Flow Sensing in the Deep Sea: Morphology of the Lateral Line System in Stomiiform Fishes

The deep sea is characterized by extreme environmental conditions including limited light availability, which makes non-visual sensory capabilities quite important. In addition to the visual, auditory, olfactory and gustatory systems found in all vertebrates, all fishes have a mechanosensory lateral line system. This system is composed of neuromast receptor organs on the skin and in bony canals on the head and trunk and is sensitive to unidirectional water flows and low frequency vibrations. Our knowledge about the lateral line system (LL) in deep-sea fishes is limited. Of the taxa in which the LL has been described, there appears to be two morphologies: widened LL canals with large canal neuromasts, and a reduced LL canal system with a proliferation of superficial neuromasts (SNs). However, the one published description of the LL in a species of the prominent midwater order Stomiiformes, suggests that there is a third LL morphology defined by a reduction in canals on the head accompanied by just a few, small SNs. The goal of Chapter 1 was to use traditional and modern morphological methods to provide the first detailed description of the LL system in two groups of stomiiform fishes (Argyropelecus [Family Sternoptychidae] and Cyclothone [Family Gonostomatidae]) as well as other fishes in the Families Gonostomatidae, Phocichthyidae, and Stomiidae in order to test the hypothesis that reduced canals and a reduced number of SNs is a strategy in the evolution of the LL in stomiiform fishes. A total of 27 species in 17 genera in four families were studied using one or more morphological approaches (including histology and micro computed tomography, or μCT). In depth analysis of four Argyropelecus species (Sternoptychidae) revealed the presence of several incompletely ossified LL canals on the head, including the supraorbital (SO), mandibular (MD) and preopercular (PO) canals. Examination of whole preserved specimens of other taxa also revealed the presence of SO, MD, PO, and IO (infraorbital) canals with varying degrees of ossification. Few species had all canals typically found in bony fishes, but all taxa, with the exception Cyclothone species, had some cranial LL canals. A proliferation of SNs was revealed using histological material and scanning electron microscopy in fishes in the families Sternoptychidae, Gonostomatidae, and Stomiidae. These domed, circular structures (= “white dots”) stood out against darkly pigmented skin and were visible under low magnification in very well preserved specimens. They appeared to be innervated, and had a similar morphology and distribution on the head and/or body in all species. Thus, it appears that there is a proliferation of SNs in many stomiiform genera, that the initial description of just a few SNs in Argyropelecus is incorrect, and that the LL is more important to these deep-sea fishes than has been previously suggested. The discovery of a proliferation of SNs in these fishes made it necessary to be able to distinguish them from the numerous bioluminescent photophores and other structures found in in the skin of these fishes. In Chapter 2, specimens were imaged under different wavelengths of light as a new tool to distinguish SNs from both complex and simple photophores. A total of 34 stomiiform species (in four families) and representatives of two other deep-sea taxa (Myctophidae, Melamphaidae) were examined under brightfield illumination, and at three different excitation wavelengths (390, 470, 545 nm). Complex photophores were visible under all wavelengths, while SNs and simple photophores were only visible at 390 and 470 nm. SNs and simple photophores were brighter under different wavelengths, allowing for SNs to be differentiated from photophores when illuminated under different wavelengths of light. This technique also revealed variation in the size, distribution, density, and orientation (direction of light emission) of numerous, minute, complex photophores on the head and trunk of stomiids. Additionally, the fragile, gelatinous coating found in some stomiid fishes and the simple photophores found within it were described in Chauliodus sloani and Stomias boa ferox.


Introduction
Approximately 65% of the surface of the earth lies at depths below 200 m, in the dark waters of the deep sea, making this habitat one of the largest on our planet. Yet, we know less about the deep sea than we do about the surface of the moon (Nouvian, 2007).
This prominent habitat is characterized by extreme environmental conditions ( Fig. 1.1).
The waters of the deep sea reach near freezing temperatures and experience high pressures (increasing 1atm per 10 m depth increase). Furthermore, only 1% of visible sunlight from surface waters penetrates below 200 m into the mesopelagic zone (~200 -1,000 m), while no detectable light extends into the bathypelagic zone (~1,000 -4,000 m). This light limitation prevents photosynthesis from occurring in the deep sea, resulting in lower biomass in this environment compared to that in shallow water habitats (Angel, 1997;Haedrich, 1997;Montgomery et al., 2014). Additionally, low-light levels define the sensory environment, limiting the detection of visual stimuli and thus challenging processes of prey and mate location and predator avoidance, which are otherwise dependent on vision in shallow water. In order to cope with these extreme environmental conditions, fishes have evolved some interesting sensory adaptations, such as enlarged olfactory structures and eyes capable of detecting lower levels of light including those generated by bioluminescence. While adaptive visual morphologies have been fairly well described in several deep-sea fish taxa (i.e. Gagnon et al., 2013;Busserolles et al., 2014), little information is available on the non-visual sensory systems including the mechanosensory lateral line system.
The mechanosensory lateral line system is a primitive vertebrate sensory system found in all fishes (and in larval and adult aquatic amphibians), that detects hydrodynamic stimuli (water flows and vibrations). It is known to mediate many important behaviors including prey detection, predator avoidance, and communication (reviewed in Webb et al., 2014a). The sensory organs of the lateral line system, the neuromasts, are composed of directionally sensitive hair cells that respond to unidirectional water flows and vibrations. Neuromasts are located either superficially, on the skin surface (superficial neuromasts), or within cylindrical pored canals (canal neuromasts), which run through a conserved set of dermal bones of the skull and the lateral line scales on the trunk (Webb, 1989). Variation in the morphology of the lateral line system is defined by variation in cranial canal phenotype (narrow, narrow with widened tubules, narrow with branched tubules, widened, or reduced; Webb, 2014b;Fig. 1.2), development and placement of the trunk canal, morphology of individual neuromasts, and the size, shape, and location of these neuromasts within canals or on the surface of the body (Webb, 1989;2014b).
This morphological variation is thought to contribute to variation in flow sensing capabilities among fishes Montgomery et al., 2014). For example, it has been shown that superficial neuromasts detect flow velocity, whereas, canal neuromasts detect flow acceleration and pressure gradients (McHenry and Liao, 2014). Further, neuromasts in widened lateral line canals are more sensitive to hydrodynamic stimuli than neuromasts in narrow canals, but exhibit slower response times (Denton & Gray, 1989;Schwalbe et al., 2012). It has been suggested that such variation in the morphology of the lateral line system is adaptive, and correlates with variation in the behavioral and ecological characteristics of fishes.

The Lateral Line System in the Deep Sea
The morphological specializations of the lateral line system found among diverse groups of fishes living in extreme environments, such as the deep sea, can be used to explore patterns of adaptive evolution of the lateral line system. In an environment with lower light levels, fishes need to rely more heavily upon non-visual sensory modalities.
The deep sea is an environment with relatively low hydrodynamic noise in which a more sensitive lateral line system can function effectively. Therefore, it is reasonable to expect the lateral line system to be well-developed and relatively important for mediating behaviors in deep-sea fishes The lateral line systems of shallow water taxa have been studied extensively, but data on deep-sea fish lateral line systems are scattered in the literature and are often incomplete (reviewed by Webb, 2014). The lateral line system has been described in a small number of deep-sea taxa (i.e. Saccopharyngidae, Nielsen & Bertelsen, 1985; Myctophidae, Lawry, 1972a,b;Macrouridae, Marshall, 1965, Fange et al., 1972 Lophiiformes, Caruso, 1989, Marshall, 1996Trachichthyiformes, Jakubowski, 1974, Moore, 1993Melamphidae and Melanonidae, Marshall, 1996). These studies suggest that the lateral line system in deep-sea fishes demonstrates two evolutionary strategies, which have each arisen convergently in multiple fish taxa.
Neuromasts in such widened canals are more sensitive than those in narrow canals, but also react slower to flow stimuli (Denton & Gray, 1988;. Fishes with widened canals feed in low light or darkness and have been shown to feed more successfully (i.e. Eurasian ruffe) in light-limited environments than fishes with narrow canals (discussed in Schwalbe et al., 2012). These studies have led to the hypothesis that widened canals evolved convergently as an adaptation for feeding in light-limited environments (Coombs et al., 1988;Janssen, 1997).
The absence of canals may be explained by selection for reduced bones, including those bones in which cranial lateral line canals are typically found. Bone reduction serves to reduce overall density of a fish providing it with more buoyancy that reduces sinking rates allowing fish to remain in the water column ( Fig.1. 4). Although these fishes lack cranial lateral line canals, they have a proliferation of superficial neuromasts on the head and body ( Fig. 1. 3D-F;Marshall, 1996). In the anglerfishes (i.e. Phrynicthys wedli; Fig.   1.3D) and in the snipe eels and gulper eels ( Fig. 1.3E -F), these superficial neuromasts sit upon papillae or stalks that extend into the environment, which possibly increases sensitivity to flows. At least some neuromasts are found in locations where cranial canals would typically be present (i.e. above and below the eye, along the lower jaw and the operculum, and around the nares), and likely represent "replacement neuromasts", where canals have been lost (Coombs et al., 1988). Handrick (1901) reported what appears to be a third morphological strategy in the hatchetfish, Argyropelecus hemigymnus (Sternoptychidae: Stomiiformes; Fig. 1.5), where a reduction of lateral line canals is accompanied by only a very small number of superficial neuromasts. The only other observations on lateral line morphology in these fishes were made by Marshall (1954), who made incidental comments on the "lateral line organs, which are particularly numerous on top of the head" in hatchetfishes. He also stated that "Cyclothone spp., the commonest bathypelagic fishes, have both canal organs and free-ending organs" , and suggested that "compared to [myctophids], the other fishes of the twilight zone the [stomiids] have small or poorly developed lateral line organs" (Marshall, 1954). With these few, conflicting reports on the lateral line system of stomiiform fishes, it is unclear if Argyropelecus has a reduced canal system with few neuromasts (as suggested by Handrick, 1901) and further, if this morphology is characteristic of other taxa within the Family Sternoptychidae, or more broadly within the Order Stomiiformes.
Two of the most abundant and well-known stomiiform fishes are those of the sister genera (as per Harold and Weitzman, 1996) Argyropelecus (hatchetfishes: Sternoptychidae) and Cyclothone (bristlemouths: Gonostomatidae). Both genera are widely distributed, frequently collected in midwater trawls, and commonly found in museum collections, making specimens readily available for detailed morphological analyses. Cyclothone is often referred to as the most numerous vertebrate genus on the planet (Grey, 1960;Maynard, 1982).
Both Argyropelecus and Cyclothone are delicate fishes with many "larval" characteristics (Maynard, 1982;Marshall, 1984). They lack scales and have a very thin epithelium (only 1 -2 cell layers thick; personal observation). As is common in stomiiforms and in deep-sea fishes in general, they exhibit extreme cranial bone reduction, including the loss or reduced size of many lateral line canal-bearing bones ( Fig. 1.4B-C). Weitzman (1967Weitzman ( , 1974 and Fink (1985) provide detailed osteological descriptions of several stomiiform genera and make reference to lateral line canals or to canal pores in the canal-bearing bones in some species (Table 1.1). Based on descriptions and figures in these papers, it appears that stomiiforms have few cranial canals, but there are no explicit references to canal or neuromast morphology. Furthermore, these studies say little about the lateral line canals in Argyropelecus in particular, and say nothing at all about those in Cyclothone. However, Cyclothone does have enlarged centers in the brain that suggest heightened importance of the lateral line system (Maynard, 1982) and Marshall (1977) mentions the presence of superficial and canal neuromasts. However, in a detailed study of the ecology of Cyclothone spp., Maynard (1982) was unable to locate superficial neuromasts when closely inspecting specimens.
The goal of this project was to use a combination of traditional and modern methods to carry out the first detailed description of the lateral line system in Argyropelecus (Sternoptychidae) and Cyclothone (Gonostomatidae) species in order to test the hypothesis that reduced canals and a reduced number of superficial neuromasts, as reported by Handrick (1901), is a strategy in the evolution of the lateral line system in deep-sea fishes. Furthermore, a comparison of lateral line morphology in Cyclothone and Argyropelecus to that in other stomiid genera will shed light on potential adaptive significance of variation in lateral line morphology among Stomiiformes.

Specimen Collection and Identification:
Study specimens were obtained from the Museum of Comparative Zoology at Harvard University (MCZ) or were collected on several cruises and provided for this study. See Table 1.2 for a complete list of species used and their collection data and Specimens were identified to species level using several sources.  was used to identify stomiiforms to family level. Specimens were then identified to genus and species level using other, more specific keys. Stomiids from the Western North Atlantic were identified using keys in Fishes of the Western North Atlantic  and were referenced against fishbase.org to check for recent name changes.
Fishbase.org was used to determine geographic ranges of the specimens from the Eastern Pacific. In most cases, species had a circumglobal distribution (so identification using Fishes of the Western North Atlantic was sufficient).
Identification of sternoptychids was carried out using keys and species descriptions in Schultz (1961) and Baird (1971). Phylogenetic revisions of Sternoptychidae done after these keys had been published tended to group taxa into a single species (Harold, 1993). Thus, fishbase.org and Harold (1993) were used to find the most recently accepted species distinctions.
Gonostomatid taxonomy is not as straightforward as Sternoptychid taxonomy.
Gonostoma specimens collected in the Western North Atlantic were identified using Fishes of the Western North Atlantic . Identification of Cyclothone specimens was relatively challenging because a key including all 13 named species was not available. Furthermore, locality and depth data were not available for all specimens used in this study, restricting the use of regional Cyclothone identification guides. Thus, data from Mukhacheva (1966), DeWitt (1972, DeWitt andCailliet (1972), Badcock (1982), Bond and Tighe (1984), , Miya (1994a, b),  were used to form a character matrix (Appendix 1) which was then used to construct a new dichotomous key including ten of the 13 species of Cyclothone (Appendix 2). This is now the most complete dichotomous key of the genus to my knowledge, incorporating 10 of the 13 recognized Cyclothone species. However, it is not locality-specific and should be referenced against species distributions when collection locality is known.

Morphological Analysis:
A combination of standard histological protocols was used to describe lateral line canal and neuromast morphology. Specimens were sampled from collections mentioned above focusing on those specimens in good condition. This resulted in a detailed morphological analysis of a relatively small number of specimens in a limited range of species.

Examination of Whole Specimens:
All specimens were first examined and imaged using a dissecting microscope min (depending on the batch and age of hematoxylin used). Specimens were removed from the hematoxylin when the putative neuromasts had been stained, but before the skin began to stain, allowing for the differentiation of neuromasts from general epithelium.
The location of putative neuromasts was noted and they were counted using ImageJ (nih.gov).

Clearing and Staining:
Cranial lateral line canals within dermatocranial bones were examined using specimens enzymatically cleared and stained for bone and cartilage (Pothoff, 1984). In addition, a modified clearing and staining procedure (Song and Parenti, 1995) using Sudan Black B was used to visualize nerve innervation (due to staining of lipids in the myelin sheath) in order to locate innervated superficial neuromasts in the skin.

Histology:
Histological material was used to study canal morphology and to locate and count both canal and superficial neuromasts. Observations indicated that the canals in many of the species examined had what appear to be incompletely developed canals or canal segments. Thus, canal morphology was described using the stages of canal morphogenesis as described by Webb (2014a) and Bird and Webb (2014;Fig. 1.6 Inc.). The preopercular canal (and therefore, its canal neuromasts) runs dorso-ventrally so methods used to measure length and diameter were reversed. Observations of superficial neuromasts in whole preserved specimens revealed that they are round, so length (along the rostro-caudal axis) was used to approximate neuromast diameter.
The number and distribution of superficial neuromasts were determined in histological material by digitally superimposing a 14x14 box grid (generated by SPOT 5.2, Diagnostic Instruments, Inc.) on the image on the computer screen. The center of the grid was placed at a stable landmark (i.e. in the center of a large bone or mid-dorsal on the head). The location of each neuromast was noted with reference to the landmark at the center of the grid, to ensure that sections of one neuromasts in sequential sections were identified appropriately, especially when neuromasts were located in very close proximity to each other in certain areas on the specimen.
µCT Imaging: Micro computed tomographic (µCT) imaging was utilized to visualize the presence/absence and morphology of partially and fully ossified canals as determined by the presence of canal pores or open troughs, which represented incompletely ossified canals ( Fig. 1.6). Specimens were imaged at the Museum of Comparative Zoology at Harvard (Bruker SkyScan 1173, pixel size range = 14.9 -54.7µm). Canal presence or absence was also confirmed by examining 2-D slices. 3-D µCT data were reconstructed using OsiriX (v3.6.1 64 bit, volume rendering).

Scanning Electron Microscopy:
Scanning Electron Microcopy (SEM) was used to assess superficial neuromast morphology (neuromast shape and size, sensory strip shape, and hair cell orientation).
Specimens were dehydrated in an ascending ethanol series at room temperature, critical point dried out of liquid CO 2 (Tousimis Samdri-780A), mounted on aluminum stubs with adhesive carbon discs, and coated with 15 nm platinum (Leica MED 020). Specimens were viewed at 3 KV at a working distance of ~10 mm on a Zeiss NTS Supra 40VP or JEOL 5900 LV SEM to generate digital images.

Results
The morphology of the lateral line canals and canal and superficial neuromasts was examined in 27 species representing 17 genera in four families of stomiiform fishes.
Of these, all but one genus (Cyclothone) had cranial lateral line canals (Table 1.4).
Variation in the subset of canals present and the morphology of canals ( Fig. 1.6 affinis, A. hemigymnus, and A. lychnus). They followed the same pattern as those in nerve stained specimens and seemed to be closely associated with superficial neuromasts, suggesting that they are innervating these structures.
Given the similarity of the distribution and morphology of superficial neuromasts in A. aculeatus to the white dots and nerve-stained structures in other Argyropelecus species, it is concluded that all Argyropelecus species examined have a proliferation of superficial neuromasts.
The distribution of superficial neuromasts was highly dependent on specimen condition. The epithelium of whole preserved specimens appeared to be easily damaged, and both scales and epithelium were easily flaked off the surface of the fish. Histology revealed that the epithelium in A. aculeatus is indeed very thin (1 -2 cell layers thick;

Fig. 1.9) and SEM revealed that the skin in both A. aculeatus and A. lychnus was fragile
and was apparently completely smooth with no evidence of cellular boundaries or the microridges characteristic of the epithelium of teleost fishes. Although round, domed structures were found medial to the longitudinal bony ridges on the dorsal surface of the head in the same location as superficial neuromasts had been observed prior to specimen preparation, no superficial neuromasts could be detected using SEM. It is possible that there was tissue covering the superficial neuromasts, or that the epithelium in which they were contained was entirely missing, thus removing any traces of superficial neuromasts.
Finally, the variability in specimen condition made it difficult to accurately determine the exact number and distribution of superficial neuromasts in all specimens of a given species. However, a comparison of superficial neuromast distributions in multiple specimens of the same species revealed that as many as ~357 superficial neuromasts are on one side of the head and body of A. aculeatus and ~521 superficial neuromasts are on one side of the head and body of A. hemigymnus (Table 1.5). This is in stark contrast to the report of only 17 superficial neuromasts and 6 canal neuromasts by Handrick (1901).

Family Gonostomatidae
The fishes in the family Gonostomatidae (the bristlemouths) have elongate bodies and rows of ventral and lateral photophores. Two of the eight gonostomatid genera were examined here: Cyclothone and Gonostoma. Both are characterized by elongate jaws with numerous, fine teeth. They are similar in morphology, but Cyclothone spp. have much smaller eyes and smaller body size than Gonostoma spp. . "White dots", the small, domed (~55 -80 µm in diameter), circular structures found in Argyropelecus, were also found covering the head and trunk of well-preserved Cyclothone spp. These were more apparent on the darkly pigment skin of deeper dwelling Cyclothone species (i.e. C. acclinidens and C. microdon), but could still be visualized in lighter colored, more transparent specimens (i.e. C. braueri, C. pseudopallida, C. signata. Furthermore, the white dots stained positively with hematoxylin, appearing darker than the surrounding epithelium ( Fig. 1.14A, D; 15A). In whole preserved specimens, and in hematoxylin stained specimens, these structures appeared to be innervated by nerves.
Histological analysis of two C. microdon and SEM of 1 -2 individuals from each of 2 species (C. braueri, C. microdon) revealed superficial neuromasts in the same locations as the white dots in whole preserved specimens. These appeared smaller than those in Argyropelecus and had a mean diameter of only ~53 µm ( microndon, C. signata) revealed superficial neuromasts in these same locations.
Interestingly, SEM did not reveal neuromasts in C. microdon specimens that had obvious superficial neuromasts that had stained positively with hematoxylin prior to preparation for SEM. Instead, SEM revealed concave, circular depressions in the skin, which were sometimes filled with debris with no signs of hair cells. These depressions were numerous and prominent against the otherwise smooth skin, and were found in the same pattern as the superficial neuromasts observed in the whole specimen prior to being prepared for SEM. This suggested that the superficial neuromasts were damaged during specimen preparation, leaving behind only the depressions in the skin seen using SEM.
C. acclinidens stained for nerves revealed vertical nerve branches on the trunk, with approximately one branch per body segment. They started near the spinal column and moved towards the skin surface. In whole preserved Cyclothone, opaque, white, branching filaments were visible against darkly pigment skin or through lighter, transparent skin and followed the same pattern as the nerves identified in the nerve stained C. acclinidens ( Fig. 1.15C). They appear at or near the horizontal septum with branches extending both dorsally and ventrally. The ventral branches diverge around the lateral photophores and then around the ventral photophores, forming a pentagon midventrally. A similar pattern is found where the dorsal branches extend dorsally from the horizontal septum. Dorsally, the branch diverges, connecting with another horizontal segment, which is visible laterally at the dorsal portion of the trunk. In whole preserved specimens, superficial neuromasts are found in discrete lines either directly aligned or in close association with these putative nerves ( Fig. 1.15C).
Cyclothone specimens were very fragile and the description of superficial neuromast morphology (using SEM) and distribution (using observations of whole preserved specimens, histology, and hematoxylin staining), was dependent on specimen condition. The epithelium is very thin (1 -2 cell layers thick; Fig. 1.14) and easily damaged during collection and handling. Using SEM the skin appeared smooth with neither cellular boundaries nor microridges, and only a few superficial neuromasts could be located that were in good enough condition to allow determination of hair cell orientation. A more complete picture of neuromast distribution was determined using several C. microdon individuals ( Fig. 1.15B) revealing ~500 superficial neuromasts on one lateral side of the body with ~6 superficial neuromasts between each of the ventral photophores (those between the pectoral and pelvic fins), for a total of ~66 superficial neuromasts ventrally. Thus, it is estimated that there may be a total of over 1066 neuromasts on both sides of the head and body of a single fish.

Gonostoma elongatum
Examination They are domed, circular structures that stand out against the darkly pigment skin, are visible under low magnification on a dissecting microscope, and appear to be innervated.
They are smaller than those in Argyropelecus and Cyclothone (~30 -40 µm in diameter), but have a similar distribution and are found around the nares, on the opercular region, in rows along the upper and lower jaws, and in vertical lines caudal to the orbit ( Fig. 1.17B).
On the trunk, they are found in vertical lines that run around the circumference of the body, with approximately one vertical line per body segment (defined by the myomeres;

Family Phosichthyidae
Species in the seven phosichthyid genera are elongate with lateral and ventral series of photophores with a body form similar to that of the gonostomatids .

Ichthyococcus ovatus
The cranial osteology of Ichthyococcus ovatus was observed using µCT reconstructions only, which revealed both fully and partially ossified cranial lateral line

Family Stomiidae
The barbeled dragonfishes of the family Stomiidae are diverse, with approximately 286 species in 27 genera . Most have an elongate body, a mental barbel, and ventral and lateral photophore series in addition to numerous minute photophores on the head and body. Lateral line morphology is described here in representatives of 14 genera based on examination of whole preserved specimens, histology, and/or µCT imaging.

Aristostomias tittmani
The cranial osteology of Aristostomias tittmani was observed in µCT reconstructions only, revealing SO, MD, and PO canals (Table 1.4; Fig. 1.19). An obvious, and mostly ossified SO canal is present and a longitudinal bony ridge, with which the SO canal appears to be associated, extends dorsally from the frontal bone ( The canal appears to be rather narrow, yet the canal pores in the bone are relatively large. In the lower jaw, canal pores were found in the rostral portion of the dentary bone ( Astronesthes spp. Numerous superficial neuromasts were visible in histological material ( Fig. 1.20A -D, G). These were smaller than canal neuromasts and densely placed, with multiple superficial neuromasts (morphologically distinct from photophores) sitting side by side in a vertical line ( Fig. 1.22A, C). Caudal to the orbit, as many as 13 superficial neuromasts were found in a vertical line in a single histological section. The analysis of sequential histological sections suggests that they are found in discrete vertical lines. A layer of pigment is found below the neuromasts and the basement membrane and pigment appear as depressions in the epithelium below the superficial neuromast. In whole preserved specimens, white dots were not seen on the skin (in contrast to Argyropelecus spp., Cyclothone spp., and Gonostoma elongatum), but brown depressions in the skin were present (like those described in G. elongatum) in the same locations as superficial neuromasts seen in histological material. These were found in discrete vertical lines on the head as well as on the trunk. They had a distribution like that in G. elongatum (Gonostomatidae), with one line running around the circumference of the trunk in each body segment. Thus, brown depressions are interpreted as superficial neuromasts that were damaged in handling or otherwise removed.

Bathophilus filifer
Canal and neuromast morphology was described in one preserved Bathophilus Small, circular, brown depressions in the epithelium are found on the head and trunk, running in vertical lines around the circumference of the trunk ( Fig. 1.21B). These appear similar to those seen in other species (e.g. Gonostoma elongatum, Gonostomatidae and other stomiids). Histology revealed no distinguishing characteristics of these structures and SEM revealed that the skin is very thin and smooth, without microridges and cellular boundaries, observed in other fish taxa. The brown depressions could be clearly distinguished from photophores, which are convex and rise above the surrounding epithelium ( Fig. 1.21C). The majority of the depressions were covered by debris ( Fig.1. 21C -D), but others were entirely clear of surface details with no identifying characteristics ( Fig. 1.21E). Thus, it appears that the brown depressions are not superficial neuromasts, but likely represent the locations of superficial neuromasts that were damaged and removed during handling.

Echiostoma barbatum
Examination of a single whole preserved Echiostoma barbatum specimen and 3-D µCT reconstructions of three specimens revealed lateral line canal pores in the epithelium and bones, respectively (Table 1.

Eustomias hulleyi
One whole preserved specimen of Eustomias hulleyi was available for study.
Small canal pores in the epithelium indicate the presence of several cranial lateral line canals (Table 1.4). A longitudinal bony ridge extends dorsally from the frontal bone.
Rostral to the orbit, this ridge is separated into two ridges (inner and outer), which merge into a single ridge at the level of the orbit (as in Gonostoma elongatum and Echiostoma barbatum; Fig. 1

Flagellostomias boureei
One whole preserved specimen of

Idiacanthus antrostomus
Examination of two whole preserved specimens revealed the presence of lateral line canals (Table 1. The SO canal runs in close association with a longitudinal bony ridge extending dorsally from the frontal bone ( Fig. 1.12C Fig. 1.24). A SO canal with two bony pores is located rostral to the orbit and appears to terminate caudal to the orbits ( Fig. 1.24A). A trough is found in the rostral portion of the dentary bone, which is interpreted as a partially ossified MD canal (St. 2b or 3, Fig. 1.24B). A trough is also present in the preopercle, which is interpreted as a partially ossified PO canal (St. 2b or 3; Fig.1. 24A).
Neonesthes spp.  Fig. 1.25). A SO canal, with four canal pores, begins rostral to the orbit ( Fig. 1.25A). The rostral portion of the MD canal is well ossified and has five or six bony canal pores ( Fig. 1.25C). More caudally, the canal is represented by a trough, indicating that the more caudal portion of the MD canal is not fully ossified (St. 2b or 3). The PO canal has three or four pores with two or three located ventrally on the preopercle and one or two pores located more dorsally and is interpreted as being fully enclosed and ossified (St. 4; Fig. 1 the orbit but never fuse ( Fig. 1.12E). The SO canal is closely associated with these two ridges, and the three most anterior SO pores sit medial to the two ridges. The fourth SO canal pore is medial to both bony ridges and more caudally, another three pores are found medial to the two ridges. The canal then appears to move through the bony ridges so that the next two pores sit lateral to the outer longitudinal bony ridge.  1.12C). The SO canal originates rostral to the anterior naris and medial to the bony ridge.
More caudally a SO pore is found medial to the posterior naris and two more pores are found dorsal to the orbit and medial to the bony ridge. The SO canal appears to move through the bony ridge so that three pores caudal to the orbit sit lateral to the bony ridge.
The SO canal appears to be completely ossified. Caudal to the last SO canal pore, two smaller pores sit slightly ventral to the SO canal, indicating the presence of a canal caudal to the SO canal (likely the OT canal), which appears to be partially ossified (

Discussion
The morphology of the lateral line system was described in 17 genera of stomiiform fishes representing four different families using a combination of methods (histology, clearing and staining, nerve staining, hematoxylin staining, SEM, and µCT imaging). This is the first description of the lateral line system in seven of those genera, (Astronesthes, Cyclothone, Gonostoma, Ichthyococcus, Idiacanthus, Opostomias, and Neonesthes). Prior to this study, our understanding of the mechanosensory lateral line system in stomiiform fishes was limited to incidental comments by Marshall (1954;, references or illustrations alluding to the presence of cranial canals based on the presence of pores in cranial bones in several stomiiform taxa (Fink, 1985;Weitzman, 1967;, and a single detailed description of the innervation of the lateral line system in Argyropelecus hemigymnus (Handrick, 1901). Here, the use of multiple methods has revealed that the lateral line system is more elaborate than suggested by reports in the literature, consisting of ossified or partially ossified canals and a proliferation of superficial neuromasts.
All but one stomiiform genus (Cyclothone) had at least some cranial lateral line canals, which appeared to be narrow (as opposed to widened). The majority of species were observed on one side of the head and body.
The estimated numbers of superficial neuromasts provided here are conservative, but indicate that these fishes have superficial neuromast numbers comparable to those found in shallow water species described as having a proliferation of superficial neuromasts. For instance, similarly sized gobies (family Gobiidae), known for superficial neuromast proliferation on the head and body (Asaoka et al., 2010(Asaoka et al., , 2011Sumi et al., 2015), had fewer superficial neuromasts than many stomiiform specimens examined in this study (  Marshall (1984) reports in these fishes (including eye, swimbladder, kidney, and muscle morphology).
It should be noted that stomiiform fishes are not the only deep-sea taxa reported to have a proliferation of superficial neuromasts. In addition to the taxa referenced earlier that have a reduced canal system (anglerfish: Lophiiformes, snipe eel: Notocanthiformes, and gulper eel: Saccopharyngiformes; Fig. 1.3), it appears that other groups with cranial lateral line canals also have numerous superficial neuromasts. Halosaurids (Notocanthiformes), melamphaids (Stephanoberyciformes), searsiids (Argentiniformes), myctophids (Myctophiformes), evermannelids, ipnopids (Aulopiformes), and macrourids (Gadiformes) are among the groups with well-developed canals (either wide or narrow) that also have a proliferation of superficial neuromasts Marshall and Staiger, 1975;. Thus, it appears that all deep-sea fishes examined have a fairly well-developed lateral line system with variation in cranial canal morphology In Argyropelecus and Cyclothone, superficial neuromasts were often visible in the epithelium that had otherwise been damaged, suggesting that superficial neuromasts are easily removed during collection or handling. If this is the case, it makes sense that seemingly well-preserved specimens may have lost the epithelium in which superficial neuromasts would sit, leaving behind only the underlying pigmented tissue that had sat beneath the neuromast. In whole specimens, this pigment stands out as numerous, tightly spaced, depressions in the skin that are a different in color (brown) than the surrounding epithelium. It is therefore concluded that the brown depressions on the head and trunk of stomiiform specimens represent the former locations of superficial neuromasts.
The number and distribution of superficial neuromasts was highly dependent on specimen condition. All specimens examined, even the most pristine, were slightly damaged, so estimates of the distribution and number of superficial neuromasts is fairly conservative. Even with the conservative number of superficial neuromasts reported here, it was found that Argyropelecus hemigymnus has over 20 times more superficial neuromasts on the head and body than were previously illustrated (Handrick, 1901), mostly likely due to the quality of the specimen previously described by Handrick. This work has shown that our ability to accurately describe the morphology of the lateral line system in deep-sea fishes is dependent upon the acquisition and examination of highquality, well-preserved, intact specimens. The current work was limited by the amount of high quality material available and it is likely that stomiiform fishes have a greater degree of superficial neuromast proliferation than accounted for here. Future studies on a higher number of specimens from a larger number of taxa is needed to determine the total number of superficial neuromasts in these fishes, the nature of inter-and intraspecific variation in lateral line morphology, and to confirm if a proliferation of superficial neuromasts is characteristic of all taxa in the order.
Given the abundance of stomiiform fishes circumglobally and the variation in depth ranges among species (i.e. Marshall, 1954, these fishes make an interesting analyses that suggest that Argyropelecus and Chauliodus feed selectively (Merrett and Roe, 1974) and video of Argyropelecus sp. (MBARI) showing them striking at prey (rather than filter feeding), it is likely that these fishes may be using the lateral line system (in addition to other senses) to locate prey. Furthermore,  suggested that sexually dimorphic taxa (e.g., Cyclothone and Argyropelecus) may follow flows and chemical concentration gradients back to the source of pheromones produced by potential mates. Additionally, behavioral observations of various stomiiform fishes suggest that they spend most of their time suspended fairly motionless within the water column. This behavior, while possibly a mechanism to reduce metabolic demands, may also be useful for reducing self-generated flows, which may be easily detected by the numerous neuromasts over the head and body. The highly proliferated superficial neuromasts in these fishes, described here for the first time, suggests the importance of the lateral line system for the potential detection of hydrodynamic stimuli from biotic sources such as predators, prey, and mates.

Introduction
Bioluminescence, the production of light by living organisms, is a widespread phenomenon in the poorly lit waters of the deep sea. An estimated 80% of all eukaryotes living in the deep sea are bioluminescent (Davis et al., 2014), including everything from bacteria, to invertebrates and fishes. Bioluminescence has been recorded in 43 families of fishes in 12 orders (including one family of cartilaginous fishes) and has evolved at least 27 times in 14 major clades of teleost fishes (Haddock, 2010;Davis et al., 2014;Davis et al., 2016). Although difficult to study in situ, observations of bioluminescence in shallow water fishes (i.e. Porichthys notatus) and terrestrial organisms (i.e. fireflies) has led to several hypotheses about the possible functions of bioluminescence in the deep sea (reviewed in Haddock et al., 2010;Widder, 2011). Bioluminescence is thought to play roles in prey location (i.e., lures or illuminating prey), predator avoidance (i.e., eliciting a startle response, creating a burglar alarm, or by using counterillumination for camouflage), and communication (i.e., to attract mates or for use in territorial displays). It is likely that bioluminescence is used for multiple behaviors in any given individual (Copeland, 1991).
In teleost fishes, bioluminescence is produced either by bacterial symbionts (i.e. in anglerfish lures, or subocular organ in flashlight fish) or endogenously in complex photophores or in luminous tissue (Mensinger, 2011;Davis et al., 2014). Photophores come in a variety of sizes, shapes, and morphologies (Mensinger, 2011) and may cover up to 12% of the total body surface of a fish (Cavallaro et al., 2004). Photophores can be simple, made up only of groups of photocytes (photogenic cells), or may be more complex and composed of photocytes grouped inside a photogenic chamber, with a lens (that concentrates the light and guides it to the external opening of the photophore), a reflector layer (with cells that reflect light emitted by the photocytes towards the lens), and a pigmented layer that surrounds the entire structure (Cavallaro et al., 2004).
Among the fishes in the mesopelagic zone of the deep sea, Myctophiformes and Stomiiformes, are known to have diverse arrangement of photophores. Myctophiform fishes (254 species in 36 genera and 2 families) are known for their species-specific photophore arrangements (found in all but the neoscopelids) and other sexually dimorphic luminous organs. The relatively high speciation rates in Myctophiformes that have species-and sex-specific photophore arrangements, compared to other bioluminescent teleosts in the deep sea, suggests that the observed variation in photophore morphology and arrangements may have driven the radiation of myctophiforms (Davis et al., 2014). Stomiid fishes (Stomiiidae: Stomiiformes) have undergone a similar degree of diversification and are incredibly speciose given the young age of the clade (286 species in 27 genera). This incredible species diversity may be explained by the variety of photophores and species-specific bioluminescent mental barbels driving a radiation similar to that seen in the myctophiforms (Davis et al., 2014;Davis et al., 2016;. Fishes in the order Stomiiformes are all characterized by bioluminescence, and those in the family Stomiidae have a particularly diverse assemblage of photophores (Fig. 2.1;Tchernavin, 1953;O'Day, 1973;Haddock et al., 2004 Cavallaro et al., 2014). The simple photophores that are found on the head and body of many stomiids (as well as the gonostomatid, Gonostoma) are composed only of groups of photocytes but lack other structures characteristic of complex photophores (i.e. lens, and pigment layer).
Due to the difficulties of studying bioluminescence in live stomiid fishes, it is hard to determine the functions of these different types of photophores. The incredible diversity of photophores on a single individual further complicates the matter, and many species have numerous simple photophores in close association with the minute complex photophores (Tchernavin, 1953;Nicol, 1960). There are only a few reports of bioluminescence in stomiid fishes kept alive for short periods of time (discussed in Nicol, 1960). Most of these studies describe light produced by the conspicuous, complex orbital photophores. Additional observations in live fishes have described luminescence apparently produced by the simple photophores that cover the head, body, and fins of Astronesthes, Chauliodus, Idiacanthus, and Stomias (Nicol, 1960;, O'Day, 1973. In studies describing light produced by the photophores on the trunk, the source of the luminescence was either undetermined or unclear, but the color of the bioluminescence is described as blue, green, or yellowish (Marshall, 1960;Nicol, 1960;Copeland, 1970;O'Day, 1973). Few studies have quantitatively described the emission spectrum of the photophores in stomiid fishes, but they appear to emit light between 460 and 485 nm (with the exception of the red light produced by genera discussed above; reviewed in Herring, 1983). Experiments have shown that the major serial photophores in Argyropelecus hemigymnus and Chauliodus sloani are capable of producing light that could function in counterillumination based on their angular distribution (ventrally directed light; Denton et al., 1972). Widder (2011) suggests that the numerous simple photophores in Melanostomias bartobeani function as a "burglar alarm" for predator avoidance, but, others suggest only that they do not appear to function in counterillumination (Tchernavin, 1953;Nicol, 1960). Little is known about the distribution or function of the numerous, minute photophores, which highlights the need for additional studies.
A closer examination of these diverse photophores is particularly needed given the new interpretation of superficial neuromast proliferation in these fishes (see Ch. 1).
The superficial neuromasts in stomiiform fishes are roughly the same size as some of the minute photophores and are found in close association with photophores ( Fig. 2.2). Thus, it is not surprising that the superficial neuromasts have been previously misidentified as minute photophores in some stomiids . This study uses single wavelength light illumination as a tool to distinguish superficial neuromasts from different types of photophores and other luminous tissue and describes the morphology and distribution of these light-producing structures in several stomiiform fishes.

Methods
Thirty-four species of stomiiform fishes from four families were studied. The general morphology and distribution of photophores in stomiiform fishes was compared to those in other deep-sea fishes of the orders Myctophiformes and Stephanoberyciformes.

Imaging:
Specimens were imaged using discrete wavelengths of light (390, 470, 545 nm) as a tool to characterize the distribution, density, and size of photophores on the head and body and to differentiate the different types of photophores from one another and from superficial neuromasts. All specimens were imaged using a Nikon SMZ1500 dissecting microscope with a SPOT RT3 25.2 2 MP camera and SPOT 5.2 imaging software (Diagnostic Instruments, Inc.). Images in multiple planes of focus were combined into a single image using Helicon Focus (Helicon Soft LTD.). Specimens were imaged using bright field illumination and when illuminated with three different epifluorescence filter sets (BFP-B excitation λ = 390 nm; GFP-B excitation λ = 470 nm; Ds-Red excitation λ = 545 nm; refer to Table 2.1 for all filter properties).

Histology:
Photophores were examined histologically in four species belonging to three families to further characterize their morphology.
Specimens were decalcified in Cal-Ex (Thermo Fisher Scientific) for 2 hours, dehydrated in ascending concentrations of ethanol and t-butyl alcohol, embedded in Paraplast (Thermo Fisher Scientific), serially sectioned (transverse sections, 8µm thickness), and mounted on slides subbed with 10% albumin in 0.9% NaCl. Sections were stained with a modified HBQ stain (Hall, 1986) and coverslipped with Entellan Other specimens (or in some cases portions of specimens) were embedded in glycol methacrylate plastic resin (Technovit 7100, Electron Microscopy Sciences) and sectioned at 5µm (on a Leica 4M2265 microtome). Every 3 rd section was mounted out of Thysanactis dentex (MCZ 96771; 55 -85 mm SL; n = 5; 470 nm only)

Results
The examination of photophore type (i.e. simple or complex), size (i.e. large or minute), distribution, orientation, and response to different wavelengths of light (390, 470, or 545 nm) in 34 stomiiform taxa revealed that all species had photophores on their head and trunk, but that the diversity of photophores varied among species. Complex photophores (i.e. large serial and minute photophores) and simple photophores, each have distinct morphologies, distributions, and responded differently when exposed to the different wavelengths of light. The superficial neuromasts present in the skin were morphologically distinct from minute photophores and were visible when illuminated by 390 nm and 470 nm wavelengths of light.
Photophore response to illumination at these wavelengths was most obvious in darkly pigmented specimens as the dark epithelium provided a stark contrast to fluorescing structures. In lightly pigmented specimens, the epithelium or underlying muscle reflected light so that photophores did not easily stand out. All of the photophores and neuromasts appeared blue when illuminated with light at 390 nm, appeared green at 470 nm, and appeared red at 545 nm. Storage solution (70% ethanol, 100% ethanol, or 10% formalin in saltwater) did not appear to have an effect on the response of any structures to single wavelength illumination.

Complex Photophores
Large serial photophores were found in all stomiiforms examined and had similar distributions among species. All large serial photophores were directed ventrally and were visible when exposed to all three excitation wavelengths used (390, 470, and 545 nm). A ventral and a lateral series on the trunk were found with approximately 1 per body segment, and a single lateral row, caudal to the origin of the anal fin (as per Harold, 2004).
All stomiiform fishes examined have one or more complex photophores near their eye ( Fig. 2.3A, C -D), which reflected light at all three wavelengths as the large serial photophores did (Fig. 2.3). In some species, multiple photophores, including both an antorbital and a postorbital photophore, were present. These two photophores responded differently from one another when illuminated under different wavelengths in some species. In Tactostoma macropus, the antorbital photophore was visible when illuminated with 390 and 470 nm light, but emitted only very dim light at 590 nm. The postorbital photophore was much brighter and was visible at all three wavelengths (Fig. 2.4). It is interesting to note that in living or freshly caught specimens, the antorbital and postorbital photophores appear to be different colors and may produce light of different wavelengths (see images in Haddock et al., 2010).
Numerous minute photophores were observed to cover the skin of many stomiid species (Fig. 2.2). Unlike the larger serial photophores, the minute photophores were not visible to the naked eye in whole preserved specimens. They were variable in distribution, size, and orientation within an individual and between species. In contrast, sternoptychids, phosichthyids, and Cyclothone spp. lacked minute photophores. Among the gonostomatids, only Gonostoma elongatum had minute photophores. When compared to those in the stomiids, the minute photophores in Gonostoma elongatum were less numerous and less diverse, but were all the same size and oriented directly outward (laterally if located on the lateral side of the body, or dorsally if located on the dorsal side of the body).
All species examined appeared to have at least two different sizes of minute photophores. Their distribution varied among species, but appeared to be relatively constant within a species and was found in each body segment on the trunk with more photophores located ventrally rather than dorsally. The photophores located more ventrally tend to be directed ventrally, but several species had some photophores directed The complex photophores all appear to respond similarly when illuminated by the three different excitation wavelengths (Fig. 2.4). All complex photophores appeared brightest when illuminated with light at 470 nm. Photophores appeared the next brightest at 545 nm (if visible at all) and appeared much dimmer at 390 nm. Photophore type (i.e alpha, beta, or gamma) did not appear to correlate with visibility when illuminated at the three different wavelengths. Species with small photophore diameter (i.e. Cyclothone microdon) appeared dimmer (Fig. 2.4), which seems to be more a factor of the size of the photophore (reducing light emission) rather than photophore type.

Simple Photophores
In addition to the numerous complex photophores, many stomiids (and the gonostomatid Gonostoma) have simple photophores on their head and trunk, but these were not present in sternoptychids or phosichthyids. These light producing tissues lack a ring of dark pigment, are smaller, and have a less defined shape, appearing occasionally globular in shape, and are located more superficially in the skin compared to the complex photophores. They appear whitish in color and have previously been described as "fattyappearing material" (in Astronesthes niger, Bigelow et al., 1964, pp. 380; Fig. 2.4C).
They are often found in groups, either densely placed in well-defined areas or found more diffusely over the head and trunk, on fin rays, and around the anal and genital openings in many species. The arrangement of simple photophores appears to be species-specific.
Although lacking numerous discrete simple photophores as seen in Gonostoma and the stomiids, some Cyclothone species appeared to have a luminous organ along the cheek and at the base of both the dorsal and anal fins. Gonostoma and all stomiids appeared to have some type of simple photophore, with more elaborate distributions found in the stomiids.
Simple photophores responded differently than complex photophores when illuminated by the three different wavelengths of light ( Fig. 2.4 -7). All simple photophores were visible when illuminated by 390 and 470 nm light, but none were visible when illuminated by longer wavelength, 545 nm light. In contrast to the complex photophores, the simple photophores appeared brightest at 390 nm, and were the brightest of the visible structures at this wavelength.
In some species, there are other tissues that appear to be luminescent and are likely simple photophores with different morphologies. Simple photophores appear to be suspended within a gelatinous coating in Chauliodus and Stomias and a luminous "gland" was seen in Cyclothone spp. Each of the structures will be described separately.

Gelatinous coatings in Stomias spp. and Chauliodus sloani
Among stomiids, two genera (Chauliodus, Stomias) are reported to have a thick gelatinous coating covering their head and/or body (i.e. , which is fragile, easily damaged during collection, and shrinks when specimens are placed in ethanol. The gelatinous coat was observed in specimens of well-preserved Chauliodus sloani (105 mm SL, n = 1) and Stomias boa ferox (110 mm SL, n = 1). These are the only two stomiiform genera with this gelatinous coat and which are both characterized by hexagonshaped scales (Fig. 2.1; i.e. . The gelatinous coat of Chauliodus sloani is not present on the head, but begins on the trunk, caudal to the gills, becomes thicker caudally, and then narrows towards the caudal fin. At its thickest point, it appears to make up almost a quarter of the diameter of the trunk and is transparent so that complex photophores in the skin are visible through the gelatinous coat using bright field illumination and single wavelength illumination (as described above, Figs. 2.6, 2.9).
Small, spherical structures are suspended within the gelatinous coat and appear to be innervated (Fig. 2.4 -7). These appear to be simple photophores based on reports of their luminescence (Tchernavin, 1953;Nicol, 1960;O'Day, 1973). In specimens in which the gelatinous coating was missing, these structures are found collapsed against the skin surface, falling into neatly clumped areas between groups of complex ventral photophores in the skin ( An almost fully intact gelatinous coating was examined in one specimen of Stomias boa ferox (Fig. 2.8), which as in Chauliodus, does not cover the head, but is present caudal to the gills, becoming thicker caudally, and thinning out towards the caudal fin.
Opposed to the seemingly homogenous gelatinous coating with few simple photophores found in Chauliodus, Stomias appears to have many more densely placed simple photophores composing its gelatinous coat. As other simple photophores, these structures are opaque white and "fatty-appearing". These, however, are more oval-shaped than the spherical simple photophores found in the gelatinous coat of Chauliodus. They are found both dorsally and ventrally, but not on the lateral surface of the trunk. They appear loosely connected to the skin and move easily when disturbed by fluid motion.
When the internal structure of the gelatinous coat is visible (on areas of the specimen where the superficial layer of the gelatinous coating was damaged), these elongate simple photophores appear to point away from the surface of the trunk so that they resemble microvilli emerging from the surface of the epithelium. A very thin layer of tissue sits superficial to the simple photophores, connecting these numerous structures. When observed in ventral or dorsal view, they are only visible through the superficial epithelial layer of the gelatinous coat as small, opaque circles that obstruct the visibility of the underlying complex photophores in the skin.
Nerves or blood vessels can be seen running to these structures and extend through the gelatinous coat. Similarly, darkly pigmented nerves running from the surface of the skin through the gelatinous coat have been observed in other Stomias species (C. Kenaley, personal communication). On the lateral surface of the trunk (where simple photophores are absent) the gelatinous coat appears transparent and thread-like structures that appear to be nerves are visible running from the surface of the scales but do not appear to run to any obvious structures. In several cases, these appear to extend beyond the gelatinous covering into the external environment and may be free nerve endings.
They were only seen on the second horizontal row of scales (counting from dorsal to ventral).
In well-preserved specimens that lacked an intact gelatinous coat, the simple photophores were still visible, but were collapsed into a longitudinal row mid-ventrally (observed in S. atriventur, n = 5; S. boa ferox, n = 1). When not connected to the gelatinous coating, they fall flat and do not stand upright as they did in the intact gelatinous coat and appear similar to the collapsed circular simple photophores in the Chauliodus specimens that lacked an intact gelatinous coat.
Additional simple photophores are also found dorsally between the nares and orbits, and extend onto the pectoral and pelvic fins, in areas not enveloped by the gelatinous coating. These appear similar to those seen in Chauliodus sloani. More caudally on the head, they gradually become more elongate, eventually becoming confluent with the gelatinous coat caudal to the gills. All of the simple photophores in the gelatinous coat were visible when illuminated by 390 and 470 nm light. With 545 nm light, the simple photophores were not visible but obscured the glow of the underlying complex photophores in the skin (Fig. 2.8).

All thirteen species of Cyclothone are morphologically similar in appearance and
Cyclothone spp. can be found in all major oceans. Species found in the same geographic range are often distributed vertically. Deeper dwelling species of Cyclothone (i.e. C. acclinidens, C. atraria, C. microdon, and C. pallida) are described as having white, "fatty-appearing" tissue in a winding pattern along the cheek and on the trunk at the base of the dorsal and anal fins. These structures have been described as "luminous glands" (Mukhacheva, 1966), however there is no evidence that these structures emit light.
These structures stained positively with Sudan Black B (a lipid stain) in all three Cycothone species. They also stained positively with hematoxylin (a nuclear stain) in both whole specimens and in histological material, confirming that the "luminous glands" are cellular in nature. In cross-section, they appear to have a highly nucleated circumference but an empty center with no apparent nerve innervation.

Superficial Neuromasts
Superficial neuromasts were found on the head and trunk of all stomiiform species examined (see Chapter 1). wavelength.

Discussion
The complement of photophore types found among stomiiform fishes was made much more apparent when illuminated with different wavelengths of light. This study found that photophore types respond differently when illuminated with different wavelengths of light and that superficial neuromasts respond differently than both simple and complex photophores. The morphology and distribution of photophores were described in 34 species of stomiiform fishes.
Illumination of specimens with different wavelengths of light proved to be a useful tool in the differentiation among types of photophores and other luminous tissue in stomiiform fishes. Under illumination at 390 nm, complex and simple photophores and superficial neuromasts were visible, but simple photophores were the brightest of these structures. At 470 nm, simple and complex photophores and superficial neuromasts were all visible. At 545 nm, only complex photophores were visible. While different complex photophore types (alpha, beta, or gamma) were not distinguishable from one another using the methods in this study (Fig. 2.4), some preorbital and antorbital photophores appeared to have variable responses to different wavelengths of light. Furthermore, the photophore lens was sometimes difficult to distinguish from surrounding tissue under brightfield illumination, but it was very distinct under illumination at 470 nm, for example.
Stomiiformes are found from the upper mesopelagic into the bathypelagic and many species migrate upwards into the epipelagic at night (discussed in Marshall, 1960).
These oceanic zones show considerable variation in light conditions (intensity and spectrum) and likely affect the evolution of bioluminescence. A comparative study of the size of different photophores in species found throughout the water column will begin to allow us to use the Stomiiformes as a model to further determine ecological significance of bioluminescence in a variety of oceanic habitats.
This study revealed an impressive variety of minute photophores, which are more diverse than the larger photophores currently used for identification of stomiiforms. A more detailed study of these photophores may reveal patterns that can be used in species identifications, as is currently done for only Chauliodus and Stomias (i.e. . The larger serial photophores are likely used for counterillumination and are known to produce intensities of light capable of this function (Tchernavin, 1953;Denton, et al., 1972;O'Day, 1973). However, given the variation in size, distribution, and orientation of the minute photophores, it is possible that they serve one or more different functions. A more detailed examination of variation in their density, distribution, and orientation may elucidate the behavioral and ecological significance of these structures.
For instance, there appears to be a higher density of photophores on the ventral surface of the fish (noted in Harold, 2004). Their ventral orientation would suggest a role in counterillumination, while the laterally and dorsally directed photophores are less likely to be used for this purpose and may function in communication, as hypothesized in myctophids.
Particularly interesting, perhaps, is the function of the simple photophores found in diverse patterns among stomatiiforms. As observed in some species, these produce a constant blueish glow around the fish. It seems unlikely that they aid in counterillumination (Tchernavin, 1953;Nicol, 1960;O'Day, 1973). This is especially true when considering the high density of simple photophores found extending onto dorsal side of the head and trunk, which do not seem likely to aid in counterillumination and those extending onto the pelvic and anal fins and around the anal/genital region, which may act more as a signal for mating. It is possible that the unique arrangements of simple photophores serve for species recognition. Simple photophores were absent in basal stomiids (i.e. sternoptychids and less elaborate in the gonostomatids; personal observation), which are also less speciose groups. The elaboration of both simple and complex photophores in the speciose stomiids suggests that there may be a connection between the number, distribution, and diversity of photophores and the radiation of these fishes in the deep sea.
It is important to note that illumination of photophores using single wavelength illumination, as revealed in this study does not mean that the tissues are necessarily bioluminescent or auto-fluorescent; therefore, results from this method should be interpreted with caution (Johnsen, 2006;Haddock et al., 2010). Although not directly examined here, epifluorecence imaging may provide insights into tissue compositions of different photophore types based on the differential reflectance of tissues under Finally, this study provided some of the first detailed descriptions of structures within the gelatinous coats of Chauliodus and Stomias. Because of its delicate nature, the gelatinous coat has only been described in a few freshly caught specimens (Tchernavin, 1953;Nicol, 1960;O'Day, 1973;Schnell & Johnson, 2012). Observations of live specimens confirm what appear to be simple photophores in the gelatinous coat that produce a bluish glow (Tchernavin, 1953;Nicol, 1960;O'Day, 1973). In this study, histological sections through the simple photophores of C. sloani provide a comparison of the differences between the simple photophores suspended within in the gelatinous coating and the complex photophores in the epithelium composing the skin (Fig. 2.6A -D); however, additional investigation is necessary to fully determine the structural and functional differences among these different groups of photophores. It is also interesting that Chauliodus and Stomias are the only two stomiid genera which are reported to have a gelatinous coat and are also the only two genera with obvious scales. It is unclear what the function of the gelatinous coat is, but it seems as if the sole function cannot be to house the simple photophores, as both genera have additional simple photophores arranged on their head and fins as seen in other stomiids. Certainly, additional investigation is needed to ascertain the function of this intriguing feature found in these two genera.    showing several minute photophores (mp) on the skin surface as well as a row of them directed at the eye. Minute photophores are interspersed with superficial neuromasts (black arrowheads). The rostral portion of the antorbital photophore (aop) is ventral to the orbit (*). Scale = 200 μm. B) A large serial photophore in C. microdon (52 mm SL) with visible photogenic chamber (pc), lens (l) and pigment layer (p). Scale = 25 μm. C) The eye (*) in A. niger, caudal to section in C showing minute photophores (mp) directed into the eye, and one attached to the eye. Scale = 200 μm. D) The antorbital photophore in A. niger (ventral to orbit [*], same plane of section as in C but at higher magnification). Photogenic chamber (pc), reflector (r), and pigment layer (p) are visible in the antorbital photophore with a smaller minute photophore visible to the right. Scale = 100 μm. E) An elongate, complex ventral photophore in A. aculeatus (28.5 mm SL) with photogenic chamber (pc), lens (l), reflector (r), pigment layer (p), and gelatinous layer (gl). Scale = 250μm. Histological material courtesy of Christopher P .Kenaley.           Combination of species locations, depth ranges, and proportional sizes (according to largest size record) for Cyclothone spp. from the Pacific, Atlantic, Southern Ocean and Mediterranean Sea. Note that no data are available for the Indian Ocean. Image from .