TL;DR
This paper introduces a novel approximate Bayesian method to untangle the complex demographic histories of rapidly expanding invasive species using genetic data. By applying this method to the cane toad in Australia, they reveal distinct expansion dynamics between eastern and northern regions, highlighting differences in population size, founder events, and migration rates.
What Problem Does This Paper Address?
Understanding how invasive species spread and establish themselves is crucial for ecological management. However, inferring the detailed demographic processes driving these expansions (like migration patterns, population sizes, and the number of founding individuals) from genetic data is incredibly challenging. Traditional methods often struggle with the complexity of recent, rapid expansions, where factors like founder effects and ongoing gene flow interact dynamically. This paper aims to bridge this gap by developing and applying a robust, model-based approach to analyze genetic variation in spatially expanding populations, using the cane toad (Bufo marinus) in Australia as a case study. The cane toad's well-documented introduction and ongoing spread across diverse regions provide a unique opportunity to test these new inferential methods.
Methods
The core of this paper is the development and application of an approximate Bayesian computation (ABC) method, specifically a rejection-regression algorithm. This approach is designed to handle complex demographic scenarios where traditional likelihood-based methods are computationally intractable.
Here's a breakdown of the key methodological components:
- Model-Based Inference: The researchers defined five distinct demographic models representing different spatial expansion scenarios, including variations in migration patterns (isolation-by-distance vs. stepwise), foundation events (presence/absence of founder effects), and the influence of geographic distance on these processes.
- Approximate Bayesian Computation (ABC): Instead of directly calculating likelihoods, ABC methods simulate data under various model parameters and compare these simulated datasets to the observed genetic data.
- Rejection-Regression Algorithm: This specific ABC technique involves two main steps:
- Rejection Step: Parameter sets that generate simulated summary statistics (like Fst values, allele numbers, heterozygosity, and allele size variance) sufficiently close to those observed in the real data are accepted. A Euclidean distance metric is used to quantify the similarity between simulated and observed summary statistics.
- Regression Adjustment: A local linear regression is then applied to the accepted parameter sets to refine the posterior distributions, effectively correcting for the approximation inherent in the rejection step and improving the accuracy of parameter estimates.
- Summary Statistics: A comprehensive set of summary statistics derived from microsatellite data was used, including pairwise Fst values between all sampled populations, mean number of alleles, mean gene diversity, and mean allele size variance.
- Coalescent Simulation: The demographic models were simulated using a generation-by-generation coalescent approach, tracing gene lineages backward in time from the sampled populations to the initial introduction events.
This approach is novel in its ability to formally test multiple complex demographic models and infer parameters for scenarios involving recent, rapid spatial expansions, which are notoriously difficult to analyze with existing methods.
Datasets
The study utilized genetic data from 10 microsatellite loci genotyped in cane toad (Bufo marinus) populations sampled along transects in two distinct expansion areas in Australia: the Eastern Expansion Area (EEA) and the Northern Expansion Area (NEA).
- EEA: Samples were collected from populations in New South Wales, representing an expansion that began around 1960.
- NEA: Samples were collected from populations in the Northern Territory, also originating from an introduction around 1960.
The genetic data was supplemented with extensive demographic and historical information for both expansion areas, including dates of first sightings, rates of spread, and geographical locations of populations. This historical and demographic data was crucial for defining the sampling schemes and informing the prior distributions for the demographic models.
Key Insights and Ideas
- Model Selection is Crucial: The study demonstrates that different demographic models can yield vastly different inferences. Formally testing these models is essential for accurate interpretation of genetic data from expanding populations.
- Stepwise Expansion Dominates: The "stepwise migration-foundation model with founder events" (Model 4) was overwhelmingly supported across both expansion areas, suggesting that new populations are primarily founded by individuals from adjacent populations, with an initial bottleneck (founder event).
- Distinct Regional Dynamics: Despite a common expansion model, the EEA and NEA exhibit significantly different demographic characteristics. This highlights that local environmental factors and historical introductions can lead to divergent evolutionary trajectories even for the same species.
- Founder Events Matter: The inclusion of founder events in the models was critical. The data strongly suggests that bottlenecks associated with colonization are a significant factor shaping genetic diversity in these expanding populations.
- Migration is Key, But Complex: While gene flow is substantial, its dynamics differ. The NEA shows higher migration rates and a larger effective number of migrants, suggesting a more dynamic and interconnected expansion compared to the EEA.
- Robustness of the Method: The study found that their inferences were relatively robust to variations in prior assumptions, suggesting the method is reliable for analyzing complex demographic scenarios.
Key Results
- Model 4 (Stepwise Migration-Foundation with Founder Events) was the best-supported model in both the Eastern Expansion Area (EEA) and the Northern Expansion Area (NEA), with posterior probabilities of ≥ 0.96.
- EEA vs. NEA Differences:
- Stable Effective Population Size (Nes): The NEA had significantly larger Nes values (point estimate ~1698) than the EEA (point estimate ~693).
- Effective Number of Founders (Nef): The EEA had lower Nef values (point estimate ~74) compared to the NEA (point estimate ~310), indicating fewer founding individuals in the east.
- Founding Ratio (FR = Nes/Nef): The founding ratio was higher in the EEA (point estimate ~9.2) than in the NEA (point estimate ~5.4), suggesting a greater disparity between stable population size and founder number in the east.
- Migration Rate (m) and Effective Number of Migrants (Nesm): The NEA showed substantially higher posterior values for both m and Nesm (modal values of 0.163 and 144, respectively) compared to the EEA (modal values of 0.006 and 4.1, respectively). This indicates more extensive gene flow in the north.
- Genetic Diversity: The EEA showed a significant positive correlation between allele number/size variance and population age, suggesting a stronger reduction in genetic diversity during expansion compared to the NEA.
- Population Structure: Pairwise Fst values were significantly higher in the EEA (FST = 0.0799) than in the NEA (FST = 0.0132), indicating greater genetic differentiation among populations in the east.
Limitations and Future Work
- Model Simplification: The authors acknowledge that their linear expansion-foundation model is a simplification of reality. A two-dimensional model, while more realistic, would be significantly more complex to implement.
- Discretization: The models discretize the expansion process into generations and distinct populations, which might not fully capture the continuous nature of spread.
- Computational Intensity: The ABC approach, while powerful, is computationally intensive, limiting its widespread availability and application. The authors hope for the development of more accessible programs.
- Transitory Population Booms: The models do not explicitly account for temporary population booms, which can influence demographic parameters. Future work could explore methods to characterize these transient events.
- Future Directions: The authors suggest developing two-dimensional models for spatial expansion and exploring how these demographic inferences can be integrated into predictive models for the evolution of adaptive quantitative traits in invasive species. They also highlight the need for more computationally efficient methods.
Summary updated