Premium
A New Estimator to Correct for Bias from Tag Rate Expansion on Natural‐Origin Fish Attributes in Mixed‐Stock Analysis Using Parentage‐Based Tagging
Author(s) -
Delomas Thomas A.,
Hess Jon E.
Publication year - 2021
Publication title -
north american journal of fisheries management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 72
eISSN - 1548-8675
pISSN - 0275-5947
DOI - 10.1002/nafm.10537
Subject(s) - estimator , hatchery , fishery , fish <actinopterygii> , stock (firearms) , statistics , maximum likelihood , restricted maximum likelihood , fish stock , econometrics , biology , mathematics , geography , archaeology
In fisheries where hatchery‐ and natural‐origin conspecifics occur as mixed stocks, it is often important to estimate both the natural‐origin proportion of the mixture and the composition of attributes within the natural‐origin portion (e.g., genetic stock, sex, and age‐class). These estimates are facilitated by parentage‐based tagging, which allows large numbers of hatchery fish to be efficiently tagged and later identified. When tag rates are less than 100%, the untagged fish in the mixture include both untagged hatchery fish and natural‐origin fish. Unbiased estimation of the abundance and attribute composition of the natural‐origin portion requires an estimator that accounts for tag rates. Two estimators are described: one “accounting‐style” estimator similar to previously described approaches and one maximum likelihood method. These estimators were evaluated and compared using simulations mimicking estimation of the composition of fish migrating past a dam. The two estimators performed similarly at reducing bias when tag rates were high, but the maximum likelihood method had smaller mean square error when tag rates were low. We provide an R package to allow usage of this estimator in a wide variety of fisheries applications.