The pace of harvest and recovery in geoduck clam stocks fifty years into the fishery

Highlights

• Slower than expected recovery for a geoduck fishery in Washington.

• Harvest rates from updated model are mismatched with observed recovery patterns.

• Recovery rates had strong spatial patterns and were weakly related to current speed.

• Episodic recruitment events likely the primary driver of post-harvest recovery.

• A switch in management to a tract-recovery based strategy is recommended.

Abstract

The commercial fishery for subtidal Pacific geoduck clams (Panopea generosa) in Washington State, USA, is substantial both by the amount of biomass extracted (2 million kg in 2019) and by the economic value (US $50 million annual). Management for this fishery, which began in 1970, is challenged by this species’ long lives, cryptic behavior, and recruitment variability. Current management, including a 2.7 % annual harvest rate, is based on an equilibrium yield model most sensitive to a natural mortality parameter estimated using sampled age distributions. We collected a large sample for aging from four new sites in Puget Sound to update the yield model and explore its applicability to individual management regions. Based on the new age distributions, natural mortality varies among regions, but is similar overall (0.0231 yr⁻¹) to the value used in original formulation of the yield model (0.0226 yr⁻¹), indicating that a change to the harvest rate is not needed if following the outputs of this model. However, concerns remain about the sustainability of the current harvest rate, particularly the time for harvested tracts (delineated harvest areas ranging in size from 0.06 to 1.45 km²) to return to their preharvest density. We calculated tract recovery rates from serial post-harvest density surveys, compared them to the expected recovery rate derived from the age-based equilibrium model, and investigated possible mechanisms behind fast or slow tract recovery. Using multiple scuba surveys spaced over four decades from 38 tracts in the South Puget Sound, we estimate that tracts recover on average at 0.03 geoducks per square meter per year. The projected average time to recovery is 55 years, compared to a projection of 39 years when the yield model was developed. There is some support for the hypothesis that tract recovery rates have slowed in recent decades. Although there were strong spatial patterns, with higher recovery in central basins compared to within inlets, there was only weak support for the hypothesized mechanisms behind tract recovery such as mean current speed. Our inability to clearly relate recovery to physical variables such as habitat, paired with extreme interannual variability in recruitment suggested by the age distributions, supports the hypothesis that the strength, timing, and spatial extent of episodic recruitment events is the primary driver of tract recovery. Currently, harvestable biomass estimation and fishery management assume that all previously harvested tracts will eventually recover to preharvest densities. However, in the South Puget Sound, only 40 % of tracts are likely to recover on a 50-year time frame relevant to management. Despite our validation of the key parameter used in the yield model, we advocate switching geoduck management to a tract-recovery-based strategy in which harvest is paced to known tract recovery, and biomass estimation is restricted to only those areas likely to hold renewable resource. This study highlights the importance of evaluating recovery patterns of stocks once data over appropriate timescales are available, and comparing to the management model to determine if fishery goals are being met.

Introduction

Fishery managers often use model outputs as the basis for decisions on harvest rules. If the harvest rate from a model is implemented, there later needs to be an empirical-based evaluation to assess the impact of model-based rules to the fishery. The initial target may need to be adjusted according to the realities of the fishery and post- harvest observations. The use of reference points to create harvest rules has been a common default for managers across many fisheries (Smith et al., 1993; deYoung et al., 1999; FAO, 1996; Richards and Maguire, 1998; Haigh and Sinclair, 2000). Harvest rate management is the most basic form of management by reference points, and the USA Washington State geoduck fishery falls under this umbrella. However, there are times when these reference points may not be useful. Mangel et al. (2002) identifies some of these instances, including for species with limited mobility and when there are long-term productivity changes. The implications of limited mobility are discussed in Orensanz and Jamieson (1998), who summarize that traditional fishery science tools such as population models and stock recruitment analysis which assume populations are self-contained can be problematic for stocks with “strong and persistent” spatial structure (such as for bivalves and abalone). Geoducks experience long-term productivity changes in the northeast Pacific with their long, sedentary life history, and are an example of stocks that are affected by climatic regime shifts (Strom et al., 2004). For these types of stocks, Parma (2002) points out that rather than using complicated rules that rely on uncertain and changing assessments which use biological reference points to maintain sustainable stocks, basic decision rules made by a monitoring feedback may be more effective.

The slow-paced dynamics of Pacific geoduck, Panopea generosa, populations present challenges to fishery managers. This species is long-lived (160+ years; Shaul and Goodwin, 1982; Goodwin and Shaul, 1984; Noakes and Campbell, 1992; Bureau et al., 2002; Strom, 2003), has a very low natural mortality rate (1.4–5.4 %; reviewed in Orensanz et al., 2004; Zhang and Campbell, 2004), exhibits large recruitment variability (Valero et al., 2004; Zhang and Campbell, 2004; Zhang and Hand, 2006; Becker et al., 2012) and their populations are difficult to survey over large areas due to the intense nature of the methods (enumeration by scientific divers) and the fact that many individuals are obscured beneath the sediment (Bradbury et al., 2000). It may take decades before any kind of feedback of population dynamics is gained after harvest rules for this geoduck fishery are implemented given that managers are making observations on a much shorter timescale (usually annually) than the slow changes in abundance that are occurring (Orensanz et al., 2004). Despite these challenges, it is prudent for fishery managers to monitor populations and measure recovery of harvested areas. Orensanz et al. (2004) emphasized the value of monitoring and feedback within the geoduck fishery management and of decision rules based on empirical data rather than relying on numerical modeling outputs. Our research objective of evaluating observed recovery rates within the Washington State geoduck fishery and making a comparison to the currently used model-derived harvest rates could provide an important perspective of fishery management of this species. This type of analysis would help fishery managers determine if, for example, harvest rates are too high to support a sustainable fishery.

Pacific geoducks are large burrowing clams with an average size of nearly a kilogram that are found throughout the Salish Sea on sand, mud, and gravel substrates at 0–90 m depth. This manuscript addresses the U.S. portion of the Salish Sea, identified here as all inland marine waters of Washington State, including the Straits of Georgia and Juan de Fuca (Fig. 1). We use the term Puget Sound throughout this manuscript to refer to these waters. Subtidal geoduck populations support an important commercial fishery in Washington State which began in 1970. Shortly after the commercial fishery started, there was a dramatic increase in total landings that peaked in 1977 at 3.92 million kg (Fig. 2). Ex-vessel price per pound of geoduck was initially very low, until the 1990s when price increased tenfold. The price again doubled in the late 2000s, with record value occurring in 2012. The management strategy of this fishery has gone through several phases over time. Annual harvest rate was initially set at 10 % of the virgin (unfished) stock estimate. The current management regime began in 1998 with the adoption of an age-based equilibrium yield model (Bradbury and Tagart, 2000) and a 2.7 % harvest rate which aims to reduce the spawners-per-recruit to 40 % of virgin biomass (an F₄₀ strategy). This harvest rate is used with current commercial geoduck biomass estimates to calculate a Total Allowable Catch for six active management regions – South Puget Sound, Central Puget Sound, North Puget Sound, Hood Canal, San Juan Islands and Strait of Juan de Fuca (see Bradbury et al., 2000 for overall management plan).

The Puget Sound fishery is jointly managed by many Native American Tribes, Washington Department of Fish and Wildlife (WDFW) and Washington Department of Natural Resources as parties to treaties signed in the mid-1850s and following the 1994 Rafeedie court decision. The current biomass is estimated at about 88 million kilograms on 206 tracts (areas identified by mangers that have suitable geoduck habitat with some density of geoducks) with a total area of about 107 km² (WDFW unpublished data). The 2.7 % harvest rate is applied to the commercially available biomass – those areas that are surveyed, are outside of Department of Health prohibited areas, and are between 5.5 and 21 m in depth (Bradbury et al., 2000). Several life history characteristics previously described above contribute to the need for low harvest mortality of this species, including its longevity, recruitment variability and low natural mortality rate. The overall strategy (as outlined in Bradbury et al., 2000) is to harvest a small number of individual tracts down to low densities, place them in recovery for several decades, and rotate harvest effort to new tracts. A cornerstone of geoduck fishery management is to conduct surveys on tracts before and after harvest. After harvest is closed on a tract, the tract receives a post-harvest survey and is placed into recovery status. On a subset of such tracts, a series of additional postharvest surveys can continue so that managers can track the status of these recovering tracts. Historically there has not been a consistent time interval for subsequent post-harvest surveys; they occurred opportunistically based on available resources and management needs. However, managers are considering a schedule in which all harvested tracts receive a survey every 10 years until recovery is reached.

The fishery has been operating for over 20 years with the current harvest strategy, and managers have not formally re-evaluated it during that time. This re-evaluation became a pressing need given concerns among managers and harvesters about the apparent longer than expected length of time for harvested areas to recover and the result of fewer areas that could be identified and opened up for harvest. To accomplish a management strategy review, there were three main study objectives: first, to update some of the yield model parameters with recent data to evaluate possible changes over time; second, to calculate tract recovery rates after harvest; and third, to evaluate how recovery rates may differ over space and time and how they may relate to physical attributes. For the first study objective we aimed to update growth parameters and natural mortality rates with contemporary age data and input them into the currently used model to compare the resulting fishery yield outputs to the 2.7 % harvest rate that was adopted in 1998. The 2.7 % harvest rate was derived from an age-based equilibrium yield model using parameter estimates from age data collected in 1979–1981 and from various literature sources (Bradbury and Tagart, 2000). The natural mortality estimate used in the Bradbury and Tagart model was derived from age data of 2,157 geoducks that came from 14 sites in four regions of Puget Sound. The paired age and size data used for the growth parameters in the Bradbury and Tagart model, reported in Hoffmann et al. (2000), came from 11 of those 14 sites but only used 234 total geoducks. Site-based natural mortality rates were not possible for that study since the sample size at each site was too low, so a Puget Sound natural mortality rate was calculated based on all sites combined. The current study aimed to make growth parameter estimates based on a much larger sample size and to calculate regional natural mortality rates. Work done by Valero (2011) showed that regional harvest rates may be an improvement to the current single, statewide rate, due to spatial variability in both recruitment and recovery rates.

Additionally, more recovery data are available from harvested geoduck tracts than were available to the fishery managers who adopted the current harvest strategy back in 1998. With these data on hand, our second objective was to calculate tract recovery rates and compare them to the expected recovery rate derived from the age-based equilibrium model. If these recovery rates align (indicating that the harvest rate removes less than or equal to the amount of biomass gained through growth and recruitment), this would support that the current harvest strategy is sustainable. If they do not align, it would indicate that the harvest rate is not appropriate for sustainable management of geoduck in Puget Sound. By comparing the calculated recovery rate from the yield model with observed recovery rates from the fishery that has operated for the last 50 years, managers would be able to begin to update and improve the management strategy for the geoduck fishery.

Another goal of this research was to uncover some of the mechanisms behind the variable recovery rates observed across tracts in Puget Sound. Several factors related to harvest practices, abiotic conditions, space, and time could influence the delivery of competent larvae, post-settlement survival, and eventual growth and survival of juveniles into fishable size – “recruitment”. Of course, biotic interactions could influence recruitment as well, but in general we do not have the appropriate data to probe the relationships between tract recovery and the abundance of other species. We do have information on some of the abiotic (e.g., current speeds, seafloor substrate) and fishery-related (e.g., geoduck density after harvest) variables to test some related hypotheses. In this research, we hypothesized that the following factors may influence the rate at which geoduck tracts recover to preharvest densities after harvest:

• H1: Geoduck tracts with a higher percentage of preferred substrate (i.e. pure sand) will recover faster than tracts with less.

This is related to the observation that some substrate types hold higher densities of geoduck than others (data presented herein, see online appendix table S3). Of course, this density difference is not necessarily related to the speed of recovery, but we hypothesize that the two will be related.

• H2: Geoduck tracts with higher average current speeds will recover faster than tracts with slower average current speeds.

Goodwin (1990) originally observed higher geoduck recruitment in locations with medium velocity currents. It could be that geoducks would both recruit and grow faster where moderate currents allowed for the delivery of settlers, oxygenated water, and phytoplankton food compared to slower currents, but unlike faster currents did not move sand to the extent that habitat was removed or buried, and did not inhibit settlement and feeding. In this effort, we apply Goodwin’s original hypothesis to geoduck tracts only, recognizing that high current areas scoured of sand or mud are generally not designated as geoduck tracts. Therefore, we hypothesize that our higher-current geoduck tracts are in effect intermediate-current areas when compared to the entire Puget Sound, and simply test for increasing recovery with increasing currents.

• H3: Geoduck tracts that had higher pre-harvest densities will recover faster than those with lower preharvest densities.

This hypothesis is related to H1 and H2, in that it assumes that whatever conditions, including substrate and current, led to the accumulation of dense adult geoduck on a tract prior to harvest would also lead to fast recovery after harvest.

• H4: Geoduck tracts that had high postharvest densities remaining after fishing will recover faster than those with lower postharvest densities.

This hypothesis assumes gregarious settlement of geoduck for which there is some evidence (Fyfe, 1984; Goodwin and Shaul, 1984), and assumes that moderate densities of adult geoduck do not inhibit settlement and growth of juveniles (through, for instance, their filtration activities).

• H5: Geoduck tracts that had a lower overall density, or proportional density removed, will recover at a faster rate than those in which a higher density or percentage was removed.

This hypothesis tests whether fishing activity, as measured by the amount or proportion of geoduck removed, impacts recovery. Geoduck excavation could displace sediments or favorable organisms that inhibit settlement or growth of juveniles. If so, tracts that had a lower amount of harvest impact would recovery faster than tracts that were more heavily impacted.

• H6: Average geoduck tract recovery rates have not changed over the last 40 years.

This is a null hypothesis against the possibility that recovery rates are speeding up or slowing down over time. Rates could speed up due to changing ocean conditions (increased growth due to more or different assemblages of phytoplankton) or increased available habitat space freed up through harvest. Rates could slow due to those same changing ocean conditions (acidified water having lethal or sublethal effects on larvae; see Timmins-Schiffman et al., 2019) or a decrease in fertilization during spawning due to the decreased density of adults as tracts are serially depleted.

• H7: Geoduck tract recovery rates will not differ between the first and later parts of the recovery period.

This is a null hypothesis of constant recovery rate against any observed shape to the recovery curve. It is interrelated with several hypotheses above. Space and resources freed up through harvest could lead to a faster initial recovery rate, which decreases as the density of adults slowly increased on a tract over time. Conversely, if settlement is gregarious, or if some remaining adults stabilize the habitat and help retain the preferred sand substrate, recovery rates could be initially slow, but increase after some adult density threshold is reached.

• H8: Geoduck tracts will have similar recovery rates to nearby tracts

This last hypothesis tests the scale of variation in recovery. Will the features that influence recovery be shared by neighboring tracts or will recovery operate on some other scale? Of particular interest are the differences among central basins and blind inlets within the South Puget Sound study area. We tested for a difference in recovery rate among post-hoc, contiguous groupings of tracts within different inlets or basins.