Temperature And Myogenesis In Embryos Of The Atlantic Herring Clupea Harengus

The first vertebrate muscle fibres are formed in the rostral somites of the axial musculature; in birds and mammals they are bipolar mononucleated cells which express the same myosin heavy chain (Holtzer et al. 1957). The myoblasts destined to give rise to appendicular muscles migrate from the somites and are probably specified by their position in the somite (see Stockdale, 1992, for a recent review). Families of myoblast types have been identified in birds and mammals that can only be isolated at specific stages of development (embryonic, foetal and adult) and are capable of differentiating into muscle fibres with distinct physiological properties (Stockdale and Miller, 1987; Stockdale, 1992).

Much less is known about muscle development and growth in ectothermic vertebrates. The first muscle fibres in the toad Xenopus laevis are formed from single mononuclear myoblasts which span the somites (Blackshaw and Warner, 1976), and multinuclear muscle cells do not occur until metamorphosis (Muntz, 1975). In zebrafish (Brachydanio rerio), myoblastic cells proliferate following segregation of the somites (van Raamsdonk et al. 1974). Myotube formation begins in a sub-population of the myoblasts lying adjacent and lateral to the notochord which elongate to span the whole somite length (van Raamsdonk et al. 1974; Hanneman, 1992). These primary muscle fibres, sometimes called muscle pioneer cells, express one or more members of the engrailed homeobox gene family (Hatta et al. 1991) and are the first fibres to be innervated by the pioneer motor neurones (Kimmel et al. 1991). Myotubes in the mid-somite are formed by the fusion of another sub-population of myoblasts, to give multinucleated muscle fibres (Waterman, 1969; van Raamsdonk et al. 1974; Hanneman, 1992). In contrast to mammals, new muscle fibres continue to be produced in fish until they reach around 70 % of their ultimate body length (Weatherley and Gill, 1985).

In Atlantic herring (Clupea harengus), the pectoral fin buds, axial muscle fibres and pronephric tubules are formed relatively later in embryos incubated at 5 °C than at higher temperatures (Johnston, 1993). However, fish reared at different temperatures may not hatch at the same developmental stage (Blaxter, 1988). In zebrafish, staging by somite number has been shown more accurately to predict, for example, the development of primary trigeminal sensory neurones in the head (Metcalfe et al. 1990) or primary motor neurones in the trunk (Eisen et al. 1986) than does staging the embryo by elapsed time after fertilisation (Westerfield, 1994). In the present study on herring embryos, the influence of temperature on the timing of muscle development has therefore been investigated in relation to somite stage and the formation of other organs. The early life stages of herring and other marine fish are subject to massive mortality (Heath and MacLachlan, 1987). Variations in larval phenotype attributable to development temperature are therefore of particular interest because of their potential to influence survival and hence recruitment to the adult population.

Embryonic development was studied in spring-spawning Atlantic herring (Clupea harengus L.) in 1993 and 1994. Mature fish, 32–35 cm total length, were caught on the Ballantrae Bank in the Firth of Clyde, Scotland, during March, their gonads dissected at sea and transported on ice to the Dunstaffnage Marine Laboratory, Oban, Scotland. In 1993, eggs from 20 females were fertilised with the milt from eight males and the embryos were incubated in sea water at temperatures of 5, 8 and 12 °C in constant-temperature rooms (temperature range ±0.5 °C). Three different batches of fish were examined in 1994.

In the first batch, caught in early March, the eggs from six females were fertilised with the sperm from five males. The second and third batches, caught in late March and comprising eight females and six males, were fertilised 12 h apart. During the 1994 season, fertilised eggs were incubated in recirculated seawater systems maintained at 5, 8, 12 or 15 °C (range ±0.5 °C). 15 °C was chosen as it is close to the upper temperature for the normal development of spring-spawning herring (I. A. Johnston and R. S. Batty, unpublished results). The unfertilised eggs become sticky on contact with sea water. Eggs were scattered at uniform density onto labelled glass microscope slides and immersed in a bath of sea water containing the sperm for about 10 min. The fertilisation rate was generally 90–95 %. Glass slides were removed from the incubation tanks at intervals and examined in Petri dishes of sea water at the same temperature.

The chorion in herring is transparent, facilitating direct observations of development. By studying batches of eggs fertilised at different times of the day, it was possible to observe living embros at 2–5 h intervals throughout development. Embryos were observed under bright-field and dark-field illumination with a Wild M3 stereo microscope. The number of somites segregating at different stages of development was counted in 12–15 embryos from each batch and incubation temperature. Once spontaneous movements had started, the embryos were removed from the egg capsules and anaesthetised with either a 1:5000 (v/v) solution of benzocaine in sea water or a 1:10 000 (w/v) solution of bicarbonate-buffered MS222 in sea water prior to counting the somites.

For light and transmission electron microscopy studies, 6–10 embryos were sampled in each fixative at the 1, 5, 7–10, 12–15, 20–25, 28–31, 37–38, 40, 42–5, 47–50, 53–55 and 59–61 somite stages. Once somitogenesis was complete, samples were taken approximately every 12 h until 80 % development time (DT; time from fertilisation to 90 % larvae hatching) and then at hatching. The fixatives used for light microscopy were Bouin’s fluid (3–24 h) and neutral buffered formalin (24 h). In order to improve penetration of the fixative, the embryos were removed from the egg capsules. Serial sections were cut at 7 μm thickness both transversely and longitudinally to the long body axis and mounted on gelatin-coated slides. Sections were either stained with Haemotoxylin–Eosin or dewaxed with xylene and used for immunocytochemistry.

Dechorionated embryos were fixed for electron microscopy for 3–24 h in either 0.5 % gluteraldehyde, 0.5 % paraformaldehyde, 1 % sucrose, 2 mmol l⁻¹ CaCl2, 100 mmol l⁻¹ NaCl, 100 mmol l⁻¹ sodium cacodylate, pH 7.4 (at 4 °C), or 2.5 % gluteraldehyde, 2.5 % paraformaldehyde, 1 % sucrose, 2 mmol l⁻¹ CaCl2, 100 mmol l⁻¹ NaCl, 100 mmol l⁻¹ sodium cacodylate, pH 7.4 (at 4 °C). No consistent difference in the quality of tissue preservation was noted for the fixatives. Specimens were washed in the same solutions minus the fixative and post-fixed for 2 h in a solution containing 2 % osmium tetroxide in 100 mmol l⁻¹ sodium cacodylate, pH 7.4 (at 4 °C). Samples were subsequently washed in buffer, dehydrated through alcohol, stained en bloc with uranyl acetate in 70 % ethanol and embedded in Araldite resin. Semi-thin sections (0.5–1.0 μm) were cut on a Reichert OMU2 ultramicrotome and stained with Toluidine Blue in order to determine the orientation of the specimens. Ultrathin sections (50–80 nm thick) were mounted on pyroxyline-coated or uncoated copper grids of 75–300 μm mesh size and stained with aqueous saturated uranyl acetate and Reynold’s lead citrate. In some cases, grids were prestained with 0.5 % potassium permanganate in order to improve contrast. Sections were viewed with a Philips 301 transmission electron microscope at 60 kV.

Glass slides with attached embryos were broken into small pieces containing around 20 individuals. Cell proliferation was investigated by the incorporation of 5-bromo-2′-deoxyuridine (BrdU). BrdU, which is incorporated into replicating DNA, was subsequently localized with a mouse anti-BrdU monoclonal antibody using a commercial kit (Amersham International, UK). The cell-labelling solution contained BrdU and 5-fluoro-2′-deoxyuridine in a 10:1 molar ratio. The latter compound inhibits thymidilate synthetase and increases BrdU incorporation by lowering competition from endogenous thymidine. The labelling solution (2 % v/v in sea water) was injected into the egg capsule using a 10 μm tip diameter microelectrode and a Picospritzer II pressure injection system (General Valve Corporation, New Jersey, USA). The glass pieces were subsequently incubated for 1, 2, 3 and 5 h in a 2 % (v/v) solution of the labelling reagent in sea water. 1-day-old and 3-day-old larvae were transferred to 2 % BrdU labelling reagent in sea water at their respective rearing temperatures and killed by an overdose of anaesthetic after 1–5 h. BrdU-labelled embryos and controls were dechorionated prior to fixation. Embryos and larvae were fixed in Bouin’s fluid and embedded in wax. Sections were dewaxed and DNA-incorporated BrdU was detected with a mouse anti-BrdU primary antibody, which was in turn detected with a peroxidase-conjugated goat anti-mouse secondary antibody. Peroxidase activity was revealed using diaminobenzidine (DAB) in the presence of cobalt and nickel, which resulted in black staining at sites of BrdU incorporation. Sections were counterstained with 0.5 % (w/v) Neutral Red to visualise the unlabelled nuclei.

Embryos were dechorionated and the yolk-sac removed, which usually resulted in skinning of the anterior half of the embryo. In order to improve penetration of substrates further, the embryos were incubated in a solution of 1 % saponin in phosphate-buffered saline (PBS) for 2 h at 4 °C and then washed thoroughly in PBS. Whole embryos were stained for acetylcholinesterase by the direct colouring method of Karnovsky and Roots (1964).

Results from the different temperatures were compared using a Student’s t-test. All data are presented as means ± S.D.

Embryonic development time (DT) was taken as the time from fertilisation to the point at which approximately 90 % of the larvae in each batch had hatched. In 1993, DT was 34 days at 5 °C, 19.5 days at 8 °C and 8 days at 12 °C. The DTs in 1994 were 28 days at 5 °C, 16 days at 8 °C, 9 days at 12 °C and 8 days at 15 °C.

The anterior–posterior body axis was unambiguously established after 150 h at 5 °C, 100 h at 8 °C and 50 h at 12 °C (1993 experiments). The increase in the length of the embryo (longest linear dimension) was not linear with respect to time (Fig. 1) and was greatest following somite segregation (Fig. 2). The average rates of body elongation during somitogenesis were 0.20 mm day⁻¹ at 5 °C and 0.38 mm day⁻¹ at 12 °C, whereas after somitogenesis growth rates were 0.27 mm day⁻¹ at 5 °C and 1.21 mm day⁻¹ at 12 °C. The Q10 for the rate of body elongation over this temperature range was 2.86. At any given somite stage, the embryos reared at different temperatures were of similar length (Fig. 2). The rate of yolk utilisation was substantially faster at 12 °C than 5 °C (Fig. 3); however, when plotted against percentage DT or somite stage, the average yolk volume at the different temperatures fell on the same curve (not illustrated). Around 80 % of the yolk volume was depleted during the embryo stage (Fig. 3).

In 1993, the first somites were formed 151 h post-fertilisation (PFT) at 5 °C and 45 h PFT at 12 °C. Somitogenesis started earlier and was more advanced as a percentage of development time at 5 °C than at 12 °C (not illustrated). In 1994, the first somites were formed 120 h PFT at 5 °C, 69 h PFT at 8 °C and 41 h PFT at 12 and 15 °C (Fig. 4). The rate of somite addition was approximately linear between the 5-and the 55-somite stages (Fig. 4). Somite formation was highly temperature-dependent, with a Q10 of 3.38 between 5 and 15 °C. The time required to form each somite was calculated using regression analysis from the data in Fig. 4 and found to be 176 min at 5 °C, 120 min at 8 °C, 65 min at 12 °C and 52 min at 15 °C. The formation of somites as a function of development time is illustrated in Fig. 5 (only 5 and 15 °C are illustrated for clarity). Somites were formed relatively earlier in embryogenesis at 5 °C than at 15 °C (P<0.001) (Fig. 5). For example, for 32 % DT, there were 41.6±1.9 somites at 5 °C,37.8±2.3 somites at 12 °C and 25.4±2.2 somites at 15 °C (mean ± S.D., N=12–15). At all temperatures, somitogenesis was complete half-way through the embryo period (50 % DT) when there were 61.6±0.9 somites at 5 °C, 60.7±1.0 somites at 8 °C, 60.8±0.9 somites at 12 °C and 62.0±1.0 somites at 15 °C (mean ± S.D., N=12–15 fish at each temperature) (Fig. 5).

Somitogenesis proceeds in a rostral to caudal direction. The somites are formed from undifferentiated cells of the paraxial mesoderm. Somites first begin to segregate after the neural plate and early notochord are formed (Fig. 6A). A sagittal semi-thin section taken at the level of the segregating somite front in a 10-somite embryo is shown in Fig. 6B. The dermatome does not segment, but remains as a continuous sheet of cells overlying the segregating somites. The dermatome and myotomal layers of the myomeres were clearly separated, with no close membrane contacts visible at the electron microscope level (not illustrated). Close membrane contacts were found between cells in the presumptive myotome region, and sometimes densely stained material was observed in the gaps between opposing membranes (not illustrated). At the boundary of the segmenting somites, the myomeres were sausage-shaped in sagittal section and contained tightly packed cells. These presumptive myoblastic cells contained a large central nucleus and numerous mitochondria. Mitotic bodies were frequently observed, indicating that the cells were actively dividing (Fig. 6A,B).

Myotubes also formed in a rostral to caudal direction and were first observed in 17-to 23-somite embryos (Fig. 6C,D). The somites took on a V-shaped appearance in sagittal section around 5–6 myomeres behind the segregating front of somites, and the presumptive myoblastic cells became more spindle-shaped, with their long axis aligned parallel to the notochord (Fig. 6D). At the point at which the somites became V-shaped, 2–5 cells spanned the somites (Fig. 6D). The nuclei at the posterior boundary were aligned towards the membrane and stained darker than the other nuclei in the somite (Fig. 6D). Adjacent and lateral to the notochord there was a small number of myoblasts (2–6), which elongated to span the entire somite width and were among the first to differentiate. These mononuclear cells, which divide the ventral and dorsal halves of the myomere (Fig. 6E), correspond to the muscle pioneer fibres described in zebrafish (Hanneman, 1992). The majority of muscle fibres were formed by the fusion of 2–5 spindle-shaped myoblasts (Fig. 6E). Early myotubes contained numerous mitochondria and large accumulations of electrondense granules. BrdU labelling experiments with 20-, 30-, 40, 50- and 60-somite embryos showed that, following fusion, the myoblasts left the cell cycle and that the myotube nuclei were no longer capable of dividing (a 60-somite embryo is illustrated in Fig. 6F). Close membrane contacts were occasionally observed between myotubes (not illustrated).

The myotubes contained large accumulations of mitochondria (Fig. 7A). For most of the length of the inter-somite border there was an appreciable gap between myotubes in adjacent myomeres (Fig. 7B), with just occasional regions where the opposing membranes came into close contact (Fig. 7C). The membrane gaps often contained densely staining material, and in some cases a multi-layered structure was evident as described for ‘gap’ junctions (McNutt and Weinstein, 1974). The anchoring of the myofibrils to the end of the myotube is illustrated for a 55-somite embryo in Fig. 7B;note that, at this stage, the myoseptal connective tissue matrix has not developed. Since myogenesis proceeds in a rostral to caudal direction, all stages of myogenesis are present simultaneously in 50-to 60-somite embryos, i.e. myotubes without contractile filaments, myotubes containing myofibrils and immature muscle fibres.

In order to investigate the effects of temperature on the relative timing of myogenesis, sections through somites 6–10 were examined at each developmental stage. Contractile filaments were first evident in myotubes at the 42-somite stage at 5 °C (Fig. 8A). Electron-dense material of the Z-line and thin filaments were observed scattered around the periphery of the myotubes (Fig. 8A). Sarcomeres containing both thin (actin) and thick (myosin) filaments were present from the earliest stages. Contractile filaments were present at earlier somite stages at higher temperatures: at the 38-somite stage at 8 °C and at the 27-somite stage at 12 °C (Fig. 8B). Contractile proteins were first synthesised in the axial muscles after the formation of the fin folds at 5 °C (Fig. 8C), but before these structures appeared at 8 and 12 °C (Fig. 8D,E). Complete myofibrils were observed to span the myomere at the 54-to 55-somite stage at 5 °C, the 42-to 43-somite stage at 8 °C and the 35-to 38-somite stage at 12 °C (see Fig. 14). Slow rhythmic contractions of the anterior part of the body, indicating the presence of functional muscle fibres, were observed at approximately the 55-, 45-and 40-somite stages at 5, 8 and 12 °C, respectively. Distinct superficial and inner muscle fibre types were apparent in transverse sections at the 62-somite stage at 5 °C, at the 48-somite stage at 8 °C and at the 40-somite stage at 12 °C (Fig. 9).

The development of muscle endplates was studied at 5 °C (Fig. 10). At 5 °C, a generalised staining for acetylcholinesterase activity was first observed in the somites towards the head at the 48-to 50-somite stage (Fig. 10A), progressing further down the body and intensifying in 52-to 54-somite embryos (Fig. 10B). Bands of staining at the myoseptal boundaries, corresponding to muscle endplates, were observed in the first 12–13 somites at the 55-somite stage at 5 °C (Fig. 10C). Around 27 myotomes had endplates by the end of somitogenesis, 286 h PFT at 5 °C, and 46 myotomes had endplates at 334 h PFT (61-somite stage, Fig. 10D). Endplates were progressively formed in myotomes towards the tail at the rate of 8–10 somites per day, such that all segments had bands of acetylcholinesterase staining after 360–370 h PFT at 5 °C.

At hatching, the larvae were observed to swim continuously to avoid sinking. Two muscle fibre types could be distinguished in the myotomes using ultrastructural criteria. Immediately beneath the skin there was a single layer of fibres with smaller diameters containing higher volume densities of mitochondria and lower volume densities of myofibrils than the underlying muscle mass (see Fig. 11 in Johnston, 1993). The myotomes just behind the yolk-sac contained around 100 superficial muscle fibres and 200–300 inner muscle fibres (based on 1993 experiments) (Figs 11, 12). The total cross-sectional area of superficial muscle was 34 % greater in larvae reared at 5 °C than at 8 °C (P<0.001) as a result of a 36 % higher average fibre cross-sectional area (P<0.005). There was no significant difference in the number of superficial fibres at the the different rearing temperatures (Fig. 11). In contrast, there were, respectively, 24 % and 43 % more inner muscle fibres at 8 °C and 12 °C than at 5 °C (P<0.02) (Fig. 12). At 5 °C, the average cross-sectional area of inner muscle fibres was almost double that at either 8 or 12 °C (P<0.0001). Since there were fewer, but larger, muscle fibres at 5 °C than at 12 °C, the total cross-sectional area of inner muscle was not significantly different at the two temperatures (Fig. 12). The total cross-sectional area of inner muscle was, however, 39 % less at 8 °C than at 5 °C (P<0.002) (Fig. 12).

At hatching, labelled nuclei were observed in the myotomes, associated with the myosepta and with a population of small-diameter cells situated between the muscle fibres (Fig. 13A,B). The cells between the muscle fibres almost certainly correspond to the presumptive satellite cells observed at the electron microscope level (Fig. 13C,D).

The eye and otic vesicle were first apparent in living embryos and histological sections at the 9-to 12-somite stage at 8 and 12 °C and at the 13-to 15-somite stage at 5 °C (Fig. 14). The gut was evident in histological sections at the 15-somite stage at 8 °C and at around the 22-somite stage at 5 ° and 12 °C (Fig. 14). The heart and presumptive dorsal aorta were observed at the 38-to 40-somite stage at all temperatures (Fig. 14). The heart began to beat spontaneously at around the 50-somite stage at 5 °C and at the 45-somite stage at 12 °C (Fig. 14). The dorsal and ventral fin folds were visible at the 38-to 40-somite stage at all temperatures, whereas the pectoral fin buds appeared in the order 8 °C before 12 °C before 5 °C (Fig. 14). The pronephric tubules showed the most marked change in relative timing of appearance of all the tissues examined, occurring after the completion of somitogenesis at 5 °C (464 h PFT), at the 47-to 50-somite stage at 8 °C and at the 40-somite stage at 12 °C (Fig. 14). Pronephric tubules were formed at about the same time or shortly after the differentiation of distinct muscle fibre types (Figs 9, 14). The appearance of the notochord and spinal cord at the somite stage at which pronephric tubules and distinct fibre types were first observed differed markedly between the different temperatures (Fig. 9). At 5 °C, the notochord was significantly larger and the spinal cord more highly differentiated than at 8 and 12 °C. The eyes became fully pigmented shortly before hatching at all temperatures.

Some asynchrony was apparent in the development of herring embryos, and at any particular time after fertilisation not all embryos were at the same stage. A similar asynchrony is also found in the development of embryos derived from the spontaneous spawning of a single clonal strain of zebrafish (Streisinger et al. 1981), and is therefore not entirely due to genetic variation. In the present study using somite stage as the reference point, we have shown that temperature affects the relative timing of the appearance of some morphological characters but not others (Fig. 14). The formation of the pectoral fin buds, pronephros, initial synthesis of contractile proteins and the differentiation of distinct muscle fibre types occurred at different somite stages at 5, 8 and 12 °C (Fig. 14). Somite formation itself was found to start relatively earlier in embryogenesis with respect to hatching as temperature was increased from 5 to 12 °C (Fig. 5). There are other reports of changes in the relative timing of development in fish at both the morphological (Fukuhara, 1990) and molecular (Crockford and Johnston, 1993) levels. For example, the sequence of pectoral fin formation, mouth opening and pigmentation of the eye has been shown to vary with environmental temperature in several tropical fish species (Fukuhara, 1990). Crockford and Johnston (1993) found that, at hatching, herring larvae reared at 5 °C expressed only embryonic troponin T (TNT) isoforms, whereas the muscle fibres from larvae reared at 10 and 15 °C contained a mixture of embryonic and larval TNT isoforms. Although the number of isoforms of TNT declined throughout the yolk-sac stage, the inner muscle fibres of larvae had a unique combination of myofibrillar proteins at each temperature (Crockford and Johnston, 1993).

In herring and other fish, it has long been recognised that meristic characters, such as myotome counts and number of vertebrae, vary with environmental factors, particularly temperature and salinity (Tåning, 1952; Hempel and Blaxter, 1961). Temperature also influences muscle cellularity in herring (Figs 11, 12). For example, at 12 °C, there were 43 % more inner muscle fibres of smaller average cross-sectional area that at 5 °C (Fig. 12). Broadly similar results were obtained in experiments conducted in spring 1991; the number of muscle fibres in myotomes immediately posterior to the yolk-sac was 311±41 at 15 °C, 257±22 at 10 °C and 187±22 at 5 °C (mean ± S.D., 12–15 fish at each temperature) (Vieira and Johnston, 1992). Since the average diameter of the fibres decreased with increasing temperature, the total cross-sectional area of the myotome was independent of temperature (Vieira and Johnston, 1992). In contrast, in the present study the total cross-sectional area of myotomal muscle increased in the series 5 °C>12 °C>8 °C (Fig. 12). The influence of temperature on muscle cellularity is somewhat variable even for the same stock and species (Johnston, 1993). This variability may reflect genetic variation and/or differences in the spawning condition of the parents which could result in changes in the concentrations of mRNAs, growth factors or hormones in the egg yolk. Temperature has also been shown to influence muscle cellularity in the larval stages of Atlantic salmon Salmo salar (Stickland et al. 1988) and plaice Pleuronectes platessa (Brooks and Johnston, 1993). Altantic salmon larvae had fewer, but larger-diameter, muscle fibres at hatching at 10 °C than at 1.6 °C, with lesser differences at earlier stages of development (Stickland et al. 1988).

In the present study, evidence was obtained for the presence of several classes of myoblast in the herring embryo. First, a relatively small number of myoblasts in each myotome elongated to span the whole somite width, giving rise to mononuclear muscle fibres (Fig. 6E). Cells with similar morphological characteristics and behaviour have been identified in zebrafish (van Raamsdonk et al. 1974; Hanneman, 1992) and are distinguished by the transitory expression of one or more members of the engrailed homeobox gene families (Hatta et al. 1991). These cells are thought to have a role in guiding the growth cones of the pioneer motor neurones and in patterning the characteristic chevron shape of the myotome (Kimmel et al. 1991). A second class of myoblasts gave rise to multinucleated myotubes (Fig. 6C,D). Cell culture studies are required to determine whether the myoblasts giving rise to the superficial and inner muscle fibre types in herring have different lineages. HNK-1 and tubulin antibody staining of herring embryos revealed that the motor root for each somite runs along the caudal myoseptum and that outgrowth of secondary motor neurones does not occur until the majority of the myoblasts in the somite have fused to form myotubes (J. Hill and I. A. Johnston, unpublished results). The formation of synapses is dependent on a complex pattern of signalling between motor neurones and muscle fibres (Pike et al. 1992; Sepich et al. 1994). Acetylcholinesterase activity, consistent with functional motor endplates, is not observed in the rostral myotomes until the 53-to 55-somite stage in herring embryos at 5 °C (Fig. 10), at about the same time that spontaneous movements were first observed. In the present study, BrdU labelling experiments have shown that just prior to hatching another population (class) of somitic myoblasts starts to divide on the surface of the muscle fibres (Fig. 13A,B). BrdU-labelled cells are scattered throughout the myotome and almost certainly correspond to the presumptive satellite cells observed at the electron microscope level (Fig. 13C). The pattern of myoblast proliferation in herring contrasts with that observed during the larval stages of sea bass Dicentrarchus labrax (Veggetti et al. 1990), plaice Pleuronectes platessa (Brooks and Johnston, 1993) and turbot Scophthalmus maximus (S. Gibson and I. A. Johnston, in preparation). In these species, new muscle fibres are initially formed in distinct germinal zones at the lateral apices of the myotomes, and only later in the early juvenile stage, when these zones are exhausted, is fibre hyperplasia observed throughout the myotome. BrdU labelling was also observed at the myosepta in yolk-sac herring larvae (Fig. 13B). These cells are probably fibroblasts and cell types associated with the formation of the myosepta, which presumably requires strengthening as the volume density of myofibrils increases and the larvae become free-swimming. New muscle fibre production in herring larvae and the juvenile stages of other species is reminiscent of that described for mammalian limb muscles. The fusion of embryonic myoblasts in the limb bud gives rise to primary myotubes which run from tendon to tendon of the presumptive muscle (Dennis et al. 1981). The primary myotubes act as a scaffold for the formation of secondary myotubes (Duxton et al. 1989). In small mammals, 5–10 secondary myotubes appear over the course of several days (Duxton et al. 1989). They initially form in the vicinity of the primary myotube endplate by fusion of two mononuclear myoblasts and subsequently grow along its surface as more nuclei are added by the fusion of additional myoblasts (Duxton et al. 1989; Wilson et al. 1992). The ratio of secondary to primary myotubes is an order of magnitude greater in large mammals; for example, 70:1 for sheep tibialis muscle (Wilson et al. 1992). In fish, new fibres are formed in close contact with mature fibres which were formed at an earlier developmental stage, giving the muscle a mosaic appearance (Weatherley and Gill, 1985; Veggetti et al. 1990).

Initially, the larval myoblasts in herring are not incorporated within the basal lamina of muscle fibres (Fig. 13C). The basal lamina does not form until later in development, for example, only becoming apparent 41 days after hatching in sea bass (Veggetti et al. 1990). In 1-day-old herring, the number of larval myoblasts per mm² muscle fibre cross-sectional area was found to vary by around twofold between different rearing temperatures (Johnston, 1993). Recent experiments have shown that herring embryos incubated at 5 °C produce more muscle fibres than embryos reared at 12 °C following transfer to a common temperature (I. A. Johnston, V. L. A. Vieira and L. Palmer, unpublished results). These results indicate that post-larval growth characteristics are determined, at least in part, by the thermal experience of the embryo, perhaps via altered cellularity either of muscle stem cells and/or cells of the neuroendocrine system involved in growth regulation.

The early life stages of herring and other fish species are associated with massive mortality of up to 20 % day⁻¹, largely due to starvation and predation (Heath and MacLachlan, 1987; McGurk, 1984). A major finding of this paper and our earlier studies (Johnston, 1993; Crockford and Johnston, 1993) is that changes in environmental temperature of only a few degrees celsius result in considerable morphological and biochemical variation in the larval stages. Whilst the significance, if any, of much of this variation remains unknown, it does offer the possibility of differential physiological performance between groups of larvae, for example in growth rate and/or predator avoidance. By such mechanisms, the temperature experienced during early development could have a significant impact on larval survival and hence recruitment to the adult population.

This paper is a contribution from the Inter-Universities Marine Research Initiative (IMRI) on the effects of temperature change in marine fish involving the NERC Dunstaffnage Marine Laboratory (DML) and the Universities of St Andrews, Stirling and Dundee. We thank the Marine Science and Technology Board of the Natural Environment Research Council for financial support and Dr A. Bullock, Dr

R. S. Batty and other staff at DML for their hospitality. We are also grateful for the expert technical assistance of Mr Ron Stuart and Mr Irvine Davidson. V.V.L.A.’s visits to Scotland were supported by CN Pq (Brasil) in 1993 and the Scottish Association of Marine Science in 1994.