Evaluation of Organochlorine Pesticides and Sex Steroids in Lower Niagara River Lake Sturgeon Free

Lake sturgeon are endemic to the Laurentian Great Lakes, upper Mississippi River, and Hudson Bay drainages in North America (Scott and Crossman 1973). Fisheries overexploitation and habitat degradation led to a range-wide decline of lake sturgeon populations by the mid-1900s, followed by increased conservation interest in this species (Carlson 1995). Restoration and management efforts over the past several decades have included the closure of lake sturgeon fisheries, the initiation of stocking programs in both the United States and Canada, and the designation of lake sturgeon as threatened in New York state and the province of Ontario (Peterson et al. 2007; Chalupnicki et al. 2011). More recently, the conservation status of lake sturgeon has improved across its range, though many populations remain imperiled (Jelks et al. 2008).

Historically, the lower Niagara River supported a large spawning population of lake sturgeon that by 1980 had declined to near-extirpation (Carlson 1995). By the early 2000s, a population assessment by Hughes et al. (2005) found evidence of a small naturally reproducing population composed primarily of juvenile to subadults. As part of continued population monitoring efforts for lower Niagara River lake sturgeon, determination of the sex and stage of these animals is highly desirable, especially as year classes initially documented by Hughes et al. (2005) reach maturity. Information about the reproductive status of these fish would benefit managers by informing quantitative population assessments and management decision-making. To address this need, we initiated a study to develop a sex and stage classification function based on blood plasma sex steroid concentrations for the lower Niagara River lake sturgeon population following methods described by Webb et al. (2002) and Allen et al. (2009). After developing such a classification function, blood can be collected from captured sturgeon prior to their live release, and sex and stage can be assigned based on the concentration of sex steroids in the blood plasma.

A potential confounding factor of sex-steroid–based reproductive classification is the disruption of typical reproductive physiology and plasma sex steroid concentrations by environmental contaminants (endocrine disruption). The Niagara River is the site of an Environmental Protection Agency–designated Great Lakes Area of Concern, where industrial and municipal, point-source and non–point-source contamination has significantly degraded water quality. Organochlorine contaminants (OCs) including dichlorodiphenyltrichloroethane (DDT) and its metabolites (dichlorodiphenyldichloroethylene [DDE] and dichlorodiphenyldichloroethane [DDD]) and dieldrin, as well as polychlorinated biphenyls (PCBs), are among a list of persistent toxic chemicals identified for remedial action in the Niagara River by the Niagara River Toxics Management Plan. As of 2005, the recombined whole-water concentrations (dissolved plus particulate phases) of these chemicals were above New York Department of Environmental Conservation water-quality standards at the mouth of the lower Niagara River (Hill and Klawunn 2010). These and other persistent organic pollutants have been shown to adversely impact the reproductive biology of sturgeons and other fishes (Nicolas 1999; Feist et al. 2005; Hinck et al. 2008) and may potentially impact the lower Niagara River lake sturgeon population. This population of lake sturgeon occupies the lower Niagara River during rearing, foraging, and spawning; therefore, exposure to and bioaccumulation of endocrine disrupting OCs and PCBs may have important implications, as much for the efficacy of sex and stage classification methods as for reproductive success and population recovery.

The concentration of OC contaminants found in blood plasma has been shown to correlate significantly with OC concentration in the body tissues of white sturgeon Acipenser transmontanus from the Columbia River (Gundersen et al. 2008), leading to atypical sex- and stage-specific sex steroid concentrations (Feist et al. 2005). The purpose of this study was to test for the influence of endocrine-disrupting contaminants on circulating sex steroid concentrations of lake sturgeon in the lower Niagara River, and to evaluate the efficacy of sex-steroid–based sex and stage classification for this population. Multiple use of blood samples for analysis of contaminants and coincident sex steroid concentrations is highly desirable, because this offers a nonlethal means of assessing reproductive status, measuring persistent organic pollutants, and identifying uncharacteristic plasma sex steroid concentrations.

We quantified the blood plasma concentration of 17 organochlorine pesticides, testosterone (T), and 17β-estradiol (E2) of lake sturgeon from the lower Niagara River to test whether contaminants were present at high enough levels to cause endocrine disruption, and to test for atypical sex steroid concentrations by sex and stage. We tested the hypothesis that plasma OC concentration in lower Niagara River fish was equal to or greater than levels shown in the literature to cause adverse reproductive effects in sturgeon, and more qualitatively screened for the presence of PCBs. We then tested two hypotheses regarding plasma T and E2 concentrations that, if supported, would indicate no significant endocrine disruption in the lower Niagara River lake sturgeon population: 1) that the occurrence of detectable levels of E2 in male lake sturgeon was no higher than in other populations; and 2) that lake sturgeon T concentrations were no lower than other populations.

Lake sturgeon (n  =  63) were captured from presumed staging and spawning aggregations in the lower Niagara River (43.262722°N, 79.07064626°W) during April and May of 2012 when water temperatures were between 4°C and 11°C. We captured fish using baited set lines following methods of Thomas and Haas (1999) and Hughes et al. (2005). We used heparinized vacutainers to collect 6-mL blood samples from the caudal vasculature, centrifuged blood at 5,000 revolutions per minute for 5 min, and removed and stored plasma at −20°C until analysis. We retrieved 0.5-cm³ gonad tissue samples through either a 1.5-cm incision (n  =  9), or through a larger incision used for implantation of telemetry tags for a separate study (n  =  30) to confirm sex and stage. Incisions were located to the left side of the ventral midline between pectoral fin and anus, and sealed with 2–4 sutures. We held fish in a submerged net-pen along shore until recovery to self-stabilization. We observed no immediate mortality and all fish swam away under their own power.

We analyzed the concentration of 17 OC contaminants in lake sturgeon blood plasma (Table 1) based on methods described by the USEPA (1980; Reference S1, Supplemental Material) and Gundersen et al. (2008). We transferred thawed plasma samples (5 mL) to culture tubes, and extracted them with 8 mL of hexane (pesticide grade) using a rotary mixer (50 revolutions per minute for 2 h). Samples were centrifuged at 2,000 revolutions per minute for 5 min to separate the organic phase from the aqueous phase, and the organic phase was separated and dried with anhydrous sodium sulfate and reduced in volume using a warm-water bath and a stream of pure nitrogen. We analyzed concentrated plasma extracts using a Varian CP-3800 gas chromatograph equipped with a 63Ni electron capture detector, a CP-8200 AutoSampler, a Star Chromatography Workstation (version 5), and a SPB-608 fused silica capillary column (30 mm × 0.25 mm × 0.25 µm film thickness, Supelco, Bellefonte, PA). Gas chromatographic conditions were as follows: carrier gas helium (1.5 mL/min), makeup gas nitrogen, detector temperature 300°C, injector temperature 290°C, and oven temperature 150°C (4 min) to 290°C (10 min) at 8°C/min. Contaminants were quantified from individually resolved peak areas with corresponding peak areas of external standards (AccuStandard). Quality assurance measures included the analysis of reagent blanks, duplicates, and matrix spike samples. Percent recoveries for all of the 17 OC contaminants in matrix spikes ranged between 87% and 105%; therefore, we did not correct sample extracts for percent recovery. Method detection limits for individual chlorinated pesticides were 2 ng/mL (wet weight). All plasma OC concentrations are expressed in parts per billion (ng/mL wet weight).

Our extraction and chromatography methods would also result in the extraction and chromatographic signal detection of PCBs from blood plasma (Feist et al. 2005). Though we did not attempt to quantify the concentration of PCBs, as we did with OCs, we visually compared plasma chromatographs against chromatographs of 16 PCB congener standards from highest priority PCB congener groups identified by McFarland and Clarke (1989). The presence of PCB peaks would indicate whether further investigation into these compounds might also be necessary.

We embedded gonadal tissue in paraffin, sectioned it at 5 µm, and stained it by Periodic Acid Schiff stain (Luna 1968) for histological assessment of sex and stage. We examined slides under a compound microscope (10–100×, Leica DM2000), and scored the germ cells for stage of maturation according to Webb and Erickson (2007). Histology indicated that tissue samples for 29 fish contained germ cells, whereas 11 tissue samples contained no germ cells. For the purposes of this paper, we defined mature males as males with spermatozoa, mature females as those with late vitellogenic ovarian follicles, and immature females as those with oocytes in the previtellogenic stage.

The steroids T and E2 were extracted from plasma following the method of Fitzpatrick et al. (1987). Briefly, we extracted 100 µL of plasma twice with 2 mL of diethyl ether. We vortexed tubes vigorously with ether, and removed the aqueous phase by snap-freezing in liquid nitrogen. We dried combined extracts under a stream of nitrogen, resuspended them in 1 mL of phosphate-buffered saline with gelatin, and assayed 50 µL for each steroid. We determined recovery efficiencies for all steroids (T  =  92–94% and E2  =  79–81%) by adding tritiated steroids to tubes containing plasma (n  =  4), which we extracted as described above. We corrected all steroid assay results for recovery.

Plasma concentrations of T and E2 were measured by radioimmunoassay as described in Fitzpatrick et al. (1986) and modified by Feist et al. (2005). We analyzed all samples in duplicate. We used a slightly more concentrated charcoal solution (6.25 g charcoal and 4.0 g dextran/L phosphate-buffered saline with gelatin) for all assays to reduce nonspecific binding. The lower limit of detection was 0.1 ng/mL for the T assay and 0.16 ng/mL for the E2 assay. The intra- and interassay coefficients of variation for all assays were <5% and 10%, respectively. Steroid levels were validated by verifying that serial dilutions were parallel to standard curves.

We compared circulating blood plasma contaminant concentration of lower Niagara River lake sturgeon against thresholds of blood plasma contaminant concentrations derived from relationships between plasma OC and sex steroid concentrations in Columbia River white sturgeon, published in studies using comparable contaminant and steroid analysis methods (Feist et al. 2005; Gundersen et al. 2008). For analysis, we grouped OCs into two categories: total DDT (sum of DDD, DDE, and DDT) and Pesticide (Table 1 compounds minus DDD, DDE, and DDT), following the convention used by Feist et al. (2005). Feist et al. (2005) reported androgen suppression in white sturgeon males at liver concentrations (reported in ug of contaminant per g of lipid in liver tissue) of >9.5 ug/g lipid total DDT and >5.6 ug/g lipid Pesticide, and at gonad concentrations of >11.6 ug/g lipid total DDT and >3.7 ug/g lipid Pesticide. We converted lipid and gonad tissue thresholds from Feist et al. (2005) into blood plasma pesticide-screening benchmarks (Table 2) using a linear regression equation between blood plasma total DDT concentration (ng/mL) and tissue concentrations (ug/g lipid) reported in Gundersen et al. (2008). We assumed that relationships among tissue concentrations of total DDT from Gundersen et al. (2008) are identical to those of the Pesticide group. We then tested whether plasma concentrations of total DDT and Pesticide in our lake sturgeon were below the most stringent of these benchmarks using single-sample student's t-tests (1-tailed, α  =  0.05). Statistical tests were conducted on the natural logarithm of plasma contaminant concentrations, which were normally distributed.

We conducted a literature search for studies using comparable methods to analyze plasma T and E2 concentrations in sturgeon species during spring and compared sex steroid information by sex and stage among populations, including the lower Niagara River lake sturgeon population herein. Five sturgeon populations from other studies using common analytical methods were included for comparison: Columbia River white sturgeon from Webb et al. (2002; known impacted by contaminants), Saint Clair River lake sturgeon from Craig et al. (2009; unknown contaminant impacts), Winnipeg River lake sturgeon from Allen et al. (2009; presumed nonimpacted), Lake Winnebago lake sturgeon from Allen et al. (2009; presumed nonimpacted), and Namakan Reservoir lake sturgeon from Shaw et al. (2012; presumed nonimpacted). The data we gathered for this meta-analysis included the mean, standard deviation, and number of fish with detectable steroid levels by sex and stage for each population. To investigate estrogenic effects, we tested whether mature male and immature female lower Niagara River lake sturgeon had elevated prevalence of detectable levels of E2 (proportion of samples with detectable E2) compared with prevalence of detectable E2 in mature males and immature females from other populations using a chi-square test. We then tested for differences in plasma T concentration among sturgeon populations using analysis of variance (ANOVA). If significant differences among populations were found, we investigated which populations were different from the lower Niagara River lake sturgeon population using Bonferroni-adjusted pair-wise comparisons. There was some skewness in the distribution of plasma T concentration in our lower Niagara River immature female class, though plasma T was normally distributed among mature fish classes. We decided to use parametric statistics to compare plasma T across all groups because ANOVA is robust to some nonnormality, and because this allowed direct and consistent comparisons with mean and standard error of plasma steroid concentrations published in other studies in our meta-analysis.

Blood plasma samples were collected from 63 lake sturgeon during spring of 2012 (Table S1, Supplemental Material). Twenty-one fish had detectable levels of DDE, averaging 18.24 ng/mL, with a standard deviation of 9.81 ng/mL and ranging from 7.20 to 47.82 ng/mL; one fish had 15.38 ng/mL γ-BHC and no other contaminants were detectable. Total DDT and Pesticide benchmarks used for contaminant screening were 48.89 ng/mL and 27.65 ng/mL, respectively. The natural logarithm of DDE, in our case also representing total DDT, was significantly below the natural logarithm of our initial screening benchmark (t  =  −9.526, n  =  21, P < 0.001). No individual fish had DDE concentrations above the benchmark of 48.89 ng/mL. The only other OC compound we detected was γ-BHC (in the Pesticide group), which was found in only one fish at a lower concentration than our 27.65 ng/mL screening benchmark. We attempted to identify any chromatograph peaks that did not correspond to OC pesticide standards in Table 1 by gas chromatography–mass spectrometry, but no other compounds were present in high enough concentration for our methods to detect (Figure 1). Qualitative analysis of chromatographs did not identify any peaks that corresponded to standards peaks from toxicologically significant PCBs (Figure 1).

Because we could not positively identify any immature male lake sturgeon in the lower Niagara River, we could only test for elevated E2 levels in immature females and mature males. Further, because no verified immature females were reported in Craig et al. (2009) and because the maximum mature male E2 concentration from Craig et al. (2009) was well below the detection limits of mature male sturgeon from other studies, only the remaining five populations (lower Niagara River, Namakan Reservoir, Winnipeg River, Lake Winnebago, and Columbia River) were comparable. One of 12 immature female lake sturgeon from the lower Niagara River had detectable levels of E2 using radioimmunoassay methods, while the Columbia River white sturgeon population was the only other population with detectable E2 (0.2 ng/mL lower detection limit; Webb et al. 2002) in immature females (Table 3). A chi-square test found no significant differences among the four lake sturgeon populations (χ²₃  =  7.324, P  =  0.062). When the Columbia River white sturgeon population was included, differences among proportions were much more apparent (χ²₄  =  18.360, P  =  0.001), reflecting the slightly more elevated rates of E2 detection in immature females of that population. Results were similar among mature male sturgeon. Two of 10 mature male lake sturgeon from the lower Niagara River had detectable E2, while E2 was detected in two of 11 Namakan Reservoir lake sturgeon (0.2 ng/mL lower detection limit; Shaw et al. 2012) and 21 of 24 Columbia River white sturgeon (Table 3). There was no significant difference in the proportion of detectable E2 in mature males among the four lake sturgeon populations (χ²₃  =  7.401, P  =  0.060). As with immature females, when mature male Columbia River white sturgeon were included, the elevated E2 levels in that population drove statistically significant among-population differences in the proportion of mature male fish with detectable E2 (χ²₄  =  53.531, P < 0.001).

Analysis of variance and pair-wise post hoc test results indicated that the lower Niagara River lake sturgeon population differed in plasma T concentration, but that it did not have reduced plasma T levels (Figure 2). Mature male and female fish had much higher levels of T than the other populations compared (F_5,83  =  19.504, P < 0.001, F_5,77  =  65.654, P < 0.001, respectively). Significant differences in immature female T concentrations were found among populations (F_4,204  =  12.127, P < 0.001), though lower Niagara River lake sturgeon grouped with the Winnipeg River population, and was significantly different from Lake Winnebago, Namakan Reservoir, and Columbia River sturgeon (Bonferroni-adjusted α  =  0.005; Figure 2). In the immature male class, there were no fish from the lower Niagara River or the Saint Clair River to compare, and there were no significant differences among the remaining populations (F_3,123  =  0.958, P  =  0.415).

Webb and Doroshov (2011) stated the importance of investigating the influence of environmental contaminants on a per-population basis while using blood plasma T and E2 concentration assess population structure (sex and stage), or else risk high misclassification rates. For instance, negative relationships between plasma T concentration and those of liver and gonad contaminants in male white sturgeon were noted by Feist et al. (2005), resulting in a marked reduction in plasma T levels at DDT and OC concentrations between 3 and 12 ug/g lipid, which could strongly affect steroid-based classification. Similar relationships that have been established based on plasma contaminant concentration and that of other tissues (Bernhoft et al. 1997; Henriksen et al. 1998; Keller et al. 2004), but we used androgen suppression thresholds from Feist et al. (2005) and converted them to blood plasma concentration using results from Gundersen et al. (2008) because they were sturgeon-specific. Other benchmarks for endocrine disruption in aquatic animals include 0.23 ug/g DDE, 0.0256 ug/g DDT, and 0.15 ug/g PCBs by total body wet weight (Beckvar and Lotufo 2011; Wenning et al. 2011), but the lack of established relationships between whole-body wet mass and blood plasma contaminant concentration prevents the use or comparison of such benchmarks for blood plasma hormone analysis.

The low levels of OCs in our blood plasma samples relative to those reported in Gundersen et al. (2008) indicate that OCs likely do not strongly influence the reproductive physiology of this lake sturgeon population in the lower Niagara River. The average plasma DDE concentration for lake sturgeon in the lower Niagara (18.24 ng/mL) was similar to the concentration of total DDT found in Columbia River white sturgeon from the estuary, John Day Reservoir, and The Dalles Reservoir, but fish caught in Bonneville Reservoir had much higher plasma DDT levels than Niagara River fish (Gundersen et al. 2008). White sturgeon captured in Bonneville Reservoir were found to have decreased condition factor, gonad size, plasma androgens, and plasma triglycerides, as well as an increased incidence of gonadal abnormalities associated with these higher contaminant concentrations. More broadly, compounds such as DDE and other DDT congeners have been shown to have a negative impact on blood plasma vitellogenin, E2, and 11-ketotestosterone on black bass Micropterus spp. and common carp Cyprinus carpio, though effects differed by sex and species (Hinck et al. 2008).

The lack of significantly large PCB-associated peaks from GC chromatographs indicates that PCBs are also unlikely to affect the lower Niagara River lake sturgeon population (Figure 1). In the Detroit River, lake sturgeon plasma DDT concentration (2.07 ng/g) was lower than the average of lower Niagara River fish and total PCB concentration (10.41 ng/g), though low, was higher than total DDT (Li et al. 2003). Though Li et al. (2003) only analyzed samples from a single lake sturgeon, and despite low OC and PCB concentrations found in its tissues, Li et al. (2003) found significant levels of several hydroxylated PCB congeners, a diverse class of potentially endocrine-disrupting metabolic breakdown products of PCBs (Buckman et al. 2006). Though the response of adult lake sturgeon reproductive biology to these compounds is unknown, if a more in-depth analysis of Niagara River lake sturgeon contaminant burdens were to be undertaken, investigation of hydroxylated PCB congeners may be informative.

Our comparison of plasma T and E2 levels among the lower Niagara River lake sturgeon population and other sturgeon populations from the literature indicated two things: 1) E2 levels in immature female and mature male fish did not appear elevated, especially when compared with those of Columbia River white sturgeon (a known pollutant-impacted population); and 2) the lower Niagara River population did not exhibit reduced T. In fact, the lower Niagara River population appeared to have elevated overall T levels when compared with other sturgeon populations. Despite our attempts to narrow search criteria to minimize the influence of individual-level or temporal variation in plasma hormone levels, the strong influences of time of year (Webb and Erickson 2007) and sex and maturity (Webb et al. 2002) no doubt differentially affect hormone comparison results, due to among-population variation in sample size, environmental conditions, and sturgeon capture dates. However, if endocrine-disrupting compounds were broadly affecting the population, we may expect to find greater-than-average rates of plasma E2 detection, and lower-than-average plasma T. Androgen levels did not appear suppressed, as was found in Feist et al. (2005), and E2 detection rates did not appear abnormal; therefore, sex steroid levels indicated no evidence of broad endocrine disruption. Though plasma T levels in lake sturgeon from the lower Niagara River were unexpectedly high, additional research is necessary to investigate whether this is response to endocrine disrupting compounds, or simply the result of inter-population variability.

Studies have shown somewhat high levels of variation among laboratories when quantifying absolute T and E2 levels of common reference samples (McMaster et al. 2001; Feswick et al. 2014). We attempted to follow recommendations by Feswick et al. (2014) in our comparison of absolute sex steroid concentrations among independent studies by limiting inclusion of studies to those with published inter- and intraassay coefficients of variation <15%, and by limiting methodological differences among studies we compared. Each of the five studies we used to compare sex steroids reported inter- and intraassay coefficients of variation of <15%, with the exception of interassay variation in T of 16% from Craig et al. (2009). Laboratory-specific methodological differences among the studies we compared were minimal, as radioimmunoassay was used in each case. Analyses from three of our five studies were conducted by Molly Webb at the U.S. Fish and Wildlife Service Bozeman Fish Technology center, in Bozeman, Montana, USA (Allen et al. 2009; Shaw et al. 2012; this study) and one other was conducted by Molly Webb at Oregon State University, in Corvallis, Oregon, USA (Webb et al. 2002). The low T and E2 concentrations of mature male and female lake sturgeon from the Saint Clair River (Craig et al. 2009) may be due to a laboratory effect, but we do not believe this would adversely affect our results.

Our results indicated the presence of low levels of DDE, the general absence of other investigated OCs, and the apparent absence of important PCB congeners in the blood plasma of lake sturgeon from the lower Niagara River. However, we cannot conclude whether lake sturgeon tissue contaminant burdens or Niagara River contaminant loadings are without impact. In white sturgeon, the concentration of contaminants in gonad and liver tissues has been shown to be orders of magnitude greater than that found in the blood plasma (Gundersen et al. 2008), and it would not be unexpected to find fish with moderately high levels of whole-body contaminants without detectable levels of OCs in the blood. The purpose of our study was to screen for plasma contaminant levels that may disrupt reproductive physiology, which we define as having plasma OC concentrations greater than or equal to those of impacted populations described in the literature. We stop well short of quantifying contaminant body burdens, tissue allocation, persistence in edible tissues, or contaminant loadings in likely habitats.

Much of the concern over OC and PCB contaminants in animal populations relates to reproductive health and persistence (Fry 1995; Monosson 2000). The absence of significantly high levels of persistent bioaccumulative contaminants such as OCs and PCBs is a positive sign for any animal population, but especially for those considered threatened or endangered. The few contaminants we quantified in blood plasma of lower Niagara River lake sturgeon did not appear great enough to adversely affect circulating plasma steroid levels. These results indicate that this population is likely not significantly affected by the presence of OC or PCB contaminants. With the exception of T in mature fish, blood plasma sex steroid concentrations were comparable to those of other sturgeon populations and neither sex steroid among all categories appeared influenced by contaminants. In the absence of discernable contaminant impacts on circulating sex steroids, reproductive biology of lake sturgeon in the lower Niagara River can be expected to exhibit patterns similar to other healthy wild populations (Allen et al. 2009; Shaw et al. 2012), which should allow low error rates associated with plasma steroid-based classification of sex and maturation and may simplify future management efforts for this population.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Reference S1. United States Environmental Protection Agency [USEPA]. 1980. Manual of analytical methods for the analysis of pesticides in humans and environmental samples. Research Triangle Park, North Carolina: U.S. Environmental Protection Agency, Heath Effects Research Laboratory, USEPA No. 600/8-80-038.

Found at DOI: 10.3996/072013-JFWM-048.S1 (20 MB PDF).

Table S1. Sex steroid and organochlorine contaminants data from analysis of lower Niagara River lake sturgeon blood plasma, collected during April and May of 2012. Variables are an individual lake sturgeon identifier (Fish), the maturation stage of the fish (Stage), the sex of the fish (Sex), the blood plasma concentration of testosterone in ng/mL where ‘nd’ denotes nondetectable levels of T and blank cells denote no data (T), the blood plasma concentration of 17β-estradiol in ng/mL where ‘nd’ denotes nondetectable levels of E2 and blank cells denote no data (E2), the blood plasma concentration of γ-BHC where ‘nd’ denotes nondetectable levels of γ-BHC (g-BHC), and the blood plasma concentration of p,p′-DDE where ‘nd’ denotes nondetectable levels of p,p′-DDE (DDE).

Found at DOI: 10.3996/072013-JFWM-048.S2 (10 KB XLSX).

We thank Peter Allen, Stephanie Shaw, and Steven Chipps for providing access to their data and for reviewing this manuscript. We also thank Robert Muth and two anonymous reviewers for their helpful comments. This paper was much improved by their contributions. Mariah Talbott, Eli Cureton, Michelle Casto-Yerty, Betsy Trometer, Todd Duvall, and many others helped assist with fieldwork and sample processing. Mention of specific products does not constitute endorsement by the U.S. Government. This study was funded by the Great Lakes Restoration Initiative.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.