Dietary Changes Cause Temporal Fluctuations in Polychlorinated

Stable isotope analysis ([15N]/[14N]) of herring gull eggs provided evidence ..... Saskatoon) and C. van Kessel (University of Saskatchewan) conducted...
0 downloads 0 Views 288KB Size
Environ. Sci. Technol. 1997, 31, 1012-1017

Dietary Changes Cause Temporal Fluctuations in Polychlorinated Biphenyl Levels in Herring Gull Eggs from Lake Ontario C. E. HEBERT,* J. L. SHUTT, AND R. J. NORSTROM Canadian Wildlife Service, National Wildlife Research Centre, 100 Gamelin Boulevard, Hull, Que´bec, Canada K1A 0H3

After adjusting Lake Ontario herring gull egg PCB concentrations for the influence of time, an analysis was conducted to explain the remaining variation in annual egg PCB concentrations. In years with cold winters and/ or high alewife abundance, egg PCB concentrations were greater than predicted. PCB levels were also greater than predicted in years when alewife condition was low. Increasing the proportion of alewives in the gull’s diet may lead to increased PCB levels in eggs. Stable isotope analysis ([15N]/[14N]) of herring gull eggs provided evidence supporting this hypothesis. Consumption of alewives by gulls (as influenced by gull metabolism, alewife abundance/ condition, and alewife overwinter mortality) and alewife population characteristics (growth rates and age distribution) may be the keys to explaining fluctuations in Lake Ontario herring gull egg PCB levels.

Introduction Since the mid-1980s, polychlorinated biphenyl (PCB) concentrations in herring gull (Larus argentatus) eggs from most colonies in the Great Lakes have shown little change (1, 2). This apparent lack of change may be, in part, the result of the difficulty in assessing recent temporal trends. This difficulty arises because of irregular perturbations (i.e., unexpected increases or decreases) in herring gull egg contaminant levels. Smith (3) showed that these temporal fluctuations were similar among Great Lakes gull populations and deduced a generalized mechanism to explain them. He speculated that, in warm years, contaminant levels would be less than expected because phytoplankton growth rates would be more rapid. This phenomenon would result in less contaminated plankton because the rate of phytoplankton growth would exceed the rate at which chemicals would be incorporated into the plankton. If this effect was to be transferred through the food web, lower than expected contaminant levels would be found in herring gull eggs laid that spring. We do not believe that this is a tenable hypothesis given that Great Lakes herring gulls lay their eggs very early in the spring (mid April-early May) and aquatic primary/secondary production at that time is low (4). Smith’s hypothesis would require an unrealistically rapid transfer of the “phytoplankton effect” through the food web such that herring gull eggs in the same year would be affected. Alternative hypotheses may be more plausible. If we are to assess our progress toward achieving the virtual elimination of persistent contaminants in the Great Lakes, we need to understand the factors that cause these temporal * Corresponding author phone: (819) 953-3904; fax: (819) 9536612; e-mail: [email protected].

1012 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997

fluctuations in contaminant levels. Greater understanding of the processes involved in regulating temporal fluctuations in contaminant levels may allow us, in the future, to account for the influence of these factors and to enhance our ability to detect temporal changes in Great Lakes contaminant levels. In this paper, we examine the influence of changes in dietary composition on levels of polychlorinated biphenyls (PCBs) in herring gull eggs from Lake Ontario. We have focussed on the polychlorinated biphenyls because of their toxicological importance.

Materials and Methods Herring Gull Egg Collection. Herring gull eggs were collected annually from two colonies on Lake Ontario: Toronto Harbour (1974-1975, 1977-1996) and Snake Island near Kingston in the eastern part of the lake (1977-1996). Homogenates of 10 whole eggs per colony were pooled on an equal weight basis for each year. All of the samples were stored at -40 °C until analyzed. Details regarding sample collection, storage, and processing have been reported previously (5, 6). Organochlorine Analysis of Herring Gull Eggs. Organochlorine analyses were completed according to the methodology described in Norstrom et al. (7). Total PCB concentrations reported in this paper are 1:1 Aroclor 1254:1260 estimates. In herring gull eggs, Aroclor estimates of total PCB levels are known to overestimate total PCB levels calculated using the sum of individual congeners by approximately 2.2 times (6). Stable Isotope Analysis of Herring Gull Eggs. Stable isotope analyses ([15N]/[14N]) were conducted on subsamples of the same herring gull egg pools that were used for organochlorine analysis. Analyses were conducted on samples collected from both Snake Island (1982-1992) and Toronto Harbour (1982-1993). Isotopes of 15N and 14N were measured to provide an indication of gull trophic position. Higher [15N]/ [14N] levels indicate feeding at higher trophic levels. Methodology is described in Hobson (8). Weather Information. Heating degree days (HDD) from November to April (1974-1995) were used as a means of comparing degree of coldness among years (1996 data were not available at the time of analysis). Heating degree days were calculated daily using the equation: HDD ) 18 °C - xj daily temperature ( °C). If the value was negative, then HDD were zero. As the degree of cold increases so does the number of HDD. For Snake Island, HDD were obtained from Kingston, Ontario, approximately 5 km from the colony. For Toronto Harbour, HDD were obtained from Toronto Island, which is located within the harbour. The HDD data were grouped into an autumn/winter period (November-February) and a winter/spring period (March and April). This grouping was done to determine if the weather immediately preceding egg laying (winter/spring period) had a greater influence on egg PCB levels than weather during the rest of the winter. Estimates of Alewife Condition and Biomass. In Lake Ontario, the pelagic forage fish community consists almost entirely of three species: the alewife (Alosa pseudoharengus), the rainbow smelt (Osmerus mordax), and the slimy sculpin (Cottus cognatus) (9, 10). Only the smelt and the alewife frequent the surface waters of the lake, thereby increasing their vulnerability to predation by surface-feeding birds. The alewife, however, is much more abundant than the smelt (10). From 1978 to 1985, estimates of the relative population size of alewife and rainbow smelt indicated that the alewife was over 10 times more abundant; alewife remain the primary prey fish species in the lake (10, 11; R. O’Gorman, unpublished data). Hence, factors that regulate alewife availability to the

S0013-936X(96)00408-7 CCC: $14.00

Published 1997 by the Am. Chem. Soc.

herring gull would be extremely important in regulating total fish consumption by herring gulls, which are facultative piscivores. Annual indices of alewife biomass (1978-1995) and condition (1977-1995) were obtained from spring trawling surveys conducted by the National Biological Service and the New York Department of Environmental Conservation (9, 10, 12; R. O’Gorman, unpublished data). Spring (April/May) alewife condition was estimated by calculating the mean weight of a 165-mm alewife from a length-weight regression. Statistical Analysis of PCB Trends in Herring Gull Eggs. For data from both colonies, statistical analyses (13) were undertaken in three steps: (1) Temporal trends in log10 contaminant levels were examined, and a single exponential model with a non-zero asymptote was fitted to the data. (2) Residuals (PCB residuals) were calculated for each of the years as the difference between the measured concentration and that predicted by the model. The magnitude and sign (positive or negative) of the residuals were used to indicate whether the PCB concentration for a particular year was lower or greater than expected based upon time alone. Residuals from both colonies were normally distributed (Shapiro-Wilks’ W test). Residuals were not autocorrelated (Box-Ljung statistic). The residuals from this analysis were then used to examine the factors that contributed to temporal fluctuations in egg PCB levels. (3) Forward stepwise multiple regression was used to examine the relationship between these residuals and factors possibly influencing the availability of alewife to the gull (autumn/winter HDD, winter/spring HDD, alewife biomass and condition, and rainbow smelt biomass). Because alewife abundance data were only available after 1977, this analysis only pertains to the years 1978-1995 (N ) 18; 1996 data were not available at the time of analysis). F values (F g 2) were used to determine which independent variables were included in the regression. The redundancy of independent variables in the regression was tested using a tolerance statistic defined as one minus the squared multiple correlation of each variable with all other independent variables included in the regression equation. Therefore, the larger the tolerance of a variable (the closer it was to one) the less redundant was its contribution to the regression. Standardized regression coefficients indicated which independent variables contributed most to the prediction of PCB residuals. Correlation analysis was used to examine the relationship between fluctuations in gull diet as indicated by stable isotopes (15N/14N) and PCB residuals.

Results and Discussion PCB levels in herring gull eggs from Snake Island and Toronto Harbour declined through time (Figure 1a,b). A single exponential model with a non-zero asymptote fit the data well (Snake Island, r 2 ) 0.88; Toronto Harbour, r 2 ) 0.91). Stow (2) also found this model to be most appropriate for describing temporal declines in PCB levels in herring gull eggs from Lake Ontario. Residuals were calculated after fitting this model. The use of the non-zero asymptote model should not be misconstrued as indicating that PCB concentrations in Lake Ontario herring gull eggs have stopped declining. Levels continue to decline but at a slower rate than observed in the 1970s and early 1980s. Eventually an asymptote may be reached, but there is no evidence to suggest that this has occurred at the present time. Tolerance statistics, standardized and raw regression coefficients, standard errors for the raw regression coefficients, and probability levels resulting from the stepwise multiple regression analyses are shown in Table 1, Sections A (Snake Island) and B (Toronto Harbour). Table 2 summarizes the amount of variance explained by the independent variables retained in each of the regressions. At Snake Island, four variables were retained within the regression: winter/spring

a

b FIGURE 1. Temporal trends in log10 PCB concentrations in herring gull eggs from Lake Ontario. A single, exponential model with a non-zero asymptote has been fitted to the data from both colonies: (a) Snake Island, 1977-1996 and (b) Toronto Harbour, 1974-1996. HDD, alewife biomass, autumn/winter HDD, and alewife condition. This regression explained 76% of variance in the PCB residual data. Both alewife biomass and condition were retained in the regression because these two variables were not significantly correlated during the period (1978-1995) examined in this study (n ) 18, p ) 0.3). At Toronto Harbour, the same four variables were retained within the regression (winter/spring HDD, autumn/winter HDD, alewife biomass, and alewife condition). A total of 64% of the variance in the PCB residual data was explained by this regression. At both colonies, autumn/winter and winter/spring HDD were positively correlated with PCB residuals (Table 1). Alewife biomass was also positively correlated with egg PCB residuals (Table 1). A negative correlation was observed between alewife condition and PCB residuals (Table 1). Winter/spring HDD contributed the most to predicting PCB residuals as indicated by standardized regression coefficients (Table 1). The relationships between log10 transformed PCB residuals (difference from predicted PCB level) and winter/ spring HDD are shown in Figures 2, panels a (Snake Island) and b (Toronto Harbour). Winter/spring HDD were correlated with PCB residuals, probably because of changes in adult metabolic rate in response to changes in ambient temperature. In any year, the metabolic rate of the gull will respond even to small changes in ambient temperature below a lower critical temperature. An increase in metabolic rate, and hence energy utilization, in response to cold would result in an increase in prey consumption rates. This would increase the adult’s exposure to PCBs. This may be of particular importance during the period immediately preceding egg formation and may have been partially responsible for the higher PCB levels in eggs laid after cold winter/springs. Although winter/spring HDD were most highly correlated with PCB residuals, the amount of variance explained by this one variable was relatively low (Figure 2a,b). Examination of the other variables included in the multiple regression

VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1013

TABLE 1. Variables Retained in the Forward, Stepwise Multiple Regression Predicting PCB Residuals from Herring Gull Eggs step

variable

1 2 3 4

intercept winter/spring HDD alewife biomass autumn/winter HDD alewife condition

1 2 3 4

intercept winter/spring HDD autumn/winter HDD alewife biomass alewife condition

tolerance

standard reg coeff

raw reg coeff

std error

t(13)

p

-1.0003 0.0007 0.0032 0.0002 -0.0124

0.3022 0.0002 0.0008 0.0001 0.0059

-3.31 4.43 3.82 2.24 -2.11

0.006 0.0007 0.002 0.04 0.05

-1.26 0.0009 0.0003 0.0013 -0.3629

0.3631 0.0003 0.0001 0.001 0.3247

-3.46 3.43 2.10 1.32 -1.12

0.004 0.005 0.056 0.21 0.28

(a) Snake Island 0.946 0.870 0.872 0.893

0.62 0.56 0.33 -0.31 (b) Toronto Harbour

0.874 0.848 0.905 0.864

0.61 0.38 0.23 -0.20

TABLE 2. Summary of the Forward, Stepwise Regression Examining the Relative Contribution of Winter/Spring and Autumn/Winter HDD and Alewife Biomass and Condition to Predicting PCB Residuals from Snake Island and Toronto Harbour Herring Gull Eggs Snake Island

Toronto Harbour

step

variable

R2

R 2∆

p

variable

R2

R 2∆

p

1 2 3 4

winter/spring HDD alewife biomass autumn/winter HDD alewife condition

0.31 0.61 0.69 0.76

0.31 0.30 0.08 0.07

0.02 0.01 0.11 0.05

winter/spring HDD autumn/winter HDD alewife biomass alewife condition

0.46 0.54 0.61 0.64

0.46 0.08 0.07 0.03

0.003 0.14 0.13 0.28

(autumn/winter HDD, alewife biomass, and alewife condition) sheds further light on additional processes contributing to fluctuations in egg PCB concentrations. The positive relationship between autumn/winter and winter/spring HDD and PCB residuals (colder years had higher than expected PCB levels in eggs) and the negative relationship between alewife condition and PCB residuals (in years when alewife condition was low, PCB levels were greater than expected) indicated that PCB levels in herring gull eggs may have been related to the degree of overwinter mortality of alewives. Winter/spring HDD were better correlated with PCB residuals than autumn/winter HDD, possibly because overwinter mortality of alewife only occurs after a cold threshold is reached. Therefore, the relationship between autumn/winter HDD and PCB residuals would not be as obvious. The alewife is an exotic species that was first recorded in Lake Ontario in the 1870s (14). When water temperatures fall below 6 °C their locomotory ability declines and below 3 °C alewife mortality may occur (15). In Lake Ontario during cold winters, alewives occupy depths where they are frequently exposed to potentially lethal temperatures (16). Previous studies have shown the importance of winter severity and alewife condition in causing alewife overwinter mortality (15, 17). Gull predation on dead/dying alewife affected by the severe 1983-1984 winter was observed in the offshore waters of eastern Lake Ontario in February of 1984 (R. O’Gorman, unpublished data). Thus, overwinter mortality of alewives may be important in regulating alewife availability to the herring gull. Alewife die-offs have been documented in Lake Ontario during the winters of 1976-1977, 1981-1982, and 1983-1984 (10). Cold weather and alewife condition were implicated in these events (10). In addition, it has been speculated that smaller overwinter mortalities may have occurred in 1991/92 and 1992/93 (18, 19). These years correspond to some of the years with greater than expected egg PCB levels (Figure 1, panels a and b). Although other fish species may also suffer greater mortality in cold winters, their relative abundances compared to the alewife are much lower and, therefore, their contribution to increased PCB levels in gull eggs would also be less important. The predominance of fish in the Lake Ontario/Erie herring gull diet during incubation and chick rearing has been

1014

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997

demonstrated (20, 21). Their importance in the herring gull’s winter diet has also been shown (22). However, herring gulls are opportunists and will consume other food items. These include garbage, vegetation, worms, insects, crayfish, amphibians, herbivorous mammals, and birds (20, 22). Hence, the annual/seasonal composition of the gull diet may vary significantly in response to changes in the relative availability of these food types (20, 22), and weather may be an important factor affecting their availability. In severe winters, direct mortality of alewives may increase or their behavior might be altered (e.g., reduced locomotory ability) such that they are more vulnerable to gull predation. Increased gull predation on alewives would result in an increase in the proportion of the gull diet consisting of fish. If the proportion of fish in the diet increased, an increase in [15N]/[14N] values would also be expected. This would occur because the fish species the herring gull consumes (primarily the alewife) are carnivorous; therefore, their [15N]/[14N] values would be greater than those found in many of the other possible components of the gull diet. This was tested by measuring stable isotope [15N]/[14N] values in herring gull eggs. The utility of nitrogen isotope ratios in quantifying the trophic position of an organism has been documented (23). The use of avian eggs in this regard has also been demonstrated (8, 24). Results from the stable isotope analysis revealed no significant relationship between [15N]/[14N] values and PCB residuals in gull eggs from Toronto Harbour (p ) 0.68). However, at Snake Island there was a marginally significant positive relationship between [15N]/[14N] values and PCB residuals (r 2 ) 0.34, p ) 0.06) (Figure 3). The lack of a relationship at Toronto Harbour and the marginally significant relationship at Snake Island may have been the result of the limited resolution of the analysis as a result of using pooled egg samples. Within herring gull colonies, substantial individual variation in foraging habits exists during the period prior to egg-laying (25). This could affect nitrogen ratios in eggs. Nitrogen turnover rates in eggs are on the order of weeks (24), much faster than turnover rates for PCBs that have half-lives in fish and birds measured in months (26, 27). Therefore, the potential for individual variation in [15N]/ [14N] values is much greater than that for PCBs. The coefficient of variation for mean [15N]/[14N] values during the study period was greater at Toronto Harbour (9.4%) than at Snake Island

a

FIGURE 3. Relationship between PCB residuals and [15N]/[14N] values in herring gull eggs from Snake Island, 1982-1992.

b

FIGURE 2. Relationship between log10 PCB egg concentrations adjusted for time (PCB residuals) and winter/spring coldness (heating degree days): (a) Snake Island and (b) Toronto Harbour. (4.6%), indicating that there may have been greater variation in the foraging habits of gulls breeding in Toronto Harbour as a result of the proximity of anthropogenic food sources. This variation may, in part, explain the intercolony differences in the relationship between [15N]/[14N] values and PCB residuals. The positive relationship between [15N]/[14N] and PCB residuals at Snake Island was probably the result of an increased proportion of fish in the gull diet as a result of weather-induced winter mortality of alewives. Of the three winters for which major alewife mortalities have been documented (1977, 1982, and 1984), [15N]/[14N] data are only available from 1982 and 1984. For the 11 years in which stable isotope ratios were measured in Snake Island herring gull eggs, [15N]/[14N] values in 1984 and 1982 ranked second and fourth, respectively. Corroborating this result was the observation that the frequency of occurrence of alewives in herring gull pellets collected from Lake Ontario colonies in April 1982 was greater than that observed in 1983 (April), 1990 (December), or 1991 (February) (22). PCB concentrations in gull eggs collected in all 5 years in which alewife mortalities may or were known to have occurred were greater

FIGURE 4. Temporal trends in Lake Ontario alewife biomass and condition. Data provided by the U.S. National Biological Service, Oswego, NY. than expected (Figure 1a). Therefore, it seems that as the composition of the herring gull diet changed in response to weather conditions these dietary shifts were reflected in changes in [15N]/[14N] ratios (at Snake Island) and PCB concentrations in herring gull eggs (at both colonies). In addition to the role that alewife condition may have played in regulating alewife overwinter mortality, it also has the potential to affect rates of food consumption in the herring gull, thereby affecting PCB accumulation. Alewife condition declined greatly over the course of this study (Figure 4). Brekke and Gabrielsen (28) found that food assimilation efficiency in seabirds was lower when the lipid content of their fish prey was low. When prey nutrient quality declines, there could be further increases in prey consumption rates to compensate for the lower caloric content of each prey item. Existing evidence for Lake Ontario salmon supports this contention. From 1978 to 1990, the estimated energy density of alewife in Lake Ontario changed dramatically (11). Energy densities in alewives peaked in 1979 (6259 J/g wet wt) after which they declined erratically to approximately 4600 J/g wet wt in 1985. From 1986 to 1990, energy densities remained low. Through the application of a bioenergetics model, these reductions in energy density were shown to increase prey consumption

VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1015

rates in the chinook salmon (Oncorhynchus tshawytscha). This increase in prey demand may have caused the observed declines in salmonine condition observed during this period, as a result of increased foraging costs and reduced food conversion efficiencies (11). These same effects would be expected in the herring gull if it was as closely linked to the aquatic food web as the salmon. However, the gull has the ability to shift its diet in response to the relative abundance of prey. Although fish are the preferred prey from a caloric standpoint (29), the foraging behavior of the gull can vary to maximize the efficiency with which energy is acquired along with other nutritional requirements (25, 29). Other years (1975, 1978, and 1981) that deviate from the expected relationship between PCB residuals and winter/ spring HDD (Figure 2a,b) can be explained in terms of temporal changes in alewife population characteristics (Figure 4). One aspect of this lies in the positive relationship between alewife abundance and PCB residuals. When alewife abundance is high, there is an increased probability of predatory encounters occurring between gulls and alewives. Increased liklihood of encounters may result in a greater proportion of the gull’s diet consisting of alewives. Ca. 1990, PCB concentrations in alewives (mean 0.43 µg g-1 wet weight, D. M. Whittle; Fisheries and Oceans Canada, Burlington, Ontario, unpublished data) were greater than those found in the gull’s other major foods, such as small mammals, invertebrates, and plant material (30). With respect to garbage as a source of PCBs to the gull, PCB levels in edible refuse should reflect guidelines set for human consumption. The guideline values for poultry and other meats are also lower than concentrations found in Lake Ontario alewives (31). Therefore, in years of high alewife abundance, increased alewife consumption by gulls will lead to increased PCB exposure to the adult gull and increased PCB accumulation in the adult and its eggs. In addition, other factors such as alewife growth rates and age may also affect gull PCB levels. Unfortunately, there are no alewife data for years prior to 1978, so we cannot examine the possible mechanism behind the 1975 anomaly. However, in 1978, alewife abundance was very low and alewife condition was extremely high (Figure 4). This phenomenon was the result of the massive overwinter alewife mortality that occurred during the winter of 1976-1977 (9, 10). This winter mortality resulted in increases in growth rates and condition of the survivors in 1978 as density-dependent limitations on the alewife population were reduced. The lower than expected PCB levels in gull eggs from 1978 may have been the result of low alewife PCB levels because of growth dilution combined with the low probability of predatory encounters occurring between gulls and alewife. By 1981, the alewife population had recovered and abundance was extremely high (Figure 4). In fact, it was so great that many of the fish stopped growing (9, 32, 33). This reduction in growth may have resulted in higher PCB concentrations in the alewife. The high probability of predatory encounters occurring between gull and alewife combined with higher alewife PCB concentrations might have resulted in the greater than expected PCB levels observed in herring gull eggs laid that year. Although 1978 and 1981 are the most obvious examples of deviations from the relationship between PCB residuals and winter/spring HDD, it is possible that shifts in alewife abundance played some role in regulating gull egg PCB concentrations throughout the entire 1974-1996 period. Large changes in alewife abundance lead to shifts in alewife growth rates and in the age and size structure of the population (9, 12). For example, from 1984 to 1986 the age distribution of alewives in Lake Ontario changed dramatically. The proportion of the population age four and older decreased from 87% in 1984 to 45% in 1985 to 26% in 1986 (33). Older alewife would be expected to have higher PCB concentrations (34), so the shifting age distribution from 1984 to 1986 probably played some part in the greater than expected PCB levels in

1016

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997

1984 and the lower than expected levels in 1986. Unfortunately, there is no continuous data set examining age and growth rates in Lake Ontario alewives during the period examined in this study, so their influence on PCB levels in gulls cannot be examined systematically. Annual changes in the age distribution of the Lake Ontario alewife population have been proposed as a mechanism explaining temporal fluctuations in lake trout (Salvelinus namaycush) PCB concentrations (34). Hence, alewife dynamics may be an important factor regulating PCB concentrations in both these terminal predators. The results from this study emphasize the unique nature of the processes that may be occurring in each of the Great Lakes. The mechanisms relating weather to PCB accumulation in Lake Ontario herring gulls may be specific to that lake, particularly those that are controlled by alewife dynamics. PCB levels in herring gull eggs from the other Great Lakes also show relationships with weather, but the significance of this relationship varies among colonies (3; CWS, unpublished data). Other mechanisms that may be responsible for regulating egg organochlorine concentrations include changes in herring gull migratory behavior in response to weather and changes in the bioavailability of contaminants as a result of weather-mediated physical processes (35). These possibilities are being explored. If we are to enhance our ability to interpret temporal trends in contaminant concentrations inferred from the Herring Gull Surveillance Program, we must first have a sound understanding of herring gull biology and the factors that regulate contaminant exposure. Short-term fluctuations in herring gull egg contaminant concentrations may not reflect changes in the level of ecosystem contamination but may reflect the ecological, physiological, and behavioral adaptations of the gull in response to its dynamic environment. This study emphasizes the importance of continuing long-term monitoring programs but with the recommendation that ensuing results be interpreted in an integrated, ecosystemic fashion.

Acknowledgments The work of D. V. Weseloh (CWS, Ontario Region) and R. O’Gorman (National Biological Service, Oswego, NY) was instrumental in completing this study. Thanks to K. A. Keenleyside for her insightful contributions. D. Russell and K. Timm prepared the egg samples for chemical analysis, and H. Won conducted the analyses. C. Kocot (Environment Canada, Atmospheric Environment Service) provided the heating degree data. K. Hobson (CWS, Saskatoon) and C. van Kessel (University of Saskatchewan) conducted the stable isotope analyses. D. Smith and two anonymous reviewers improved the manuscript. Environment Canada’s Great Lakes Action Plan provided funds for this study.

Literature Cited (1) Weseloh, D. V.; Pekarik, C. Abstracts of Papers, Second World Congress of the Society of Environmental Toxicology and Chemistry, Vancouver, BC, 1995. (2) Stow, C. A. Environ. Sci. Technol. 1995, 29, 2893-2897. (3) Smith, D. W. Environ. Sci. Technol. 1995, 29, 740-750. (4) Wetzel, R. G. Limnology; W. B. Saunders Company: Toronto, Ontario, 1975; p 743. (5) Mineau, P.; Fox, G. A.; Norstrom, R. J.; Weseloh, D. V.; Hallett, D. J.; Ellenton, J. A. In Toxic Contaminants in the Great Lakes; Nriagu, J. O., Simmons, M. S., Eds.; J. Wiley and Sons: New York, 1984; pp 425-452. (6) Turle, R.; Norstrom, R. J.; Collins, B. Chemosphere 1991, 22, 201213. (7) Norstrom, R. J.; Simon, M.; Muir, D. C. G. Environ. Sci. Technol. 1988, 22, 1063-1071. (8) Hobson, K. A. Condor 1995, 97, 752-762. (9) O’Gorman, R.; Schneider, C. P. Trans. Am. Fish. Soc. 1986, 115, 1-14. (10) O’Gorman, R.; Bergstedt, R.; Eckert, T. H. Can. J. Fish. Aquat. Sci. 1987, 44, 390-403.

(11) Rand, P. R.; Lantry, B. F.; O’Gorman, R.; Owens, R. W.; Stewart, D. J. Trans. Am. Fish. Soc. 1994, 123, 519-534. (12) Jones, M. L.; Koonce, J. F.; O’Gorman, R. Trans. Am. Fish. Soc. 1993, 122, 1002-1018. (13) StatSoft. In Statistica Procedures Manuals. Vol. 1-5; StatSoft Inc.: Tulsa, OK, 1995. (14) Smith, S. H. Trans. Am. Fish. Soc. 1970, 99, 754-765. (15) Colby, P. J. In Responses of fish to environmental changes; Chavin, W., Ed.; C. C. Thomas: Springfield, 1973. (16) Bergstedt, R. A.; O’Gorman, R. Trans. Am. Fish. Soc. 1989, 118, 687-692. (17) Brown, E. H., Jr. J. Fish. Res. Board Can. 1972, 29, 447-500. (18) Ontario Ministry of Natural Resources. In Annual Report. Lake Ontario Fisheries Research Station; OMNR: Glenora, Ontario, 1993. (19) National Biological Service. In Status of prey fishes in the U.S. waters of Lake Ontario, 1993; Department of the Interior: Lake Ontario Biological Station, Oswego, NY, 1994. (20) Fox, G. A.; Allan, L. J.; Weseloh, D. V.; Mineau, P. Can. J. Zool. 1990, 68, 1075-1085. (21) Belant, J. L.; Seamans, T. W.; Gabrey, S. W.; Ickes, S. K. Condor 1993, 95, 817-830. (22) Ewins, P. J.; Weseloh, D. V.; Groom, J. H.; Dobos, R. Z.; Mineau, P. Hydrobiologia 1994, 279/280, 39-55. (23) Peterson, B. J.; Fry, B. Annu. Rev. Ecol. Syst. 1987, 18, 293-320. (24) Hobson, K. A. Abstracts of Papers, Second World Congress of the Society of Environmental Toxicology and Chemistry, Vancouver, BC, 1995. (25) Pierotti, R.; Annett, C. A. Ecology 1991, 72, 319-328. (26) Baumann, P. C.; Whittle, D. M. Aquat. Toxicol. 1988, 11, 241257.

(27) Hebert, C. E.; Norstrom, R. J.; Zhu, J.; Macdonald, C. R. J. Great Lakes Res. Submitted for publication. (28) Brekke, B.; Gabrielson, G. W. Polar Biol. 1994, 14, 279-284. (29) Pierotti, R.; Annett, C. A. In Foraging Behavior; Kamil, A. C., Krebs, J. R. Jr., Pulliam, H. R., Eds.; Plenum Publishing: New York, 1987; pp 417-442. (30) Hebert, C. E.; Weseloh, D. V.; Kot, L.; Glooschenko, V. Arch. Environ. Contam. Toxicol. 1994, 26, 356-366. (31) Wong, M. P. In Chemical Residues in Fish and Wildlife Harvested in Northern Canada; Indian and Northern Affairs Canada: Ottawa, Ontario, 1985. (32) O’Gorman, R.; Barwick, D. H.; Bowen, C. A. In Age and Growth of Fish; Iowa State University Press: Ames, IA, 1987; pp 203210. (33) Madenjian, C. P.; Whittle, D. M.; Elrod, J. H.; O’Gorman, R.; Owens, R. W. Environ. Sci. Technol. 1995, 29, 2610-2615. (34) Borgmann, U.; Whittle, D. M. J. Great Lakes Res. 1991, 17, 368381. (35) Ludwig, J. P.; Auman, H. J.; Kurita, H.; Ludwig, M. E.; Campbell, L. M.; Giesy, J. P.; Tillitt, D. E.; Yamashita, N.; Tanabe, S.; Tatsukawa, R. J. Great Lakes Res. 1993, 19, 96-108.

Received for review May 8, 1996. Revised manuscript received September 25, 1996. Accepted October 7, 1996.X ES960408P X

Abstract published in Advance ACS Abstracts, February 1, 1997.

VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1017