Per Gunnar Fjelldal, Institute of Marine Research, Matre, NO-5984 Matredal, Norway. E: email@example.com
This study examined vertebra formation, morphology, regional characters, and bending properties of the vertebral column of Atlantic cod throughout its life cycle (0–6 years). The first structure to form was the foremost neural arch, 21 days post hatching (dph), and the first vertebra centrum to form – as a chordacentrum – was the 3rd centrum at 28 dph. Thereafter, the notochord centra developed in a regular sequence towards the head and caudal fin. All vertebrae were formed within 50 dph. The vertebral column consisted of 52 (± 2) vertebrae (V) and could be divided into four distinct regions: (i) the cervical region (neck) (V1 and V2), characterized by short vertebra centra, prominent neural spines and absence of articulations with ribs; (ii) the abdominal region (trunk) (V3–V19), characterized by vertebrae with wing-shaped transverse processes (parapophyses) that all articulate with a rib; (iii) the caudal region (tail) (V20–V40), where the vertebra centra have haemal arches with prominent haemal spines; (iv) the ural region (V41 to the last vertebra), characterized by broad neural and haemal spines, providing sites of origin for muscles inserting on the fin rays – lepidotrichs – of the tail fin. The number of vertebrae in the cervical, abdominal and caudal regions was found to be constant, whereas in the ural region, numbers varied from 12 to 15. Geometric modelling based on combination of vertebra lengths, diameters and intervertebral distances showed an even flexibility throughout the column, except in the ural region, where flexibility increased. Throughout ontogeny, the vertebra centra of the different regions followed distinct patterns of growth; the relative length of the vertebrae increased in the cervical and abdominal regions, and decreased in the caudal and ural regions with increasing age. This may reflect changes in swimming mode with age, and/or that the production of large volumes of gametes during sexual maturation requires a significant increase in abdominal cavity volume.
During embryonic and early larval stages of teleost fish, the notochord, through its hydroskeletal properties, provides axial body support. Eventually, the notochord itself is the template for a functional vertebral column as it nucleates the development of the vertebrae by the formation of ring-shaped structures of mineralized notochord sheath – known as chordacentra – along the notochord; each of these are the primordia of a vertebra centrum (Arratia et al. 2001; Grotmol et al. 2003). Conforming to the same segmental pattern, two chains of paired serially homologous bones associate with the chordacentra: on the dorsal side neural arches are formed, whereas ventrally, depending on region, rib-bearing parapophyses or haemal arches develop. The individual structures that make up the fully developed vertebrae arise through specific developmental processes, and attain regional anatomical characters that define the functional properties of the backbone as a whole – a complex organ facilitating body support and propulsion.
Due to the swim bladder, the teleost backbone does not need to counteract gravity (bear weight), hence, its prime task is to support propulsion (see review by Wardle et al. 1995). Functional demands, and thus the morphological adaptations to these, vary from region to region within the vertebral column. Such adaptations to function are recognized as metameric patterns of variations in number, and size and shape of the vertebra centra. Regarding vertebra shape, the ratio between the cranial-caudal length and lateral-lateral diameter has been used as a measure of axial elongation in teleosts; in an evolutionary perspective, elongation of the vertebral column has been associated with an increase in the number of vertebrae, an increase in the length of the vertebral centra, or a combination of both (Ward & Brainerd, 2007). In snakes, differences in this ratio between species have been suggested to relate to function and mode-of-life (Johnson, 1955). Also, the ratio between the cranial-caudal length and dorso-ventral diameter may change with age in teleosts; in Atlantic salmon (Salmo salar) this ratio changes as the fish increases in size during the post-smolt stage of development (Fjelldal et al. 2005). Further, the morphology of the vertebral column – the relative size of the individual vertebra – may change between life stages; in Atlantic salmon (Salmo salar), the caudal vertebrae grow relatively faster than the trunk vertebrae during parr-smolt transformation (Fjelldal et al. 2006). Although species differences in the number of vertebrae have been studied in a wide range of species within the order Gadiformes (Fahay & Markle, 1984), there are no single within-species studies on how the shape or the relative size of the vertebrae changes during ontogeny in this order.
The coastal Atlantic cod (Gadus morhua L.), like many other marine teleosts, has a complex life history encompassing larval, juvenile and adult phases that differ in their morphology, physiology and behaviour – each stage being adapted to a specific habitat and source of food. Eggs and larvae inhabit the surface waters of the pelagic zone; the embryo grows curved around the yolk-sac, and at hatching the body axis straightens. Towards the end of the yolk-sac stage, when the yolk is almost consumed, the larvae start to swim to catch plankton and escape predation. Later, the juveniles, which display the characters of the adult (Colton & Marak, 1969), seek shallow water, near the tidal zone. As they grow, reaching a length of about 10 cm, they gradually move to deeper waters, where they remain relatively stationary until they mature at an age of 2 years or more (Salvanes, 2001). However, it remains to be elucidated whether the morphology of the vertebral column changes with age and size in Atlantic cod, as well as in other marine teleosts. Here, our hypothesis was that the development of the vertebral column plays a key role by passing through a sequence of specializations. To test this, we studied the cranial-caudal length to dorso-ventral and lateral-lateral diameter ratios, and the relative size of vertebrae in cod with an age ranging from 0 to 6 years sampled in a fjord in Western Norway. We address the growth dynamics and functional anatomy of the backbone, and discuss our results in relation to morphological adaptations to habitat and sexual maturation. In addition to elucidating functional anatomy, basic knowledge on how differential growth determines the morphology of specific structures of the vertebrae throughout the axis of the backbone may contribute to the understanding of pathologies, especially the deformities of farmed cod.
Materials and methods
Culture conditions, sampling, and bone staining of larvae and juveniles
Larvae and juveniles of Atlantic cod (Gadus morhua L.) were collected from four different land-based aquacultural production facilities. All the facilities kept wild-caught broodstock, and artificial light regimes were employed to induce spawning at different times of the year; the fish spawned naturally in the tanks. The eggs were collected from the outlet water and kept in small aerated coned containers where hatching took place. Thereafter, the larvae were transferred to 30-m3 indoor start-feeding tanks. Each facility employed different fish densities, feeding protocols and temperature regimes. A series of developmental stages was analysed from each facility. Samples were taken at 2-day intervals from the day on which the larvae hatched, until 90 days post-hatching (dph). Each sample consisted of 30 fish collected at random from different depths in the tanks. To study the development of bony structures, 20 fish from each sample were processed as whole specimens and stained with Alizarin after a modified protocol of Taylor & van Dyke (1985). Before staining, the specimens were fixed for at least 48 h in 4% formaldehyde in phosphate buffer (pH 7.4). Depending on their size, the fish were immersed for up to 5 days in a buffered enzyme solution that contained 0.4 mg mL−1 trypsin (Sigma Chemical Co., London, UK), 0.4 mg mL−1 proteinase K (E. Merck, Darmstadt, Germany), 0.4 mg mL−1 pronase (E. Merck) and 0.4 mg mL−1 lipase (Sigma Chemical Co.). Specimens were then washed in buffer and bleached in 0.3% H2O2 in 1% KOH for 1–4 h. The bone matrix was then stained with 0.1 mg mL−1 Alizarin red S (CI 58005) in 1% KOH for 24–48 h, and then dehydrated by passage through a series of baths with increasing glycerol concentration, and stored on glycerol. The larvae were photographed using a Sony DXC-S500 camera (Sony Inc., Tokyo, Japan) on a Leica MZ16 A stereomicroscope (Leica Microsystems Ltd., Heerbrugg, Switzerland), and drawn.
Sampling of wild cod
Eighteen wild cod (~ 50 cm) were caught in cod traps in Matredal (Norway). Fifteen of these were used to establish an anatomical basis for regionalization of the vertebral column, and three were employed for acquiring measurements needed for the calculation of the angle between vertebrae.
A further 45 wild cod of various sizes were caught using different types of fishing gear: the smallest with a beach seine, those of medium size with eel traps, and the largest in cod traps. The fish were alive at the time of sampling, but were euthanized by an overdose of anaesthetic (400 mg L−1, Finquel® vet., ScanAqua AS, Årnes, Norway). Prior to analyses, their length was measured. For determination of age, otoliths were dissected and examined (Rollefsen, 1933), and the fish were accordingly divided into seven age groups (Table 1). Fish that were younger than 1 year were classified as ‘0 year’. To assess age-related changes in vertebra morphology, vertebral columns were dissected for measurement of vertebra centrum morphology.
Table 1. Total length (mean ± SE) of the fish employed to assess age related changes in vertebra morphology
Total length (cm)
0 – September
8.0 (± 0.3)
0 – October
16.2 (± 0.6)
22.8 (± 2.1)
35.3 (± 1.6)
49.0 (± 2.2)
51.0 (± 3.5)
76.7 (± 4.8)
The fish utilized for anatomical characterizations of different regions were warmed in water to 60 °C for approximately 2 h, after which all the soft tissue was removed carefully using tweezers and brushes under a binocular microscope. The vertebrae were separated, and the anatomy of structures such as neural and haemal arches, parapophyses and ribs were described. The vertebrae were photographed and drawn.
The fish used to acquire parameters for calculations of the angles between vertebrae at full flexion were, immediately after euthanasia, radiographed in the lateral and horizontal planes. For the measurement of intervertebral distances, vertebra centrum cranial-caudal lengths and lateral-lateral diameters, multiple radiographs were taken along the cranial-caudal axis of each fish to obtain images where the angle between the x-ray source was perpendicular to the vertebrae to be measured; therefore, only distances between vertebrae in the mid region of each image were measured. A portable x-ray apparatus (HI-Ray 100; Eickenmeyer Medizintechnik für Tierärzte e.K., Tuttlingen, Germany) and 30 × 40 cm film (Fujifilm IX 50; Fujifilm Corp., Tokyo, Japan) was employed. The film was exposed twice at 50 mAs and 40 kV (distance x-ray apparatus – film, 100 cm), and developed using a manual developer [Cofar Cemat C56D; Arcore (MI), Italy] with Kodak Professional manual fixer and developer (Kodak S.A., Paris, France). The radiographs were digitalized by scanning (Epson Expression 10000 XL; Seiko Epson Corp., Nagano-Ken, Japan). The cranial-caudal length, lateral-lateral diameter of each vertebra centrum, in addition to the intervertebral distance, was measured by means of image analysis software (image-pro plus, version 4.0; Media Cybernetics, Silver Spring, MD, USA). The datasets were manually smoothened to avoid noise. The angle between vertebrae at full flexion was calculated accordingly (see Fig. 6A for details on abbreviations and different measurements):
Cos CӨ = (2D2 − 2d2)/(2 * D2), where D represent the ‘vertebral lateral-lateral diameter’ of the largest of two adjacent vertebral end-plates, Ө the ‘angle at full flexion’ between the two vertebrae, and 2d the ‘intervertebral distance’ between them at full flexion.
For the evaluation of age-related changes in morphology, the cranial-caudal length, dorso-ventral and lateral-lateral diameters of each vertebra were measured by means of digital callipers under a binocular microscope. In most of the fish, the number of vertebrae was 52. To compare fish of different lengths, measurements of vertebra length (Ls) were standardized: the cranial-caudal length of each vertebra was divided by the sum of vertebrae cranial-caudal lengths of the fish (). To standardize measurements referring to shape of the vertebra centra in different regions and between fish sizes (ages), measurements of vertebra dorso-ventral (DDs) and lateral-lateral (LDs) diameter were standardized: the diameters of each vertebra were divided by its cranial-caudal length (DDs = DDv/Lv, LDs = LDv/Lv). The shape of the vertebra cross-section (SCS) was measured as the dorso-ventral diameter divided by the lateral-lateral diameter (SCS = DDv/LDv).
Due to the large differences in the number of individuals in each age-group, a non-parametric method was employed to test the differences in vertebra morphology. Differences in Ls, DDs, LDs and SCS within each region, and between different age-groups, were tested with a one-way Kruskal–Wallis test.
The statistical analyses of Ls were based on the length of each region, i.e. length of region 2 was measured as the sum of the Ls of all the vertebrae in region 2 (∑V3−19). Differences in DDs, LDs and SCS within each vertebra region, and between different age groups, were tested with a one-way Kruskal–Wallis test. The statistical analyses were based on mean values within each region. P < 0.05 was considered a statistical significant difference.
The first structure to form in the vertebral column was the bone of the foremost neural arch at 21 dph – 4 mm total length (Fig. 1A). Thereafter, the remaining neural arches were formed, successively in a cranial to caudal direction (Fig. 1B). The first chordacentrum to mineralize was no. 3 at 28 dph – 6 mm total length. Thereafter, the chordacentra mineralized in a regular sequence from here towards the head and tail fin (Fig. 1C,D). All the vertebrae were mineralized 50 dph.
Regionalization and modulated flexibility
The vertebral column (Fig. 2) comprised 52 (± 2) vertebrae. Based on morphology of the individual vertebrae, the backbone was divided into four distinct regions: the cervical region (neck) comprised V1 and V2 (two vertebrae). Here, the vertebra centra were short, ribs were absent, whereas neural spines were prominent (Fig 3A,B). V1 was closely attached to the occipital bone. The abdominal region (trunk) comprised V3–V19 (17 vertebrae). Here, all the vertebrae were articulated with ribs – the ribs being attached to broad wing-shaped transverse processes (parapophyses) (Fig 3C–H). The caudal region (tail) comprised V20–V40 (20 vertebrae). These had fused haemal arches with prominent haemal spines (Fig 4A–J). Finally, the ural region (caudal fin) comprised the remaining vertebrae: V41 and onwards. The number of vertebrae in this region varied between 12 and 15, accounting for the variation in total number. The neural and haemal spines in this region were broader in the medial plane than those of other regions, providing a site of origin of muscles inserting on the lepidotrichs (fin-rays) of the caudal fin (Fig 5A–D).
A schematic drawing of a vertebral column superimposed on a graph that depicts calculated full flexion is shown in Fig. 6C. Calculations of the theoretical maximum angles between vertebrae – the column at full flexion (Fig. 6A,B) – yielded a vertebral column with a circular shape in the cervical through the caudal region (Fig. 6C). Thereafter, in the ural region, the flexibility between vertebrae increases, the column attaining a spiral-like geometry (Fig. 6C).
Morphology of the vertebra centra
In the cervical region (V1–V2) of all year classes, the standard vertebra cranial-caudal length (Ls) of V2 was shorter than that of V1 (Fig. 7A). For the youngest fish (0 year), Ls increased gradually in the caudal direction throughout the abdominal region (V3–V19), and decreased successively throughout the caudal and ural regions (V20 and onward). With increasing age, Ls increased significantly (Kruskal–Wallis, P < 0.05) in the cervical and abdominal regions, and decreased significantly in the caudal and ural regions (Fig. 7A). Figure 7B illustrates schematically how this change in vertebra lengths may affect body proportions.
Standard dorso-ventral (DDs) and lateral-lateral (LDs) diameters were large in the cervical region, decreasing rapidly along the cranial portion of the abdominal region, to become constant towards the tail (Fig. 8A,B). Furthermore, with age, DDs and LDs increased significantly (Kruskal–Wallis, P < 0.05) (Fig. 8A,B).
The cross-section of the vertebra centra (SCS) had an oval shape. However, the axis of the oval differed between regions; generally, DD was lower than LD (SCS < 1) in vertebrae nos. 1–3, while thereafter DD was greater than LD (SCS > 1) (Fig. 8C). Only the oldest fish (age 6 years) exhibited a different trend; here, on average, 28 vertebrae caudally to no. 3 had SCS < 1 (Fig. 8C). Most of these vertebrae were located caudally in the abdominal region (V11–19) and cranially in the caudal region (V20–29).
This study describes the ontogeny, functional anatomy and regional identity of different vertebra regions, and attempts to relate morphology of the vertebral column to the life history of cod.
Regional patterns in vertebral design
In most teleosts the vertebral column consists of two main parts: the trunk and caudal regions (Zhang, 2009). As in Atlantic salmon (Kacem et al. 1998), rainbow trout (Kacem et al. 2004), zebrafish (Ferreri et al. 2000; Bensimon-Brito et al. 2012) and halibut (Lewis et al. 2004), we have divided the vertebral column of the Atlantic cod into four distinct regions. In zebrafish, however, based on the same characteristics as in the present study, a subdivision into five regions has been proposed (Morin-Kensicki et al. 2002). Here, the additional region consists of 0–2 vertebrae of intermediate morphology, at the junction between the rib-articulated and haemal arch bearing vertebrae; these vertebrae possess both a haemal arch and are articulated with ribs (Morin-Kensicki et al. 2002). Such a transitional region was not found in cod. Indeed, the number of pre-caudal and caudal vertebrae has been determined in a wide range species within the order Gadiformes (Fahay & Markle, 1984); among gadids there is a general increase in total number of vertebrae from phycines (47–55) to gadines (52–58) to lotines (62–66), caused by an increase in number of precaudal vertebrae, except in Brosme brosme (Markle, 1982). The present study is, however, the first to do a more detailed description on the regional identity of vertebra in the order Gadiformes.
Vertebra shape and intervertebral distance infers flexibility
Differences in the design of the vertebrae in specific regions of the column may reflect mechanical demands and modalities of locomotion. Flexion along the backbone relates to shape and distance between vertebrae, and may be predicted through employing simple geometry. Here, the geometric combination of vertebra cranial-caudal lengths, lateral-lateral diameters and intervertebral distances for all regions, except the ural region, yields an even, circular curve for theoretical maximal backbone flexion. The model, however, does not account for potential restrictions on flexibility imposed by anatomical structures such as ligaments, muscle and skin, thus possibly overestimating whole body flexibility.
Cod changes its swimming behaviour from subcarangiform swimming to kick-and-glide to increase swimming speed, accompanied by a transition from steady state aerobic metabolism to anaerobic metabolism (Lurman et al. 2007). The kick-and-glide swimming behaviour consists of cyclic kicks of swimming movements, followed by a glide phase in which the body is kept motionless and straight (Weihs & Webb, 1983; Videler, 1993). During the initial phase of the burst, fast twitch contractions occur, producing high amplitude flexion along the body axis. Whether an even flexibility throughout trunk and tail is advantageous in kick acceleration remains to be elucidated. Cod utilize their caudal fin to brake by setting it perpendicular to the body axis (Videler, 1981); hence the spiral geometry of the theoretical maximum flexion graph may reflect an ability to produce refined caudal fin movements.
In the cervical region, vertebrae are short, have large diameters and the distance between them is small; hence, deflexion in this region is lower than in others. This may serve to stabilize the transition between the cranium and backbone, and may maintain a hydrodynamic body contour, important when propulsive forces act on trunk and tail.
In the present study, we only modulated the flexibility in adult cod. The fact that vertebra shape changes during ontogeny may suggest that the flexibility of the vertebral column also changes. However, this issue needs to be investigated further.
Variation in number of vertebrae is restricted to the ural region
The total number of vertebrae counted in our study is within the range described in earlier reports (Løken & Pedersen, 1996). Among teleost species, variation in the number of vertebrae may reside in different backbone regions: in halibut, variation is found in the cephalic, pre-haemal and caudal regions, but not in the haemal region (corresponds to the caudal region in this study) (Lewis et al. 2004), whereas in zebrafish, variation may be found in all regions except the cervical (Morin-Kensicki et al. 2002). Cod, therefore, appear to display a different pattern of variability in all regions except the ural, encompassing a constant number of vertebrae.
Water temperatures, salinity and oxygen levels during early life stages may affect the number of vertebrae of teleosts (Blaxter, 1969). Such environmental parameters may also have an impact on Atlantic cod, as counts of vertebrae vary between specific natural populations (Swain et al. 2001). Brander (1979) reported that there is an inverse relationship between vertebral number and water temperature during early development in Atlantic cod in the North Atlantic, and also some dependence of temperature and vertebral number between different year classes in the North Sea. Lear & Wells (1984) found that the vertebral averages for age-groups 0 and 1 cod from Notre Dame Bay were inversely correlated with mean temperatures of the 0–100-m water layer at a station in the Labrador Current off eastern Newfoundland. Moreover, Løken & Pedersen (1996) incubated fertilized cod eggs from different stocks at constant temperatures of 2, 4 and 6 °C and found an inverse relationship between temperature and vertebral counts, with the average temperature response being −0.50 vertebrae (oC)−1. Here, we analyzed cod from a restricted geographical region and hence the variability in vertebra number probably reflects that present within a specific population.
The relative length of the vertebrae increased in the cervical and abdominal regions, and decreased in the caudal and ural regions with increasing age in the present study. In the Atlantic salmon, the vertebrae with the relatively largest cranial-caudal lengths and dorso-ventral diameters are found in the caudal region throughout life (Fjelldal et al. 2005, 2006, 2009a). Here the relative cranial-caudal length of the caudal vertebrae increases during smoltification (Fjelldal et al. 2006), which may provide a more slender body morphology (Winans & Nishioka, 1987). In comparison, the regional changes in vertebra growth in cod are more pronounced, which may suggest that cod either change their swimming mode drastically with age, or that the production of large volumes of gametes (roe) during sexual maturation may require a significant increase in the volume of the abdominal cavity. Indeed, in the fjord where the cod in the present study were sampled (Masfjorden, Norway), 20% spawned after 2 years, 52% after 3 years, and 82% after 4 years (Salvanes, 2001). Salvanes & Ulltang (1992) determined the age of 603 Atlantic cod collected in Masfjorden by different fishing techniques; the oldest fish was 8 years, and only eight fish were older than 6 years. Hence, the fish used in this study can be regarded to represent the whole life-cycle for this cod population.
Relevance to fish farming
Vertebra deformities are commonly observed in farmed cod. These deformities typically develop as curvatures within the cervical region (Grotmol et al. 2005) or in the mid-caudal region (Fjelldal et al. 2009b); the high diameter to length ratio, and compact morphology of the cervical vertebrae and relatively large size of the mid-caudal vertebrae in juvenile cod indicates that, during early life stages, these regions receive high mechanical loads. Similarly, curvatures in the caudal region of the vertebral column have been observed in other cultured marine teleosts such as sea bream (Sparus auratus) and sea bass (Dicentrarchus labrax), induced by high swimming activity (Chatain, 1994; Kihara et al. 2002). Similarly, in Atlantic salmon, deformities tend to develop in the vertebrae with the largest cranial-caudal length and dorso-ventral diameter (Fjelldal et al. 2009a), which are found in the caudal region. A thorough understanding of the normal ontogeny and anatomy of the vertebral column of cod, provides a necessary reference, permitting recognition of pathologies.
This study was supported by the Research Council of Norway (no. 159665/S40) and University of Bergen. We thank Britt Sværen Daae, Grethe Thorsheim, Nina Ellingsen, Irene Heggstad and Teresa Cieplinska for expert technical assistance. There are no conflicts of interest among the authors of this manuscript.