Experimental support towards a metabolic proxy in fish using otolith carbon isotopes

Metabolic rates reflect the energy used in a system for fuelling biological processes (Gillooly et al., 2001; Nagy, 2005) and are widely used to understand the biology of individuals (performance, growth, reproduction and excretion), populations (population dynamics and interactions) and ecosystems (trophic dynamics and food availability) (Brown et al., 2004; Chabot et al., 2016). As such, metabolic measures can address a wide range of important ecological questions. However, field metabolic rates can be difficult to measure in fish. The doubly labelled water method, which involves measuring the elimination of an introduced oxygen isotope signature, is used for most air-breathing animals but is unsuitable for fish (Nelson, 2016; Treberg et al., 2016). Metabolic rates have therefore been measured using electromyogram telemetry (Cooke et al., 2004; Quintella et al., 2009), heart rate monitoring (Clark et al., 2005; Thorarensen et al., 1996) or accelerometry (Metcalfe et al., 2016). However, these approaches only offer short-term ‘snapshots’ of metabolic rates in live fish. Records conserved in hard-calcified structures offer a valuable alternative to uncovering long-term and retrospective histories. Furthermore, biogeochemical techniques can reconstruct histories when direct measures are not possible, such as historical or extinct species using archaeological samples (Disspain et al., 2016; Wurster and Patterson, 2003), or inaccessible species, such as deep-sea fish (Shephard et al., 2007; Trueman et al., 2013). Otoliths (ear stones) are calcium carbonate (CaCO₃) structures in teleost fishes which permanently retain lifetime chemical signatures that can correspond to ambient environments and physiology (Campana, 1999; Izzo et al., 2018; Sturrock et al., 2015). Alternating translucent and opaque increments form annually in otoliths and arise from seasonal differences in growth rates. Otolith increments provide estimates of somatic growth rate and age, allowing chemical signatures to be matched to specific age and calendar years (Campana and Thorrold, 2001). Additionally, otolith archives are maintained in research institutes and museums worldwide, providing easy access to a wide range of species, locations and years.

Stable carbon isotopes (δ¹³C) in otoliths show potential as a biogeochemical tracer of metabolic rates in fish. Otolith carbon arises from two sources: (1) dissolved inorganic carbon (DIC) in environmental water incorporated via the gill and/or intestinal interfaces and (2) metabolically sourced carbon via cellular respiration of food (Chung et al., 2019a; Kalish, 1991a; McConnaughey et al., 1997). As metabolic demand increases, a concurrent increase in respiration and metabolic CO₂ increases the proportion of metabolically sourced carbon (M_oto) in the blood and endolymph, which is subsequently deposited onto the otolith. Metabolically sourced carbon is significantly depleted in ¹³C compared with DIC in water, and so distinctive changes in δ¹³C in otoliths are thought to primarily reflect shifts in the proportional contributions of carbon sources to otoliths (Dufour et al., 2007; Høie et al., 2003; Kalish, 1991a). The influence of environmental water and diet is considered minimal as δ¹³C in ocean DIC is typically uniform and substantial changes to diet are required to significantly alter isotopic values in otoliths (Shephard et al., 2007; Sherwood and Rose, 2003; Trueman et al., 2016). Previous experimental studies uncovered that δ¹³C correlates with M_oto (Høie et al., 2003; Kalish, 1991a; Solomon et al., 2006). Field studies have also revealed that δ¹³C relates to measures of swimming capacity (Sherwood and Rose, 2003), temperature-driven metabolic cycles (Wurster and Patterson, 2003) and ontogenetic reductions in mass-specific metabolic rate (Trueman et al., 2016; Wurster and Patterson, 2003). To date, only one study has directly related δ¹³C in otoliths to measures of metabolic rate, uncovering a significant negative relationship between δ¹³C and standard metabolic rate (SMR) in Atlantic cod (Chung et al., 2019b). Consequently, a gap exists in understanding how δ¹³C relates to other metabolic parameters. SMR represents the minimum energy usage needed to sustain life, while maximum metabolic rate (MMR) represents the upper limit of metabolic capacity (Chabot et al., 2016, Treberg et al., 2016). Field metabolic rate falls in the middle of these two measures, and is composed of the energy totals from SMR, specific dynamic action (SDA, the postprandial increase in metabolism) and activity metabolism (Smith, 1980; Treberg et al., 2016). Quantifying the relationship between MMR and δ¹³C may provide insight into the influence of activity metabolism on otolith values. Additionally, absolute aerobic scope (AAS) indicates fitness and performance capacity through calculating the difference between MMR and SMR. Linking δ¹³C to MMR or AAS in addition to SMR could help pinpoint the exact metabolic drivers of changes to metabolically sourced carbon into otoliths. Furthermore, metabolic rates and biogeochemical incorporation are highly species-specific because of somatic characteristics and physiological thresholds, so a need exists to understand how this relationship co-varies in a range of species. Additionally, minimal work has been completed to understand δ¹³C as a metabolic proxy across a wide thermal range, so little is known about the relationship under thermal stress.

Oxygen isotopes (δ¹⁸O) in otoliths are a potential proxy of environmental water temperatures. This proxy relationship arises as δ¹⁸O in otoliths are primarily in equilibrium with δ¹⁸O in ambient water, which is influenced by temperature and salinity (Darnaude et al., 2014; Kitagawa et al., 2013; Moreira et al., 2018). Species-specific validations are an important precursor for use of δ¹⁸O as a temperature proxy in the field and have not been assessed in Australasian snapper. Additionally, as fish are ectothermic, temperature is a primary driver of metabolic rate (Clarke and Johnston, 1999; Gillooly et al., 2001). Consequently, there is an inherent relationship between temperature and metabolic rate that is frequently observed between δ¹³C and δ¹⁸O (Geffen, 2012; Kalish, 1991b; Wurster and Patterson, 2003). Reconstructing temperature histories using δ¹⁸O can uncover environmental mechanisms driving metabolic shifts, and aids in our interpretation of δ¹³C and M_oto as metabolic biomarkers.

To advance the development of a metabolic proxy in fish, we experimentally investigated the relationship between δ¹³C and oxygen consumption (as a proxy measure of metabolic rate) in an iconic and valuable fishery species, Australasian snapper (Chrysophrys auratus). Our objectives were to (1) quantify values of δ¹³O and δ¹³C in otoliths and SMR, MMR and AAS of snapper under different temperature treatments, and (2) investigate the functional form and strength of the relationships between isotopic and metabolic parameters.

Australasian snapper, Chrysophrys auratus (Forster 1801), is an iconic long-lived demersal species that supports important commercial and recreational fisheries in the Australasia and Indo-Pacific region (Fowler et al., 2016). Wild snapper eggs were collected in Cockburn Sound, WA, Australia, reared at 25°C and 35 psu at Challenger TAFE, WA, Australia, for 2 months and then transported to the University of Adelaide, SA, Australia, for experiments. Fish were batch marked via immersion in an alizarin complexone (C₁₉H₁₅NO₈) solution (40 mg l⁻¹ of tank water) for 24 h (van der Walt and Faragher, 2003). Fingerlings were separated into 16 tanks (4 treatments×4 replicate tanks) which contained 40 l of locally sourced coastal seawater at 25°C and 40 psu as well as a submerged filter and aerator. After fish had acclimated to local conditions for 7 days, temperature was adjusted by 1.5°C per day until experimental temperatures (20, 24, 28 and 32°C) were reached. Water temperature was measured daily throughout the experiment with a handheld probe (HI-98127, Hanna Instruments, Keysborough, VIC, Australia). Salinity was tested twice weekly using a refractometer (RF-1, Vertex, Toronto, ON, Canada) and maintained at 35 psu via the addition of aged bore water. To maintain water quality, ammonia levels were regularly tested and kept below 0.25 ppm using commercial test kits (API, Chalfont, PA, USA) with 50% water changes occurring weekly. Fingerlings were fed to satiation twice daily with commercial fish food (Skretting Protec, Stavanger, Norway). Fish were exposed to experimental conditions for up to 61 days (Table 1). All animal procedures were approved by the University of Adelaide animal ethics committee (project no. S-2015-161).

Intermittent-flow respirometry was used to measure fish oxygen consumption. Fish were fasted for 24 h prior to respirometry to limit the influence of specific dynamic action (Chabot et al., 2016). The respirometry system consisted of four chambers with a maximum of three chambers at a time used to measure fish oxygen consumption [mg l⁻¹ (ppm)] and the fourth used to measure background respiration (Fig. S1). Each chamber was a closed system attached to a pump (approximately 250 ml s⁻¹), to recirculate water through the system loop, and a FireSting O₂ Optical Oxygen Meter (Pyroscience, Aachen, Germany), which was used to measure oxygen levels (Fig. S1). Logging software (Pyro Oxygen Logger v3.1) was used to record oxygen and time. The oxygen sensor system was calibrated prior to each run. To measure MMR, fish were exerted through a combined exhaustive chase and air exposure procedure (Gilmore et al., 2018; Roche et al., 2013). Fish were then immediately placed into the respirometry chamber and then left for 24 h to reach a resting state. Each respirometry cycle consisted of three stages: a 3 min flush stage where the pump pushed oxygenated water through each system; a 30 s wait stage where the system was closed, awaiting oxygen measurement; and a 12 min measuring period where the pump was off, the loop was closed and oxygen was measured.

Two-factor permutational univariate analysis of variance (ANOVA) was used to determine differences in metabolic rates and isotope values in snapper between temperature treatments (PRIMER v6.0/Permanova). Data were normalised prior to constructing resemblance matrices based on Euclidean distance dissimilarity and analysed by unrestricted permutation of the data for all tests. Within the two-factor design, temperature was a fixed factor and nested replicate tanks a random factor. When significant differences were detected, post hoc pair-wise comparisons were used to identify the source of differences in metabolic rates and isotopic ratios between temperature treatments.

Rearing conditions remained stable throughout the experimental period, with desired temperature levels maintained for each treatment (Table 1). Mortality rate increased sharply at 32°C and no fish survived at this temperature, so were therefore not included in the respirometry experiments. DIC in tank water, fish size (mass and length) and Fulton's condition factor K did not vary among temperatures (Table 1, Table S1). Linear relationships were observed between physiological and isotopic markers (Fig. S2).

SMR ranged from 173.4 to 417.9 mg O₂ kg⁻¹ h⁻¹ (Fig. 1). Significant temperature effects were found with the differences occurring between the 20 and 24°C temperature treatments and the 28°C temperature treatment (Table S1). MMR ranged from 275.4 to 621.1 mg O₂ kg⁻¹ h⁻¹ (Fig. 1). MMR significantly increased with increasing temperature (Table S1). AAS ranged from 80.74 to 393.0 mg O₂ kg⁻¹ h⁻¹ with a mean of 178.7 mg O₂ kg⁻¹ h⁻¹ (Fig. 1). No significant differences between temperature treatments were detected for AAS (Table S1).

Values of δ¹³C in otoliths significantly decreased with increasing temperature (Table S1). Carbon isotope values ranged from −8.13±0.09 at 20°C, to −8.97±0.09 at 24°C and −10.03±0.22 at 28°C (Fig. 2). Values of δ¹⁸O in otoliths also significantly decreased with increasing temperature (Table S1). Oxygen isotope values ranged from −0.07±0.2 at 20°C, to −0.82±0.2 at 24°C and −1.78±0.15 at 28°C (Fig. 2).

A logarithmic curve was modelled between δ¹³C and SMR (Fig. 3A), with a decreasing rate of −2.05±0.41 (t=−5.03, P<0.001) and an intercept (passing through SMR=1) of 2.60±2.27 (t=−1.14, P=0.26). A logarithmic curve was also modelled between δ¹³C and MMR (Fig. 3B), with a decreasing rate of −2.54±0.45 (t=−5.60, P<0.001) and an intercept (passing through MMR=1) of 6.61±2.76 (t=2.4, P<0.05). Logarithmic, linear, curvilinear and exponential models were investigated between δ¹³C and AAS, but the relationship was found to not be suitable for statistical regressions. However, based on average means among temperature treatments, δ¹³C was observed to decrease with an increase in aerobic scope (Fig. 3C). Although high variation was present in the 28°C temperature treatment, the lowest values of δ¹³C were observed in this treatment and corresponded with the highest AAS values. Additionally, while M_oto did not significantly increase with temperature (Table S1, Fig. S3), strong relationships were seen with metabolic measurements. An exponential decay curve (increasing form) was observed between M_oto and measured SMR (Fig. 4A), with a decay constant of 0.007±0.004 (t=1.72, P=0.097) and an upper bound of 27.46±5.8 (t=4.75, P<0.001). However, a stronger exponential curve was observed between M_oto and theoretical SMR (Fig. 4B, Fig. S4), with a decay constant of 0.005±0.002 (t=2.07, P<0.05) and an upper bound of 32.02±7.5 (t=4.75, P<0.001). The relationship between M_oto and measured MMR was non-significant, with a decay constant of 0.002±0.002 (t=1.15, P=0.26; Fig. 4C).

Isotopes in otoliths display exciting potential as retrospective biomarkers in fish. Experimental research is critical to deepen our understanding of these isotopic relationships while minimising confounding factors. We experimentally reared Australasian snapper under different temperature regimes and found a significant negative logarithmic relationship between δ¹³C and both SMR and MMR. Furthermore, an exponential decay curve was found between M_oto and SMR, which was stronger for theoretical SMR based on size and temperature, predicted from the metabolic theory of ecology. Also, δ¹⁸O significantly decreased with increasing temperature, suggesting its potential as a temperature proxy in snapper. This study advances the use of carbon isotopes as a chemical proxy of metabolic rates, establishing confidence in its functioning across a wider range of species and thermal performance ranges.

Our results support previous studies that uncovered relationships between indirect metabolic parameters and δ¹³C (Kalish, 1991a; Sherwood and Rose, 2003; Trueman et al., 2016; Wurster and Patterson, 2003). Carbon isotopes in otoliths are derived from two sources: (1) DIC in environmental water and (2) metabolically sourced carbon from diet (Kalish, 1991a; Solomon et al., 2006; Tohse and Mugiya, 2008). The values of δ¹³C in diet and water in our study remained constant across treatments yet δ¹³C in otoliths were significantly altered, suggesting that physiology was driving the changes. As temperature increased, metabolic demands, higher respiration and food consumption likely increased M_oto, the proportion of metabolically sourced carbon into otoliths (Chung et al., 2019a). M_oto in Australasian snapper has been previously shown to be regulated by temperature and contribute 21–28% to the carbon pool in otoliths (Martino et al., 2019). As such, more negative values of δ¹³C in snapper otoliths were indicative of a higher incorporation of M_oto in fish with higher metabolic rates. A logarithmic curve between δ¹³C and oxygen consumption has been observed directly in Atlantic cod and modelled in other species (Chung et al., 2019b; Kalish, 1991b). The relationships assessed in this study were similarly revealed to be logarithmic curves. Our decreasing rate in logarithmic models of SMR (−2.05) and MMR (−2.54) were within the range found in the literature (−3.52, Kalish, 1991b; and −1.38, Chung et al., 2019b).

Metabolic performance and life history traits in ectotherms can often reflect a bell-shaped curve with increasing temperature (Neuheimer et al., 2011; Pörtner et al., 2017; Schulte, 2015; Pörtner and Farrell, 2008), with metabolic rate increasing until an optimal thermal maximum, after which biological functioning declines rapidly. Although it should be noted that there is debate on whether oxygen is the main limiting factor of ectothermic performance at high temperatures (Jutfelt et al., 2018). Previous experimental research investigating metabolic relationships with δ¹³C focused on experimental and metabolic ranges before or approaching the optimum thermal apex, while field study estimates likely reflected fish functioning in their comfortable thermal ranges in natural environments (Chung et al., 2019b; Kalish, 1991b; Thorrold et al., 1997). Our study presents an investigation into δ¹³C as a metabolic proxy across the thermal performance range, with the highest treatments likely representing conditions beyond the optimum thermal apex (pejus temperature). The distinctive shallowing of the metabolic and δ¹³C relationship at higher metabolic values was likely caused by physiological thresholds. There is a likely upper maximum proportional contribution of metabolically sourced carbon that can be accommodated into otoliths and as the fish approach this value, the relationship between δ¹³C and metabolic rates would evidently plateau (Chung et al., 2019b). The upper threshold of M_oto and, subsequently, the shape of the functional form between δ¹³C and metabolic rates is likely to be species specific as a result of differences in life-history traits and physiology regulations. Consequently, this experimental study is important in corroborating not only the general utility of δ¹³C as a metabolic proxy, but provides insight into species-specific variation through direct experimental measurements of only the second species to date, after Atlantic cod (Chung et al., 2019b). Further research is needed to understand the driving physiological or biochemical mechanism of carbon thresholds in otoliths.

Evidence of the highest temperature treatments (28 and 32°C) being beyond the pejus temperature for snapper is indicated by lower survival rates, suggesting a rapid depression in functioning as a result of thermal stress. Poor survival suggests that 28°C is beyond the optimum thermal range for the species but provides insight into carbon isotopes as a metabolic biomarker beyond the optimum thermal range; however, some intra-individual variability in the 28°C treatment was evident. It is unclear whether this variability is derived from the low sample size or whether fish in this treatment are more variable because of greater individual variability in the response to thermal stress.

This is the first study to investigate how δ¹³C relates to a range of metabolic measurements, which is important in separating out and identifying the best metabolic drivers of changes in M_oto into otoliths. Values of δ¹³C were strongly related to both SMR and MMR, but not AAS. These relationships suggest that both basal energy needed for subsistence living and activity metabolism contribute towards incorporation of δ¹³C into otoliths. As field metabolic rate is composed of SMR, activity metabolism and specific dynamic action, this study supports the appropriate use of δ¹³C as a proxy of metabolic rate in field settings. More work is needed to understand the influence of specific dynamic action on incorporation of δ¹³C into otoliths. Furthermore, while regressions were not suitable for the relationship between δ¹³C and AAS, there was a general trend indicating lower metabolic rates accompanied higher metabolic scopes and providing evidence that our fish were at the higher end of thermal functioning. However, the functional form of the relationship between δ¹³C and AAS is complex and context dependent (Chung et al., 2019b), and further experimental validation is needed to understand this relationship.

The relationship between M_oto and metabolic rates was stronger for theoretical SMR, obtained from estimates derived from the metabolic theory of ecology, than from measured SMR or MMR, derived from measured oxygen consumption of experimental fish. This may be due to measured oxygen consumption reflecting a short window of time (24 h) whereas isotope values reflect a longer period of time – ca. 2 months in this study. SMR may alter slightly between days due to natural fluctuations in behaviour or physiological responses. Fish are known to have different personalities, with individual differences in behaviour such as aggressiveness or shyness leading to differences in movement or acquisition of food (Mittelbach et al., 2014). Personality differences may contribute to differences in M_oto, which is reliant on oxygen and food consumption. Consequently, while we still indicate the strong exponential decay relationship between δ¹³C and metabolic rates, theoretical metabolic rate may be a better indicator for the average fish per mass and temperature range when accounting for a longer time frame and averaging of personality differences between individuals. More precise estimates of metabolic rates could be determined using alternative analytical approaches, such as secondary-ion mass spectrometry (SIMS). Compared with micromilling, SIMS can target more precise regions in otoliths and can create high-resolution lifetime profiles of isotopes, which is particularly valuable when identification of incremental changes or high spatial resolution is warranted for the research question (Hanson et al., 2010; Matta et al., 2013; Weidel et al., 2007).

Additionally, our study suggests the suitability of δ¹⁸O as a temperature proxy in snapper, with δ¹⁸O in otoliths significantly decreasing with temperature. However, more experimental work is needed to explicitly validate this relationship in snapper, particularly with a greater manipulation and assessment of the influence of salinity. A validated δ¹⁸O proxy of temperature would provide valuable insight into metabolic histories. Used in combination, metabolic and temperature histories could expose physiological changes in fish populations in response to environmental change, such as ocean warming. This could be particularly valuable for fish populations that cannot be directly monitored.

Using δ¹³C as a proxy of field metabolic rate requires several considerations. We provide a brief summary of key limitations and solutions, while a comprehensive review is available in Chung et al. (2019a). While significant changes in diet or δ¹³C of ambient water are needed to significantly alter carbon isotopes in otoliths as indicated by simulations (Chung et al., 2019b), some effort can be made to improve the effectiveness of the metabolic tool. Confident use of the metabolic proxy is best for marine environments, particularly deep-sea environments, because of the general uniformity in DIC in these waters (Becker et al., 2016; Kroopnick, 1985; Schmittner et al., 2013; Tagliabue and Bopp, 2008). However, the Suess effect, which is an anthropogenic increase in carbon affecting carbon isotope values in the ocean since the 1800s, is an increasingly important consideration (Chung et al., 2019a; Eide et al., 2017; Keeling, 1979). Model calibrations to predict or reconstruct δ¹³C in ocean waters may assist with using this metabolic tool (Chung et al., 2019a). A metabolic proxy in freshwater fish may be possible (Gerdeaux and Dufour, 2015; Solomon et al., 2006; Weidel et al., 2007), but further experimental work is needed to validate the functional form of the relationship between δ¹³C and metabolic rates, and field studies will need to consider the high variability of δ¹³C in freshwater systems (Bade et al., 2004). A further consideration is that metabolic rate in fish is subject to mass-specific decreases with age which will be observed as increasing δ¹³C across the lifetime of most fish (Fidhiany and Winckler, 1998; Peck and Moyano, 2016; Rosenfeld et al., 2015). Consequently, intra-species comparisons may be limited to comparing rates of metabolic decline across a lifetime or with direct comparisons mostly restricted to within a life-history stage.

We have provided experimental evidence for the quantitative use of otolith δ¹³C as a metabolic proxy in wild fish, by linking δ¹³C in otoliths directly to oxygen consumption across a wide thermal performance range. This is also the first study to explore the relationship between δ¹³C and both MMR and AAS, which was valuable in identifying metabolic drivers of changes to M_oto. We reveal that energy for basic functioning and activity metabolism, both core components of field metabolic rates, relates to incorporation of δ¹³C into otoliths and consequently supports the use of δ¹³C as a metabolic proxy in field settings. Biogeochemical reconstructions using hard-calcified structures like otoliths offer an inexpensive approach to develop long-term metabolic histories in wild fish. These chemical approaches have particular value for populations where direct monitoring is unfeasible, particularly historical or extinct species and inaccessible deep-sea species. Metabolic histories provide insight into physiological responses to environments and trophic interactions. This information will enable ecologists and fishery scientists to identify physiological stressors and, when necessary, to develop effective countermeasures (Wikelski and Cooke, 2006). Consequently, δ¹³C in otoliths can be a powerful intrinsic proxy of field metabolic rate for wide use in conservation and management of fish populations.

We thank Mark Rollog, Robyn Williamson and Grzegorz Skrzypek for providing valuable advice on sample preparation and analysis. We also thank Rob Michael at Challenger TAFE for collecting wild snapper eggs and rearing through to larval fish. We thank our anonymous reviewers for their constructive comments and suggestions which improved the manuscript.