• Stefan Gavriliuc, Faculty of Veterinary Medicine, Department of Ecosystem and Public Health, University of Calgary
• Salman Reza, Faculty of Veterinary Medicine, Department of Ecosystem and Public Health, University of Calgary
• Chanwoori Jeong, Faculty of Veterinary Medicine, Department of Ecosystem and Public Health, University of Calgary
• Philip McLoughlin, Department of Biology, University of Saskatchewan
• Jocelyn Poissant, Faculty of Veterinary Medicine, Department of Ecosystem and Public Health, University of Calgary

Introduction Genomic technologies have become increasingly widespread through their decreases in costs, but have remained inaccessible for wildlife and conservation genetics studies. The main impediment for applying genomics to these fields has been the high input requirements of DNA which entails invasive sampling. Recently, target enrichment approaches have been shown to yield genome-wide genotype data with minimal DNA input, but these methods have yet to be tested on sample tissue obtained non-invasively such as feces. Concurrent advances in genome imputation have enabled the inference of missing genotypes, increasing genotype density to the size of a reference population and in turn improving coverage and statistical power for genomic studies. Methods Our goals were to 1) test whether a novel target enrichment method can provide genotype information comparable to the common standard of DNA microarrays and 2) whether genome imputation can further increase genotype density with high accuracy. Results We collected fecal swabs from a population of feral horses, and genotyped 48 swabs at 279 loci using the NuGEN Allegro Targeted Genotyping kit. Output genotypes were compared to the same set of loci genotyped using standard equine DNA microarrays for coverage and agreement. For imputation, we masked genotypes to random subsets (1000 – 20000) and varied the number of individuals in the reference population. Target enrichment a mean coverage and accuracy of 85% and >99% per sample, respectively, while genome imputation accuracy approached unity when using >10k SNPs. Conclusion This work serves as a proof-of-concept for non-invasive genotyping, helping enable large-scale genomics for natural populations.