Submission Type
Oral Presentation
Abstract
Hybridization is a common problem in conservation. It can be difficult to identify hybrid individuals but doing so is often necessary in order to make conservation recommendations. For example, botanical gardens and seed banks want to identify if individuals are hybrids of two or more species to determine if offspring could be used for conservation or restoration, and to analyze the risk of possible extinction via introgression. Our project looks at how well the program STRUCTURE can detect and cluster hybrids, their parent species, and other pure species, which we created using simulations. We build on past studies, which typically only examined hybrids between two species, by detecting hybrids in systems with four through eight species, while also changing the numbers of individuals sampled per species, and marker number and types (microsatellites and SNPs). This situation in which many species within a genus can hybridize is common in plants. Because we use simulations, we are able to detect when structure makes errors in classification. Our results found five errors that can be made by STRUCTURE: a hybrid is labeled as a pure species, a pure species is labeled as a hybrid, a pure individual is labeled as unknown, a hybrid is labeled as unknown, and two pure species are confused as each other which was the most common. We found that STRUCTURE is generally accurate when comparing smaller numbers of species (four); the more species that it needs to compare, the more errors occur. Somewhat surprisingly, sampling more individuals or more loci did not reduce error rates when using microsatellites; we are currently assessing error rates when using SNPs.
Included in
Genetics Commons, Other Ecology and Evolutionary Biology Commons, Plant Biology Commons, Population Biology Commons
How Accurately STRUCTURE Can Identify Hybrids When Comparing Different Numbers of Species and Marker Types and Numbers
Hybridization is a common problem in conservation. It can be difficult to identify hybrid individuals but doing so is often necessary in order to make conservation recommendations. For example, botanical gardens and seed banks want to identify if individuals are hybrids of two or more species to determine if offspring could be used for conservation or restoration, and to analyze the risk of possible extinction via introgression. Our project looks at how well the program STRUCTURE can detect and cluster hybrids, their parent species, and other pure species, which we created using simulations. We build on past studies, which typically only examined hybrids between two species, by detecting hybrids in systems with four through eight species, while also changing the numbers of individuals sampled per species, and marker number and types (microsatellites and SNPs). This situation in which many species within a genus can hybridize is common in plants. Because we use simulations, we are able to detect when structure makes errors in classification. Our results found five errors that can be made by STRUCTURE: a hybrid is labeled as a pure species, a pure species is labeled as a hybrid, a pure individual is labeled as unknown, a hybrid is labeled as unknown, and two pure species are confused as each other which was the most common. We found that STRUCTURE is generally accurate when comparing smaller numbers of species (four); the more species that it needs to compare, the more errors occur. Somewhat surprisingly, sampling more individuals or more loci did not reduce error rates when using microsatellites; we are currently assessing error rates when using SNPs.
Comments
Sean Hoban, PhD - The Morton Arboretum
Austin Koontz, MSc - The Morton Arboretum
Emily Schumacher, MSc - The Morton Arboretum