Department of Fisheries and Wildlife Quantitative Fisheries Center

Parameter estimation performance of a recapture-conditioned integratedtagging catch-at-age analysis modelMatthew T. Vincenta,∗, Travis O. Brendenb, James R. BencebaSecretariat of the Pacific Community, Oceanic Fisheries Program, BP D5, Noumea, New Caledonia 98848bQuantitative Fisheries Center, Department of Fisheries and Wildlife, Michigan State University, 375 Wilson Rd., UPLARoom 101, East Lansing, Michigan, USA, 48824-1101AbstractRecapture-conditioned models are infrequently used to analyze tag-recovery data, but have been proposedas an alternative to release-conditioned models for estimating movement from tagging studies when tag-lossprocesses (e.g., tag reporting, tag shedding) can be assumed constant and estimates of these processes arenot available. Through simulations, we investigated the performance (bias and precision) of a recapture-conditioned integrated tagging catch-at-age analysis (ITCAAN) under varying model complexities and in-termixing rates and compared the results to those from a release-conditioned ITCAAN. We also investigatedhow misspecification of natural mortality, parity in population productivities, tag shedding, and spatially-varying reporting rates affected model estimates. At low intermixing rates, estimates of total abundanceand spawning population abundances were accurate and precise, with precision decreasing when naturalmortality was estimated for the recapture-conditioned ITCAAN. Accuracy and precision of individual popu-lation abundances declined with higher intermixing rates, with the largest bias and lowest precision occurringwhen estimating relative reporting rates. Assuming reporting rates were spatially constant in the ITCAANwhen they varied regionally in the operating model led to biased estimates of movement rates and pop-ulation abundances for both ITCAANs; attempting to estimate relative reporting when reporting variedspatially greatly improved parameter estimates compared to assuming spatially constant reporting. Whentag shedding was simulated to occur, the recapture-conditioned ITCAAN yielded unbiased estimates of to-tal abundance without additional data on the tag-shedding rate, whereas the release-conditioned ITCAANestimates were dependent on the quality of the tag-shedding estimates. For most scenarios investigated,the release-conditioned ITCAAN estimates were less biased and/or variable compared to the recapture-conditioned models. However, both models performed poorly in estimating population specific abundancesfor scenarios when intermixing rates were high and that assumed regionally constant reporting rates in theITCAAN but varying rates in the operating model.∗Corresponding authorEmail address:mtvincen@vt.edu(Matthew T. Vincent)Preprint submitted to ElsevierApril 30, 2019

Keywords:tag integrated assessment, recapture-conditioned, tagging, catch-at-age, ITCAAN, simulationanalysisIntroduction1Spatially-explicit population assessment models simultaneously estimate abundances, mortalities, and2movement rates of populations that are exploited as mixed stocks during the fishing season (Goethel et al.,32011). Herein we define a population as an interbreeding group of fish that are self-sustaining and share4similar life history characteristics. We define a stock as an exploited fishery unit delineated by region5of harvest (Cadrin et al., 2004). Therefore, a mixed stock is comprised of individuals from two or more6populations that are exploited as a single unit. Mixed stocks create overharvest risks for less productive7populations depending on how stocks are managed (Ying et al., 2011; Guan et al., 2013; Hulson et al.,82013; Molton et al., 2013; Li et al., 2014). Integrated tagging and catch-at-age analysis (ITCAAN) models9(Maunder, 2001; Goethel et al., 2015b; Vincent et al., 2017), which incorporate tag-recovery data within a10statistical catch-at-age assessment model, are spatially-explicit assessments that can assess and help manage11mixed stocks.12Two approaches are generally used to analyze tag-recovery data. The most common approach is for13tag recoveries to be conditioned on the number of tags released (release-conditioned framework) (Brownie14et al., 1987; Hoenig et al., 1998; Frusher and Hoenig, 2003; Latour et al., 2003; Jiang et al., 2007). We15refer to this approach as a release-conditioned framework as this accurately describes the denominator of16the recovery probability, but it has also been called the tag-conditioned model (McGarvey and Feenstra,172002; McGarvey, 2009; McGarvey et al., 2010). The number and probability of tags never recovered are18an important component of a release-conditioned framework. The probability of never recovering a tag is19influenced by several tag-loss processes including tag reporting, tagging mortality, and tag shedding; these20tag-loss processes must be accounted for in a release-conditioned framework to prevent biased parameter21estimates (Hampton, 1997; Denson et al., 2002; Cowen et al., 2009; Brenden et al., 2010; Vandergoot et al.,222012). The other tagging framework is to condition tag recoveries on the total number of recoveries (McGar-23vey and Feenstra, 2002; McGarvey, 2009). To remain consistent with published literature, we refer to this24method as the recapture-conditioned framework; however, the likelihood formula uses terminal tag recover-25ies (i.e. tags that were caught and returned to the tagging agency). The recapture-conditioned framework26was proposed to eliminate the need to account for tag-loss processes (e.g., tag reporting) when estimating27movement rates from tag-recovery data (McGarvey and Feenstra, 2002). Removing the need to account for28tag-loss processes can be beneficial because studies to accurately estimate parameters associated with these292