Modeling larval dispersion of rockfish: A tool for marine reserve design?

Marine reserves have been suggested as an important tool for rockfish management and conservation in the northeast Pacific Ocean. One issue confronting effective reserve design is to ensure that larvae released within a reserve system are not lost through dispersal but actually contribute to the population within the reserve areas and beyond. As a first attempt to address this issue for marine reserves in the Aleutian Islands (AI) and Gulf of Alaska (GOA), we modified the particle-tracking module of a three-dimensional circulation model for the northeast Pacific to incorporate simple larval behaviors such as diel vertical migration. We used the model to simulate dispersal of rockfish larvae during peak months of larval release from a suite of potential reserve locations in the AI and GOA. Because larval behavioral patterns are unknown for most rockfish species, we incorporated several alternative behavioral models in the simulations. We also addressed intraand interannual variation in dispersal by repeating the simulations with larval release occurring during several different months for two different years. Model results indicate that retention of larvae near release sites is greatest for sites in the AI and least for sites in the GOA. However, we regard these results as preliminary and as a demonstration of the modeling approach rather than as an actual basis for selecting reserve areas. 252 Stockhausen and Hermann—Modeling Larval Dispersion of Rockfish Introduction Rockfishes (Sebastes spp.) represent one of the most diverse (over 70 species) and economically valuable multispecies resources for commercial and recreational fisheries along the Pacific coast of the United States and Canada (Love et al. 2002). Rockfish have been taken commercially since 1875 in California and since the early 1900s off Alaska. In Alaska waters, a large fishery for Pacific ocean perch (POP, Sebastes alutus) by the USSR and Japan developed in the early 1960s. Catches quickly peaked in the mid-1960s at nearly 500,000 t, but were followed by a precipitous decline at the end of the 1960s that continued into the next decade (Love et al. 2002). More recently, estimated stock abundance has rebounded to ~50% of that in the early 1960s, apparently due to a shift to environmental conditions that favor recruitment success (Heifetz et al. 1999, Ito et al. 1999, Hanselman et al. 2005, Spencer et al. 2005). Directed fisheries currently exist in Alaska for POP, northern rockfish (S. polyspinis), and dusky rockfish (S. ciliatus). All other rockfish species are considered nontarget species and can only be retained as certain percentages of the targeted species catch. Under current harvest strategies, no federally managed rockfish species in Alaska are considered to be overfished (NPFMC 2004a,b). However, there is concern that existing harvest strategies for rockfishes may be inadequate (e.g., Clark 2002, Dorn 2002, Ianelli 2002, Berkeley et al. 2004). The life history characteristics of many rockfish species make individual stocks particularly vulnerable to overexploitation and slow to recover. In particular, many species mature slowly (age at 50% maturity greater than 5-10 years) and are long-lived (50-150 years) (Love et al. 2002). Recruitment success can be exceedingly intermittent (Ralston and Howard 1995, Love et al. 2002). Further, at least some species show evidence of genetic divergence and stock structure on small spatial scales (Withler et al. 2001, Buonaccorsi et al. 2002, Matala et al. 2004). Marine reserves (i.e., harvest refugia) have been suggested as an important tool for rockfish management and conservation in the Aleutian Islands (AI) and Gulf of Alaska (GOA), as well as along the West Coast of the United States (Murray et al. 1999, Soh et al. 2001, Berkeley et al. 2004). Potential advantages posited for the use of marine reserves in conjunction with existing harvest strategies include protection from stock depletion and prevention of serial overfishing of substocks (Soh et al. 2001), maintenance of complex population age and spatial structure (Berkeley et al. 2004), and conservation of essential fish habitat (O’Connell et al. 1998). The criteria for designating areas as marine reserves depend on the management goals the reserves are intended to address. For reserves whose purpose is to protect a species from depletion, one element 253 Biology, Assessment, and Management of North Pacific Rockfishes of effective design is to ensure that local populations protected in reserves are self-sustaining; that is, they remain viable in the absence of recruitment from outside the reserve system. Consequently, larvae released within the reserve system must not be completely lost through dispersal, but must contribute to the population within reserve areas as adults (Roberts 2000, Warner et al 2000). In this respect, areas that display a high degree of larval retention may be preferred as reserve sites to areas that have low larval retention. As a demonstration of one approach to address the issue of larval retention for rockfish stocks in the AI and GOA, we coupled an individual-based model (IBM) that incorporated simple larval behaviors such as vertical migration to a three-dimensional circulation model for the northeast Pacific to assess the extent of local retention at potential sites of larval release. We used the model to simulate dispersal of rockfish larvae during peak months for larval release (i.e., parturition) from a suite of potential reserve locations in the AI and GOA. We addressed temporal variation in dispersal by simulating hydrodynamic currents for two different years and three different release periods within each year. Also, because larval behavioral patterns are uncertain for most rockfish species, we repeated model runs using several alternative behavioral models. Among the sites considered, we regarded release sites that exhibited strong retention patterns that were robust to variation in release period and larval behavior as the best candidates for reserve location. We regard the results presented here as a demonstration of the modeling approach and preliminary at best. They should not be used as the basis for reserve selection. Materials and methods The early life history of rockfishes in Alaska waters is generally characterized by a lack of species-specific information for most species, and information is sparse even for the best-studied species. Difficulty in identifying larval Sebastes to species level confounds understanding of species-specific patterns and behavior (Matarese et al. 2003). As such, our simulation model reflects an amalgam of details drawn from studies of disparate individual species or from early life history characteristics of rockfishes classified only to the generic level. Larval IBM We developed a very simple IBM for rockfish larval behavior that incorporated the ability to actively migrate vertically to occupy a preferred depth range. Modeled larvae that were outside their preferred depth range at any time immediately began to swim vertically (up or down) at a prescribed rate until they entered their preferred depth range. In the current IBM configuration, preferred depth ranges could differ between 254 Stockhausen and Hermann—Modeling Larval Dispersion of Rockfish daytime and nighttime to model diel vertical migration, a type of active larval behavior that could result from feeding behaviors, light sensitivity, or predator avoidance but could also enable larvae to utilize vertical current shear to modify dispersal patterns from those of passive particles (Neilson and Perry 1990). Unlike many marine fish species, rockfish do not undergo a planktonic egg stage. Rockfish are a primitive viviparous group, with females extruding larvae rather than eggs in a process known as “parturition” (Love et al. 2002). Extruded larvae are ready to begin feeding and, at 3-7 mm standard length, are comparable in size to first-feeding larvae of species with planktonic eggs (Kendall and Lenarz 1987). In Alaska waters, parturition occurs primarily during the spring and summer for most species (see references in Wyllie Echeverria 1987, Love et al. 2002). Depending on species, larvae may be released near the bottom or in midwater (Love et al. 2002). Recent studies of POP in British Columbia suggest that, for this species, adult females migrate to the mouths of submarine canyons and release their larvae at depth (500-700 m); the larvae subsequently remain at depth for a month or more prior to moving to shallower water (Love et al. 2002). However, most rockfish larvae are typically found above the pycnocline at relatively shallow depths (Ahlstrom 1959, Boehlert et al. 1985, Sakuma et al. 1999, Matarese et al. 2003). Matarese et al. (2003) give the duration of the larval stage as 1-2 months (see also Kendall and Lenarz 1987, Laidig et al 1991, Sakuma and Laidig 1995, Plaza Pasten et al. 2003). After completion of the larval stage following growth to 20-30 mm SL, most rockfish species undergo a pelagic juvenile stage lasting several weeks to months before transitioning to a demersal existence (Love et al. 2002). Little is directly known of larval behavior of rockfish species in the AI and GOA. Studies that examined rockfish larvae focused on temporal patterns of larval abundance or on broad spatial patterns, not on vertical position or diel behavior (e.g., Doyle et al. 2002, Matarese et al. 2003). Complicating these studies is an inability to distinguish most rockfish larvae at the species level (Matarese et al. 2003). More work has been done along the west coast of North America. Ahlstrom (1959) found rockfish larvae off California and Baja California above or in the thermocline (<100 m) with no consistent day/night differences in vertical distribution. Barnett et al. (1984) and Moser and Boehlert (1991) obtained similar results off southern California. Boehlert et al. (1985) found Sebastes larvae off Oregon distri