Isotopic incorporation rates for shark tissues from a long-term captive feeding study

Sharks are present throughout estuarine and pelagic habitats, but many species have an elusive lifestyle, which prevents accurate characterization of their diet and habitat use (Baum, 2003; Dulvy et al., 2000; Myers et al., 2007). In addition, shark species may undergo ontogenetic or seasonal shifts in diet or habitat use, and traditional biological methods (i.e. direct observation, stomach/gut content and tagging methods) often require large sample sizes and incur considerable costs to collect robust and unbiased data. A better understanding of diet and habitat use can contribute to conservation and management strategies for rapidly declining shark populations (Myers et al., 2007; Dulvy et al., 2008). One potentially useful tool is stable isotope analysis. However, the biological parameters needed to interpret field isotopic data for large ectotherms are not yet well understood.

Stable isotope analysis depends on naturally occurring variations in stable isotope ratios (i.e. ¹³C/¹²C and ¹⁵N/¹⁴N) to trace energy and nutrient flow from prey to consumers. The isotopic value of a consumer's tissues reflects baseline variations (due to seasonal shifts in primary production and/or location) and trophic level. Carbon and nitrogen isotopes are partitioned (or fractionated) during metabolic processes, such that most consumer tissues are enriched in the heavier isotope (¹³C and ¹⁵N) relative to prey (Martínez del Rio et al., 2009). This difference between consumer and prey is often referred to as a trophic enrichment or discrimination factor (Martínez del Rio et al., 2009). In addition, metabolically active tissues (i.e. blood and muscle) integrate the isotopic composition of resources at incorporation rates that depend on tissue type and organism characteristics (Martínez del Rio et al., 2009). Stable isotopes are often used to study the temporal variation of diet and habitat use in animals, sometimes exploiting tissues with different turnover rates (Dalerum and Angerbjörn, 2005). Among vertebrates, birds, mammals and teleost fishes have been the subjects of most research and controlled feeding experiments (Dalerum and Angerbjörn, 2005).

To estimate tissue-dependent incorporation rates, an experiment should include a switch between two isotopically distinct diets and a time series of tissue samples. Previous feeding experiments with this design estimated average carbon residence times in muscle ranging from 17 days in quail to 555 days in whitefish (Hobson and Clark, 1992; Hesslein et al., 1993). This large difference is the result of contrasting rates of protein incorporation for growth and catabolic turnover between fishes (ectotherms) and birds (endotherms) (Dalerum and Angerbjörn, 2005; Martínez del Rio et al., 2009). Martínez del Rio et al. (Martínez del Rio et al., 2009) proposed that these rates depend on the organisms' body size and metabolic activity – including protein turnover. Incorporation rates seem to be slower in larger than in smaller animals (MacAvoy et al., 2006; Reich et al., 2008; Trudel et al., 2011), and faster in endotherms than in ectotherms (Martínez del Rio et al., 2009). For example, ectothermic lizards incorporate carbon in their blood at fractional rates that are much lower than those of endotherms of the same size (Warne et al., 2010).

Better understanding the dependence of isotopic incorporation on body size will facilitate the use of stable isotope analyses on large animals because incorporation rates can be estimated from body size data (Carleton and Martínez del Rio, 2005). Weidel et al. (Weidel et al., 2011) found a clear allometric relationship between body mass and muscle incorporation rate for teleosts. However, their data include only relatively small fishes (<100 g), and it is unclear whether the relationship holds for other aquatic ectothermic vertebrates. There is only one long-term diet switching experiment (>1 year) featuring an ectotherm over 1000 g – a marine turtle (Reich et al., 2008).

To determine the carbon and nitrogen isotope incorporation rates for elasmobranch plasma, red blood cells and muscle, we conducted a 1250-day diet-switching experiment in leopard sharks (Triakis semifasciata Girard 1855). We expected incorporation rates for carbon and nitrogen to be slower for elasmobrachs than for endotherms, and incorporation rates to differ among tissues. More specifically, we expected the incorporation of both carbon and nitrogen to be more rapid in plasma than in red blood cells and muscle. Finally, our data set allowed us to examine whether Weidel et al.'s (Weidel et al., 2011) allometric relationship applies to larger elasmobranchs.

Leopard sharks were chosen because they are relatively easy to maintain in captivity, locally abundant and not threatened – they are an IUCN species of Least Concern (Carlisle and Smith, 2004). Sharks were caught August 2005–January 2006 in San Francisco Bay during regular otter trawls by the Marine Science Institute (Redwood City, CA, USA). This facility also acclimated the sharks to captivity for at least 1 week; sharks were then transported to Long Marine Laboratory (LML) at the University of California, Santa Cruz (UCSC). The sharks were maintained as pairs in polyethylene tanks (2.3 m diameter, 1.2 m water depth) with a continuous flow of filtered seawater from Monterey Bay (temperature range: 13–17°C; salinity range: 30–34). All sharks were fed squid (Loligo opalescens) from their arrival at LML until 1 August 2007 (approximately 565 days). Six sharks were chosen for the experimental group and switched to a tilapia diet (Oreochromis sp., 300–500 g individuals, farm-raised in Taiwan). Three other sharks remained on the squid diet as the control group. The isotopic values of the control sharks are presented in Kim et al. (Kim et al., 2011). Headless and tail-less tilapia were cut into 20–40 g pieces and given to sharks three times a week. Sharks consumed from 3 to 5% of their body mass at each feeding. The care and sampling protocol for leopard sharks in this study was approved by the UCSC Chancellor's Animal Research Committee (CARC) and is in accordance with Institutional Animal Care and Use Committee (IACUC) standards (permit number Kochp0901).

Plasma, red blood cells and muscle were sampled approximately every 21 days from February 2006 to May 2009 following methods detailed in Kim et al. (Kim et al., 2011). Briefly, sharks were anesthetized in a smaller tank containing 30–50 mg l^–1 of tricaine methanesulfonate (MS-222) dissolved in water and buffered to pH 8. Once sharks lost mobility, measurements of total length (TL; cm) and mass (kg; when possible) were recorded. Blood was taken from the caudal vein (0.7 cc), immediately centrifuged in a no-additive Vacutainer® (BD, Franklin Lakes, NJ, USA) to prevent coagulation, and plasma was transferred into another glass test tube for storage. A muscle biopsy was taken above the lateral line between the two dorsal fins and stored in a glass test tube. All samples were frozen at –20°C until preparation for stable isotope analysis.

Plasma and red blood cell samples were freeze-dried and homogenized before isotopic analysis, but muscle samples were lipid and urea extracted according to methods detailed in Kim and Koch (Kim and Koch, 2011). To extract urea from muscle, biopsies were freeze-dried overnight then loaded into an Accelerated Solvent Extractor cell (ASE®; Dionex, Sunnyvale, CA, USA) between glass fiber filters (GF/F, Whatman, Piscataway, NJ, USA). Each cell was treated with two rinses of petroleum ether for lipid extraction (Dobush et al., 1985) and three rinses of de-ionized water to remove urea. Each rinse was 9 ml of solution at 50°C, 10,342 kPa, and 60% volume for 5 min (Kim and Koch, 2011). After these treatments, muscle samples were dried in an oven set to 50°C or freeze-dried overnight.

Tilapia pieces fed to the sharks were randomly selected throughout the experiment and their isotopic composition was measured. Sub-samples of tilapia (N=16) were individually homogenized in a blender and freeze-dried overnight. Then, ∼10 g (dry mass) were transferred to a glass scintillation vial, decalcified with 10 ml of 0.5 mol l^–1 hydrochloric acid overnight at 4°C, and lipid extracted with three rounds of petroleum ether (10 ml) via sonication for 10 min. After treatments, tilapia samples were further homogenized with a mortar and pestle before isotopic analysis.

Accurate mass measurements required sharks to be out of the water and distressed for prolonged periods of time. Therefore, masses were estimated from TL measurements once a relationship between body mass and length was established. Based on mass, individual growth rates were estimated based on exponential and linear models. Traditional tissue-incorporation models assume exponential growth rates, but we also developed a model to reflect linear growth rates. To determine the best-fit model, we used Akaike's information criterion (AIC) (Anderson et al., 1998), which evaluates the model based on the number of parameters and model fit. Because most controlled feeding studies have limited observations relative to estimated parameters, the small-sample AIC (AIC_c) is more appropriate (Martínez del Rio and Anderson-Sprecher, 2008). The model best supported by data was chosen based on the lower AIC (or AIC_c) score, but relatively small differences between scores (<2) indicate similar support of data for the two models (Martínez del Rio and Anderson-Sprecher, 2008). Overall, the exponential growth model was better supported and, therefore, it is described in detail.

Tilapia, the final diet, had mean (±s.d.) δ¹³C and δ¹⁵N values of –23.2±0.9‰ and 7.9±0.4‰, respectively. These values were significantly different from those of the initial squid diet (mean δ¹³C and δ¹⁵N values equaled –18.5±0.5‰ and 13.3±0.7‰, respectively; Welch's two sample t-test, δ¹³C, t=–19.76, d.f.=18.09, P<0.0001, δ¹⁵N, t=–39.24, d.f.=46.62, P<0.0001). Based on the δ^hX_∞ values from the average exponential growth model and average tilapia isotope values, discrimination factors were calculated for individual tissues (Table 1).

The relationship between mass and TL was well described by a power function (mass=0.00392±0.00258×TL^3.03±0.148, r²=0.82) with an exponent not significantly different from that expected from geometric similarity (≈3, d.f.=123, t=0.20, P=0.84; Fig. 1). Exponential and linear growth models for carbon and nitrogen incorporation were supported by the data equally well (ΔAIC_c ranged from –1 to 3). Therefore, we chose to use the exponential growth model for isotopic incorporation as it has a more transparent interpretation. Mass-specific growth rates ranged from 0.000237 to 0.000785 day^–1 (Fig. 2).

In all cases, a single compartment isotope incorporation model was better supported by the data than a two-compartment model (ΔAIC_c ranged from –55 to –1). Indeed, in a few cases, especially for plasma, the two-compartment model failed to converge. The one-compartment exponential model for both carbon and nitrogen described data adequately well and explained from 52 to 99% of the variation in isotopic values (Table 2). Carbon and nitrogen incorporation rates differed significantly among tissues (repeated-measures ANOVA, carbon, F_1,5=11.31, P=0.0200, nitrogen, F_2,10=9.22, P=0.0054; Figs 3, 4, Table 2). Average incorporation rates differed significantly between plasma, red blood cells and muscle [Tukey's honestly significantly different (HSD) test, P<0.05 for both carbon and nitrogen], but the incorporation rates of carbon and nitrogen did not differ significantly between red blood cells and muscle (Tukey's HSD, P>0.05; Table 2). As predicted, red blood cells and muscle had slower incorporation rates than plasma for both carbon and nitrogen.

The contribution of growth was generally higher in tissues with lower incorporation rates, such as muscle, than in those with higher incorporation rates, such as plasma and red blood cells (Table 3). However, the relationship between growth and incorporation rates was not statistically significant for carbon and nitrogen in any tissues (r²<0.44, P>0.05), except for plasma carbon (r²=0.86; ANOVA, F_1,5=25.43, P=0.0073). The incorporation rates of carbon in five individuals were within the 95% prediction interval of Weidel et al.'s (Weidel et al., 2011) allometric relationship (Fig. 5).

Despite their unique physiology and relatively large size for aquatic ectotherms, the sharks in this study exhibited isotopic incorporation patterns that followed those observed in other organisms (Dalerum and Angerbjörn, 2005). The rate of isotopic incorporation was faster in their plasma than in their red blood cells and muscle (Dalerum and Angerbjörn, 2005 and references therein). As observed in other ectotherms, growth contributed significantly to total isotopic incorporation and the contribution of growth to incorporation rate was higher in plasma than in muscle and red blood cells (Carleton and Martínez del Rio, 2010).

Although many of the patterns observed in this study were similar to those reported previously, we made a few unexpected observations. We observed diet-dependent differences in tissue to diet nitrogen discrimination (Δ¹⁵N_{tissues–diet}). In addition, the incorporation rates of sharks were lower than those observed in other aquatic ectotherms (Hesslein et al., 1993; MacAvoy et al., 2006; Trueman et al., 2005; MacNeil et al., 2006). The contribution of growth to incorporation rates varied among tissues and individuals, and was generally lower than previously reported for aquatic ectotherms (Hesslein et al., 1993; MacAvoy et al., 2001; Trueman et al., 2005; MacNeil et al., 2006). Here we consider these unexpected results.

Leopard sharks had lower carbon and nitrogen discrimination factors during their initial equilibration to a squid diet than when fed on tilapia (Table 1) (Kim et al., 2011). Previous studies have suggested that carbon and nitrogen discrimination factors increase with protein content (Pearson et al., 2003) and decrease with protein quality (Florin et al., 2010). Although sharks are carnivorous, and therefore should not experience substantial variation in protein quantity in their diet, there may be nutritional differences between the proteins of a marine invertebrate (squid) and a freshwater teleost (tilapia). Florin et al. (Florin et al., 2010) varied dietary protein quality in a feeding experiment with rats and found an empirical relationship between the diets' most limiting amino acid and Δ¹⁵N values. Furthermore, the proxy of methionine amino acid content as a percentage of protein for protein quality accurately predicted Δ¹⁵N values of mammals and birds from other previously published studies [see Florin et al. (Florin et al., 2010) and references therein]. The Δ¹⁵N values for a tilapia diet are predicted to be less than for a squid diet [2.0 versus 2.9‰, respectively; protein and methionine data are from the USDA National Nutrient Database (USDA Agricultural Research Service, 2011)], which is contrary to our results. The discrepancy between the Florin et al. (Florin et al., 2010) Δ¹⁵N prediction and our study may occur because methionine requirements differ between omnivorous mammals and birds versus carnivorous sharks. In contrast to Δ¹⁵N values, the mechanism relating Δ¹³C values and protein has not been studied. A comparison of individual amino acid concentrations between sharks and their prey may illuminate why discrimination factors differ among diets (Martínez del Rio et al., 2009).

Tissue-specific incorporation rates were lower than previously reported on other aquatic ectotherms (e.g. Reich et al., 2008; Trudel et al., 2011), including other elasmobranchs (sharks, skates and rays) (e.g. MacNeil et al., 2006; Logan and Lutcavage, 2010). Two previous studies estimated elasmobranch tissue incorporation rates, but tissues did not reach steady state with experimental diets because of the short duration (∼60 days) of these experiments (MacNeil et al., 2006; Logan and Lutcavage, 2010). The reliability of incorporation rate estimates is crucially dependent on whether the experiments have been conducted until tissues are close to an asymptotic value (Martínez del Rio et al., 2009).

In addition to these methodological differences between our study and previous research on elasmobranchs, there are biological reasons to expect lower isotopic incorporation rates in the leopard sharks that we studied. Allometric relationships between body size and isotopic incorporation have been documented in birds (Carleton and Martínez del Rio, 2005) and mammals (Bauchinger and McWilliams, 2009). Weidel et al. (Weidel et al., 2011) found a clear allometric relationship between the rate of ¹³C incorporation into muscle and body mass in freshwater teleosts (Fig. 5). All previously published relationships predict lower isotopic incorporation rates in larger animals compared with smaller animals (Carleton and Martínez del Rio, 2005). Although the leopard sharks in this study were at least four times larger than the fish in the Weidel et al. (Weidel et al., 2011) data set, five out of six individuals were within the estimated 95% prediction interval of the size versus carbon isotope incorporation relationship established for muscle (Fig. 5). The success of Weidel et al.'s (Weidel et al., 2011) relationship in predicting the rate of muscle carbon incorporation of leopard sharks is reassuring. By giving a first-order estimate of the isotopic incorporation in muscle from body mass data, this allometric relationship can yield estimates that will aid future ecological studies that need to interpret field isotopic data for large ectotherms.

Many ectotherms exhibit indeterminate growth: after a rapid growth phase following birth that sometimes continues into the juvenile phase, growth slows (but does not drop to zero) as animals increase in size (Angilletta et al., 2004). Previous studies on turtles (Reich et al., 2008) and lizards (Warne et al., 2010) found slower incorporation rates in older individuals. The leopard sharks in this study had variable mass-specific growth rates (Fig. 2), and although smaller individuals generally had faster growth rates, similar-sized individuals sometimes had different growth rates (i.e. BL versus DL and ES, Fig. 2). In addition, the relationship between growth and incorporation rates was not significant, except for carbon in plasma. These observations suggest that: (1) developmental stage and growth rate should be considered before applying captive experimental results to wild subjects, and (2) there is likely individual variation in incorporation rates that are independent of growth. However, it is important to note that the contribution of growth to tissue incorporation rates was lower in this study than has been previously reported for aquatic (Hesslein et al., 1993; MacAvoy et al., 2001; Trueman et al., 2005; MacNeil et al., 2006) and terrestrial (Reich et al., 2008; Warne et al., 2010) ectotherms. Because growth contributed a relatively small portion to tissue incorporation rates, it is unlikely that elasmobranch isotopic incorporation rates are homogenized by growth (e.g. Reich et al., 2008).

We thank J. Adams, A. Bennett, M. Gorey, L. Krol, S. Perry, S. Rumbolt, A. Sjostrom, A. Thell and C. Spencer for their assistance maintaining and sampling the sharks throughout the duration of this study; the Marine Science Institute (Redwood City, CA, USA) for donating specimens from their educational program for this project; the Monterey Bay Aquarium for donating squid fed to the sharks during their initial equilibrium; H. Lee for help with the data analysis; and two anonymous reviewers for their constructive feedback.