Model-based exploration of the variability in lake trout (Salvelinus namaycush) bioaccumulation factors: The influence of physiology and trophic relationships
Abstract
Because dietary consumption of fish is often a major vector of human exposure to persistent organic pollutants (POPs), much effort is directed toward a quantitative understanding of fish bioaccumulation using mechanistic models. However, many such models fail to explicitly consider how uptake and loss rate constants relate to fish physiology. We calculated the bioaccumulation factors (BAFs) of hypothetical POPs, with octanol–water partition coefficients ranging from 104.5 to 108.5, in lake trout (Salvelinus namaycush) with a food-web bioaccumulation model that uses bioenergetics to ensure that physiological parameters applied to a species are internally consistent. We modeled fish in 6 Canadian lakes (Great Slave Lake, Lake Ontario, Source Lake, Happy Isle Lake, Lake Opeongo, and Lake Memphremagog) to identify the factors that cause the BAFs of differently sized lake trout to vary between and within lakes. When comparing differently sized lake trout within a lake, larger fish tend to have the highest BAF because they allocate less energy toward growth than smaller fish and have higher activity levels. When comparing fish from different lakes, the model finds that diet composition and prey energy density become important in determining the BAF, in addition to activity and the amount of total energy allocated to growth. Environ Toxicol Chem 2019;38:831–840. © 2019 SETAC
INTRODUCTION
Because fish consumption is one of the primary pathways for human exposure to persistent organic pollutants (POPs; Dougherty et al. 2000; Darnerud et al. 2006; Gasull et al. 2011), there has been extensive work for many decades to understand the bioaccumulation of POPs in fish (Gobas et al. 1988; Rasmussen et al. 1990; Gobas 1993; Bentzen et al. 1996; Luk and Brockway 1997; Fisk et al. 1998; Mackay and Fraser 2000; Arnot and Gobas 2004; Gewurtz et al. 2006; LeBlanc et al. 2006; Barber 2008; Burtnyk et al. 2009; Powell et al. 2009; Ng and Gray 2011; De Hoop et al. 2013). An important element of these efforts is the development and application of models that mechanistically describe contamination in fish as resulting from the balance of uptake and elimination processes (e.g., Clark and Mackay 1991; Arnot and Gobas 2004; Gewurtz et al. 2006; Burtnyk et al. 2009; Powell et al. 2009; De Hoop et al. 2013). The physiologically determined rates of respiration, consumption, and growth are key parameters in model parameterization; yet not all models explicitly account for the interdependence between them and may rely on physiological rates that are inconsistent with each other (Quinn et al. 2019). Although fortuitous cancellation of errors can still lead to good model performance, integrating bioenergetics considerations into the parameterization of contaminant bioaccumulation models can aid in identifying and reducing such errors (Barber 2008). Norstrom et al. (1976) presented one of the first bioenergetic-bioaccumulation models for polychlorinated biphenyls (PCBs) and methylmercury in yellow perch (Perca flavescens) in the late 1970s. Kitchell et al. (1974, 1977) developed a simplified bioenergetics equation, including a novel method to account for fish activity, which has become the basis for other bioenergetic-bioaccumulation models (e.g., Hanson et al. 1997; Trudel and Rasmussen 2006; Quinn et al. 2019). Luk and Brockway (1997) later used combined models to predict levels of 2 PCB congeners in Lake Ontario lake trout. Other researchers have used PCB elimination and dietary uptake kinetics to derive bioenergetic data in lake trout (Borgmann and Whittle 1992; Paterson et al. 2006, 2005).
Fish from different lakes can vary widely in their contamination with POPs, even if they belong to the same species (Rasmussen et al. 1990; Madenjian et al. 1994; Bentzen et al. 1996; Guildford et al. 2008; Ontario Ministry of the Environment and Climate Change 2017); contamination levels can also vary substantially for one fish species within a lake (Paterson et al. 2016). Although differing concentrations of POPs in water are partly responsible, differences can also be attributable to how effectively fish bioaccumulate POPs, which can be expressed quantitatively by differences in the bioaccumulation factor (BAF). Although some differences in bioaccumulation potential must be rooted in differences in trophic relationships (Rasmussen et al. 1990; Bentzen et al. 1996), Houde et al. (2008) showed that correcting for differences in trophic and lipid levels between lake trout populations does not totally account for variations in PCBs and organochlorine pesticide concentrations. Accounting then for the differences in physiology of fish in different lakes may prove to be fruitful.
Bioenergetically balanced fish bioaccumulation models have been used to assess the factors driving differences in bioaccumulation potential between male and female fish (Ng and Gray 2011; Madenjian et al. 2013), across (Luk 2000) and within (Paterson et al. 2016) species of fish. Paterson et al. (2016) attributed variations in contaminant bioaccumulation in Lake Huron trout to differences in growth rate and prey abundance. Gewurtz et al. (2006) compared the bioaccumulation potential of Arctic and temperate fish species by correcting individual physiological rates for temperature but did not assure that parameterization was bioenergetically consistent. Higher predicted concentrations in temperate lake trout than Arctic char were attributed largely to differences in lipid content (Gewurtz et al. 2006), rather than temperature. Ng and Gray (2011) used a bioenergetically balanced bioaccumulation model approach to predict how climate change might affect contaminant cycling and PCB bioaccumulation in Lakes Superior and Erie. Using this model approach allowed them to consider some of the indirect effects of changing temperatures on fish, including prey availability, foraging costs, and changes to consumption and respiration rates. So far, no model-based study appears to have focused on how the bioaccumulation potential of chemicals changes with respect to fish size in different lakes.
Bioaccumulation models require fish growth to be specified. Many bioaccumulation models (e.g., Arnot and Gobas 2004; Gewurtz et al. 2006; McLeod et al. 2016) use an allometric equation developed by Thomann (1981) that relates growth to fish weight and temperature. Because this equation was originally derived from measuring the growth of particles in the ocean (Sheldon et al. 1973), there is no reason to expect that it is adequate for predicting the growth of different fish species in freshwater lakes. Multiple studies have reported weight-at-age data, which display differences in growth between and within species (Jonsson et al. 1988; Rubin 1993; Shimose et al. 2009; Loewen et al. 2010; Hansen et al. 2016). Using the Thomann equation to predict growth carries the implicit assumption that all fish species, regardless of their maximum growth potential, grow at the same rate (for a given weight and temperature). Species-specific physiological parameters are necessary to improve understanding of the differences in the bioaccumulation potential between and within species.
We present a version of the aquatic food-web bioaccumulation model AQUAWEB (Arnot and Gobas 2004) that has been modified to assure that the physiological rates for all fish are plausible and internally consistent. This is achieved by solving a bioenergy balance for each organism (Quinn et al. 2019). We then apply this model to describe bioaccumulation in lake trout, Salvelinus namaycush, also known as lake char, from 6 Canadian lakes. We can do this because the trophic relationships in those 6 lakes are known (Halfon et al. 1996; Pazzia et al. 2002; Arnot and Gobas 2004), and Pazzia et al. (2002) also have established the bioenergetics data for those 6 lake trout populations. Specifically, we use the modified AQUAWEB model to predict BAFs of hypothetical persistent chemicals with a log octanol–water partitioning coefficient (KOW) between 4.5 and 8.5 in lake trout. The goal of the present study was to quantify the extent of BAF variability that can be expected for a single fish species and identify the factors that are responsible for such variability. The present study thus is placed within the context of a larger effort to recognize organisms and populations with high exposure susceptibility, that is, to identify those that are particularly vulnerable to achieving high contaminant concentrations (Undeman et al. 2010).
METHODS
Bioaccumulation factors were calculated for hypothetical neutral chemicals in lake trout from 6 Canadian lakes that vary widely in size: Lake Ontario, Great Slave Lake, Lake Memphremagog, Lake Opeongo, Source Lake, and Happy Isle Lake (see Supplemental Data, Section S1, Figure S3). Although the focus of the present study is on lake trout, bioenergetics-based calculations were conducted for all organisms in the different food webs. The hypothetical neutral chemicals were intended to mimic typical POPs, and therefore, the logarithm of the KOW ranged from 4.5 to 8.5 at 25 °C (quarter log unit increments). The KOW was corrected for the temperature of each lake using the van't Hoff equation (Li et al. 2003) and an energy of phase change (ΔUOW) of –26.3 kJ mol−1, which is the average of values reported for different PCBs by Li et al. (2003). Biotransformation was also assumed to be negligible in all organisms, and the corresponding rate constant (km, see below) was set to zero.
(1)
(2)
(3)The user has 2 options when applying the bioenergetics version of the AQUAWEB model: 1) specify the feeding rate and calculate an “energetically” consistent growth rate, or 2) specify the growth rate and calculate an “energetically” consistent feeding rate. For the present simulations, we specify the growth rates for lake trout and prey species (Supplemental Data, Section S2) and calculate the corresponding feeding rate. Lake-specific allometric equations for lake trout activity multipliers (Act) as a function of fish weight (kilograms; Supplemental Data, Table S6) are also calculated. Differences in environmental and biological conditions across the lakes are thus accounted for to a great extent without explicitly specifying all factors. This is because fish grow in response to the abiotic and biotic conditions they encounter in their environment (Paterson et al. 2016; van Poorten and Walters 2016) and grow when they have energy available to do so. Complete details of the parameterization of the modified AQUAWEB model are provided in Supplemental Data, Sections S2 and S3, in particular with respect to the parameterization of 1) the modeled lakes, 2) the bioenergetics model, and 3) the contaminant bioaccumulation model. The model is implemented as a Microsoft Excel spreadsheet.
A preliminary evaluation of model performance was also conducted using data for PCBs in Lake Ontario, Lake Opeongo, and Great Slave Lake (Oliver and Niimi 1988; M. Evans, Environment and Climate Change Canada, Saskatoon, SK, Canada, personal communication; D. Muir, Environment and Climate Change Canada, Burlington, ON, Canada, personal communication). The results of this evaluation are reported in Section S4 of the Supplemental Data. It is important to note that the model is not calibrated using these field observations, and the exercise was conducted to determine whether the model could produce reasonable results.
We used the bioenergetic-bioaccumulation model to assess the relative differences in the bioaccumulation potential of lake trout in 6 lakes. Four differently sized fish weighing 100 g (S Trout), 500 g (M Trout), 1 kg (L Trout), and 3 kg (XL Trout) were modeled for each lake, except in Source and Happy Isle Lakes where the maximum size of lake trout found is approximately 500 g. The diet composition of each fish modeled is dependent on both its size and its lake. We do not model dynamically the fish's entire life, but we assume steady-state conditions at each of the 4 sizes. We note that the time to steady state is generally quite short relative to the life span of the fish, with high KOW compounds in XL trout being the only exception. Furthermore, any potential bias attributable to the steady-state nature of the model calculations is the same across all lakes (with same-size fish) and thus has little impact on the conclusions. The use of the steady-state assumption also means that seasonal changes, for example, in terms of water temperature or fish lipid content, are not taken into account.
RESULTS
The BAF values, as a function of log KOW and the energy balance, obtained by the AQUAWEB model for lake trout populations are summarized in Figure 2 for Lakes Ontario, Memphremagog, Great Slave, and Opeongo and in Figure 3 for Happy Isle and Source Lakes. Figures 2A and C and 3A and C display calculated BAF values as a function of log KOW for the lake trout and its prey species, respectively. They all show a bell-shaped profile, with maximum BAFs at approximately log KOW of 6.5 and much lower BAFs for compounds with log KOW below 5 and above 8 (Pitt et al. 2017). Chemicals with a log KOW below 5 tend not to bioaccumulate greatly in fish because respiration is an effective elimination pathway for these relatively water-soluble substances. With increasing log KOW the gut uptake efficiency (Ed) declines and sorption to dissolved organic matter in the water increases, causing bioavailability of the chemical (fbioavailable) to decline: when log KOW is less than 5.5, Ed and fbioavailable remain unchanged with KOW; however, when the log KOW of a chemical is greater than 5.5, there is a steady decline in Ed and fbioavailable (see Figure S2 and Figure S10 in Supplemental Data Sections S1 and S3). Highest BAFs occur for intermediate log KOW values (6–7) where efficient dietary uptake combines with slow excretion (Fisk et al. 1998).
Figures 2B and 3B show how each lake trout population allocates the energy obtained from its diet. Metabolic costs, in green, comprise the energy the fish uses on respiration and take into account the activity levels; in blue is the amount of energy spent on food processing (SDA and U); finally, the energy remaining for growth is shown in red. If less energy is allocated to growth, BAF values are expected to increase because of reduced growth dilution.
Although in some lakes, such as in Memphremagog and Great Slave, all lake trout consume the same prey species, regardless of size, in Lakes Ontario, Opeongo, and Source the diet compositon of the lake trout changes with size (Figures 2D and 3D). This is significant because the energy density of the prey species can differ and could in conjunction with the BAF of the prey explain variations in lake trout BAF.
The model results suggest that the BAF of differently sized lake trout within a lake can vary by as much as 1 log unit for highly lipophilic chemicals. For the same-sized trout in different lakes, we calculate differences in BAF as high as 1.30 log units for chemicals with high log KOW values. Overall, for a chemical with log KOW of 6.5 we calculate BAFs that range over an order of magnitude for the same species, from 2.7 × 105 L kg−1 in small trout from Happy Isle Lake to 3.3 × 106 L kg−1 in very large trout in Lake Ontario.
Additional model results showing the relationship between consumption rates and energy intake for lake trout and how altering invertebrate diet can affect changes in the BAF of lake trout are reported in Section S5 of the Supplemental Data.
DISCUSSION
Factors which influence BAF
Using the combined bioenergetics-bioaccumulation modeling approach, we are able to identify 4 main factors contributing to differences in lake trout BAFs within and between lakes. The first 2 relate directly to energy allocation: the amount of energy a fish allocates to growth and to metabolic processes. The other 2 relate to lake trout diet: diet composition and the energy density of the prey.
The amount of energy a fish allocates to growth is in some cases defined as the growth efficiency, the proportion of total energy ingested by a fish used for growth (Pazzia et al. 2002). In the present study we compare the fraction of the total energy absorbed by lake trout allocated to growth. Within a given lake, as fish size increases, lake trout expend less energy on growth, so the effect of growth dilution decreases. The effects of growth efficiency on bioaccumulation have been observed before in lake trout and other salmonid species (Debruyn and Gobas 2006; Trudel and Rasmussen 2006; Paterson et al. 2016).
Energy for metabolic processes is heavily influenced by fish activity (as expressed by the population-specific activity multiplier). Activity levels of fish can vary widely, attributable, for example, to differences in foraging costs, prey densities, or predator evasion. Higher activity levels (expressed by higher activity multipliers) imply increased metabolic costs and total energy intake to maintain a given growth rate. For all modeled trout, over 50% of the total energy intake is used to maintain metabolic functions. Fish that have higher activity multipliers need to consume more to meet their energy needs and although activity multipliers increase with fish size, routine metabolic rates decrease (per gfish). To understand the overall effect of metabolic costs on the BAF, we consider the total energy intake of the fish (Figures 2B and 3B). Luk (2000) identifies fish activity levels as one of the main drivers for PCB intake in fish.
The energy density of its prey influences the consumption rate of a trout. For example, lake trout consuming cisco, a prey with high energy density, need to consume less than those feeding on energy-poor invertebrates. Although prey species with higher energy densities have higher lipid or nonlipid organic matter content, they do not always have the highest BAF.
Prey species common to different lakes (e.g., rainbow smelt and cisco) were assumed to have similar uptake and loss rate constants. Differences in prey species BAFs between lakes are attributable to differences in chemical bioavailability in the food chain and gut assimilation efficiencies. Prey fish feed directly on zooplankton or phytoplankton, so their chemical uptake is controlled by the fraction of freely dissolved chemical in the water; invertebrates, feeding lower in the food chain (see Figure S6 in Supplemental Data Section S2), have the lowest BAFs. Young-of-the-year (YOY) perch are extremely fast-growing, and growth dilution reduces the BAF in this prey species. Cisco and rainbow smelt both have similar growth rates, but cisco with higher lipid and nonlipid organic matter content retain more of the chemical. Alewife are extremely slow-growing fish and have high nonlipid organic matter levels (comparable to cisco), so chemical dilution is minimal.
Lake trout diet changes with size within lakes and can differ between lakes. The contamination level of the prey species in some cases has considerable influence on the BAF of the modeled trout. Trout consuming more contaminated prey generally have higher predicted BAFs. However, this works in concert with the energy density of the prey species.
In the next section we explain how the 4 identified factors combine to explain differences in BAF between differently sized trout in the same lake and between similarly sized trout from different lakes.
Differences between differently sized trout eating the same prey in the same lake
The BAFs predicted for very large trout in Lake Memphremagog and Great Slave Lake are higher than those of the 3 smaller size classes by factors of 1.9 and 1.4, respectively. Because all size classes eat the same prey, dietary composition cannot serve as an explanation. Figure 2B shows that it is the fraction of energy allocated to growth that sets the large trout apart from the others. The largest trout in Lake Memphremagog and Great Slave Lake allocate less energy to growth than the smaller trout. Because this difference is smaller in Great Slave Lake (7%) than in Lake Memphremagog (13–14%), the difference in BAF is also smaller in Great Slave Lake. Energy allocation by the 3 smaller size classes of trout is very similar (Figure 2B). The very large trout, while spending proportionately less energy on growth, also have elevated metabolic requirements. This is because the metabolic requirements of fish increase with size, so the total energy intake increases. Very large trout in Great Slave Lake also have an increased activity multiplier compared to the smaller class sizes (Figure 2B). However, the growth rate of trout in Great Slave is similar for all class sizes; increased BAFs for the 3-kg trout is then not attributable to a change in growth rate but to increased energy intake, because of higher activity, which results in reducing the fraction of energy intake allocated for growth.
Differences between differently sized trout eating different prey in the same lake
Although the diet of the 3 larger trout in Lake Opeongo (Figure 2) consists solely of cisco, the diet of small trout includes some invertebrates (20%), leading to a less energy-dense diet. As a result, the small trout has a slightly higher consumption rate than if it were to feed solely on cisco. However, because the contamination of invertebrates is much lower than that of cisco, this difference in diet has a minimal net effect on the BAF of the small trout. Indeed, in Lake Opeongo, we see a pattern similar to that in Lake Memphremagog and Great Slave Lake, discussed in the preceding section: the very large lake trout have elevated BAFs compared to the smaller class sizes because of a 17 to 18% reduction in the amount of energy allocated to growth, relative to the 100- and 500-g trout. The large trout (1 kg) also have a slightly elevated BAF compared to the smaller trout because the fraction of energy spent on growth is lower by 4%.
The largest lake trout (3 kg) in Lake Ontario has double the BAF of the smaller trout. Multiple factors contribute to this difference. The largest size class trout in Lake Ontario again allocates relatively less energy to growth (by 13–14%). In addition, they eat more contaminated prey with higher BAFs than the 3 smaller size classes, namely more alewife and no slimy sculpin (Figure 2D). The largest trout is also the most active, with an activity multiplier of 5, whereas smaller size class trout in Lake Ontario have activity multipliers of approximately 4; this elevates the total energy intake of the largest trout and, thus, demands higher consumption rates (see Figure S15 of Section S5 in the Supplemental Data). As a result, the difference in the BAF between the largest trout and the 3 smaller size classes is larger than in the other lakes. The effect of diet variation on PCB loading was also observed in Lake Michigan rainbow trout (Madenjian et al. 1994).
The situation in Source and Happy Isle Lakes is similar to Lake Ontario. The larger of the 2 size classes present has 3 to 4 times the BAF of the small trout because 1) it allocates 17 to 20% less to growth; 2) it begins consuming some YOY perch, which has a somewhat higher BAF than the invertebrates, which are the sole prey of the smallest trout; and 3) it has a much higher activity multiplier, resulting in a much higher energy intake compared to the smaller trout. In Source Lake the increase in BAFs between the 100- and 500-g trout is primarily attributable to an increased activity multiplier, which causes the total amount of energy being spent on metabolic processes to increase by a factor of about 7.3, whereas the same amount of energy is spent on growth. In fact, the amount of energy the 500-g trout is using to sustain metabolic processes alone is 4.2 times that of the total energy intake of the 100-g trout, and consumption rates for the 500-g trout are approximately 5.7 times higher than that of the 100-g trout. Thus, for the 500-g trout in Source and Happy Isle Lakes we see a large BAF increase (by 0.47 log units) relative to the smaller trout.
Differences between similarly sized trout from different lakes
We cannot only compare the BAF of trout within the same lake (different lines in one graph of Figures 2A and 3A) but also similarly sized trout in different lakes (lines of the same color in Figures 2A and 3A).
All trout in Lake Ontario have BAFs that are higher than those of the same-sized fish in other lakes. Lake Ontario trout have higher energy intake (Q × δd; Figure 2B) when compared to the same-sized fish in other lakes, as a result of more energy used to maintain metabolic processes (because of higher activity multipliers) and high growth rates. Higher energy intake is associated with higher consumption rates (see Supplemental Data Section S5, Figure S15); and because Lake Ontario trout are feeding on more contaminated prey species, this further drives up their BAFs.
The small 100-g trout in Source and Happy Isle Lakes have the lowest BAFs among all of the small trout. Although their total energy intake is very similar to that of small trout in Lake Opeongo and Great Slave Lake, their consumption rates are higher because they are feeding on invertebrates, which have relatively low lipid content and therefore are very energy-poor (see Figure S7 of Section S2 in the Supplemental Data). Invertebrates are less contaminated than most other prey species (with the only exception of YOY perch), so despite high consumption rates, the 100-g trout in the smaller lakes have very low BAFs because of their prey.
Whereas the BAFs of the 500-g trout in Happy Isle Lake are similar to those of the same-sized fish in other lakes, the 500-g trout in Source Lake have higher BAFs than the same-sized trout in other lakes, with the exception of Lake Ontario. At 500 g, trout in Source and Happy Isle Lakes are still feeding on small prey species, invertebrates, and YOY perch; and fish feeding on smaller prey are known to have lower growth rates (Giacomini et al. 2013) because they have less energy available to spend on growth. Smaller prey tend to be energy-poor, so feeding on smaller prey means that trout have to spend more time foraging for food and thus have increased activity multipliers (Dodrill et al. 2016). It is also possible that the prey density (the amount of prey species in a given area) in these smaller lakes is lower, which increases the amount of time spent foraging for larger fish (Giacomini et al. 2013), elevating the activity multiplier and subsequently the energy needed for metabolic processes. The 500-g trout in Source Lake has a much higher activity multiplier than the 500-g trout in Happy Isle Lake and thus uses less energy on growth (2%) compared to trout in Happy Isle Lake (5%).
The BAFs of trout in Lakes Memphremagog, Great Slave, and Opeongo are generally similar for all trout sizes modeled. However, this is not because of similar bioenergetics. The distribution and amount of energy used for metabolic processes and growth are different, and therefore consumption and growth rates for the same-sized trout in these 3 lakes also differ. For example, in Lakes Opeongo and Great Slave, the BAFs of the 100-g trout are very similar, and BAFs of the 100-g trout in Lake Memphremagog are different by merely 0.1 log units at log KOW 6.5. However, trout in Lakes Opeongo and Memphremagog grow 1.7 and 2.5 times faster, respectively, compared to the same-sized trout in Great Slave Lake. Lake Memphremagog trout have the greatest total energy intake of these 3 trout and allocate the most energy to growth. Memphremagog trout feed on rainbow smelt, which has a lower energy density, relative to cisco, which is the primary prey species for trout in Lakes Opeongo and Great Slave. The latter 2 trout therefore have decreased consumption rates and subsequently a lower dietary uptake rate constant. Interestingly, despite having different uptake and loss processes, the difference between total chemical uptake and total chemical loss is equal between the 100-g trout in Lake Opeongo and Great Slave Lake. In this case, trout with very different growth rates have the same BAFs because of differences in their bioenergetics.
The 3-kg trout in these 3 lakes display larger differences in the calculated BAFs. Trout in Great Slave Lake have lower BAFs compared to trout in Lakes Memphremagog and Opeongo and in Lake Memphremagog relative to Lake Opeongo. However, total energy intake is greatest for Great Slave Lake trout, then Opeongo trout. Trout in Great Slave Lake and Opeongo both feed only on cisco, and although a 3-kg trout in Lake Opeongo allocates only 3% of its total energy intake to growth, this fraction is 10% in Great Slave Lake. With a higher activity multiplier, Great Slave Lake trout use more energy to maintain metabolic processes relative to the same-sized fish in Lake Opeongo. As a result, the total energy intake is very similar in both lakes, and because they feed on the same prey species, so is their consumption rate. In this instance we see that because lake trout in Lake Opeongo use less energy on growth, they have increased BAFs relative to the same-sized trout in Great Slave Lake. This may explain why lake trout with similar trophic levels and food webs display different BAFs, as observed by Rasmussen et al. (1990).
The 3-kg trout in Lake Memphremagog has a higher BAF than the same-sized trout in Great Slave Lake, yet both spend 10% of their total energy intake on growth. With a higher activity multiplier, the Great Slave Lake trout has higher total energy intake. Memphremagog trout are consuming less fatty rainbow smelt compared to trout in Great Slave Lake eating lipid-rich cisco, which means that the trout in Memphremagog actually has a higher consumption rate because it needs to feed more to meet its energy requirements; Great Slave Lake trout are feeding on an energy-rich diet and do not need to eat as much to meet the same energy requirements. This causes the dietary uptake rate constant for Great Slave Lake trout to be lower than in Lake Memphremagog, resulting in lower BAFs.
Luk (2000) examined PCB bioaccumulation in 3 salmonid species, lake trout, brown trout, and chinook salmon, in Lake Ontario. They found that the rate of chemical intake increases so quickly with the size and age of the fish that no effect of growth dilution on chemical concentrations was observed. This is clearly apparent when the growth rate of trout in all lakes is compared. Lake Ontario trout, which generally is the fastest-growing (the only exception being at 3 kg, where Great Slave Lake trout have slightly faster growth rates), also has the highest predicted BAF because it is feeding on a more contaminated diet, has a higher energy intake, and allocates more energy to metabolic processes because of a higher activity multiplier. As a result, the 100-g trout from Lake Ontario has a BAF similar to that of the 3-kg trout from Great Slave Lake, which is feeding on very lipid-rich cisco. Despite being slow-growing, the trout in Source and Happy Isle Lakes had low BAFs because they were consuming invertebrates which are relatively clean. It is apparent that it is important to consider not only the growth rate of lake trout in different lakes but also the trophic levels and the bioaccumulation potential and energy density of the prey species.
Implications
Bioenergetics plays a significant role in determining the bioaccumulation potential of trout in lakes. In particular, the activity multiplier proved to be important in determining the total energy intake of a fish, which in turn influences the BAF. Experimental work and models, which do not adequately take this parameter into account, may be underestimating the bioaccumulation potential of chemicals in some food webs. For example, in caged fish or mesocosm bioaccumulation studies, the fish may be using less energy for metabolic processes because they are unable to swim freely (Brown et al. 2002; Miller et al. 2014), which could mean that BAFs are biased low. Conditions in tank and laboratory experiments are also often standardized and designed so that fish experience very little stress. They do not need to hunt for food or evade predators and, therefore, will have reduced activity levels, relative to an organism in the wild. For example, Debruyn and Gobas (2006) observed that laboratory studies have underestimated maximum biomagnification factors for fish, which they attributed to increased growth efficiency (total energy spent on growth) because tank fish have decreased activity levels and are provided food.
In 5 of the lakes, water temperatures ranged between 6.9 and 7.3 °C, reflecting the preferences of trout; only in Great Slave Lake was temperature lower, at 5.1 °C. These cooler temperatures had only a small impact on the calculated routine metabolic rate, which was countered by elevated activity multipliers. Overall, temperature effects on the BAFs of trout in these lakes appear to be minimal. However, indirect effects related to climate change may have a more dramatic effect on the BAFs of chemicals. Increasing lake temperatures could limit the range and thermal habitat of lake trout and other cold-water species (Guzzo et al. 2017). In warmer waters, some lake trout may lose access to littoral feeding areas, which can limit their prey availability (Guzzo et al. 2017). Climate change may also have more complex effects on lake trout ecosystems, for example, when changes in food-web structure and predator–prey dynamics arise from species migrating north into warming waters.
A change to the food-web structure in Lake Ontario, namely the introduction of alewife, has already been observed to alter contaminant concentrations in lake trout (Borgmann and Whittle 1991). The addition of invasive species with ballast water and fisheries can also have dramatic effects on lake trout bioaccumulation. The introduction of bass into lakes across North America created competition for lake trout feeding on pelagic fish, forcing them to shift their diet to smaller zooplankton (Van der Zanden et al. 1999). We might expect such trout to increase their foraging (i.e., increased activity) and consumption because they are feeding lower on the food chain. Limits on their growth potential could ultimately lead to increased BAFs, in a scenario not unlike the one in the present study for Source Lake and Happy Isle Lake.
Lake trout are a highly diverse species, and in the last decade multiple polymorphs of lake trout have been identified to exist within the same lake (Muir et al. 2016). Lake trout polymorphs can have a completely different life history, which includes different growth rates, lipid levels, and maturation rates. They also live within different habitats and niches within a single lake and have different feeding preferences (Muir et al. 2016). Some polymorphs have been declared to be genetically distinct (Zimmerman and Krueger 2009). In the present study, we show that fish physiology can strongly influence the bioaccumulation potential of POPs, which suggests that within- and between-lake variability in measured BAFs of trout could be attributable to differences in the life history of different lake trout polymorphs.
CONCLUSIONS
We identified 4 main factors which can increase the bioaccumulation potential in lake trout: 1) a low fraction of total energy intake allocated to growth, 2) high energy intake attributable to higher activity levels (activity multipliers), 3) high prey species contamination, and 4) feeding on prey species with low energy density. These factors work in conjunction with each other, such that all of these factors need to be considered to explain the calculated differences in BAFs. For example, even trout with relatively fast growth rates can have high bioaccumulation potential, and trout with very different physiology and bioenergetics can have similar BAFs.
This understanding was made possible by using a combined bioenergetic-bioaccumulation modeling approach. Although constraining the parameterization of the bioaccumulation model with bioenergetics is shown to be extremely valuable, the field-based data required for model parameterization is often not available, particularly population-specific activity multipliers. Nevertheless, model calculations spanning plausible ranges of activity multipliers for different fish species (Kitchell et al. 1977; Hanson et al. 1997; Giacomini et al. 2013) can be conducted to quantify the potential influence of this parameter on model output. Data characterizing the ecological status of the aquatic ecosystem (e.g., oligotrophic vs eutrophic, fish abundance) may also provide useful guidance for parameterizing combined bioenergetic-bioaccumulation models for different scenarios.
Supplemental Data
The Supplemental Data are available on the Wiley Online Library at DOI: 10.1002/etc.4368.
Acknowledgment
We thank D. Muir and N. Mandrak for helpful feedback and suggestions and J. Rasmussen for insights on fish growth rates. Thank you also to D. Muir and M. Evans for providing data for model evaluation. We are grateful for funding from the Natural Sciences and Engineering Research Council of Canada.
