Mercury concentrations in lake sturgeon from four river systems in Manitoba, Canada

Abstract

Lake sturgeon (Acipenser fulvescens) are large, long-lived fish, characteristics conducive to the bioaccumulation of mercury. This study measured total mercury (THg) in the axial musculature of 319 lake sturgeon captured between 2002 and 2022 from six river reaches on the Winnipeg, Nelson, Churchill, and Hayes Rivers in the province of Manitoba, Canada. Since the 1900s, all sampled rivers except the Hayes were affected by hydroelectric development, which can increase THg concentrations in fish for several decades following inundation. Concentrations of THg in lake sturgeon measuring 171–1,435 mm fork length ranged from less than the detection limit (0.005 ppm) to 0.698 ppm wet weight. Total mercury concentrations increased significantly with fish length for all waterbodies analyzed, except for the Hayes River, where the sample size and length range were small. Mean length-standardized THg concentrations ranged from 0.069–0.177 ppm and differed significantly between waterbodies. Concentrations were lowest where the majority of samples were collected from fish that had been hatchery-reared and stocked. A paucity of historical data precludes the isolation of effects of hydroelectric impoundment on THg levels in lake sturgeon, but concentrations are not particularly high in any of the waterbodies examined herein. Of the 35 lake sturgeon also analyzed for diet composition, 34 had identifiable gastrointestinal contents. Ten different prey taxa were identified, including fish. Chironomids dominated dietary numerical abundance. The number of taxa increased with fish length to include larger, and therefore potentially more, THg‐rich organisms.

Introduction

Lake sturgeon (Acipenser fulvescens) have a historical range covering much of central North America outside of Mexico (Bruch et al., 2016). Within Canada, the lake sturgeon range extends from Alberta in the west to the St Lawrence River estuary (Québec) in the east. Within Manitoba, lake sturgeon are distributed throughout the majority of large river systems, including the Red, Assiniboine, and Winnipeg Rivers in the south and the Churchill, Burntwood, Nelson, Hayes, and Gods Rivers in the north. As with other populations in Canada, lake sturgeon populations within Manitoba have experienced sharp declines beginning in the late 1800s due in part to overexploitation and habitat loss and alterations (Haxton & Findlay, 2008; Manitoba Economic Development, Investment, Trade and Natural Resources, 2024). In 2017, the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) assessed the Designatable Units that encompass Manitoba populations of lake sturgeon as “endangered” (COSEWIC, 2017).

The majority of river systems inhabited by lake sturgeon within Manitoba have been affected by flooding as the result of hydroelectric development beginning in the early 1900s on the Winnipeg River. The Winnipeg River flows from Ontario to Manitoba. Six generating stations (GSs) are located within the Manitoba portion of the river, including Pointe du Bois (1926), Slave Falls (1948), Seven Sisters (1952), MacArthur Falls (1955), Great Falls (1928), and Pine Falls (1952; Figure 1; Manitoba Hydro, n.d.). The Churchill River diversion was completed in 1976 and diverts the majority of the flow of the Churchill River (which flows from Alberta and Saskatchewan into northwestern Manitoba) into the Rat River-Burntwood River-Nelson River system to augment power generation potential (Manitoba—Manitoba Hydro, 2015). This project has decreased water levels and flows within the lower Churchill River system while increasing flows within the Rat River-Burntwood River-Nelson River system, flooding a total area of 759 km². The Lake Winnipeg Regulation Project was completed in the same year as the Churchill River diversion, to regulate Lake Winnipeg outflows into the Nelson River, flooding an additional 65 km² (Manitoba—Manitoba Hydro, 2015). Six GSs are located on the Nelson River (Figure 1), including Jenpeg (1979), Kelsey (1961), Keeyask (2022), Kettle (1974), Long Spruce (1979), and Limestone (1992). The extent of flooding varied for each structure, from as little as 2.1 km² for the Limestone GS to as much as 221 km² for the Kettle GS (Manitoba—Manitoba Hydro, 2015). The Hayes River is the only unregulated river system sampled in our study. Located entirely within northeastern Manitoba, it flows just under 500 km from Molson Lake to Hudson Bay.

Flooding of terrestrial areas (often a result of hydroelectric projects) is known to increase the bioavailability of mercury, mainly by enhancing bacterial production of methylmercury (Bodaly et al., 2004; St Louis et al., 2004), which promotes the bioaccumulation and biomagnification of mercury through the food chain (Bilodeau et al., 2017; Bodaly et al., 2007). Dietary mercury exposure can negatively affect fish reproduction and growth (Crump & Trudeau, 2009; Depew et al., 2012; Simoneau et al., 2005), including sturgeon species (Webb et al., 2006). Lake sturgeon primarily consume benthic invertebrates (e.g., Chironomidae, Ephemeroptera, and Trichoptera), and larger individuals often feed at higher trophic levels, with diets including crayfish and some fish (Barth et al., 2013; Braun et al., 2018; Bruestle et al., 2019; Smith et al., 2016) that likely have elevated mercury concentrations compared with lower trophic levels. Further, lake sturgeon attain a large size over the course of long lifespans, a characteristic that is conducive to the bioaccumulation of mercury, because older and, normally, larger fish have higher concentrations than younger, smaller fish (Evans et al., 2005; Green, 1986). For all these reasons, lake sturgeon are particularly vulnerable to the adverse effects of mercury.

Although a national Canadian fish mercury database contains more than 1,000 entries on lake sturgeon collected between 1967 and 2010 (Depew et al., 2013), to our knowledge, no study with a primary focus on mercury levels in lake sturgeon in Manitoba exists. A recent publication presented mercury levels in lake sturgeon from other geographic locations, but this study analyzed mercury in the blood of sturgeon as opposed to the axial musculature (Lescord et al., 2024). Over the past two decades, incidental sturgeon mortalities that have occurred during environmental monitoring and stewardship initiatives conducted in Manitoba have been sampled and analyzed for total mercury (THg) in the axial musculature. These data were collected by Manitoba, Manitoba Hydro, and the Keeyask Hydropower Limited Partnership. Samples were collected from six reaches of three regulated (Winnipeg, Nelson, and Churchill) and one unregulated (Hayes) river system(s). With the inclusion of biometric data for each fish sampled for mercury, the recently compiled data from the Manitoba, Manitoba Hydro, and the Keeyask Hydropower Limited Partnership initiatives likely have more relevance and utility than previous records, which are based mainly on composite samples (i.e., a triplicate sample taken from the homogenized fillets of five fish) from commercial captures in Manitoba between 1970 and 1983 (Department of Fisheries and Oceans [DFO], 1987).

This study is based on THg concentrations for 319 lake sturgeon collected over a 20-year period from six river reaches within Manitoba, Canada. Specifically, we evaluated whether mercury concentrations in lake sturgeon correlate with fish length and whether they differ among river systems (including rivers regulated by hydroelectric development and an unregulated river) and over time within Manitoba, Canada. For a small subset of sturgeon (n = 35) for which diet analysis was conducted, we explored whether variation in THg concentration among individuals could be related to diet.

Material and methods

Sample collection

Samples were collected from incidental sturgeon mortalities during gillnetting studies conducted by Manitoba and Manitoba Hydro as part of the Coordinated Aquatic Monitoring Program (CAMP, 2017, 2023), by Manitoba Hydro as part of environmental monitoring studies associated with the Conawapa and Point du Bois GSs and the Lake Sturgeon Stewardship and Enhancement Program and by the Keeyask Hydropower Partnership for the Keeyask GS (Table 1). Lake sturgeon were obtained from six river reaches within four Manitoba rivers between 2002 and 2022, including the Winnipeg River, Nelson River, Churchill River, and Hayes River (the only completely unregulated river considered in this study; Table 1). All sampling sites are located within the boreal forest ecozone. For the purposes of analysis, the 671 km long Nelson River was split into three reaches: upper (Lake Winnipeg outlet to Kelsey GS), middle (Kelsey GS to the Keeyask GS), and lower (Keeyask GS to the Nelson River estuary) sections based on marked habitat differences among the sections and spatial isolation of sturgeon populations between them (Gosselin et al., 2015). Six samples obtained from the lower Burntwood River were included with the middle Nelson River for analysis because lake sturgeon are known to move between the two areas and do not represent genetically distinct populations (Gosselin et al., 2015). Fork length (± 1 mm) and weight (± 1–25 g depending on fish size) were obtained in the field and a sample of axial muscle weighing approximately 10–40 g was removed just anterior to the caudal fin. The skinned muscle sample was frozen and sent to an analytical laboratory. For a few smaller lake sturgeon, mercury samples were not taken in the field but rather were removed in the laboratory from previously frozen fish. These whole lake sturgeon were thawed to the point where the entire gastrointestinal tract (GIT) could be removed intact for analysis of the diet.

Mercury laboratory analysis

Because of the approximately 20-year time span of data reported herein, as well as the consolidation of multiple data sources, it is important to note that the analytical methodologies were not identical for the entire data set. Between 2002 and 2006, analyses of sturgeon samples were performed in the same laboratory (metals laboratory of the Freshwater Institute in Winnipeg, Manitoba, DFO Canada) and with few personnel changes. Total mercury (THg) analysis of the 20 samples from sturgeon collected between 2002 and 2006 was performed using a modified method described by Hendzel and Jamieson (1976) that increased the efficiency of sample digestion by use of an aluminum “hot block,” and analyzed THg using cold-vapor atomic absorption spectrometry with a detection limit (DL) of 0.050 ppm. The 299 lake sturgeon samples collected between 2007 and 2022 were analyzed by ALS Environmental (Waterloo, Ontario and Winnipeg, Manitoba). Total mercury in digested muscle tissue was determined using cold vapor atomic fluorescence spectrophotometry following U.S. Environmental Protection Agency (USEPA) Method 1631E (USEPA, 2002) on a PSA Millenium Merlin 10.025 device (PS Analytical) or by cold-vapor atomic absorption spectrometry applying a modification of USEPA Method 200.3/1631E and using a Teledyne Leeman M-7600 mercury analyzer (Teledyne Leeman Labs, Hudson, NH). Quality control (QC) results were within the control limits for the QC sample (ALS Data Quality Objective). The method DL ranged from 0.010–0.040 ppm for samples analyzed between 2009 and 2016 and 0.0010 ppm for samples analyzed between 2017 and 2022. Method blanks were consistently low (< 0.001 ppm), and no blank correction was applied.

In all cases, THg was determined as a proxy for methylmercury. The concentrations of THg reported in this article are generally much higher than the DL of the analytical methods used, and no obvious bias existed in the association of certain sampling locations or years with analytical methods or laboratories. Thus, it is contended that the data presented herein should be robust to the slight methodological variations that occurred.

Diet determination

Diet analysis of 35 lake sturgeon captured between August and October 2011–2017 was based on the content of the entire GIT of a subsample of lake sturgeon (range of 97–745 mm fork length) analyzed for THg. The entire GIT was used, because previous studies have shown that the percentage of lake sturgeon containing food in their stomach and esophagus is less than 65% (Barth et al., 2013; Nilo et al., 2006). Gastrointestinal tracks placed into a Petri dish on top of a frozen gel pack were dissected starting from the esophagus. Diet items were removed, placed under a dissecting microscope for identification, and enumerated. Identification of organisms was to the lowest taxonomic level feasible while also considering the state of digestion, trophic status, contribution to sturgeon feeding ecology, and approximate THg concentration based on known literature values. Numbers of individuals of each taxon were counted, including estimates based on the number of insect head capsules or other uniquely identifying structures. The percentage of numerical abundance was calculated for each taxon as the number of individuals consumed by all fish divided by the total number of individuals found in the GIT of all fish. The percentage of frequency of occurrence was calculated as the number of lake sturgeon that had consumed the prey item divided by the total number of sturgeon with food in their GIT.

Data analysis

Most fish species, including lake sturgeon, have been observed to accumulate mercury over their life, with concentrations increasing with size (Haxton & Findlay, 2008; MacCrimmon et al.,1983; Simoneau et al., 2005). To account for this variability and facilitate comparisons between locations, total mean concentrations for each reach were standardized for fish length, based on the significant relationship between logarithmic transformations of muscle THg (µg g⁻¹) and fork lengths (mm) of each individual. For each reach, a linear regression was fit between THg and fork length, and a mixed-effect model was fit with fork length as a fixed covariate and reach as a random effect to predict THg concentration for each individual fish at 700 mm fork length. A standard length of 700 mm fork length was chosen to calculate length-standardized mean mercury concentrations for each reach (referred to as the standard mean) as it was close to the mean length of sturgeon in the present data set and has been used in previous analyses (i.e., CAMP, 2017). The residuals from the fixed-effect model were then added back to the predicted values to calculate the standardized THg concentration for each individual fish (Eagles-Smith et al., 2016; Eagles-Smith & Ackerman, 2014). Length standardization was consistent with that of many other studies (e.g., Bodaly et al., 2007; Eagles-Smith et al., 2016; Simoneau et al., 2005). All mercury results are reported as wet weight concentrations of THg (µg·g⁻¹ or parts per million [ppm]).

Differences in mean fork length, arithmetic, and length-standardized mercury concentration of sturgeon between locations were assessed using a one-way analysis of variance (ANOVA) and Dunn-Sidak’s pairwise multiple comparison tests. If normality of data distribution or equality of variances was not achieved by data transformation, Kruskal-Wallis one-way ANOVA on ranks was performed, applying Dunn’s method for pairwise multiple comparisons. In all cases, significance was established at p ≤ 0.05.

The relationship between length-standardized mean mercury concentrations per waterbody and time over the sampling period was assessed for each river reach using analysis of covariance (ANCOVA), with standardized THg included as the independent variable, year as a dependent continuous covariate, and reach as a dependent categorical variable. A model was first fit that allowed for separate lines for each river reach (i.e., a nonparallel slope model). If slopes were parallel, a second model was fit that allowed for the same lines (i.e., a parallel slope model), but different intercepts for each reach. If there was no relationship with year and no differences in slopes or intercepts among years, a third “global” model was fit between pooled mean annual length-standardized THg per year and year. For all models, a Type III (marginal) test was used to test the hypothesis of significant and/or parallel slopes and significant and/or different intercepts. In all cases, significance was established at p ≤ 0.05.

Lake sturgeon THg concentration as a function of diet was examined using a forward stepwise linear regression approach. Variables considered were fish length, taxa number, and the percent numerical abundance of each taxon, with inclusion conditional on a p < 0.05.

Statistical analyses were run using XLSTAT 2023.1.6 (Lumivero, 2023) and R Ver. 4.2.2 (R Core Team, 2022). Mixed-effect modeling was completed using the R package lme4 (Ver. 1.1–14, Bates et al., 2015).

Results and discussion

Comparison of mercury concentrations among waterbodies

A total of 319 lake sturgeon with individual mercury data were available for this analysis. The majority came from the Upper, Middle, and Lower Nelson River (45%), followed by the Winnipeg River (40%), the Churchill River (13%), and the Hayes River (2%; Table 1; see online supplementary material Table 1). Individual THg concentrations measured between 0.005 ppm (concentration measured below DL, censured value equivalent to 50% of the DL) and 0.698 ppm (Figure 2). Mean arithmetic THg concentrations ranged from 0.041 ppm in the upper Nelson River to 0.159 ppm in the Hayes River (Table 2; see online supplementary material Table 1). Mean arithmetic THg differed significantly among waterbodies (ANOVA, F_1,5 = 3.51, p = 0.004). Total mercury concentrations were significantly lower in the Upper Nelson River than in the Hayes, Lower Nelson, and Middle Nelson Rivers but did not differ significantly from the Churchill or Winnipeg Rivers (Figure 3). However, some of these differences may be explained by differences in length distributions.

Lake sturgeon fork length ranged between 171–1,435 mm (Figure 2) with average length ranging from 480 mm in the upper Nelson River to 816 mm in the middle Nelson River (Table 3; see online supplementary material Table 1). Mean fork length was significantly lower in the Winnipeg and upper Nelson Rivers than in the Churchill, lower Nelson, and middle Nelson rivers (ANOVA, F_1,5 = 3.51, p = 0.004; Figure 3). Total mercury concentrations increased significantly with fish length for all waterbodies except in the Hayes River, likely related to the small number of available samples (Table 3) and the narrow range of sturgeon lengths (Table 2; see online supplementary material Table 3). As a result, THg concentrations were standardized by fish length for all waterbodies.

Mean length-standardized THg concentrations (standard means) ranged from 0.069 ppm in the Upper Nelson River to 0.177 ppm in the Hayes River and differed significantly between waterbodies (Table ANOVA, F₁,₅ = 13.18, p < 0.0001). The standard mean for the Upper Nelson River was lower than for any of the other waterbodies sampled (Figure 3). The majority of lake sturgeon sampled in the Upper Nelson River were the progeny of brood stock collected at the Landing River (a tributary to the Upper Nelson) and were reared in the Grand Rapids Hatchery until age 12–18 months prior to being stocked into the Upper Nelson River at Sea Falls and Pipestone Lake (McDougall et al., 2020). Mercury concentrations of the lake sturgeon at the time of hatchery release is unknown, but based on their diet of commercially raised chironomids (Cheryl Klassen, Environmental Specialist, Manitoba Hydro, Winnipeg, personal communication) and their generally faster growth compared with wild fish (McDougall et al., 2020), concentrations may be lower than would be expected of similar-aged wild conspecifics. Mean THg concentration of three Grand Rapids Hatchery food chironomid samples were measured for this study and ranged from 0.005–0.014 ppm. Total mercury concentrations of chironomids sampled in the wild are generally higher (discussed below). In combination with presumed mercury growth dilution in which rapid growth from high-quality food consumption can reduce mercury accumulation (Essington & Houser, 2003; Simoneau et al., 2005), the significantly lower standard mean for lake sturgeon from the Upper Nelson River compared with all other waterbodies suggests that fish stocked at age 1+ maintain lower mercury concentrations compared with similar-sized fish from other locations, even several years after hatchery release.

The ANCOVA and linear regression analyses used to assess the relationship between length-standardized THg and time (year) suggested there were no changes in THg concentrations over the up to 20-year (see online supplementary material Table 1) sampling period for any of the reaches. For the different line ANCOVA, there was no significant relationship between size-standardized THg concentrations and year for any of the river reaches nor did the slopes differ significantly among reaches (ANCOVA interaction term p = 0.77; see online supplementary material Table 2; Figure 1). Similarly, for the main effect and global linear regression models, there was no significant relationship between standardized mercury concentrations and year for any of the river reaches or when all reaches were pooled (ANCOVA year term p = 0.76; linear regression year term p = 0.75; see online supplementary material Table 2). The Hayes River intercept in the main effect model was significantly higher than the Winnipeg River and the Upper Nelson River when year was accounted for.

Standard means were highest in the Hayes River and Lower Nelson River (Figure 3). Although drivers of mercury accumulation in fish are complex, studies have suggested that waterbody type (i.e., lake, river, or reservoir) may significantly influence THg concentrations in fish (Eagles-Smith et al., 2016; Kamman et al., 2005). The Hayes River is the only system in the current study that is not affected by hydroelectric development, and it may be expected that THg concentrations would be lower in this reach than in other areas; however, this was not the case. Similarly, Haxton & Findlay (2008) found no differences in lake sturgeon muscle THg concentrations between impounded and unimpounded reaches of the Ottawa River, and Lescord et al. (2024) reported higher blood mercury levels in lake sturgeon from an unaffected river compared with conspecifics from within a hydroelectric complex on a nearby river. The latter authors suggested that the old age of the GSs, differences in sturgeon trophic ecology and growth, or river morphology, catchment characteristics and limnology were likely responsible for the higher mercury concentrations in the unaffected river. Wetlands are typically an important source of mercury to downstream aquatic ecosystems (see Selvendiran et al. 2008 and references therein) and percentage of wetland was the most important predictor of dragonfly mercury concentrations among landscape variables in a continental-scale modeling study (Kotalik et al., 2025). However, the proportion of wetlands in the watershed of each river reach in our study likely did not differ markedly. Many other factors including local habitat type and water chemistry, levels of eutrophication, topographic variation, location within a watershed, climate, and human influence (such as mining and forestry operations) may also play a role in influencing THg concentrations in freshwater fish (Eagles-Smith et al., 2016; Kozak et al., 2021; Lescord et al., 2019; Lucotte et al., 2016; Thomas et al., 2020). It is likely that several of these factors may have contributed to the differences in lake sturgeon THg concentrations among river reaches. It must also be noted that only a small number (n = 7) of lake sturgeon were sampled from the Hayes River and these fish varied little in size (543–771 mm fork length). Thus, the results for this river reach are somewhat tenuous and potentially prone to stochasticity.

Most of the Manitoba river systems with contemporary lake sturgeon populations have hydroelectric development. However, except for the Keeyask GS (in operation since 2022) on the Middle Nelson River, the construction of GSs and their associated reservoirs in Manitoba dates back 30 years or more. The flooding of organic soils during reservoir creation results in a rapid increase to peak THg concentrations within 4–8 years (depending on the amount of flooding and the fish species considered). Thereafter, concentrations typically decline slowly, taking up to 20–40 years to reach preflooding levels (Bilodeau et al., 2017; Bodaly et al., 2007). As such, it is perhaps unsurprising that lake sturgeon mercury levels were not particularly high in any of the rivers examined in our study. However, similar to observations for other large-bodied fish species (Bodaly et al., 2007), it is possible that mercury concentrations in lake sturgeon from Manitoba waters were elevated for some time soon after hydroelectric development.

Historical data on THg concentrations in lake sturgeon in Manitoba is scarce. Data from northern Manitoba, published in Derksen (1978a, 1978b), included 26 composite samples collected between 1970 and 1978 from commercially caught fish from Mud (n = 1), Playgreen (n = 7), Cross (n = 1), Sipiwesk (n = 2), and Split (n = 2) lakes, and the lower Nelson River (n = 13). These composite samples of relatively large sturgeon (because commercial harvest tends to target adult sized fish) had mean THg concentrations of 0.10–0.18 ppm with a range of 0.06–0.49 ppm (Derksen, 1978a, 1978b, 1979; DFO, 1987), similar to the concentrations reported in our study. However, these data mainly predate the 1978–1986 period when substantially elevated THg concentrations were observed in other fish species from northern waterbodies affected by flooding (Bodaly et al., 1984, 2007). Sturgeon mercury data from the period of peak elevation in other species are limited to the following: three individuals from the Fox River (a Hayes River tributary not influenced by hydroelectric development) captured in 1979 with a mean concentration of 0.26 ppm; five individuals from the Lower Nelson River caught in 1982 with a mean concentration of 0.22 ppm; and two commercial samples from the Lower Nelson River caught in 1982 and 1983, both with a concentration of 0.12 ppm (DFO, 1987; DFO, 2023). The mean THg concentration for individual sturgeon from the Lower Nelson River in 1982 (i.e., 0.22 ppm) is more than twice the weighted average of 0.10 ppm for the 13 commercial samples from the same area in 1970–1978, suggesting an increase in concentrations consistent with the timeline for increases observed for other species. However, the relatively high mean of the fish collected in 1982 is mainly due to one sturgeon with a THg concentration of 0.6 ppm, whereas the other four fish ranged from 0.10–0.15 ppm. Thus, stochasticity likely associated with small sample sizes, the lack of fish lengths for the commercial samples, and uncertainties regarding the exact capture locations preclude robust conclusions regarding temporal trends/increases in sturgeon mercury concentrations in the 1970s and 1980s potentially related to the effects of impoundment of Manitoba rivers.

Overall, current THg concentrations in lake sturgeon from Manitoba appear to be generally similar if not slightly lower compared with their conspecifics sampled elsewhere in Canada. Haxton and Findlay (2008) reported muscle THg concentrations of 0.06–0.68 ppm in 48 lake sturgeon from the Ottawa River measuring approximately 500–1,130 mm total length. More recently, Lescord et al. (2024) reported whole blood THg concentrations of 0.11–5.46 ppm dry weight for 46 lake sturgeon measuring approximately 80–130 mm fork length from two northern Ontario rivers. Similar to our study, both of these authors found a significant increase in THg with fish length despite considerable variability among individuals. A significant positive relationship between THg concentration and fish length was also obtained for 89 lake sturgeon measuring between approximately 900 –1,090 mm total length from the Moose River, Ontario (Threader & Broussaeu, 1986). Based on a regression model, these sturgeon exceeded mercury concentrations of 0.5 ppm at lengths >1,020 mm. Although mercury concentrations of up to 2.9 ppm have been reported in some lake sturgeon measuring ≥1,300 mm total length (Figure 9 in Haxton & Findlay, 2008), the median concentration for approximately 1,000 individuals of 955 mm median total length in a Canadian database is 0.23 ppm (Depew et al., 2013).

Sturgeon diet and mercury concentrations

Of the 35 lake sturgeon with THg concentrations that were also analyzed for diet composition, 34 had identifiable GIT contents. Fish from three river systems contributed diet information, the majority coming from the Winnipeg River and the Upper Nelson River (including five hatchery reared fish). Ten different taxa of aquatic invertebrates and fish were identified (Figure 4). Chironomids were present in the GIT contents of more than 90% of individuals sampled and also dominated (81%) the numerical abundance of the 4,764 counted diet items (Figure 4). All GITs contained very little plant and mineral material, with the main exception of a few caddisfly cases.

The number of taxa found in the GIT significantly increased with fish length (taxa number = 0.586 + 0.289 length, r² = 0.32, p < 0.001), whereas no significant relationship was observed between the total numerical abundance of food organisms and fish length. However, the percentage of numerical abundance of chironomids significantly decreased with sturgeon length (percentage of abundance = 98.7–0.097 length; r² = 0.18, p = 0.01). When sturgeon length, taxa number, and the percentage of numerical abundance of each taxon were included in a stepwise regression with THg concentration as the dependent variable, only taxa number significantly (p = 0.04) explained variability in THg beyond the effect of fish length (p < 0.001). Ontogenetic diet shifts and prey selectivity in lake sturgeon have been previously reported (Guilbard et al., 2007; Nilo et al., 2006; Sandilands, 1987; Smith et al., 2016), which suggests individuals may broaden their diet spectrum to include larger prey. Consistent with this hypothesis, we found the diet of the smallest and youngest (age 0+ and 1+) sturgeon to consist mainly of chironomids. The diet of bigger sturgeon more often included larger food items, such as other insects, particularly burrowing mayflies (genus Hexagenia), crayfish (Orconectes virilis), and fish. The literature suggests that these organisms have higher THg concentrations (Mathers and Johansen, 1985 for Hexagenia: 0.047 ppm; Pennuto et al., 2005 for Orconectes virilis: range of 0.023–0.55 ppm) compared with chironomids (Göthberg, 1983: mainly < 0.020 ppm; Watras et al., 1998: 0.023 ppm mean for 15 waterbodies). Preferential feeding on invertebrates or fish with relatively high mercury content could contribute to higher mercury concentrations in sturgeon relative to individuals feeding mainly on chironomids.

It must be cautioned that our results based on 34 fish are not fully representative of the 319 lake sturgeon with known mercury concentrations. With a mean fork length of 386 mm (range 97–745 mm), the sturgeon with diet composition data were smaller than the mean fork length of all sturgeon analyzed for THg (663 mm), and their average THg concentration (0.047 ppm) was also much lower. The sturgeon size-classes with the largest range in THg concentration were entirely missing from the sample analyzed for diet composition. Also, our interpretation of the results is somewhat speculative because analysis of diet composition provides just a single point in time measurement, which may not be indicative of longer-term feeding preferences of individuals. Additional information on feeding patterns of sturgeon spanning a larger size range and measurements of THg in prey items is needed to more robustly evaluate trophic effects on THg concentrations.

Supplementary material

Supplementary material is available online at Environmental Toxicology and Chemistry.

Data availability

Data used in this report are the property of and were provided by the Coordinated Aquatic Monitoring Program (CAMP), Manitoba and/or Manitoba Hydro, and/or the Keeyask Hydropower Limited Partnership under license/by permission. Data will be shared under reasonable request to the corresponding author with permission of the Coordinated Aquatic Monitoring Program (CAMP), Manitoba and/or Manitoba Hydro and/or the Keeyask Hydropower Limited Partnership.

Author contributions

Wolfgang Jansen (Conceptualization, Data curation, Investigation, Methodology, Writing—original draft, Writing—review & editing) and James Aiken (Data curation, Formal analysis, Methodology, Software, Writing—review & editing)

Funding

Funding for manuscript preparation was provided by Manitoba Hydro’s Lake Sturgeon Stewardship and Enhancement Program.

Conflicts of interest

The authors declare no conflicts of interest. The Coordinated Aquatic Monitoring Program (CAMP), Manitoba and/or Manitoba Hydro and/or the Keeyask Hydropower Limited Partnership are not responsible for the analysis or conclusions of this report.

Acknowledgments

We thank all anonymous reviewers for their guidance and comments on earlier manuscript drafts. When required by provincial legislation, fish were collected under Manitoba scientific collection permits.

References