Overall Research Project Description Curbing the effects of pathogens and securing the survival of endangered or commercially exploited species is of national interest. Collecting genetic data on these species has become standard over the last decade. Unfortunately, these types of data need translation and statistical analyses to be useful for policy decisions. It is common practice to use the data to calculate probabilities for particular models that describe the recent and ancient history of the species or population of interest. These models are founded on theoretical population genetics and are often not very flexible. A particular aspect addressed in this research is the assumption that the populations under study have a relatively constant number of offspring per generation. From observation, we know that this assumption is incorrect. For example, some SARS-CoV-2 strains are more successful in infecting people than others, suggesting that the ancestor with a new mutation has many more 'offspring' than others.
This research is generalizing this common assumption and constructs a framework that allows improving the analysis of population models with better accuracy and less bias.
A product of this research is a software tool. This tool uses genomic sequence data of a sample of individuals of the populations or species of interest to statistically compare different population models allowing the reconstruction of the history of the samples. Researchers will use this software to establish more accurate population size and genetic diversity estimates of species of interest. Accurate estimates will lead to better regulation of catch quota for commercial fishing, improved maintenance of endangered species, or better control of pathogen outbreaks.

Natural populations live in heterogeneous environments, and individuals in these populations have different chances to produce offspring. In contrast, current inferences of population genomic data are coalescence-based and assume environmental homogeneity within a population. Additionally, the commonly used coalescent framework assumes a variance of relatively narrow offspring numbers because it is derived from the Wright-Fisher or Moran population model. Heterogeneity can occur at very different scales: small, for example, different cell types in a human can lead to the differential success of a virus, or large, such as changes in microhabitats for a vertebrate species on a regional or even continental scale. This research explores the effect of heterogeneity of offspring production on the genealogy of individuals using (1) a theoretical framework that can handle heterogeneity and the development of software to infer this heterogeneity from genomic data. This framework is based on the fractional coalescent expanded to multiple, structured populations. The research extends a single-population derivation of the fractional coalescent that incorporates offspring variability as a random variable. This change will be well suited to tackle the variability of offspring numbers induced by environmental heterogeneity within and among different locations. These new methods will be incorporated into the widely-used open-source computer software MIGRATE. The new approach will then be compared with multi-merger coalescent methods using artificial data. These data are generated using (2) a simulator taking into account environmental quality changes within and among populations affecting the number of offspring an individual can have. A preliminary single-population simulator suggests that heterogeneity affects the time to the most recent common ancestor in a way that renders analyses with the standard coalescent questionable. The simulator will be used to generate large numbers of diverse scenarios that will be analyzed using standard summary statistics and complete probabilistic coalescence-based inference methods. (3) Analyses of the effect of heterogeneity for many biological datasets over a broad range of species with different life histories: from viruses to humpback whales and from small geographic scale to large scales. These datasets will be analyzed in collaboration with practical scientists. Software and tutorials will be reported on http://popgen.sc.fsu.edu and https://peterbeerli.com.