Determination of water balance maintenance in Orcinus orca and Tursiops truncatus using oxygen isotopes Free

Life in the sea requires many physiological, morphological and behavioural adaptations that are acquired during the process of marine colonisation, through the exaptation or the acquisition of new features (Mazin and de Buffrénil, 1996; Houssaye and Fish, 2016; Motani and Vermeij, 2021). One of the greatest challenges for marine vertebrates is to maintain water balance in an environment where total dissolved salt concentrations are approximately 300 times higher than in bodies of fresh water (Irving et al., 1935; Ortiz, 2001; Rash and Lillywhite, 2019). To cope with the salt in their environment, which can be toxic at high concentrations, marine vertebrates have adopted various osmoregulatory strategies including evolutionary adaptations aimed at the conservation of fresh water, thereby avoiding dehydration while excreting excess electrolytes (Thewissen et al., 1996; Ortiz, 2001; Ewing et al., 2017; Costa, 2018; Rash and Lillywhite, 2019). To obtain fresh water, fully aquatic marine vertebrates can hydrate themselves using free water from their prey and water produced endogenously during the breakdown of organic molecules (carbohydrates, proteins, lipids) within the cells (metabolic water), and by voluntary hydration from lenses on the surface of ocean waters following precipitation events or near estuaries (Lillywhite et al., 2014a,b; Rash and Lillywhite, 2019), but also, as suggested but never proven, by consuming ice for polar species. Rash and Lillywhite (2019) recently reviewed the drinking behaviours and water balance maintenance of marine vertebrates (fish, amphibians, marine reptiles, seabirds, pinnipeds, cetaceans, etc.). Fish and marine reptiles, except sea snakes (Lillywhite et al., 2014a,b; Rash and Lillywhite, 2019), drink surrounding seawater (Smith, 1930; Holmes and McBean, 1964; Taplin, 1984; Marshall and Cooper, 1988; Reina et al., 2002). Osmosis is regulated by sea turtles, which excrete the excess salt through their paired lacrimal salt glands (Wyneken et al., 2013; Davenport, 2017), while marine bony fishes remove ions through renal, rectal and branchial excretion (Smith, 1932; Evans and Claiborne, 2008). In cartilaginous fishes, salt excretion occurs through the rectal gland (Fänge and Fugelli, 1963; Silva et al., 1996). These physiological adaptations, which allow salt excess to be excreted, enable these organisms to voluntarily consume salt water and thus maintain their water balance through this source of water. In contrast, cetaceans, which do not have salt glands (Pfeiffer, 1997; Janech et al., 2002; Costa, 2018), maintain their water balance in a very different way.

Studying the water balance of cetaceans in their natural habitat constitutes a challenging task mainly for practical reasons. Early twentieth-century studies of cetacean water balance were mostly performed on small fasted dolphins and porpoises, and it was considered that cetaceans did not drink salt water (Fetcher, 1939; Fetcher and Fetcher, 1942). Later, Telfer et al. (1970) demonstrated that the body water of fasted dolphins was a combination of ingested salt water and metabolic water, while Hui (1981) and Andersen and Nielsen (1983) considered the surrounding salt water to be the main water input in fasted cetaceans. Prey body water and water derived from their metabolism (i.e. metabolic water) are currently considered as the two main sources of water for cetaceans (Ridgway, 1972; Ortiz, 2001; Rash and Lillywhite, 2019) but their respective contributions remain unknown, especially in fed animals.

Box-modelling combined with oxygen isotope data can be helpful to estimate the contribution of each water source to the animal's body water (Kohn, 1996; Langlois et al., 2003; Green et al., 2018; Feng et al., 2022), when water reservoirs (i.e. environmental ocean water, metabolic water and atmospheric water vapour) have significantly different oxygen isotope compositions. To determine the sources contributing to the body water reservoir in vertebrates, which includes free water contained in tissues, cells and blood vessels, the following environmental oxygen sources have been considered in the box-modelling approach: drinking water, vapour water from respiration, food water, metabolic water produced during organic molecule metabolism, and dioxygen used in cellular respiration (Longinelli, 1984; Luz and Kolodny, 1985; Kohn, 1996; Ehleringer et al., 2008; Podlesak et al., 2008; O'Grady et al., 2012). To assess the isotopic composition of body water, the δ¹⁸O value of blood plasma water is commonly measured, as it is regarded as representative of body water isotopic values (Amiot et al., 2007; O'Grady et al., 2012; Green et al., 2018). At equilibrium, i.e. when the input fluxes are equal to the output fluxes, the oxygen isotope composition of the body water can be estimated by considering the amount of oxygen coming from the different sources, their respective oxygen isotope composition and the oxygen isotope fractionations of each flux associated with metabolism as well as input and output fluxes. The most recent study focusing on cetaceans and using the box-modelling approach showed that the oxygen isotope composition of body water is strongly driven by dietary and metabolic water, with more than 50% of the total oxygen flux coming from metabolic water (Feng et al., 2022).

In this study, we tested the hypothesis that the main source of fresh water in cetaceans is dietary free water (prey body water) and the contribution of metabolic water is dependent on the composition of the diet in terms of percentage of lipids. We therefore measured the oxygen isotope composition of blood plasma from specimens of two species of cetaceans raised under human care, the killer whale Orcinus orca (Linnaeus 1758) and the common bottlenose dolphin Tursiops truncatus (Montagu, 1821) and compared them with the oxygen isotope composition of their urine, their surrounding water and their dietary free water during one year at regular intervals. The well-controlled environment of the studied organisms and the significant oxygen isotope differences between each water source allowed the design of a mass balance box model predicting the contributions of water input and output fluxes that control the body water of cetaceans. In the light of the measured dataset and the model output, we clarified the contribution of each water source to the body water of killer whales and common bottlenose dolphins.

The diet of each cetacean was determined by the veterinary team and was prepared and controlled every day by the training teams of the marine zoo Marineland at Antibes (France). The composition of the diet and the mass of each fish constituting the meals of the killer whales and bottlenose dolphins, as well as the seasonal changes in diet required to meet the nutritional needs of the cetaceans, are reported in Table S1. When changes in diet composition occurred in the last 3 days before sampling, the quantities for each species were averaged, as body water is estimated to be completely replenished after 3–5 days (Hui, 1981; Rimbach et al., 2021). Without any change in diet within 3 days before sampling, only the meals of the day before were considered. The oxygen isotope composition of the free body water of each fish species was measured from one specimen randomly chosen from the batch throughout the experiment. Assessment of the nutritional value (percentage water, percentage lipids, percentage proteins) of each fish species and batch was carried out by a private company (Eurofins) as part of this study. Samples of both fish gelatine and ice cubes made from Antibes (France) tap water and given daily to cetaceans as treats were also analysed for their oxygen isotope composition. During all experiments, pool water was collected and its oxygen isotope composition was measured in order to compare it with the oxygen isotope composition of the blood plasma and urine from the cetaceans. In addition, surface water temperature was measured daily for each pool throughout the experiment.

Body fluids (blood plasma and urine) from three adult (OO1, OO2 and OO3) and one subadult (OO4) killer whale Orcinus orca (one female and three males) as well as from seven adult (TT1, TT3, TT4, TT5, TT6, TT7 and TT8) and two subadult (TT2 and TT9) common bottlenose dolphins Tursiops truncatus (four females and five males), hosted at the marine zoo Marineland for several years, were sampled and analysed for their oxygen isotope composition (δ¹⁸O). The common bottlenose dolphins are housed at the zoo in two groups in two different pools (pool 1: TT1–TT5; pool 2: TT6–TT9).

Blood plasma from killer whales and common bottlenose dolphins was sampled once or twice a month from November 2020 to December 2021. Blood samples were taken after overnight fasting, using positive operant conditioning training for killer whales and common bottlenose dolphins. Blood samples were collected aseptically after cutaneous disinfection with povidone iodine (Vetedine solution, Vetoquinol) and medical 70% alcohol, and collected from the ventral peri-arterial venous rete, using a 20 G×3/4 inch or 21 G×3/4 inch winged epicranial micro-fuser (Mirage PIC) for killer whales and common bottlenose dolphins, respectively, mounted on a 20 ml syringe (20 ml BD Luer-Lok, BD Plastipak). After sampling, blood was transferred into a 9 ml lithium heparin tube (Vacuette, Greiner Bio One) and centrifuged for 10 min at 3000 g. Plasma was then collected and transferred to 1.8 ml Eppendorf microtubes.

Urine was collected from killer whales and common bottlenose dolphins, after overnight fasting, using positive operant conditioning training. Urination was spontaneous in killer whales (n=3 from OO1, n=3 from OO2, n=1 from OO3 and n=2 from OO4) and urine was collected manually in a 150 ml straight container with cap (Gosselin™), when the killer whales voluntarily beached themselves on demand for approximately 5 s, exposing their genital region for urine collection. In common bottlenose dolphins, urine was sampled by urethral catheterisation (n=2 on only one specimen TT1), using a CH 4.5 flexible feeding tube (B. Braun) mounted on a 10 ml syringe (10 ml BD Luer-Lok, BD Plastipak), after disinfection with povidone iodine and sterile saline (sodium chloride 0.9%, Osalia) of the vaginal mucosa. Cetacean urine was then transferred to a 150 ml straight container with cap (Gosselin^TM). All samples after collection were stored frozen at −20°C until the oxygen isotope analyses were performed.

Seventeen faecal samples were taken to assess the percentage of water in faeces and thus to estimate the amount of water loss by this way per day in the cetaceans. These samples were taken from three of the four killer whales (OO1, OO3 and OO4) and from the nine bottlenose dolphin specimens for which blood samples were taken (TT1–TT9). An additional sample was taken from a subadult common bottlenose dolphin for which only faeces were sampled (TT10 hosted in pool 2). Faeces were collected from killer whales and common bottlenose dolphins, under positive operant conditioning, with the animals positioned on their back, exposing the anal region out of the water. A single-use tube (for common bottlenose dolphins: CH.14 Levin tube, Dahlhausen medical technology, Cologne, Germany; for killer whales: 9.5 mm Portex Horse stomach tube, Smiths Medical International, Rydalmere, NSW, Australia) was coated with lubricating gel and gently inserted into the rectum of the animal. When faeces are present in the rectum, they spontaneously rise up the tube. The water content of cetacean faeces was estimated by weighing the hydrated sample, then evaporating the water from the bulk faeces in an oven at 20°C and, finally, weighing the dry sample.

Oxygen isotope analyses of the body fluids from the cetaceans, dietary free water, fish gelatine, ice cubes and pool water were performed at the Plateforme d'Ecologie Isotopique (LEHNA; UMR CNRS 5023 Villeurbanne – France). The method is based on an ISOflow™ system connected online in continuous flow mode to a precisION mass spectrometer via a centrION interface. The data processing was performed with the ionOS software suite. Three aliquots of 200 μl of each sample of body fluids from marine vertebrates, water from pools and ice cubes were loaded into LABCO Exetainer^® 3.7 ml soda glass vials and automatically reacted at 40°C with CO₂ for a minimum of 5 h to allow oxygen isotope equilibration between water and CO₂. A few milligrams of sodium azide were added to fish muscle and gelatine samples to limit fermentation processes, which have been shown to lead to poorly reproducible isotope values (Daux et al., 2005; Lazzerini et al., 2016). Oxygen isotope measurements from calibrated waters gave a typical standard deviation of 0.05‰. Calibrated waters were used to anchor the results to the V-SLAP/V-SMOW scale (where V-SLAP is Vienna standard light Antarctic precipitation and V-SMOW is Vienna standard mean ocean water). The calibrated waters used were EE1 (δ¹⁸O_V-SMOW=+6.44‰), Apollo (δ¹⁸O_V−SMOW=−10.05‰), LKD3 (δ¹⁸O_V-SMOW=−20.95‰) and LKD2 (δ¹⁸O_V-SMOW=−26.03‰). Those waters used as working standards were calibrated against waters from the Water Isotope Inter-Comparison intercalibration programme (Wassenaar et al., 2018). Aliquots of Apollo water were placed at the beginning and at the end of each analytical batch to correct for potential instrumental drift with time. Although the SI unit for isotope ratio is the Urey (Ur), for better understanding we are reporting data with delta permil values expressed with respect to V-SMOW with 1‰=1 mUr.

As normality and homoscedasticity of the oxygen isotope composition of blood plasma values from cetaceans (killer whales and common bottlenose dolphins) and oxygen isotope composition of pool water values were not validated (P>0.05; Shapiro–Wilk statistical test), we used the non-parametric Mann–Whitney–Wilcoxon test to compare median values between two observational series. Statistical tests were performed using R software (http://www.R-project.org/) and the level of significance was set at P<0.05.

To estimate the isotope composition of the cetaceans' blood plasma in a complex system of interconnected reservoirs (‘boxes’; Fig. 1), we used the R package isobxr (https://CRAN.R-project.org/package=isobxr). The calculations are based on the mass conservation principles derived from the mathematical formula described by Albarède (1996) and adapted for isotope ratio by Jaouen et al. (2019). In the present design of the model, flux is expressed in grams of oxygen per day (g O day⁻¹).

Oxygen flux in the form of dietary free water (F_FW­) and solid food (F_bio) was calculated from the daily diet (Table S1). The studied killer whales eat between 33 and 104 kg of fish, 1.25 and 3.75 kg of fish gelatine and approximately 16 kg of ice cubes each day, while the daily diet of common bottlenose dolphins is composed of 7.2–15 kg of fish and 0.5 kg of fish gelatine (see Table S1). Cetaceans consume a water-rich diet of fish composed of approximately 70% water and 30% dry matter (lipids and proteins depending of the fish species; Table 1). We estimated that 90% of dietary free water and solid food matter is incorporated into body water of the killer whales and common bottlenose dolphins (Reddy et al., 1994; Lockyer, 2007). The amount of water formed in the first part of lipid and protein catabolism (biomolecular water) is calculated from the assumption that 1 g of protein and lipid produce 0.50 g and 1.07 g of water, respectively (Brody and Lardy, 1946; Costa, 2018). So, oxygen input from free dietary water is 15,900 g O day⁻¹<F_FW<59,000 g O day⁻¹ for the killer whales and 2700 g O day⁻¹<F_FW<7200 g O day⁻¹ for the common bottlenose dolphins, and oxygen from oxygen-bearing biomolecules in dry food (F_bio) is between 2900 and 11,000 g O day⁻¹ for the killer whales and between 490 and 1400 g O day⁻¹ for the common bottlenose dolphins.

Oxygen flux associated with transcutaneous water and that for ingested pool water (F_pool) were grouped because they possess the same water origin, which is the pool water, and thus they share a similar oxygen isotope composition (δ¹⁸O_pool). A previous study has shown that transcutaneous water flux can reach 0.08–0.6 l m² h⁻¹ in Delphinus delphis (Hui, 1981), representing between 4.5 and 36 l for common bottlenose dolphins (4000 g O day⁻¹<F_pool<32,000 g O day⁻¹) and 16–210 l for killer whales considering their cutaneous body surface (14,000 g O day⁻¹<F_pool<187,000 g O day⁻¹). Fasted organisms are supposed to ingest surrounding salt water (Telfer et al., 1970; Hui, 1981) but this is challenged by other studies that consider voluntary drinking behaviour as a rare phenomenon in cetaceans (Kjeld, 2003; Rash and Lillywhite, 2019). Accidental ingestion of water can also be discounted as it was shown that harbour porpoises (Phocoena phocoena) expelled water when they ate their prey (Andersen and Nielsen, 1983). Moreover, cetaceans in this study were fed out of the water by hand by the training team, largely limiting the ingestion of pool water during feeding. In their modelling, Feng et al. (2022) estimated that the combination of these two fluxes represents approximately 10% of the total oxygen flux.

Oxygen input from inhaled air (F_cell) was estimated from experimental studies performed on killer whales and common bottlenose dolphins whose body masses are in the same range as those of the studied cetaceans (Kriete, 1994; Fahlman et al., 2015; 2016; Roos et al., 2016). Common bottlenose dolphins (body mass ∼200 kg) consume between 0.857 and 1.185 l O₂ min⁻¹ representing 1640–2270 g O day⁻¹, while the O₂ consumption of killer whales is between 4.4 and 13.5 l O₂ min⁻¹ (8430 g O day⁻¹<F_cell<25,855 g O day⁻¹) for specimens ranging from 1000 to 5000 kg (Kasting et al., 1989; Kriete, 1994; Roos et al., 2016).

The corresponding flux is between 252 and 775 g O day⁻¹ for the killer whales and between 49 and 68 g O day⁻¹ for the common bottlenose dolphins.

Oxygen losses related to exhaled water vapour (F_VAPe) are between 300 and 700 g O day⁻¹ considering experimental values for killer whales (Kasting et al., 1989) but between 927 and 2844 g O day⁻¹ in relation to body mass if the relationship F_VAPe=0.11×F_cell is considered (Feng et al., 2022). For common bottlenose dolphins and according to the previous relationship (Eqn 8), oxygen output associated with exhaled water vapour is between 180 and 250 g O day⁻¹.

The oxygen output from dejections corresponds to liquid in urine and faeces, here grouped under the output flux F_dej. Very little data about the amount of dejection excreted per day is available except that presented by Ridgway and Wong (2007). Dolphins (T. truncatus and Lagenorhynchus obliquidens) of ∼180 kg produced approximately 4620 g of urine and 1450 g of excrement in 24 h. Urine is considered to be composed of nearly 100% water but for faeces no data have been published yet. By a simple extrapolation, we estimate that a common bottlenose dolphin of 200 kg will lose 5800 g O day⁻¹ through dejections and a killer whale between 43,000 and 103,000 g O day⁻¹ depending on its body mass.

The δ¹⁸O value of atmospheric molecular oxygen is +23.5‰ (Kroopnick and Craig, 1972). Marine mammals have a high oxygen efficiency of X=0.9 because of morphological and physiological adaptation to their environment (Walker, 2007; Wartzok, 2009). According to the relationship between oxygen efficiency in marine mammals and the oxygen isotope composition of inhaled air measured by Epstein and Zeiri (1988), δ¹⁸O of cellular water (δ¹⁸O_cell) approximates as +22.8±0.1‰ V-SMOW. The oxygen isotope composition of atmospheric water vapour is assumed to be δ¹⁸O_VAPi=−15.5±2.7‰ V-SMOW (Uemura et al., 2010) and that of food biomolecules to be δ¹⁸O_bio=+19.2±1.3‰ V-SMOW (Chesson et al., 2011; Table S2). Oxygen isotope composition of pool water (δ¹⁸O_pool) and dietary free water (δ¹⁸O_FW) were, respectively, measured and estimated in this study. Urine has an oxygen isotope composition equal, or close to, that of the blood plasma (Schoeller et al., 1986; Wong et al., 1988; Bryant and Froelich, 1995; Langlois et al., 2003), and faecal water is also assumed to be isotopically unfractionated relative to body water (O'Grady et al., 2012) and thus equal to the urine oxygen isotope composition of urine. Therefore, we assumed for the following box model that δ¹⁸O_dej is equal to δ¹⁸O_urine (Schoeller et al., 1986; Wong et al., 1988; Bryant and Froelich, 1995; Langlois et al., 2003).

Diet composition of the four killer whales and the nine common bottlenose dolphins is detailed in Table S1. The diet of the killer whales was significantly richer in lipids than that of common bottlenose dolphins (1–5% versus ∼3–8%; Mann–Whitney Wilcoxon test, P<0.001), except for one specimen (TT2; ∼10%) whose diet contained a similar percentage of lipids to that of killer whales (Fig. S1). The δ¹⁸O values from fish body water differed according to species and fishing location (Table 1), ranging from −7.69‰ (Oncorhynchus mykiss, rainbow trout, n=1) to 1.43‰ (Mullus barbatus, red mullet, n=1). The δ¹⁸O values of fish gelatine and ice cubes were −8.37‰ (n=1) and −7.70±0.10‰ (±1 s.e.m., n=6), respectively (Table 1). The oxygen isotope composition of water from the prey free water (δ¹⁸O_FW) ingested by each individual was estimated using a mass balance taking into account the percentage of water present in each fish species (Table 1), the oxygen isotope composition of the body water of each of these species (Table 1) and the quantity of each fish species ingested per day (Table S1). The oxygen isotope composition of the three pools in which the killer whales and common bottlenose dolphins (pools 1 and 2) were housed was repeatedly measured from December 2020 to December 2021 and ranged from 0.87‰ to 1.78‰ with a mean value of 1.38‰ (n=21; Table 2). The δ¹⁸O values of the of the killer whale pool ranged from 1.19‰ to 1.67‰ (n=8) during the experiment, while those of the common bottlenose dolphin pools ranged from 0.87‰ to 1.78‰ (n=9) for pool 1 and from 1.04‰ to 1.46‰ (n=5) for pool 2. Water temperature of the three pools ranged from 10.2 to 20.5°C and is also reported in Table 2.

The oxygen isotope composition of body fluids from Cetacea during the experiment revealed inter- and intra-species differences (Table 3). The mean δ¹⁸O_{blood plasma} value of killer whales and common bottlenose dolphins was, respectively, −1.35±0.07‰ (±1 s.e.m., n=29; Table 3) and −0.95±0.13‰ (±1 s.e.m., n=22; Table 3). The δ¹⁸O_urine values were equal to, or slightly lower than, those of the δ¹⁸O_{blood plasma} in both killer whales and common bottlenose dolphins (Table 3; Figs S2 and S3). Intra-individual temporal variability for both cetacean species was observed during the experiment (Figs 2 and 3). Killer whale intra-individual δ¹⁸O_{blood plasma} variability through time was between 0.56‰ (OO3) and 1.30‰ (OO1) while that of common bottlenose dolphins ranged from 0.15‰ (TT3) to 0.74‰ (TT1). The average water content of killer whale and bottlenose dolphin faeces was 83.6% and ranged from 76.9% to 88.8% for killer whales and from 73.2% to 30.1% for common bottlenose dolphins (Table 3). No relationship between faecal water content was observed at the species level.

The δ¹⁸O values of cetacean blood plasma and urine were compared with the δ¹⁸O_pool values of their respective pool. Cetacean δ¹⁸O_{blood plasma} and δ¹⁸O_urine values were significantly lower than the δ¹⁸O_pool values (Mann–Whitney–Wilcoxon test, P<0.001) for both killer whales and common bottlenose dolphins and a non-significant relationship between cetacean δ¹⁸O_{blood plasma} values and δ¹⁸O_pool values was observed (R²=0.01, P=0.70; Fig. S2A). However, a significant relationship (R²=0.39, P=2.4E⁻⁷; Fig. S2B) and covariation was observed between measured δ¹⁸O_{blood plasma} with δ¹⁸O_FW values over time for both killer whales (Fig. 2) and common bottlenose dolphins (Fig. 3). For each blood sample taken, a positive shift in δ¹⁸O values between δ¹⁸O_{blood plasma} and δ¹⁸O_FW was observed in killer whales (Fig. 2). This positive difference in the killer whale specimens was significantly different from and greater than that calculated for common bottlenose dolphin specimens (Mann–Whitney–Wilcoxon statistical test, P<0.001; Fig. 4A) and varied among specimens (Fig. 4B). A significant relationship was observed between ¹⁸O-enrichment values and the percentage of lipids in the diet in Cetacea (R²=0.51, P<0001; Fig. 5).

Model results indicated a strong contribution of dietary free water (61–69%) and metabolic water (here F_bio and F_cell, 26–35%) to the body water of killer whales and common bottlenose dolphins, while more than 98% of the outputs were related to exhaled CO₂ and liquids present in urine and faeces (Table 4). The contributions of inhaled water vapour (VAPi) and surrounding water (pool) were low (<5%). The oxygen isotope compositions of blood plasma predicted by the model closely matched those measured in both killer whales and common bottlenose dolphins (Fig. 6). The differences between measured and predicted values ranged from 0.03‰ to 1.23‰ for killer whales (Fig. 6A) and from 0.01‰ to 0.64‰ for common bottlenose dolphins (Fig. 6B). The linear regression between modelled and measured δ¹⁸O_{blood plasma} values for killer whales and common bottlenose dolphins showed a slope close to 1 (0.98 for killer whales and 1.11 for common bottlenose dolphins; Fig. S3) and an intercept near to 0 (−0.19 for killer whales and −0.06 for common bottlenose dolphins; Fig. S3), thus revealing the robustness of the box-modelling approach developed in this study (P<0.001).

The absence of a relationship between body fluid and pool water δ¹⁸O values indicates that cetaceans do not directly drink surrounding salt water, or at least that surrounding salt water is not their main source of water and oxygen intake as proposed in the box-model (Table 4; Ridgway, 1972; Ortiz, 2001; Rash and Lillywhite, 2019). The δ¹⁸O_{blood plasma} values of killer whales and common bottlenose dolphins were compared with the oxygen isotope composition of their dietary free water (δ¹⁸O_FW), which includes water from fish, fish gelatine and ice cubes. The similar temporal evolution patterns between δ¹⁸O_FW and δ¹⁸O_{blood plasma} in killer whales and common bottlenose dolphins demonstrate that dietary free water is an important source of water in cetaceans (Figs 2 and 3). These observations agree with previous studies using other determination methods which concluded that dietary free water is one of the most important sources of fresh water in cetaceans (Ridgway, 1972; Ortiz, 2001; Rash and Lillywhite, 2019). Nonetheless, we observed that measured δ¹⁸O_{blood plasma} from killer whales was systematically higher than δ¹⁸O_FW estimates (Fig. 2), contrary to common bottlenose dolphin data for which δ¹⁸O_{blood plasma} and δ¹⁸O_FW estimates were very close each other, except for individual TT2 (Fig. 3). Given the δ¹⁸O values of the environment surrounding the cetaceans studied here, this ¹⁸O-enrichment of killer whale blood plasma could be explained either by the ingestion of swimming pool water or by a greater intake of metabolic water.

Regarding the first explanation, pool water can be voluntarily or accidentally ingested by cetaceans during feeding, or through transcutaneous water flux. In the killer whales and common bottlenose dolphins studied here, the food was directly taken from the hand of the feeder, so pool water ingested during feeding was limited and can be considered as minor. Moreover, it has been shown that cetaceans with their large muscular tongue can occlude their oesophagus and eject seawater, thus limiting the ingestion of surrounding water when they swallow their prey underwater (Fetcher and Fetcher, 1942; Telfer et al., 1970; Andersen and Nielsen, 1983). Nonetheless, some studies suggest that ingestion of surrounding water may occur but the amount remains relatively low – a few litres per day for small fasted delphinids (Telfer et al., 1970; Ridgway, 1972; Hui, 1981; Rash and Lillywhite, 2019). Although absorption of water through the skin was thought unlikely to occur in cetaceans (Telfer et al., 1970), later studies reported that the skin is a major in route of water, and that delphinids may experience net gains of fresh water in hypoosmotic habitats (Andersen and Nielsen, 1983; Ridgway and Venn-Watson, 2010). Hui (1981) considered that transcutaneous water is a major input and may account for as much as 75% of the total water flux in a fasting Delphinus delphis. Based on this assumption, ¹⁸O-enrichment should also be observed in studied common bottlenose dolphins, which was not the case.

The second explanation for the ¹⁸O-enrichment observed in killer whales relates to the contribution of water produced by metabolic reactions that use oxygen bound to organic matter for generating H₂O during glycolysis and β-oxidation (biomolecular water production; F_bio) and water formed from cellular metabolism during oxidative phosphorylation (metabolic water; F_met), where inhaled dioxygen (i.e. atmospheric dioxygen, O₂) is the terminal acceptor and is combined with hydrogen to produce water. The oxygen of this ‘metabolic water’ derives from both the oxygen originally presents in the ingested biomolecules of food and the inspired O₂, which have respective δ¹⁸O values of +19.2‰ V-SMOW and +22.8‰ V-SMOW (Kroopnick and Craig, 1972; Gat et al., 1996; Chesson et al., 2011). Experimental studies show that metabolic water production increases when organisms have a lipid-rich diet instead of a protein-rich one (Depocas et al., 1971; Frank, 1988). A second hypothesis could explain this difference in terms of ¹⁸O-enrichment and deals with body mass. This hypothesis has not been further explored, as the common bottlenose dolphin TT2 had an enrichment value close to that of killer whales, while its mass was equivalent to that of the other common bottlenose dolphin specimens (P>0.05, Kendall statistical test). Thus, the most likely explanation for the high ¹⁸O-enrichment value of killer whale body fluids is the percentage of lipids present in the diet, which leads to higher metabolic water production.

The main difference between killer whales and common bottlenose dolphins concerns the contribution of metabolic water to the total body water budget (from biomolecular water and cellular water). The contribution of the water formed from metabolism was higher in killer whales than in common bottlenose dolphins and probably explains the specific ¹⁸O-enrichment observed in the blood plasma values of killer whales. The comparison with previous studies should be made with caution because the considered fluxes are not always the same, especially when organisms were fasted during the study period, which can range from a few hours to the whole day (Hui, 1981; Andersen and Nielsen, 1983). The difference in fasting time probably influences the ingestion of surrounding water, as it was observed that the flow of environmental water was greater in fasting specimens (Hui, 1981; Andersen and Nielsen, 1983). This could be related to the need for the organisms to maintain correct electrolyte levels, partly lost during urination. Thus, this could explain why the ingestion of surrounding water appears greater in previous studies than in the present one, where the fasting time is limited (overnight). It should also be noted that surrounding water can also be ingested when animals are socialising or handling objects. The results of this study lead to markedly different conclusions from those obtained by Feng et al. (2022) where metabolic water was proposed to be the main oxygen input in cetaceans (49%), followed by dietary free water (37%) and surrounding water (8%). This difference in terms of the contributions of each water source can be explained by the more accurate estimates of the dietary free water obtained in the present study with the measurement of the oxygen isotope composition of the free water of each fish species. Indeed, in their study, Feng et al. (2022) took a mean value corresponding to the mean value of the δ¹⁸O of the oceans for all cetaceans whose geographical origin was different. A second possible explanation concerns the feeding frequency of the organisms. In this study, orcas and common bottlenose dolphins were fed daily. In the study by Feng et al. (2022), data come from wild organisms whose feeding frequency was probably different (Barros, 1990; Kastelein et al., 2002). Fasting for shorter or longer periods could explain differences in the contribution of prey water and metabolic water between these studies, and in particular a greater contribution from the metabolic component (Iverson et al., 1993; Iverson, 2009).

The significant differences in oxygen sources in terms of isotopic composition allowed us to determine the contribution of each water source to the body water of cetaceans, notably the respective contributions of water coming from dietary free water, water formed through metabolism and surrounding salt water. We showed that dietary free water is the main source of water for cetaceans and that cetaceans do not drink seawater. Beyond these physiological implications, this work raises some questions about the use of isotopic data from cetaceans as environmental proxies of the oxygen isotope composition of present and past oceanic water. Indeed, the phosphate oxygen isotope composition of cetacean bones and teeth (δ¹⁸O_p) is used to determine the oxygen isotope composition of present and past oceans (Yoshida and Miyazaki, 1991; Barrick et al., 1992; Amiot et al., 2008; Ciner et al., 2016), to differentiate marine and freshwater habitats of extinct and extant cetaceans (Clementz and Koch, 2001; Clementz et al., 2006) and to study the geographical distribution of cetaceans and their movements across oceans (Matthews et al., 2016, 2021). These studies are based on the strong assumption that the δ¹⁸O value of cetacean body water is equal to that of the surrounding marine water (Kohn, 1996; Thewissen et al., 1996; Clementz and Koch, 2001; Newsome et al., 2009). For the study of wild organisms for which the δ¹⁸O of dietary free water is equal to that of the surrounding salt oceanic water δ¹⁸O_SW, this working hypothesis is acceptable for cetaceans but also for organisms that drink sea water, such as sea turtles. However, our results show that precautions should be taken when organisms with a lipid-rich diet are used as this may lead to underestimation of body water δ¹⁸O values and thus underestimation of those of the surrounding oceanic water. Furthermore, wild cetaceans do not have to consume food as regularly as captive animals do (Barros, 1990; Kastelein et al., 2002). In this case, water balance is probably maintained by a higher contribution of metabolic water products following the remobilisation of stored lipids in blubber (Iverson et al., 1993; Iverson, 2009). This would suggest that in the wild, cetaceans with a varying species-dependent blubber thickness (Worthy and Edwards, 1990; Iverson, 2009; Favilla and Costa, 2020) would have body water δ¹⁸O values significantly higher than those of the surrounding ocean water. Further studies focusing on the body water δ¹⁸O values during fasting periods would help to define the importance of metabolic water production from blubber in the water balance and clarify the relationship between δ¹⁸O_{blood plasma} and δ¹⁸O_SW for wild cetaceans.

The authors thank the training teams for assistance during sampling and discussions concerning studied organisms, B. Choux, Marineland's Legal Coordinator, for administrative tasks and P. Picot, director of Marineland, for making this study possible. We also thank the Plateforme d'Ecologie Isotopique du Laboratoire d'Ecologie des Hydrosystèmes Naturels et Anthropisés (LEHNA, UMR5023, Université Claude Bernard Lyon 1, Lyon, France) for supporting the oxygen isotope measurements. N.S. also warmly thanks A. Fahlman and A. Allen for their help with cetacean respiratory physiology. Finally, we thank the Editor Kathleen Gilmour and the three anonymous referees for their helpful comments and suggestions.