Environ. Sci. Technol. 2001, 35, 14-20
Biomagnification of DDT through the Benthic and Pelagic Food Webs of Lake Malawi, East Africa: Importance of Trophic Level and Carbon Source KAREN A. KIDD,* HARVEY A. BOOTSMA,† AND RAYMOND H. HESSLEIN Department of Fisheries and Oceans, Freshwater Institute, 501 University Crescent, Winnipeg, Manitoba, R3T 2N6 Canada DEREK C. G. MUIR AND ROBERT E. HECKY‡ Environment Canada, National Water Research Institute, 867 Lakeshore Road, P.O. Box 5050, Burlington, Ontario, L7R 4A6 Canada
Lake Malawi, an East African Rift Valley lake, is internationally renowned for having the highest diversity of fish species in the world, and these cichlids are highly specialized in their dietary habits. In this lake, tissue stable carbon (δ13C) and nitrogen (δ15N) isotopes can be used over several trophic levels to distinguish those consumers relying upon carbon fixed by either benthic or pelagic primary producers. As such, it was possible to contrast the biomagnification of persistent organochlorines through the benthic and pelagic food webs. In 1996 and 1997, foodweb organisms were collected from Lake Malawi and analyzed for organochlorines, δ13C and δ15N to determine the factors that affect the biomagnification of contaminants in a tropical lake. The pesticide DDT was the most predominant pollutant in the biota from Lake Malawi and was found at the highest concentrations in the largest and fattiest fish species. As observed in temperate systems, logtransformed ΣDDT concentrations in food-web organisms were significantly predicted by δ15N or log lipid (r 2 ) 0.32 and 0.40, respectively). In addition, the slope of the regression of log ΣDDT versus δ15N was significantly higher in the pelagic than the benthic food web. These results indicate that pelagic organisms are at greater risk of accumulating these pollutants than biota relying upon benthic primary production.
Introduction The most southerly of the East African Rift Valley lakes, Lake Malawi is internationally recognized for having the highest diversity of fish species in the world. Most of the fish present in the lake are from the family Cichlidae; these species have * Corresponding author phone: (204)983-5226; fax: (204)984-2404; e-mail:
[email protected]. † Current address: Great Lakes WATER Institute, University of WisconsinsMilwaukee, Milwaukee, WI 53204. ‡ Current address: Department of Biology, University of Waterloo, Waterloo, Ontario, Canada. 14 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
undergone dramatic evolutionary radiation in neurocranial morphology, and it is thought that this morphological diversity may allow them to have specific dietary habits and partition food resources and niches within the lake. Current estimates of the number of species of cichlids range from 700 to 1000 (1), with species densities being greatest in the nearshore rocky habitats. These nearshore areas of Lake Malawi are highly productive with C fixation rates of up to 1000 mg C m-2 day-1, and they can support more than 500 fish in a 50 m2 area (2, 3). In contrast, the offshore pelagic food web is less species-rich and is based on much lower areal rates of algal photosynthesis. Stable nitrogen and carbon isotope ratios of biota can be used to characterize an organism’s trophic position and carbon source, respectively. These isotopic signatures integrate dietary habits over a period of months to years for slower-growing species (4). The enrichment of the heavier isotope of nitrogen from prey to predator (3-5‰ (5)) provides a continuous variable with which to quantify and contrast the biomagnification of organochlorines in aquatic food webs (6, 7). Little change in isotope ratios is observed when carbon is incorporated into a consumer from its diet, and, for this reason, these ratios are used to trace carbon flow from primary producers to tertiary consumers (8). In temperate and arctic lakes, predicting the importance of benthic or pelagic carbon sources in the fate of organochlorines is confounded by the integration of isotopic signatures in upper-trophic-level biota. Despite distinctive δ13C signatures (differences of up to 20‰) at the base of these food webs, increasing omnivory in primary through tertiary consumers results in an integration of both benthic and pelagic production in top predators (8). In contrast, habitat and resource partitioning of fishes in Lake Malawi provides a unique opportunity to examine the importance of benthic versus pelagic carbon in the trophic transfer of organochlorines. Stable isotope analyses of cichlids and their prey from this lake indicate that the unique isotopic signatures of the primary producers are conserved through several trophic levels (8). Little overlap is observed in the isotopic signatures of species (3), which enables contrasts to be made between the food webs supported by these primary producers. The current understanding of factors affecting organochlorine biomagnification is based on studies done mainly on mid- to high-latitude systems. Organisms higher in lipid and trophic position, larger in size (6) and from lakes with longer underlying food webs (e.g. ref 9) are known to have higher tissue concentrations of organochlorines. Persistent pesticides continue to be used in East Africa for the protection of human health (e.g. DDT and dieldrin to control malaria and sleeping sickness) and to meet the increasing agricultural demands [e.g. DDT, hexachlorcyclohexane, heptachlor, and aldrin/dieldrin for cash crops]. It is estimated that Africa uses less than 5% of the world production of pesticides (10), but this may be changing due to increasing food production requirements. Pesticide import surveys for Malawi indicate that chlorinated pesticides such as aldrin and endosulfan continue to be used in the 1990s (11). Though high organochlorine concentrations have been observed in the top predators from East African Lakes (12, 13), little is known about the factors affecting concentrations of persistent organochlorines in these species-rich food webs. The objectives of this study were to determine the factors that affect DDT concentrations in organisms from Lake Malawi and to contrast the biomagnification of DDT through the food webs based upon pelagic and benthic carbon. 10.1021/es001119a CCC: $20.00
Published 2001 by the Am. Chem. Soc. Published on Web 11/17/2000
TABLE 1. Mean Weight, Sample Size, Percent Lipid and Concentrations (ng g-1 Weight Wet in Biota; pg L-1 in Water, (SD) of ΣDDT, p,p′-DDE, p,p′-DDT, and p,p′-DDD [and Ratio (p,p′-DDE + o,p′-DDE)/ΣDDT] in Fish, Invertebrates, Algae and Water Samples (15) from Lake Malawi (1996 and 1997) species
Bagrus meridionalis Buccochromis nototaenia Clarius sp. Ctenopharynx pictus Dimidiochromis kiwinge Diplotaxodon argenteus Docimodus evelynae Engraulicyprus sardella Genyochromis mento Labeotropheus fuelleborni Nimbochromis polystigma Opsaridium microlepis Oreochromis spp.a Petrotilapia sp. “fuscous” Protomelas taeniolatus Pseudotropheus sp. “tropheops” Pseudotropheus xanstomachus Rhamphochromis ferox Synodontis njassae Tyrannochromis macrostoma
common name
wt (g) (no. of fish)
percent lipid
Kampango 1298 ( 1244 (4) 472 (2) 133 (1) 10 ( 9.4 (5) 10 ( 9.0 (5) 48.5 (2) Mbuna 4.8 (1) Usipa 4.6 ( 2.1 (6) Mbuna 13 ( 1.7 (5) Mbuna 20 ( 12 (6) 28 ( 17 (3) Mpasa 491 ( 153 (7) Chambo 633 ( 208 (6) Mbuna 24.5 ( 11.1 (7) Mbuna 13 ( 2.8 (6) Mbuna 16 ( 6.7 (6) Mbuna
12 ( 6.8 (6)
Ncheni
516 (1) 85.8 ( 35.9 (5) 20.4 ( 10.7 (5)
ΣDDTb
p,p′-DDE
p,p′-DDT
p,p′-DDD
DDE/ΣDDTb
5.9 ( 4.5 1.7, 2.6 2.9 3.7 ( 2.2 1.8 ( 0.3 0.9, 1.7 0.7 3.3 ( 1.3 1.0 ( 0.7 3.3 ( 1.3 2.7 ( 1.1 9.1 ( 6.9 1.0 ( 1.0 2.1 ( 1.2 5.0 ( 1.8 3.6 ( 1.6
10.2 ( 11.7 3.4 1.4 2.0 ( 0.9 2.9 ( 1.3 3.7 1.1 5.2 ( 3.3 1.1 ( 0.97 1.2 ( 0.45 3.5 ( 0.20 34 ( 16 0.59 ( 0.38 1.1 ( 1.0 3.5 ( 1.1 1.3 ( 0.34
10.9 ( 9.5 1.92, 1.21 0.90 1.12 ( 0.69 1.93 ( 1.08 2.27, 1.13 0.67 2.52 ( 1.12 0.68 ( 0.55 0.49 ( 0.17 2.35 ( 0.15 27.4 ( 15.0 0.42 ( 0.28 0.46 ( 0.52 1.90 ( 0.51 0.62 ( 0.15
1.36 ( 1.18 1.77, 1.33 0.36 0.38 ( 0.29 0.63 ( 0.12 2.52, 0.81 0.11 1.39 ( 1.64 0.20 ( 0.31 0.43 ( 0.18 0.78 ( 0.05 1.38 ( 0.49 0.12 ( 0.09 0.39 ( 0.40 0.83 ( 0.19 0.41 ( 0.15
0.84 ( 0.68 0.27, 0.24 0.12 0.38 ( 0.30 0.35 ( 0.10 0.37, 0.22 0.29 0.62 ( 0.44 0.12 ( 0.14 0.14 ( 0.11 0.24 ( 0.21 1.73 ( 1.26 0.04 ( 0.02 0.15 ( 0.11 0.65 ( 0.66 0.23 ( 0.17
0.77 ( 0.11 0.54, 0.38 0.63 0.55 ( 0.13 0.63 ( 0.08 0.52, 0.44 0.62 0.56 ( 0.15 0.66 ( 0.20 0.45 ( 0.11 0.69 ( 0.05 0.72 (0.09 0.80 ( 0.12 0.42 ( 0.15 0.56 ( 0.06 0.48 ( 0.10
2.3 ( 1.4
1.9 ( 1.3
0.75 ( 0.25 0.37 ( 0.17 0.77 ( 1.07 0.46 ( 0.13
2.2 12.8 ( 3.7 2.0 ( 0.7
8.3 58 ( 39 3.4 ( 3.3
2.88 4.80 0.43 0.35 47.2 ( 30.0 7.74 ( 6.18 1.59 ( 0.88 0.83 ( 0.02 2.26 ( 2.32 0.83 ( 0.74 0.25 ( 0.24 0.65 ( 0.06
Fish
Invertebrates and Algae epilithic algae Chaoborus edulis crab mussels snail sp. A snail sp. B trichopteran zooplankton blanks (ng‚g-1) waterc (pg‚L-1)
(2) (2) (1) (3) (1) (1) (1) (6) (7) (4)
0.15, 0.13 5.43, 4.83
0.17, 0.08 2.96, 1.56 0.23 1.51 ( 0.34 0.12 ( 0.07 0.70 0.69 0.52 0.41 6.70 2.35 0.44 ( 0.05 0.13 ( 0.01 0.06 ( 0.05 28 ( 38
0.02, 0.01 0.97, 0.27 0.16 0.26 ( 0.17 0.53 0.33 1.15 0.03 ( 0.01 0.03 ( 0.01 21 ( 42
0.08, 0.01 1.53, 1.01 0.02 0.16 ( 0.14 0.04 0.01 0.62 0.04 ( 0.01 0.02 ( 0.03 5.2 ( 6.2
0.01, 0.01 0.40, 0.26 0.01 0.10 ( 0.06 0.05 0.02 0.24 0.01 (0.01 0.00 1.5 ( 2.9
0.18, 0.17 0.33, 0.18 0.78 0.45 ( 0.05 0.84 0.86 0.51 0.27 (0.08 0.76 ( 0.47
O. lidole and squamipinnis. b ΣDDT includes both o,p′- and p,p′-isomers. Concentrations of o,p′DDE and o,p′-DDD were generally much lower than the corresponding p,p′- isomers and are therefore not shown. DDE/ΣDDT includes both isomers. c Large volume (70-100 L) filtered (GFF) samples extracted with XAD-2 resin. Average (blank corrected) for samples collected in February 1996, 1997 and 1998 from surface waters of the south basin (15). a
Experimental Section Lake Malawi is the most southerly of the East African Rift Valley lakes and is located between 9° and 14° S. It is a meromictic, oligotrophic lake that has a surface area of 28 800 km2, a maximum depth of 700 m and a flushing time of 750 years (8, 14). In May to September of 1996 and in May of 1997, openwater fish species were purchased from local fishermen or from a commercial fishery operating near Monkey Bay, Malawi. In May 1997 SCUBA was used to collect rock-dwelling fish and benthic invertebrates from the Maleri Islands (13°54.376′ S, 34°37.480′ E) in the southwestern part of the lake. Most fish species were from the Family Cichlidae (16 of 21 species) but several species from the Families Clariidae (Clarias gariepinus), Bagridae (Bagrus meridionalis), Mochokidae (Synodontis njassae) and Cyprinidae (Engraulicypris sardella, Opsaridium microlepis) were also collected. All fish were identified to species when possible (see Table 1 for complete list and n of species), weighed and measured. No aging structures were collected because of the lack of seasonal influences on these tissues in tropical systems. Most fish were frozen whole in Whirlpaks; however, muscle fillets (including skin) for the larger species (B. meridionalis, Oreochromis squamipinnis, O. lidole, O. microlepis, Buccochromis nototaenia, Rhamphochromis ferox, Clarius sp.) were removed, wrapped in solvent-rinsed aluminum foil and then frozen. Snails, trichoptera, mussels, crab and epilithic algae (rock scrapings) were collected from the nearshore areas, pooled by major taxa, removed from shells or cases if present,
put into precleaned glass jars in polyethylene (Whirlpaks) bags, and frozen shortly after collection. Zooplankton samples were collected using vertical tows or trolling of a 71 µm Nitex net in May of 1996 and 1997. Subsamples of the zooplankton indicate that these samples were composed primarily of cyclopoids (62-81%), and calanoids (9-14%) and cladocerans (5-29%) to a lesser extent. Adult lakeflies, Chaoborus edulis, were collected during emergences on an opportunistic basis. A parallel study, conducted at the same time as biota sampling, collected lake surface water samples (70-100 L, GFF filtered) at 1 m depth in the southern basin of Lake Malawi (15). To assess the isotopic signatures at the base of the food web, epilithic algae scraped from rocks at 2, 5, and 10 m depth at the Maleri Islands were collected several times over a year and filtered onto GF/C filters. Additional bulk samples were collected for organochlorine analyses by scraping algae into solvent-rinsed glass jars. Settled particulate matter was also collected from sediment traps set at 100 m depth to characterize the basal pelagic isotopic signatures. Stable Isotope Analysis. Fish muscle (except for whole E. sardella), foot tissue from individual mussels, shelled pooled whole snails and trichoptera, and whole Chaoborus and zooplankton were analyzed for δ13C and δ15N using the automated methods and quality assurance procedures detailed in Kidd et al. (6). Invertebrate and algal samples were fumed with concentrated HCl prior to analysis to remove surficial carbonates. VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
15
Carbon isotope signatures at the base of the food web were used to determine the relative importance of benthic and pelagic autochthonous carbon in the diets of the biota. Epilithic particulate material collected between 2 and 10 m had an annual mean (( SD) δ13C of -8.32 ( 4.98‰ (n)72, range -13.36 to -4.15‰) and δ15N of -0.87 ( 1.35 ‰ (-3.06 to 1.89‰; n)59). POM collected monthly over eight months in sediment traps deployed at 100 m had a mean δ13C of -26.08 ( 1.28‰ (n)10) and a mean δ15N of 0.04 ( 1.51 (3.84 to -2.05‰, n)38) (H. Bootsma, unpublished data). Zooplankton samples had mean δ13C and δ15N values of -24.5 ( 1.02 and 3.18 ( 0.52 (n)6), respectively. Mean carbon isotope signatures of the zooplankton (-24.5‰) and epilithic algae (-8.32‰) were used as endpoints to calculate the relative contributions of pelagic versus benthic carbon in the diets of biota using a mixing model similar to that described in Hobson et al. (16). Zooplankton and not sediment trap material was chosen as the pelagic endpoint because largerbodied organisms are less seasonally variable in their isotopic composition than smaller-bodied organisms (17). The proportion of pelagic carbon in fishes and invertebrates was calculated as follows:
Ppelagic C ) (δ13Corganism- δ13Cbenthic algae)/ (δ13Czooplankton - δ13Cbenthic algae) Contrasts of organochlorine biomagnification between the pelagic and nearshore benthic food webs were made with organisms with Ppelagic C greater than 70% (n ) 44 of 105; δ13C values more depleted than -19.7‰) or less than 30% (n)34 of 105; δ13C values more enriched than -13‰), respectively. In addition, δ15N values were adjusted to account for isotopic differences at the base of these food webs (see algal data above). δ15N of biota from the benthic food web were increased by 0.87‰, whereas δ15N values of pelagic biota were decreased by 0.04‰ such that the base of both food webs corresponded to a mean value of 0‰. Organochlorine Analysis. Most fish were analyzed whole for persistent organochlorines, but only muscle (including skin) was analyzed for the larger fish (see above). No comparison between whole fish and muscle concentrations was done. E. sardella were pooled by date into six size classes of 4-12 fish each that ranged in mean weight from 2.4 to 8.2 g. Prior to extracting the samples, fish tissues were homogenized with dry ice, and invertebrates were freeze-dried. Subsamples of invertebrates were dried to constant weight to determine % H2O. Algal organochlorine measurements were converted to wet weight assuming 80% water (18). In brief, tissues were homogenized with 20-40 g of anhydrous Na2SO4 (heated at 600 °C for 16 h prior to use), and internal standards of PCB30 and octachloronaphthalene (OCN) were added to the samples prior to extraction. Samples were then extracted using pressurized fluid extraction (ASE, Dionex Instruments; 2000 psi; 100 °C) and dichloromethane (DCM)/hexane (1:1). Percent lipid in these tissues was determined gravimetrically using one tenth of the extract. Lipids were removed using automated gel permeation chromatography. The extracts were then fractionated on Florisil to separate most DDT related compounds (except p,p′-DDE) from PCBs. Extracts were analyzed by GC-ECD using a 60 m × 0.25 mm id DB-5 column as described in Kidd et al. (6) and Karlsson et al. (15). Cod liver oil SRM 1588 (National Institute of Standards and Technology) or CRM-349 (BCR, Brussels, Belgium) was used as reference materials. Reagent blanks were run with every 15-20 samples, and tissue concentrations were corrected using the mean blank values of individual congeners. Recoveries for internal standards averaged (( SD) 80 16
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
( 12 and 77 ( 15% for PCB30 and OCN, respectively. Minimum detectable levels (MDL) were calculated as described in Long and Winefordner (19) and averaged 0.02 ng g-1 for most OC pesticides. Results from a subset of samples were confirmed using GC-MS analysis. Data Analysis. For results below the MDL, random numbers were generated between 0.01 and 0.0001 for individual congeners; all data were included in statistical analysis. DDT concentrations and percent lipid were logtransformed because standard deviations were proportional to means and transformation reduced the heteroscedacity of the data. Stepwise multiple regression analysis was conducted using the general linear model in SAS (20) to examine whether lipid (wet weight), trophic position (δ15N) and carbon source (δ13C) were significant predictors of DDT concentrations (wet weight) in these biota. Coefficients of partial determination and comparisons between linear and multiple regression coefficients were used to assess multicollinearity of these independent variables. Factors were retained in the model if they explained a significant proportion of the variability (p < 0.05) over and above all other factors (p < 0.05 of Type III SS). Residuals of models were examined to ensure normality. Specific comparisons of DDT biomagnification in the pelagic and benthic food webs (see above for delineations) were conducted using Analysis of Covariance (ANCOVA). Concentrations of organochlorines were adjusted for the covariate lipid using the common regression equation, and residuals after it determined that the DDT-lipid slopes were not significantly different between the pelagic and benthic food webs (p ) 0.49, this study (21, 22)). Lipid-adjusted organochlorine concentrations and lipid were not significantly related (p ) 0.28). Relationships between lipid-adjusted DDT concentrations and δ15N or δ13C were then examined through the benthic and pelagic food webs using ANCOVA.
Results and Discussion Food web interrelationships of the biota from Lake Malawi are shown in Figure 1. These biota represented several trophic levels (range in mean δ15N 1.8-9.1‰), and the range in carbon at the base of the food web (-26.6 to -8.3‰ for the pelagic and benthic primary producers) was conserved over several trophic levels, as has been observed previously in this lake (8). The fish ranged in dietary habits from benthic herbivores (e.g. L. fuelleborni with mean δ15N of 4.4‰, epilithic algae ranging from -3.0 to 1.7‰) to zooplanktivores (e.g. E. sardella with mean δ15N of 5.82‰, zooplankton with mean δ15N of 3.18‰) and zoobenthivores (e.g. P. taeniolatus with mean δ15N of 5.6‰ consistent with a diet of snails, δ15N of 2.2‰) to piscivores (e.g. the pelagic fish Clarius and Rhamphochromis with δ15N of 9.06 and 8.67‰, respectively). Fewer top predators were collected from the benthic than pelagic food web but included N. polystigma and T. macrostoma (mean δ15N values of 7.2 and 7.6‰, respectively). Stable isotope results (Figure 1) were generally consistent with known dietary habits of these species or limited gut content analyses that were conducted as part of this study. Ramphochromis spp., B. meridionalis, Buccochromis spp., and Opsaridium spp. are pelagic piscivores with mean δ15N ranging from 6.97 to 8.65‰ and tend to feed mainly upon small cichlids or Engraulicyprus sardella (mean δ15N of 5.83‰ for whole fish, muscle was an average of 0.5‰ more enriched in δ15N than whole bodies, n ) 6 (23), this study). However, the stable isotope data suggests that Opsaridium spp. (δ15N of 6.97) relies more heavily upon zooplankton or Chaoborus (mean δ15N of 3.19 and 3.58‰, respectively). The stable isotope and gut content analyses concur that E. sardella are mainly zooplanktivores; δ15N increased from the smaller (5.22‰) to larger (6.33‰) fish supporting a shift in diet from
predominant group (average of 97, 85 and 84% of sum of p,p′+ o,p′-DDT, -DDE or -DDD). Therefore we have not tabulated levels of o,p′-DDE, o,p′-DDE, and o,p′-DDD. ΣDDT was also the most predominant pesticide found in biota from Lake Kariba, Zimbabwe (12), likely due to its regional use to control mosquitoes and tse tse flies. In relation to other studies in East Africa, concentrations of ΣDDT in fish from Lake Malawi were much lower than in comparable fish species from Lakes Kariba (12) and McIlwaine (13) in Zimbabwe. For example, catfish of similar sizes (500-1000 g) from Lakes McIlwaine (Clarias gariepinus (13)) and Malawi (B. meridionalis) contained 63 and 10.2 ng ΣDDT.g-1 wet weight, respectively. These geographic differences suggest that DDT use in the riparian countries around Lake Malawi is lower than in other areas of East Africa or that the large size of the lake acts to dilute DDT and reduce biotic exposure.
FIGURE 1. Mean δ13C and δ15N ((SD) of algae, invertebrates and fish from Lake Malawi, East Africa. BM - Bagrus meridionalis, BN Buccochromis nototaenia, Ch - C. edulis (lakeflies), CL - Clarias gariepinus, CP - Ctenopharynx pictus, Cr - crab, DA - Diplotaxodon argenteus, DE - Docimodus evelynae, DK - Dimidiochromis kiwinge, Ea - epilithic algae, ES - Engraulicypris sardella, GM - Genyochromis mento, LF - Labeotropheous fuelleborni, Mu mussels, NP - Nimbochromis polystigma, OL - Oreochromis lidole, OP - Opsaridium microlepis, OS - Oreochromis squamipinnis, PF - Petrotilapia fuscous, POM - sediment trap material, PS Pseudotropheus sp. “tropheops type”, PT - Protomelas taeniolatus, PX Pseudotropheus xanstomachus, RF - Rhamphochromis ferox, SnA - snail, SnB - snail, SY - Synodontis njassae, TM Trematocranus microstoma, Tr - trichopteran, Zo - zooplankton. some phytoplankton (0.04‰) to mostly zooplankton (3.19‰) with increasing size (this study; 24). Tyrannochromis macrostoma is piscivorous and feeds upon rock-dwelling cichlids (23), which is supported by its more enriched δ13C (-12.86‰) and depleted δ15N (7.23‰) when compared to pelagic piscivorous fishes (Figure 1). Oreochromis spp. consume phytoplankton as well as copepod zooplankton (25, 26, this study); their mean δ15N (5.33 and 4.88‰ for O. lidole and O. squamipinnis, respectively) was consistent with this type of diet, but their δ13C (-19.68 and -17.32 ‰, respectively) was more enriched than would be expected from a strict planktivorous diet. D. kiwinge (mean δ13C and δ15N of -21.66 and 6.44‰, respectively) feeds upon small surface-dwelling fishes (23) which may include the herbivorous larval stages of E. sardella (24) that would be lighter in δ15N than the fish analyzed in this study. L. fuelleborni, P. taeniolatus, Pseudotropheus spp. and Petrotilapia spp. scrape algae off of rocks in the nearshore area (2, 23). Though their δ13C values (-6.27, -8.83, -3.99, -4.32‰, respectively) are consistent with a diet of epilithic algae (δ13C range of -4.15 to -13.36, mean δ15N of -0.87‰), the δ15N values for these species (4.64, 5.61, 4.34, and 4.32‰, respectively) are slightly more enriched than would be expected from this diet and suggests some reliance on heavier sources of N (such as benthic invertebrates). Some fish with unusual dietary habits were also analyzed: Genyochromis mento and Docimodus evelynae are fin and scale eaters (1); no such tissues were analyzed for stable isotopes for comparison. The pesticide DDT was the most predominant pesticide (up to 60 times higher in concentration than other organochlorine pesticidessthese data not shown) in the biota from Lake Malawi (Table 1) and was generally highest in piscivorous fish (such as S. njassae) and lowest in fish species that fed directly on epilithic algae (e.g. L. fuelleborni). The p,p′ isomers of DDT, DDE or DDD were always the
Concentrations of ΣDDT in the top predators from Lake Malawi were similar to what has been found in piscivorous fish from some temperate and arctic lakes. For example, lake trout from Lake Superior (7) and remote Peter Lake, Northwest Territories (22) had mean p,p′-DDE concentrations of 52 (9.1% lipid) and 35 (3.6% lipid) ng g-1 wet weight, respectively. Species of similar trophic positioning and percent lipid from Lake Malawi had p,p′-DDE concentrations of 47.2, 27.7 and 10.9 ng g-1 wet weight (S. njassae, O. microlepis and B. meridionalis, respectively; Table 1). In contrast, concentrations of p,p′-DDE in trout from Lake Ontario were considerably higher (1159 ng g-1 wet weight) than the other lakes and may be partially explained by the higher lipid (16%) in these individuals. In cold-water lakes, fish with higher concentrations of persistent organochlorines tend to be the fattier, older and larger individuals within a species and in a food web (6, 9, 22). In Lake Malawi, ΣDDT concentrations were highest in the fattiest species (S. njassae and O. microlepis; Table 1) and higher in the larger (and presumably older) individuals within several species (log ΣDDT vs log weight, p < 0.05, r 2 ) 0.57 to 0.98; B. meridionalis, Oreochromis spp., E. sardella, Opsaridium spp., L. fuelleborni, P. sp. “tropheops”). These results suggest that some of the same factors that affect organochlorine concentrations in fish from high-latitude lakes are also important in tropical ecosystems. Organochlorine concentrations are significantly related to tissue δ15N and lipid content of biota from the Laurentian Great Lakes (7, 28) and from lakes in the Canadian Arctic (22, 27). We also found that δ15N was a significant predictor of ΣDDT concentrations through the entire food web of Lake Malawi.
log ΣDDT (ng g-1 ww) ) 0.20((0.03) δ15N 0.87 ((0.18), r 2 ) 0.32, p ) 0.0001 log p,p′-DDE (ng g-1 ww) ) 0.26((0.03) δ15N 1.49 ((0.18), r 2 ) 0.43, p ) 0.0001 These results indicate that δ15N can also be used to contrast the biomagnification of persistent contaminants across lakes that differ considerably in climatic regimes. Indeed, the slope of the relationship between p,p′-DDE and δ15N in Lake Malawi was similar in magnitude to what has been observed in temperate and arctic lakes (e.g. Lake Baikal slope 0.25, Lake Laberge, Yukon slope 0.27) and higher than was found for Lake Ontario (slope 0.15 (29)). ΣDDT and p,p′-DDE concentrations were also significantly related to percent lipid (Figure 2) and, for the former, this variable explained more of the variance in contaminant concentrations than did trophic position. VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
17
FIGURE 2. Relationship between log ΣDDT (ng g-1 wet weight) and log lipid in biota from Lake Malawi (see Figure 1 for legend).
log ΣDDT (ng g-1 ww) ) 0.71 ((0.09) log lipid 0.07 ((0.06), r 2 ) 0.40, p ) 0.0001 log p,p′-DDE (ng g-1 ww) ) 0.67 ((0.10) log lipid 0.19 ((0.07), r 2 ) 0.32, p ) 0.0001 Similar results were found in Lake Superior, where lipid was a better predictor of PCB concentration (r 2 ) 0.81) than was δ15N (r 2 ) 0.52) (7). In addition to lipid and δ15N, δ13C was a significant (p ) 0.04) but poor (r 2 ) 0.04) predictor of ΣDDT concentrations through the entire Lake Malawi food web, with the benthic biota tending to have lower contaminant concentrations than pelagic biota. In some (7, 28) but not all (22) freshwater food webs, percent lipid increases with increasing δ15N of organisms. The relative importance of lipid and δ15N to the food-web transfer of organochlorines is difficult to assess given the collinearity of these variables. However, organochlorine concentrations adjusted or normalized for the effects of lipids remain significantly related to biotic δ15N (e.g. ref 22), indicating that trophic position has a significant effect over and above that due to lipid alone. We assessed the importance of δ15N, δ13C and log-lipid on ΣDDT concentrations in the Lake Malawi food web using multiple regression analyses. Results indicated that lipid, δ15N and δ13C in combination were significant predictors of ΣDDT concentrations in these biota, explaining 64% of the variability in these data. The residuals from the model relating ΣDDT concentrations to lipid and δ13C were significantly related to δ15N (see the following equation), indicating that trophic position remained a significant predictor of pesticide concentrations after variability associated with lipid and carbon had been accounted for.
log ΣDDTresiduals of lipid & δ13C(ng g-1 ww) )
0.14 ((0.02) δ15N - 0.85 ((0.13), r 2 ) 0.31, p ) 0.0001 Little change in regression coefficients was observed from simple to multiple regression analyses (0.19-0.16 for δ15N; 0.71-0.64 for log-lipid), suggesting that these variables did not exhibit significant collinearity in the Lake Malawi food web. Coefficients of partial determination indicated that loglipid, δ15N, and δ13C explained 45, 33 and 8% of the variance in the regression, respectively, over and above that explained by the other variables. These results also demonstrate that lipid is a more important predictor of organochlorine 18
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
FIGURE 3. Relationships of (a) log ΣDDT (ng g-1 wet weight) versus δ15N and (b) lipid-adjusted log ΣDDT versus δ15N through the benthic and pelagic food webs of Lake Malawi (see Experimental Section for description of how food webs were distinguished). concentrations in the entire food web than either trophic positioning or carbon source. DDT was not the predominant OC pesticide in filtered lake water, where the more soluble HCH isomers as well as chlordane related compounds (heptachlor, and cis-/transchlordane) predominated (15). The ratios of the main derivative of DDT, DDE, to its parent congeners are used to determine the age of the pesticide accumulating in aquatic biota. A higher ratio of DDE/(DDD+DDE+DDT (o,p′+p,p′isomers)) is indicative of an older source of DDT to those organisms. With few exceptions (e.g. Ramphochromis ferox), the ratios of DDE/ΣDDT tended to be higher in those fishes feeding upon pelagic carbon (e.g Oreochromis spp. 0.80) when compared to a benthivorous species at a similar trophic level (e.g. Labeotropheus fuelleborni 0.45). This difference may be due to a preferential accumulation of DDE through the pelagic food web or to an older source of DDT at the base of the pelagic food web. In lake water p,p′-DDE was also the predominant DDT related isomer (Table 1), although concentrations ranged from
