Grandparent Inference from Genetic Data: The Potential for Parentage‐Based Tagging Programs to Identify Offspring of Hatchery Strays

Abstract

Fisheries managers routinely use hatcheries to increase angling opportunity. Many hatcheries operate as segregated programs where hatchery‐origin fish are not intended to spawn with natural‐origin conspecifics in order to prevent potential negative effects on the natural‐origin population. Currently available techniques to monitor the frequency with which hatchery‐origin strays successfully spawn in the wild rely on either genetic differentiation between the hatchery‐ and natural‐origin fish or extensive sampling of fish on the spawning grounds. We present a method to infer grandparent–grandchild trios by using only genotypes from two putative grandparents and one putative grandchild. We developed estimators of false‐positive and false‐negative error rates and showed that genetic panels containing 500–700 single‐nucleotide polymorphisms or 200–300 microhaplotypes are expected to allow application of this technique for monitoring segregated hatchery programs. We discuss the ease with which this technique can be implemented by pre‐existing parentage‐based tagging programs and provide an R package that applies the method.

The Genetic Composition of Wild Steelhead Based on Spatial Proximity to a Hatchery

“…Our results illustrate the complexities inherent in using genetic markers to infer demographic membership of individuals in the presence of historical, legacy, and contemporary gene flow, yet the genetic character of the group can still be distinguished. Additionally, we note that analytical techniques available in the near future (e.g., grandparentage analysis; Delomas and Campbell 2022) have the potential to quantify stray rates by hatchery offspring, and samples used in this study can be regenotyped at new marker panels to perform such analyses. Moving forward, genetic tools will continue to play an important role in understanding the patterns and processes at play between hatchery and wild populations of steelhead.…”

“…Additional genetic techniques will become available in the near future to examine hatchery influence in wild populations. Parentage-based tagging involves genotyping complete sets of hatchery broodstock to genetically tag their offspring (Hargrove et al 2021), and specific forms of parentagebased tagging, including grandparentage analysis, may be able to directly quantify hatchery straying (Delomas and Campbell 2022). Recent simulation work showed that grandparent-grandchild trios can be identified using genotypes from two putative grandparents (i.e., two hatchery broodstock) and one putative grandchild (i.e., juvenile sampled on the landscape), and statistical packages have already been developed to perform these analyses (Delomas and Campbell 2022), but marker panels with sufficient variability to facilitate grandparentage analysis have yet to be developed.…”

“…Hatcheries that exclude the use of natural‐origin fish as broodstock (i.e., segregated broodstock) can lead to genetic divergence between natural and hatchery‐origin stocks (Delomas and Campbell 2022), which may ultimately affect the population genetic structure of wild stocks (Pearse et al. 2011; Pearse and Garza 2015; Powell and Campbell 2020; although see Hansen 2002; Gow et al.…”

Single‐parentage assignments reveal negative‐assortative mating in an endangered salmonid

“…The false positive rate, particularly for unrelated pairs, is typically small enough that this simple estimator is computationally infeasible. Previous approaches to this issue have used importance sampling (Baetscher et al, 2018 ), but we implemented an alternative approach of stratified sampling to reduce Monte Carlo variance similar to the approach utilized by Delomas and Campbell ( 2021 ) for grandparent–grandchild trios. The domain of the estimator was stratified by the number of observed MIs in a potential parent–offspring pair.…”

“…This is especially true when close relatives of the parent (e.g., full‐siblings and half‐siblings) are also present in the dataset. Therefore, we build upon existing methods and use an approach that performs single‐parentage assignments through a pairwise process of calculating the likelihood ratio for a parent–offspring relationship to that of the pair being unrelated (Anderson & Garza, 2006 ; Baetscher et al, 2018 ; Marshall et al, 1998 ; SanCristobal & Chevalet, 1997 ; Thompson, 1976 ) and combine it with estimates of the associated error rates for these assignments using stratified sampling (Delomas & Campbell, 2021 ). This methodology is implemented in the R package grandma ( https://github.com/delomast/gRandma ).…”