Modeling the impacts of two age‐related portfolio effects on recruitment variability with and without a marine reserve

Abstract

Many rockfish species are long-lived and thought to be susceptible to being overfished. Hypotheses about the importance of older female rockfish to population persistence have led to arguments that marine reserves are needed to ensure the sustainability of rockfish populations. However, the implications of these hypotheses for rockfish population dynamics are still unclear. We modeled two mechanisms by which reducing the proportion of older fish in a population has been hypothesized to influence sustainability, and explored whether these mechanisms influenced mean population dynamics and recruitment variability. We explored whether populations with these mechanisms could be managed more sustainably with a marine reserve in addition to a constant fishing mortality rate (F) than with a constant F alone. Both hypotheses can be seen as portfolio effects whereby risk of recruitment failure is spread over a "portfolio" of maternal ages. First, we modeled a spawning window effect whereby mothers of different ages spawned in different times or locations (windows) with local environmental conditions. Second, we modeled an offspring size effect whereby older mothers produced larger offspring than younger mothers, where length of a starvation period over which offspring could survive increased with maternal age. Recruitment variability resulting from both models was 55-65% lower than for models without maternal age-related portfolio effects in the absence of fishing and increased with increases in Fs for both models. An offspring size effect caused lower output reproductive rates such that the specified reproductive rate input as a model parameter was no longer the realized rate measured as the reproductive rate observed in model results; this quirk is not addressed in previous analyses of offspring size effects. We conducted a standardization such that offspring size effect and control models had the same observed reproductive rates. A comparison of long-term catch, the probability of falling below a biomass threshold, and recruitment variability over a range of exploitation rates for models with an age-related portfolio effect showed no benefit of a marine reserve implemented in addition to a constant F (as compared to a constant F alone) for populations with sedentary adults and sedentary or mobile larvae.

Saving large fish through harvest slots outperforms the classical minimum‐length limit when the aim is to achieve multiple harvest and catch‐related fisheries objectives

“…In relation to a positive effect of maternal size on offspring quality, our findings are in agreement with several previous modelling studies that implied size‐dependent offspring quality effects are not of sufficient importance to independently justify alternative size‐selectivity during harvest (Arlinghaus et al, ; Calduch‐Verdiell, MacKenzie, Vaupel, & Andersen, ; McGilliard, Punt, Hilborn, & Essington, ; O'Farrell & Botsford, ; Shelton et al, ). Experimental studies with artificial insemination conducted under controlled conditions (e.g.…”

Spatial variation in exploited metapopulations obscures risk of collapse

“…Previous studies have also investigated spatial mismatches in fisheries contexts to understand consequences stemming from spatial mismatches among biological processes and available data, as well as the spatial implementation of fisheries for yield and measure of population status assessments (Cope and Punt , McGilliard et al. , , Spies et al. ).…”

“…Our models address this gap by building on existing research with increased biological realism to allow for (1) adult migration rates among subpopulations rather than only dispersal associated with recruitment (e.g., Cope and Punt , McGilliard et al. , , Spies et al. ); (2) a range of complex spatiotemporal correlations in the stochastic populations dynamics (but see McGilliard et al.…”

“…), spatially complex density dependence (McGilliard et al. ), Allee effects, cost, and data limitations for spatial management). We also ignore many alternative spatial allocation strategies that may be explicitly designed to maximize long‐term yields or minimize spatiotemporal variability in part because these approaches would require a spatial assessment model, which is out of the scope of this study.…”

“…The numerical model (Model 3), on the other hand, is more detailed but outcomes are context dependent. Moreover, both models ignore many biological and management scenarios that are likely to further complicate spatial patterns (e.g., behaviorally or geographically complex migration; MacCall et al 2018, Rogers et al 2018), spatial and age-specific mortality (McGilliard et al 2011), spatially complex density dependence (McGilliard et al 2017), Allee effects, cost, and data limitations for spatial management). We also ignore many alternative spatial allocation strategies that may be explicitly designed to maximize long-term yields or minimize spatiotemporal variability in part because these approaches would require a spatial assessment model, which is out of the scope of this study.…”