Comparing isotope signatures of prey fish: does gut removal affect δ¹³C or δ¹⁵N?

Abstract

Stable isotope analysis is a quick and inexpensive method to monitor the effects of food web changes on aquatic communities. Traditionally, whole specimens have been used when determining isotope composition of prey fish or age-0 recreational fishes. However, gut contents of prey fish could potentially alter isotope composition of the specimen, especially when recent foraging has taken place or when the gut contains non-assimilated material that would normally pass through fishes undigested. To assess the impacts of gut content on prey fish isotope signatures, we examined the differences in isotopic variation of five prey fish species using whole fish, whole fish with the gut contents removed, and dorsal muscle only. We found significant differences in both δ¹⁵N and δ¹³C between the three tissue treatments. In most cases, muscle tissue was enriched compared to whole specimens or gut-removed specimens. Moreover, differences in mean δ¹⁵N within a species were up to 2‰ among treatments. This would result in a change of over half a trophic position (TP) based on a 3.4‰ increase per trophic level. However, there were no apparent relationships between tissue isotope values in fish with increased gut fullness (more prey tissue present). We suggest that muscle tissue should be used as the standard tissue for determining isotope composition of prey fish or age-0 recreational fishes, especially when determining enrichment for mixing models, calculating TP, or constructing aquatic food webs.

Keywords:

• stable isotope analysis
• rainbow smelt
• gizzard shad
• yellow perch
• white bass
• spottail shiner

Introduction

Stable isotope analysis (SIA) offers a useful approach for quantifying energy flow within food webs (Kling et al. 1992; Cabana and Rasmussen 1994). SIA offers time-integrated analysis of carbon (δ¹³C) and nitrogen (δ¹⁵N) reflective of energy assimilated by consumers and provides an analytical approach for determining the trophic position (TP; Kling et al. 1992; Cabana and Rasmussen 1996; Vander Zanden et al. 1999). TP is a continuous variable that accounts for omnivory and better quantifies matter and energy flow within a food web (Kling et al. 1992; Vander Zanden and Rasmussen 1996; Post 2002). One key assumption in developing food web models and calculating TP is developing a known measure of enrichment (Δ), or the magnitude of increase of a predator's isotopic value relative to the isotopic value of its prey (i.e., Δ = δ¹⁵N_consumer − δ¹⁵N_diet; Vanderklift and Ponsard 2003). Many studies have attempted to incorporate Δ variability to improve the efficiency of mixing models; however, these models are contingent on accurate estimates of Δ and further emphasize the need to develop accurate Δ estimates when using SIA in food web and energy flow studies (Phillips and Koch 2002).

Many variables affect Δ; these variables include, but are not limited to, the biochemical form of nitrogenous waste, feeding behaviors of the species, taxonomic classes, and the type of ecosystem under consideration (see Vanderklift and Ponsard (2003) for a review). One variable shown to influence Δ is the type of tissue used; however, results of these studies were conflicting (Yoneyama et al. 1983; Hobson and Clark 1992; Vanderklift and Ponsard 2003). For example, in a review by Vanderklift and Ponsard (2003), Δ was examined using both whole fish and dorsal muscle, but no clear evidence was established regarding the effect of using whole fish on Δ, likely because of small sample sizes of whole fish (n = 1, Δ = 3.2).

The use of whole fish in SIA studies has many potential benefits to that of dorsal muscle as prey fish vulnerable to predation by larger fish or age-0 recreational fishes are generally small, making the removal of enough dorsal muscle tissue for isotope analysis difficult. However, isotopic signatures derived from whole fish or pooled muscle tissue from multiple small specimens may alter inherent isotopic variability found in prey fish or age-0 recreational fish samples (Yoshioka et al. 1994; Mitchell et al. 1996; Vander Zanden et al. 1998; Johnson et al. 2002; McIntyre et al. 2006). Moreover, no information is available on the isotopic differences found between whole fish and dorsal muscle and the impacts of removing gut contents of prey fish on their isotopic signatures. Additionally, small variations in Δ may produce significant differences in the output of isotopic mixing models (Ben-David and Schell 2001). Due to the growing use of SIA in food web reconstruction and the limited information that exists on isotopic differences between whole fish and dorsal muscle, the objective of our study was to determine if differences exist in δ¹³C and δ¹⁵N between whole fish, whole fish with the gastrointestinal tract (i.e., gut) removed, and dorsal muscle in five prey fish species and to determine if gut fullness had an effect on isotopic signatures.

Methods and materials

Thirty specimens from five different prey species [rainbow smelt (Osmerus mordax), gizzard shad (Dorosoma cepedianum), yellow perch (Perca flavescens), white bass (Morone chrysops), and spottail shiner (Notropis hudsonius)] were collected from Lake Oahe, South Dakota, in late August 2008 using standard shoreline seining procedures. All specimens of a single species were collected within a single location to minimize potential isotopic disparities caused by differences in feeding patterns or isotopic baselines spatially (McKinney et al. 1999). Upon capture, all fish were placed on ice and transported to the laboratory. All specimens were rinsed with distilled water to remove external matter, and total length and mass were recorded. We excised and weighed the gastrointestinal tract from all individuals. Ten individuals of each species were then randomly assigned to each of the three groups for isotopic analysis. Groups included whole fish (WF – the entire fish used), whole fish with the gut removed (GR), and dorsal muscle tissue (DM – only dorsal muscle used). Samples were placed in individually labeled aluminum trays, dried for 72 h, then ground and homogenized using a mortar and pestle. Samples were analyzed for δ¹⁵N and δ¹³C using a Europa 20-20 mass spectrometer.

We tested for differences in δ¹⁵N and δ¹³C between tissue treatments using an analysis of variance (ANOVA) with a Tukey's significance test to determine if significant differences existed between treatments. In addition, since total length may affect isotopic signatures due to ontogenetic diet shifts and isotopic ratios of δ¹⁵N tend to increase at higher trophic levels (Kling et al. 1992), δ¹⁵N values should be positively correlated with total length (Vander Zanden et al. 1998). To account for potential differences in size or feeding patterns between groups, we used an ANOVA with a Tukey's significance test to examine differences in total length between treatment groups. As gut fullness could also alter isotope signatures of treatment groups, we used an ANOVA with a Tukey's significance test to examine differences in gut weight to total weight ratios between the treatment groups of each species. For all comparisons, we used a Bonferroni correction for multiple comparisons and set the significance level at α = 0.001.

Results

In general, the DM treatment tended to have higher δ¹⁵N and δ¹³C compared to both WF and GR groups for three of the five species studied (i.e., yellow perch, gizzard shad, and white bass; Figure 1). Spottail shiner and rainbow smelt were the only species that had similar isotopic values (both δ¹⁵N and δ¹³C) between the DM and GR treatments. Significant differences in δ¹³C were found between tissue types for white bass and spottail shiner (Table 1). White bass was the only species that showed consistent enrichment in DM for both the δ¹³C and δ¹⁵N between treatments. Other than the DM and GR comparisons for spottail shiner and rainbow smelt, in all other significant pair-wise comparisons, the DM treatments were enriched compared to WF or GR groups.

Figure 1. ANOVA results for differences among tissues (whole fish [WF] – dark bars, gut removed whole fish [GR] – hashed bars, and dorsal muscle tissue [DM] – open bars) in (A) δ¹⁵N and (B) δ¹³C. Those tissues with different letters are significantly different (p < 0.01). Species include gizzard shad (GZD), rainbow smelt (RBS), spottail shiner (STS), white bass (WTB), and yellow perch (YLP) collected from Lake Oahe, South Dakota. Error bars represent one unit of standard error.

Table 1. ANOVA results for differences between whole fish (WF), gut removed whole fish (GR) and dorsal muscle (DM) treatments in δ¹⁵N and δ¹³C for gizzard shad, rainbow smelt, spottail shiner, white bass, and yellow perch collected from Lake Oahe, South Dakota.

		δ^15N		δ^13C
Species	df	F-value	p-value	F-Value	p-value
Gizzard shad	2,27	6.12	0.006	2.51	0.100
Rainbow smelt	2,27	2.96	0.069	0.19	0.830
Spottail shiner	2,27	5.54	0.010	5.947	0.007
White bass	2,27	8.29	0.002	10.33	<0.001
Yellow perch	2,27	19.28	<0.001	1.82	0.181
Note: Due to multiple comparisons, α = 0.001.

No significant differences were observed in total length between any tissue groups (Table 2; Figure 2). Two of the five species showed significant differences in gut weight to total weight (GW/TW). Yellow perch in the DM treatment group had greater GW/TW compared to the WF or GR treatments. For white bass GW/TW ratios were significantly different among all three tissue treatments. However, no relationships were found relating differences in isotope signatures between tissues to differences in GW/TW in any species.

Figure 2. ANOVA results for differences among tissues (whole fish [WF] – dark bars, gut removed whole fish [GR] – hashed bars, and dorsal muscle tissue [DM] – open bars) in (A) total length and (B) the ratio of gut weight to total weight (GW/TW). Those tissues with different letters are significantly different (p < 0.01). Species include gizzard shad (GZD), rainbow smelt (RBS), spottail shiner (STS), white bass (WTB), and yellow perch (YLP) collected from Lake Oahe, South Dakota. Error bars represent one unit of standard error.

Table 2. ANOVA results for differences between whole fish (WF), gut removed whole fish (GR) and dorsal muscle (DM) treatments for total length and gut weight to total weight for gizzard shad, rainbow smelt, spottail shiner, white bass, and yellow perch collected from Lake Oahe, South Dakota.

		Total length		Gut weight/total weight
Species	df	F-Value	p-value	F-Value	p-value
Gizzard shad	2,27	1.21	0.315	1.15	0.332
Rainbow smelt	2,27	0.60	0.556	3.80	0.035
Spottail shiner	2,27	1.09	0.349	2.65	0.089
White bass	2,27	0.38	0.687	62.05	<0.001
Yellow perch	2,27	0.40	0.671	13.20	<0.001
Note: Due to multiple comparisons α was set at 0.001.

Discussion

We found significant differences in isotopic values between DM, WF and GR prey fish samples in three of five species, suggesting that a substantial error can be accrued when analyzing different tissue types for stable isotope analyses and subsequent Δ calculation. For instance, the use of WF yellow perch for Δ estimates of δ¹⁵N on a Lake Oahe food web study would result in a difference of 2‰ [i.e., over half a TP based on 3.4‰ Δ by Post (2002)] compared to the DM of yellow perch. This could significantly alter mixing models and estimation of TP using this species and tissue type. Similarly, if white bass were an important component in the food web, the significant differences in signatures of both isotopes between treatment types could dramatically impact the interpretation of energy flow, food webs and mixing models (Vander Zanden and Rasmussen 2001). Therefore, when trying to minimize the error associated with Δ, a standard tissue type should be used.

Across species, the WF treatment was depleted in δ¹³C for four of five species compared to both DM and GR treatments. Additionally, the WF treatment was depleted in δ¹⁵N for four of five species relative to DM treatments. Several potential explanations may exist for WF treatments being depleted in δ¹⁵N and δ¹³C compared to DM or GR. One reason could be that food items or undigested waste product altered whole prey fish isotopic signatures relative to the DM or GR treatments. However, this comparison was only statistically significant in five (of 10) WF to DM comparisons and one WF to GR comparison. Another reason for the consistent isotopic depletion in WF could be related to the gut weight to fish weight ratios. However, in our study GW/TW ratios did not appear to significantly alter isotopic signatures.

We saw no differences in TL between tissue groups for any species, which was expected. However, we did find significant differences in gut fullness (GW/TW). White bass showed significant differences in gut fullness, as the DM treatment had greater GW/TW compared to the WF treatment and the GR treatment had at least twice the ratio as that of the other two treatment groups. However, due to the randomization process, the fullest guts were in DM and GR treatments and not in the WF treatment. Some of the isotope signature of the whole fish is likely the undigested waste (explaining the lower signature of whole fish relative to other treatment groups), which would not likely be incorporated into predators’ diets (Vanderklift and Ponsard 2003; Jardine et al. 2005; Caut et al. 2009). If gut contents did in fact decrease specimen isotope values, we did not reveal this scenario, since the WF treatment had the lowest GW/TW compared to the other treatments. Including individuals with full stomachs in WF treatments should be considered in future studies. In addition, gut evacuation rates vary among species (Brooke et al. 1996; Irigoien 1998; Miyasaka and Genkai-Kato 2009) by water temperature (Chipps 1998) and by diet (Targett and Targett 1990). Though we did not quantify gut evacuation rates, we collected specimens in August when warmer water temperature likely facilitated high gut evacuation rates. This coupled with differences in gut evacuation rates between species could aid in the non-significant isotopic depletion in the WF treatments relative to other tissues.

Another explanation for the observed differences between WF, GR, and DM treatments may be the incorporation of lipid-rich tissues into the isotopic signatures of WF and GR groups. Generally, white muscle is more δ¹⁵N-enriched compared to the heart, liver, and red muscle, a likely result of increased tuarine or other amino acids in white muscle tissue (Wilson and Poe 1974; Pinnegar and Polunin 1999). By simply removing the gastrointestinal tract, we left those tissues with decreased lipids and potentially lowered the δ¹⁵N signature of the specimen compared to muscle alone. Moreover, lipids are relatively depleted in δ¹³C (DeNiro and Epstein 1977), which could explain the relative similarities in δ¹³C between WF, GR, and DM treatments compared to the differences observed in δ¹⁵N. The DM group was only significantly enriched in δ¹³C for two of five species. In spottail shiners, the DM group was not significantly different than GR, which may be a function of gut contents being assimilated into the δ¹³C signature of the specimen. Future research may focus on removing liver, heart, and other lipid-depleted tissues to determine whether differences are still observed between treatments.

Our results solidify the need to standardize the tissue use in SIA for food web reconstruction, calculating TP, and the development of mixing models. Small differences in Δ can reflect large changes in mixing models (Caut et al. 2009), and one form of variability could be minimized by the use of muscle tissue only (or a standardization of tissues use). Although we recognize that prey fish and age-0 recreational fishes are often too small for SIA, researchers should try to use muscle tissue when possible. If whole fish are to be used, we recommend caution when interpreting the effects on food webs, TP estimates, or mixing models.

Acknowledgments

We thank R. Hanten, K. Potter, K. Edwards, N. Poole and A. Leingang from the South Dakota Department of Game, Fish and Parks, and B. Kline, D. Clay, D. Willis, M. Wuellner, and D. James from South Dakota State University for assisting with specimen collection, technical support, and review of earlier drafts. Funding for this study was provided by Federal Aid in Sport Fish Restoration, Project F-15-R, Study 1515, administered through South Dakota Department of Game, Fish, and Parks. Any use of trade names is for descriptive purposes only and does not imply endorsement by the US government.