Environ. Sci. Technol. 1999, 33, 3768-3773
Variation in Net Trophic Transfer Efficiencies among 21 PCB Congeners C H A R L E S P . M A D E N J I A N , * ,† LARRY J. SCHMIDT,‡ SERGEI M. CHERNYAK,† ROBERT F. ELLIOTT,§ TIMOTHY J. DESORCIE,† RICHARD T. QUINTAL,† LINDA J. BEGNOCHE,† AND ROBERT J. HESSELBERG† U.S. Geological Survey, Biological Resources Division, Great Lakes Science Center, 1451 Green Road, Ann Arbor, Michigan 48105, U.S. Geological Survey, Biological Resources Division, Upper Mississippi Science Center, 2630 Fanta Reed Road, LaCrosse, Wisconsin 54603, and U.S. Fish and Wildlife Service, Green Bay Fishery Resources Office, 1015 Challenger Court, Green Bay, Wisconsin 54311
We tested the hypothesis that the efficiency with which fish retain polychlorinated biphenyl (PCB) congeners from their food strongly depends on Kow and degree of chlorination of the congener. We used diet information, determinations of concentrations of individual PCB congeners in both coho salmon (Oncorhynchus kisutch) and their prey, and bioenergetics modeling to estimate the efficiencies with which Lake Michigan coho salmon retain various PCB congeners from their food. The retention efficiency for the tetrachloro congeners averaged 38%, whereas retention efficiencies for higher chlorinated congeners ranged from 43 to 56%. Not including tetrachloro congeners, we found neither decreasing nor increasing trends in the efficiencies with which the coho salmon retained the PCB congeners from their food with either increasing Kow or increasing degree of chlorination of the PCB congeners. We concluded that (a) for PCB congeners with 5-8 chlorine atoms/molecule, Kow and degree of chlorination had little influence on the efficiency with which coho salmon retained the various PCB congeners in their food, and (b) the efficiency with which coho salmon retained tetrachloro PCB congeners in their food appeared to be slightly lower than that for higher chlorinated PCB congeners.
Introduction The efficiency with which piscivorous fish retain contaminants from their prey is one of the most important factors governing contaminant accumulation in these fish (1, 2). Moreover, such information can be used in risk assessment models to predict future risk to both people and wildlife eating contaminated fish (3). Trophic transfer efficiency of contaminants within Great Lakes ecosystems has been addressed in numerous modeling exercises (4-12). * Corresponding author telephone: (734)214-7259; fax: (734)9948780; e-mail:
[email protected]. † U.S. Geological Survey, Great Lakes Science Center. ‡ U.S. Geological Survey, Upper Mississippi Science Center. § U.S. Fish and Wildlife Service. 3768 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 21, 1999
Gross trophic transfer efficiency refers to the efficiency with which the contaminant in the food ingested by the predator is taken up through the gut wall of the predator. Once a quantity of contaminant has been taken up through the gut wall, a portion of this quantity may eventually be eliminated (or excreted) from the fish’s body and/or a portion of this quantity may be metabolically transformed into another chemical compound. Net trophic transfer efficiency refers to the efficiency with which the contaminant in the food ingested by the predator is retained by the predator, including the losses due to elimination and metabolic transformation (11). One of the few studies that has attempted to quantify elimination rates and/or rates of metabolic transformation of individual PCB congeners in fish was the investigation by Niimi and Oliver (13). Although the net trophic transfer efficiency of total PCBs to coho salmon (Oncorhynchus kisutch) and to lake trout (Salvelinus namaycush) from their prey has been estimated in previous studies using field data (2, 11, 14), the efficiencies for individual PCB congeners have yet to be estimated using field data. Some PCB congeners have been shown to be substantially more toxic than others (15, 16), so estimation of net trophic transfer efficiencies by PCB congener would be especially useful for risk assessment purposes. Furthermore, it has been suggested that gross trophic transfer efficiency of organic contaminants to piscivorous fish from their prey is a function of lipid affiliation, as measured by the octanol-water partition coefficient Kow (17, 18). Despite the high degree of variation in the gross trophic transfer efficiency estimates of organic contaminants to predator from prey, as presented by Thomann (18), there was a suggestion that these efficiencies peaked between log Kow values of 5 and 6, and there was a suggestion of a decline in these efficiencies as log Kow increased from 6 to 10. In contrast, Burreau et al. (19) concluded, based on a laboratory study using northern pike (Esox lucius) fed with rainbow trout (O. mykiss) contaminated with several different types of organochlorines, that gross trophic transfer efficiency was not influenced by log Kow as log Kow increased from about 5.5 to 8.5. Would net trophic transfer efficiencies, as calculated from actual field observations of PCB congener concentrations in fish, for various PCB congeners show any pattern as log Kow of the congener increased? The answer to this question has yet to be examined in detail. Our intent in this study was to determine if chemical properties, such as Kow and degree of chlorination, of PCB congeners have a detectable effect on the net trophic transfer efficiency of these congeners to coho salmon from their prey. More specifically, we estimated the net trophic transfer efficiency of the congener to coho salmon from their prey, using field data and bioenergetics modeling and then assessed the relationship between log Kow and net trophic transfer efficiency of the congener and the relationship between degree of chlorination and net trophic transfer efficiency of the congener.
Methods During 1994-1995, the Great Lakes National Program Office (GLNPO) of the U.S. Environmental Protection Agency sponsored the Lake Michigan Mass Balance (LMMB) project. One of the objectives of this study was to develop a model for flow of PCBs through the Lake Michigan ecosystem. Coho salmon was chosen to represent one of the top predators of the ecosystem (2). Diet and PCB concentration of coho salmon of various sizes and at numerous locations throughout Lake Michigan were investigated during April-November of 10.1021/es9903882 Not subject to U.S. copyright. Publ. 1999 Am. Chem.Soc. Published on Web 09/24/1999
TABLE 1. Logarithm of the Octanol-Water Partition Coefficient (log Kow) and Number of Chlorine Atoms per Molecule for the 21 PCB Congeners Selected for Our Studya no. of PCB chlorine atoms congener per molecule 66 74 85 99 101 105 107 114 128 151 158
4 4 5 5 5 5 5 5 6 6 6
log Kow 6.20 6.20 6.30 6.39 6.38 6.65 6.71 6.65 6.74 6.64 7.02
no. of PCB chlorine atoms congener per molecule 171 172 174 177 180 183 194 198 201 202
7 7 7 7 7 7 8 8 8 8
log Kow 7.11 7.33 7.11 7.08 7.36 7.20 7.80 7.62 7.62 7.24
a Values of log K ow were taken from Hawker and Connell (29). Numbering of congeners was according to the IUPAC numbering system of PCBs.
1994 and 1995. Additionally, PCB determinations were made on prey fish of various sizes and from several nearshore locations in Lake Michigan during spring, summer, and fall of the same time period. The five prey fish species chosen for the LMMB study included alewife (Alosa pseudoharengus), bloater (Coregonus hoyi), rainbow smelt (Osmerus mordax), slimy sculpin (Cottus cognatus), and deepwater sculpin (Myoxocephalus thompsoni). For both coho salmon and prey fish, the concentrations of 80 different PCB congeners were determined. Thus, data from the LMMB project were well suited for the calculation of net trophic transfer efficiency of various PCB congeners to coho salmon from their prey. In an earlier study, Madenjian et al. (2) estimated the net trophic transfer efficiency of total PCBs to coho salmon from their prey. In the present study, we applied the identical procedure used in the earlier study to estimate the net trophic transfer efficiency of 21 different PCB congeners to coho salmon from their prey. From the earlier study, PCB data were available from 67 coho salmon composites, 132 alewife composites, 73 rainbow smelt composites, 139 bloater composites, 69 slimy sculpin composites, and 74 deepwater sculpin composites. We chose the 21 PCB congeners that were among the most frequently detected in the prey and coho salmon composite samples. These 21 congeners were measured in detectable quantities in all but 7 composite samples, and these 7 composites were excluded from our estimation procedure. PCB congeners that coeluted with other PCB congeners were not considered as potential candidates for our study. The 21 congeners ranged in log Kow from 6.2 to 7.8, and the range in number of chlorine atoms per molecule was from 4 to 8 (Table 1). These 21 congeners included two tetrachloro congeners (congeners 66 and 74), six pentachloro congeners (congeners 85, 99, 101, 105, 107, and 114), three hexachloro congeners (congeners 128, 151, and 158), six heptachloro congeners (congeners 171, 172, 174, 177, 180, and 183), and four octachloro congeners (congeners 194, 198, 201, and 202). For details on the field sampling for the coho salmon and the prey fish, the processing of the coho salmon stomachs, and the analysis of the coho salmon diet data, refer to Madenjian et al. (2). Coho salmon diet in Lake Michigan was summarized by Madenjian et al. (2). During the second year in the lake, about 80% of the coho salmon diet is composed of alewives (2). For PCB determinations, coho salmon were composited into groups of 3-5 fish of the same age, same sex, similar size, and similar time and location of sampling (2). Prey fish were composited into groups of 5 fish by species, length, location, and time of sampling (2).
Concentrations of the 21 selected PCB congeners in each of the coho salmon composites and the prey fish composites were determined using gas chromatography/mass spectrometry, following the general procedure outlined by Schmidt and Hesselberg (20). Appropriate quality control measures (blanks, matrix spikes, surrogate standards, and duplicates) were used to ensure accuracy and precision of the analyses (21). Recovery tests for fish sample surrogates ranged from 70 to 120% (20). Matrix spike recoveries ranged from 75 to 110% (21). For each of the 21 congeners and for each species of fish (coho salmon and the five prey species), we used regression analysis to examine the relationship between concentration of the PCB congener and fish total length. As reported by Madenjian et al. (2) in their analysis of total PCB data, we found that a straight line fit was most appropriate for the prey species and an exponential fit was most appropriate for coho salmon. These regression equations were then used in estimating net trophic transfer efficiencies. Estimation of Net Trophic Transfer Efficiencies. To estimate the efficiency of net trophic transfer of a particular PCB congener from prey to coho salmon, we used the equation developed by Jackson and Schindler (11):
γ)
[PCBcoho]R
(1)
[PCBprey]
where γ ) net trophic transfer efficiency of the PCB congener from the prey to the predator, [PCBcoho] ) PCB congener concentration of coho salmon (ng/g), [PCBprey] ) PCB congener concentration of the predator’s food (ng/g), and R ) gross growth efficiency (gge) of the predator. Gross growth efficiency (gge) was equal to the weight gained by the predator divided by the amount of food eaten by the predator. Note that the PCB congener concentration of the predator’s food represented the average concentration in the food of the predator during the time period over which net trophic transfer efficiency was estimated. The main assumption for this estimation technique was that practically all of the PCBs taken up by the predator were from the diet, whereas direct uptake of PCBs from the water was negligible (11). For each of the 21 selected PCB congeners, we estimated γ for 24- and 29-month-old coho salmon. We then averaged these two estimates to generate an estimate of γ for a particular PCB congener. Averaging of net trophic transfer efficiencies across several ages has been used in previous studies to arrive at an overall net trophic transfer efficiency during the fish’s duration in the lake (2, 14). We also calculated the standard error associated with each of these 21 averages. To estimate γ for each PCB congener, we followed the same procedure detailed in Madenjian et al. (2). Estimates of gross growth efficiency (gge) of the coho salmon in Lake Michigan were taken from Madenjian et al. (2). These estimates were based on application of the coho salmon bioenergetics model developed by Stewart et al. (22) and modified by Stewart and Ibarra (23). Estimates of gge from salmon bioenergetics models are expected to be accurate to within 10% of the true gge, and the model estimates of gge are unbiased (24). See Madenjian et al. (2) for more details on application of the bioenergetics model. The PCB congener concentrations of coho salmon of ages 24 and 29 months were estimated by substituting average lengths at these ages into our regression equations of PCB congener concentration in coho salmon as a function of coho salmon total length. To estimate the PCB congener concentration of prey, the modal length of each prey fish category was substituted into the appropriate regression equation to calculate the PCB congener concentration of that prey category. Refer to Madenjian et al. (2) for more details on this procedure. VOL. 33, NO. 21, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3769
Data on PCB congener concentrations of aquatic invertebrate prey, including adult Mysis, adult Diporeia, and Bythotrephes cederstromi in Lake Michigan during the 19941995 period were provided by researchers at the University of Minnesota (D. Swackhamer and A. Trowbridge, unpublished data; see Supporting Information Table 1). The PCB congener determinations in the aquatic invertebrates were part of the LMMB study and, therefore, were subjected to quality assurance measures similar to those for the fish determinations. Data on the PCB congener concentrations of terrestrial insects, an important prey category for newly planted coho salmon in Lake Michigan, were not available, although data on total PCB concentration in terrestrial insects were available (2). Therefore, we assumed that, for each of the 21 congeners, the ratio of PCB congener concentration with total PCB concentration in benthic invertebrates from Lake Michigan was identical to that for terrestrial insects. The ratio of PCB congener concentration with total PCB concentration in benthic invertebrates was known from previous work (D. Swackhamer and A. Trowbridge, unpublished data), and the total PCB concentration in terrestrial insects was estimated at 70 ng/g (2). From these two pieces of information, the concentration of each of the 21 PCB congeners in terrestrial insects could be estimated. Over the course of 12-17 months of life in Lake Michigan, terrestrial insects compose only a very minor (