Normal stages of embryonic development of a brood parasite, the rosy bitterling Rhodeus ocellatus (Teleostei: Cypriniformes)

Abstract Bitterlings, a group of freshwater teleosts, provide a fascinating example among vertebrates of the evolution of brood parasitism. Their eggs are laid inside the gill chamber of their freshwater mussel hosts where they develop as brood parasites. Studies of the embryonic development of bitterlings are crucial in deciphering the evolution of their distinct early life‐history. Here, we have studied 255 embryos and larvae of the rosy bitterling (Rhodeus ocellatus) using in vitro fertilization and X‐ray microtomography (microCT). We describe 11 pre‐hatching and 13 post‐hatching developmental stages spanning the first 14 days of development, from fertilization to the free‐swimming stage. In contrast to previous developmental studies of various bitterling species, the staging system we describe is character‐based and therefore more compatible with the widely‐used stages described for zebrafish. Our bitterling data provide new insights into to the polarity of the chorion, and into notochord vacuolization and yolk sac extension in relation to body straightening. This study represents the first application of microCT scanning to bitterling development and provides one of the most detailed systematic descriptions of development in any teleost. Our staging series will be an important tool for heterochrony analysis and other comparative studies of teleost development, and may provide insight into the co‐evolution of brood parasitism.


| INTRODUCTION
The bitterlings are a group of freshwater teleost fish in the family Acheilognathidae (Teleostei: Cypriniformes) which have a brood parasitic life-history . With their elongated ovipositor, female bitterlings lay eggs in their preferred mussel host species through the exhalant siphon of the mussel (Boeseman et al., 1938;Duyvené de Wit, 1955;Olt, 1893;Rouchet et al., 2017;Wiepkema, 1962). The eggs are introduced directly into the gill chamber by the ovipositor (Chang, 1948). They come to occupy the lumen of the water tube between the gill lamellae (Kim et al., 2008;Kim et al., 2017;Tankersley, 1996). Later, male bitterlings release their sperm near the inhalant siphon (Boeseman et al., 1938;Smith & Reichard, 2013). The sperm are carried into the mussel by the water flow and fertilize the eggs; subsequently the eggs undergo their early stages of development inside the gill chamber (Liu et al., 2006;Reichard et al., 2007). When the larvae are capable of swimming, they migrate into the exhalant cavity and emerge from the host; this marks the end of the parasitic phase of their life (Aldridge, 1999).
Because of their interesting life history, bitterlings have been intensively studied by ecologists and behavioral biologists (Boeseman et al., 1938;Smith et al., 2004;Wiepkema, 1962). Aspects of the bitterling life cycle that have been studied include the parental behavior of bitterlings  and bitterling-mussel co-evolution (Mills & Reynolds, 2003). For studies of bitterling and mussel phylogeny, see Chang et al. (2014) and Kawamura et al. (2014). While there have been many studies of adult bitterlings, their development is less wellknown. This is largely because of the difficulties of accessing the early embryos inside the mussel host. Furthermore, when early embryos are dissected out of the mussel, it is difficult to stage them because the precise time of fertilization is unknown (Duyvené de Wit, 1955;Olt, 1893).
The unique early life history of bitterlings means that the embryos are subject to a protected but physiologically challenging environment, which may result in deviations from typical teleost embryonic development (Aldridge, 1999). The first stage description of the early life history of the European bitterling, Rhodeus amarus, was given by Olt (1893). Olt (1893) noticed the peculiar forms of the yolk and presented the changing shapes of the yolk stage by stage in illustrations, whereas he failed to give precise developmental age of these stage. Fortunately, the techniques of in vitro fertilization and time-controlled in vitro incubation were used to obtain embryos of the most widely-distributed oriental bitterling, the Rosy bitterling, Rhodeus ocellatus (Chang, 1948;Chang & Wu, 1947;Kim & Park, 1985;Nagata & Miyabe, 1978;Park & Han, 2018). Previous studies provided detailed descriptions (Chang, 1948;Chang & Wu, 1947), as well as schematic developmental atlases of bitterling embryonic stages based on external morphological characters (Kim & Park, 1985;Nagata & Miyabe, 1978;Park & Han, 2018). Nagata and Miyabe (1978) described 30 developmental stages of R. ocellatus: 14 stages before hatching; and 16 post-hatching, "prelarvae" stages. Unfortunately, the illustrations provided by Nagata and Miyabe (1978) are not annotated. Kim and Park (1985) described 20 pre-hatching stages with special emphasis on the period from yolk plug closure to presence of the tailbud (stages O-S in their series). They successfully illustrated the migration and elongation of the rostral end towards the ventral side but did not indicate the rostral-caudal axis in their illustrations, which leaves some descriptions ambiguous.
The developmental age was recorded by Nagata and Miyabe (1978) in terms of hours post fertilization (hpf). Kim and Park (1985) and Park and Han (2018) also used the term hpf for pre-hatching stages, but used "hours post hatching" (hph) and "days post hatching" (dph) for later stages. Those three studies differed in the timing that they assigned to certain developmental events. For example, "hatching" was at 27.8 hpf, 39 hpf, and 50 hpf in Nagata and Miyabe (1978), Kim and Park (1985) and Park and Han (2018), respectively. We suggest that this difference in the time of hatching recorded in these first two studies is because the rearing temperature was different, namely: 22 ± 1 C in Nagata and Miyabe (1978), 17-25.5 C in Kim and Park (1985). It is well-known that temperature influences developmental timing F I G U R E 1 Rhodeus ocellatus, the development timeline. Abbreviations: hpf, hours post fertilization; YSEs, yolk sac extensions (Werneburg, 2009). The greatest difference in time of hatching is between Park and Han (2018) and Nagata and Miyabe (1978), although the rearing temperature in Park and Han (2018) is 21.5 ± 1 C, more similar to Nagata and Miyabe (1978). It is because the newly-hatched embryo illustrated by Park and Han (2018) corresponds to a later developmental stage in Nagata and Miyabe (1978). Because Kim and Park (1985) used a relatively wide temperature range, we have chosen to use here the developmental age in Nagata and Miyabe (1978) as a guide. In Nagata and Miyabe (1978) and Kim and Park (1985), hatching occurred at the same morphological stage, namely when the embryo has 6-10 somites, Kupffer's vesicle is present, and the tailbud is not yet free from the yolk extension.
In the literature on teleost development, it is sometimes stated that the embryo becomes a larva at hatching (e.g., Ali et al., 2011;Ballard, 1981). Kunz (2004) noticed that bitterlings have an ostraphilic reproductive habit (laying eggs in mussels) and have a nidicolous (nest-dwelling) type of hatching. This means that they hatch at a relatively early age and are not capable of independent living by means of, for example, free-swimming and foraging (Aldridge, 1999;Li & Arai, 2010). Therefore, the term "postembryo" is suggested by Kunz (2004) to describe the fish after hatching until the yolk is completely absorbed. After that, when exogenous feeding begins, the fish is termed a larva.
Here, we shall define the embryonic period of bitterlings as beginning at fertilization and ending when the embryos are capable of swimming out of their mussel host. Post-hatching individuals are termed "embryos" instead of "larvae" in this study ( Figure 1). Hatching, in this view, implies the breaking of the chorion, and the embryo-tolarva transition is a nutritional definition (when endogenous feeding transitions to exogenous feeding).
In the current study, we make detailed descriptions of developmental stages in R. ocellatus in order to extend the published studies.
For post-hatching stages, we use microCT (X-ray microtomography) to reveal the internal structure of the embryos in tomographic sectional view, and to provide 3-D (three-dimensional) visualization of the external morphology of development. In preliminary studies (data not shown) we found that the perivitelline space of bitterlings is quite narrow which means that there is little room to perform dechorionation with forceps without damaging the yolk or embryo. Therefore, we decided not to dechorionate the eggs. Because the chorion is highly impermeable to many reagents (Masuda et al., 1986;Masuda et al., 1992), we assumed that the contrast medium needed Note: Intensity is the "light" intensity that reaches the detector camera. Typically, the exposure time was set so that the intensity was at least 5000 in the darker parts of the sample. Abbreviations: exp, exposure; hpf, hours post fertilization; s, second.
for microCT would not penetrate. Therefore, we decided not to use microCT for the pre-hatching stages.
Advantages of the microCT technique are that it is less timeconsuming than conventional histological sectioning, it does not destroy the samples, and larger specimens can be studied than is possible with, for example, confocal microscopy (Bassi et al., 2015;Huisken & Stainier, 2009;Weber et al., 2014). The main disadvantages of microCT are a lack of cellular resolution, the inability to use special stains to identify particular structures or molecules, and the inability to perform in vivo tracing. Metscher (2009) has pioneered the application of microCT to developmental biology by developing a soft-tissue staining protocol. MicroCT has been used to study mouse development (see for example the 3-D mouse embryo atlas (Wong et al., 2012), compare phenotypic variation of larval and juvenile zebrafish at the histological level (Ding et al., 2019), and quantitative morphometric analysis of adult teleost fish (Weinhardt et al., 2018).
By analyzing a developmental series of the rosy bitterling, with 3-D and sectional views, we hope to provide a practical guide to staging bitterlings embryos in the lab and field, and provide a knowledge foundation for future research that focusses on development, comparative embryology, evolutionary developmental biology (evo-devo), and gene expression patterns. Our study may also serve as a model in the emerging discipline of eco-evo-devo, or ecological evolutionary developmental biology, which aims to integrate evolution and development with ecology (Abouheif et al., 2014;Gilbert et al., 2015).
temperature and water temperature were kept at 22.5 ± 1 C and the fish were checked every day by caretakers. Fish were fed with frozen chironomid larvae (Ruinemans Aquarium B.V., Montfoort, NL) daily. Duck mussels (Anodonta anatine, Linnaeus, 1758) and swan mussels (Anodonta cygnea, Linnaeus, 1758) were obtained from Vijver-centrum Enschede, Aquaria Veldhuis; Enschede, NL, and kept indoors with natural light from a window, in a shallow filtration tank without feeding. Water from the filtration tank was led into the fish tanks to stimulate the bitterling mating behavior.

| In vitro fertilization
Embryos with synchronized development were obtained by in vitro fertilization following the method of Nagata and Miyabe (1978).
Briefly, sexually mature parental fish were chosen based on the bright mating color of the male and the elongated ovipositor of the female. Eggs were expressed from 35 females into a clean, dry 10 cm Petri dish by gentle abdominal compression. Sperm was also harvested from 25 males by gentle abdominal compression. We used a narrowmouthed pipette to distribute the sperm evenly over each batch of eggs in a clean Petri dish. Fresh aquarium water was then added, so as to synchronously activate embryonic development. Embryos were raised 20 per Petri dish containing embryo water (Kimmel et al., 1995) changed every 24 h. The Petri dish was kept in an incubator with stationary shelves at 22.5 ± 1 C.

| Time-lapse videography
For all pre-hatching stages, we used time-lapse videography of embryos at room temperature with epi-illumination from a fiber-optic lamp (Schott KL 1500 LCD). Photos were taken every 5 min with a CCD (chargecoupled device) camera (Nikon DS-Fi1-L2) connected to stereo microscope (Nikon SMZ1500 sealed with paraffin oil and parafilm, and stabilized in a polystyrene tube during scanning (see Figure 2).
The raw data for 3-D imaging of the samples were acquired using an Xradia 520 Versa 3-D X-ray microscope (Zeiss). The X-ray source was set to 80/7 or 40/3 (keV/W). A thin LE1 filter was used to avoid beam hardening artifacts. To obtain high resolution images, a CCD optical objective (4×) was used. The acquisition parameters were set according to the developmental stage of the sample stages (   Our results are divided into: (i) pre-hatching stages; and (ii) posthatching stages ( Figure 1). The pre-hatching stages begin at fertilization, and include cleavage, blastula, gastrula and neurula periods, and end at hatching. These pre-hatching stages were all studied by time-lapse videography in live embryos. The developmental age was calculated from the time-lapse videos. The post-hatching stages include the somitogenesis, pharyngula, and organogenetic periods. These periods were originally applied to zebrafish development by Kimmel et al. (1995). The definition of these "periods" is arbitrary, but useful for organizing the stages and comparative with zebrafish staging series. Cleavage is meroblastic, as in other teleosts including Danio rerio (Kimmel et al., 1995). At the 2-cell stage, the blastodisk becomes divided symmetrically, forming two equally-sized blastomeres ( Figure 4a). This stage is comparable to Kimmel stage 2-cell.

STAGE 3: Blastula, 3.7 hpf
The blastula stage (Figures 4b and 5) is characterized by the proliferation of blastomeres so that they come to form cells of many layers deep. A distinct border, the yolk syncytial layer (YSL), appears between the blastodisk and yolk. In late blastula stages, epiboly movements start so that the blastodisk spreads towards the vegetal pole, engulfing the underlying yolk ball. The animal-vegetal (A-V) axis becomes shortened and the shape of the embryo changes from pearshape to ellipsoid (compare Figure 4a,b). This stage is comparable to Kimmel stage 256-cell.

STAGE 4: 50% epiboly, 15 hpf
Epiboly is coordinated by three morphogenetic movements: spreading, convergence, and extension (Xiong et al., 2014). First, the blasto-  There are three somite pairs. The head has assumed its definitive location at the end of the wide bulb-end of the chorion. During neurulation the neural ectoderm develops into the neural plate, which forms the neural keel by primary neurulation (Lowery & Sive, 2004). The neural keel is triangular in cross-section, and initially solid; it later forms the neural rod which has a circular cross-section and is also solid. The eye field, a common primor-

| Segmentation period
During the somitogenesis period, segmentation of somites continues and rhombomeres develop. The somite number is a quantal (discrete) staging character and is therefore particularly useful in comparative developmental studies (Battle, 1940;Furutani-Seiki & Wittbrodt, 2004;Iwamatsu, 2004;Signore et al., 2009;Tsai et al., 2013). The elongation of the tail during the somitogenesis period is a useful staging character.
Embryos before tailbud protrusion have <12 somite pairs (Figure 8a). There are 28 somite pairs. The YSEs are increasingly narrowed at their tips (Figures 8c and 12a). Kupffer's vesicle is no longer visible ( Figure 12j). The tailbud is flexed dorsally at its caudal end ( Figure 8c). The median fin fold is now visible as a continuous ridge extending the length of the tail (Figures 8c and 12j). The cephalic flexure of the neural tube is now apparent, dorsal to the hypothalamus (Figure 8b). The neural tube is completely hollow (Figure 12a).
The midbrain-hindbrain boundary (mhb) is s a shallow constriction (isthmus) of the neural tube (Figure 12a). The midbrain and hindbrain ventricles are becoming expanded (Figure 12a). Rhombomeres

| Pharyngula period
In defining this period as the "pharyngula" period, we are following the lead of the Kimmel stages (Kimmel et al., 1995). The term "pharyngula" was introduced by Ballard (1981)      The tips of the wing-like YSEs are directed caudally, and covered by skin warts (Figure 16b). The caudal yolk extension is markedly tapered at its caudal end (Figure 16b). The median fin fold in the tail becomes taller than the caudal yolk extension (Figure 16b). At the ventral base of the caudal fin fold, the prospective caudal fin rays are appearing ( Figure 17b). The median fin fold at dorsal extends rostrally, its anterior-rostral margin approaching the axial level of myotomes 6-8 ( Figure 19p).
In the bitterling, the telencephalic ventricle undergoes eversion, as it does in other ray-finned fishes including the zebrafish (Mueller & Wullimann, 2009;Wullimann & Puelles, 1999). This is in contrast to F I G U R E 2 0 Rhodeus ocellatus, stage 2-ovl, microCT images, virtual sections. (a-q) Transverse section views, dorsal towards the top, sections from rostral to caudal, direction of section plane from (a) to (f) indicated in (r); section plane in (g)-(q) indicated in (s) (Folgueira et al., 2012). Therefore, instead of two lateral ventricles uniting in the midline, the bitterling has a large, fan-shaped telencephalic ventricle everted dorsoventrally (Figures 15b and 19a). The isthmic constriction (mhb) is much deeper than in previous stages (compare Figure 15a,b).
The olfactory placodes are oval, with their long axes parallel to the rostrocaudal axis (Figure 15b). The optic tectum expands to the lateral side, overlying the optic cups (Figure 15b).
A pair of common cardinal veins (ducts of Cuvier) is present, and contains blood flowing from the yolk sac to the inflow tract (sinus venosus). The heart has a regular heart beat and blood circulation. A solid endodermal rod, the primordium of the gut, is appearing F I G U R E 2 1 Rhodeus ocellatus, stage 1-ovl, microCT images, virtual sections. (a)-(q) are transverse section view, dorsal towards the top, sections go from rostral to caudal, direction of section plane from (a) to (f) indicated in (r), section plane from (g) to (q) indicated in (s); (r)   The pericardial cavity bulges prominently from the surface of the yolk sac ( Figure 20a). In live specimens the common cardinal vein is red and contains flowing blood (Figure 16c'). Gill rudiments appear as shallow furrows rostral to the otic vesicle ( Figure 16c). The cerebellum is clearly distinguishable at the axial level of r1 (Figures 15c and 20g).
In sagittal microCT virtual sections, the epiphysis appears as a swelling in the midline of the diencephalic roof plate (Figure 20r)

| Organogenetic period
Throughout the development of the previous (pharyngula) period, the body plan of the embryo was established. In the current period, In dorsal view, the pectoral fin bud is dome-shaped (Figure 23a).
The height of the pectoral fin bud is equal to its dorsoventral width (Figures 23a and 29n). The apical ectodermal ridge is discernible The olfactory bulb is forming (Figure 29a). What we presume to be the inner plexiform layer of the retina is distinct (Figure 29c,d). There is no mouth opening; the mouth is indicated by a shallow groove  (Figures 22b and 30s). Olfactory pit a shallow groove (Figures 22b and 30a). The mouth a small opening, not yet gaping (Figures 30c,d,s and 28a). No gill filaments present on the branchial arches (Figures 30g and 28a). Rudiments of pharyngeal teeth appearing on the 5th branchial arch (Figure 30h). Branchial clefts not yet open (compare Figures 30h and 31h). Cells in the liver have the histological features of hepatocytes (Figure 30n). A common chamber of the saccule and lagena appears (Figure 26b). This stage corresponds to Kimmel high-pec stage, on the basis of the morphology of the pectoral fin bud.
STAGE 23: Long-pec, 235 hpf (9.8 dpf) The apical ectodermal ridge of the pectoral fin develops into the fin fold (Figures 23c and 31o). Chondrocytes are differentiating in the pectoral girdle (Figure 31o). The YSEs shrinks to vestigial bumps (Figures 22c and 31q). Melanophores are differentiating in a rostrocaudal gradient in the skin, and are most prominent on the dorsal surface of the head (Figure 22c'). The entire retinal pigment layer is pigmented (Figure 22c').
The olfactory epithelium of the bowl-shaped olfactory pit is connected to the olfactory bulb by a distinct olfactory nerve with sequences in other non-parasitism teleost based on parsimony analysis (Ito et al., 2019;Jeffery et al., 2005). Sequence heterochrony (changes in the order in which events occur) is an important mechanism for the evolution of development (Bininda-Emonds et al., 2002;Mabee et al., 2000). Our study demonstrates the value of microCT in developmental biology. In addition to being relatively time-efficient compared with routine histology, it is a non-destructive technology. For species that were previously difficult to study because of limited material, microCT scans provide a wealth of morphological data and readily yield 3-D information.

| The body direction in relation to the polarity of the chorion
According to Suzuki (1958), the time-window in which bitterling eggs can be fertilized is about 30 min after the egg has been activated when contacting water; sperms remain viable for only 7 min after contact with water. Egg activation is an irreversible process and is independent of the presence of sperm (Kunz, 2004). Once activated, the chorion becomes inflated and lifts from the egg surface ( Figure 3). A funnel-shaped micropyle, a specialization of the chorion, connects the chorion and egg surface, and serves as a passage for the sperms to fertilize the egg (Suzuki, 1961). When the female bitterling oviposits eggs inside the host mussel, these eggs are inevitably activated.
Therefore, a successful fertilization requires that the male bitterling releases sperm near the host mussel within 30 min and the sperms find their way to the micropyle.
The chorion of the rosy bitterlings eggs is bulb-shaped (Figures 3 and 4). The micropyle is always at the narrower stalk pole (Suzuki, 1961). As we observed, during the hatching period, the head always emerges from the chorion at its wider (bulb) side

| Notochord vacuolization is not necessary for body straightening
During the process of body elongation, especially from stages s-18 to s-32, the notochord elongates and expands in diameter because of vacuolization of the inner layer of notochordal cells. The notochord first has a typical "stack-of-coins" appearance at the s-18 stage (Adams et al., 1990;Koehl et al., 2000), suggesting that the subsequent vacuolization of notochord plays an important role for body elongation by providing a mechanical force needed for straightening of the body axis along the RC axis (Stemple, 2005). Ellis et al. (2013) argued against this hypothesis, providing in vivo experimental evidence indicating that the notochord is not necessary for embryonic straightening.
Our Segmented surface view of the digestive system the previously solid endodermal rod develops into an alimentary canal and gives rise to liver, gall bladder and pancreas buds, as well as the endodermal lining of the swim bladder. Abbreviations: bd, basidorsal cartilage; gb, gall bladder; ib, intestine bulb; li, liver; no, notochord; p, pancreas; pd, pronephric duct; ph, pharynx; pi, posterior intestine; pt, pronephric tubule; sb, swim bladder; sc, semicircular canal; y, yolk. Scale bar, 1 mm

| The yolk extension forms independently of body straightening
In the zebrafish, the straightening of the body axis (from its original conformation of being curved over the yolk sac) overlaps temporally with the formation of the yolk extensions (Virta, 2009 (Easter Jr. & Nicola, 1996;Schmitt & Dowling, 1994, 1999, we find that the developmental timeline of early eye morphogenesis between the zebrafish and bitterling is broadly similar. The degree of ventral displacement of the optic primordium is similar between the bitterling at the 10-somite stage (Figure 9c) and the zebrafish at the 11-12 somite stages. In the 16-17 somite stages of zebrafish, the lens placodes appear, and this event takes place at the corresponding 28-somite stage of the bitterling ( Figure 12d). However, in the zebrafish, the lens is completely detached from the surface ectoderm at the prim-5 stage (24 hpf); F I G U R E 2 9 Rhodeus ocellatus, stage pec-bud, microCT images, virtual sections. (a)-(r) are transverse section view, dorsal towards the top, sections go from rostral to caudal, direction of section plane from (a) to (f) indicated in (s), section plane from (g) to (r) indicated in (t); (s) and (t) (Easter Jr. & Nicola, 1996). These responses not only require retinal pigmentation, but also retinal lamination (Figure 24), and formation of extraocular muscles (Figure 25)  we find that they take place at a comparable stage, although the morphogenesis of the pars inferior of the inner ear is strikingly pre-displaced in bitterling development. Specifically, in the bitterling, the induction of the otic placode is at the 10-somite stage (Figure 9d), the same as in the zebrafish (9-10 somite stages; (Haddon & Lewis, 1996;Whitfield et al., 2002).    (Bever & Fekete, 2002;Whitfield et al., 2002).
The third otolith (asteriscus of the lagena), starts to forms at the pec-fin stage in the bitterling, whereas the same event happens much later (9-17 dpf) in different studies of zebrafish depending on the strain (Bever & Fekete, 2002;Riley & Moorman, 2000;Whitfield et al., 2002). According to one study of otolith development and vestibular function in the zebrafish (Riley & Moorman, 2000), the utricular otolith is necessary and sufficient for vestibular function and survival in the zebrafish, whereas the sacculus and lagena otoliths function primarily in hearing. Therefore, we expect a pre-displacement of hearing development in the bitterling, which may be related to brood parasitism life and development in a dark environment where hearing is more useful than vision. This in turn would also explain why visual development appears to be delayed in the bitterling.

| MicroCT in developmental biology
The bitterling is not an easy species to study. Its YSEs and its large, opaque yolk mass are in contrast to the small, transparent early stages of zebrafish development. It is therefore much more difficult to observe with optical microscopy. We found that it is not feasible to manipulate differential interference (DIC) optics to count somite numbers during the somitogenesis period or trace the migration of lateral line primordia during the pharyngula period, which are key characters of staging in zebrafish embryos. In addition, the yolk mass becomes brittle when fixed, making it difficult to perform routine histological.
For these reasons, 3-D reconstruction from serial sections is not the optimal technique for studying bitterling development.
This study has shown that application of microCT is a highly efficient technique for studying rosy bitterling development. The volume rendering of X-ray tomography is sufficient to virtually display the staging features of interest (compare for example the left and right columns in Figure 8).
Virtual slices provide microanatomical tissue details (e.g. retinal lamination; Figure 24), and provide us with the ability to reconstruct complex 3-D structures that were previously only visible through dye-injection methods (e.g., the semicircular canal and alimentary tract in Figure 27).

| CONCLUSIONS
This paper represents one of the first detailed studies of development in any teleost species using microCT. To define stages, we have used numeric characteristics such as somite number and prim-number, which facilitate comparison with zebrafish stages, and more broadly facilitate the study of evolution and development in other teleosts.