Skip to main content
PLOS ONE logoLink to PLOS ONE
. 2021 Apr 1;16(4):e0248711. doi: 10.1371/journal.pone.0248711

Differences in metabolic rate between two Atlantic cod (Gadus morhua) populations estimated with carbon isotopic composition in otoliths

Szymon Smoliński 1,2,*, Côme Denechaud 1,3, Gotje von Leesen 4, Audrey J Geffen 1,3, Peter Grønkjær 5, Jane A Godiksen 1, Steven E Campana 4
Editor: David P Gillikin6
PMCID: PMC8016290  PMID: 33793572

Abstract

The isotopic composition of inorganic carbon in otoliths (δ13Coto) can be a useful tracer of metabolic rates and a method to study ecophysiology in wild fish. We evaluated environmental and physiological sources of δ13Coto variation in Icelandic and Northeast Arctic (NEA) cod (Gadus morhua) over the years 1914–2013. Individual annual growth increments of otoliths formed at age 3 and 8 were micromilled and measured by isotope-ratio mass spectrometry. Simultaneously, all annual increment widths of the otoliths were measured providing a proxy of fish somatic growth. We hypothesized that changes in the physiological state of the organism, reflected by the isotopic composition of otoliths, can affect the growth rate. Using univariate and multivariate mixed-effects models we estimated conditional correlations between carbon isotopic composition and growth of fish at different levels (within individuals, between individuals, and between years), controlling for intrinsic and extrinsic effects on both otolith measurements. δ13Coto was correlated with growth within individuals and between years, which was attributed to the intrinsic effects (fish age or total length). There was no significant correlation between δ13Coto and growth between individuals, which suggests that caution is needed when interpreting δ13Coto signals. We found a significant decrease in δ13Coto through the century which was explained by the oceanic Suess effect-admixture of isotopically light carbon from fossil fuel. We calculated the proportion of the respired carbon in otolith carbonate (Cresp) using carbon isotopic composition in diet and dissolved inorganic carbon of the seawater. This approach allowed us to correct the values for each stock in relation to these two environmental baselines. Cresp was on average 0.275 and 0.295 in Icelandic and NEA stock, respectively. Our results provide an insight into the physiological basis for differences in growth characteristics between these two cod stocks, and how that may vary over time.

Introduction

Otoliths are calcium carbonate structures forming part of the acoustic-lateralis system in fish. They grow by accretion of new material across the outer surface, a continuous process throughout the life of the individual. Seasonal changes in the growth rate of fish induce the formation of translucent and opaque zones in the otolith, which together typically form annual increments. The incremental growth allows for the estimation of fish age and for attribution of chemical signals to specific and discreet time periods (e.g. calendar years or seasons). Otolith growth is strongly correlated with individual somatic growth [1], and the growth record and chemical composition reflect the physiological state of an individual fish during its life and its responses to changes in the environment [2].

Stable carbon isotope ratios in otoliths (δ13Coto) can provide information about the metabolic state of fish [3]. The carbon in otolith aragonite is drawn from dissolved inorganic carbonate (DIC) in the ambient seawater and metabolically derived carbon released to the blood stream from the respiration [4]. Carbon from metabolic sources is significantly 13C depleted (e.g. from -20 to -17‰) when compared to seawater DIC (e.g., from 0 to 2‰). With knowledge of δ13C in otoliths and environmental sources, it is possible to estimate Cresp—the proportional contribution of the respired (metabolic) carbon in otolith. The relationship between Cresp (estimated to range from 0 to 0.95 [3, 5, 6]) and oxygen consumption has been described for several species, so Cresp is increasingly used as a proxy of field metabolic rate, an important measure of physiological performance in free-ranging organisms. Oxygen consumption or metabolic rate are challenging to measure in field conditions, particularly in aquatic environments. Most of the traditional methods are dedicated to laboratory-based research and their adoption for free-swimming fish is logistically difficult [7]. Thus, the proxy provided by δ13Coto is a valuable tool for ecophysiology [8, 9].

Field metabolic rate is the sum of three components: standard metabolic rate (minimum metabolic rate needed to sustain life at a specified temperature), specific dynamic action (associated with the cost of processing food), and activity metabolism (associated e.g. with swimming, feeding, etc.) [8]. Estimates of field metabolic rate, calculated as Cresp, combined with the estimates of assimilated energy, may provide information on the amount of energy available for growth and reproduction—two biological processes which have a profound influence on population resilience [8, 10]. Thus, Cresp as a proxy of field metabolic rates can be used for the investigation of how fish grow and reproduce under different environmental conditions [9]. It provides a unique method to study physiological ecology in fish at the individual level and across the whole lifetime [3]. Moreover, considering the availability of historical otoliths in archives worldwide, time series of Cresp calculated based on δ13Coto can provide a unique opportunity for long-term retrospective assessment of the physiological performance of fish [3, 8] and changes in the important attributes, e.g. diet variability or environmental tolerance [11].

δ13Coto is affected both by intrinsic and extrinsic processes [12], which can complicate its interpretation and the estimation of field metabolism. Age and growth rate, as well as temperature, trophic position, and depth distribution, are among the factors that have been shown to influence δ13Coto through the changes in the metabolic activity of fish, or δ13C values in fish diet and seawater DIC [13]. The contribution of the intrinsic and extrinsic factors needs to be evaluated in order to interpret δ13Coto changes and to use Cresp as a metabolic proxy [3]. Studies of the chemical composition of otoliths that simultaneously control for environmental (e.g. temperature) or physiological effects (e.g. changes of the growth rate of individual fish) help to disentangle different sources of variation and improve the ecological interpretation of chemical composition of otoliths [1416].

In this study, we measured the carbon isotopic composition of Icelandic and Northeast Arctic (NEA) cod (Gadus morhua) otoliths over a century time scale (1914–2013) and analyzed the influence of intrinsic and extrinsic factors on δ13Coto. Icelandic and NEA cod have ranked among the largest stocks of Atlantic cod in the world and are important components of the marine ecosystem [17]. Icelandic cod occur primarily on the Icelandic shelf and spawn around the island [18], but their main spawning ground is located off southwest Iceland [19]. NEA cod inhabit the Barents Sea and spawn mainly around the Lofoten archipelago along the northwest coast of Norway [20] (Fig 1). The utilization of growth and otolith isotopic data from two different, well-characterized, stocks enabled us to test a wider range of environmental conditions across a significant time period.

Fig 1. Icelandic and Northeast Arctic cod (Gadus morhua) distribution areas.

Fig 1

Sampling regions (dark grey polygons) and isobaths (the light grey lines) are marked. The bounding box of the enlarged map is indicated with a solid polygon. The dashed polygons show the areas from which predicted δ13CDIC values were gathered. Annual mean temperatures at depth for each location, extracted respectively from the selected station (red dot) and the Kola section (solid blue line), are presented at the bottom corners. The map was created based on the bathymetric [21] and shoreline [22] data.

The aims of this study were i) to evaluate environmental and physiological sources of variation in δ13Coto; ii) to investigate the relationship between δ13Coto and fish growth; iii) to estimate Cresp as a metabolic proxy based on δ13Coto. We used univariate and multivariate mixed-effects models and estimated conditional correlations between carbon isotopic composition and growth of fish controlling for different intrinsic and extrinsic effects on both otolith traits (δ13Coto and increment width). We used a stable isotope mixing model to estimate the mean proportion of metabolically derived carbon in otolith carbonate—a metabolic proxy that is corrected for the differences between stocks in the isotopic composition of diet and environment. Our results provide an insight into the physiological basis for differences in growth characteristics between these two cod stocks, and how that may vary over time.

Materials and methods

Otolith sampling and processing

Otoliths of Icelandic and NEA cod were collected from the archives of the Marine and Freshwater Research Institute in Hafnarfjörður, Iceland, and the Institute of Marine Research in Bergen, Norway. All otoliths came from the sampling of the commercial catches and from scientific surveys in the spawning areas in southwest Iceland (1929–2015) and around the Lofoten archipelago (1933–2015) (Fig 1). Using archival information on fish age at capture, we aimed at a random collection of 3 individuals caught at age 10+ per sampling year for each stock (S1 Table of S1 File). Information about the location of catch (detailed geographical coordinates or fishing area), date of catch, and biological parameters (total length and sex) were available for most of the individuals. In total, otoliths of 436 fish (213 from Icelandic and 223 from NEA stock) were collected for the isotopic analysis, representing fish life history over a century time scale (1914–2013).

The otoliths were embedded in epoxy resin and ~1 mm-thick transverse sections were cut through the core to reveal the concentric layers of the annual growth increments. Otolith sections were photographed, and the width of each annual increment was measured from the core to the outer edge along the distal axis following a standard protocol (Fig 2). Age was estimated by one reader for each stock in order to maintain consistency and these estimates (not archival) were used throughout analysis. Each increment was assigned to the year of formation by counting back from the known date of capture and accounting for marginal increment interpretation. Additionally, the majority of otolith increments from both stocks were visually assessed and designated as “spawning zones” (the distinctive increments which are believed to be formed after the onset of sexual maturation) using standard procedures for NEA cod age reading [23, 24]. The first and last increment widths were excluded from further analysis since they may not reflect a whole year of growth. Further details on the otolith sampling, processing, and measurement can be found in [25, 26].

Fig 2. Example of a cod otolith section showing the increment measurement axis (blue) and milled regions (growth increments formed at age 3 and 8 are located between red crosses).

Fig 2

Carbonate powder was milled from individual annual growth increments formed at age 3 and age 8 (Fig 2). In a few cases it was not possible to mill these increments and samples from increments formed at age 2 (0.2% of samples) or age 7 (3.5% of samples) were collected (S2 Table of S1 File). Typically, two carbonate samples were obtained from each otolith. Otoliths were milled with high precision using a computer-controlled Merchantek New Wave MicroMill equipped with a digital camera. Translucent and opaque growth zones were combined in one sample to provide a pooled isotope record for the entire year. Samples were drilled up to 750 μm depth (20–25 passes with a drill depth of 30 μm/pass) to obtain approximately 50 μg of otolith carbonate powder in each sample. Further details on the carbonate powder sampling procedure can be found in [27].

Analysis of carbon isotope ratios was conducted on a MAT 253 mass spectrometer at the Institute of Geosciences, University of Bergen, Norway. Carbon isotopic ratios were reported as in parts per thousand or per mil (‰) units with respect to the Vienna Pee Dee Belemnite (VPDB) scale using NBS-19 (δ13C = 1.95‰) and NBS-18 (δ13C = -5.01‰) standards [28]

δ13C=[(RsampleRstandard)1]×103,

where R is the 13C/12C. The long-term reproducibility (1σ precision) of the equipment was δ13C ≤0.4‰ for sample sizes greater than 13 μg based on replicate measurements of an internal carbonate standard over a period of months. Data were not corrected for the aragonite acid fractionation.

Temperature data

Annual mean sea temperatures for the Icelandic spawning ground were extracted from Hadley Centre EN4.2.1 dataset for the location 63°00’ N 22°00’ W (Fig 1) over the depths 0–200 m [29]. The subsurface temperature in this area was assumed to be a proxy of the thermal conditions experienced by the Icelandic cod. For NEA cod annual mean sea temperature data was obtained from the stations 3–7 of the Kola section (from 70°30’ N 33°30’ E to 72°30’ N 33°30’ E, Fig 1) over the depths 0–200 m [3032]. NEA cod are resident year-round in the Barents Sea as juveniles, and once mature they only leave for their annual spawning migration lasting a few months. Since most of the temperature variations in the Barents Sea and Lofoten spawning area are driven by the same large-scale climatic factors associated with Atlantic water masses, the Kola section is a good representation of the climatological temperatures within the area occupied by the NEA cod throughout their whole life [20, 33, 34]. Both temperature time series covered the entire period represented by the otolith growth and isotopic composition measurements.

Modeling of otolith traits (δ13Coto and increment width)

Evaluation of environmental and physiological sources of variation in δ13Coto and growth

A linear mixed-effects modeling framework [35, 36] was applied in order to take into account repeated measurements (both δ13Coto and width of otoliths growth increment) from the same individual or year of formation (Table 1). This method allows for the partitioning of the variance observed in the traits of cod otoliths with consideration of the intrinsic and extrinsic sources of variation [37, 38]. Two separate univariate mixed-effects models were developed, with either δ13Coto (representing metabolic processes) or increment width (representing somatic growth) as the response variable. Stock, Age, Sex, and body total length (TL, in cm) were treated as intrinsic effects in both models. The Age term was included in the δ13Coto model as a fixed factor with two levels (3, 8), while in the increment width model Age was included as a continuous covariate. Sex was included in both models as a fixed factor with males, females, and individuals of unknown sex (lack of archival information) as separate groups. The TL term was incorporated in order to correct model estimates and test the potential effects of certain phenotypes (e.g., fish with larger body size at age). We allowed for the interaction between TL and Stock, as well as three-way interaction between Age, Sex, and Stock. The characterization of increments as “spawning zones” was included with the binary term (SZ: yes/no). Only relatively simple random structures, containing random intercepts for individual fish (FishID) and year of otolith increment formation estimated for each stock (StockYear), were tested in the model due to the limited number of observations. This allowed us to assess the magnitude of the variation associated with between-individual and between-year differences [39]. Increment width and Age were log-transformed before the analysis to meet model assumptions of normality and homogeneity of variance. Fixed covariates (Age, TL, AnomT, Year) were mean-centered in order to facilitate model convergence [37].

Table 1. General workflow and variables used in the consecutive steps of the statistical analysis.
Step Aim Modeling framework Response variable Random structure Fixed intrinsic effects Fixed extrinsic effects Mix sources
1. Evaluation of environmental and physiological sources of variation in δ13Coto and growth Univariate models δ13Coto or otolith increment width Fish ID, StockYear Stock, Sex, TL, Age, SZ AnomT, Year
2. Investigation of the relationship between δ13Coto and growth Bivariate model δ13Coto and otolith increment width Fish ID, StockYear Stock, Sex, TL, Age AnomT, Year
3. Estimation of Cresp as a metabolic proxy Stable isotope mixing model δ13Coto Fish ID, StockYear δ13Cdiet, δ13CDIC

Fish ID—unique identifier of the fish individual; StockYear—stock-specific effect of the year; TL—total body length of fish; SZ—otolith “spawning zone” (included only in the univariate modeling of δ13Coto); AnomT- temperature anomaly for the given area.

Mixed-effects models with different levels of complexity were compared using the Akaike Information Criterion corrected for small sample sizes (AICc) to select the best base model describing variation in otolith traits. The optimal random structure was selected by comparison of models fitted using restricted maximum likelihood (REML) and the most complex fixed structures [37, 38]. The optimal fixed effects were selected by comparison of models fitted using maximum likelihood and previously identified optimal random structures. In an additional analysis, the SZ term was added to the optimal intrinsic model to test for differences in δ13Coto between increments assigned as “spawning zones” or not to test for metabolic signals associated with reproduction. Further, the same optimal intrinsic model was extended by using the anomaly of annual mean temperature calculated separately for each stock (AnomT; based on the data gathered from the profile in SW Iceland and the Kola section). We allowed for the interaction between AnomT and the stock term to test possible stock-specific effects of temperature on otolith traits. Additionally, a continuous Year effect was added to allow for the interaction with the stock term, in order to test for long-term stock-specific trends in the otolith traits. Alternative models were again compared using AICc and the best-ranked model was then refitted with REML, allowing unbiased parameter estimates [35]. Assumptions of the final models were checked and satisfied with standard diagnostics. The significance of fixed effects was assessed based on conditional F-tests with Kenward-Roger approximation for the degrees of freedom [40]. Predicted effects of the variables selected during comparisons of AICc were estimated and visualized. The intraclass correlation coefficient (ICC) was calculated in order to assess between-individual and between-year differences in otolith traits [41, 42]. The conditional and marginal R2 metrics were calculated for both models to assess the amount of variance in otolith traits explained by random effects alone and both fixed and random effects, respectively [43].

Investigation of the relationship between δ13Coto and growth

The variables considered in the analysis (δ13Coto and growth increment width) represent repeated measures of different otolith traits that are not independent of one another [14]. Therefore, bivariate linear mixed-effects models [44] with both measured otolith traits as response variables were developed. Bivariate linear mixed-effects models allow for the simultaneous estimation of the variance of each response and the covariance between them, at group levels specified within the random effects structure [45]. Potential correlations of otolith traits can be decomposed into within-individual, between-individual, and between-year correlations through the partitioning of the variance at different levels of random effects [46]. Importantly, correlation estimates between otolith traits obtained with multivariate mixed-effects models are unbiased and derived with adequate quantification of uncertainty in specified random effects [47].

All bivariate linear mixed-effects models were fitted with the optimal random and fixed effects identified during the development of univariate models [14]. Series of bivariate models were built in order to attribute potential covariances to the set of intrinsic or combined intrinsic and extrinsic fixed effects [46]. Therefore, correlations between increment widths and δ13Coto were tested within the bivariate framework which included i) random effects only, ii) random effects and intrinsic fixed effects, and iii) random effects, intrinsic and extrinsic fixed effects.

Measurements of increment width were available for all years of fish life, while δ13C in the otolith carbonate was measured only from annual growth increments formed at age 3 and 8. However, multivariate linear mixed-effects models are able to deal with missing values for response variables [44]. The bivariate model with both otolith traits as response variables was fitted to the data using a Bayesian approach and Markov Chain Monte Carlo methods [44]. Prior to the fitting, response variables were standardized to ensure similar scale, and growth increment measurements were log-transformed to meet the assumptions of multivariate normality [14]. The estimates of parameters were evaluated based on the model run with parameter-expanded priors for 120,000 iterations with a burn-in phase of 20,000 and a thinning interval of 10. Model assumptions were checked by visual inspection of residuals and by analysis of autocorrelation of the chains. The resulting matrices of within-individual, between-individual, and between-year (co)variance of both otolith traits were used to calculate point estimates and 95% credible intervals, which were further compared between models with different effects incorporated. Correlations where credible intervals did not overlap zero were considered significant.

Estimation of Cresp as a metabolic proxy

Long-term trends in biogenic carbonate δ13C measurements are subject to mis-interpretation due to the Suess effect, i.e. decrease of δ13CDIC in the seawater due to the penetration of isotopically light fossil fuel CO2 into the oceans [48]. Therefore, following recommendations in previous historical and paleoclimate studies [3], the δ13Coto values were corrected prior to the estimation of Cresp with the Bayesian stable isotope mixing model using a Year slope estimated during univariate modeling.

The mean proportion of respiratory carbon in otolith carbonate (Cresp) was estimated through a Bayesian stable isotope mixing model [49], using a two-source input mass balance equation [50]:

δ13Coto=Cresp*δ13Cdiet+(1Cresp)*δ13CDIC+ε,

where δ13Coto is the corrected δ13C measured in the otolith increments, while δ13Cdiet and δ13CDIC are the average δ13C values of the diet and DIC in seawater. Since Cresp represents proportion, it is a unitless quantity. The ε term is the total net isotopic fractionation during carbon exchange (between DIC and blood and between blood and endolymph in which the otolith is formed), which was set to 2.7‰ [13, 51]. The values of δ13CDIC within the depth range occupied by cod in each study area [5256] were estimated based on the apparent oxygen utilization values [57, 58] obtained from the Global Ocean Data Analysis Project version 2—GLODAPv2 database [59] (see S1 File for the detailed description). The values of δ13Cdiet were approximated based on the published information [6064] (see S1 File for the detailed description).

Bayesian models simultaneously consider isotopic variations of both sources (δ13CDIC and δ13Cdiet) and their mixture (δ13Coto) and allow for the proper inclusion of uncertainty [65, 66]. Cresp was estimated separately for the Icelandic and NEA fish, taking into account the hierarchical structure of the data and repeated measurements by including FishID and Year as random effects. The models were run with uninformative priors for 100,000 iterations with a burn-in phase of 50,000 and a thinning interval of 50. Standard Gelman-Rubin and Geweke diagnostics were used for the evaluation of models’ performance.

All analyses were conducted using the R scientific computing language [67] and the following packages: lme4 [68], MCMCglmm [44], MixSIAR [49], MuMIN [69].

Results

Sources of variation in δ13Coto

A univariate model was developed to identify intrinsic and extrinsic effects on δ13Coto (see S1 File for the detailed results of the model selection procedure), which were later used to estimate the correlations between δ13Coto and growth. Otoliths of Icelandic cod had higher δ13Coto than those of NEA cod (Stock effect p<0.001). Fish at age 8 showed higher δ13Coto values than fish at age 3 in both stocks, but the differences were not statistically significant (main Age effect p = 0.409), nor was there a significant sex-related effect (term was excluded during model selection based on AICc). δ13Coto decreased with TL at capture, but the effect was not statistically significant (p = 0.306). Overall, there was a decreasing trend in δ13Coto over the last century (main Year effect p<0.001), with a larger decrease observed in Icelandic cod (mean values of δ13Coto decreased by 0.7‰ over the past 100 years) compared to NEA cod (mean values decreased by 0.3‰ over the past 100 years) (Table 2, Fig 3). There was no significant difference in δ13Coto between “spawning zones” and normal increments (term was excluded during model selection). Temperature did not have a significant effect on δ13Coto (AnomT term was excluded during model selection). Both FishID (ICC = 0.418), and the Year random effects (ICC = 0.111) explained a significant portion of the variance. Overall, random effects (FishID and StockYear) explained the majority (conditional R2 = 0.61), while fixed effects explained only a small fraction of δ13Coto variance (marginal R2 = 0.17) (Table 2).

Table 2. Parameter estimates of the optimal univariate models for otolith carbon isotope ratios and increment width selected with AICc.

Estimates are given for all fixed effects with confidence intervals (CI) and significance (p). For the random effects residual variance (σ2), the variance associated with tested effects (τ00) and their intraclass correlation coefficient (ICC) are given. The number of observations used to fit model and the amount of variance explained (marginal and conditional R2) are specified.

Predictors δ13Coto Increment width
Estimates CI p Estimates CI p
Intercept -1.563 -1.648 –-1.477 <0.001 5.146 5.130–5.162 <0.001
Age (factor)* 0.036 -0.048–0.119 0.409
Age (factor)*:Stock** 0.071 -0.043–0.186 0.223
Year -0.007 -0.010 –-0.004 <0.001
Year:Stock** 0.004 -0.000–0.008 0.080
TL -0.203 -0.591–0.185 0.306 0.301 0.216–0.386 <0.001
Stock** -0.404 -0.522 –-0.286 <0.001 0.041 0.018–0.063 <0.001
AnomT 0.020 -0.004–0.043 0.102
Age (continuous) -0.649 -0.669 –-0.630 <0.001
Age (continuous):Stock** 0.133 0.105–0.161 <0.001
Random Effects
σ2 0.144 0.058
τ00 0.128 FishID 0.005 FishID
0.034 StockYear 0.001 StockYear
ICC 0.418 FishID 0.076 FishID
0.111 StockYear 0.021 StockYear
N of measurements 836*** 4243
N of random groups 436 FishID 436 FishID
183 StockYear 194 StockYear
Marginal R2 / Conditional R2 0.166 / 0.608 0.604 / 0.642

*estimates of the coefficient for increment formed at age 8 in relation to age 3.

**estimates of the coefficient for NEA stock in relation to Icelandic stock.

*** for 36 individuals only 1 measurement of δ13C was available.

Fig 3. Effects of the variables predicted by the univariate model of carbon isotope ratios selected based on the AICc.

Fig 3

The estimated Year slopes (c) which were explained by the oceanic Suess effect were used to correct the δ13Coto values prior to the estimation of Cresp.

Sources of variation in growth

The effects on growth rate were investigated with a univariate model (see S1 File for the detailed results of the model selection procedure). Important intrinsic and extrinsic effects were identified and included in the further step of the analysis in order to estimate correlations between δ13Coto and fish growth rates conditioned on these effects. Fish growth, as represented by increment width, decreased significantly with age (p<0.001), but there were different age-related growth patterns in the two stocks (Age:Stock interaction p<0.001) (Fig 4a). The Icelandic cod grew faster as young fish, and slower as old fish when compared to NEA cod, but there was no detectable effect of sex (this term was excluded during model selection). No significant long-term linear trends were found in the growth data (Year term was excluded during model selection). There was a positive, albeit not statistically significant, effect of temperature anomalies (AnomT p = 0.102) on growth, which was supported by the model selection based on AICc values (Table 2, Fig 4c). A higher proportion of variance in growth was associated with the random effect of FishID (ICC = 0.076) when compared to random effects of Year (ICC = 0.021), but most of the variance in the growth was explained by the fixed effects (marginal R2 = 0.60, conditional R2 = 0.64) (Table 2).

Fig 4. Effects of the variables predicted by the univariate model of otolith increment width selected based on the AICc.

Fig 4

Correlation between δ13Coto and growth

In the bivariate models δ13Coto was significantly correlated with growth within-individuals (R = -0.12; 95% credible interval, CI: -0.22 –-0.02) and between-years (R = -0.30; CI: -0.56 –-0.03). This can be attributed to the intrinsic factors since the correlations diminished after incorporation of intrinsic or both intrinsic and extrinsic factors in the bivariate model. There was no significant correlation of δ13Coto and growth between-individuals (R = -0.06; CI: -0.38–0.26) (Fig 5).

Fig 5. Correlations between δ13Coto and growth increment width (mean value and 95% credible intervals) estimated within-individual, between-individual, and between-years with the models incorporating different sets of effects (indicated with colors).

Fig 5

Significant correlations where credible intervals do not overlap zero are indicated with asterisks.

Estimated Cresp

The estimated average Cresp was 0.275 (±0.006) for Icelandic cod and 0.295 (±0.007) for NEA cod and the difference between stocks was significant (95% credible intervals of the estimated difference did not overlap zero). There was noticeable interannual variability in the Cresp, but the Year random effect estimates were characterized by the high uncertainty. There was no statistically significant synchrony in the estimated mean annual Cresp between the Icelandic and NEA cod (R = -0.17, p = 0.117) (Fig 6).

Fig 6. Year-specific (random effect; left panel; mean±sd) and global (right panel) estimates of the proportion of metabolically derived carbon in otolith carbonate (Cresp) for Icelandic and NEA cod.

Fig 6

Discussion

In Icelandic cod δ13Coto was significantly higher than in NEA cod. Differences in the mean δ13Coto between stocks inhabiting distant areas are expected due to the differences in the oceanographic and ecological conditions, as well as the metabolic activity of the fish [12]. Previous studies, for example in eastern Newfoundland, Canada, indicated that there was a general similarity in δ13Coto within a more limited geographic area [13]. However, even in this relatively limited geographical space of eastern Newfoundland, different areas were characterized by particular lifetime patterns of δ13Coto indicating some degree of separation of individuals throughout their lives [13].

Although the δ13Coto from increments formed at age 3 were generally lower than from increments formed at age 8, the differences among the ontogenetic stages were not statistically significant. Previous studies on Atlantic cod and other species have shown a general increase in δ13Coto with age [12, 50, 70]. This increase has been attributed to varying fish diet or decreasing size-specific metabolic rate as fish get older, which means a lower contribution of the metabolically derived carbon depleted in 13C [3, 5, 8]. It has been noted that δ13Coto values increase through the early period of the life of fish, then reach a maximum around the time of sexual maturation, and this period is followed by a stable plateau or decreasing trend as the fish grow older [50]. A decrease in δ13Coto of mature cod was previously attributed to the migration of the fish to deeper waters characterized by a lower δ13C in DIC [13, 50]. It is probable that a large portion of the samples in our study collected from the annual increments formed at age 8 represents a period where the δ13Coto values had already declined with a changed depth distribution, and therefore, differences between age groups were not highly pronounced.

We also observed a decrease in the δ13Coto with increasing TL. Intrinsic genetic variation in maximum body size and growth among individuals or life-history types are linked with metabolic rate [71]. Because most of the fish analyzed in this study were caught at age 10 or 11, the TL effect reflects differences in life-long growth rate. The faster growing individuals are characterized by lower δ13Coto, suggesting higher incorporation of metabolically derived carbon (depleted in 13C) in addition to the DIC source during otolith calcification [72]. Therefore, the observed trend supports a relationship between δ13Coto and the individual’s metabolic rate, which has been reported for different fish species based on field and experimental observations [6, 8, 9].

There were declining trends in δ13Coto over the last century in both cod stocks. A much stronger decline was observed in Icelandic cod (0.7 over the past 100 years), than for NEA cod (0.3). We found no significant temporal trends in the otolith growth data, suggesting that the trend in δ13Coto was associated with effects other than long-term changes in growth rate. The observed trends in δ13Coto were consistent with direct observations and modeling studies showing a continuous decline in δ13C in the seawater DIC over the postindustrial period, caused by the penetration of isotopically light fossil fuel CO2 into the oceans, referred to as the Suess effect [48]. A strong Suess effect has been observed in well-mixed waters, and the North Atlantic Ocean is among the areas showing the strongest (more than 0.6‰ per century) decrease of δ13C in seawater [48]. A similar decrease (0.5‰ per century) can be calculated based on the δ13Coto values reported for the years 1919–1992 in a previous study on NEA cod otoliths [73]. We mathematically corrected for the Suess effect using estimated linear year trends. Proper calibration of long-term otolith carbon isotopic data concerning postindustrial depletion in seawater 13C is needed before they can be utilized for the estimation of the changes in metabolic rates of fish [3]. Neglecting this effect may seriously hamper the interpretation of the changes in δ13Coto and lead to an overestimation of the metabolic rates in the most recent periods [74]. Because the decrease in the seawater δ13C has a nonlinear form, with a stronger decreasing trend over the last two decades, more precise methods for the correction for the Suess effect than linear approximation should be applied [75].

We found no statistically significant effect of temperature on δ13Coto, but the range of thermal conditions observed in this study may be too small to reveal temperature effects on δ13Coto. We hypothesized that metabolic factors would influence δ13Coto and thus we would expect that increased temperature, leading to higher metabolic rates and oxygen demands, should be reflected in decreased δ13Coto [8]. However, previous experimental and field studies have not consistently shown the same correlation patterns between temperature and δ13Coto [6]. For example, δ13Coto was not significantly affected by water temperature in laboratory-reared plaice (Pleuronectes platessa) [72], a negative correlation was observed in laboratory-reared Atlantic croaker (Micropogonias undulutus) larvae [76], a positive correlation was observed in cod populations in the Northeast Atlantic [70], and a negative correlation in various other fish species in the field [6]. These inconsistencies in the observed relationships suggest a more complex set of controls on δ13Coto variation than temperature alone and apparently reflect a combination of feeding and physiological processes, in addition to temperature [70]. The impact of temperature on δ13Coto would not be expected to be direct, but rather indirect through the influence on fish growth and metabolism [6, 50].

We controlled for different effects on lifelong growth rate in order to properly estimate correlations between δ13Coto and fish growth rates and identify the intrinsic or extrinsic factors which control them [14]. Our predictions from univariate models corroborate previous findings. The growth of fish decreased significantly with age, which is a well-recognized effect [38]. The otolith increment growth proxy indicates that Icelandic cod grew faster in their early ages (1–6) compared to the NEA cod, but the pattern was reversed in the older ages (7 and older). Individuals with higher growth rates throughout their whole life have wider otolith increments and we corrected for this effect by the inclusion of TL term, as higher length at age at point of capture identified faster growing fish. We found positive, albeit not significant, effects of temperature on otolith growth, which indirectly indicates a relationship between fish metabolism and otolith growth [2].

Inter-individual variation in δ13Coto was approximately 4 times higher than interannual variation in δ13Coto. Fish have different personalities expressed as individual behavioral differences, such as aggressiveness or shyness, which can be linked to differences in migration patterns or food consumption [9]. The interpretation of environmental effects can be complicated by different physiological responses and diversity of life histories of individual fish [2]. In our models, a high level of variance associated with the FishID effect, and low sample size, reduce the statistical power to detect environmental effects and estimate population-level changes in δ13Coto in relation to the environmental variability. Potential differences in the δ13Coto between fish from the same locations but characterized by certain life-history types would lead to valuable insights into ecophysiological processes over individual’s whole lifetime. In this study, only the migratory NEA cod were analyzed, and their stock identity confirmed by otolith morphology [77], but the Icelandic samples may have included fish of different ecotypes [78].

We found correlations between δ13Coto and growth within-individuals, which reflect intrinsic processes and the changes in the assimilation of δ13C associated with growth rate differences through the ontogeny. However, we found no significant correlations of δ13Coto and growth between-individuals. Similarly to our results, δ13Coto was not influenced by growth variability within or across nursery sites of herring in Icelandic waters [79]. However, previous studies showed that δ13Coto was negatively related to the growth rate of juvenile plaice (Pleuronectes platessa) reared at 11 °C and tended to increase with growth at 17 °C [72], while δ13Coto was negatively related to the growth rate of cod reared at 6 and 10 °C [5]. The absence of between-individual correlation between δ13Coto and growth and the observed high interindividual variance in δ13Coto suggest that metabolic effects may be obscured in the field studies by a complex set of intrinsic and extrinsic effects. These effects were not fully controlled here in these wild populations, as compared to experimental studies [5, 72].

Individual fish experience periods of higher and lower metabolic activity and growth through the year [80], yet growth and δ13Coto in this study were measured at an annual resolution [13]. Due to uneven otolith accretion rates, annual otolith carbonate samples used in our study represent a weighted mean [27]. Therefore, stronger correlations between δ13Coto and growth may have been obscured. Sampling with higher temporal resolution would be necessary to investigate seasonal changes in δ13Coto associated with differences in growth [81, 82]. These links between growth and the isotopic composition of otoliths have to be considered in order to properly reconstruct past environmental histories based on the analysis of δ13Coto [5].

We found a significant negative correlation between δ13Coto and growth between-years which was attributable to intrinsic factors. Since our otoliths were in general sampled in an age-balanced manner through the study period, we could assign the observed differences in δ13Coto directly to body size, representing a life-time growth rate. We observed lower δ13Coto in otolith carbonates deposited in years characterized by more intensive fish growth. These negative correlations indicate depletion of 13C in otoliths at higher respiration rates [5] and corroborate previous findings and model predictions for cod and other fish species [6, 50]. Typically higher metabolism drives higher consumption and growth, however, higher metabolism can exist also where feeding opportunities are poor and growth is lower [83]. In some cases, despite differences in consumption and metabolic rates, similar growth rates can be observed between fish of varying life-history types [84]. Since growth is a complex physiological phenomenon involving the transformation of food into tissue and the transport of e.g. amino acids, proteins, or lipids in the blood, the chemical composition of otoliths appears to be also an important predictor of growth rate [2].

We calculated Cresp using approximated values of δ13CDIC and δ13Cdiet, which allowed us to correct our estimates for each stock in relation to these two environmental baselines [3]. We observed a higher mean Cresp for the NEA cod stock relative to the Icelandic stock. NEA cod is known for its intensive migratory behavior [20, 85]. The long-term physiological state of NEA cod associated with its migratory nature (i.e. elevated locomotor activity, respiration rate, and oxygen consumption [86]) may therefore explain its higher metabolic rate and Cresp which was inferred through the analysis of δ13Coto. Our results (Cresp = 0.275 and Cresp = 0.295 for Icelandic and NEA cod, respectively) corroborate well with previous studies. For example, the mean Cresp estimated based on δ13Coto for cod specimens collected from different locations in the eastern North Atlantic was 0.2 [70]. A Cresp in the range of 0.07–0.43 was reported for cod in the northeastern Scotian Shelf, Atlantic Canada [50]. The proportion of metabolically derived carbon in the otoliths of larvae and early juvenile cod reared in a controlled laboratory experiment was estimated to be 0.28–0.32 [5]. High agreement of the results obtained in these independent studies suggests that carbon isotope composition in fish otoliths reflect the level of aerobic activity and foraging patterns of wild fish [8, 87]. Retrospective analysis of metabolic history can provide information on important lifestyle attributes, e.g. diet variability, environmental tolerance, or population performance [11].

We based our Cresp estimations on i) measured δ13Coto, ii) predicted δ13CDIC using data on measured AOU, iii) δ13Cdiet assumed based on the information available in the literature, iv) fractionation factor (ε term) adopted from the previous cod otolith studies, and v) mathematical correction for the long-term trend attributed to the oceanic Suess effect. Each element of this methodological approach introduces unavoidable uncertainties to the analysis, but it is the fractionation factor that seems to be the most critical in the estimation of Cresp [81, 88, 89]. The application of different ε term values can cause important differences in calculated Cresp values [88]. We applied ε = 2.7% in order to make our findings comparable with previous results [5, 13, 80], but further studies are needed to investigate the specific values of fractionation factor among species and minimize the potential bias of Cresp estimations [3].

This study assessed different sources of variation in the carbon isotopic composition of otoliths in the two biggest cod stocks of the Atlantic cod. Our results show high inter-individual variation in the δ13Coto signals, which make population-level inferences very difficult. We emphasize the need to consider these inter-individual differences in the analysis of δ13Coto data. Interpretation of δ13Coto signals remains challenging because of the intrinsic effects that can influence δ13Coto, besides environmental conditions. Nonetheless, the consistent differences between Icelandic and NEA cod in their δ13Coto and Cresp provide evidence for underlying physiological basis for the well-documented growth differences between these two stocks. Similarly, observed within-individual and between-years correlations of δ13Coto and growth indicate a link between the metabolic state of fish and the carbon isotopic composition. Analysis of δ13Coto has the potential to indicate changes in the aerobic activity of wild fish [87], but more detailed knowledge on the relationships between fish metabolism and δ13Coto is necessary before δ13Coto and Cresp proxy can be applied to reconstruct the history of the metabolic state of the wild fish populations.

Supporting information

S1 File

(PDF)

Acknowledgments

We thank David P. Gillikin and two anonymous reviewers for their helpful comments on the manuscript.

Data Availability

The datasets generated and analyzed during the current study are available at https://doi.org/10.21335/NMDC-493502289.

Funding Statement

Funding for this work was provided by the Icelandic Research Fund (grant 173906-051) to SEC. This work was also supported by The Eimskip University Fund (project number: 1535-1533127) to GvL. Establishment funding for FARLAB (Facility for advanced isotopic research and monitoring of weather, climate, and biogeochemical cycling) by the Research Council of Norway (grant 245907) is also acknowledged. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Campana SE. Chemistry and composition of fish otoliths pathways, mechanisms and applications. Mar Ecol Prog Ser. 1999;188: 263–297. [Google Scholar]
  • 2.Grønkjær P. Otoliths as individual indicators: a reappraisal of the link between fish physiology and otolith characteristics. Mar Freshw Res. 2016;67: 881–888. 10.1071/MF15155 [DOI] [Google Scholar]
  • 3.Chung M-T, Trueman CN, Godiksen JA, Grønkjær P. Otolith δ13C values as a metabolic proxy: approaches and mechanical underpinnings. Mar Freshw Res. 2019;70: 1747–1756. 10.1071/MF18317 [DOI] [Google Scholar]
  • 4.Kalish JM. Oxygen and carbon stable isotopes in the otoliths. Mar Biol. 1991;110: 37–47. [Google Scholar]
  • 5.Høie H, Folkvord A, Otterlei E. Effect of somatic and otolith growth rate on stable isotopic composition of early juvenile cod (Gadus morhua L) otoliths. J Exp Mar Bio Ecol. 2003;289: 41–58. 10.1016/S0022-0981(03)00034-0 [DOI] [Google Scholar]
  • 6.Kalish JM. 13C and 18O isotopic disequilibria in fish otoliths: metabolic and kinetic effects. Mar Ecol Prog Ser. 1991;75: 191–203. 10.3354/meps075191 [DOI] [Google Scholar]
  • 7.Treberg JR, Killen SS, MacCormack TJ, Lamarre S, Enders EC. Estimates of metabolic rate and major constituents of metabolic demand in fishes under field conditions: Methods, proxies, and new perspectives. Comp Biochem Physiol Part A. 2016;202: 10–22. 10.1016/j.cbpa.2016.04.022 [DOI] [PubMed] [Google Scholar]
  • 8.Chung M-T, Trueman CN, Godiksen JA, Holmstrup ME, Grønkjær P. Field metabolic rates of teleost fishes are recorded in otolith carbonate. Commun Biol. 2019;2: 24. 10.1038/s42003-018-0266-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Martino JC, Doubleday ZA, Chung MT, Gillanders BM. Experimental support towards a metabolic proxy in fish using otolith carbon isotopes. J Exp Biol. 2020;223. 10.1242/jeb.217091 [DOI] [PubMed] [Google Scholar]
  • 10.Vasilakopoulos P, Marshall CT. Resilience and tipping points of an exploited fish population over six decades. Glob Chang Biol. 2015;21: 1834–1847. 10.1111/gcb.12845 [DOI] [PubMed] [Google Scholar]
  • 11.Martino JC, Doubleday ZA, Gillanders BM. Metabolic effects on carbon isotope biomarkers in fish. Ecol Indic. 2019;97: 10–16. 10.1016/j.ecolind.2018.10.010 [DOI] [Google Scholar]
  • 12.Begg GA, Weidman CR. Stable δ13C and δ18O isotopes in otoliths of haddock Melanogrammus aeglefinus from the northwest Atlantic Ocean. Mar Ecol Prog Ser. 2001;216: 223–233. 10.3354/meps216223 [DOI] [Google Scholar]
  • 13.Jamieson RE, Schwarcz HP, Brattey J. Carbon isotopic records from the otoliths of Atlantic cod (Gadus morhua) from eastern Newfoundland, Canada. Fish Res. 2004;68: 83–97. 10.1016/j.fishres.2004.02.009 [DOI] [Google Scholar]
  • 14.Grammer GL, Morrongiello JR, Izzo C, Hawthorne PJ, Middleton JF, Gillanders BM. Coupling biogeochemical tracers with fish growth reveals physiological and environmental controls on otolith chemistry. Ecol Monogr. 2017;87: 487–507. 10.1002/ecm.1264 [DOI] [Google Scholar]
  • 15.Izzo C, Reis-Santos P, Gillanders BM. Otolith chemistry does not just reflect environmental conditions: A meta-analytic evaluation. Fish Fish. 2018;19: 441–454. 10.1111/faf.12264 [DOI] [Google Scholar]
  • 16.Reis-Santos P, Vasconcelos RP, Tanner SE, Fonseca VF, Cabral HN, Gillanders BM. Extrinsic and intrinsic factors shape the ability of using otolith chemistry to characterize estuarine environmental histories. Mar Environ Res. 2018;140: 332–341. 10.1016/j.marenvres.2018.06.002 [DOI] [PubMed] [Google Scholar]
  • 17.Link JS, Bogstad B, Sparholt H, Lilly GR. Trophic role of Atlantic cod in the ecosystem. Fish Fish. 2009;10: 58–87. 10.1111/j.1467-2979.2008.00295.x [DOI] [Google Scholar]
  • 18.Jónsdóttir IG, Campana SE, Marteinsdottir G. Stock structure of Icelandic cod Gadus morhua L. based on otolith chemistry. J Fish Biol. 2006;69: 136–150. 10.1111/j.1095-8649.2006.01271.x [DOI] [Google Scholar]
  • 19.Brickman D, Marteinsdottir G, Logemann K, Harms IH. Drift probabilities for Icelandic cod larvae. ICES J Mar Sci. 2007;64: 49–59. 10.1093/icesjms/fsl019 [DOI] [Google Scholar]
  • 20.Ellertsen B, Fossum P, Solemdal P, Sundby S. Relation between temperature and survival of eggs and first-feeding larvae of northeast Arctic cod (Gadus morhua L.). Rapp P-v Réun Cons int Explor Mer. 1989;191: 209–219. [Google Scholar]
  • 21.Amante C, Eakins B. ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Tech Memo NESDIS NGDC-24. 2009. 10.7289/V5C8276M [DOI] [Google Scholar]
  • 22.Wessel P, Smith WHF. A global, self-consistent, hierarchical, high-resolution shoreline database. J Geophys Res Solid Earth. 2004;101: 8741–8743. 10.1029/96jb00104 [DOI] [Google Scholar]
  • 23.Rollefsen G. The otoliths of the cod. Fisk Skr Ser Havundersøkelser. 1933;4(3): 1–14. Available: https://core.ac.uk/download/pdf/52050284.pdf [Google Scholar]
  • 24.Irgens C, Folkvord A, Otterå H, Kjesbu OS. Otolith growth and zone formation during first maturity and spawning of Atlantic cod (Gadus morhua). Can J Fish Aquat Sci. 2020;77: 113–123. 10.1139/cjfas-2019-0033 [DOI] [Google Scholar]
  • 25.Smoliński S, Deplanque‑Lasserre J, Hjörleifsson E, Geffen AJ, Godiksen JA, Campana SE. Century‑long cod otolith biochronology reveals individual growth plasticity in response to temperature. Sci Rep. 2020;10:16708. 10.1038/s41598-020-73652-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Denechaud C, Smoliński S, Geffen AJ, Godiksen JA, Campana SE. A century of fish growth in relation to climate change, population dynamics and exploitation. Glob Chang Biol. 2020;26: 5661–5678. 10.1111/gcb.15298 [DOI] [PubMed] [Google Scholar]
  • 27.Von Leesen G, Ninnemann US, Campana SE. Stable oxygen isotope reconstruction of temperature exposure of the Icelandic cod (Gadus morhua) stock over the last 100 years. ICES J Mar Sci. 2020;77: 942–952. 10.1093/icesjms/fsaa011 [DOI] [Google Scholar]
  • 28.Brand WA, Coplen TB, Vogl J, Rosner M, Prohaska T. Assessment of international reference materials for isotope-ratio analysis (IUPAC Technical Report). Pure Appl Chem. 2014;86: 425–467. 10.1515/pac-2013-1023 [DOI] [Google Scholar]
  • 29.Good SA, Martin MJ, Rayner NA. EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates. J Geophys Res Ocean. 2013;118: 6704–6716. 10.1002/2013JC009067 [DOI] [Google Scholar]
  • 30.Bochkov Y. Water temperature in the 0–200m layer in the Kola-Meridian in the Barents Sea, 1900–1981. Sb Nauchn Tr PINRO. 1982;46: 113–122 (in Russian). [Google Scholar]
  • 31.Tereshchenko VV. Seasonal and year-to-year variations of temperature and salinity along the Kola meridian transect. ICES CM 1996/C11. 1996; 24 pp. Available: https://www.ices.dk/sites/pub/CMDoccuments/1996/c/1996_c11.pdf [Google Scholar]
  • 32.PINRO. PINRO official website. 2019 [cited 10 Jan 2019]. www.pinro.ru
  • 33.Dippner JW, Ottersen G. Cod and climate variability in the Barents Sea. Clim Res. 2001;17: 73–82. 10.3354/cr017073 [DOI] [Google Scholar]
  • 34.Righton DA, Andersen KH, Neat F, Thorsteinsson V, Steingrund P, Svedäng H, et al. Thermal niche of Atlantic cod Gadus morhua: Limits, tolerance and optima. Mar Ecol Prog Ser. 2010;420: 1–13. 10.3354/meps08889 [DOI] [Google Scholar]
  • 35.Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM. Mixed effects models and extensions in ecology with R. New York: Springer; 2009. [Google Scholar]
  • 36.Harrison XA, Donaldson L, Correa-cano ME, Evans J, Fisher DN, Goodwin CED, et al. A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ. 2018;6:e4794: 1–32. 10.7717/peerj.4794 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Morrongiello JR, Thresher R. A statistical framework to explore ontogenetic growth variation among individuals and populations: a marine fish example. Ecol Monogr. 2015;85: 93–115. [Google Scholar]
  • 38.Weisberg S, Spangler G, Richmond LS. Mixed effects models for fish growth. Can J Fish Aquat Sci. 2010;277: 269–277. 10.1139/F09-181 [DOI] [Google Scholar]
  • 39.Smoliński S, Mirny Z. Otolith biochronology as an indicator of marine fish responses to hydroclimatic conditions and ecosystem regime shifts. Ecol Indic. 2017;79: 286–294. 10.1016/j.ecolind.2017.04.028 [DOI] [Google Scholar]
  • 40.Halekoh U, Højsgaard S. A Kenward-Roger approximation and parametric bootstrap methods for tests in linear mixed models—the R package pbkrtest. J Stat Softw. 2015;59: 1–30. 10.18637/jss.v059.i09 [DOI] [Google Scholar]
  • 41.Wolak ME, Fairbairn DJ, Paulsen YR. Guidelines for estimating repeatability. Methods Ecol Evol. 2012;3: 129–137. 10.1111/j.2041-210X.2011.00125.x [DOI] [Google Scholar]
  • 42.Smoliński S. Sclerochronological approach for the identification of herring growth drivers in the Baltic Sea. Ecol Indic. 2019;101: 420–431. 10.1016/j.ecolind.2019.01.050 [DOI] [Google Scholar]
  • 43.Nakagawa S, Schielzeth H. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol. 2013;4: 133–142. 10.1111/j.2041-210x.2012.00261.x [DOI] [Google Scholar]
  • 44.Hadfield JD. MCMCglmm: MCMC Methods for Multi-Response GLMMs in R. J Stat Softw. 2010;33: 1–22.20808728 [Google Scholar]
  • 45.Houslay TM, Wilson AJ. Avoiding the misuse of BLUP in behavioural ecology. Behav Ecol. 2017;28: 948–952. 10.1093/beheco/arx023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Dingemanse NJ, Dochtermann NA. Quantifying individual variation in behaviour: Mixed-effect modelling approaches. J Anim Ecol. 2013;82: 39–54. 10.1111/1365-2656.12013 [DOI] [PubMed] [Google Scholar]
  • 47.Hadfield JD, Wilson AJ, Garant D, Sheldon BC, Kruuk LEB. The misuse of BLUP in ecology and evolution. Am Nat. 2010;175: 116–125. 10.1086/648604 [DOI] [PubMed] [Google Scholar]
  • 48.Eide M, Olsen A, Ninnemann US, Johannessen T. A global ocean climatology of preindustrial and modern ocean δ13C. Global Biogeochem Cycles. 2017;31: 515–534. 10.1002/2016GB005473 [DOI] [Google Scholar]
  • 49.Stock BC, Jackson AL, Ward EJ, Parnell AC, Phillips DL, Semmens BX. Analyzing mixing systems using a new generation of Bayesian tracer mixing models. PeerJ. 2018;6: e5096. 10.7717/peerj.5096 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Schwarcz HP, Gao Y, Campana S, Browne D, Knyf M, Brand U. Stable carbon isotope variations in otoliths of Atlantic cod (Gadus morhua). Can J Fish Aquat Sci. 1998;55: 1798–1806. 10.1139/f98-053 [DOI] [Google Scholar]
  • 51.Solomon CT, Weber PK, Cech JJ Jr., Ingram BL, Conrad ME, Machavaram M V, et al. Experimental determination of the sources of otolith carbon and associated isotopic fractionation. Can J Fish Aquat Sci. 2006;63: 79–89. 10.1139/f05-200 [DOI] [Google Scholar]
  • 52.Pálsson ÓK, Thorsteinsson V. Migration patterns, ambient temperature, and growth of Icelandic cod (Gadus morhua): evidence from storage tag data. Can J Fish Aquat Sci. 2003;60: 1409–1423. 10.1139/f03-117 [DOI] [Google Scholar]
  • 53.Begg GA, Marteinsdottir G. Environmental and stock effects on spatial distribution and abundance of mature cod Gadus morhua. Mar Ecol Prog Ser. 2002;229: 245–262. 10.3354/meps229245 [DOI] [Google Scholar]
  • 54.Stensholt BK. Cod migration patterns in relation to temperature: Analysis of storage tag data. ICES J Mar Sci. 2001;58: 770–793. 10.1006/jmsc.2001.1067 [DOI] [Google Scholar]
  • 55.Ottersen G, Michalsen K, Nakken O. Ambient temperature and distribution of north-east Arctic cod. ICES J Mar Sci. 1998;55: 67–85. 10.1006/jmsc.1997.0232 [DOI] [Google Scholar]
  • 56.Strand KO, Sundby S, Albretsen J, Vikebø FB. The northeast Greenland shelf as a potential habitat for the Northeast Arctic cod. Front Mar Sci. 2017;4: 1–14. 10.3389/fmars.2017.00304 [DOI] [Google Scholar]
  • 57.Kroopnick PM. The distribution of 13C of ∑CO2 in the world oceans. Deep Sea Res Part A Oceanogr Res Pap. 1985;32: 57–84. Available: 10.1016/0198-0149(85)90017-2. [DOI] [Google Scholar]
  • 58.Filipsson HL, McCorkle DC, Mackensen A, Bernhard JM, Andersson LS, Naustvoll LJ, et al. Seasonal variability of stable carbon isotopes (δ13CDIC) in the Skagerrak and the Baltic Sea: Distinguishing between mixing and biological productivity. Palaeogeogr Palaeoclimatol Palaeoecol. 2017;483: 15–30. 10.1016/j.palaeo.2016.11.031 [DOI] [Google Scholar]
  • 59.Olsen A, Kozyr A, Lauvset SK, Hoppema M, Pérez FF, Steinfeldt R, et al. The Global Ocean Data Analysis Project version 2 (GLODAPv2)–an internally consistent data product for the world ocean. Earth Syst Sci Data. 2016;8: 297–323. [Google Scholar]
  • 60.Pálsson ÓK, Bjrnsson H. Long-term changes in trophic patterns of Iceland cod and linkages to main prey stock sizes. ICES J Mar Sci. 2011;68: 1488–1499. 10.1093/icesjms/fsr057 [DOI] [Google Scholar]
  • 61.Bogstad B, Haug T, Mehl S. Who eats whom in the Barents Sea? NAMMCO Sci Publ. 2000;2: 98–119. 10.7557/3.2975 [DOI] [Google Scholar]
  • 62.Holt RE, Bogstad B, Durant M. Barents Sea cod (Gadus morhua) diet composition: long-term interannual, seasonal, and ontogenetic patterns. ICES J Mar Sci. 2019;76: 1641–1652. [Google Scholar]
  • 63.Thompson DR, Lilliendahl K, Solmundsson J, Furness RW, Waldron S, Phillips RA. Trophic relationships among six species of Icelandic seabirds as determined through stable isotope analysis. Condor. 1999;101: 898–903. [Google Scholar]
  • 64.Tamelander T, Renaud PE, Hop H, Carroll ML, Ambrose WG, Hobson KA. Trophic relationships and pelagic-benthic coupling during summer in the Barents Sea Marginal Ice Zone, revealed by stable carbon and nitrogen isotope measurements. Mar Ecol Prog Ser. 2006;310: 33–46. 10.3354/meps310033 [DOI] [Google Scholar]
  • 65.Yanes Y, Asta MP, Ibáñez M, Alonso MR, Romanek CS. Paleoenvironmental implications of carbon stable isotope composition of land snail tissues. Quat Res (United States). 2013;80: 596–605. 10.1016/j.yqres.2013.08.010 [DOI] [Google Scholar]
  • 66.Parnell AC, Inger R, Bearhop S, Jackson AL. Source partitioning using stable isotopes: Coping with too much variation. PLoS One. 2010;5: 1–5. 10.1371/journal.pone.0009672 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.R Core Team. R: A language and environment for statistical computing. Vienna, Austria; 2018. Available: https://www.r-project.org/ [Google Scholar]
  • 68.Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2014;67: 1–48. 10.18637/jss.v067.i01 [DOI] [Google Scholar]
  • 69.Bartoń K. Package MuMIn: Multi-Model Inference. CRAN; 2019.
  • 70.Weidman CR, Millner R. High-resolution stable isotope records from North Atlantic cod. Fish Res. 2000;46: 327–342. Available: 10.1016/S0165-7836(00)00157-0 [DOI] [Google Scholar]
  • 71.Rosenfeld J, Van Leeuwen T, Richards J, Allen D. Relationship between growth and standard metabolic rate: Measurement artefacts and implications for habitat use and life-history adaptation in salmonids. J Anim Ecol. 2015;84: 4–20. 10.1111/1365-2656.12260 [DOI] [PubMed] [Google Scholar]
  • 72.Geffen AJ. Otolith oxygen and carbon stable isotopes in wild and laboratory-reared plaice (Pleuronectes platessa). Environ Biol Fishes. 2012;95: 419–430. 10.1007/s10641-012-0033-2 [DOI] [Google Scholar]
  • 73.Kalish J, Nydal R, Nedreaas K, Burr G, Eine G. A time history of pre- and post-bomb radiocarbon in the Barents Sea derived from Arcto-Norwegian cod otoliths. Radiocarbon. 2001;43: 843–855. [Google Scholar]
  • 74.Lorrain A, Pethybridge H, Cassar N, Receveur A, Allain V, Bodin N, et al. Trends in tuna carbon isotopes suggest global changes in pelagic phytoplankton communities. Glob Chang Biol. 2020;26: 458–470. 10.1111/gcb.14858 [DOI] [PubMed] [Google Scholar]
  • 75.Estrella-Martínez J, Schöne BR, Thurstan RH, Capuzzo E, Scourse JD, Butler PG. Reconstruction of Atlantic herring (Clupea harengus) recruitment in the North Sea for the past 455 years based on the δ13C from annual shell increments of Arctica islandica. Fish Fish. 2019; 1–15. 10.1111/faf.12362 [DOI] [Google Scholar]
  • 76.Thorrold S, Campana S, Jones C, Swart P. Factors determining δ13C and δ18O fractionation in aragonitic otoliths of marine fish. Geochim Cosmochim Acta. 1997;61: 2909–2919. [Google Scholar]
  • 77.Denechaud C, Smoliński S, Geffen AJ, Godiksen JA. Long-term temporal stability of Northeast Arctic cod (Gadus morhua) otolith morphology. ICES J Mar Sci. 2020;77: 1043–1054. 10.1093/icesjms/fsz259 [DOI] [Google Scholar]
  • 78.Bardarson H, McAdam BJ, Thorsteinsson V, Hjorleifsson E, Marteinsdottir G. Otolith shape differences between ecotypes of Icelandic cod (Gadus morhua) with known migratory behaviour inferred from data storage tags. Can J Fish Aquat Sci. 2017;74: 2122–2130. 10.1139/cjfas-2016-0307 [DOI] [Google Scholar]
  • 79.Macdonald JI, Drysdale RN, Witt R, Cságoly Z, Marteinsdóttir G. Isolating the influence of ontogeny helps predict island-wide variability in fish otolith chemistry. Rev Fish Biol Fish. 2019;30: 173–202. 10.1007/s11160-019-09591-x [DOI] [Google Scholar]
  • 80.Wurster CM, Patterson WP. Metabolic rate of late Holocene freshwater fish: Evidence from δ13C values of otoliths. Paleobiology. 2003;4: 492–505. [Google Scholar]
  • 81.Hanson NN, Wurster CM, Eimf, Todd CD. Reconstructing marine life-history strategies of wild Atlantic salmon from the stable isotope composition of otoliths. Mar Ecol Prog Ser. 2013;475: 249–266. 10.3354/meps10066 [DOI] [Google Scholar]
  • 82.Weidel BC, Ushikubo T, Carpenter SR, Kita NT, Cole JJ, Kitchell JF, et al. Diary of a bluegill (Lepomis macrochirus): daily δ13C and δ18O records in otoliths by ion microprobe. Can J Fish Aquat Sci. 2007;64: 1641–1645. 10.1139/f07-157 [DOI] [Google Scholar]
  • 83.Sherwood GD, Rideout RM, Fudge SB, Rose GA. Influence of diet on growth, condition and reproductive capacity in Newfoundland and Labrador cod (Gadus morhua): Insights from stable carbon isotopes (δ13C). Deep Res Part II Top Stud Oceanogr. 2007;54: 2794–2809. 10.1016/j.dsr2.2007.08.007 [DOI] [Google Scholar]
  • 84.Morinville GR, Rasmussen JB. Early juvenile bioenergetic differences between anadromous and resident brook trout (Salvelinus fontinalis). Can J Fish Aquat Sci. 2003;60: 401–410. 10.1139/f03-036 [DOI] [Google Scholar]
  • 85.Meager JJ, Fernö A, Skjæraasen JE. The behavioural diversity of Atlantic cod: insights into variability within and between individuals. Rev Fish Biol Fish. 2018;28: 153–176. 10.1007/s11160-017-9505-y [DOI] [Google Scholar]
  • 86.Chabot D, Steffensen JF, Farrell AP. The determination of standard metabolic rate in fishes. J Fish Biol. 2016;88: 81–121. 10.1111/jfb.12845 [DOI] [PubMed] [Google Scholar]
  • 87.Sherwood GD, Rose GA. Influence of swimming form on otolith δ13C in marine fish. Mar Ecol Prog Ser. 2003;258: 283–289. 10.3354/meps258283 [DOI] [Google Scholar]
  • 88.Dufour E, Gerdeaux D, Wurster CM. Whitefish (Coregonus lavaretus) respiration rate governs intra-otolith variation of δ13C values in Lake Annecy. Can J Fish Aquat Sci. 2007;64: 1736–1746. 10.1139/f07-132 [DOI] [Google Scholar]
  • 89.Elsdon TS, Ayvazian S, McMahon KW, Thorrold SR. Experimental evaluation of stable isotope fractionation in fish muscle and otoliths. Mar Ecol Prog Ser. 2010;408: 195–205. 10.3354/meps08518 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 File

(PDF)

Data Availability Statement

The datasets generated and analyzed during the current study are available at https://doi.org/10.21335/NMDC-493502289.


Articles from PLoS ONE are provided here courtesy of PLOS

RESOURCES