature doesn't tell us how the barrier functions or why some species are affected while others are not. This paper tackles the challenge of moving beyond descriptive accounts of barriers to quantitatively assess their impact on entire communities. Specifically, it aims to:

  1. Synthesize phylogeographic data across multiple species to understand community-wide patterns of genetic divergence.
  2. Quantify the influence of species-specific traits (like how they move and regulate body temperature) on their response to a barrier.
  3. Evaluate the statistical power of different demographic models in distinguishing between various modes of speciation and gene flow, especially in the context of complex barriers.

The Cochise Filter Barrier, separating the Sonoran and Chihuahuan deserts, serves as an ideal case study due to its known environmental gradient and the presence of many taxa that cross it.

Methods

The study employs a multi-pronged approach:

  • Literature Synthesis: They compiled and analyzed phylogeographic data from 99 published studies covering 68 taxa. This involved extracting information on divergence timing, gene flow, phylogeographic structure, and inferred modes of speciation.
  • Trait Characterization: Each taxon was categorized based on its locomotion (walking, flying, etc.) and thermoregulation (ectotherm vs. endotherm). Elevational preference was also considered.
  • Spatially-Explicit Simulations: Using the SLIM 3.1 software, they simulated genetic data under four demographic models (single population, pure isolation, isolation-with-migration, and secondary contact) with and without isolation-by-distance (IBD). These simulations were run across different time scales (6,000 to 1,000,000 generations) to mimic various environmental conditions and divergence histories.
  • Machine Learning for Model Selection: A neural network was trained on the simulated data to classify demographic models, IBD presence, and divergence times. This approach was used to assess the identifiability of different evolutionary scenarios and the challenges in distinguishing them, particularly when genetic data is limited or divergence is recent.

Datasets

The primary dataset for the empirical analysis was a synthesis of 99 published phylogeographic studies encompassing 68 taxa. This included data from various molecular markers (mitochondrial, chloroplast, nuclear loci, SNPs, microsatellites, etc.) and phenotypic assessments.

For the simulation component, simulated genetic data was generated using the SLIM 3.1 software. The landscape for these simulations was informed by an ecological niche model (ENM) of the Bell's Vireo (Vireo bellii), a species known to be restricted by the Cochise Filter Barrier.

Key Insights and Ideas

  • Barriers are not monolithic: The Cochise Filter Barrier doesn't act as a single, uniform filter. Instead, its impact varies significantly across species, highlighting the concept of "semi-permeability" where different taxa experience different degrees of isolation and gene flow.
  • Traits matter: Species' functional traits are crucial mediators of barrier effects. Locomotion, in particular, strongly correlates with the degree of phylogeographic structure observed, with flying species generally showing less structure, suggesting higher dispersal ability.
  • Divergence is complex: While allopatric isolation is often invoked as the primary mode of speciation, the study found evidence for other mechanisms like isolation-by-environment, hybrid speciation, and sexual selection, underscoring the multifaceted nature of diversification.
  • Simulation and ML are powerful, but not perfect: The simulation and machine learning pipeline demonstrated the ability to identify demographic models and IBD presence with reasonable accuracy, especially for older divergences. However, it also revealed that distinguishing between certain models (e.g., isolation vs. secondary contact, or isolation-with-migration vs. single population) can be challenging, particularly with shorter divergence times.

Key Results

  • Community-wide structure: Approximately 62% of the studied taxa showed evidence of phylogeographic structure across the Cochise Filter Barrier.
  • Divergence timing: Divergence times ranged from the Miocene to the Pleistocene, with a significant portion of splits occurring in the Pleistocene.
  • Locomotion and structure: Flying species had the lowest proportion of taxa with structure (~47%), while walking species had the highest (~70%).
  • Thermoregulation and divergence: While not statistically significant, ectotherms showed a higher proportion of structured taxa and tended to have older divergences compared to endotherms.
  • Model classification accuracy: The neural network achieved high accuracy (92-100% for training) in classifying the presence of IBD. Classification of divergence age was also good (83-85% for training), but classifying the true demographic model was more variable, with accuracy ranging from 28-62% for validation/testing.
  • Misclassification patterns: Younger divergences were more prone to misclassification between demographic models, making it harder to infer evolutionary history.

Limitations and Future Work

  • Taxa restricted to one side: The study acknowledges a limitation in not assessing taxa that are completely restricted to one side of the barrier, which could provide further insights into what makes the CFB a potent barrier. Future work could focus on sampling such endemic species.
  • Sampling bias: The authors note that many studies included in their synthesis have incomplete sampling, particularly from the transition zone of the CFB. Targeted sampling with genome-level data is suggested to resolve discrepancies.
  • Model identifiability: While simulations and machine learning are powerful, the study highlights that some demographic models remain difficult to distinguish, especially at younger divergence times. Future work could explore more complex simulation scenarios and analytical techniques to improve model identifiability.
  • Trait complexity: The study focused on locomotion and thermoregulation. The authors suggest that many other traits (e.g., habitat preferences) could also play a role in mediating divergence across barriers.
  • Dynamic environments: The simulations did not account for changes in population size, mutation rate, or environmental factors over time, which can influence demographic inferences. Incorporating paleoclimate models into future simulations could provide a more realistic representation of past environmental dynamics.

Summary updated

Abstract

Abstract The study of biogeographic barriers have been instrumental in understanding the evolution and distribution of taxa. Now with the increased availability of empirical datasets, it is possible to infer emergent patterns from communities by synthesizing how barriers filter and structure populations across species. We assemble phylogeographic data for a barrier and perform spatially-explicit simulations to quantify temporal and spatial patterns of divergence, the influence of species traits on these patterns, and understand the statistical power of differentiating alternative diversification modes. We incorporate published datasets to examine taxa around the Cochise Filter Barrier, separating the Sonoran and Chihuahuan deserts of North America, to synthesize phylogeographic structuring across the community with respect to organismal functional traits. We then use a simulation and machine learning pipeline to assess the power of phylogeographic model selection. Taxa distributed across the Cochise Filter Barrier show heterogeneous responses to the barrier in levels of gene flow, phylogeographic structure, divergence timing, barrier width, and divergence mechanism. These responses vary concordantly with locomotor and thermoregulatory traits. Many taxa show a Pleistocene population genetic break, often with introgression after divergence. Allopatric isolation and isolation-by-environment are the primary mechanisms purported to structure taxa. Simulations reveal that in spatially-explicit isolation-with-migration models across the barrier, age of divergence, presence of gene flow, and presence of isolation-by-distance can confound the interpretation of evolutionary history and model selection by producing easily-confusable results. By synthesizing phylogeographic data for the Cochise Filter Barrier we show a pattern where barriers interact with species traits to differentiate taxa in communities over millions of years. Identifying the modes of differentiation across the barriers for these taxa remains challenging because commonly invoked demographic models may not be identifiable across a range of likely parameter space.
BibTeX
@misc{provost2020comparative,
  author = {Kaiya L. Provost and Edward A. Myers and Brian Tilston Smith},
  title = {{Comparative phylogeography reveals how a barrier filters and structures taxa in North American warm deserts}},
  year = {2020},
  doi = {10.1101/2020.06.17.157842},
  howpublished = {bioRxiv},
}

Loading related content...

Figure