Can selective breeding reduce the heavy metals content of Pacific oysters (Crassostrea gigas), and are there trade-offs with growth or survival?

ABSTRACT Oyster producers in the Pacific Northwest region of the United States export large quantities of oysters to international markets. Proposed changes to international limits for heavy metals content in shellfish could drastically curtail exports and impact the viability of this environmentally-friendly industry. "Supply-side" solutions such as moving oyster farms to uncontaminated sites or short-term depuration would incur substantial costs in terms of labor and infrastructure and displace workers in already economically challenged coastal communities, whereas selective breeding could benefit both producers and consumers within the current infrastructure. We studied the feasibility of selective breeding to reduce heavy metal content through a quantitative genetic analysis of heavy metals accumulation in the Pacific oyster, Crassostrea gigas by opportunistically sampling a factorial mating experiment initiated in 2000 to study the effects of parent size on offspring growth and survival. The experiment consisted of all possible crosses among six males (sires) and five females (dams). At harvest, we measured four performance traits (yield, survival, whole oyster live weight and shucked meat dry weight) and the accumulated levels of four heavy metals (copper, lead, zinc and cadmium). Analyses of variance testing for sire and dam effects and bootstrap estimates of heritability showed that all of these traits have a genetic basis. Further, half-and full-sib family means correlations revealed genetic trade-offs between copper and cadmium content and performance traits. Preliminary indications are that selective breeding to reduce heavy metals accumulation is possible, but that genetic trade-offs between metal content and performance should be taken into account in designing any program of selective breeding in this species.

KEY WORDS: Crassostrea gigas, cadmium, copper, lead, zinc, heavy metals contamination, depuration, quantitative genetics, genetic correlation

INTRODUCTION

Oyster aquaculture is an important economical activity in the Pacific Northwest region of the United States where traditional finfish and shellfish capture fisheries are generally overexploited and increasingly regulated. Since the early part of the 20th century, after the collapse of the fishery for the native Olympia oyster, Ostrea concaphila, oyster production in this region has been dominated by the normative Pacific oyster, Crassostrea gigas. Prior to the 1970s, the west coast oyster industry was heavily dependent on shipments of oyster seed on cultch from Japan, but with the development of hatchery technology on the west coast (see Clark & Langmo 1979 for a review) and increased regulation over the movement of live shellfish, the industry has evolved to rely almost entirely on hatchery production. Oysters grown in the Pacific Northwest region are marketed internationally, and foreign markets are particularly profitable for growers (W. Dewey, X. Liu, P. Taylor pers. comm to MDC).

Bivalve mollusks in general, and Pacific oysters in particular, are efficient bioaccumulators of heavy metals such as lead, copper, zinc and cadmium (Shulkin et al. 2003), potentially at concentrations toxic to humans who consume them in large quantities. Concerns over heavy metals, especially cadmium, in shellfish have recently lead to the closure of oyster farms in heavily contaminated areas of France (Geffard et al. 2002). The Codex Alimentarus Commission, created in 1963 by the Food and Agriculture Organization and the World Health Organization (http://www.codexalimentarius.net/) is currently considering new international standards for cadmium content in shellfish and other agricultural products of 1 [micro]/g wet tissue weight (ftp://ftp.fao.org/codex/ccfac33/fa01_28e.pdf). Typically, once an international limit is set by Codex, member countries, including the United States, adopt the same standards for not only international, but also domestic interstate sales, and any product exceeding the standards is considered adulterated. Because oysters grown in the Pacific Northwest region of the United States sometimes contain levels of cadmium above the proposed limits (Christy et al. 2004 see also http://www.pacshell.org/Cadmium.htm), these new regulations could threaten export markets for United States producers. Also, the USFDA established 3.7 [micro]/g of cadmium in shellfish as a "level of concern" in 1993, and domestic consumers are increasingly uneasy about the risks associated with heavy metal contamination.

Several strategies to limit the heavy metal content of oysters reaching these markets are possible. Perhaps the simplest approach from the biologic perspective would be to control the availability of metals in the growing environment. Oysters clearly acquire the heavy metals that they accumulate from the environment, and "supply-side" solutions aimed at controlling the environmental availability of heavy metals could be effective if they can be implemented. Reinfelder et al. (1997), for example, found that American oysters (Crassostrea virginica) assimilated approximately 70% of the cadmium, selenium and zinc contained in their food. Ke and Wang (2001) studied the bioaccumulation of the same three metals in Crassostra rivularis and Saccostrea glomerata, and found that most of the retained metals came from ingested particles rather than from dissolved sources and that the assimilation efficiencies for all three metals ranged from about 25% to 80%. These data clearly argue that this approach warrants serious investigation because of the strong influence of heavy metal availability (especially in food organisms) on accumulation. Other researchers are collecting baseline environmental data to describe spatial variation in the environmental availability of cadmium and the levels accumulated by oysters in the Pacific northwest (Christy et al. 2004 see also http://www.pacshell.org/ Cadmium.htm). This parallel effort should reveal the magnitude of variation in heavy metal availability and its spatial patterning. However, managing the availability of cadmium and other metals in the environment is likely to require either large-scale changes in watershed management and land use that contribute to heavy metal availability or the concentration of oyster farms in areas with low availability accompanied by restrictions on the future activities allowed in these areas. Because these are matters policy rather than biology, we will not address them further here, except to point out that implementing this approach would clearly require substantial changes to either where oysters are farmed or what other activities are allowable in nearby areas.

Depurating or otherwise decontaminating metal-containing shellfish before they reach markets could also provide safer oysters to markets. The use of depuration strategies, however, is difficult to assess with the available information. Greig and Wenzloff (1978) and Zaroogian (1979) found that in C. virginica, depuration of experimentally contaminated oysters for 48 and 56 wk respectively had no effect on cadmium concentrations (though Zaroogian did observe decreases in the total quantity of cadmium accompanied by a decrease in tissue weight). Geffard et al. (2002) studied the kinetics of heavy metal elimination from Pacific oysters (C. gigas) grown in a contaminated estuary in France and found that this species retains accumulated heavy metals for long periods after being transferred to cleaner areas. In Geffard's study, cadmium was eliminated more quickly than copper or zinc, but still, 4 mo of deputation was necessary to reduce cadmium to safe levels. In contrast, van Dolah et al. (1987), and Okazaki and Panietz (1981) found that cadmium, copper and zinc could be reduced quickly through depuration. Wallner-Kersanach et al. (2000) showed that the efficacy of depuration depends on the details of their exposure, which could explain the discrepancies between these studies. Oysters exposed briefly (60 days) to high levels of copper were able to eliminate 30% of their copper content in 30 days, whereas resident oysters from the same area eliminated only 9% of their copper content in the same period. In addition, van Dolah et al. (1987), and Amiard-Triquet et al. (1991) noted that the kinetics of metal elimination are influenced by environmental factors such as salinity, temperature and humic acid concentration. Geffard et al. (2002) argue that these variable results can be explained by the facts that (a) environmental factors alter the distribution of accumulated metals among various tissues and the degree to which they are bound to cytosolic ligands versus in solution and (b) the soluble fraction is more easily eliminated through depuration. If this view is correct, then growing oyster at sites contaminated with heavy metals for long periods could make subsequent depuration impractical. However, even if depuration is possible, implementing this approach would minimally require identifying appropriate sites and regularly moving vast quantities of materials to these sites, at considerable expense.

In this study, we address the feasibility of selective breeding to reduce the heavy metal content of cultured Pacific oysters with the idea that if oyster strains that naturally accumulate less heavy metals were available, growers would need only to change their source of broodstock whereas other approaches require substantial changes to policy and/or infrastructure. Due the urgency of this issue for the cultured oyster industry and the time that is required to initiate a new experiment, we opportunistically sampled an existing quantitative genetic field experiment to assay the heavy metal content of oysters with known genetic relationships. Because this experiment was initially designed to examine the relationship between parental size and the growth and survival of their progeny (Evans 2004), the male parental animals consisted of large and small animals, sampled nonrandomly, and this could potentially bias the results. Nonetheless, this initial examination of the degree to which heavy metals accumulation is under genetic control provides important and timely insights regarding the prospects for manipulating heavy metal accumulation through selective breeding and has important implications for ongoing selective breeding efforts and commercial producers.

MATERIALS AND METHODS

Experimental Design

The parents came from a previous experiment (Brooks 2000) in which individually tagged adult oysters from unselected pedigreed oyster families were deployed in lantern nets at two locations within Yaquina Bay, Oregon (44.6[degrees]N, 124.1[degrees]W) during Summer 1998. The first site was located approximately 3 km from the mouth of Yaquina Bay ("downriver"), the second site was located approximately 15 km up the Yaquina River at a commercial shellfish lease-site ("upriver"). As a continuation of research initiated by Brooks (2000), body weights of all tagged oysters were periodically measured from plant-out in Spring 1998 until harvest in Summer 2001. For this experiment, which was originally designed to test the hypothesis that the size of parents affects the performance of their progeny (Evans 2004), a set of 30 families was created by crossing the three largest males and the three smallest males from the downriver population with five females randomly selected females from the downriver population (i.e., 6 male x 5 female factorial cross). Crosses among full-siblings were avoided. Table 1 shows the details of this mating design.

Hatchery and Nursery Protocol

During Summer, 2001, these parents were held in 18[degrees]C sand-filtered seawater and fed a mixture of Isocrysis galbana (Iso) and Cheatoceros calcitrans (Cc) at a concentration of approximately 50,000-80,000 cells [mL.sup.-1] until ready to spawn. Animals were stripped-spawned as per Langdon et al. (2003) in Fall 2001. Fertilized eggs were allowed to develop into veliger larvae (D-larvae) for 24 h in cross-specific 20-L containers filled with 25[degrees]C, 0.2-[micro]m filtered seawater. D-larvae from each cross were then stocked into family-specific 100-L larval culture containers at a concentration of 10 larvae [mL.sup.-1]. Larvae were fed daily with a mixture of Iso and Cc at concentrations ranging from 30,000-80,000 cells [mL.sup.-1], depending on age (Breese & Malouf 1975). Larval tanks were drained through sieves and refilled with 25[degrees]C 0.2 [micro]m filtered seawater twice per week. For the first week, larvae were retained on 37 [micro]m sieves. During the second week, larvae were retained on 80 [micro]m sieves and densities reduced to 1 larva [mL.sup.-1]. During the third week, larval cultures were drained through 243 [micro]m and 80 [micro]m sieves. All larvae retained on 243 [micro]m sieves were exposed to 2 x [10.sup.-4] M epinephrine to induce metamorphosis (Coon et al. 1986).

Successfully metamorphosed spat were transferred to family-specific 15-cm diameter upwellers. Upwellers were held in a semi recirculating system that received approximately 6 exchanges [d.sup.-1] of 25[degrees]C UV-irradiated 1 [micro]m filtered seawater. Once all larvae had metamorphosed (approximately 4 wk post harvest), the number of spat per upweller was randomly thinned to 10,000. Spat were allowed to grow until retained on a 1.4 mm sieve, then transferred to family-specific 28-cm diameter upwellers in a larger upwelling system. These larger upwellers were supplied with 18[degrees]C 1-[micro]m filtered seawater delivered at approximately 2.8 1 [min.sup.-1] and fed an Iso/Cc mixture at a final concentration of approximately 50,000-80,000 cells [mL.sup.-1]. Once all animals were transferred from the 15-cm upwellers, the number of oysters per 28-cm upweller was randomly thinned to 5000. Oysters were allowed to grow until retained on a 6.4 mm sieve, before being transferred to family-specific spat bags (2-mm mesh) held in storage tanks. Storage

tanks received ambient 1-[micro]m filtered seawater (mean 9.5[degrees]C; range 7.0[degrees]C to 13.68[degrees]C) and batch-fed to a final concentration of approximately 100,000 cell [mL-.sup.1] of a Cc/Iso mixture for 8 h, twice per week. The reduced temperature and limited feeding was intended to slow oyster growth, minimizing variation in spat weight within and between cultures prior to planting in the field (Langdon et al. 2003). After approximately 80% of the oysters were sieved from the 28-cm upwellers, spat in the storage tank were counted and weighed for subsequent planting in the field. Because of variable performance in the nursery, some crosses did not produce enough spat to plant-out and evaluate in the field and only 26 of 30 crosses were deployed in the field (Table 1).

Field Deployment

In Spring 2002, 50 randomly selected seed oysters from each family were weighed and stocked into two replicate compartments in each of five vertical blocks in 10-tier lantern nets (0.3-m diameter, 2 mm mesh) for a total of 10 replicates per family. Blocking was conducted to account for potential variation in performance traits caused by water depth. Variable survival in the nursery resulted in some families having fewer than desired individuals, and therefore < 10 replicates. Stocked lantern nets were suspended from a raft at the upriver field site. In July of 2004 (after 866 days in the field) all live oysters from each replicate were cleaned of biotic and abiotic fouling, counted, and their collective weight measured to the nearest gram. These data allowed for replicated estimates of yield and survival in the field. Survival was measured as the proportion of the 50 seed oysters planted that were alive at harvest. Yield was measured as the total weight of the surviving oysters in each lantern net compartment because all lantern net compartments were initially stocked with exactly 50 seed oysters. After these compartment-level data were collected, all of the surviving oysters were pooled within blocks, returned to the Hatfield Marine Science Center, and held in flow-through tanks, whereas subsequent measurements were taken (2-3 days). For the purposes of this study, we randomly sampled 10 animals from each family from two of the five blocks (the deepest and shallowest), and individually weighed them to the nearest gram. We then carefully opened their shells using knives dipped in weak acid and thoroughly rinsed with running distilled water to prevent cross-contamination, removed the meats, blotted them dry using paper towels, and weighed individual meats to the nearest 0.1 g. We then froze the meats at -80[degrees]C in individually marked plastic bags.

Heavy Metals Assays

In late July and August 2004, we freeze-dried the oyster meats by removing them from the freezer, quickly opening their individual bags, and placing them into the vacuum chamber of a Labconco freeze dryer still in their individual bags. When they were completely dried, we crushed each meat to a fine powder inside its individual bag using a Seward Stomacher 80 blender. We later used small subsamples of the powdered tissues for heavy metals analysis using inductively coupled plasma--optical emission spectrophotometry (ICP-OES). The microwave digestion of oyster tissue in preparation for ICP-OES analysis began by combining 500 mg freeze-dried oyster tissue and 15 mL 6 M nitric acid in a 100 mL Teflon vessel. The vessel was then loaded into an Ethos Labstation microwave (Milestone, Inc., Shelton, Connecticut) and digested using Milestone application method #04-012. After digestion, the vessel was removed from the microwave and brought up to a final volume of 45 mL with Milli-Pore deionized water. Quantification of copper (Cu), Zinc (Zn), Lead (Pb) and Cadmium (Cd) in the digested oyster tissue was preformed using a Perkin Elmer 4300 ICP-OES. The detection limits for Cd, Cu, Pb and Zn in oyster tissue were 0.05, 1.5, 0.001 and 1.0 mg [kg.sup.-1] d. wt., respectively. The instrument detection limit for Cd, Cu, Pb and Zn were 0.02, 0.02, 0.001 and 0.001 mg [L.sup.-1], respectively. The tissue sample recovery rate for Cd, Cu, Pb and Zn were all within 100% [+ or -] 10%. The raw data were then expressed as [micro]g of metal [g.sup.-1] dry tissue weight.

Statistical Analyses

We first tested for significant effects of blocks (= depths), sires, dams and sire-by dam interaction on each of the eight characters measured using 2-way analysis of variance (ANOVA) performed using the PROC GLM procedure in the SAS statistical package version 8 (http://v8doc.sas.com/sashtml/). In these analyses, we treated sire and dam effects as fixed factors because the sires were a nonrandom sample from the entire population. Extrapolating from these analyses to the entire population is, therefore tenuous because the parents we used could be a nonrepresentative sample of the larger population from which they came.

We next estimated the broad-sense heritabilities of each of the traits we measured by treating the data as a set of 26 full-sib families, essentially ignoring the factorial structure of the matings. The sharing of parents among full-sib families is most likely to reduce the variation among families and therefore provide conservative estimates of broad-sense heritabilities. We calculated these estimates and their standard errors using the h2boot program written by Patrick Phillips [http://www.uoregon.edu/~pphil/software.html; (Phillips & Arnold 1999)]. This program uses a distribution-free bootstrapping approach to obtain point estimates and standard errors of the heritability of each trait as well as a nonparametric P value. These P values are the proportion of bootstrap samples, which distributions overlap zero. For these analyses, we first removed variance attributable to blocks by performing 1-way ANOVA on the raw data and performing the bootstrap analyses on the block-corrected residuals. All of these analyses used 5,000 bootstrap resamplings of the data.

Finally, we estimated the genetic correlations among the variables as standard product-moment correlations among the family means for all pairs of traits (Falconer & Mackay 1996) For these estimates, we again used the block-adjusted data, and reduced these data to the means of 2 different types of families: (1) the six paternal half-sib families, which provides an estimate of the purely additive or strict sense genetic correlation and (2) the 26 full-sib family means, which, because of the inclusion of the nonadditive influences of dominance and epistasis, estimates the broad-sense genetic correlation.

RESULTS

The eight characters we studied are described in Table 2, and summary statistics are presented in Table 3. Several overall points are worth mentioning here. First, overall, survival was low by industry standards, averaging 27% and indicating that the conditions in the Yaquina Bay during the period of this experiment were relatively stressful for Pacific oysters. Second, most of the traits we measured have quite high levels of overall phenotypic variation, with coefficients of variation ranging from about 0.3-0.75. It should be noted that the levels of heavy metals that we report here are on a dry weight basis whereas the regulatory limits discussed in the Introduction are expressed in terms of wet weight. Overall, dry meat weight averaged 19.8% of wet meat weight. Finally, because this experiment was originally designed for a different purpose, we did not collect any data on cadmium availability in either the hatchery or the field. However, all the animals were reared in the same conditions at the same time, so any variation in metals content is attributable to differential uptake rather than availability.

Table 4 shows the results of the 2-way analyses of variance for each character. Blocks had significant effects only on dry meat weight and cadmium content. There were significant main effects of sires for all response variables except zinc content and significant dam effects on all the traits we measured. In addition, there were significant sire-by-dam interactions for live weight, dry meat weight and all four metals. It should be noted, however, that all of the characters for which we found significant sire-by-dam interactions were measured on individuals and that the error degrees of freedom for the bag-level measurements and the individual-level measurements are very different (66 vs. 468). As a result, the power of the tests for interaction effects is much greater for the measurements taken on individuals.

Bootstrap estimates of broad-sense heritability range from 0.325-1.29 and all the traits we measured have heritabilities that differ significantly from zero (Table 5). In theory, heritabilities, being the proportion of total variance attributable to genetic causes, are constrained to fall between zero and one, but statistical estimates can fall outside of this interval caused by sampling errors and the fact that heritability is calculated by multiplying the observed among-family variance component by the reciprocal of the coefficient of relatedness to estimate the numerator of the proportion (Falconer & Mackay 1996), magnifying estimation errors. All of the heritability estimates greater than the theoretical maximum are for traits measured not on individuals but on growing compartments (i.e., yield and survival). The inability to measure these traits on individuals effectively reduces the data to compartment means, resulting in smaller sample sizes for these traits. Furthermore, the inability to estimate the variance among individuals within compartments for these traits has two other effects. First, it is unclear what the coefficient of relatedness is for compartments that share both parents. Whereas it is clear that the individuals within compartments are full-sibs with a coefficient of relatedness of 0.5, taken as a group, replicate compartments are essentially resamplings of the same full-sib families and thus might best be viewed as genetic clones with a coefficient of relatedness of one. In this case, the heritability estimates for yield and survival presented in Table 5 should be divided by two. Second, reducing the data to compartment means reduces the total phenotypic variance estimate because the sampling variance among individuals cannot be estimated and is necessarily excluded. This is expected to upwardly bias the heritability estimates of traits that cannot be measured on individuals because the total phenotypic variance appears as the denominator in heritability estimates.

Turning to the genetic correlations among traits and first examining correlations estimated using paternal half-sib family means, if (taking into account that these correlations use only six data points) we accept a liberal P value of <0.1 as suggestive of a significant relationship, Table 5 shows positive relationships between yield and survival and between dry meat weight and both survival and live weight. For heavy metal accumulation, there is evidence that copper content is positively correlated with yield, survival and zinc content but no evidence that lead content is correlated with any other traits. Cadmium in contrast may be positively correlated with yield and survival but negatively correlated with meat dry weight.

The correlations among the 26 full-sib family means provide more powerful tests of significance. Using the standard criterion of P < 0.05 for statistical significance and looking first at the correlations among performance traits, our data indicate that yield is positively correlated with survival; survival is positively correlated with dry meat weight; and dry meat weight and live weight are positively correlated. The correlations between the levels of bioaccumulated heavy metals and performance traits vary among the four metals assayed. Copper content is positively correlated with yield and survival but negatively correlated with the dry weight of the shucked meat. Lead content, in contrast, is uncorrelated with any of the performance traits we measured. Zinc is uncorrelated with yield, survival or live weight but negatively correlated with dry meat weight. Cadmium content is positively correlated with yield and survival, uncorrelated with live weight and negatively correlated with dry meat weight. Copper, zinc and cadmium levels are mutually positively correlated but uncorrelated with lead levels.

Comparing the half- and full-sib family means correlations reveals a high level of agreement between these two estimates of the genetic correlation. Full-sib family means correlations produce, as expected, more significant results, but in all cases where the full-sib family means correlation is significant the signs of the full and half-sib correlations agree. In addition, there are no significant half-sib family correlations that are not also significant when estimated as full-sib correlations.

DISCUSSION

Because of the limited number of parents used in this study, the results must be interpreted with caution. Even so, our data support the hypotheses that all of the traits we measured are under considerable genetic control and that populations of C. gigas are likely to harbor substantial genetic variation for all of these traits. The significant sire-by-dam interactions for live weight, dry meat weight and the four metals we measured further indicate that non-additive genetic effects also affect these characters and that the broad-sense heritabilities we estimated using full-sib families probably overestimate the narrow-sense heritabilities.

That we found high heritabilities for performance traits agrees well with previous work in a variety of oyster species. Lannan (1971, 1972) first reported the heritability of growth traits in Pacific oysters, and found that the early larval growth and survival measured at 12 mo postspawning had heterabilities >0.2, but reduced heritabilities at 18 mo. Newkirk et al. (1977) reported that the heritability for larval growth rate in eastern oysters was in the range of 0.25-0.5. In Pacific oysters, Hedgecock et al. (1993) reported a heritability of 0.2 for size at the time of harvest. Degremont et al. (2002) reported high heritability (>1) for survival of adult Pacific oysters in France, and Ernande et a1.(2003) found that whereas most larval survival and size traits were heritable, growth rate was not, nor were juvenile characters. Researchers in Maine have reported a 25% reduction in the time required for C. virginica to reach market size through artificial selection (Barber et al. 1998, Davis & Barber 1999), and an ongoing selection program in Pacific oysters has resulted in substantial single-generation genetic gain of approximately 9.5% for yield compared with unselected controls and a realized heritability estimated variously from 0.01-0.32 depending on the cohort analyzed (Langdon et al. 2003).

This study is, however, to the best of our knowledge, the first to report heritability estimates for the bioaccumulation of heavy metals in shellfish and the first evidence that there are substantial genetic correlations between heavy metal accumulation and performance characters.

The positive full-sib genetic correlations we observed between copper and cadmium content and survival and yield, imply that natural selection or selective breeding for either survival or yield will likely produce a correlated response that raises the tissue concentrations of both of these heavy metals. Single-trait selection to reduce the copper and/or cadmium content of oyster meats is likely to simultaneously reduce survival and yield. This makes intuitive sense if bioaccumulation of heavy metals by incorporation into insoluble complexes is an important mechanism by which bivalves mitigate the toxic effects of heavy metals exposure and genotypes more able to set heavy metals safely aside in less reactive forms better avoid the toxic effects of exposure.

On the other hand, copper, cadmium and zinc all show negative genetic correlations with individual dry meat weight. Thus selection for larger meats could lower the concentrations of these three metals and selection for lower concentrations of copper, cadmium or zinc could result in not only fewer surviving oysters and lower yield as mentioned earlier but also larger meats. These results seem at first to he counterintuitive in that we might expect that animals that devote more metabolic energy to detoxifying heavy metals would do so at the expense of individual growth. One possibility is that fast-growing oysters somehow process particulate matter in the water differently than slow-growing animals such that high-metal particles are eliminated as pseudofeces rather than ingested. It would be interesting, therefore, to compare the distribution of heavy metals among different sized particles in nature and to test whether genotypes of oysters that differ in metal content also differ in their utilization of different-sized particles.

Another, perhaps more far-fetched, possibility is that the differences in metals concentration are related to variance in reproductive effort. C. gigas develop gonads and gametes, but only rarely spawn in the cold-water temperatures of most Pacific Northwest estuaries, including Yaquina Bay. However, the summer of 2004 was unusually hot and dry in the Yaquina Bay area. If some of the families we used in this experiment were more prone to spawning, and if heavy metals are less concentrated in gonadal tissue or gametes, then oysters that had spawned would have smaller total meat weights but similar overall metals content resulting in increased metals concentrations. Unfortunately, we made no effort to record the reproductive status of the oysters used in this experiment and, therefore, cannot address this issue.

Whereas the genetic mechanisms that underlie these patterns are beyond the scope of this study, it is worth noting that molecular-level studies also support the hypothesis that heavy metals bioaccumulation is genetically determined in bivalve mollusks. For example, observations in two species of clams have shown that among-site differences in heavy metal concentrations are reflected in allele frequencies at 2 genetic markers (phosphoclucomutase and glucosephosphate isomerase) and differences in protein level responses (metallothionein) to heavy metals challenges, strongly supporting the hypothesis that long-term exposure to heavy metals contamination can alter allele frequencies in bivalve mollusk populations (Moraga et al. 2002). Further, metallothionein genes, which encode essential metal binding proteins that likely play a major role in metals detoxification (Cherian & Goyer 1978) have been cloned and characterized in Pacific oysters (Tanguy et al. 2001) and immunochemical assays for environmental contamination have been developed based on the idea that the levels of metallothionein in the gills and/or digestive gland of Pacific oysters reflect the environmental availability of heavy metals through the induction of transcription (Boutet et al. 2002). Curiously, however, European flat oysters (Ostrea edulis) showed no such induction of metallothionein genes when exposed to copper and cadmium (Tanguy et al. 2003). As well, the somewhat misnamed heat shock protein gene family, Hsp70, which is involved in various stress responses (Feder & Hofmann 1999), shows altered patterns of expression in gill and digestive gland tissue of O. edulis in response to experimental cadmium exposure (Boutet et al. 2003). These more mechanistic studies of how bivalves respond to heavy metals contamination and the relationship between environmental availability of metals and the expression of genes associated with detoxification also justify optimism that further work could identify molecular markers that would allow direct selection on genotypes through marker-assisted selection. In fact, Tanguy et al. (2002), have showed that polymorphisms in the exons of metallothionein genes are related to heavy metal induced mortality in Pacific oysters, and it could be straightforward to extend this work to heavy metals bioaccumulation.

Regardless of the specific mechanisms that underlie the genetic correlations we observed, however, breeding programs seeking to simultaneously improve growth and survival and reduce heavy metals should estimate and take into account the genetic correlation structure among these characters. Because genetically rigorous selective breeding of oysters is still in its infancy, and since the development of hatchery technology in the 1950s and its widespread commercialization in the 1970s, commercial growers have necessarily been making informal choices of which animals to use as broodstock, and these choices have, quite sensibly, emphasized yield--a combination of growth and survival. In addition, recently initiated formal selective breeding efforts, in response to growers' stated desires, have also emphasized improving yields (Langdon et al. 2003). If our finding that survival and yield are positively genetically correlated with copper and cadmium accumulation can be generalized to these semidomesticated populations of Pacific oysters, then these efforts are likely to also have raised the heavy metals content of farmed oysters as an indirect, correlated response.

Further research is needed to refine these estimates, and be cause genetic correlations are potentially variable among populations (Falconer & Mackay 1996), they must be estimated in the specific population of interest. Nonetheless, preliminary indications are that selective breeding to reduce heavy metals accumulation is possible, but unless the genetic correlations between heavy metals content and performance characters are explicitly taken into account in designing the selection scheme, undesirable correlated effects are likely. The genetic correlations we found between heavy metals content and performance traits are all moderate (r = 0.4-0.7), and if these estimates hold for other populations, properly designed selection strategies for multitrait improvement such as index selection or independent culling (Falconer & Mackay 1996) should be able to produce strains with genetic improvement in performance and a reduction in heavy metals content. However, the rate of improvement is likely to be substantially slower than single-trait selection on any single character.

On a more cautionary note, designing a multitrait selection scheme is complex, and its implementation is a costly, long-term endeavor. For index selection, what is required is not only genetic-level information on the degree to which desirable traits are under additive genetic control, a reliable method for evaluating potential broodstock for all traits and the physical capacity to handle large numbers of genetically distinct families, but also appropriate weightings for the contributions of each trait to the selection index (Falconer & Mackay 1996). In most agricultural situations, these weightings are determined by the relative economical value of incremental improvement in each trait. Larger, more rigorous quantitative genetic analyses can provide better predictions of how much improvement is expected under different selection schemes, but not the information required to establish economical priorities. This must come from producers.

ACKNOWLEDGMENTS

The authors thank Nicholas Sanchez and Machelle Nelson for their diligence in preparing and processing samples for heavy metals analysis and Alan Barton, Chris Emerson, Dave Jacobson, Sean Matson, Drew Mosher, Ebru Onal, David Stick, and Dan Troop for their work in the hatchery and field. We also thank Dr. Chris Langdon for advice on experimental design and oyster husbandry. This research was supported with funding from the USDA Agricultural Research Service (CRIS Project #5358-31000-001-00D), from Anja Robinson Fellowship at Oregon State University's Hatfield Marine Science Center, and from a USDA CSREES Special Grant to the Molluscan Broodstock Program (CRIS Project #ORE00359G).

LITERATURE CITED

Amiard-Triquet, C., B. Berthet & R. Martoja. 1991. Influence of salinity on trace metal (Cu, Zn, Ag) accumulation at the molecular, cellular, and organismal level in the oyster (Crassostrea gigas) Thunberg. Biol. Met. 4:144-150.

Barber, B. J., C. V. Davis & M. A. Crosby. 1998. Cultured oysters, Crassostrea virginica, genetically selected for fast growth in the Damariscotta River, Maine, are resistant to mortality caused by Juvenile Oyster Disease (JOD). J. Shellfish Res. 17:1171-1175.

Boutet, I., A. Tanguy, M. Auffret, N. Mujdzic & D. Moraga. 2003. Expression of HSP 70 in experimentally metal-exposed European fiat oysters Ostrea edulis. J. Shellfish Res. 22:763-766.

Boutet, I., A. Tanguy, M. Auffret, R. Riso & D. Moraga. 2002. Immunochemical quantification of metallothioneins in marine mollusks: Characterization of a metal exposure bioindicator. Environ. Toxicol. Chem. 21:1009-1014.

Breese, W.P. & R.E. Malouf. 1975. Hatchery manual for the Pacific oyster. Corvallis, OR: Oregon State University. Special report/ Agricultural Experiment Station, Oregon State University; no. 443 22 pp.

Brooks, C. 2000. Is the laboratory growth of Pacific oyster spat, Crassostrea gigas, exposed to varying salinities, predictive of their growth in the estuary? MS thesis. Oregon State University, Corvallis, OR.

Cherian, M. G. & R. A. Goyer. 1978. Metalothioneins and their role in the metabolism and toxicity of metals. Life Sci. 23:1-10.

Christy, A., D. P. Chrney & R. A. Elston. 2004. Characterization of the cadmium health risk, concentrations and ways to minimize residues in shellfish (abstract). Pacific Coast Shellfish Growers Association 58th Annual Conference, Tacoma, WA.

Clark, J. E. & R. Langmo. 1979. Oyster seed hatcheries on the US West Coast: an overview. Marine Fisheries Review Volume 41, issue 12, pp. 10-16.

Coon, S. L., D. B. Bonar & R. M. Weiner. 1986. Chemical production of cultchless oyster spat using epinephrine and norepinephrine. Aquaculture 58:255-262.

Davis, C. V. & B. J. Barber. 1999. Growth and survival of selected lines of eastern oysters, Crassostrea virginica (Gmelin 1791) affected by juvenile oyster disease. Aquaculture 178:253-271.

Degremont, L., E. Bedier, J. L. Martin, P. Soletchnick, J. P. Joly, M. Ropert, A. Huvet, J. Moal, J. F. Samain & P. Boudry. 2002. Genetic basis of survival in juvenile cupped oysters. 7th World Congress on Genetics Applied to Livestock Production, Montpelier, France. pp. 481-484.

Ernande, B., J. Clobert, H. McCombie & P. Boudry. 2003. Genetic polymorphism and trade-offs in the early life-history strategy of the Pacific oyster, Crassostrea gigas (Thunberg, 1795): a quantitative genetic study. J. Evol. Biol. 16:399-414.

Evans, S. 2004. Improving Pacific oyster (Crassostrea gigas) production through selective breeding in the Pacific Northwest. Ph.D. Thesis, Oregon State University, Corvallis, OR. 137 pp.

Falconer, D. S. & T. F. C. Mackay. 1996. An introduction to quantitative genetics. Essex, England: Addison Wesley Longman Limited. 464 pp.

Feder, M.E. & G.E. Hofmann. 1999. Heat-shock proteins, molecular chaperones, and the stress response: evolutionary and ecological physiology. Annu. Rev. Physiol. 61:243-282.

Geffard, A., J. C. Amiard & C. Amiard-Triquet. 2002. Kinetics of metal elimination in oysters from a contaminated estuary. Comp Biochem. Phys. C. 131:281-293.

Greig, R. A. & D. R. Wenzloff. 1978. Metal accumulation and depuration by the American oyster, Crassostrea virginica. Bull. Environ. Contam. Toxicol. 20:499-504.

Hedgecock, D., K. Cooper, W. Hershberger & X. Guo. 1993. Body size at harvest, sex ratio, and mantle color of pedigreed Pacific oysters (Crassostrea gigas) from controlled crosses. Aquaculture 111:299.

Ke, C. & W.-X. Wang. 2001. Bioaccumulation of Cd, Se, and Zn in an estuarine oyster (Crassostrea rivularis) and a coastal oyster (Saccostrea glomerata). Aquat. Toxicol. 56:33-51.

Langdon, C., F. Evans, D. Jacobson & M. Blouin. 2003. Yields of cultured Pacific oysters Crassostrea gigas Thunberg improved after one generation of selection. Aquaculture 220:227-244.

Lannan, J.E. 1971. Experimental self-fertilization of the Pacific oyster, Crassostrea gigas, utilizing cryopreserved sperm. Genetics 68:599-601.

Lannan, J. E. 1972. Estimating heritability and predicting response to selection for the Pacific oyster, Crassostrea gigas. Proceedings of the National Shellfisheries Association 62:62-66.

Moraga, D., E. Mdelgi-Lasram, M. S. Romdhane, A. El Abed, I. Boutet, A. Tanguy & M. Auffret. 2002. Genetic responses to metal contamination in two clams: Ruditapes decussatus and Ruditapes philippinarum. Mar. Environ. Res. 54:521-525.

Newkirk, G. F., L. E. Haley, D. L. Waugh & R. Doyle. 1977. Genetics of larvae and spat growth rate in the oyster Crassostrea virginiea. Mar. Biol. 41:49-52.

Okazaki, R. K. & M. H. Panietz. 1981. Depuration of twelve trace metals in tissues of oysters, Crassostrea gigas and Crassostrea virginica. Mar. Biol. 63:113-120.

Phillips, P. C. & S. J. Arnold. 1999. Hierarchical comparison of variance-covariance matrices. I. Using the Flury hierarchy. Evolution 53:1506-1515.

Reinfelder, J. R., W. X. Wang, S. N. Luoma & N. S. Fisher. 1997. Assimilation efficiencies and turnover rates of trace elements in marine bivalves: a comparison of oysters, clams and mussels. Mar. Biol. 129: 443-452.

Shulkin, V. M., B. J. Presley & V.la. Kavun. 2003. Metal concentrations in mussel Crenomytilus grayanus and oyster Crassostrea gigas in relation to contamination of ambient sediments. Environ. Int. 29:493-502.

Tanguy, A., I. Boutet, F. Bonhomme, P. Boudry & D. Moraga. 2002. Polymorphism of metallothionein genes in the Pacific oyster Crassostrea gigas as a biomarker of response to metal exposure. Biomarkers 7:439-450.

Tanguy, A., I. Boutet, R. Riso, P. Boudry, M. Auffret & D. Moraga. 2003. Metallothionein genes in the European flat oyster Ostrea edulis: a potential ecological tool for environmental monitoring? Mar. Ecol. Prog. Ser. 257:87-97.

Tanguy, A., C. Mura & D. Moraga. 2001. Cloning of a metallothionein gene and characterization of two other cDNA sequences in the Pacific oyster Crassostrea gigas (CgMT1). Aquat. Toxicol. 55:35-47.

van Dolah, F.M., T.C. Siewicki, G.W. Collins & J.S. Logan. 1987. Effects of environmental parameters on the elimination of cadmium by eastern oysters Crassostrea virginica. Arch. Environ. Contam. Toxicol. 8:85-95.

Wallner-Kersanach, M., H. Theede, U. Eversberg & S. Lobo. 2000. Accumulation and elimination of trace metals in a transplantation experiment with Crassostrea rhizophorae. Arch. Environ. Contain. Toxicol. 38:40-45.

Zaroogian, G. E. 1979. Studies on depuration of cadmium and copper by the American oyster, Crassostrea virginica. Bull. Environ. Contain. Toxicol. 23:117-122.

MARK D. CAMARA, (1) * STEPHEN M. GRIFFITH (2) AND SANFORD EVANS III (3)

(1) USDA, Agricultural Research Service, Hatfield Marine Science Center, 2030 SE Marine Science Dr., Newport, Oregon 97365; (2) USDA, Agricultural Research Service, Forage Seed Production Research Center, 3450 SW Campus Way, Corvallis, Oregon 97331; (3) Oregon State University, Coastal Oregon Marine Experiment Station, Hatfield Marine Science Center, 2030 SE Marine Science Dr., Newport, Oregon

* Corresponding author. E-mail: [e-mail] Any use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement or approval by the United States Department of Agriculture or the Agricultural Research Service of any product or service to the exclusion of others that may be suitable.

TABLE 1.
Mating design. Three large sires and three small sires were crossed
with the same five dams, resulting in 30 matings, 26 of these
matings were used in this experiment (X) and four did not produce
enough progeny (-).

                         Small Sires
                    (Live Body Weight, g)

                   Sire 1   Sire 2   Sire 3
      Live Body
Dam   Weight (g)   61.59    70.34    73.82

1     144.01         X        X        X
2     147.58         X        X        --
3     148.57         X        X        X
4     155.16         X        X        X
5     172.46         X        X        X

           Large Sires
      (Live Body Weight, g)

      Sire 4   Sire 5   Sire 6

Dam   138.22   138.56   158.35

1       X        --       X
2       --       X        X
3       X        X        X
4       X        --       X
5       X        X        X

TABLE 2.
Characters measured in this experiment.

Character         Experimental Units Measured   Description (units)

Yield             Lantern net compartments      Wet weight of whole
                                                  live oysters
                                                  harvested from 50
                                                  seed(kg).
Survival          Lantern net compartments      Proportion of seed
                                                  surviving to harvest
                                                  (none)
Live Weight       Individual oysters            Whole weight of live
                                                  animals (g)
Dry Meat Weight   Individual oysters            Weight freeze-dried
                                                  oyster meats (g)
Copper Content    Individual oysters            Copper content/dry
                                                  meat weight
                                                  ([micro]g/g)
Lead Content      Individual oysters            Lead content/dry meat
                                                  weight ([micro]g/g)
Zinc Content      Individual oysters            Zinc content/dry meat
                                                  weight ([micro]g/g)
Cadmium Content   Individual oysters            Cadmium content/dry
                                                  meat weight
                                                  ([micro]g/g)

TABLE 3.
Means, sample sizes and standard deviations of the eight
characters measured.

Variable           N     Mean     Std Dev     C.V.

Yield              93      1.82      0.983    0.541
Survival           93      0.27      0.147    0.546
Live weight       493    140.79     43.892    0.312
Dry meat weight   497      5.39      2.098    0.389
Copper            495     35.01     12.255    0.350
Lead              495      0.18      0.136    0.748
Zinc              495    277.63     82.992    0.299
Cadmium           495      1.55      0.494    0.319

TABLE 4.
Results from analyses of variance for each character. See text for
explanation. P-values <0.05 are in bold face.

   Source of
   Variance       DF     Type III SS    Type III MS     F     Prob >F

Yield
  Blocks            1          0.7752       0.77520    2.43    0.1235
  Sires             5         46.8509       9.37017   29.43   <.0001
  Dams              4          8.1903       2.04757    6.43    0.0002
  Sire x Dam       16          6.0890       0.38056    1.20    0.2955
  Error            66        21.01657       0.31843
Survival
  Blocks            1          0.0238       0.02381    3.39    0.0702
  Sires             5          1.0306       0.20612   29.33   <.0001
  Dams              4          0.2316       0.05791    8.24   <.0001
  Sire x Dam       16          0.1139       0.00712    1.01    0.4556
  Error            66         0.46389       0.00703
Live weight
  Blocks            1       123.04795     123.04795    0.08    0.7841
  Sires             5     23956.50601    4791.30120    2.93    0.0130
  Dams              4     64894.84899   16223.71225    9.91   -0.0001
  Sire x Dam       16     89198.41276    5574.90080    3.40   <.0001
  Error           466    763080.97160    1637.51280
Dry meat weight
  Blocks            1        36.13681      36.13681   10.12    0.0016
  Sires             5        64.83015      12.96603    3.63    0.0031
  Dams              4       203.28882      50.82220   14.23   <.0001
  Sire x Dam       16       204.36883      12.77305    3.58   <.0001
  Error           470      1678.45899       3.57119
Copper
  Blocks            1        24.56160      24.56160    0.21    0.6473
  Sires             5      3598.11827     719.62365    6.14   <.0001
  Dams              4      5887.97669    1471.99417   12.56   <.0001
  Sire x Dam       16      8499.23578     531.20224    4.53   <.0001
  Error           468     54849.49127     117.19977
Lead
  Blocks            1         0.00006       0.00006    0.01    0.9409
  Sires             5         1.02634       0.20527   19.28   <.0001
  Dams              4         2.01972       0.50493   47.42   <.0001
  Sire x Dam       16         1.07453       0.06716    6.31   <.0001
  Error           468         4.98371       0.01065
Zinc
  Blocks            1     19438.59360   19438.59360    3.45    0.0639
  Sires             5     39544.39210    7908.87840    1.40    0.2217
  Dams              4    297046.80630   74261.70160   13.18   <.0001
  Sire x Dam       16    309618.09620   19351.13100    3.43   <.0001
  Error           468   2637827.76300    5636.38400
Cadmium
  Blocks            1         0.82398       0.82398    4.70    0.0307
  Sires             5        13.73373       2.74675   15.65   <.0001
  Dams              4        10.39830       2.59958   14.81   <.0001
  Sire x Dam       16        15.31493       0.95718    5.45   <.0001
  Error           468        82.12644       0.17548

TABLE 5.
Heritabilities and genetic correlations for the eight traits
measured in this study. Diagonal elements of the matrix (shaded)
are the single-trait heritabilities (bootstrapped standard errors).
Off-diagonal elements are Pearson correlation coefficients (r).
Correlations above the diagonal are narrow-sense genetic correlations
(family means correlations among the six paternal half-sib families).
Those below the diagonal are broad sense genetic correlations (family
means correlations among the 26 full-sib families). All estimates are
based on block-adjusted data. Significance tests for heritabilities
are based on the proportions of bootstrap re-samplings distributions
that overlap zero; those for correlations are standard parametric
hypothesis tests.

                 Yield         Survival       Live Wt

Yield         1.28 (0.185)       0.97          -0.07
                  ****           ****            ns
Survival          0.99       1.29 (0.186)      -0.24
                  ****           ****            ns
Live Wt          -0.48          -0.58       0.30 (0.099)
                   ns             ns            ****
Dry Meat Wt      -0.70          -0.75           0.75
                   ns             *              *
Copper            0.90           0.86          -0.19
                  ***             **             ns
Lead              0.33           0.26           0.25
                   ns             ns             ns
Zinc              0.58           0.55           -0.1
                   ns             ns             us
Cadmium           0.78           0.78          -0.43
                   *              *              ns

              Dry Meat Wt       Copper          Lead

Yield            -0.20           0.47           0.21
                   ns             **             ns
Survival         -0.37           0.56           0.24
                   **             **             ns
Live Wt           0.70          -0.17          -0.14
                  ****            ns             ns
Dry Meat Wt   0.32 (0.104)      -0.66          -0.13
                  ****           ****            ns
Copper           -0.47       0.42 (0.110)       0.25
                   ns            ****            ns
Lead              0.38           0.58       0.94 (0.301)
                   ns             ns            ****
Zinc             -0.21           0.81           0.57
                   ns             *              ns
Cadmium          -0.76           0.49          -0.19
                   *              ns             ns

                  Zinc         Cadmium

Yield             0.18           0.46
                   ns             **
Survival          0.29           0.56
                   ns            ***
Live Wt          -0.17          -0.29
                   ns             ns
Dry Meat Wt      -0.54          -0.66
                  ***            ****
Copper            0.85           0.70
                  ****           ****
Lead              0.14           0.27
                   ns             ns
Zinc          0.34 (0.099)       0.53
                  ****           ***
Cadmium          0.047       0.53 (0.118)
                   ns            ****

**** P <0.001; *** P < 0.01; ** P <0.05; * P < 0.01; ns P > 0.1.