Data and code for: Selecting puppies for underrepresented alleles conserves genetic diversity: a forward simulation in three dog breeds

Complete data and code package for a forward simulation study comparing four keeper-selection strategies for managing genetic diversity in closed purebred dog populations: outlier index (OI), average genetic relatedness (AGR), internal relatedness (IR), and random selection. Founder populations of 200 real dogs with 33-locus STR microsatellite genotypes were bred forward for 20 generations, 50 replicates per breed, across three breeds with distinct bottleneck histories: Standard Poodle, Doberman Pinscher, and Flat-Coated Retriever. Selection acts on which puppies are kept, not on which parents are bred. Contents manuscript/ — manuscript (DOCX, authoritative; Markdown extraction provided for accessibility), submission-resolution figures (Figures 1–4 and Supplementary Figures S1–S2), supplementary tables S1–S6 in XLSX, and the Supplementary Figure S2 source data (per-dog outlier index, average genetic relatedness, and Mid-Century Bottleneck ancestry percentage for 632 Standard Poodles under anonymous study codes) simulation/ — the main study runner (outlier index, average genetic relatedness, internal relatedness and random keeper selection) and the threshold-sensitivity runner behind Supplementary Table S6, per-replicate and per-generation results for all 50 replicates across three breeds, and 33-locus STR genotypes for the 4,835 drawn founders (3,540 Standard Poodles, 859 Doberman Pinschers, 436 Flat-Coated Retrievers), identified by study codes only Anonymisation. All released data are anonymised. Dogs carry study codes only; no registered names, owner details, database identifiers, or other identifying fields appear in any file. Founder genotypes and the Supplementary Figure S2 dataset use separate, unlinked code spaces. Licensing. Code is released under the MIT License; data, manuscript, and figures under CC-BY-4.0. See LICENSE-CODE and LICENSE-DATA in the archive. Known artifact. The band_infr_delta, band_norm_delta and band_hfreq_delta columns in the replicate files subtract two band measures computed on different denominators and are not interpretable. This is documented in README.md. All other delta columns are valid.