Landscape genetics is a field that has expanded rapidly in recent years, but that doesn’t mean that it hasn’t gone without criticism. Perhaps the largest problem with landscape genetics (LG) studies is one of timing. If you observe genetic differentiation between two populations, is it truly due to the contemporary landscape or a more-historical process? This is often described as “time lag”, the number of generations it takes for a population split/coalescence to manifest itself in genetic data. Although the strategies for incorporating genetic data with landscape variables have blossomed with the increase in LG studies, the strategies for separating these historical and contemporary landscape effects have not.
An upcoming review in Molecular Ecology by Clinton Epps and Nusha Keyghobadi lays the time lag issue on the table, reviews current work that proposes solutions, and makes recommendations for future strategies.
The first step is asking the question: What affects an investigator’s ability to detect a pattern in genetic data that is driven by landscape?
- The parameter you measure. Measures of inbreeding, heterozygosity changes more slowly than something like Fst, but there are other alternatives like conditional genetic distance and the proportion of shared allele distance
- Not just the parameter (response variable), but also the analytical method that calculates them.
- The molecular marker under the microscope. Specifically, how quickly do certain groups of genetic loci change over time. What about loci under selection?
- Generation time of the taxon
- Direction of change. Equilibrium will happen slower when populations are fragmented than the opposite, when populations are reconnected following the removal of a barrier.
- Dispersal, pop sizes, structure, dynamics….the list goes on, you get it
If these are the causes for genetic lag, what are the solutions?
Epps and Keyghobadi lay out some detailed approaches that have been incorporated by other researchers, including strategies for when historical landscape information is know and when the landscapes of the past are a mystery.
Getting some idea of the historical landscape is the most helpful strategy to control for effects of past landscape on observed genetic patterns. You can look for historical data in traditional sources of ecological knowledge, like fire history maps, archival maps, and vegetation surveys. Alternatively, you can piece together an idea using combinations of past climate data, geological records, and ecological niche models.
Secondly, varying the type of analysis or molecular marker can provide at least a broad idea of differences in time scale between inferences of connectivity. One example would be combining microsatellite data and mtDNA data to assess connectivity at contemporary (BayesAss), historical (Migrate-n), and even more historical time scales (mtDNA divergence). Simulations are suggested as an important tool for creating expectations of time lag for multiple markers and various methods of analysis, adding this review to the “simulations are underused in molecular ecology” folder.
Finally, if you’re lucky, just have samples from the present and past.
After all of these approaches, the authors provide a unique spin on the time lag problem. That is, considering time lags as the measurement of interest:
We propose that an as-yet little exploited approach could be to take advantage of time lags in genetic structure to establish baselines for connectivity conservation. For instance, where known barriers to species dispersal have recently been constructed, rather than conducting LG analyses to determine whether an effect on genetic structure can be detected, LG analyses that consider and estimate time lags could show where the disconnect between pre- and post-fragmentation connectivity is greatest.
Epps, C. W., & Keyghobadi, N. (2015). Landscape genetics in a changing world: disentangling historical and contemporary influences and inferring change. Molecular Ecology.