Stable isotope turnover rates and fractionation in captive California yellowtail (Seriola dorsalis): insights for application to field studies

Ethics statement

All procedures used in these experiments were in accordance with protocol SW1401 of the SWFSC Animal Care and Use Committee. YOY California yellowtail were collected under California Department of Fish and Wildlife Scientific Collection Permit #SC-12372. Experimental protocols were approved by the National Oceanic & Atmospheric Administration Southwest Fisheries Science Center (NOAA SWFSC) Animal Care & Use Committee. Reporting of methods and results were in compliance with ARRIVE guidelines for animal research⁵³.

Collection and captive husbandry of yellowtail

YOY California yellowtail (14–19 cm) were collected from offshore floating kelp mats near San Diego, CA, USA on September 12, 2012. Fish were caught using unbaited sabiki bait rigs, then immediately transferred to an onboard flow-through holding tank. On land, fish were transported in the holding tank on the trailered vessel to NOAA’s SWFSC Experimental Aquaria Facility in La Jolla, CA where they were transferred via dipnet to holding tanks. Fish were first held in 300 × 150 × 90 cm oval tanks (~ 3200 L) with flow-through, filtered local seawater at local ambient seawater temperature (~ 18 °C) and reared on a diet of Bio-Oregon BioTrout feed pellets. Pellets were presented immediately to newly transferred fish and fish began feeding 0–7 days after capture. Yellowtail were fed pellets 6 days/week to apparent satiation. After 525 days, the now larger yellowtail (42–50 cm) were transferred to a larger circular tank (diameter 3.7 m, capacity 9600 L) with the same filtered seawater at the same ambient temperatures (~ 18 °C) and tagged in the dorsal musculature with uniquely numbered and colored Floy plastic spaghetti tags. Yellowtail were switched to a diet of Pacific mackerel (Scomber japonicus) and market squid (Doryteuthis opalescens), both sourced off the coast of southern California (McRoberts Sales Co.). Yellowtail were fed mackerel and squid (by mass: 62% mackerel, 38% squid) 6 days/week to apparent satiation.

Yellowtail muscle was sampled at t = 0 (the day of diet switch), followed by sampling intervals ranging from 27 to 119 days (mean interval 60 ± 27 days) depending on perceived condition of yellowtail and conditions for sampling. Samples were collected by lowering tank water levels, capturing yellowtail in a rubber knotless net, then transferring fish to vinyl cradles. Biopsy punches (Cook Quick-Core G07821) were used to remove 0.1–0.2 g of white muscle from the dorsal musculature. When possible, fork length (FL; cm) was measured at the time of sampling. Muscle tissue was also collected from the diet (dorsal musculature from mackerel and mantle tissue, with the outer membrane removed, from squid) throughout the study for SIA. Sampling continued until yellowtail were removed due to poor condition, suffered natural mortality, or became too large to remain in the holding tank (t = 753 days after diet switch, the endpoint of the study).

YOY yellowtail (13.7–18.8 cm, 15.7 cm ± 1.7; 0.02–0.08 kg; 0.04 kg ± 0.02) fed well on pellet diet, increasing ~ threefold in length and ~ 40-fold in mass over 525 days. After 525 days, 21 similar-sized (42–50 cm FL, 46.0 cm ± 2.5; 1.0–2.2 kg, 1.7 kg ± 0.3) individual yellowtail were selected for the diet switch experiment to fish/squid (Fig. 1). During the course of the diet switch experiment, 11 individual fish were removed from captive conditions before reaching apparent isotopic steady-state due to poor physical condition, lack of feeding, or natural mortality. A total of 10 yellowtail fed consistently in captivity for a long enough period to reach apparent steady-state with new diet (595–753 days) allowing calculation of individual turnover rates in these fish. This allowed for estimates of Δ¹⁵N_low and Δ¹³C_low and population-wide isotopic turnover estimates from 21 fish subjected to the diet switch, and Δ¹⁵N_high and Δ¹³C_high and individual yellowtail turnover estimates from 10 individuals in captive conditions for 595–753 days that reached steady-state with new diet.

SIA of yellowtail and diet

Yellowtail dorsal muscle tissue and prey muscle tissue samples were immediately stored in cryovials at  − 20 °C. Feed pellets were analyzed whole. All samples were then frozen at  − 80 °C and subsequently lyophilized and ground to a homogenous powder for isotope analysis. The δ¹³C and δ¹⁵N values of all samples were determined at the University of Hawaii using an on-line C–N analyzer coupled with a Delta XP isotope ratio mass spectrometer. Replicate reference materials of atmospheric nitrogen and V-PDB were analyzed every 10 samples, and analytical precision was < 0.2‰ for δ¹³C and δ¹⁵N. Isotope ratios are described by:

where q is the isotope of interest, X is the element of interest, R_A is the ratio of the rare to the common isotope, and R_standard is the isotope standard Air or V-PDB. Isotope values are reported as per mille (‰).

Arithmetic corrections of δ¹³C values

While both chemical and arithmetic lipid extractions have been shown to be effective methods to correct bias in δ¹³C values due to lipid-content, arithmetic corrections preserve sample integrity and simplify sample preparation⁵⁴. Since studies have noted effects of chemical lipid extraction on δ¹⁵N^54,55,56 and suggested separate treatment for δ¹⁵N and δ¹³C analyses, and biopsy samples did not always provide adequate material for such treatment (especially from smaller yellowtail), we chose to arithmetically correct for lipid content. δ¹³C values of yellowtail muscle and diet items (mackerel and squid) were arithmetically lipid-normalized based on mass C:N ratios using muscle- and organism type-specific lipid normalization algorithms⁵⁵. For yellowtail and mackerel, we used a muscle-specific lipid correction algorithm derived from a suite of fish species:

where δ¹³C’ is the arithmetically-corrected δ¹³C value, C:N is the C/N ratio by mass of the specific sample, and P and F are parameter constants based on measurements by Logan et al.⁵⁵. For squid, we used the same equation, with parameters P and F derived from invertebrates which included shortfin squid Illex illecebrosus. Relatively low C:N ratios in squid (3.4 ± 0) led to minimal differences in δ¹³C and δ¹³C’ in squid. Since pellet composition was unknown and no appropriate arithmetic δ¹³C correction was available, bulk δ¹³C values are reported for pellets.

Calculating DTDF

We calculated two DTDFs (Δ¹⁵N and Δ¹³C), one for pellet diet and one for fish/squid diet. For both DTDFs, Δ¹⁵N and Δ¹³C were calculated from the mean difference between yellowtail muscle and respective diet δ¹⁵N and δ¹³C values when yellowtail were at isotopic steady-state. Δ¹⁵N_low and Δ¹³C_low were calculated before the diet switch, after yellowtail fed on pellets for 525 days. Δ¹⁵N_high and Δ¹³C_high were calculated after yellowtail that had reached steady-state with the fish/squid diet using the weighted (by proportion mass in diet) mean δ¹⁵N and δ¹³C values of fish/squid diet. DTDF values were calculated according to the equation:

where Δ_diet represents the diet- (‘low’ or ‘high’) and isotope-specific DTDF, δ_yellowtail is the δ¹⁵N or δ¹³C value of yellowtail that reached steady-state with diet, and δ_diet is the mean δ¹⁵N or δ¹³C value of the food (pellet or fish/squid). For the fish/squid diet, δ¹³C values were arithmetically lipid-corrected⁵⁵ and weighted by the proportional mass of each item in the diet (62% and 38%, respectively). For Δ¹⁵N_low and Δ¹³C_low, we assumed YOY yellowtail had reached steady-state with pellet diet after 525 days. This was supported by the length of time (525 days) and relative growth (mass_final/mass_initial =  ~ 2 kg/0.05 kg =  ~ 40) of yellowtail on pellet feed, both of which are substantially higher than what is demonstrably necessary for small fish to reach steady-state with diet.

Since we calculated two different DTDFs that were diet-dependent, as has been previously demonstrated^14,15, we compared our experimentally-derived DTDF values to the diet-dependent DTDF algorithms reported by Caut et al. 2009 (Δ¹⁵N and Δ¹³C) and Hussey et al. 2014 (Δ¹⁵N only), both of which used DTDFs for fish to derive linear equations for diet-based estimates of DTDFs. We used the fish white muscle equation from Caut et al. 2009 for Δ¹⁵N and Δ¹³C:

and the fish (muscle and/or whole) equation from Hussey et al. 2014 for Δ¹⁵N:

and compared those estimated DTDFs to our experimentally-derived values.

Time-based isotopic turnover

Sequential sampling of individual yellowtail allowed for quantification of turnover rate in (1) all yellowtail that reached steady-state with new diet and (2) the pooled population of captive yellowtail. We used exponential fit models and to quantify yellowtail muscle tissue turnover rate of δ¹³C and δ¹⁵N, as used previously^{24,57,58,59,60}:

where δ_t is the stable isotope value at time t, a and c are parameters derived from the best fit, and λ is a data-derived first-order rate constant. Parameters a and c represent specific parameters: a = difference (‰) between initial and final steady-state values and c is the model-estimated final isotope steady-state value^24,60. The isotope-specific half-life (t_0.5) was then calculated:

for different λ values derived for δ¹⁵N and δ¹³C, for both individual yellowtail and the grouped population. We used a modified equation from Buchheister and Latour²¹ to calculate the time needed to obtain a given percentage (α) of complete turnover:

where t_α/100 is the time needed to attain α% turnover and λ is the data-derived first-order rate constant.

Growth-based isotopic turnover

Fish length (fork length or FL; cm) was recorded at t₀, various time steps throughout the experiment concurrent with tissue sampling, and t_f. Direct measurements of yellowtail mass (kg) were taken at t₀. Some mass measurements were taken with FL during the experiment; however direct measurements of mass during sampling were not always possible due to the difficulty of weighing large, active fish and the priority to minimize stress on captive fish during sampling events. When only length was available for individual yellowtail at specific timesteps, mass at time t (W_t) was estimated using the length–weight equation from Baxter 1960:

Relative gain in mass (W_R, hereafter referred to as ‘relative growth’) was then calculated:

where W_f is the measured final mass and W_i is the initial mass estimate from SL. Using the equation from Ricker⁶¹ for W_f:

where k’ is the group specific growth-rate constant, we derive k’:

and can obtain the growth rate constant k’ for individual fish using relative growth (W_R) and time in captivity t. Hesslein et al.⁶² describes the isotope value of a fish at time t (δ_t) as:

where δ_f is the final isotope value at steady-state with diet, δ_i is the initial isotope value before the diet switch, m is the metabolic turnover constant. This is a modification of Eq. (3), where δ_f = c, (δ_i – δ_f) = a, and (k’ + m) = λ. Thus we calculate λ from Eq. (3), k‘ from Eq. (13), and use Eq. (14) to calculate the metabolic constant m based on turnover rates of δ¹⁵N and δ¹³C²⁴. We can also calculate the amount of relative growth needed to achieve α percent turnover of δ¹³C and δ¹⁵N^21,24:

and growth-based turnover can be calculated:

where G_0.5 is the growth-based half-life and λ is the data-derived rate constant from growth-based model fits to yellowtail muscle δ¹³C and δ¹⁵N. We estimated the proportion of isotopic turnover due to growth (P_g) and the proportion of turnover due to metabolism (P_m) as the proportion of k’ and m, respectively, of the overall isotopic turnover constant λ^21,24,61:

We applied Eqs. 13–20 to δ¹³C and δ¹⁵N values in yellowtail muscle tissue and report growth turnover rate constants and overall estimated contribution of growth and turnover to observed δ¹⁵N and δ¹³C turnover in captive yellowtail.

Applicability to field data

Typical applications of δ¹³C and δ¹⁵N data from field-collected animals include approaches that assume isotopic steady-state with diet (e.g. Bayesian mixing models) and approaches that utilize isotopic values after an assumed or inferred shift in diet and/or habitat (e.g. isotopic clock approaches). Sequential sampling of individual yellowtail allowed for investigation of variability in δ¹³C and δ¹⁵N values at steady-state with pellet feed, at various time intervals after a diet switch, and at steady-state with fish/squid diet. Sequential sampling reduced the influence of individual variability in assessment of isotopic parameters. We compared time after diet switch to three metrics of variability (variance, range, standard deviation) of δ¹³C and δ¹⁵N throughout the experimental period to assess the robustness of approaches assuming steady-state and/or changing δ¹³C and δ¹⁵N values in wild predators.

Since tissue δ¹³C and δ¹⁵N values represent a time-integrated signature of prior feeding, Bayesian mixing model approaches represent both prior and current diet until consumers reach steady-state. To allow comparison of isotopic turnover rates and the time required for Bayesian mixing models to adequately represent current diet after a diet switch, we applied the Bayesian mixing model MixSIR¹⁷ to all timesteps of the diet switch experiment. For inputs, we used δ¹³C and δ¹⁵N from sampled yellowtail, two different DTDFs calculated here (one for pellet feed, one for fish/squid; see Results), and mean δ¹³C and δ¹⁵N values of pellets and fish/squid as diet inputs. For each timestep we ran 10⁴ iterations and uninformative priors. Mixing model outputs allowed for estimation of the time required for the model to represent the actual current diet of yellowtail following the diet switch (100% fish/squid).

Source: Ecology - nature.com