Ontogenetic patterns in bluefish (Pomatomus saltatrix) feeding ecology and the effect on mercury biomagnification

Abstract

In this study, bluefish (Pomatomus saltatrix; age 0–7, n = 632) and their prey (forage fish, macroinvertebrates, zooplankton; n = 2,005) were collected from the Narragansett Bay estuary (RI, USA), and total Hg concentration was measured in white muscle and whole-body tissues, respectively. Bluefish Hg concentrations were analyzed relative to fish length, prey Hg content, and ontogenetic shifts in habitat use and foraging ecology, the latter assessed using stomach content analysis (n = 711) and stable nitrogen (δ¹⁵N) and carbon (δ¹³C) isotope measurements (n = 360). Diet and δ¹³C analysis showed that age 0 bluefish consumed both benthic and pelagic prey (silversides, sand shrimp, planktonic crustaceans; δ¹³C = − 16.52‰), whereas age 1 + bluefish fed almost exclusively on pelagic forage fish (Atlantic menhaden, herring; δ¹³C = − 17.33‰). Bluefish total Hg concentrations were significantly correlated with length (mean Hg = 0.041 and 0.254 ppm wet wt for age 0 and age 1 + bluefish, respectively). Furthermore, Hg biomagnification rates were maximal during bluefish early life stages and decelerated over time, resulting in relatively high Hg concentrations in age 0 fish. Rapid Hg accumulation in age 0 bluefish is attributed to these individuals occupying a comparable trophic level to age 1 + bluefish (δ¹⁵N = 15.58 and 16.09‰; trophic level = 3.55 and 3.71 for age 0 and age 1 + bluefish, respectively), as well as juveniles having greater standardized consumption rates of Hg-contaminated prey. Finally, bluefish larger than 30 cm total length consistently had Hg levels above the U.S. Environmental Protection Agency criterion of 0.3 ppm. As such, frequent consumption of bluefish could pose a human health risk, and preferentially consuming smaller bluefish may be an inadequate strategy for minimizing human dietary exposure to Hg. Environ. Toxicol. Chem. 2011; 30:1447–1458. © 2011 SETAC

INTRODUCTION

Mercury is a toxic environmental contaminant that accumulates in aquatic ecosystems, and its pervasiveness has led to health concerns for wildlife and humans 1. The organic form of Hg, methylmercury (MeHg), is of particular concern because it biomagnifies in aquatic food webs, and thus, concentrates in the tissues of high trophic level organisms 2, 3. Moreover, chronic exposure to MeHg can have adverse effects on biota, including neurological and cardiovascular disorders 3. From a human health perspective, MeHg exposure is mainly the result of dietary intake of contaminated fish and shellfish 1, 4, and MeHg accounts for more than 95% of the total Hg in fish muscle tissue 5–7. Furthermore, most fish consumed by the human population are of marine origin 8, thus underscoring the importance of research focused on MeHg biomagnification in coastal marine fisheries.

Estuarine and coastal sediments are repositories for Hg, receiving substantial external inputs from both direct atmospheric deposition and watershed transport 9. Aquatic sediments are also a principal location for methylation, the transformation of inorganic Hg to MeHg, which is mediated by sulfate-reducing bacteria under anoxic conditions 10, 11. The methylation process is favored in estuarine and coastal sediments because these systems often have high loadings of anthropogenically derived Hg 12–14, host active bacterial communities 15, and maintain biogeochemical conditions conducive to MeHg production, such as frequent anoxia and elevated levels of organic matter and sulfate 16. Recent investigations further suggest that estuarine and coastal sediments are the primary source of MeHg to resident biota 10, 17, and fish inhabiting contaminated environments experience increased MeHg concentrations 18, 19.

Methylmercury uptake and biomagnification in fish depends on Hg input to the ecosystem and biogeochemical conditions that affect MeHg production and mobilization 20. Moreover, physicochemical and biological variables influence the initial incorporation of MeHg into the food web and its transfer through successive trophic levels 20. Fish MeHg concentrations, for example, are positively correlated with body size and age when dietary intake of the contaminant exceeds depuration rates 21. Diet history and feeding ecology also impact MeHg concentrations, such that MeHg accumulation rates are increased in fish feeding at higher trophic levels 22. Accordingly, MeHg biomagnification in fish needs to be examined concurrently with intraspecies life history traits, such as body size, age, and ontogenetic shifts in dietary preference and habitat use 23. Research focused in this manner will improve predictions of fish MeHg concentrations and assist in the management of this important dietary resource.

Current research in the Narragansett Bay estuary (RI, USA) is focused on toxicant concentrations (total Hg and MeHg) in surface sediments and biota, with the latter including fishes that are of recreational and commercial value 6, 7. The products of this synoptic examination, in turn, provide a meaningful assessment of fish Hg contamination that constitutes an environmental and human health risk. This report is an extension of the ongoing research in the Narragansett Bay and specifically addresses the biomagnification of total Hg in bluefish (Pomatomus saltatrix), a highly migratory, coastal marine species that constitutes a premier fishery throughout its geographic distribution 24. The specific objectives of this investigation were threefold: to measure total Hg concentrations in bluefish muscle tissue and analyze the results relative to body size and spatiotemporal patterns in habitat use; to examine the effects of ontogenetic shifts in diet and trophic ecology on bluefish Hg contamination, as determined from conventional stomach content analysis and stable isotope measurements; and, given the economic value of bluefish, to evaluate Hg contamination in this species relative to the U.S. Environmental Protection Agency (U.S. EPA) criterion for the safe consumption of fishery resources.

MATERIALS AND METHODS

Bluefish life history and habitat use

Bluefish are dominant apex predators in estuarine and coastal habitats during multiple life history stages 25, and thus are potentially exposed to high levels of environmental contaminants, including Hg 18, 19. Bluefish spawning and the subsequent development of early life stages (planktonic eggs and larvae) occur in offshore, continental shelf waters 25, 26. Once achieving a body size of approximately 40 to 70 mm total length (TL) (age ∼35–75 d), juvenile bluefish recruit to estuaries and coastal beaches 25, 26. The inshore recruitment of juvenile bluefish co-occurs with an ontogenetic diet shift from copepods to piscine prey 27, 28. This early onset of piscivory results in greatly increased juvenile growth rates (> 2 mm TL/d), which continue into the adult life stage (maximum weight and age ∼6.8 kg and 12 year) 25, 29. Similar to age 0 juveniles, the use of inshore habitats by adult bluefish is largely dictated by availability of prey (forage fish and cephalopods) 26.

Sample collection and preparation

Bluefish (size range = 6.5–75.0 cm TL) were collected from the Narragansett Bay from May through October 2006 to 2009, using seines and hook and line (Fig. 1). During this study and concurrent investigations 6, 7, common prey of juvenile and adult bluefish were also collected from the estuary using seines and plankton nets. Targeted prey of bluefish included finfish (Atlantic silverside [Menidia menidia], river herring [Alosa sp.], bay anchovy [Anchoa mitchilli], Atlantic menhaden [Brevoortia tyrannus], scup [Stenotomus chrysops], striped killifish [Fundulus majalis], and winter flounder [Pseudopleuronectes americanus]), squid (Atlantic long-fin squid [Loligo pealei]), macrocrustaceans (sand shrimp [Crangon septemspinosa] and grass shrimp [Palaemonetes spp.]), and planktonic crustaceans (copepods and crab/shrimp zoea) 26, 30. Bluefish and prey were immediately placed on ice after capture and frozen at −20 °C in the laboratory. In the laboratory, samples were partially thawed and measured for whole-body wet weight (g) and length (cm; total length [TL] for fish and sand shrimp, carapace length for grass shrimp, and mantle length for squid). Bluefish age was also estimated from age–length relationships reported in the literature 31, 32, and bluefish were classified as age 0 or age 1 + , using 25 cm TL as a demarcation value. According to temporal patterns in bluefish size-at-capture, age 0 fish were determined to be representative of the spring-spawned cohort 32, 33.

Figure 1. Map of the Narragansett Bay estuary (Rhode Island, USA) with points denoting collection sites of bluefish (age 0 and age 1 +) and prey.

After size measurements, bluefish and prey were processed and preserved as follows. For age 0 bluefish (n = 478), epaxial and hypaxial muscle tissue was excised from the posterior margin of the operculum to the caudal peduncle (one filet per left and right side of the fish, each ∼2 g wet wt; Fig. 2A). A random sample of age 0 bluefish (n = 120) were also treated as whole bodies to determine whether Hg concentrations in the entire body were indicative of the contaminant present in isolated muscle tissue. For age 1 + bluefish (n = 154), white muscle tissue was removed from above the operculum (D₀ = 1 biopsy per left and right side of the fish, each ∼ 2 g wet wt; Fig. 2B). To verify that the Hg concentration of the D₀ biopsy was consistent with the whole-body filet, a subsample of age 1 + bluefish (n = 14) had additional biopsy specimens excised from the dorsolateral tissue (D_1-3 and L_1-3; 6 biopsies per left and right side of the fish; Fig. 2B). The stomachs of age 0 and age 1 + bluefish were also removed for diet analysis, as described later. Prey collected in the field (n = 32–588 per species) were processed and analyzed as whole bodies. Tissue and whole-body samples were weighed (g wet wt), freeze-dried for 48 h, reweighed (g dry wt), and homogenized using a mortar and pestle. Samples were then screen-sifted using a 1-mm mesh to remove skin and scales when appropriate, and stored at room temperature in clear borosilicate 40-ml glass vials.

Figure 2. Locations of muscle tissue excision in age 0 and age 1 bluefish (A and B, respectively). Total Hg concentrations in age 0 and age 1 + bluefish (C and D, respectively) as a function of tissue type (age 0: whole body vs muscle filet, n = 516; age 1 + : muscle biopsies, n = 14, mean ± standard error). Each muscle filet or biopsy represents the average Hg content of tissue removed from the left and right side of the fish. Age 0 bluefish were grouped into 1-cm length increments for graphical purposes only, and least-squares exponential regression models were fit to the full data set.

Stomach content and diet analysis

Prey extracted from the stomachs of age 0 and age 1 + bluefish were identified to the lowest practical taxon using stereoscopic microscopes, and each prey taxon was weighed (g wet wt) with analytical balances 34. Where possible, prey items were enumerated, and individual lengths were measured as defined previously. The contribution of each prey taxon to the overall diet of age 0 and age 1 + bluefish was expressed as the frequency of occurrence (%F) and percent weight (%W). Frequency of occurrence was calculated as the number of stomachs containing a specific prey taxon divided by the total number of stomachs with prey contents, whereas percent weight was calculated as the weight of a specific prey taxon divided by the total weight of all identifiable prey types. The relative importance of each prey taxon to the diet of bluefish was also assessed using the alimentary index (%IA)

• (1)

where IA is calculated for each prey taxon i as the product of %F_i and %W_i, and n is the total number of prey taxa identified in the stomachs of age 0 and age 1 + bluefish.

Diet diversity for age 0 and age 1 + bluefish was estimated using an index of niche breadth (B) 35:

• (2)

where p_i is the proportion by weight of prey category i, and n is the number of prey categories. The defined prey categories were benthic fish (flounder, goby [Gobiosoma spp.]), benthic invertebrates (shrimp), pelagic fish (anchovy, herring, killifish, menhaden, scup, silverside), and pelagic invertebrates (squid, crab/shrimp zoea). In this exercise, the niche breadth index (B) can vary between 1 and 4, where 1 denotes a singular prey category in the diet of bluefish, and 4 represents an equal contribution of each prey category. The stomach content weights of unidentifiable fish and crustaceans were not included in the analysis.

Dietary overlap between age 0 and age 1 + bluefish was analyzed using Schoener's index 36. Unidentifiable stomach contents were excluded from this analysis, and %W of each prey category was used to assess diet overlap, such that

• (3)

where α approximates the degree of resource overlap, p_ij is the proportion by weight of the ith resource (prey type) used by age 0 bluefish (j), and p_ik is the proportion by weight of the ith resource used by age 1 + bluefish (k). The Schoener's index ranges between 0 and 1, and values greater than 0.6 denote biologically significant overlap in the utilization of a specific prey resource 37.

Stable isotope analysis

Nitrogen (¹⁵N/¹⁴N) stable isotope signatures were used to quantify the trophic level of target organisms as a function of their time-integrated feeding history 38. Carbon (¹³C/¹²C) stable isotope signatures were also used as indicators of the initial carbon source to the estuarine food web (e.g., pelagic and benthic primary production) 39. Isotope measurements of a subsample (∼1 mg dry wt) of age 0 bluefish (muscle filet; n = 20), age 1 + bluefish (D₀ biopsy; n = 20), and prey (whole body; n = 20–85) were performed by the Boston University Stable Isotope Laboratory (Boston, MA, USA) 7. Briefly, automated continuous-flow isotope ratio mass spectrometry was used to quantify isotopic signatures, and ratios of ¹⁵N/¹⁴N and ¹³C/¹²C were expressed as the relative permil (‰) difference between the samples and international standards (atmospheric nitrogen,¹⁵N_air, and Vienna Pee Dee Belemnite, ¹³C_V-PDB, respectively)

• (4)

where X = ¹⁵N or ¹³C and R = ¹⁵N/¹⁴N or ¹³C/¹²C. The precision of the continuous-flow isotope ratio mass spectrometry method, as determined by the analysis of internal reference material (peptone and glycine), was 0.2‰ and 0.4‰ for nitrogen and carbon, respectively.

Nitrogen isotope signatures were used to estimate the trophic position of bluefish and prey, using the following equation modified from Piraino and Taylor 7:

• (5)

where 2 is the assumed trophic position of a defined primary consumer (i.e., blue mussel Mytilus edulis), δ¹⁵N_consumer and δ¹⁵N_mussel are the respective nitrogen isotope signatures of a consumer of interest and the blue mussel, and δ¹⁵N_enrichment is the constant nitrogen isotope enrichment (‰) per trophic level (2.9‰ and 3.2‰ for prey whole bodies and bluefish white muscle tissue, respectively) 40. The blue mussel was selected as the reference primary consumer because it is phytoplanktivorous 41, and a concurrent investigation measured the δ¹⁵N signature of blue mussels collected from the Narragansett Bay (δ¹⁵N = 10.62‰; n = 20) 7.

Mercury analysis

Total Hg concentrations of bluefish and prey were measured using a direct Hg analyzer (DMA-80), with a detection limit of 0.01 ng Hg (typical working range: 0.05–600 ng). The instrument was calibrated using standard reference materials (prepared by the National Research Council Canada, Institute of Environmental Chemistry, Ottawa, Canada), and included TORT-1 (lobster hepatopancreas) and DORM-2 (dogfish muscle). Calibration curves were highly significant (mean R² = 1.00; range R² = 0.99–1.00; p < 0.0001), and the recovery of the TORT-1, DORM-2, and PACS-2 (marine sediment) standard reference materials ranged from 96.1 to 102.6% (mean = 97.9%). All samples were analyzed as duplicates, and an acceptance criterion of 10% was implemented. Duplicate samples with less than 10% error were averaged for subsequent analysis (mean absolute difference between duplicates ∼4%). Samples with more than 10% error were reanalyzed to achieve the acceptance criterion or were eliminated from further analysis. For additional quality control, blanks were analyzed every 10 samples to assess instrument accuracy and potential drift. Finally, total Hg data were converted to wet weight using a wet/dry ratio measured for all bluefish and prey (water content range = 75.6–79.4%), and hereafter all units are presented as ppm (mg Hg/kg wet wt). Note that total Hg in muscle tissue is composed mostly of MeHg (> 95%) for upper-trophic-level fish 6, 7, 42, justifying the use of total Hg as a proxy for MeHg concentrations. Percent MeHg in organisms feeding at lower trophic levels is variable, however, and total Hg results for prey should be interpreted with caution 43.

Statistical analysis

Initial statistical analyses examined variations in the total Hg concentration of age 0 and age 1 + bluefish as function of tissue type and biopsy location (Fig. 2A, 2B). Specifically, for age 0 bluefish, a two-way analysis of variance model was used to test for differences in Hg bioaccumulation rates between muscle filets and whole bodies (interaction effect between bluefish length and tissue type). The analysis of variance model tested least-squares exponential regression models (semi-logarithmic transformation) fit to sample groups of comparable sizes (bluefish size range = 6.5–15.8 cm TL). For age 1 + bluefish, mean Hg concentrations among dorsolateral muscle biopsies were compared with an analysis of variance model using biopsy location (D_0-3 and L_1-3) as a fixed factor. Note that the Hg content of a muscle filet (age 0 bluefish) or specific biopsy (age 1 + bluefish) represents the average contaminant concentration measured for the left and right side of the fish, and total Hg data were log₁₀-transformed to meet assumptions of normality and homogeneity of variance.

Interspecies differences in total Hg concentrations, isotopic signatures (δ¹⁵N and δ¹³C), and trophic levels were analyzed with one-way analysis of variance models using bluefish age class and prey taxa as fixed factors. Moreover, mean differences in Hg content (prey only), δ¹⁵N, δ¹³C, and trophic levels across 13 levels of bluefish and prey were contrasted with Ryan-Einot-Gabriel-Welsch (Ryan's Q) multiple comparison tests. Direct statistical comparisons of Hg values among bluefish and prey were not made because of the different tissues analyzed between the groups (muscle vs whole-body tissues), and thus the Ryan's Q multiple comparison test was applied to prey taxa only.

Last, stepwise multiple linear regression models were used to examine the effects of several abiotic and biological parameters on bluefish total Hg concentrations and isotopic signatures. The variables included in the regression model were year (2006–2009), date of capture (day of year), place of capture (latitude and longitude; decimal degrees), and bluefish total length (cm; log_e transformed). For a given explanatory variable, the significance level for entry into the regression model was set at p < 0.05.

RESULTS AND DISCUSSION

Bluefish diet and stable isotope analysis

Direct visual analysis of stomach contents affirmed that bluefish foraging habits are age specific and consistent with documented ontogenetic shifts in diet 30. In this study, the stomachs of 590 age 0 and 121 age 1 + bluefish were examined, of which 66.1% and 52.0% contained prey (Table 1). For age 0 bluefish, 10 unique prey taxa were identified in stomach contents (six teleost and four invertebrate taxa), and age 1 + bluefish consumed seven different prey taxa (five teleost and two invertebrate). The broader diet of age 0 bluefish, relative to older conspecifics, was reflected in the niche breadth index (B = 2.06 and 1.05; Table 1). Specifically, age 0 bluefish consumed vertebrate and invertebrate prey from both the pelagic and benthic realm, whereas age 1 + bluefish fed almost exclusively on pelagic forage fish.

Table 1. Stomach contents of age 0 and age 1 + bluefisha

Prey taxon	Age 0			Age 1 +
	%F	%W	%IA	%F	%W	%IA
a Values represent the percent frequency of occurrence (%F), percent wet weight (%W), and alimentary index (%IA; Eqn. 1). Squared brackets indicate major taxon subtotals for %W and %IA. Indices of niche breadth within life stage (B; Eqn. 2) and dietary overlap across life stages (α; Eqn. 3) are also presented.
Fish		[79.0]	[87.7]		[97.5]	[99.8]
Bay anchovy	0.5	0.2	0.00	3.7	1.0	0.10
Goby	0.9	0.8	0.03	0.0	0.0	0.00
Herring	9.3	14.3	4.92	31.5	42.2	35.9
Killifish	3.7	4.2	0.58	0.0	0.0	0.00
Menhaden	0.0	0.0	0.00	50.0	45.2	61.0
Scup	0.0	0.0	0.00	1.9	0.5	0.03
Silverside	34.7	30.4	39.2	5.6	2.8	0.42
Winter flounder	2.3	3.2	0.27	0.0	0.0	0.00
Unidentified fish	44.4	25.9	42.7	14.8	5.8	2.32
Crustaceans		[18.5]	[12.2]		[1.3]	[0.19]
Copepods and crab/shrimp zoea	23.6	6.3	5.52	0.0	0.0	0.00
Grass shrimp	5.1	1.4	0.27	0.0	0.0	0.00
Sand shrimp	18.5	8.9	6.14	5.6	1.3	0.19
Unidentified decapods	3.7	1.9	0.26	0.0	0.0	0.00
Cephalopods		[2.5]	[0.13]		[1.2]	[0.06]
Squid	1.4	2.5	0.13	1.9	1.2	0.06
Stomachs examined (n)		590			121
Stomach with unidentified contents (%)		29.5			9.4
Empty stomachs (%)		33.9			48.0
Niche breadth index (B)		2.06			1.05
Dietary overlap index (α)	0.26

The dominant prey of age 0 bluefish, with respect to frequency of occurrence (%F), were silversides, planktonic crustaceans, and sand shrimp (Table 1). Collectively, these prey were observed in 18.5 to 34.7% of the stomachs analyzed. Teleost fish accounted for the largest percentage by weight of age 0 bluefish stomach contents (%W = 79.0%), followed by crustaceans (%W = 18.5%) and squid (%W = 2.5%). The order of importance of identifiable prey to the diet of age 0 bluefish, as expressed by the alimentary index (%IA), were as follows: silversides, sand shrimp, planktonic crustaceans, and herring. The remaining identified prey had %IA less than 1% and included striped killifish, juvenile winter flounder, grass shrimp, squid, and gobies. For age 1+ bluefish, Atlantic menhaden and herring were the most utilized prey resource, collectively accounting for 81.5, 87.4, and 96.9% of the %F, %W, and %IA, respectively (Table 1). The remaining identified prey, silversides, sand shrimp, bay anchovy, squid, and scup, were of secondary importance (%IA < 1%).

Although there was no indication of significant competition for prey resources between age 0 and age 1 + bluefish (α = 0.26) 37, appreciable levels of dietary overlap occurred for herring, and to a lesser extent, silversides, sand shrimp, and squid. Resource partitioning still occurred among communal prey, however, such that age 0 bluefish consumed smaller individuals within a defined prey taxon (fish and squid < ∼5 cm < ∼4 cm length). This prey size selectivity by age 0 bluefish agrees with prior accounts from field research and laboratory experiments 27, 44, 45. Moreover, results from this study are consistent with previous analyses of the foraging habits of bluefish collected from the western North and Mid-Atlantic, including Maine (USA) inshore waters 46, New York–New Jersey (USA) estuaries 47, 48, 49, Chesapeake Bay and coastal Virginia (USA) 50, 51, and the inner continental shelf 30, 34. Geographic and seasonal differences in bluefish diet are also reported in the literature, however, and are likely attributable to ontogenetic shifts in habitat use and spatiotemporal variations in prey availability 30, 52.

The visual analysis of a predator's stomach contents provides valuable information on diet composition but is limited to only recent foraging activity. Therefore, nitrogen stable isotope (δ¹⁵N) signatures also were used in this study to assess time-integrated feeding history and trophic positioning of bluefish and prey. Mean δ¹⁵N values and trophic levels differed significantly across taxa (Tables 2 and 3). Among prey, mean δ¹⁵N values ranged between 11.91‰ and 14.91‰, corresponding to trophic levels between 2.44 and 3.48 (Table 2). Scup, silversides, and anchovy had significant ¹⁵N enrichment (δ¹⁵N > 14‰), and occupied relatively high trophic positions in the food web (trophic level = 3.30–3.48). Conversely, δ¹⁵N values were significantly depleted for killifish, menhaden, and planktonic crustaceans (δ¹⁵N < 13‰), with calculated trophic levels between 2.44 and 2.75. Spatiotemporal dynamics in environmental conditions (e.g., nutrient loading and nitrogen biogeochemistry) can affect δ¹⁵N signatures of prey, as determined in the Narragansett Bay 53. In the present investigation, however, prey were collected over comparable spatial and temporal scales, and thus relative differences in interspecies δ¹⁵N signatures are attributed to trophic processes.

Table 2. Comparison of total length, wet weight, nitrogen and carbon isotope (δ¹⁵N and δ¹³C) signatures, estimated tropic level, and total mercury (Hg, ppm wet wt) in bluefish and preya

Species	n	Length (cm)	Wet wt (g)	δ^15N (‰)b	δ^13C (‰)b	Trophic level	Hg (ppm)
a Length was measured as total length for fish and sand shrimp, carapace length for grass shrimp, and mantle length for squid. Trophic levels were calculated, using Equation 5. Sample sizes (n) and mean values (± standard error) are also presented. b Sample size for isotope analysis was n = 20–85 per species. c Portion of data set published in Payne and Taylor 6 or Piraino and Taylor 7.
Bluefish
Age 0	478	13.1 (0.1)	32.4 (7.3)	15.58 (0.22)	−16.52 (0.17)	3.55 (0.07)	0.041 (0.001)
Age 1 +	154	51.2 (0.9)	1278 (151)	16.09 (0.22)	−17.33 (0.11)	3.71 (0.07)	0.254 (0.007)
Prey
Anchovyc	39	7.3 (0.1)	2.72 (0.12)	14.39 (0.13)	−18.34 (0.11)	3.30 (0.05)	0.035 (0.001)
Grass shrimp	192	1.0 (0.0)	0.40 (0.02)	13.60 (0.20)	−16.46 (0.34)	3.03 (0.07)	0.030 (0.002)
Herringc	200	6.1 (0.1)	2.37 (0.15)	13.42 (0.19)	−19.73 (0.47)	2.97 (0.07)	0.028 (0.001)
Killifish	91	4.1 (0.1)	0.85 (0.04)	12.79 (0.38)	−16.45 (0.65)	2.75 (0.13)	0.016 (0.001)
Menhadenc	32	4.2 (0.2)	0.72 (0.13)	12.08 (0.13)	−18.77 (0.11)	2.50 (0.04)	0.016 (0.002)
Plankton	133	-	-	11.91 (0.19)	−19.28 (0.30)	2.44 (0.06)	0.005 (0.001)
Sand shrimpc	297	3.8 (0.1)	0.49 (0.02)	13.70 (0.18)	−16.37 (0.38)	3.06 (0.06)	0.021 (0.001)
Scupc	58	10.0 (0.5)	23.0 (2.2)	14.91 (0.22)	−17.09 (0.22)	3.48 (0.08)	0.032 (0.002)
Silverside	588	8.2 (0.1)	4.46 (0.14)	14.52 (0.14)	−17.02 (0.27)	3.34 (0.05)	0.054 (0.002)
Squidc	62	8.7 (0.4)	29.1 (2.4)	13.44 (0.26)	−17.90 (0.22)	2.97 (0.09)	0.037 (0.004)
Winter flounderc	313	5.4 (0.1)	3.11 (0.40)	13.82 (0.17)	−16.32 (0.62)	3.10 (0.06)	0.017 (0.001)

Table 3. Summary statistics for one-way analysis of variance (ANOVA) models testing for mean differences in nitrogen and carbon stable isotope (δ¹⁵N and δ¹³C) signatures, trophic levels, and total mercury (Hg) concentrations among bluefish and prey. Statistical results are also presented for ANOVA models examining the effects of tissue type (muscle vs whole body; Fig. 2A) and tissue location (multiple muscle biopsies; Fig. 2B) on age 0 and age 1 + bluefish total Hg concentrations, respectively.

Variable		F	Degrees of freedom	p
Response	Explanatory
δ^15N	Bluefish and prey	36.49	12, 284	< 0.0001
δ^13C	Bluefish and prey	12.19	12, 284	< 0.0001
Trophic level	Bluefish and prey	30.58	12, 284	< 0.0001
Hg	Bluefish tissue type	31.96	1, 516	< 0.0001
Hg	Bluefish tissue location	0.76	6, 97	0.605
Hg	Bluefish age-class	1662.3	1, 631	< 0.0001
Hg	Prey	157.76	14, 1958	< 0.0001

Total length and date of capture were the most significant factors affecting bluefish δ¹⁵N signatures, such that fish length and day of year accounted for a cumulative R² of 0.285 in the stepwise multiple linear regression model (Table 4). Moreover, estimated coefficients for both variables were positive (Table 4), and are explained by ¹⁵N enrichment in age 0 bluefish during early ontogeny (Fig. 3A). Specifically, age 0 bluefish between 8 and 12 cm TL experienced a 2.5‰ increase in δ¹⁵N values (∼ 78% of one trophic level). This rapid change in δ¹⁵N is attributed to the seasonal recruitment of juvenile bluefish into the estuary and the subsequent diet shift from copepod to piscine prey 27, 28. Indeed, stomach content analysis from this investigation showed that age 0 bluefish smaller than 12 cm TL consumed zooplankton (δ¹⁵N = 11.91‰), whereas silversides were the dominant prey of larger age 0 bluefish (δ¹⁵N = 14.52‰). This observed ontogenetic diet shift in age 0 bluefish would increase trophic position by approximately 81% of one full level.

Table 4. Parameter estimates (standard error [SE]) and summary statistics for stepwise multiple linear regression models used to analyze bluefish nitrogen (δ¹⁵N) and carbon (δ¹³C) stable isotope signatures and total mercury (Hg) concentrationa

Relationship	Model parameters		F	Degrees of freedom	R^2	p
	Estimate	SE
a Response variables were analyzed as a function of year, date of capture (day of year), place of capture (latitude and longitude; decimal degrees), and total length (cm), and only parameters significant at p < 0.05 were included in the model and therefore displayed in the table. b Loge(x) transformed.
δ^15N
Lengthb	0.545	0.200	8.12	1, 39	0.176	< 0.01
Day	0.014	0.006	5.63	1, 39	0.109	< 0.05
Intercept	11.236	1.296	-	-	-	-
δ^13C
Day	0.017	0.004	13.37	1, 39	0.260	< 0.001
Lengthb	−0.392	0.138	8.08	1, 39	0.133	< 0.01
Intercept	−19.191	0.894	-	-	-	-
Hg
Lengthb	0.151	0.003	1858.02	1, 631	0.747	< 0.0001
Day	−0.001	0.000	196.57	1, 631	0.060	< 0.0001
Longitude	−0.177	0.037	23.30	1, 631	0.007	< 0.0001
Intercept	12.596	2.617	-	-	-	-

Figure 3. Nitrogen (δ¹⁵N) and carbon (δ¹³C) stable isotope signatures of bluefish (A and B, respectively; n = 40) as a function of total length. Least-squares logarithmic regression models were fit to the full data sets.

Age 0 and age 1 + bluefish had comparable mean δ¹⁵N signatures (δ¹⁵N = 15.58 and 16.09‰), and bluefish across multiple life stages occupied similar trophic levels (trophic level = 3.55 and 3.71) (Table 2). The similar trophic status of age 0 and age 1 + bluefish is explained by the trophic positioning of their respective prey. Age 0 and age 1 + bluefish feed on prey with collective mean trophic levels of 2.89 and 3.11, respectively. Reevaluating the data for only preferred prey (%IA > 35%), however, revealed that age 0 bluefish (> 12 cm TL) consumed prey of higher trophic status than the preferred prey of age 1 + bluefish (mean trophic level = 3.22 and 2.73 for small silversides and large herring/menhaden, respectively). The reduced trophic position of age 1 + bluefish is attributed to the dietary contribution of menhaden (52, this study); a phytoplanktivorous forage fish that occupies a low trophic level in the estuary (trophic level = 2.50) 26.

Carbon stable isotope (δ¹³C) signatures were used to differentiate among sources of carbon to the Narragansett Bay ecosystem. Mean δ¹³C values of bluefish and prey ranged from −19.73‰ to −16.32‰ and varied significantly among taxa (Tables 2 and 3). Evidence of distinct δ¹³C isotopic groupings within the data (−18 to −19‰ and −16 to −17‰) indicated two principal sources of primary production to the estuary. Specifically, depleted δ¹³C values denoted phytoplankton-based primary production, whereas enriched isotope values represented a benthic macroalgal carbon source 39. Assuming an increase of −0.5‰ per trophic level, the depleted δ¹³C signatures of planktonic crustaceans, planktivorous forage fish, and squid suggested that these prey derived carbon from a phytoplankton (pelagic) source (Table 2). Conversely, decapod crustaceans and demersal/benthic fish had enriched δ¹³C values consistent with a benthic carbon source.

The mean δ¹³C signatures of age 0 and age 1 + bluefish (−16.52‰ and −17.33‰) indicated that carbon was derived from different trophic pathways (Table 2). The more depleted mean δ¹³C signature of age 1 + bluefish implies that older fish have an increased reliance on pelagic prey. This supposition is supported by the stomach content data, whereby planktivorous forage fish (herring, menhaden, and anchovy) and squid accounted for more than 97% of the estimated %IA for age 1 + bluefish (Table 1). Conversely, the more enriched δ¹³C signature of age 0 bluefish is explained by the relative dietary contribution of shrimp and demersal/benthic fish (predominantly silversides), with a cumulative %IA of 46.5%. Moreover, in the stepwise multiple linear regression model, date of capture and total length explained the most variability in bluefish δ¹³C signatures (cumulative R² = 0.393; Table 4; Fig. 3B). These results again indicate a subtle ontogenetic diet shift toward pelagic prey as the bluefish aged.

Mercury contamination in bluefish and prey

The initial statistical analysis of bluefish Hg data examined the effects of tissue type (muscle vs whole body) and location (multiple muscle biopsies) on contaminant concentrations (Fig. 2). For age 0 bluefish, total Hg bioaccumulation rates were significantly faster in muscle filets relative to whole bodies (Table 3; Fig. 2A, 2C), thus precluding direct comparisons across the different tissue types. For age 1 + bluefish, no significant difference was seen in the mean total Hg concentration among muscle biopsies performed along the dorsolateral surface (Table 3; Fig. 2B, 2D); thus, Hg concentrations in the D₀ biopsy are consistent with levels measured in other locations of the filet. Based on these results, Hg in age 0 and age 1 + bluefish are hereafter reported for isolated muscle filets and D₀ biopsies only.

Total Hg concentrations of bluefish collected from the Narragansett Bay ranged from 0.011 to 0.616 ppm (n = 632), with mean values equal to 0.041 and 0.254 ppm for age 0 and age 1 + bluefish, respectively (Table 2). These observed Hg concentrations differ from values reported for bluefish from other geographic areas. Bluefish from the Long Island Sound and the greater New York Bight (Long Island Sound: mean TL = 40.6 cm; n = 46; New York Bight: mean TL = 53.3 cm; n = 14), for example, had mean MeHg and total Hg concentrations equal to 0.137 ppm (Hg ∼ 0.130 ppm assuming %MeHg = 95%) and 0.102 ppm, respectively 19, 54. These contaminant levels are 62 to 145% lower than the predicted Hg content of bluefish from the Narragansett Bay; bluefish of consistent sizes had projected Hg concentrations of 0.210 to 0.250 ppm (Table 2; Fig. 4). In contrast to these findings, bluefish sampled from coastal New Jersey had increased Hg levels (mean Hg = 0.345 ppm; mean TL = 46.3 cm; n = 206) relative to the same-sized fish from the Narragansett Bay (0.229 ppm; Table 2; Fig. 4) 55.

Figure 4. Total Hg concentrations of bluefish muscle tissue (n = 632) as a function of total length. A least-squares logarithmic regression model was fit to the full data set.

Previous investigations report that bluefish Hg concentrations vary over relatively broad spatial and temporal scales 18, 19, 55, which is attributed to localized differences in Hg loadings to the environment, as well as in situ sediment MeHg production and mobilization 20. Moreover, the extent to which MeHg biomagnifies in estuarine and marine food webs is affected by spatiotemporal dynamics in resident prey populations. This is especially relevant to bluefish because of their expansive coastal migrations and ontogenetic shifts in diet and habitat use. Accordingly, a complete evaluation of Hg contamination in this species would benefit from a synoptic examination over broad spatial scales and protracted temporal periods 55.

Bluefish muscle Hg concentrations were affected by its physical attributes. Foremost, total Hg increased significantly with bluefish length, despite concentrations being highly variable in age 1 + fish (stepwise regression: length; partial R² = 0.747; Table 4; Fig. 4). Comparable Hg-length (or weight) trends have been reported for bluefish in the Long Island Sound and coastal New Jersey 18, 19, 54, as well as other estuarine and coastal fishes from the northwestern Atlantic 6, 7, 54, 56. The significant positive correlation between bluefish muscle Hg and length affirms that Hg biomagnifies in this tissue, and is likely attributable to the rapid uptake of the contaminant relative to its excretion from the body 21, 57. Alternatively, Hg elimination rates may be negatively correlated with bluefish body size, as demonstrated in several freshwater and marine finfish species 57. Bluefish total Hg concentrations also varied over relatively small spatiotemporal scales in the estuary (longitude and date of capture; Table 4), although these variables accounted for only 6.7% of the observed variation in fish Hg content.

The biomagnification of Hg in bluefish was maximal during early life stages and decelerated over time (Fig. 4). Similarly, Burger 55 observed high rates of Hg bioaccumulation in bluefish between 11 to 25 cm fork length (∼ 17–36 cm TL), after which Hg concentrations remained relatively constant for fish 26 to 50 cm fork length (∼37–68 cm TL). Burger 55 also observed rapid Hg accumulation in bluefish larger than 50 cm fork length, but this larger size class was not examined in the current study. Nevertheless, the pattern of Hg biomagnification for bluefish 6 to 70 cm TL is atypical of estuarine and marine fish, where the standard response is an exponential increase in Hg throughout the entire life history of the fish 6, 7, 42, 54. Consequently, early-stage bluefish have increased Hg concentrations relative to similarly aged estuarine and coastal fishes, with possible deleterious effects to both the individual (e.g., compromised behavior and growth) and population (e.g., reduced survival and recruitment success) 18. The unusually high rate of Hg biomagnification in juvenile bluefish is attributed to its diet and high trophic status in the estuary, as discussed previously.

Results from previous investigations in the Narragansett Bay indicate that dietary preference and trophic processes are key factors affecting Hg bioaccumulation in the estuary 6, 7. Accordingly, Hg is transferred through the estuarine food web via predator–prey interactions, and high trophic level fishes have increased contaminant exposure because they feed on Hg-enriched prey 58. To evaluate the effect of food preference on bluefish Hg concentrations, the main prey of bluefish were collected from the estuary and analyzed for whole-body Hg content. Prey common to the diet of both age 0 and age 1 + bluefish were divided into specific length categories (demarcation length = 5.0 cm for herring, silversides, and squid and 4.0 cm for sand shrimp), such that specimens smaller and larger than the demarcation length were defined as age 0 and age 1 + bluefish prey, respectively. Mean Hg concentrations of bluefish prey varied significantly across species (Tables 2 and 3; Fig. 5). Collectively, age 0 bluefish consumed prey with a lower mean Hg concentration than the prey of age 1 + bluefish (0.021 and 0.035 ppm, respectively). With respect to preferred prey (%IA > 35%; Table 1), however, the Hg concentration of age 0 and age 1 + bluefish prey were equivalent (0.021 ppm for small silversides and large herring/menhaden; Fig. 5). These results strongly suggest that age 0 bluefish are highly responsive to prey Hg contamination 18, 58, which in turn partly explains the rapid accumulation of Hg during the juvenile life stage.

Figure 5. Total Hg concentrations (mean + standard error) of prey whole bodies: Atlantic silverside (n = 471 and 117 for small and large, respectively), long-finned squid (n = 54 and 8 for small and large, respectively), bay anchovy (n = 39), scup (n = 58), sand shrimp (n = 96 and 155 for small and large, respectively), herring (n = 40 and 160 for small and large, respectively), grass shrimp (n = 192), winter flounder (n = 313), killifish (n = 91), Atlantic menhaden (n = 32), and zooplankton (n = 133). Unique letters above histogram bars indicate significant differences in mean prey Hg concentrations (Ryan's Q multiple comparison test).

CONCLUSIONS

This investigation evaluated Hg biomagnification in bluefish in relation to intraspecies life history traits, such as body size and ontogenetic patterns in diet. The total Hg content of bluefish muscle tissue was positively correlated with length, affirming that Hg biomagnifies in this tissue. Moreover, Hg accumulation rates were greatest for age 0 bluefish and decreased as the fish aged, resulting in relatively high Hg concentrations in the juvenile life stage. Accelerated Hg biomagnification in age 0 bluefish, relative to older conspecifics, is attributed to three key biological and ecological conditions. First, age 0 juvenile bluefish are obligate users of estuarine or coastal beach habitats 47, 59, and thus are potentially exposed to elevated levels of environmental Hg 18, 20. Second, the initial inshore recruitment of juvenile bluefish coincides with an ontogenetic shift from planktivory to piscivory 27, 28. This early switch to a piscine-dominated diet results in age 0 bluefish having a comparable trophic status to age 1 + bluefish, and concomitantly, feeding on relatively Hg-enriched prey. Third, juvenile bluefish have among the highest evacuation (5–7 h) and consumption rates (2–30% body wt/d) documented for temperate fishes 29. As such, bluefish standardized feeding rates decline exponentially with increasing age 60, leading to increased age 0 bluefish dietary uptake and exposure to Hg 18.

Finally, bluefish support a major fishery along the northeastern United States, accounting for approximately 10% of total recreational landings from 2005 to 2009 (National Marine Fisheries Service, Fisheries Statistics Division, http://www.st.nmfs.noaa.gov/st1/). The importance of bluefish as an edible species suggests that they are a potential source of Hg for fish-consuming humans. In this study, 26% of age 1 + bluefish had Hg levels exceeding the U.S. EPA criterion of 0.3 ppm 61. The Hg-length regression model also predicted that bluefish obtain Hg levels of 0.3 ppm at approximately 75 cm TL (Table 3; Fig. 4). However, bluefish demonstrate substantial variability in Hg contamination, and individuals larger than 30 cm TL consistently have Hg levels above the U.S. EPA threshold. The cumulative data presented herein therefore suggest that frequent consumption of bluefish could pose a human health risk, and preferentially consuming smaller bluefish may be an inadequate strategy for minimizing human dietary exposure to Hg.

Acknowledgements

We are grateful to J. McNamee, J.C. Powell, and S. Olszewski (Rhode Island Division of Fish and Wildlife, Jamestown, RI), K. Henry (University of Rhode Island, Narragansett, RI), and E. Payne, M. Piraino, J. Linehan, S. Helming, L.F. Ho, M. Gardner, and B. Bourque (Roger Williams University, Bristol, RI) for assistance in sample collection and preparation. We thank R. Michener (Boston University, Boston, MA) for stable isotope analysis. The project described was supported by the Roger Williams University Foundation Fund Based Research Grant and by Award P20RR016457 from the National Center for Research Resources. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health.