The Deepwater Horizon (DWH) oil spill was a catastrophic event for the Gulf of Mexico ecosystem resulting in a total volume discharged of approximately 780,000 m³ between April 22 and July 15, 2010 (McNutt et al. 2012). Persistence of oil compounds in the water column (Allan et al. 2012; Reddy et al. 2012; Sammarco et al. 2013), impacts to plankton communities (Ortmann et al. 2012; Almeda et al. 2013), and flocculent particulate organic matter that transferred oil compounds to the benthos (Passow et al. 2012; Mason et al. 2014) have been well documented. Direct oiling and toxicological effects of the DWH oil spill on higher taxa, including fishes, sea turtles, birds, and marine mammals, also have been reported (Whitehead et al. 2012; Barron 2012; Schwacke et al. 2013; Murawski et al. 2014). However, little information exists to infer food web effects at higher trophic levels.

We examined changes in the diet and trophic position of Red Snapper Lutjanus campechanus following the DWH oil spill, which was only possible due to baseline data from a study that actually began 18 months prior to the spill but with very different objectives than to examine oil spill impacts. The original study was focused on examining differences in reef fish ecology on natural versus artificial reefs in the northern Gulf of Mexico. Understanding the relationship between artificial reefs and associated fish communities has important implications for whether artificial reefs are more likely to enhance reef fish production or merely aggregate fish, thus making them more vulnerable to fishing mortality (Bohnsack 1989; Lindberg 1997; Perkol-Finkel et al. 2006). Few fisheries issues in the northern Gulf are as controversial as the ecological versus fishery function of artificial reefs (Cowan et al. 2011), and this debate has mostly centered on Red Snapper.

Red Snapper was an ideal model species given their ecological and economic importance in the region, as well as their abundance on both artificial and natural reefs (Dance et al. 2011; Patterson et al. 2014). Previous studies have been conducted on Red Snapper feeding ecology, but nearly all of the research on this topic has been focused on their ecology on artificial reefs with little to no comparison to fish occupying natural reefs (e.g., Szedlmayer and Shipp 1994; McCawley and Cowan 2007; Dance et al. 2011). Others have clearly demonstrated that comparisons between artificial and natural habitats are critical for assessing the effects of artificial reefs on reef fish ecology (e.g., Carr and Hixon 1997; Perkol-Finkel et al. 2006). Therefore, the original focus of this study was to examine the feeding ecology of Red Snapper at artificial versus natural reefs in the northern Gulf. Analysis of potential habitat effects also required explicitly accounting for the effect of ontogeny given previous reports of ontogenetic shifts in Red Snapper feeding ecology (Ouzts and Szedlmayer 2003; McCawley and Cowan 2007; Wells et al. 2008).

In spring 2010, the occurrence of the DWH oil spill became an unplanned factor in this study. Food web impacts have been reported following previous large-scale spills (Teal and Howarth 1984; Jackson et al. 1989; Moreno et al. 2013), although chronic ecological impacts may not be apparent for many years following such events (Peterson et al. 2003). Environmental effects of the DWH oil spill have been reported across a variety of habitats and taxa, but effects of the spill on fishes have mostly been restricted to inshore or coastal species (e.g., Fodrie and Heck 2011; Whitehead et al. 2012; Pilcher et al. 2014). Our study occurred over a broad area (about 8,000 km²) of the northcentral Gulf of Mexico continental shelf. The Red Snapper samples collected there prior to the DWH oil spill provided a unique opportunity to examine the effects of the spill on their diet and trophic ecology. This is significant, not only due to the ecological and economic importance of Red Snapper, but also because relatively few baseline data sets exist to examine the impacts of the DWH oil spill on upper trophic level taxa in the northern Gulf.

FIGURE 1.

(A) Location of the study area in the northern Gulf of Mexico, showing the location of the Macondo wellhead (star), and (B) the natural (green circles [n = 33]) and artificial (blue triangles [n = 27]) reef sites where Red Snapper were sampled before and after the Deepwater Horizon oil spill; MS = Mississippi, AL = Alabama, FL = Florida. Isobath depths in m.

METHODS

Red Snapper were sampled between June of 2009 and August of 2011 from 33 natural and 27 artificial reefs south of Dauphin Island, Alabama, to southeast of Destin, Florida (Figure 1). Sampling was conducted onboard chartered fishing boats (home ports of Orange Beach, Alabama, and Pensacola and Destin, Florida) and also onboard the state of Florida's RV Bellows. Hook-and-line fishing was performed with two-hook rigs composed of 9/0, 12/0, or 15/0 Mustad circle hooks on 60-lb fluorocarbon leaders. Bait was kept uniform and consisted of cut squid, Loligo spp. or Mackerel Scad Decapterus macarellus, which allowed for easy identification and segregation from food during diet analysis.

Upon capture, Red Snapper fork length (FL) and total length (TL) were measured (mm) and sex was determined by examination of gonads. A sample (≥50 g) of white muscle tissue was removed above the pectoral fin of each fish and placed in a sealed plastic bag on ice for transport to the laboratory, then stored in an ultralow freezer at — 80°C until processed. Fish stomachs were also removed at sea and immediately fixed in 10% buffered formalin. A two-factor ANOVA was computed to test for significant differences in Red Snapper total length between habitat types and pre-DWH versus post-DWH. A priori, the experiment-wise error rate (α) was set to 0.05 for all statistical analyses.

Diet analysis.—Red Snapper stomachs were fixed in 10% formalin for at least 48 h and then were transferred to 70% isopropyl alcohol for preservation. Diet composition analysis was performed by identifying stomach contents to the lowest taxonomic level possible with the aid of a dissecting microscope for all stomach samples that had prey items present. Prey items were sorted by taxon and then dried for 48 h at 60°C to obtain dry mass.

Prey were sorted into one of seven categories for statistical analysis of diet: fish, decapods, cephalopods, stomatopods, gastropods, zooplankton, and other invertebrates. Mean percent dry mass was computed as the arithmetic mean among samples within a given factor, the factors being size-class (<400, 400–500, >500 mm), habitat type (natural versus artificial reefs), and DWH oil spill timing (before versus after April 22, 2010). A three-factor permutational multivariate ANOVA (PERMANOVA) was computed with the Primer statistical package (version 6; Anderson et al. 2008) to test for differences in Red Snapper diet by percent dry mass among size-classes, between habitat types, and pre versus post spill. Diet data were square-root-transformed, and a dummy variable with value = 1 was added to each sample to stabilize dispersion in the data prior to computing the Bray-Curtis similarity measure between each pair of samples (Clarke 2006). The PERMANOVA model then was computed in Primer with 10,000 permutations. Type-III sums of squares were utilized to determine significance.

Stable isotope analysis.—Stable isotope analysis was performed on a randomly selected subset of Red Snapper white muscle samples. Muscle tissue was dried at 60°C for at least 24 h, ground with a mechanical tissue grinder, and then pulverized into a fine powder with a glass or agate mortar and pestle. Between 0.2 and 0.5 g of pulverized dried tissue were placed in sterile plastic centrifuge tubes and shipped to a contract laboratory to perform analysis of δ¹⁵N, δ¹³C, and δ³⁴S with a Europa Scientific GSL/Geo 20–20 stable isotope ratio-mass spectrometer. The isotopic ratio of a sample relative to the ratio in an international standard is reported in the standard delta notation: δX = [(R _samle/R _standard) — 1] × 1,000, where X = ¹⁵N, ¹³C, or ³⁴S and R = ¹⁵N : ¹⁴N, ¹³C : ¹²C, or ³⁴S : ³²S. Standards included nitrogen in air (δ¹⁵N_Air), Pee Dee Belemnite (δ¹³C_V-PBD), and Canyon Diablo Troilite (δ³⁴Sv-CDT). International Atomic Energy Agency standard reference materials were run periodically as check samples to assess machine performance for each stable isotope ratio analyzed. Analytical precision was estimated from duplicate analysis of 20 randomly selected samples.

Trophic level was estimated for each Red Snapper sample following Post (2002): trophic level = 1 + (δ¹⁵N_fish — δ¹⁵N_prod/Δ _n , where δ¹⁵N_fish is the δ¹⁵N of a given Red Snapper muscle sample, δ¹⁵N_prod is the mean δ¹⁵N_prod of primary producers in the system, and Δ _n is trophic fractionation per trophic level. Estimating mean δ¹⁵N_prod of primary producers can be problematic if there is a diversity of primary producers present in the system (e.g., phytoplankton, benthic algae, seagrasses, marsh grasses) or if there is considerable temporal variability in δ¹⁵N. While seagrass and marsh habitats were present in the study region, they were not present on the shelf or within 30 km of the study reefs. Therefore, the predominant primary producers in the system are phytoplankton and benthic microalgae. There was no significant difference in mean δ¹⁵N between phytoplankton and benthic microalgae samples (n = 33) collected in the northern Gulf shelf prior to the DWH oil spill (two-sample t-test, P — 0.305; W. Patterson, unreported data), which is similar to the pattern reported for δ¹⁵N by Radabaugh et al. (2013) for the west Florida shelf, despite considerable variability in δ¹³C between phytoplankton and benthic algae. Given the lack of difference in δ¹⁵N among pre-DWH oil spill phytoplankton and benthic microalgae samples, the overall mean δ¹⁵N of 5.82 (SE = 0.13) among combined sample types served as our estimate of δ¹⁵N_prod. Trophic fractionation (Δ _n ) was assumed to equal 3.0‰, which is the mean measured for fishes feeding on highprotein diets (McCutchan et al. 2003; Vanderklift and Ponsard 2003; Rooker et al. 2006).

Values of δ¹³C were corrected for percent lipid with the regression equation reported by Post et al. (2007) for aquatic animals: CF = —3.32 + (0.99 × C:N), where CF is the correction factor applied to δ¹³C to account for percent lipid and C: N is a proxy for percent lipid. Hereafter, δ¹³C always indicates lipid-corrected δ¹³C. To examine basal sources of C and S, the δ¹³C and δ³⁴S values at the base of the food web were estimated following Post (2002):

where δX _base = estimated δ¹³C or δ³⁴S of the C or S source, respectively, at the base of the food web, δX _fish = δ¹³C or δ³⁴S of a given Red Snapper sample, trophic level = estimated trophic level of a given Red Snapper sample derived from δ¹⁵N, and Δ _x = trophic fractionation (increase) in 8¹³C or 8³⁴S per trophic level. A value of 1.5‰ was assumed for Δ_c, which is the mean reported by Sweeting et al. (2007) for fish muscle tissue. A value of 0.5‰ was assumed for Δ_s, which is the mean reported by McCutchan et al. (2003).

Linear regressions were computed between Red Snapper δ¹⁵N, δ¹³C, and δ³⁴S and total length to test for ontogenetic effects on those variables. Two-factor ANOVA models were computed to test the effects of habitat type and DWH oil spill timing on total length, estimated trophic level, δ¹³C_base, and δ³⁴S_base.

RESULTS

Study reefs ranged in depth from 17 to 92 m. Red Snapper ranged from 238 to 890 mm TL, the means not differing (ANOVA) between habitat types (F _{1, 706} = 0.30; P — 0.586) or between pre and post samples (F _{1, 706} = 2.45; P — 0.118).

Stable Isotope Analysis

Red Snapper white muscle samples included tissue from 32 individuals sampled between June and November 2009 (preDWH), as well as 34 fish sampled between March and August 2011 (post-DWH). There was a significant linear relationship between δ¹⁵N and total length (GLM procedure [SAS, Cary, North Carolina], P — 0.035), but regressions for δ¹³C (GLM, P — 0.981) and δ³⁴S (GLM, P — 0.130) were not significant (Figure 3). Total length of fish sampled for stable isotope analysis was significantly different between pre-DWH (mean = 452 mm, SE = 17.8) and post-DWH (mean = 529 mm, SE = 29.7) samples (Table 3). Therefore, δ¹⁵N data were length-detrended by subtracting the slope of the total length versus δ¹⁵N regression times a given sample's total length from the sample's δ¹⁵N value, and then trophic level was estimated as indicated above. The DWH oil spill timing effect was significant for Red Snapper trophic position, but there was no difference between habitat types; the interaction between habitat and DWH oil spill timing also was not significant (Table 3; Figure 4).

TABLE 1.

Mean percent diet by dry mass computed from identifiable prey items within stomachs of 343 Red Snapper sampled at natural and artificial reefs in the northern Gulf of Mexico.

Continued.

TABLE 2.

Permutational multivariate ANOVA model testing the effect of fish size-class (< 400, 400–500, and >500 mm TL), habitat type (natural versus artificial reefs), and the timing of the Deepwater Horizon (DWH) oil spill (pre versus post) on Red Snapper diet.

FIGURE 2.

Mean percent Red Snapper diet by dry mass estimated among seven prey categories observed among stomach samples. Panels demonstrate the effect of (A) habitat (natural versus artificial reefs), (B) size-class (total length in mm), (C) the timing of the Deepwater Horizon (DWH) oil spill, and (D) the interaction between habitat type and DWH timing on Red Snapper diet among samples collected in the northern Gulf of Mexico from June 2009 through August 2011.

Variables δ¹³C_base and δ³⁴S_base required a log _e transformation to meet the parametric assumption of normality (Table 3). The decrease in δ¹³C_base following the DWH oil spill was statistically significant, and the DWH oil spill timing effect was also significant δ³⁴S_base, both decreasing the year after spill (Table 3; Figure 4). Consistent with the lack of significant diet differences between natural and artificial reefs, none of the stable isotope variables displayed a significant habitat effect (Table 3).

FIGURE 3.

Scatterplots of white muscle (A) δ¹⁵N, (B) lipid-corrected δ¹³C, and (C) δ³⁴S versus total length for pre-Deepwater Horizon (DWH) and post-DWH oil spill Red Snapper samples. Regression fit to δ¹⁵N data in panel A was significant but weak, and regressions for lipid-corrected δ¹³C and δ³⁴S were nonsignificant.

DISCUSSION

Overall, data from this study demonstrate adult Red Snapper to be generalist mesopredators with broad diets. They are opportunistic foragers that consume prey along a size spectrum ranging from zooplankton to fish. However, nothing in their morphology would suggest they are able to strain or even pick zooplankton from the water column. Instead, they likely forage on swarms of plankton, such as pteropods or hyperiid amphipods (Fenwick 1978; Sanvicente-Añorve et al. 2013), when available. The phenomenon of Red Snapper subsidizing their diet with zooplankton was described previously by Ouzts and Szedlmayer (2003) and McCawley and Cowan (2007) for fish sampled at artificial reefs off Alabama in the northern Gulf of Mexico. McCawley and Cowan (2007) demonstrated that plankton consumption increased for larger Red Snapper, as is reported here, and that zooplankton consumption was highest in spring. They concluded that zooplankton consumption was probably either a response to artificial reefs being spaced too closely—thus. Red Snapper had to rely on unexpected prey resources to meet basal bioenergetics demands— or that adults subsidized their diet with zooplankton to facilitate maximum reproductive output during their protracted April to September spawning season (Jackson et al. 2007).

We found different zooplankton taxa were predominant in pre-DWH Red Snapper stomachs sampled at natural versus artificial reefs, but zooplankton percent diet by mass was nearly identical between habitat types. Therefore, it seems unlikely that artificial reef spacing explains zooplankton consumption, especially given the fact that Red Snapper density is an order of magnitude higher on artificial versus natural reefs in the system (Patterson et al. 2014). Instead, the presence of zooplankton highlights the extreme diversity of Red Snapper diet and the opportunism they display while foraging. It should also be noted that several zooplankton taxa identified in diet samples were relatively soft-bodied compared with decapods or fishes. Therefore, it is likely that zooplankton prey experienced higher digestion and evacuation rates than other prey types, thus may have constituted an even higher percentage of the Red Snapper diet than the overall 22.6% by mass observed among pre-DWH oil spill samples.

Habitat was not a significant factor with respect to Red Snapper diet or trophic position, and few reef-dependent taxa were present in Red Snapper stomach samples, regardless of habitat type. Densities of larger (>250 mm TL) reef fishes, including Red Snapper, on artificial reefs are typically 150– 200% greater than their densities on natural reefs in the northern Gulf (Dance et al. 2011; Patterson et al. 2014), which may lead to greater competition for resources and decreased food web complexity (Bohnsack and Sutherland 1985). However, among our seven broad prey categories, only subtle differences in Red Snapper diet contribution were observed between habitat types, and those differences were probably related to local abundance rather than prey selectivity. The lack of reefassociated prey in Red Snapper stomachs indicates a reliance on nonreef foraging habitat (Bradley and Bryan 1975; Ouzts and Szedlmayer 2003; McCawley and Cowan 2007), regardless of whether fish were associated with natural or artificial reefs.

Many reef fishes, including snappers, have been shown to forage away from reef structure, thus translocate energy and nutrients horizontally back to reefs (Appledoorn et al. 2009; Luo et al. 2009; Berkström et al. 2012). Data presented here and by others suggest Red Snapper also translocate energy and nutrients from surrounding open substrates back to reefs, whether natural or artificial. Lack of habitat-specific differences in Red Snapper diet may at first seem to indicate no foraging advantage conveyed by artificial reef habitat. However, based on foraging distances (typically <100 m) estimated via acoustic telemetry data (Westmeyer et al. 2007; Topping and Szedlmayer 2011), deployment of artificial reefs appears to enable Red Snapper to exploit foraging habitats on the shelf they otherwise would be unlikely to visit, especially given that natural reef habitat in the northcentral Gulf of Mexico is concentrated in deeper (>40 m) waters than where the bulk of artificial reef zones are located (Parker et al. 1983; Schroeder et al. 1988; Patterson et al. 2014).

TABLE 3.

ANOVA models testing for effects of habitat type (natural versus artificial reefs) and Deepwater Horizon (DWH) timing (pre versus post) on Red Snapper stable isotopes or stable isotope-derived variables.

We observed ontogenetic shifts in Red Snapper diet and trophic position, but the decline in their estimated trophic level with increasing size was not expected a priori. Others have reported shifts to higher trophic position and to greater piscivory during development (Bradley and Bryan 1975; Wells et al. 2008). However, those studies described feeding and trophic shifts from settlement and early life, when a clear shift from planktonic to benthic feeding was observed. Here, we report a similar pattern among sub adult and adult Red Snapper as that reported by McCawley and Cowan (2007) in which zooplankton constituted a substantial (>20%) percentage of the diet of larger, reproductively mature fish. Higher trophic-level prey, such as decapods and fish, were predominant in the diet of smaller (<400 mm TL) Red Snapper but became less predominant for larger individuals. Although decapod contribution to diet decreased as Red Snapper length increased, the mass of individual prey was notably greater in the diet of larger fish.

Analysis of δ¹⁵N corroborated a decrease in Red Snapper trophic position with increasing size, which also indicates the presence of lower trophic-level prey in the diet of larger individuals was not ephemeral, given that turnover time of fish white muscle tissue is typically weeks to months (Buchheister and Latour 2010; Nelson et al. 2010). While δ¹⁵N values and trophic level estimates derived from them confirm trends observed in diet data, δ¹³C and δ³⁴S data provide information about Red Snapper trophic ecology that may not be apparent in diet data alone. For example, northern Gulf phytoplankton typically has a δ¹³C signature of approximately —22‰ and a δ³⁴S signature of approximately 18‰, whereas benthic microalgae has a δ¹³C signature of approximately — 18‰ and a δ³⁴S signature of approximately 10‰ (Moncrieff and Sullivan 2001; Fry 2006; Rooker et al. 2006). Therefore, estimates of δ¹³C_base and δ³⁴S_basee provide information with respect to the relative contribution of pelagic versus benthic production to Red Snapper biomass, and the lack of a significant relationship between δ¹³C or δ³⁴S and length indicates neither the source of C nor S was related to fish size. The lack of a habitat effect on δ¹³C_base and δ³⁴S_base indicates the relative contribution of pelagic versus benthic production was not significantly different between natural and artificial reefs, which further indicates that Red Snapper trophic ecology does not differ between these habitat types.

The factor that had the greatest impact on Red Snapper diet and trophic level was DWH oil spill timing. Pre-DWH diet composition estimates presented here are consistent with earlier estimates reported by Ouzts and Szedlmayer (2003) and McCawley and Cowan (2007) (e.g., fish and crustaceans forming large percentages of Red Snapper diet, but zooplankton also contributing 15–20% of diet by mass), thus providing support that our pre-DWH diet data are generally representative of Red Snapper diet. Following the DWH oil spill, zooplankton consumption decreased markedly at both natural and artificial reefs, which was replaced with higher trophic level prey: decapods, cephalopods, and stomatopods at natural reefs and predominantly fish at artificial reefs. These shifts clearly were persistent for weeks to months as higher post-DWH δ¹⁵N values and lower δ³⁴S values in Red Snapper muscle samples indicated a shift to higher trophic level and more benthic prey resources (i.e., pelagic zooplankton prey replaced by benthic decapods and fishes). Toxicological, immunological, and genetic effects on plankton communities were documented after the spill (Ortmann et al. 2012; Paul et al. 2013), and a decline in zooplankton consumption is consistent with high plankton mortality and a resultant blizzard of marine snow observed in the months following the spill (Passow et al. 2012). Not only did oil toxicity impact northern Gulf plankton communities following the spill, but the release of millions of gallons of Corexit dispersant likely magnified the toxic effects of the spill on plankton (Middaugh and Whiting 1995; Ortmann et al. 2012; Paul et al. 2013). Therefore, while increased higher trophic level prey following the DWH oil spill might suggest greater abundance of those taxa, a more plausible scenario is that zooplankton were less abundant in the months following the spill which resulted in much lower zooplankton consumption.

FIGURE 4.

Mean ( ± 95% CI habitat (NR = natural reef; AR = artificial reef) and Deepwater Horizon (DWH) oil spill timing-specific values for Red Snapper (A) white muscle δ¹⁵N, (B) estimated trophic level, (C) lipid-corrected white muscle δ¹³C, (D) estimated δ¹³C at the base of the food web. (E) white muscle δ³⁴S, and (F) estimated δ³⁴S at the base of the food web for fish sampled between June 2009 and August 2011 in the northern Gulf of Mexico.

The one variable that appears inconsistent with DWH oil spill-related trends observed in diet and other stable isotope data is δ¹³C, specifically estimates of δ¹³C_base. Phytoplankton are depleted in ¹³C relative to benthic algae (Moncrieff and Sullivan 2001; Fry 2006); thus, lower δ¹³C_base values a year after the DWH oil spill should imply greater, not lower, plankton contribution to Red Snapper muscle C. However, hydrocarbons released during the DWH oil spill constitute another source of organic carbon on the shelf, and they were even more depleted in ¹³C than phytoplankton (Chanton et al. 2012; Cherrier et al. 2013). Results from δ¹³C mixing models computed by Cherrier et al. (2013) for northern Gulf particulate organic carbon (POC) sampled in summer 2011 and 2012 indicated 28–43% of POC was derived from fossil CH₄. Due to the sequestration of hydrocarbons beneath the seabed for millions of years, oil hydrocarbons are dead with respect to ¹⁴C activity (Chanton et al. 2012). Therefore, extremely low POC Δ¹⁴C values (e.g., less than —600‰) provide even stronger evidence that hydrocarbons from the DWH spill entered the food web (Cherrier et al. 2013). The fact that Red Snapper δ¹³C_base estimates were significantly lower following the spill when other evidence suggests higher values would be expected may suggest that oil carbon reached higher levels of the food web, and did so as early as the year following the spill.

Red Snapper samples collected in 2009–2010 prior to the DWH oil spill provide a unique opportunity to examine potential impacts of the spill on their diet and trophic position, as well as future examination of ecosystem resiliency. These relatively large generalist mesopredators may be an ideal species to examine for such impacts because they have such a broad diet, ranging from zooplankton to fish, thus can easily shift to foraging on locally abundant prey taxa. The corroboration of shifts observed in diet data with trophic shifts inferred from muscle stable isotope ratios provides clear evidence of DWH oil spill impacts to Red Snapper feeding ecology, and likely the northern Gulf food web in general. Where pre-DWH muscle samples or stable isotope data exist, stable isotope analysis also can be employed to examine DWH oil spill effects for other northern Gulf reef fishes for which pre-DWH oil spill diet data are unlikely to be as extensive as they are for Red Snapper.