Animals included in the study were stranded cetaceans in the Canary Islands between 2006 and 2010, mass stranded sperm whales in Italy in 2009, and two sea lions (Otaria byronia) from aquatic-parks of the Canary Islands submitted for necropsy to our institution. A total of 88 necropsies were performed on marine mammals belonging to 18 different species. Species were segregated into two broad groups: deep divers and non-deep divers, defining deep divers as those species known to dive deeper than 500 m for foraging (Kogia, Physeter, Ziphius, Mesoplodon, Globicephala, and Grampus; Gannier, 1998; Astruc and Beaubrun, 2005; Aguilar de Soto, 2006; Tyack et al., 2006; Watwood et al., 2006; West et al., 2009). These genera were further studied separately except for Ziphius and Mesoplodon, which were studied together as the family Ziphiidae. Twenty-nine out of 88 (33%) necropsied animals were deep divers.

Animals were identified by their stranding codes, represented by CET (“cetacean”) followed by the stranding number. Animals not stranded in the Canary Islands were identified by their investigation numbers, represented by I (“investigation”) followed by their corresponding numbers and the year when their necropsies were performed (Table 1). Cause(s) of death (defined as pathological entities) were determined by the Division of Histology and Animal Pathology of the Institute for Animal Health, University of Las Palmas de Gran Canaria (ULPGC; Unit of Cetaceans Research, 2006, 2007, 2008, 2009, 2010; Arbelo, 2007; Table 1).

At necropsy, putrefaction of the animal was evaluated using a morphological decomposition code from one to five according to Kuiken and García-Hartmann (1991) where one is the animal alive (becomes code 2 at death), code 2 is when the animal is extremely fresh (no bloating), code 3 is moderate decomposition (bloating, skin peeling but organs still intact), code 4 is advanced decomposition (major bloating, organs beyond recognition), and finally code 5 when no organs are present. Dissection, gas sampling and analysis were performed following procedures described by Bernaldo de Quirós et al. (2011). A total of 429 samples were recovered and analyzed from the studied animals, 208 (48%) of which belong to deep diving animals.

Samples containing blood were considered contaminated, since putrefaction gases from the blood will alter original gas sample composition. When no gas compound had a chromatographic signal higher than the detection limit, the sample was considered empty. In addition, if only one compound was present in slightly higher quantities than the established detection limit, the sample was considered small and not representative of its original composition. Finally samples with an atmospheric-air-like composition (around 79% nitrogen and 21% oxygen) were considered as “atmospheric-air polluted.” Thus, samples with blood, empty, or identified as small or atmospheric-air polluted were not included in the results. These samples represented 28% of the total. Most of them were emptied intestines or blood contaminated samples from the heart when only the vacutainer was applied. This problem was prevented later in the study by using an aspirometer (Bernaldo de Quirós et al., 2011).

The amount of gas present in veins and tissues was evaluated retrospectively using pictures and data in the necropsy reports. Gas amount was semi-quantified by giving a score to different vascular locations as well as to the presence of subcapsular gas (emphysema), defined as macroscopically visible gas found beneath the capsule of body organs (e.g., kidneys). Vascular locations studied for gas scoring were subcutaneous, mesenteric, and coronary veins as well as the lumbo-caudal venous plexus. We used the following score for these locations: grade 0 is no bubbles, grade I represented few bubbles and grade II represented abundant presence of bubbles. Prevalence of subcapsular gas, was evaluated with a similar grading system: grade 0 is no subcapsular gas, grade I represented the presence of subcapsular gas in one or two organs, and grade II represented wide distribution through the body organs. The summation of gas score (0–II) in the different vascular locations and tissues (subcapsular gas; n = 5), gave the total gas score for each animal that ranged from 0 to 10. However, because gas score was done retrospectively, some information was missing on occasion. Those cases on which only one localization could not be scored were marked with an asterisk (*), indicating that the total gas score might be up to two units higher according to the established gas score (e.g., 8*). If information from more than one localization was missing, the total gas score was not reported as it would not be a realistic estimate. Finally, there were some animals with advanced putrefaction where the veins could not be clearly distinguished from the rest of the tissues. Gas score was not undertaken in these cases.

The proportion of observations were analyzed for non-random associations between two categorical variables with the Fisher exact test. Hypothesis testing was considered significant when the corresponding P-value was less than 0.05. All data were analyzed using Sigma Stat software, version 3.5.

Intravascular bubbles were found in 51 out of 88 animals (58%). They were absent in 33 out of 88 (37%) and the observation was not correctly undertaken in 4 out of 88 animals (5%). Bubbles were statistically (P = 0.006) more frequently present in deep divers (76%) compared to non-deep divers (52%). Subcapsular emphysema (mostly peri-renal) was found in 57 out of 88 animals (65%). It was found in 66% of the non-deep diving animals and in 71% of the deep divers. This difference was statistically non-significant (P = 0.474).

A positive relationship between decomposition code and presence of intravascular bubbles and/or subcapsular emphysema was found (Figure 1). Intravascular bubbles were present in all the studied animals with decomposition code 4 or higher. However, intravascular bubbles had a higher prevalence (P = 0.006) in deep divers compared to non-deep divers regardless of decomposition code (Figures 2 and 3); they were present in 57% of the fresh (decomposition code 2) deep diving animals compared to 20% of non-deep divers (P = 0.001), and in 100% of deep diving animals with decomposition code 3 compared to 50% of non-deep divers within the same decomposition code (P = 0.019). In addition, subcapsular emphysema was found in 64% of the deep divers vs. 44% with decomposition code 2, and in 66% of the deep divers vs. 61% of the non-deep divers with decomposition code 3, although these differences were not statistically significant (P = 0.204 and P = 0.648 respectively).

In 40% of the fresh animals with subcapsular emphysema, intravascular bubbles were additionally present, while 61% of the fresh animals with intravascular bubbles also had subcapsular emphysema.

In freshly dead animals, intravascular bubbles were seen in 7 out of 14 of the cetaceans that were known to strand actively (by swimming ashore) but only in 3 out of 15 that were found floating ashore dead or dead in the beach with no signs of active stranding. Animals found dead but suspected to have stranded alive were not included in the study. Differences were not statistically significant (P = 0.128).

Gas composition from the intestines was highly variable with no trend with PM time. Nitrogen was found in high concentrations (higher than 60%) in 21% of the samples. However, nitrogen was found in very low concentrations in the rest of the samples. In 40% of the samples, nitrogen was not present. In most of the samples (67%), CO₂ was the major compound reaching values as high as 100%. Hydrogen was very frequently found in the intestine (in 64% of the samples). Therefore most of the samples were a mixture of high CO₂ levels with hydrogen and some, if any, nitrogen. Methane was only detected in Kogiidae and Physeteridae.

Subcapsular gas was composed of around 80% of nitrogen and 20% of CO₂ in the fresh animals studied (a Risso’s dolphin and a sea lion), except from one sample recovered from the peri-renal area of a common dolphin (CET 517). Gas composition of subcapsular samples from the fresh animals was clearly different from those of more decomposed animals, which mainly consist on CO₂ and hydrogen. Nitrogen was lower than 30% when present. Nitrogen was found in high concentrations in only one sample from the most decomposed animals. This sample also presented the highest concentration of oxygen (15% compared to a maximum of 4% in the rest of the samples; Figure 12).

Gas composition from the sinuses remained constant for longer PM time compared to gas recovered from other tissues. Nitrogen was always the major compound (58–87%) until decomposition code 4 was reached. Within these decomposition codes, CO₂ was present in concentrations ranging from 10 to 36%. After decomposition code 4, CO₂ increased devaluating nitrogen presence. There were two samples with atmospheric-air like composition (CET 512 and CET 504) (Figure 13)

Fifty-one out of 88 stranded animals presented macroscopic bubbles during the necropsy. This represents 58% of the animals that were studied indicating that the presence of gas bubbles within the cardiovascular system in stranded cetaceans is a common gross finding during necropsy.

In forensic human pathology the presence of intravascular gas in carcasses decomposing under different environmental conditions is well known and widely reported (Knight, 1996). Indeed, bubbles were found in all the studied animals with decomposition code 4 and 5. Putrefaction is a continual process of gradual decay and disorganization of organic tissues and structures after death that results in the production of liquids, simple molecules, and gases (intravascular gas and/or putrefactive emphysema; Vass et al., 2002; Lerner and Lerner, 2006). However, it is very unusual to find gas bubbles within the cardiovascular system in fresh necropsied domestic animals or humans (King et al., 1989; Knight, 1996). Therefore, the most interesting data are those recorded from fresh animals. Our results demonstrated that it is not uncommon to find intravascular bubbles in fresh cetaceans. Thirty-three percent of our fresh animals presented intravascular bubbles to some extent, but only two animals presented high gas score. Recent studies have demonstrated that large amounts of intravascular bubbles found within a few hours PM are not due to putrefaction processes (Bernaldo de Quirós, 2011). Indeed, these animals were the only ones diagnosed with gas embolism according to the pathological studies carried out by the ULPGC. Therefore amount of intravascular bubbles is more important than the mere presence of bubbles from a pathological point of view in stranded cetaceans, and the gas score is also an efficient tool to distinguish between gas embolism and putrefaction gases in stranded cetaceans. Although we used a simple gas scoring technique with a scale from 0 to 10 because the study was done retrospectively, we encourage using a larger gas score scale (0–27) described by Bernaldo de Quirós (2011). This gas score has a grading scoring of 0–6 for the different vascular localizations (n = 4) and a 0–3 grading scoring for subcapsular gas (0–3). Therefore, total gas score in each animal ranged from 0 to 27. A larger scale highlights better the possible differences in gas score along decomposition codes.

Subcapsular emphysema in the peri-renal area has been described in 22 of 22 alive stranded animals using B-mode ultrasound (Dennison et al., 2012). In our study, we found it in 20 out of 39 fresh animals. Although the prevalence is smaller in our study, it still suggests that subcapsular emphysema is a common finding in dead stranded cetaceans. However, we could not find a clear relationship between subcapsular emphysema and intravascular bubbles. There is a 60% chance that a fresh animal with intravascular bubbles presents subcapsular emphysema, but 60% of the fresh animals with subcapsular emphysema showed no intravascular bubbles. Therefore, intravascular bubbles can occur without simultaneous subcapsular emphysema and vice versa.

An interesting result from our data was that intravascular bubbles were more frequently found (P = 0.006) and in higher quantities in deep divers than in non-deep divers regardless of decomposition code. The higher prevalence of intravascular bubbles related to the diving behavior of the species, together with the fact that finding bubbles in necropsied domestic animals or humans is very unusual and mostly linked to iatrogenic air embolism or diving fatalities (Knight, 1996; Muth and Shank, 2000), suggest that the most parsimonious explanation for the presence of small quantities of intravascular bubbles in fresh stranded cetaceans would be diving physiology related (Tikuisis and Gerth, 2003). Thus our results further suggest that deep divers might be at a higher risk from decompression.

Intravascular bubbles and subcapsular emphysema of fresh marine mammals (decomposition code 2) showed high concentrations of nitrogen (>70%) and values of CO₂ around 20%. Similar results were obtained in animal models exposed to air embolism or compression and decompression using the same methodology (Bernaldo de Quirós, 2011). Furthermore, these values are in accordance to what has been reported in cases of air embolism and/or DCS in humans and laboratory animals (Bert, 1878; Armstrong, 1939; Pierucci and Gherson, 1968; Smith-Sivertsen, 1976; Ishiyama, 1983; Bajanowski et al., 1998; Bernaldo de Quirós, 2011). In addition, gas from pterygoideal sinuses showed high concentrations of nitrogen in animals with decomposition codes 2 and 3. This was an additional finding that needs further investigation (Figure 13).

Hydrogen, which is a putrefaction indicator (Pierucci and Gherson, 1969), was mostly found after decomposition code 3 in gas collected from the different localizations (intravascularly, subcapsularly, or in the pterygoideal sinuses). Animals with incipient autolysis (decomposition code 3), had gas samples that varied from typical gas embolism composition to putrefaction gases (Pierucci and Gherson, 1968, 1969). Samples from animals with decomposition code 4 and 5 always presented high CO₂ concentrations together with hydrogen in most of the cases and low concentration of nitrogen (mostly lower than 40%) if present. In a few decomposed samples, small quantities of oxygen were also detected. This gas composition was very similar to the gas composition of the subcapsular emphysema found in very decomposed animals and completely different from gas composition of intravascular bubbles and subcapsular emphysema in fresh animals.

These data show that detection of hydrogen and/or very high levels of CO₂ are “signal” gases indicative of putrefaction, while nitrogen decreases progressively until disappearing (not detected chromatographically) in carcasses with the highest decomposition code (code 5). Similar results have been described in humans and laboratory animals (Pierucci and Gherson, 1968, 1969; Keil et al., 1980; Bajanowski et al., 1998). In this sense, we can state that hydrogen in stranded cetaceans is also a key putrefaction marker with one exception that will be discussed later. In some cases, we have found one sample’s composition to be clearly different from the rest of the samples, presenting significantly higher nitrogen and oxygen content. Thus, we have considered as a first option, a likely contamination of atmospheric-air during sampling (see as an example the sample from CET 546 in Figure 12). The exception previously mentioned, was a Steno bredanensis (CET 473) with decomposition code 2 that presented hydrogen in the mesenteric veins. Here, it is important to remember that these are stranded animals, and in most of the cases sick animals. Therefore these gas results should be considered together with the individual diagnostic analysis (including pathology, microbiology, toxicology, etc.). Speculatively, we have considered acute gastrointestinal disease as a primary source of hydrogen or due to overlapping with other pathogenic processes, although further research is needed in order to confirm this hypothesis.

One of the most interesting and unexpected results was CO₂ concentrations of around 40–50% in the animal with the second highest gas score for fresh animals and diagnosed with gas embolism by the ULPGC (CET 483). This gas composition is not in accordance with traditional DCS theories or some of the empirical values reported for DCS (Bert, 1878; Smith-Sivertsen, 1976; Ishiyama, 1983; Tikuisis and Gerth, 2003). However Armstrong (1939) reported values of 30% of CO₂ in decompressed goats, and Bernaldo de Quirós (2011) has recently reported very similar CO₂ concentrations to our results in rabbits.

In addition there are several empirical and physical models suggesting an important role of CO₂ in bubble formation: (i) there is a higher prevalence of elevated CO₂ in bends following dives using compressed air (Behnke, 1951), (ii) a statistically significant increase in DCS risk in rats breathing elevated levels of CO₂ in either He-O₂, or N₂-O₂ mixtures during the hyperbaric exposure has been reported (Berghage et al., 1978), (iii) bubbles moving vertically through different water layers alternately saturated with air or CO₂, increase in size in the CO₂ saturated water and decreasing in the air-saturated layer (Harvey, 1945), (iv) bubbles formed with less mechanical agitation and grew at a faster rate in decompressed tubes filled with water saturated of CO₂ rather than nitrogen (Harris et al., 1945). All these findings, suggest that CO₂ might play an important role in bubble formation due to the high diffusivity of this gas. If this is the case, prolonged dives where CO₂ will build up might pose an additional risk to animals with nitrogen saturation.

Theoretical studies have predicted BWs end-dive nitrogen levels that would cause a significant proportion of DCS cases in land mammals (Houser et al., 2001; Zimmer and Tyack, 2007; Hooker et al., 2009). In addition BWs have both the deepest (3120 m) and longest (137 min) dives ever recorded from an air-breathing mammal (Schorr et al., 2011). Furthermore, BWs exposed to mid-frequency sonar playback have reacted with unusual longer and slower ascent dives (Tyack et al., 2011). All BWs included in our study presented a minimum gas score of 5, this was not found in the other studied sub groups. Although more fresh animals should be studied in order to make conclusions, our results suggest that BWs might be the most sensitive species to bubble formation. Future theoretical studies should consider CO₂ accumulation in addition to nitrogen saturation, and its possible role in bubble formation.

We have shown that gas analysis may be a valid a technique to differentiate between gas embolism and putrefaction gases in stranded cetaceans. Therefore we encourage gas analysis. To try to avoid putrefaction-masking gases, necropsy, and gas sampling must be performed as soon as possible, before 24 h PM is recommended but preferably within 12 h PM. However, if gas analyses are not possible due to logistics, we strongly recommend doing the gas scoring, which is of no additional cost.

In summary, intravascular bubbles and subcapsular emphysema are a common finding during necropsies of stranded marine mammals, although they are present in low quantities (low gas score) and with a gas composition of around 70–80% of nitrogen and 20–30% of CO₂. They are more frequently found (P = 0.006) and in higher quantities in deep divers vs. non-deep divers. They have higher prevalence in active stranded animals, although they are also present in passive stranded animals. Based on gas composition and in the highest prevalence of bubbles in deep divers, the most likely source of these bubbles is decompression-related.

In air-breathing animals, nitrogen can build up in tissues and formed bubbles. Although they only have a finite amount of air in their lungs available for nitrogen diffusion, PN₂ in the alveoli increases with compression of the thorax as the pressure increases. This results in a net exchange of nitrogen across the alveolar membrane and it is either taken up or removed by the blood and tissues depending on the partial pressure gradient. During ascent, nitrogen diffusion from tissues to blood and alveolus is slower, and if the surface interval is too short, not all of the nitrogen that has been taken up is removed. This may result in nitrogen build up in tissues during repeated dives (Ferrigno and Lundgren, 2003). When the sum of the dissolved gas tensions (oxygen, CO₂, nitrogen, helium) and water vapor exceeds the local absolute pressure, a state known as supersaturation, bubbles may form due to gas phase separation (Hamilton and Thalmann, 2003; Vann et al., 2011). Bubbles can form without negatively impacting a diving animal, but bubbles can have mechanical, embolic, and biochemical manifestations ranging from trivial to fatal (Vann et al., 2011). They are generally considered a pivotal event in the occurrence of DCS (Francis and Simon, 2003).

Decompression bubble formation may still occur even after death if the animal dies at depth and is later depressurized, a phenomenon known as post mortem off gassing (Brown et al., 1978; Lawrence, 1997; Cole et al., 2006; Lawrence and Cooke, 2006; Wheen and Williams, 2009). Post mortem off gassing has been proposed for explaining the existence of bubbles in marine mammals trapped in fishing nets (Moore et al., 2009). These animals died at depth and were later hauled out. Our studied animals are stranded marine mammals, which presumably have not died at depth. Thus, post mortem off gassing is ruled out as a plausible mechanism. The stated absence of bubbles in stranded animals in the Moore et al. study probably reflects a relative, rather than absolute absence, with bubble quantity being far more obvious in the animals drowned at depth that off-gassed post mortem. These authors probably failed to record the low prevalence of bubbles commonly seen in stranded animals in the current study (Michael Moore, personal communication). It is interesting that the Moore et al. findings likely reflect a routine supersaturation state in foraging, diving marine mammals. Analysis of the gas bubble composition of drowned bycatch would test this hypothesis.

The most likely explanation for the observed gas composition in the freshly dead animals is that bubbles were in vivo or peri mortem formed from supersaturated tissues by physiological off gassing. Indeed, a recent study has shown that intravascular bubbles and peri-renal subcapsular emphysema occur in live marine mammals and without clinical consequences (Dennison et al., 2012). The stranding itself has been proposed as a causal mechanism for bubble formation in marine mammals (Houser et al., 2010; Dennison et al., 2012). According to this hypothesis, a fast stranding will result in the inability to recompress resulting in problematic supersaturation of the tissues. In addition, as a consequence of the stranding, there would be tissues of reduced perfusion and blood pooling that would continue to off-gas, but because the vascular flow to the lung is supposed to be compromised, nitrogen elimination would not be as effective and autochthonous bubble formation would increase (Houser et al., 2010).

Another plausible stranding mechanism related would be tribonucleation. Tribonucleation occurs when two surfaces in intimate contact but separated by a viscous liquid film are suddenly separated in an abrupt way. Then, the viscosity will avoid sudden filling and cavities, which will be filled by diffusing gases, can occur (Banks and Mill, 1953; Hayward, 1967; Campbell, 1968; Ikels, 1970; Blatteau et al., 2006). This phenomenon has been proposed to happen with movements, exercise (McDonough and Hemmingsen, 1984a,b, 1985a,b), within the joints (Fick, 1911), or in the heart valves (Hennessy, 1989). The agonal actions of the beached animals have been proposed to promote bubble formation (Houser et al., 2010).

We have found intravascular bubbles in 3 out of 15 of the animals that stranded passively, where reduced perfusion and tribonucleation are not expected. Regardless of the low number of passive strandings studied, our results support the idea that bubbles can occur independently to the stranding, as reported in three BWs (one fresh, one partially autolytic, and an autolytic animal) which were recovered floating in the atypical mass stranding event of the Canary Islands in 2002 (Fernandez et al., 2005). However, since we have found a higher prevalence of bubbles in actively stranded animals (although the difference was non-statistically significant), we suggest that stranding is not a necessary factor but a contributing factor in bubble formation.

More recently, Houser et al. (2010) and Dennison et al. (2012) hypothesized that the abnormal behavior of moribund cetaceans that might spend a period of time at sea with reduced depth and repetitiveness, may allow them to wash out nitrogen prior to stranding. Few (Dennison et al., 2012) or no (Houser et al., 2010) empirical data were presented to support this suggestion. Our study animals are mostly single stranded specimens, affected by different pathologies (Table 1). According to the mentioned hypothesis, it is presumed that our case studies have washed out their nitrogen prior to stranding. Our results do not support this hypothesis; intravascular bubbles were found in 57% of the fresh deep diving animals and in 20% of the non-deep diving animals, and they were mainly composed of nitrogen (70–80%). Dennison et al. (2012) further hypothesize that the prevalence of bubbles will be less in most single stranded than mass stranded animals, if they were diving less over an extended period. Since our study cases are mostly single stranded animals, we could not test this hypothesis. Further studies are needed to confirm whether mass stranded animals have higher prevalence of bubbles compared to single–stranded animals.

If bubbles are not a consequence of the stranding, and since no embolic related lesions were found in the pathological study, it is reasonable to consider as a hypothesis that this small amount of bubbles composed of nitrogen found in fresh stranded animals, were silent bubbles. The existence of potentially asymptomatically bubbles in marine mammals has been reported before (Dennison et al., 2012). However, we studied dead animals where symptoms could not be evaluated, therefore our results are indicative but not conclusive by itself of the existence of silent bubbles in marine mammals. A recent review of evidences for gas bubble incidence in marine mammals concluded that they may deal with bubbles on a more regular basis than previously thought (Hooker et al., 2012) and suggested that our view of marine mammal adaptations should therefore change from one of simply minimizing nitrogen loading to one of management of the nitrogen load. Our study further supports the hypothesis that decompression-induced bubbles are relatively common but that marine mammals have unknown adaptations allowing them to tolerate these under natural conditions. It is important to remember that our case studies have only few bubbles and died from numerous pathological reasons different from gas embolism. Only two animals, showing a lot of bubbles widely dispersed and with no other pathological signs, were diagnosed with massive gas embolism.

Further research is needed in order to clarify questions that remain unanswered from this study. The clinical impact of subcapsular emphysema remains unclear from this study since gas composition indicates a plausible physiological decompression-related off gassing origin, but no clear relationship with intravascular bubbles was found. More studies are needed in order to better understand the formation of this gas and the plausible impact that might have on the health status of the animals. Comparison between single-active, single-passive, mass stranded, and by caught animals, might contribute to a better understanding of the causative factor of intravascular bubble formation as well as nitrogen supersaturation levels on these animals, and different species sensibilities (like deep divers vs. non-deep divers). These will provide new data that should be considered in future diving physiology models and studies.

In conclusion, this study has demonstrated that bubbles are a common finding in stranded cetaceans. The amount and composition of these bubbles in fresh animals suggests that these bubbles have formed from nitrogen-supersaturated tissues, most likely formed in vivo. This study has also demonstrated that these bubbles were more frequently found in deep divers indicating a higher risk of decompression to these species. Further research is needed in order to determine if there are any species-specific sensibilities to bubble formation as preliminary results point out, as well as the implications that CO₂ accumulation along a dive might have for these animals from a diving physiology perspective. Finally, monitoring live non-deep diving and deep diving animals with an ultrasound might help to confirm the hypothesis deduced from our results.