Year: 2026
Type: preprint
Abstract: The program STRUCTURE is widely used in conservation genetics to detect human-caused hybridization between previously isolated populations. However, its performance at this specific task is not yet known, and methods used to interpret the model output vary widely among studies. We simulated hybridization between two populations on the timescale of real-world cases and tested STRUCTURE’s ability to correctly detect hybridization based on a simple “hybrid signal”: inferring the presence or absence of a minimum of one hybrid. Overall, this returned the correct result in 100% of scenarios when Fst between the two populations was between 0.05-0.15 when using an optimal Q-threshold of 0.10. Performance was slightly reduced when divergence was very weak (Fst = 0.01, with an optimal Q-threshold between 0.30-0.35) or when the population receiving migrants was substructured, but it still returned the correct result in 89-100% of cases, depending on simulation parameters. We conclude that STRUCTURE, paired with this hybrid signal and our recommended Q-thresholds, is an appropriate tool for detecting contemporary hybridization. Article summary Human actions may be causing previously isolated populations to begin hybridizing, which can erode valuable genetic diversity. Many researchers seek to detect recent hybridization with the software STRUCTURE, which uses genotype data to classify the ancestry of sampled individuals as either pure or hybrid. Using simulations, the authors find STRUCTURE to be highly capable of correctly detecting either the presence or absence of hybridization. They provide methodological recommendations to maximize STRUCTURE’s performance at this task, which should aid conservation geneticists working to detect cases of human-caused hybridization.
Institution: Simon Fraser University
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7234415546
