Genetically reconstructed pedigrees: The costs and benefits of using full-sibling structure to constrain parentage assignments

22 We present a simple yet effective method to improve parentage assignment 23 (PA) accuracy an average of 47% compared to the PA programs PEDAPP 24 (39%), PASOS (53%), and CERVUS (50%) as measured over a wide range of 25 simulated scenarios. The method, termed sibship constraint (SC), uses the 26 results of sibship reconstruction (SR) to constrain assignments from PA output. 27 It works by assigning the PA candidates allocated to the greatest proportion of 28 offspring within a reconstructed full-sibling family to all members of that 29 family. A user-specified minimum threshold value determines which 30 candidate(s) to keep based on assignment proportions. Accuracy comparisons 31 were made between SC method and PA program outputs for the following eight 32 variables: minimum threshold value, SR program used, mating strategy, mean 33 family size, proportion of true parents sampled, number of loci used, 34 genotyping error rate, and cohort assignment error rate. The cost of using the 35 SC method was a decrease in assignments made to offspring whose true parents 36 were sampled by 9% compared to PEDAPP and PASOS, and 21% compared to 37 CERVUS. However, this cost was more than offset by the benefit of a 38 decreased number of assignments made to offspring whose true parents were 39 not sampled by 80% (PEDAPP), 82% (PASOS), and 84% (CERVUS), which 40 resulted in marked improvement to assigned accuracies. The SC method is 41 3 highly flexible in that it can use outputs from six SR and twelve PA programs, 42 with all SR-PA pairings possible. Additionally, the method is fully automated 43 within the freely-available software program PEDAGREE. 44 4