Developing approaches for linear mixed modeling in landscape genetics through landscape-directed dispersal...

Area of Science:

• Ecology
• Evolutionary Biology
• Genetics

Background:

• Dispersal influences population dynamics and genetic variation, making landscape factors crucial for connectivity.
• Genetic approaches and mixed models are increasingly used to study landscape influences on populations.
• Lack of consensus exists on optimal model selection protocols for these analyses.

Purpose of the Study:

• To develop and test landscape-directed simulations for evaluating model selection in genetic connectivity studies.
• To compare the effectiveness of different model selection indices (AIC, BIC, R²) in identifying landscape factors influencing dispersal.
• To assess the reliability of model coefficients and confidence intervals in reflecting true dispersal models.

Main Methods:

• Developed landscape-directed simulations emulating empirical genetic datasets for two species (greater sage-grouse, eastern foxsnake).
• Applied linear mixed models to relate genetic differentiation to landscape resistance measures.
• Evaluated model selection indices (AIC, BIC, marginal R²) and assessed model coefficients and confidence intervals.

Main Results:

• AIC and BIC were identified as the most effective model selection indices in simulated scenarios.
• Marginal R² values showed bias towards more complex models.
• Model coefficients for landscape variables accurately reflected underlying dispersal models when geographic distance was controlled.

Conclusions:

• AIC and BIC provide robust model selection for identifying landscape drivers of genetic connectivity.
• Linear mixed models, when properly applied, can reliably detect landscape features influencing species dispersal.
• This study establishes improved methods for using genetic data and mixed models to understand landscape-level dispersal patterns.