Printer Friendly

Long-term projections of eastern oyster populations under various management scenarios.

ABSTRACT Time series of fishery-dependent and fishery independent data were used to parameterize a model of oyster population dynamics for Maryland's Chesapeake Bay. Model parameters are (1) fishing mortality, estimated from differences between predicted and reported landings scaled to a fishery-independent estimate of exploitation rate; (2) natural mortality, estimated front the ratio of dead to live oysters in fishery-independent surveys; (3) recruitment, estimated from annual changes in the market oyster stock and total mortality rates; and (4) carrying capacity, estimated from predicted densities of oysters in the absence of either exploitation or excess mortality from parasitic diseases, multiplied by charted areas of high quality and marginal habitat. The model predicts continuing decline in population abundance and biomass in the absence of significant management intervention. Moderate decreases in fishing mortality, alone or in combination with increases in recruitment through stock enhancement, could reverse recent population trends, resulting in a larger population and improved harvests within 1-2 decades, even if currently high rates of natural mortality are sustained.

KEY WORDS: Crassostrea virginica, fishing mortality, models, natural mortality, oyster, population dynamics, recruitment

INTRODUCTION

In 2000, the inter-jurisdictional Chesapeake Bay Program (CBP) established a goal to increase the bay's population of native oysters (Crassostrea virginica) 10-fold by 2010 from a 1994 baseline. With colleagues, we developed a preliminary quantification of the oyster population in Maryland's portion of the bay, along with methods for tracking population trends and predicting annual landings with an index of relative biomass (Jordan et al. 2002). A logical next step was to examine the potential contributions of various management strategies toward meeting the 10-fold restoration goal. To this end, we used time series of fishery dependent and fishery-independent data to parameterize a model of oyster population dynamics. The model projects long-term trends in the Maryland stock of market-sized oysters ([greater than or equal to] 76 mm in shell height) with variable rates of fishing mortality (F), natural mortality (M), and recruitment (R). The model is limited to market oysters, first for simplicity, second because the relationship between oysters <76 mm in shell height (pre-recruits) and the market stock has been difficult to quantify (Jordan et al. 2002), and third, because the goals of the CBP and the state of Maryland include restoration of the oyster fishery, in addition to the population as a whole (Maryland Oyster Roundtable 1993, CBP 2002). Moreover, the standing stock of larger oysters is an important indicator of population viability, whether the objectives of the restoration program are ecological, economic, or both.

The model is stochastic, that is, for each of many iterations of the model within a simulation, F, M and R vary randomly, constrained by means and standard deviations observed from the monitoring time series or specified by the user. The principal outputs are projected annual means of market oyster stocks and landings. To estimate the uncertainty associated with the results, the model computes percentages of model iterations that achieve specified targets. Thus, a user could evaluate proposed management strategies on the basis of, for example, 90% probability of achieving a desired outcome. Although the CBP oyster restoration goal set 2010 as the year in which to achieve a 10-fold increase, we extended simulations to 2020 to present longer-range views of population trends.

Environmental forcings (e.g., variations in temperature and salinity) are not included explicitly in the model, except in the separate baseline parameter values for three salinity zones. We assume that the effects of environmental variations have been captured in the standard deviations of the baseline model parameters. The time-series data spanned 16 y, from 1986 to 2001, capturing a range of warm, cool, wet, and dry periods, along with other variations in the environment.

The model was designed to evaluate long-term management strategies in response to the CBP 10-y goal, and does not accurately reproduce or project short-term (interannual) population trends. Year-to-year tactical management decisions would be better addressed by a model similar to that developed by Klinck et al. (2001) for Delaware Bay oysters. The CBP (2002) emphasis on sanctuaries (i.e., areas permanently closed to oyster harvest) is an example of a long-term strategy appropriate for the model described here. In Maryland, sanctuaries generally are stocked with seed oysters produced in hatcheries. Consistent with the CBP plan and recent practice, we modeled the sanctuary strategy as a 10% decrease in F (to approximately 10% of productive bottom closed to the fishery), combined with a 10% increase in R (approximating potential hatchery contributions), and projected this scenario over 20 y. Another example of a long term management strategy is projecting the effects of reductions in F in the range of 10-40%, for which the model predicts that initial decreases in landings will be more than compensated by increases after a few years.

The CBP (2002) plan is concerned with selecting sites for restoration projects and how different areas of the bay should be managed to achieve the restoration goal. The models for three salinity zones should be helpful in these decisions, although an ideal model for these purposes would have much finer geographical resolution (e.g., individual oyster bars). The model reported here should be a step toward a more detailed, age-structured, geographically articulated model.

MATERIALS AND METHODS

Parameters of the model are instantaneous rates of natural mortality (M), fishing mortality (F), and recruitment (R), plus point estimates of carrying capacity (K, an upper bound on the stock of market oysters). The unit of measure is one Maryland bushel (~46 L) of oysters. The only boundary condition is an initial estimate of the market stock (in bushels), which can be derived from landings and relative biomass (as described below) for any year for which monitoring data are available (Jordan et al. 2002).

The primary sources of data for model parameterization were (1) Maryland fall dredge surveys of oyster bars (1985-2000), and (2) annual commercial oyster landings reported to the Maryland Department of Natural Resources. The fall dredge surveys provided annual estimates of natural mortality rates and relative biomass of market sized oysters and are described elsewhere (Smith and Jordan 1993, Homer et al. 1996, Maryland DNR 2001, Jordan et al. 2002).

We developed an aggregated model for all of Maryland's Chesapeake Bay and also developed separate baseline parameter values for each of three zones based on salinity data from the fall oyster surveys (Fig. 1). Salinity zones were defined by averaging salinity measurements taken from 43 dredge survey sites in October or November 1990-2000, then assigning sites to categories: high, >14 ppt; medium, 12-14 ppt: and low, <12 ppt. We used two criteria to establish the salinity ranges: (1) including roughly equal numbers of sites in each category so that each would have sufficient data, and (2) gradients in recruitment and the impacts of the oyster parasites Haplosporidium nelsoni and Perkinsus marinus on oyster mortality. At salinity <12 ppt. H. nelsoni infections occur rarely, if ever: P. marinus infections, although chronic in this zone, are associated with low to moderate mortality rates: and recruitment rates of both small and market oysters are typically very low, except in areas where natural recruitment has been supplemented by transplanted seed oysters. In the medium-salinity zone, H. nelsoni epizootics are sporadic, occurring only in drought years, mortality associated with P. marinus is moderate to high, and recruitment is variable. In the high-salinity zone, H. nelsoni infections tend to be enzootic, mortality rates associated with P. marinus infections are consistently high, and recruitment, although variable, tends to be higher than in the lower salinity zones (Jordan 1995, Giescker 2001).

[FIGURE 1 OMITTED]

The model assumes classic logistic population growth, with two enhancements. First, instead of a net rate of population change [r, as in Krebs (1994)], the model includes separate parameters for instantaneous rates of natural mortality (M), fishing mortality (F), and gross annual recruitment (R). Second, each parameter is input for each simulation in Monte Carlo fashion, as a random number from a log-normal distribution with mean and standard deviation computed from the time series of monitoring data.

Natural mortality (M) estimates were based on the ratio of articulated dead oyster shells (D = boxes) to the total of live and dead oysters (L + D) in each dredge sample: M = -loge(D)(L + D). Fishing mortality was estimated from the ratio of reported landings to landings predicted by log-log linear regression from a fishery-independent biomass index for market oysters: F = -loge[0.53 (H/[??])], where H = reported landings, [??] = landings predicted from relative biomass of market oysters, and 0.53 is a scaling constant derived from a fishery-independent estimate of the proportion of the market stock landed during the 1990-1991 season (Smith and Jordan 1993). For that season, the ratio H/[??] was approximately unity. These methods of estimating M and F were described and discussed by Jordan et al. (2002). Gross recruitment (R) was calculated as the instantaneous annual rate of change in the market oyster stock (r = dS/dt), plus total annual mortality (Z = M + F). The term dS/dt was estimated for a 1987-2001 time series by calculating stock size for each year as S = (H)(1 - [e.sup.-F]), where H is annual landings (R could not be computed for 1986, the first year of the monitoring time series).

We made four estimates of carrying capacity, one for each of the salinity zones, and one for the Maryland Chesapeake Bay as a whole. The Maryland Bay Bottom Survey (MBBS) of oyster bars was used to calculate acreage of cultch, mud with cultch, and sand with cultch bottom-type categories for each area. A variety of observations and surveys with patent tong grabs, divers, video, and acoustics indicated that only about 10% of these areas of nominal oyster habitat actually supported oyster populations (Smith et al. 2001, Maryland Department of Natural Resources, unpublished data). Even within productive areas, high densities of oysters ([greater than or equal to] 100 per [m.sup.2]) occurred only in scattered patches. Thus, we assumed that at carrying capacity, 10% of cultch areas from the MBBS would support mean densities of 10 market oysters per [m.sup.2], and 10% of sand with cultch and mud with cultch areas would support mean densities of 3 market oysters per [m.sup.2].

We attempted to decompose Maryland-wide landings data to match the three salinity zones, but values of F estimated from landings and relative biomass by zone were extremely variable and in some cases unreasonable. Although landings were reported for discrete areas (fishing regions defined by the National Marine Fisheries Service) that could be matched with monitoring sites, landings reports reflected where the oysters were landed, not necessarily where they were caught. Moreover, some landings were reported from areas that could not be assigned to salinity zones based on monitoring data. Therefore, we used the bay-wide mean F, along with zone-specific estimates of M, R, and K, to initiate the salinity zone parameterizations.

The equation for simulating the market oyster population is; [S.sub.t] = (K/1 + G * [e.sup.-[R-(M+F)]), where [S.sub.t] is the market oyster stock in year t, G = (K - [S.sub.t-1])/([S.sub.t-1]), [S.sub.t-1]is the stock in the previous year, e is the base of natural logarithms, and the other parameters are as above. Values of any of the parameters can be specified to simulate particular scenarios. Standard deviations of F, M, and R also can be varied if desired. In all simulations reported here, the model was run for 20 y with 1000 iterations per year. Preliminary simulations with 5000 iterations produced results almost identical to those with 1000 iterations. Model output is processed to generate log-mean simulated stock size and landings for each year of the simulation. Simulated means of stock and landings are back-transformed to arithmetic values and graphed as time-series plots to portray expected trends. We established three reference points for evaluating bay-wide scenarios: (1) stock collapse, defined as market stock <0.1 x [10.sup.6] bushels, (2) stock restoration, defined as a stock of at least 1.77 x [10.sup.6] bushels (i.e., 10 times the stock estimate for the oyster restoration goal baseline year 1994), and (3) fishery restoration, defined as annual landings [greater than or equal to] 2 x [10.sup.6] bushels, consistent with the magnitude of landings sustained from the 1920s through the 1970s. The model calculates percentages of simulations within each scenario that indicate stock collapse or that achieve or exceed the fishery or stock restoration targets in the final year of simulation (2020), as measures of uncertainty for the bay-wide model. Lack of resolution in the landings data precluded fisheries targets for the salinity zone models, and starting stocks were so low in the medium- and high-salinity zones that stock restoration and stock collapse reference points would have had little meaning. We set a stock restoration reference for the low-salinity zone based on the CBP (2000) oyster restoration goal (i.e., 10 times the 1994 stock).

In the simulations reported hem, we varied F and R to forecast the trends that would be expected under reasonable scenarios of management intervention (reductions in F and stock enhancement with hatchery-produced seed oysters). To simulate increased recruitment by means of stock enhancement, we made the following assumptions: hatchery production would not exceed about 2 to 3 x [10.sup.8] spat on shell per year; 30% survival of hatchery oysters from planting to market size; recruitment to the market stock 4 y after planting (Jordan et al. 2002); and 375 market oysters per bushel. These assumptions limited reasonable scenarios to a 10% increase in R. We also ran the models with baseline parameter values in both forecasting and hindcasting modes.

RESULTS

Maryland oyster landings varied by a factor of almost 20 from 1986 to 2001 (Table 1). Variations in F corresponded to annual exploitation rates of 21-73% of the market stock. Natural mortality ranged from 10-60%, total mortality 43-81%, and gross recruitment 40-72%. Annual net rates of change in the market stock varied from -66% to +48%, with a majority (8 of 15 y) of negative values. The mean annual rate of change in the market stock from 1987 to 2001 was -10%. Annual landings and natural mortality (M) for the bay-wide model, along with computed values of K, F. R, and S, are listed in Table 1.

Because the monitoring data we used to establish salinity zones did not support full coverage of oyster habitat or reported landings, our estimates of bay wide K and initial stock were greater than the sums of K and initial stocks for the three salinity zones. Mean baseline parameters, initial stocks, and K for the three salinity zones are listed in Table 2.

Means of 1000 runs of a bay-wide hindcast simulation using the baseline parameters reproduced the observed trends in stock and landings from 1986 to 2001, but not year-to-year variation (Fig. 2). Many individual iterations from this simulation generated patterns similar to the actual time series, but many others diverged widely. The two examples shown in Figure 3 illustrate the stochastic properties of the model. The single iteration of the model shown in the upper pane of Figure 3 generally matched the pattern and magnitude of reported landings, whereas the iteration shown in the lower pane of Figure 3 was an extreme deviation from the data.

[FIGURES 2-3 OMITTED]

Forward simulation with baseline parameter values in the bay-wide model indicated continuing declines in stock and landings through 2020. A reduction of fishing mortality to 0.9F (90% of baseline, or F = 0.64) moderated, but did not reverse these trends. Forward simulations produced a consistent pattern of long-term increases in stock and landings at 0.8F and lower (Fig. 4). Because of overlapping trends among the scenarios, projected landings can be seen more clearly in three dimensions (lower plot of Fig. 4). At 0.8F (F = 0.57), the model projected approximately linear increases in stock and landings, with modest reductions in landings below the 1.0F and 0.9F scenarios for the first few years. Initial reductions in landings were followed by increases in subsequent years. At 0.7F (F = 0.50), there were exponential increases in stock and landings, with the stock increasing by a factor of 5.5 and landings by 5-fold over 20 y. Early losses to the fishery were compensated within the first few years. Reducing F to 0.6F (F = 0.43) increased stock by more than 10-fold and landings by 9-fold. Greater reductions in F (50% or more) increased stock size more rapidly, but long-term landings were less than for the 0.6F scenario. Simulated long-term fishery yields were optimum at 0.6F (F = 0.43; Fig. 4).

[FIGURE 4 OMITTED]

Forty-three percent of bay-wide simulations with 1.0F predicted stock collapse to <0.1 x [10.sup.6] bushels (Fig. 5). Reducing F to 0.6F or less virtually eliminated the probability of stock collapse. Stock restoration ([greater than or equal to] 1.77 x [10.sup.6] bushels) was achieved in 9.2% of simulations at 1.0F and 98% of simulations at 0.5F. The must stringent target, fishery restoration ([greater than or equal to] 2 x [10.sup.6] bushels projected landings), was achieved in 73% of simulations for 0.4F and 0.5F (right panel of Fig. 5); greater or smaller reductions in F indicated less likelihood of achieving the fishery restoration target.

[FIGURE 5 OMITTED]

Increasing the instantaneous rate of recruitment by 10% (1.1R, the maximum we thought to be achievable based on realistic levels of hatchery production), with F held at 1.0F, stabilized stock and landings after an initial lag required for spat to reach market size (Fig. 6). A sanctuary scenario (i.e., coupling 1.1R with 0.9F) predicted approximately 3-fold increases in stock and landings within 20 y (Fig. 7). This scenario indicated a low likelihood (percentage of iterations) of stock collapse and moderate likelihoods of stock and fishery restoration (Fig. 8).

[FIGURES 6-8 OMITTED]

Hindcasts for all three salinity zones indicated exponential stock declines during the period 1986-2000 (Figs. 9-11). Simulated landings for the period were roughly consistent with reported landings for the low- and high-salinity zones, but less so for the medium salinity zone. Forward simulations for the low-salinity zone indicated that F would have to be reduced to 0.6F to achieve increasing stocks of market oysters (Fig. 12). Long-term landings in this zone were stable at 0.8F and optimal at 0.6F (lower plot of Fig. 12). In the medium-salinity zone, stock increased at 0.7F and approached carrying capacity at 0.3F or less (Fig. 13). Long-term landings in the medium-salinity zone were stable at 0.9F and optimal at 0.7F (lower plot of Fig. 13). Moderate reductions in F had little or no effect on stock or landings in the high-salinity zone (Fig. 14). The model predicted substantially increasing stocks at 0.2F and 0.1F in this zone. Simulated long-term landings in the high salinity zone attained maximum values at 0.1F, reaching 0.12 x [10.sup.6] bushels by 2020, about 80% of 1986 landings from this zone. Simulations of a sanctuary strategy (0.9F, 1.1R) slowed, but did not reverse, stock declines in the low- and high-salinity zones. In the medium-salinity zone, the sanctuary simulation predicted an increasing stock after an initial lag (Fig. 15).

[FIGURES 9-15 OMITTED]

The stock restoration target lot the low-salinity zone was an increase from 0.16 x [10.sup.6] to 1.6 x [10.sup.6] bushels by 2020. A high likelihood of achieving this target required 0.4F or less (Fig. 16).

[FIGURE 16 OMITTED]

DISCUSSION

We have quantified the population dynamics of the Maryland harvestable oyster stock and simulated the long term consequences of several management scenarios for the stock as a whole, and also for three subpopulations segregated by salinity. The results indicate that there is potential for restoring the stock and the fishery to sustainable historical levels with reasonable management interventions. The following discussion elaborates on the uncertainties in the data and assumptions that were used in the modeling process, examines management implications both prospectively and retrospectively, and offers recommendations for restoring sustainable oyster stocks in Maryland.

Uncertainties

There are at least three types of uncertainty associated with each of the model parameters: natural variation, measurement error, and inaccuracy (i.e., how closely sampling and estimation measure the true parameters). Natural variation is included in the model structure by allowing F, M, and R to vary according to observed temporal and spatial variation, and is applied to model results as shown in Figures 3 and 9. This variability permits a wide range of possible outcomes, for which we have assumed that log-normal mean trends represent the most likely. Measurement errors for the survey data and indices of relative abundance used to parameterize the model have been shown to be small relative to temporal and spatial variations in the oyster population (Jordan et al. 2002, Jordan 1995), and will not be considered further here.

Parameter values are accurate only to the extent that the assumptions made in estimating them are valid. The use of box counts to estimate annual natural mortality has been a subject of investigation (Christmas et al. 1997, Powell et al. 2001. Powell et al. 2002) and was discussed in the context of the current study by Jordan et al. (2002). We conclude that because annual natural mortality estimated from box counts was strongly correlated with trends in the live oyster population (market oyster biomass; r = -0.60, P = 0.01) and landings (r = -0.58, P = 0.02), the method is sufficiently accurate.

The estimates of fishing mortality are based on an assumption tested only by the comparison of hindcast results to landings data (Fig. 2) and by the generally reasonable nature of simulations. Fishing mortality rates reported from prior stock assessments bracket our mean estimate (F = 0.71) for 1986-2001. Cabraal (1978) reported F = 0.27 from analysis of catch and effort data for 1975-1976, with a market oyster stock of 10.4 x [10.sup.6] bushels and landings of 2.5 x [10.sup.6] bushels. Rothschild et al. (1994) calculated F = 1.3 for the 1991 Maryland oyster harvest; we used Smith and Jordan's (1993) finite exploitation rate of 0.53 (F = 0.64) for the same year to scale F from relative to absolute values. Jordan et al. (2002) showed that the Rothschild et al. (1994) estimate was probably too high because of errors in specifying natural mortality and growth. Because this model is sensitive to F, a focus of this study, the estimates need better verification. The methods used to harvest oysters in Maryland are inherently inefficient (except for diving, a minor component of the fishery), thus very high values of F probably would require more fishing effort than would be returned by the value of the catch. High rates of effort might occur locally, or for brief episodes (viz. F = 1.29 for 1989), but should not be sustained.

The two mortality terms in the model, M and F, are strictly additive and do not interact. A model by Klinck et al. (2001) of the New Jersey oyster fishery in Delaware Bay predicted higher yields by adjusting fishing seasons to compensate for high talcs of disease-related natural mortality. In effect, natural mortality could be decreased by concentrating the fishery in late spring, earlier in the year than most of the disease-related mortality. Moreover, much of the recruitment to market stocks in mid-Atlantic estuaries probably occurs in spring (April-May), when oyster growth rates tend to be high. For these reasons, stock size should peak in late spring. Klinck et al. (2001) used a daily time step and separate functional forms for M and F in their model. Unlike the New Jersey fishery, where a late spring-early summer harvest is traditional, the Maryland fishery has operated only in the fall and winter (October through March, with minor exceptions) for many decades. Although it is likely that higher yields could be achieved in Maryland at a given F with seasonal management, tactical fisheries management has not been the objective of this study. We have focused on long-term population restoration, for which an annual time step and additive mortality terms appear to suffice. Total annual mortality rate (Z = M + F) would not be affected by the compensatory mortality demonstrated by Klinck et al. (2001). From a long-term population or restoration perspective, it matters not how an oyster dies, only that it does.

We did not measure or validate directly the recruitment term (R) in the model; it was derived from year-to-year differences in the market stock, plus total mortality. Net recruitment [dS/dt = r = R - (M + F)], the rate of change in stock size, was consistent with observed long-term trends in relative stock biomass and reported landings. Gross recruitment rates were higher at higher salinity (Table 1), as expected because of higher spat settlement densities at higher salinity (Jordan 1995).

The estimates of carrying capacity (K), although based on the best available data, are uncertain and not directly verifiable. Fortunately, model results and conclusions are not especially sensitive to K, but low-mortality simulations for which the stock approaches K should be interpreted more cautiously than others. Our only verification of this parameter is that K for the bay-wide model (~17 x [10.sup.6] bushels) is reasonable with respect to historical landings, which peaked at 10 to 15 x [10.sup.6] bushels in the 1880s. Although K is a constant in this model, it certainly varies in the real world. Long-term declines in the quantity and quality of oyster habitat (factors of K) in Maryland have been reported (Rothschild et al. 1994, Smith et al. 2001).

Besides uncertainties in estimating model parameters, the governing logistic equation may not accurately represent oyster population dynamics. A particular concern is the stock-recruitment (S/R) relationship. In this model, the rates of gross recruitment (R) and net recruitment (r) are not correlated with stock size, but the average absolute number of recruits increases linearly as the stock increases. The logistic foundation we use artificially limits stock size as a function of K without directly affecting the randomly input recruitment and mortality rates, in which density-dependence would be expressed in nature. Observations in Maryland have shown that oysters can produce high densities of spat at very low population levels; for example, two of the highest spat density indices on record were in 1991 and 1997, when adult population densities were very low (Maryland DNR 2001). This fact suggests that the S/R curve could be steep at low stock sizes, and if so, the model is conservative with respect to recruitment in these cases. Conversely, two major sources of natural mortality, diseases and predation, could be more intense at higher oyster densities, suggesting that the model could overestimate recruitment at moderate stock size; that is, the S/R curve could be flatter at moderate stock size than predicted by the logistic model (at large stock sizes, approaching K, errors in R would make little difference). Estimated stock size varied from 0.18 x [10.sup.6] to 3.21 x [10.sup.6] bushels (1-19% of K) during the time series of data used to parameterize the model, so we have observed only a minor portion of the S/R relationship, if indeed there is one. During this period, gross annual recruitment (R) ranged from 40% to 80% of the stock, without any obvious pattern. We quote Cushing (1968, p. 124): "... there is very often no relation between parent stock and subsequent recruitment at those levels of stock which support fisheries."

Implications and Recommendations for Management

The model indicates that long-term average recruitment rates should be ample to replace or rebuild Maryland oyster stocks if mortality rates could be reduced sufficiently. Natural mortality cannot be managed to a significant degree so long as parasitic diseases kill large proportions of oysters. If the population is developing disease resistance through natural selection, the process has been too slow to observe. Selective breeding and introduction of disease-tolerant oyster stratus for restoration purposes have not been implemented at a scale that could make more than a local difference. Attempts to improve management by restricting movements of infected seed oysters have had no measurable effects at the population scale.

Fishing mortality, in principle, is entirely controllable. The Maryland oyster fishery is regulated by gear restrictions, seasonal and area closures, a minimum size limit, daily catch limits, license surcharges, and limited entry. These restrictions, most of which pre-date the major impacts of Dermo and MSX diseases, apparently have not been sufficient to maintain sustainable rates of fishing mortality while natural mortality rates are high. Annual indices of relative and absolute stock size (Jordan et al. 2002) are available that could be used to limit fishing mortality by setting a total allowable catch (TAC), based on a conservative F, for each year. The fishery could then be closed whenever in the season the TAC was attained, or effort could be otherwise limited so as not to exceed the TAC. The Maryland bay-wide model indicates that reducing average F by 40% (F = 0.43) to 50% (F = 0.35) would virtually assure stock restoration and an enhanced fishery within a decade.

As an example, the estimated initial stock for the 2000-2001 oyster season was 702,000 bushels (Table 1), which produced landings of 348,00l) bushels (F = 0.70, near the long-term mean). At 0.4F (F = 0.43), landings would have been about 245,000 bushels, and at 0.5F (F = 0.35), about 207,000 bushels. These losses to the fishery would be economically significant ($2-4 million in dockside value) in the short-term, but more than compensated in the long-term. With F = 0.43, the model predicts average landings of 656,000 bushels for 2001-2020 and average landings >1 x [10.sup.6] bushels in the second decade of the simulation.

Model scenarios indicated little likelihood of rebuilding the stock or increasing landings by managing fishing mortality in the high-salinity zone. Maryland uses higher salinity areas, where spat settlement is relatively high, for seed oyster production. Large plots are shelled each summer to stimulate spat settlement. The following spring, the spat and shell are transferred to lower salinity areas where disease-related mortality is less. In effect, this strategy enhances recruitment in low-salinity areas, where natural recruitment is undependable. The bay-wide and low-salinity versions of the model implicitly include this source of recruitment, because some of the low-salinity monitoring sites were supplied with transferred seed oysters in most of the baseline years of this study. Because it is unlikely that a significant harvestable stock can be restored in the high-salinity zone, seed oyster production for supplementation of low-salinity areas is probably the best management strategy for this zone. There are benefits In stimulating recruitment through hatchery spat production, but cost and logistical constraints should be expected to limit this contribution to roughly the 10% increase in recruitment we have simulated (Figs. 6, 7, and 15).

Model predictions are most promising for the medium-salinity zone. This large area benefits from moderate to high spat settlement in some years and also from episodes of low salinity that can diminish disease-related mortality for one or more years. The model indicates that a sanctuary strategy alone could have modest success in this zone, in contrast to the low- and high-salinity zones. Management for this zone could use sanctuaries, closed to harvest and stocked with seed oysters, in combination with explicit reductions in fishing mortality, to rebuild productive stocks.

A productive oyster fishery was sustained in Maryland until repeated disease epizootics drastically increased natural mortality rates beginning in the 1980s. Fishing mortality rates, meanwhile, remained constant or increased. Recruitment, whether stable or declining, could no longer keep pace with total mortality rates and the stock declined. This dynamic has continued into the 2000s. Strategies intended to increase recruitment (habitat improvement, natural and hatchery seed oyster production) have been used extensively, but they are expensive, have not stemmed the decline, and offer only marginal hope for the future.

Preliminary landings for the 2002-2003 Maryland oyster season are 51,145 bushels, about 2% of 1975 landings and 3% of 1986 bindings. At a dockside value of no more than $1.5 million, the fishery is close to economic extinction. We recommend that it is past time to reduce and control fishing mortality specifically to restore the oyster stock. With F set at 0.43 [+ or -] 0.25 (standard deviation) in 1986 and maintained at that level, mean simulated 2002-2003 landings would have been 1.96 x [10.sup.6] bushels, 38 times the reported quantity. Ecological contributions (Jordan 1987, Newell 1988, CBP 2002) would have multiplied by roughly the same factor. Taking this action in 1986 would have caused a loss of nearly $9 million in dockside value to the fishery for that year, but the loss would have been repaid many times in sustainable harvests and ecological services.
TABLE 1.
Baseline time-series data for the Maryland bay-wide model.

 Landings Stock
Year (1000 bushels) (1000 bushels) F M R

1986 1560 3211 0.665 0.106 N.A.
1987 980 1428 1.159 0.501 0.850
1988 360 909 0.504 0.589 0.641
1989 400 551 1.294 0.375 1.169
1990 410 768 0.763 0.170 1.265
1991 420 816 0.723 0.217 1.001
1992 320 707 0.603 0.470 0.929
1993 120 238 0.701 0.908 0.521
1994 80 177 0.598 0.786 1.087
1995 166 321 0.730 0.276 1.602
1996 201 452 0.590 0.350 1.281
1997 177 862 0.231 0.335 1.212
1998 285 670 0.554 0.217 0.519
1999 422 678 0.976 0.202 1.189
2000 380 802 0.642 0.367 1.177
2001 348 702 0.684 0.444 0.995
Mean 414 830 0.714 0.394 1.029
SD 368 701 0.253 0.221 0.300

Year is the year in which October-March oyster landings were reported.
F, M, and R are instantaneous rates of fishing mortality, natural
mortality, and gross recruitment to the market stock, respectively.
Landings data were rounded to the nearest 1000 bushels. R was not
computable for 1986: K = 16,964,000 bushels of market-size oysters.

TABLE 2.
Model baseline parameters and initial stock estimates
for three salinity zones.

 Zone Low Medium High

K 4221.8 3006.4 5034.8
Initial stock 231.9 488.9 39.4
F 0.714 (0.253) 0.714 (0.253) 0.714 (0.253)
M 0.165 (0.081) 0.428 (0.287) 0.719 (0.519)
R 0.739 (0.589) 1.1885 (1.046) 1.283 (2.105)

The units of K and initial stocks are bushels x [10.sup.3]. Means of
F (simulated) and M are from 1986 to 2000 time series; the mean of R
in from 1987 to 2000. Standard deviations are in parentheses.


ACKNOWLEDGMENTS

We greatly appreciate assistance, advice, and support front staff of the Maryland Fisheries Service and Sarbanes Cooperative Oxford Laboratory, especially Mark Homer, Kelly Greenhawk, Gary Smith, Mitchell Tarnowski, and Jim Uphoff. This model was made possible by the long-term data collection efforts of many people from the Maryland Department of Natural Resources. William Fisher provided a thoughtful preliminary review of the manuscript. Financial support was provided by the Maryland Department of Natural Resources, the National Oceanic and Atmospheric Administration Chesapeake Bay Office, and the U.S. Environmental Protection Agency Chesapeake Bay Program. The information in this document has been funded in part by the U.S. Environmental Protection Agency. It has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents reflect the views of the agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use. This is contribution number 1185 of the Gulf Ecology Division.

LITERATURE CITED

Cabraal, R.A. 1978. Systems analysis of the Maryland oyster fishery: production, management and economics, Ph.D. Thesis, University of Maryland, College Park, MD.

Chesapeake Bay Program. 2002. Comprehensive Oyster Management Plan. Draft November 2002. Available: http://www.chesapeakebay.net/ pubs/subcommittee/lrsc/doc-COMPnovdraft2.PDF.

Christmas, J. F., M. R. McGinty, D. A. Randle, et al. 1997. Oyster shell disarticulation in three Chesapeake Bay tributaries. J. Shellfish Res. 16:115-123.

Cushing, D. H. 1968. Fisheries biology: a study in population dynamics. Madison: The University of Wisconsin Press.200 pp.

Gieseker, C. 2001. Year 2000 Maryland oyster disease stains report. Maryland Department of Natural Resources Technical Report FS-SCOL-01-1. Oxford.

Homer, M. L., R. Scott & G. E. Krantz. 1996. Maryland oyster population status report: 1995 fall survey. Maryland Department of Natural Resources Report MDDNRSP1-96. Oxford.

Jordan, S. J. 1987. Sedimentation and remineralization associated with biodeposition by the American oyster Crassostrea virginica (Gmelin). Ph.D. Thesis, University of Maryland, College Park, MD.

Jordan, S. J. 1995. Population and disease dynamics of Maryland oyster bars: a multivariate classification analysis. J. Shellfish Res. 14(2):459-468.

Jordan, S. J., K.N. Greenhawk. C.B. McCollough, et al. 2002. Oyster biomass, abundance and harvest in northern Chesapeake Bay: trends and forecasts. J. Shellfish Res. 21:733-741.

Klinck, J. M., E. N. Powell, J. N. Kraeuter, et al. 2001. A fisheries model for managing the oyster fishery during times of disease. J. Shellfish Res. 20:977-989.

Krebs. C. J. 1994. Ecology: the experimental analysis of distribution and abundance. Fourth Edition. Addison-Wesley Longman, Inc., Menlo Park, CA. 801 pp.

Maryland DNR. 2001. Maryland Oyster Population Status Report, 19962000 Fall Surveys. Annapolis: Maryland Department of Natural Resources.

Maryland Oyster Roundtable. 1993. Maryland oyster roundtable action plan. Annapolis: Maryland Department of Natural Resources.

Newell, R. I. E. 1988. Ecological changes in Chesapeake Bay: are they the result of over-harvesting the American oyster, Crassostrea virginica?

In: M, Lynch & E. C. Krome, editors. Understanding the estuary: advances in Chesapeake Bay research. Solomons, MD: Chesapeake Research Consortium. pp. 536-546.

Powell, E. N., K. A. Ashton-Alcox, S. E. Banta, et al. 2001. Impact of repeated dredging on a Delaware oyster reef. J. Shellfish Res. 20:977-989.

Powell. E. N., K. A. Ashton-Alcox, J. A. Dobarro, et al. 2002. The inherent efficiency of oyster dredges in survey mode. J. Shellfish Res. 21:691695.

Rothschild, B. J., J. S. Ault, P. Goulletquer, et al. 1994. Decline of the Chesapeake Bay oyster population: a century of habitat destruction and overfishing. Mar. Ecol. Prog. Set. 111:29-39.

Smith, G. F. and S.J. Jordan. 1993. Monitoring Maryland's oysters: a comprehensive analysis of modified fall survey data 1990-1991. Maryland Department of Natural Resources Report CBRM-OX-93-03. Oxford.

Smith, G. F., K. N. Greenhawk, D. G. Bruce, et al. 2001. A digital representation of the Maryland oyster habitat and associated bottom types in the Chesapeake Bay. J. Shellfish Res. 20(1):197-206.

STEPHEN J. JORDAN (1) AND JESSICA M. COAKLEY (2) U.S. Environmental Protection Agency, Gulf Ecology Division, Gulf Breeze, Florida 32526; and University of Maryland, Center for Environmental Science, Sarbanes Cooperative Oxford Laboratory, Oxford, Maryland 21654
COPYRIGHT 2004 National Shellfisheries Association, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Coakley, Jessica M.
Publication:Journal of Shellfish Research
Date:Apr 1, 2004
Words:6555
Previous Article:Ecosystem influences of natural and cultivated populations of suspension-feeding bivalve molluscs: a review.
Next Article:Age structure and growth in an unusual population of purple clams, Amiantis purpuratus (Lamarck, 1818) (bivalvia; veneridae), from Argentine...
Topics:


Related Articles
Site selection for oyster habitat rehabilitation in the Virginia portion of the Chesapeake bay: a commentary.
A comparison of dredge and patent tongs for estimation of oyster populations.
Decadal scale changes in seasonal patterns of oyster recruitment in the Virginia sub estuaries of the Chesapeake Bay.
Parasitic and symbiotic fauna in oysters (Crassostrea virginica) collected from the Caloosahatchee River and Estuary in Florida.
The effect of peat moss particles on the physiology of the Eastern oyster, Crassostrea virginica.
Condition index of the Eastern Oyster, Crassostrea virginica (Gmelin, 1791) in Sapelo Island Georgia--effects of site, position on bed and pea crab...
Survey of water quality, oyster reproduction and oyster health status in the St. Lucie Estuary.
The case for sequencing the Pacific oyster genome.
Naturalized populations of oysters, Crassostrea gigas along the South African coast: distribution, abundance and population structure.
Apoptosis of the protozoan oyster pathogen Perkinsus marinus in vivo and in vitro in the Chesapeake Bay and the Long Island Sound.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters