Distribution, composition and functions of gelatinous tissues in deep-sea fishes

Many deep-sea fishes have a gelatinous layer, or subdermal extracellular matrix, below the skin or around the spine. We document the distribution of gelatinous tissues across fish families (approx. 200 species in ten orders), then review and investigate their composition and function. Gelatinous tissues from nine species were analysed for water content (96.53 ± 1.78% s.d.), ionic composition, osmolality, protein (0.39 ± 0.23%), lipid (0.69 ± 0.56%) and carbohydrate (0.61 ± 0.28%). Results suggest that gelatinous tissues are mostly extracellular fluid, which may allow animals to grow inexpensively. Further, almost all gelatinous tissues floated in cold seawater, thus their lower density than seawater may contribute to buoyancy in some species. We also propose a new hypothesis: gelatinous tissues, which are inexpensive to grow, may sometimes be a method to increase swimming efficiency by fairing the transition from trunk to tail. Such a layer is particularly prominent in hadal snailfishes (Liparidae); therefore, a robotic snailfish model was designed and constructed to analyse the influence of gelatinous tissues on locomotory performance. The model swam faster with a watery layer, representing gelatinous tissue, around the tail than without. Results suggest that the tissues may, in addition to providing buoyancy and low-cost growth, aid deep-sea fish locomotion.


MEG, 0000-0002-2026-5829
Many deep-sea fishes have a gelatinous layer, or subdermal extracellular matrix, below the skin or around the spine. We document the distribution of gelatinous tissues across fish families (approx. 200 species in ten orders), then review and investigate their composition and function. Gelatinous tissues from nine species were analysed for water content (96.53 ± 1.78% s.d.), ionic composition, osmolality, protein (0.39 ± 0.23%), lipid (0.69 ± 0.56%) and carbohydrate (0.61 ± 0.28%). Results suggest that gelatinous tissues are mostly extracellular fluid, which may allow animals to grow inexpensively. Further, almost all gelatinous tissues floated in cold seawater, thus their lower density than seawater may contribute to buoyancy in some species. We also propose a new hypothesis: gelatinous tissues, which are inexpensive to grow, may sometimes be a method to increase swimming efficiency by fairing the transition from trunk to tail. Such a layer is particularly prominent in hadal snailfishes (Liparidae); therefore, a robotic snailfish model was designed and constructed to analyse the influence of gelatinous tissues on locomotory performance. The model swam faster with a watery layer, representing gelatinous tissue, around the tail than without. Results suggest that the tissues may, in addition to providing buoyancy and low-cost growth, aid deep-sea fish locomotion.  and muscle or between muscle bundles (figure 1). Fishes in the superorder Elopomorpha (Anguilliformes, Albuliformes, Elopiformes and Saccopharyngiformes) have larvae called leptocephali in which most of the body consists of an acellular gelatinous matrix that provides structural support in the absence of a vertebral column and transparency for camouflage (e.g. [1,2]). The first known scientific record of these tissues in a fully adult fish comes from the Challenger Report description of the gelatinous blind cusk eel Aphyonus gelatinosus, in which the 'anterior half of the skin forms a large loose bag which, during life, is probably filled and distended with mucus' [3]. Gelatinous tissue is even a defining character in the genus Careproctus of the family Liparidae (snailfish), which 'best illustrates the production of pseudotissue which envelops the body and fins just beneath the skin' [4]. The tissues are sometimes referred to as the subdermal extracellular matrix, or SECM (e.g. [5,6]). More recently, such tissues have been found in hadal snailfishes in the Kermadec and Mariana trenches. In a freshly collected fish, the layer of clear gelatinous tissue is prominent (figure 1a), although as the skin is lacerated, this tissue leaks out and melts away. It is largely concentrated just behind the abdominal cavity, with a thin layer around the posterior third of the body. Although these gelatinous tissues have been noted in several deep-living adult species and can compose up to a third of the mass of a fish [5], they have not been compared across families and 6 cm) were measured. Water content. Gelatinous tissues were dried at 60°C for 3 days to ensure that all water evaporated and remaining dry mass was compared to original wet tissues mass. Osmotic pressure. A vapour pressure osmometer, Wescor 5500, was used in the laboratory for most species, and at sea for Notoliparis kermadecensis, to determine sample osmolality. Samples were homogenized with a small pestle in a microfuge tube, then centrifuged at 2000 × g for 30 min at 4°C. Ten microlitres of the resulting supernatant was measured with an osmometer. The 290 and 1000 mmol kg −1 standards were checked periodically to confirm accurate calibration. Sample preparation. A section of frozen gelatinous tissues, clear of white muscle, was cut and weighed to obtain about 0.1 g, with a precision of 0.0001 g. The section was ground in 7% perchloric acid (PCA) or 70% ethanol, added at nine times the tissues mass, to precipitate proteins. The sample was refrigerated overnight, then centrifuged for 20 min at 15 500 × g at 4°C. The supernatant, transferred to a new tube, was used for inorganic ion and organic osmolyte analyses, while the pellet was used for protein analysis. When ethanol was used to homogenize tissues, the supernatant was evaporated and the remaining powder dissolved in distilled water. The supernatants in PCA were titrated with 2 M KOH to pH 6.5-7.5. The resulting precipitate was centrifuged and the supernatant removed to a new tube. The PCA method was not used for ion analysis because of the required addition of potassium. Protein. Protein content was determined with the bicinchoninic acid protein assay [22]. Bovine serum albumin was used as a standard. Lipids. Lipid contents were analysed using the Bligh & Dyer [23] extraction and colorimetric determination of content with the sulfuric acid charring method of Marsh & Weinstein [24] with triolein as a standard. Carbohydrates. Carbohydrate analysis was conducted using phenol and sulfuric acid [25], with D-glucose as a standard, measured in a spectrophotometer (Beckman Coulter DU 730) at 480 nm. Ions. Sodium and potassium contents were analysed by atomic absorption (PerkinElmer AAnalyst 400) in 10 µl aliquots of the PCA homogenates dissolved in 10 ml of purified water. All results are presented as average ± standard deviation.

Taxonomic distribution
Records of gelatinous tissues in fishes were collected in an extensive literature search. Recent findings from coastal to hadal surveys are also presented. Anecdotally, these tissues were thought to be more common in deeper-living fishes. To test this, common depth ranges of fishes with gelatinous tissues were taken from FishBase [26]. Care was taken to avoid records that were obviously spurious or outlying, for example, several thousand metres out of all other capture and sighting records. The effects of phylogenetic relationships can confound interpretation of this type of analysis, as closely related species become a kind of pseudoreplicate [27]. To account for this potential error, and to clarify the distribution of gelatinous tissues, we compared depth trends within clades. Statistical analyses were conducted in the programming platform R [28]. Generalized linear models (GLM) using minimum and maximum depths and the median of each depth range were fitted using a Gaussian error distribution. Models were selected through optimization of Akaike information criteria.

Locomotor effects
Few studies have investigated locomotion in deep-sea fishes (e.g. [29][30][31][32]), largely due to the difficulty of direct experimentation. To test the effect of body shape change with gelatinous tissues, a robotic model was designed after the Kermadec Trench snailfish, N. kermadecensis, a good example of a neutrally buoyant species with large amounts of gelatinous tissues. This technique has become a valuable tool used to investigate swimming biomechanics in a number of shallow-living fishes (e.g. [33][34][35]) and is well-suited to deep-sea species that cannot easily be brought into a laboratory setting. The plastic (polylactic acid) body and fins were three-dimensional printed (ORION HB #58744) based on a model constructed from a photogrammetry recreation of freshly captured specimens collected on the HADES (HADal Ecosystems Studies) Cruise in April and May of 2014 (Model: MeshMixer, Slicing: Cura, 3D Printing: Repetier Host). The free-swimming, neutrally buoyant robotic model was larger (40 cm SL) than the actual hadal snailfish (known maximum 29 cm SL) due to design constraints. The model motion program was controlled by an on-board Arduino Nano microcontroller. Tail-beat frequency (0.5 Hz) was chosen to match that found through video analysis of the hadal snailfish, Pseudoliparis belyaevi, filmed in situ in the Japan Trench (described in [36]). The robot was powered by a 9 V battery with constant cycleaveraged power and swam using a servomotor connected to two piano wires that oscillated the tail region back and forth (figure 2). A silicone rubber mould was cast to simulate the posterior skeleton and musculature of the fish. Water between the silicon tail analogue and outer skin represented the gelatinous tissues, to isolate the shape effect from changes due to tail stiffness. The model was designed to test the  locomotor effects of gelatinous tissues that are directly below the skin, outside of the muscle tissue, such as in the hadal snailfish. As discussed, this positioning is not consistent across taxa and the locomotor effects may vary accordingly. In some species, such as the cusk eel Spectrunculus grandis, it is unlikely that the gelatinous tissue flows freely as water in our model would. However, in the liparids, morphological analyses suggest that gelatinous layers are displaced during movement [10,11]. This is also suggested by video of hadal snailfishes swimming in situ, which show the gelatinous tissues rippling under the skin, making water below the skin, rather than gelatine, an appropriate analogue. Two approximately 10 s swim trials for the submerged, neutrally buoyant robot were conducted with both empty and full tail 'skin'. Swim trials were filmed from above as the robot swam in a 1 m diameter tank, and body lengths per second and tail-beat amplitude were compared between trials (with the same tail-beat frequency and power) using ImageJ [37].

Taxonomic distribution
Fish species with gelatinous tissues were found in ten orders, thirteen families and approximately 200 species, presented in table 2. References to 'gelatinous tissues' or 'subdermal extracellular matrix' were included in these results. Fishes in the family Aphyonidae (recently absorbed into the Bythitidae; [42]) are described, for example: 'skin loose, transparent and gelatinous' [44]. Images of freshly caught fish confirm that this refers to the subdermal gelatinous tissues. Other occurrences of gelatinous tissues have been seen and verified by the authors in recent captures. We note that the gelatinous tissues are present in many, but not all, species of the snailfish genus Paraliparis. Additional species of the genus Lycodapus may contain gelatinous tissues as well, though this has not been confirmed [7]. Depth ranges for species with gelatinous tissues are presented in table 2. Median depths of occurrence ranged from approximately 300 to 7400 m (mean approximately 1800 m). Most species with records of gelatinous tissues typically live around or below 1000 m depth and include both benthic and pelagic species. GLM showed fishes with gelatinous tissues to have significantly deeper minimum, median and maximum depths (t = 2.40, p < 0.05; t = 3.01, p < 0.01; t = 2.95, p < 0.01; 117 d.f.) across all species, a finding confirmed by a non-parametric Kruskal-Wallis rank sum test (median and maximum depths, p ≤ 0.01, p = 0.001) across all species. This was also a significant trend within orders (e.g. Gadiformes, minimum: t = 6.70, p < 0.005, median: t = 4.75, p < 0.005, maximum: t = 3.20, p < 0.01, 27 d.f.; Pleuronectiformes, median: t = 3.0, p < 0.01, maximum: t = 3.1, p < 0.01, 16 d.f.). Species with gelatinous tissues were present across multiple clades, and represent the deeper-living species within clades (figure 3).

Locomotor effects
Gelatinous tissues change the body shape of the hadal liparid, as illustrated in figure 2. Intuitively, this changes the drag profile around the animal. In all swim trials, the robotic model performed significantly better with the gelatinous tissue analogue (0.074 ± 0.007 body lengths per second) than without (0.022 ± 0.007) with constant cycle-averaged power provided at a constant tail-beat frequency of 0.5 Hz (Welch two-sample t-test, p = 0.019). Tail-beat amplitude was 16.1 ± 0.3% of body length and did not vary significantly between treatments (p > 0.05). Films of swim trials are provided in the electronic supplementary material. Additional tests were conducted in the harbour to ensure that wall effects were not confounding swim trials, yielding similar results (data not shown).      *Those species formerly classified as family Aphyonidae [42]. **Pelagic fish, maximum depth greatly overestimated [36]. Note that some, but not all species of the genus Paraliparis have gelatinous tissues. Depth ranges presented in metres.

Distribution and composition
Proximate chemical analysis of gelatinous tissues in nine benthic and benthopelagic species showed high water content and low protein, lipid and carbohydrate content in comparison to white muscle (86.3 ± 2.7% for seven species with gelatinous tissues; [21]). Our average for water content (96.5%) is consistent with previous studies of gelatinous tissues, which found 93.3% water in C. lumpus [13], 96% in Bathylagus pacificus [14] and 97% in P. devriesi [5]. Osmolality increases with depth, in part due to higher extracellular sodium and in part because of organic osmolytes (especially trimethylamine oxide) that increase with depth to combat the negative effects of high hydrostatic pressure [64,72]. In concert with We found support for the hypothesis that gelatinous tissues in fishes are a characteristically deepsea phenomenon. Phylogenetic relationships were a potential concern; especially as there are certain genera where gelatinous tissues are more common-i.e. Aphyonus, Careproctus, Paraliparis. The method of Felsenstein [27] for investigating trends without confounding influence of phylogeny is designed for use with continuous variables, but has been met with criticism for categorical variables (e.g. [73]). Considering these concerns, we investigated depth trends within clades. The results hold true within the different orders tested: gelatinous tissues appear more often in deeper-living species.
Records of gelatinous tissues were found across multiple orders and families (table 2). In several recent phylogenetic hypotheses of major groups within the Actinopterygii based on multi-locus molecular datasets, gelatinous tissues are present in species across many clades, from basal to highly derived [74,75]. The fact that the gelatinous tissues are present across ten orders suggests the potential for the independent development of SECM tissue in multiple deep-water groups. It is possible that these tissues have evolved from different origins, given their varied locations in different species-such as directly under the skin or closer to the spine (figure 1).
Although we were thorough in our literature searches and covered a broad depth range in our surveys, it is likely that the list presented in table 2 is not exhaustive. Often, the tissue has leaked away shortly after capture or during preservation, and it is not always recorded in taxonomic descriptions or it is regarded as unimportant. This study reveals how common and multi-functional gelatinous tissues may be, and we suggest that future studies should note its presence.

Gelatinous tissues as a buoyancy mechanism
In our investigation, most gelatinous tissues did float in shipboard tests, suggesting that buoyancy is indeed a main function of these tissues, in agreement with most previous findings [13,14]. The one exception was the deepest fish tested, N. kermadecensis (hadal snailfish), which also had a significantly higher potassium content and lower per cent water than other species (table 2), indicating more intracellular components than in other species. These buoyancy and composition results suggest that the gelatinous tissue is not positively buoyant in that species. It is possible that testing at atmospheric pressure may have biased these results because these fish were collected from considerably greater depths than the other species. Observations of the swimming behaviour of these fish in situ suggest that the entire fish is slightly negatively buoyant, settling to the seabed when active swimming ceases. This swimming behaviour has been observed in multiple hadal trench liparids (N. kermadecensis, Pseudoliparis amblystomopsis, P. swirei). These fishes do not have swim bladders, and the gelatinous tissues, even if not positively buoyant, would have lower density than most other tissues (e.g. bone, muscle), so may help reduce overall body density and thus rate of sinking, as previously suggested for Chauliodus sloani, a pelagic viperfish species that also has gelatinous tissue that is not positively buoyant [12]. Additionally, the gelatinous SECM is often found in fishes with aglomerular kidneys and lacking gas bladders, such as Ateleopus japonicus, which may serve to reduce whole-animal density [6]. It is possible that aglomerular kidneys result in increased water retention and thus the accumulation of gelatinous tissues. However, the correlation between the aglomerular kidney and gelatinous tissues remains to be fully explored.
As noted earlier, previous work on mesopelagic fishes revealed lower ion concentrations in gelatinous tissue compared to blood [14]. Our osmolality values hint at a similar pattern for gelatinous tissue because they are well below osmolalities of blood and muscle of other fish species from comparable depths. Muscle osmolalities at 1000 m have been reported at approximately 400 (cf. gelatinous tissues at 312-377), at 2000 m approximately 490-500 (cf. gelatinous tissues at 355-385), at 3000 m approximately 590-600 (cf. gelatinous tissues at 467) and at 7000 m approximately 990-1000 (cf. gelatinous tissues at 945, [64,72]). It should be noted that there would be an energetic cost associated with actively maintaining the ionic gradient needed to produce greater buoyancy (e.g. ion-regulating chloride cells in gelatinous tissues of leptocephali, [76]).
While most of the gelatinous tissues tested could aid fish buoyancy, our results suggest that this might not be the only function. Importantly, gelatinous tissues are found in some species with gas-filled swim bladders (e.g. family Ophidiidae), indicating that buoyancy may not always be their primary adaptive role. Furthermore, gelatinous tissues are found in benthic flatfishes (order Pleuronectiformes; e.g. E. bathybius and M. pacificus), which would have less evolutionary pressure to develop positively buoyant tissues, as they spend more time resting on the seafloor than swimming. In several species, gelatinous tissues are concentrated ventrally, an unlikely position to provide positive buoyancy. Some bathypelagic fishes are within 0.5 and 1.2% (Gonostoma elongatum and Xenodermichthys copei) of neutral buoyancy without swim bladders or gelatinous layers, through reduced ossification and watery muscle tissue [12] and some benthopelagic fishes lacking gas bladders also have watery muscle to aid in achieving neutral buoyancy [77]. In the hadal liparids, near neutral buoyancy seems also maintained by other means, including a large fatty liver and reduced bone ossification.

Locomotor effects
Watery gelatinous tissues may be used to increase body size at lower production cost than muscle tissue, a strategy noted earlier that has been proposed for some deep-sea invertebrates (e.g. [8]) and some larval fishes (e.g. [7]). Gelatinous tissues may be an example of neoteny, where deep-sea species may have evolved to retain this low-growth-cost paedomorphic character into adulthood in a food-poor environment. In the hadal snailfish, N. kermadecensis, there does seem to be more gelatinous tissue in larger individuals, although the exact amount of tissue could not be quantified due to damage. Some deep-sea fishes, including two flatfish in this study, also have very watery muscle tissue, which further reduce growth costs, though, in this case, by sacrificing locomotory capacity [77]. The gelatinous tissues are the extreme end of this continuum. They serve as low-growth-cost bulk tissues, allowing the animal to grow large, reducing the likelihood of predation, without alteration to locomotory muscle.
Material properties of the actual gelatinous tissues should also be analysed under deep-sea, especially hadal, temperatures and pressures, as even small changes in body shape and stiffness can make a large difference in swimming performance (e.g. [33,78]). Gelatinous tissues (which melt at room temperature) are probably stiffer at hadal conditions of cold temperatures and high pressures, and could provide an even better paddle for forward propulsion. There may be an additional cost of transport to the stiffer tail, though this may improve acceleration [79]. Gelatinous tissues may change stiffness and shape with movement, as seen in other models of undulatory swimming (e.g. [80]). While further exploration of this hypothesis is needed, the improved performance of the robotic model with a gelatinous tissue analogue suggests that the presence of a subdermal gelatinous layer could enhance swimming performance. The chemical composition of the gelatinous tissues shows that they are inexpensive to form, but the benefit to structure and locomotory capacity could be significant, accounting for some of its prevalence across many deep-sea genera. However, this use of gelatinous tissue cannot be universal; when the gelatinous tissue occurs within the main musculature of a fish (e.g. figure 1b) no locomotory advantage is likely.

Conclusion
Our results suggest that gelatinous tissues are widely used by fishes, principally in deep-sea species, serving multifunctional roles both for individual fish and across families. Gelatinous tissues, which are primarily extracellular fluid, are present in fishes of very different life histories and behaviours, from the flatfish, M. pacificus, to the hadal snailfish, N. kermadecensis. The varied location of gelatinous tissues, which are present in the trunk of some eelpouts (Zoarcidae), the snout of Ateleopus japonicas (Ateleopodidae) and directly below the skin in many snailfishes (Liparidae), also calls attention to potential functional complexity. Through chemical analyses and float tests, we found support for the use of gelatinous tissues in aiding fish buoyancy. Robotic modelling supported the hypothesis that these tissues may also provide a functional role in reducing drag during swimming. Overall, gelatinous tissues seem to be a low-density, low-production-cost method to increase body size and alter body shape and size, with adaptive advantages for both swimming efficiency and buoyancy with varied functions among species.
Ethics. Sample collection followed guidelines on animal welfare in research. We followed the Institutional Animal Care and Use Committee (IACUC) guidelines on animal welfare in research.
Data accessibility. The datasets supporting this article have been uploaded as part of the electronic supplementary material.