Genetic variation in natural systems is complex and affected by a variety of processes, and this reality has contributed to the growing popularity of simulation-based approaches that can help researchers understand the processes acting in their systems. Despite the flexibility of simulation-based approaches, simulations of natural selection across a heterogeneous landscape have typically been limited to one or two loci (e.g. Landguth, Cushman, & Johnson, 2012). In a recent issue of Molecular Ecology Resources, Landguth et al. introduce an approach to model multilocus selection in a spatially-explicit, individual-based framework, implemented in the programs CDPOP and CDMetaPOP. Read the interview with lead author Erin Landguth below to learn about the challenges in developing this program, the potential of this approach to help understand complex genotype-environment associations, and the benefits of working with strong multidisciplinary team! Read the full article here.

What led to your interest in this topic / what was the motivation for this study? 

Over the last two decades, there has been an exponential increase in landscape genetic studies, and still, the methodology and underlying theory of the field are under rapid and constant development. Furthermore, interest in simulating multilocus selection, including the ability to model more complex and realistic multivariate environmental scenarios, has been driven by the growing number of empirical genomic data sets derived from next-generation sequencing. We believe many of the major questions in landscape genetics require the development and application of sophisticated simulation tools to explore the interaction of gene flow, genetic drift, mutation, and natural selection in landscapes with a wide range of spatial and temporal complexities. Our interests lie in developing such tools and providing more flexible models that are linked to theory, and that better represent complex genetic variation in real systems. For example, adaptive traits often have a complex genetic basis that interacts with selection strength, gene flow, drift, and mutation rate in a multivariate environmental context; and this module provides the ability to simulate these processes across many adaptive and neutral loci in a landscape genetic context.

What difficulties did you run into along the way? 

When developing new modules for existing software packages, my first and primary goal is to validate these modules to theory where possible. This can take some time and many decisions, questions, and trial and errors come up along the way through this very important validation process. For multilocus selection, our validation process was to match simulation output with the theoretical expected change in allele frequencies for selection models developed by Sewall Wright in 1935. If the module is placed in the wrong location in the simulation workflow (i.e., timing) or if all of the Wright-Fisher assumptions are not matched exactly, then the simulation output will not match theoretical expectations. However, once all of these pieces are lined up, there is definitely a eureka moment, and I am then confident in the module’s performance for more complex scenarios where we will not be able to evaluate against theoretical expectations.

What is the biggest or most surprising innovation highlighted in this study? 

Multivariate environmental selection can produce complex landscape genetic patterns, even when only a few adaptive loci are involved. The relatively simple “complex” example simulated in the paper illustrates how complicated the underlying relationships can be between allele frequencies and environmental conditions. Simulating these complex relationships will be essential for testing genotype-environment association methods in a more rigorous fashion than has been seen so far. Additionally, the ability to simulate realistic landscape genetic scenarios that reflect the environmental complexity of actual landscapes will be important for validating findings from empirical data sets. 

Moving forward, what are the next steps in this area of research?

Epigenetics! We of course have a number of applications in progress for this current module, but we have already started beta testing our next module for simulating epigenetic processes in landscape genetics.

What would your message be for students about to start developing or using novel techniques in Molecular Ecology? 

Starting a simulation study in landscape genetics for the first time can be daunting and intimidating. Fear not, we say! As with all software packages, there will be a learning curve, but if you persevere and get past the first few hurdles (e.g., learning the ins and outs of file formats, running the program in a potentially unfamiliar programming interface), the door will be opened to unlimited questions that can be addressed with simulations in your system. Additionally, just like any other field study or experiment, simulation modeling is most informative when coupled with specific questions and hypotheses and well-thought-out study designs.

What have you learned about methods and resources development over the course of this project? 

As we begin to add more complex modules to these simulation platforms, I am increasingly relying on multidisciplinary approaches and teams. For example, development of this current module required Brenna Forester for her expertise in landscape ecology and genotype-by-environment concepts, as well as Andrew Eckert, with his in-depth knowledge of population genetics theory, particularly the history of additive vs. multiplicative models for fitness.

Describe the significance of this research for the general scientific community in one sentence.

We have implemented a new module into the landscape genetic simulation programs CDPOP and CDMetaPOP that allows realistic multivariate environmental gradients to drive selection in a multilocus, individual-based, landscape genetic framework.

Describe the significance of this research for your scientific community in one sentence.

This new simulation module provides a valuable addition to the study of landscape genetics, allowing for explicit evaluation of the contributions and interactions between demography, gene flow, and selection-driven processes across multilocus genetic architectures and complex, multivariate environmental and landscape conditions.

References

Landguth EL, Forester BR, Eckert AJ, et al. (2020). Modelling multilocus selection in an individual-based, spatially-explicit landscape genetics framework. Molecular Ecology Resources, 20, 605–615. https://doi.org/10.1111/1755-0998.13121

Landguth, E. L., Cushman, S. A., & Johnson, N. A. (2012). Simulating natural selection in landscape genetics. Molecular Ecology Resources, 12, 363– 368. https://doi.org/10.1111/j.1755-0998.2011.03075.x

Wright, S. (1935). Evolution in populations in approximate equilibrium. Journal of Genetics, 30, 257– 266. https://doi.org/10.1007/BF02982240