Cytological analysis of integumentary and muscular adaptations in three sand‐dwelling marine teleosts, Ammodytes tobianus (Ammodytidae), Gorgasia preclara (Congridae) and Heteroconger hassi (Congridae) (Teleostei; Actinopterygii)

Abstract Sandy bottoms are a ubiquitous environment found from sea bottoms to intertidal and freshwater zones. They are inhabited by many invertebrates and vertebrates which have developed morphological and physiological adaptations to sustain life under these particular conditions. Sandy habitats exhibit three potential constraints: abrasion, hypoxia and mechanical resistance. Here, three teleost species living in sandy environments were investigated: Ammodytes tobianus (Ammodytidae), Gorgasia preclara and Heteroconger hassi (Congridae). These teleost fishes were studied for their integument and muscular systems, which are potentially subject to sand abrasion and hypoxia, respectively. Based on histochemistry and transmission electron microscopy, we found the complex mucus system of G. preclara and H. hassi consists of two types of goblet cells and one type of sacciform cell. The secretions of both species are made of complex polysaccharides. In contrast, the scaly integument of A. tobianus has only a few goblet cells and no sacciform cells. We also highlighted, by immunohistochemistry, that the epidermal cell proliferation was much higher for this latter species, potentially resulting from the high rate of sand abrasion when A. tobianus buries itself quickly in the substrate. For all species, the major muscle fibre type was revealed by histoenzymology and corresponds to fast glycolytic fibres followed by intermediate fibres with slow fibres in the lowest proportion. Ammodytes tobianus possesses the highest fast fibre proportion (about 87% for A. tobianus and 75–78% for both garden eels). Our results provide new insights into the previously poorly studied teleost species, such as G. preclara, and allow us to highlight the complex skin histology of both garden eel species. Furthermore, the previously unknown muscle typing of these three species was determined.


| INTRODUCTION
Sand is defined as a granular material, finer than gravels and coarser than silt, which is composed of particles between 0.05 and 2 mm in diameter. Sandy substrates are present in many terrestrial and aquatic biotopes covering a large part of our planet. For example, 20% of deserts are made of sand (Seely, 1991). In places free of ice, sandy shores (sand or a mixture of sand and rocks) represent 75% of the world shores (Brown, 2001). Psammophilous species have developed morphological, physiological and behavioural adaptations to such habitats (Etheridge, 2000). The number of species adapted to sandy substrates is high and includes many vertebrates, such as reptiles (Baumgartner et al., 2008;Carothers, 1986;Lamb & Bauer, 2006), amphibians (Thompson et al., 2005) and even some mammals (Reichman & Smith, 1990). Among them, numerous species have developed adaptations to cope with the constraints encountered in sandy substrates, like abrasion, hypoxia and mechanical resistance.
The granularity of the substrate could greatly influence benthic species by, for example, leading to potential physiological adaptations to maintain an oxygenated water flow (Robards et al., 2000). Sand particles can also be a source of abrasion for the integument of these benthic species, and many species have developed adaptations to cope with the mechanical properties of sand particles (Wainwright & Lauder, 2017). The friction made by these particles acts as a significant selective pressure, leading to specific morpho-physiological adaptations, such as special integument features (scales, mucus production; see Fishelson, 1996). Some species of Pleuronectiformes use thin sand layers as camouflage. Consequently, their integument has special characteristics that could retain sand particles, particularly at the dorsal epidermal level (Spinner et al., 2016).
Another constraint is the oxygen availability beneath sandy bottoms.
The oxygen level is lower in intertidal environments subject to tides than in inferior bathymetric zones, which are constantly replenished by oxygenated currents. Among the adaptations to cope with low oxygen levels, metazoans, such as the worm Arenicola marina, have developed more efficient haemoglobin with a high performing affinity to oxygen (Everaarts & Weber, 1974). The choice of a suitable sandy territory, regularly washed by oxygenated water, also constitutes a behavioural adaptation to deal with potential hypoxia periods, like for the sandeel species Ammodytes tobianus (Robards et al., 2000). Finally, another constraint found in these sandy habitats is reduced locomotive ability resulting from the mechanical resistance of the substrate. Moreover, morphological specialisations are found in benthic species, as in Heteroconger hassi, which possesses a rigid tail with a reduced musculature, allowing this species to use its tail as a tool to dig into the sand (De Schepper et al., 2007). Another species, Pisodonophis boro, belonging to the same Anguiliformes, possesses a rigid tail but also a reinforced skull with reduced eyes, allowing this species to penetrate the substrate with its head or tail. A cinematic study performed on these two species highlighted that their locomotion is made by wavy movements different from the ones found in pelagic species (Herrel et al., 2011).
The first species used in the present study is the sandeel A. tobianus (lesser sandeel) belonging to the Ammodytidae (Uranoscopiformes), (Betancur-R et al., 2017). This species lives in the North Sea, the European Atlantic coasts, and the Baltic and Mediterranean seas (O'Connell & Fives, 1995). This benthopelagic species, contrary to the garden eels that live in sandy oceanic bottoms, is found near sandy shores along intertidal zones. A. tobianus is very picky in choosing its habitat, preferring a sandy substrate regularly rinsed by interstitial water (Robards et al., 2000). The two other species we used belong to the Congridae family (Order of Anguiliformes) and more precisely to the subfamily of Heterocongrinae (Böhlke, 1957): H. hassi (spotted garden eel) and Gorgasia preclara (splendid garden eel). The spotted garden eel lives in the warm waters of the Indo-Pacific region, and may be found in tropical regions (Miller et al., 2011). The splendid garden eel shares the same geographical distribution and lifestyle (Miller et al., 2011). The literature is scarce for both garden eel species.
The aim of our study was to investigate the morphology and physiology of two major organ systems of these teleost fishes, the integument and the muscular system. The first is related to the potential abrasion caused by the constant friction of sand particles; the second is related to its locomotion capacities and indirectly to the potential adaptations to low oxygen levels inside the sand. Marinelife (Spijk, the Netherlands). All specimens of this garden eel species were captured along Indonesian coasts. The fishes were located in their sandy tube when they were caught. They were then sent from Indonesia to the Netherlands at the import De Jong facilities. Eight other unsexed adult individuals (mean total length, STL: 19.33 ± 1.53 cm) of Gorgasia preclara (Figure 1c), from the same geographical area, were obtained from the same Dutch company. Specimens of both garden eel species were sent from the Netherlands to our laboratory by UPS and the fishes were carefully placed in bags full of sea water and air rich in oxygen for transport. All the animals were euthanized by a lethal dose (250 mg l −1 of water) of MS222 (tricaine methanesulfonate). The water used for the fishes was buffered by sodium bicarbonate (500 mg l −1 ).
Each animal was considered euthanized when it lost the ability to swim in the aquarium. The animals were treated according to the guidelines specified by the Belgian Ministry of Trade and Agriculture and under the control of the UMONS Ethical Committee for the Survey of Experimental Studies and Animal Welfare (agreement LA1500021).

| Histology
After euthanasia, four individuals of each species were sectioned transversally into 1 cm thick slices that were immediately fixed by immersion in to allow histological examination. In addition to the classical stainings mentioned above, a more specific staining was performed to discriminate the different subpopulations of goblet cells. Alcian Blue staining (Alcian Blue, Mayer's Haemalum) allows the nature of the polysaccharides present in the mucus to be revealed. At pH 1, only the sulphated polysaccharides are stained, and at pH 2.5 both sulphated and carboxylated polysaccharides are. Tissue sections were observed on a research optical microscope (Leitz ® Orthoplan) equipped with a high sensitivity camera (DFC 7000 T) and photographs were recorded using specialized software (Leica Application Suite X, LAS X; Leica company, Wetlzar, Germany). To highlight the potential lipidic secretions of glandular cells, we performed a staining test with Oild RedO on frozen fixed tissue, which allows lipids to be specifically highlighted. One body section of one individual per species was used.

| Cell proliferation
Epidermal proliferating cells were detected with a monoclonal antibody raised against proliferating cell nuclear antigen (PCNA), the accessory protein associated with DNA-polymerase δ. This method has been detailed in a previous publication (Piron et al., 1998). Briefly, serial sections of the paraffin embedded body of H. hassi, Gorgasia preclara and A. tobianus individuals were used to study the proliferation of epidermal cells. Each centimetre along the fish's body, a 5 μm thick paraffin section was used for the immunohistochemistry. The unmasking of antigenic sites was performed by immerging the tissue sections in a Pyrex container filled with distilled water. This container was placed in a pressure cooker filled with 800 ml of distilled water at the bottom.
Once firmly closed, the sections were submitted to high pressure at 120 C. After 20 min, the pressure cooker was turned off and the sections were cooled at room temperature for 30 min. After rinsing with distilled water, endogenous peroxidase activity was quenched by a 5 min exposure to 0.5% H 2 O 2 . Before applying the primary antibodies, the pretreated sections were rinsed in phosphate buffered saline (PBS) Preparations were counterstained with hemalun and luxol fast blue, dehydrated and mounted with a permanent medium. Controls for the specificity of immunolabelling included the omission of the primary antibody or the substitution of nonimmune mouse serum for the primary antibody. In both cases, these controls were negative.

| Histoenzymology
Four euthanized individuals per species were used for this study. The body of each individual was transversally cut into 1 cm thick segments. All the pieces were put into a little basket filled with optimal cutting temperature glue which was itself put into an isopentane beaker. The isopentane was precooled with liquid nitrogen. All the frozen fixed tissues were stored at −80 C. The samples were cut with a thickness of 12 μm at −20 C using a CM1950 cryostat (Leica) and placed on silane-coated glass slides. The frozen sections were then stored at −20 C until their use. The slices were treated according to the following protocol, adapted from Zghikh et al. (2014). Briefly, after an incubation at 37 C (pH 7.6) for 40 min in a solution of succinate 100 mM, nitroblue-tetrazolium 1.25 mM and Meldolablau blue 8.3 mM, the slices were rinsed in physiological saline solution (0.9% NaCl). Tissue sections were pre-incubated for 10 min at pH 10.4 and rinsed in distilled water. Thereafter, histological sections were incubated for 1 h at 37 C in a solution of adenosine triphosphate adjusted to pH 9.4. Finally, the sections were rinsed three times in 1% CaCl 2 , then in 2% CoCl 2 for 3 min and rinsed again three times in distilled water. Revelation was performed by incubation in 1% (NH 4 ) 2 S for 3 min. After rinsing for 20 min in distilled water, the preparations were dehydrated and mounted with a permanent medium.

| Fibre muscle typing
For each species, the number of fibres and the area occupied by each type of muscle fibre were quantified by morphometric analysis at low magnification (objective 1.5× for A. tobianus and objective 2.5× for G. preclara and H. hassi). The procedure was based on a hardware system consisting of a Zeiss Axioplan microscope equipped with a ProgRes C10 plus colour camera (Jenoptik, Germany) connected to an IBM-compatible PC, and software designed for morphometry and colour analysis (KS 400 Imaging system, Carl Zeiss Vision GmbH, München, Germany). This imageanalysing system discriminated immunoreactive muscle fibres based on differences in colour and contrast. For each image (2080 × 1542 pixels), standardized microscopic fields (5,400,000 μm 2 by image for G. preclara and H. hassi, and 21,600,000 μm 2 for A. tobianus) were taken in order to cover the whole surface occupied by the musculature on the body transverse sections for the three species. For each fish (n = 4 per species), one transverse section was taken every centimetre along the body length (from behind the gills to the end of tail) and scanned at low magnification for morphometric analysis, as detailed above. For each species, the surfaces occupied by the three categories of muscular fibres were evaluated. These data were used to calculate the relative percentage of the musculature occupied by slow, intermediate and fast fibres.

| Scanning electron microscopy
Three small integument samples (<1 cm 3 ) were collected along the body of each species (behind the gills and before the caudal fin).
These were fixed in Duboscq-Brazil's fluid for 24 h. The fixed samples Jeol JSM-7200F (JEOL Company: Tokyo, Japan) scanning electron microscope (SEM). The mean distance (width) between two consecutive micro-ridges was measured using the aforementioned public software ImageJ (Image Processing and Analysis in Java, Java 1.6.0).

| Statistical analysis
All the statistical analyses were carried out using the R software for statistical computing (version 3.2.1). In function of their distribution, the quantitative data obtained in this study were submitted to parametric (ANOVA one-way when three or more groups were compared or Student's t-test when two groups were compared) or nonparametric (Kruskal-Wallis when three groups or more were compared or Wilcoxon when two groups were compared) tests.
P values lower than 0.05 are indicated on the graphs by *. All the boxplots show minimum, maximum, median, first quartile and third quartile values in each data subset.

| Histology
The histological structure of the integument exhibits many differences between the three psammophilous Actinopterygii species. For  H. hassi, the dermis has a low thickness and shows a loose structure (Figure 2c,d). In contrast, the dermis seems thicker and is made of a denser connective tissue in G. preclara (Figure 2e,f). Both species of garden eel have some dispersed melanocytes in the deeply dermis.
The hypodermis is less developed in both species, but we noted some white adipocyte clusters at the junction of myotomes.
The four staining methods confirmed that the epidermal goblet cells of the three studied species were stained by Masson's Trichrome   (d) The interspecific comparison was performed by the nonparametric Kruskal-Wallis test and significant differences set at P < 0.05. *Indicates the significant differences. All the boxplots show minimum, maximum, median, first quartile and third quartile values in each data subset FISH content. In the three fish species, only a fine layer of adipocytes stained by Oild RedO were present at the hypodermis level ( Figure 3f).
Finally, the Alcian Blue staining method allowed us to distinguish two populations of goblet cells for both H. hassi (Figure 4a,b) and G. preclara (Figure 4c,d). One subset of goblet cells was deeply stained and the other remained very poorly contrasted, but presented a granular content. These two cell types remained stained by the PAS on consecutive sections (data not shown), pointing to a glycoprotein content in both goblet cell subpopulations. In both garden eels, the Alcian Blue staining at pH 1 (Figure 4a (Table S1)

| Epidermal thickness
The thickness of the epidermis was statistically different between G. preclara and the two other species (Figure 6a). More precisely,

| Cell proliferation
The cells targeted by the anti-PCNA primary antibody were clearly identifiable by the DAB precipitation (brown nuclei) (Figure 6b-d).
The number of proliferating cells was very low for H. hassi

| Scanning electron microscopy and transmission electron microscopy
The SEM micrographs highlighted a thin geometrical pattern for both H. hassi and G. preclara (Figure 8a,b, respectively). These patterns looked like 'fingerprints' and were found at every location of the skin of both garden eel species (even on the fin rays). These structures have a diameter of about 10 μm for both species. A. tobianus (Figure 8c,d) presents a scaly pattern made of typical cycloid scales.
These scales had a width of about 240 μm (data not shown). The transmission electron microscopy micrographs allowed us to highlight the epidermal glandular types of both garden eel species. We highlighted a first type of goblet cells, recognizable by its numerous white-contrasted mucous secretion grains (Figure 8e-g). A second type of glandular cell was also observed, which was characterized by numerous small serous granules of secretion (Figure 8h). These granules had a round shape and exhibited a dark contrast. Sacciform cells were observed in the basal part of the epidermis (Figure 8f). These cells were characterized by a significant expansion of the rough endoplasmic reticulum ( Figure 8i). The apical keratinocytes presented regular micro-ridges clearly visible at the surface of the epidermis (Figure 8f,g).

| Muscle fibre typing
The histoenzymology performed on cryosections of our three species revealed three types of muscular fibres identifiable by their . The assessment of intraspecific differences was performed by the nonparametric Kruskal-Wallis test and significant differences set at P < 0.05. *Indicates the groups that are statistically different. All the boxplots show minimum, maximum, median, first quartile and third quartile values in each data subset colouration (Figure 9). For the three species, the fast fibres were coloured in brown, the intermediate fibres in black and the slow fibres in blue. The intense blue colouration of the slow fibres was due to the targeting of the succinate dehydrogenase located inside the mitochondria, giving a spotted pattern to these fibres. The fast-myosin ATPase activity gives the brown colour to fast muscular cells. The combination of the brown precipitate (resulting from ATPase activity) and the blue colour (associated with succinate dehydrogenase activity) gives a dark black staining pattern to the intermediate fibres. The total scanned surface of muscles fibres analysed by morphometry gave us insights about the relative proportions of fibre types. The statistical analyses showed that the surface occupied by each fibre type per microscopic field was significantly different for A. tobianus (Figure 10a). Indeed, the surface occupied inside myotomes by fast fibres was significantly higher than those of slow and intermediate fibres.
On the other hand, intermediate fibres occupied a larger surface of myotomes compared to slow fibres. The same intraspecific differences were found for G. preclara (Figure 10b) and H. hassi ( Figure 10c). For the latter, there was, in contrast, no statistically significant difference between the slow and the intermediate fibres. As the total scanned surfaces were not comparable because of the absolute size differences of the three species (A. tobianus being much broader than the garden eels), we estimated the relative proportions of each fibre type between the species (Supporting Information Table S2). The majority of muscle fibres were composed of fast fibres for all species, followed either by intermediate or slow fibres.

| DISCUSSION
Sandy substrates, because of their constraints, subject living organisms to significant selective pressures. The behavioural ecology of the genus Solea (Pleuronectiformes) is, for example, characterized by a hiding behaviour that is helped by morphological adaptations at the skin level (Spinner et al., 2016). In contrast, sand-burying species must resist sand abrasion by producing protective mucin or by developing a thick integument acting as armour.
We note that the presence of melanocytes in the epidermis is uncommon among the teleost fishes, where this kind of cell is almost exclusively located in the dermis for a lot of species. All the histological staining methods performed on the tissues of both garden eel species did not allow us to confirm the precise nature of the sacciform cell secretions. These cells remained uncoloured with all the stains and thus led us to conclude that they contain no big molecules, like sulphated or carboxylated polysaccharides. This lack of colouration could be an indication that these secretions are made of small uncoloured, hydrophilic molecules (ions, amino acids, catecholamine) or even of peculiar proteins. We highlighted, through electron microscopy, the presence of numerous dilated cisternae of rough endoplasmic reticulum within these sacciform cells. This characteristic points to a high rate of protein synthesis. According to the scarce literature on the subject, many hypotheses have been suggested concerning the nature of secretions and the physiological role played by sacciform cells. The study of Mittal and Munshi (1971) performed on three fossorial teleost species proposed a hypothesis suggesting that the sacciform cells are a holocrine gland instead of a merocrine one.
The cells that we have described are unicellular and do not match with the holocrine gland hypothesis detailed in their study, these theoretically requiring multicellular components. In G. preclara the sacciform cells are exclusively located at the basal epidermal level. In H. hassi they are found at mid-height between the basal and the apical epidermis, but generally closer to the basal epidermis. These observations contradict the Mittal and Munshi (1971) hypothesis about migrating cells, at least for our two garden eel species. It was not possible to identify any lipidic content for the paraffin embedded tissues, the fixation process eliminating all the fatty components. The frozen-fixed tissue allowed us to perform Oil RedO staining, which did not colour the sacciform cells. In light of these results, we can confirm that the sacciform cells of both garden eel species seem to be free of lipidic content (fatty acids or triglycerides). We can also exclude an albuminous secretion because of the negative results obtained with Mallory's Trichrome. Another study, by Fishelson (1996), was performed on a garden eel species, Gorgasia sillneri. In this study, the author offered many hypotheses on the role and nature of the sacciform cells and highlighted the PAS coloured cells. In light of our results, where no sacciform cells were coloured by this staining, we suppose that Fishelson (1996) wrongly named these cells, which were actually a subset of goblet cells. Our last hypothesis is that these sacciform cells are in fact epidermal club cells. These cells, which are common in the superorder of Ostariophysi, do not discharge their content until cutaneous injuries or stress occur (Chivers et al., 2007;Manek et al., 2013). The substances present in the secretions are made of amino acids or catecholamine that act as chemical alarm cues, or even small peptides acting as biocide. Beyond the Ostariophysi, some species such as Salmo trutta and Salvelinus alpinus exhibit epidermal sacciform cells which seem to contain proteinaceous compounds (Pickering & Fletcher, 1987). In this particular study, the proteinaceous material of those cells was specifically highlighted by stainings such as mercuric bromophenol. We must note that the presence of sacciform cells with a proteinaceous content,  elson (1996) and Mittal and Munshi (1971) observed goblet cells in the integument of garden eels, although they did not mention any differentiation inside this cellular type.
As the structure of the epidermis is different between the two garden eel species and the sandeel A. tobianus, we expected to observe a very different localization for the goblet cells. For A. tobianus, these cells were in small numbers and mainly localized in the overlapping zone of two consecutive scales. We hypothesise that the mucus secreted in these zones could ease the shift between the scales when the animal is swimming or digging in the sand. Moreover, this mucous film could prevent the accumulation of sand grains in these hinges when the animal is completely buried. The large number of glandular cells in both garden eels suggests that the mucus surrounding the body plays an important function in the resistance to abrasion. Mittal and Munshi (1971) and Fishelson (1996), proposed this hypothesis. The mucus could play a lubricant role and could reduce the mechanical injuries due to abrasion. Fishelson (1996) goes further, pointing out the role of the micro-ridges that he observed in G. sillneri and that we have observed in both garden eel species. These superficial micro-ridges could improve the resistance to abrasion by anchoring the mucus at the epidermal surface, prolonging its protective effect. These micro-ridges were observed at every level of the body of G. preclara and H. hassi, even at the fin rays level (not shown).
It is noteworthy that other roles are attributed to the mucus produced by the goblet cells. Among the main roles, which are not mutually exclusive, are protection and osmoregulation of the gills (Robards et al., 2000), ion and electrolyte regulation (Shephard, 1994), repulsion of predators (Fishelson, 1996) and protection against bacteria (Ebran et al., 2000).
The results obtained concerning the cellular epidermal proliferation and the epidermal thickness point to major interspecific differences. A. tobianus possesses a slightly thinner epidermis compared to G. preclara, but the sandeel exhibits a 10× higher epidermal cell turnover compared to both garden eels. In contrast to most of the vertebrates, where proliferating cells are restricted to the basal germinal layer of epidermis, in A. tobianus these cells are located on many levels, from the basal layer to the mid strata of the epidermis.
Moreover, we have brought to light that this proliferation is much greater in the anterior part of the body. We could make the assumption that this is caused by the peculiar behaviour of this species, the lesser sandeel digging into the sand using its head. Considering that the mucus production is assuredly reduced for this species, the injuries caused by abrasion are inevitable (Robards et al., 2000). The hypothesis that seems the most probable is that A. tobianus copes with this mechanical constraint, with the help of a great cell turnover, particularly at the head level. We find this kind of adaptation in human beings in which the cell proliferation rate is higher in the parts of the body where there is thick skin submitted to frequent friction, such as the palms of the hands and the soles of the feet.
Thus, the epidermis of the sole of the foot is entirely renewed every 20-30 days (Iizuka, 1994). In both garden eel species, the cellular epidermal turnover is much slower. The PNCA-positive cells were rather scarce and were limited to the stratum germinativum. We hypothesise that epidermal glandular cells produce a glycoprotein film stabilizing the walls of tunnels in which they live, which would allow the friction of the epidermis to be reduced. This would explain why we observed a lesser proportion of epidermal cellular turnover for G. gorgasia and H. hassi. The epidermis of H. hassi is thinner that of G. preclara, which could be explained by the fact that the latter possesses an epidermis with more stratification. Considering the fact that both species share the same lifestyle and present the same cellular types, we have no convincing explanation concerning this interspecific difference.
In general, the musculature represents 40-60% of the body mass of teleost fishes (Alami-Durante & Rescan, 2003). This musculature is organized in structural units called myotomes, which vary greatly according to the species, the developmental stage and even the position inside the animal (Johnston et al., 2011). This structure, in its most classical pattern (Altringham & Ellerby, 1999;Zhang et al., 1996), is found in A. tobianus, whereas both garden eels have a much more complex pattern. Inside these muscular bundles three fibre types are found: slow, intermediate and fast.
Numerous factors influence the distribution of muscular fibres among teleost fishes, particularly temperature and oxygen availability (Johnston, 2006), and many fishes are capable of great acclimation, which can affect the fibre typing in function of the location and the time at which a fish is captured (Sidell, 1980). As for the epidermis, the muscular system presents major differences between our three psammophilous fishes. A. tobianus possesses a standard myotome structure. In this species, which lives part of the time in cold and turbulent waters rich in oxygen, we could expect a higher proportion of slow and intermediate fibres compared to both garden eel species, which live in warm waters. In contrast to this assumption, the slow fibres represent only a weak proportion of muscles fibres, about 3%. On the other hand, intermediate fibres represent about 10% of the total muscular fibres. The reasons behind these numbers are not necessarily due to the environmental oxygen level but probably due to the fossorial lifestyle of A. tobianus. Once buried, the sandeels capture the small amount of oxygenated water within the sandy substrate. They are still able to live there, but they capture 40-50% less oxygen than in the water column (Behrens et al., 2007).
This weak level of available oxygen in the near substrate surrounding the animal could explain the very high proportion, more than 80%, of fast fibres, as these rely on anaerobic glycolysis and thus consume less oxygen. When discussed in the literature we see that the relative proportion of fast fibres can vary along the body from 66.3% to 82.5% for a nonpsammophilous and pelagic species, such as Stenotomus chrysops (Zhang et al., 1996). The study of Drazen et al. (2013) indicates that benthic fish species seem to possess on average a lower proportion of slow fibres (2.1%) when compared to benthopelagic species (7.6-10.2%). The study of Altringham and Ellerby (1999) also mentions the fact that constantly swimming pelagic species tend to have more slow fibres.
In comparison with our results, we see that both garden eels, considered as benthic species, exhibit a higher proportion of slow fibres than the benthopelagic species A. tobianus. The fact that we have obtained a lower proportion of slow muscles fibres for the benthopelagic and more active A. tobianus, in contradiction of the prementioned studies, pushes us towards a hypothesis linked to the poor level of oxygen in the environment. Indeed, despite being notably more "active" than both garden eels, this species has to face regularly poor oxygen levels when buried in the sand.
Moreover, the small amount of oxygen inside the blood circulation may be aimed towards vital organs (brain, heart, kidneys).
Beside hypoxia, other factors could equally play a role in favour of a rich content of fast fibres (and therefore low intermediate and slow fibres content). Coarse sand (1-2 mm diameter) induces a mechanical resistance such that the animal has to perform powerful body movements to bury itself, and the fast fibres have the ability to furnish such power. Indeed, A. tobianus is able of very quick movements reaching 5 m s −1 (Robards et al., 2000). Another possible explanation could be the intense predation to which A. tobianus is subjected, considering that they are an essential food resource for many predatory fishes and bird species (Daunt et al., 2008;Robards et al., 2000). To survive, only powerful and fast movements, potentially provided by the higher fast fibre content, are required. G. preclara and H. hassi exhibit an uncommon myotome structure, which is separated into four distinct muscle bundles. The results seem to indicate a higher proportion of slow and intermediate fibres in both species in comparison to A. tobianus. This could be a handicap in view of the lifestyle of these garden eels living in warm tropical waters, which are poorer in oxygen, despite being benthic and low active species. However, some hypotheses could be proposed to explain these results. Both G. preclara and H. hassi live in tunnel nests (Tyler & Smith, 1992) open to the water column, which gives them the advantage of having constant access to the more oxygenated pelagic waters. In contrast to A. tobianus, these species constantly live in the same place and do not seem to be hunted by larger species. Thus, with such weak predation they would not need as many fast fibres allowing as fast movements as the sandeel. In summary, the higher relative proportion of fast fibres for A. tobianus could be due to potential hypoxia, mechanical resistance and escape and hunting behaviour, all of which are nonmutually exclusive.
In conclusion, the adaptations observed in the integument and muscles are very different between the sandeel A. tobianus and the garden eels G. preclara and H. hassi. These differences are related to the very different behaviours adopted by these two psammophilous fish families.