Experimental evidence of chemokinesis in newly hatched cod larvae (Gadus morhua L.)

The swimming behaviour of laboratory-reared newly hatched cod larvae (Gadus morhua L.) was observed in a control solution of artificial seawater and in seven solutions, each with a different concentration of arginine (10⁹ to 10^-3 M). The behaviour of 20 larvae was analysed in each of the eight solutions; the individual observation time was 1 min. Individual movements were recorded on video and analyzed using a computer-assisted program. The larvae swam in straight lines (a trajectory), rested, moved and started swimming again. For the parameters analyzed, i.e., number of movements, angle between successive trajectories and straightness index, there was no significant difference between the behaviour of the larvae in the different solutions. However, for the larvae in 10^-5, 10^-4 and 10^-3 M arginine solutions, the analyzed parameters, i.e., time active, frequency of trajectories (number of movements exceeding body length), distance swum min^-1, length of individual trajectories and trajectory velocity, were all significantly lower than for the larvae in the control solution of artificial seawater and for larvae in the solutions of 10^-9, 10^-8, 10^-7 and 10^-6 M arginine. The results show that the mean distance swum by cod larvae min^-1 was two to five times longer in artificial seawater without arginine and in the four lower concentrations of arginine than in the three higher concentrations. Scanning micrographs show that newly hatched (pre-feeding) cod larvae possess olfactory organs. It seems reasonalbe to assume that the observed changes in swimming behaviour are mediated by the olfactory sense and are important in the feeding strategy of cod larvae. We suggest that the observed behaviour increases the probability of the larvae localizing patches of prey organisms and remaining in the patch once they have found it. The results show that chemokinesis is a mechanism by which the spatial distribution of fish larvae will be correlated with their prey.     

Effect of amino acids on the swimming activity of newly hatched turbot larvae (Scophthalmus maximus)

The swimming behaviour of newly hatched turbot (Scophthalmus maximus L.) larvae was observed in artificial seawater (ASW) and in solutions of 21 l-amino acids at a concentration of 10⁻⁵  M. The behaviour of 20 larvae was analysed in each solution. Each larva was observed for 1 min. Individual movements were recorded on video and analysed using a computer-assisted program. The larvae swam in convoluted, randomised three-dimensional paths, rested and started swimming again. There were large variations in the swimming behaviour of turbot larvae during ontogeny. In ASW the mean frequency of trajectories longer than a body length of 4 mm larva⁻¹ min⁻¹ increased from 1.2 at Day 1, to 10 at Day 4. Analysing the data (Dunnett's method) revealed that the frequency of swimming trajectories increased in the presence of glycine, histidine and glutamine, and decreased in the presence of proline. The total distance swum increased for glycine but decreased for proline. The threshold concentration for glycine detected by turbot larvae was 10⁻⁵  M. The straightness index did not change in the presence of the amino acids. The possible role of these changes in behaviour is discussed. Received: 12 June 1997 / Accepted: 13 January 1998     

Life in a warm deep sea: routine activity and burst swimming performance of the shrimp<Emphasis Type="Italic"> Acanthephyra eximia</Emphasis> in the abyssal Mediterranean

Measurements of routine swimming speed, tail-flip escape responses, and oxygen consumptions were made of the deep-sea shrimp Acanthephyra eximia using autonomous landers in the Rhodos Basin at depths of up to 4,400 m and temperatures of 13–14.5°C. Routine swimming speeds at 4,200 m averaged 0.18 m s^–1 or 3.09 body lengths s^–1, approximately double those of functionally similar oceanic scavengers. During escape responses peak accelerations of 23 m s^–2 or 630.6 body lengths s^–2 were recorded, with animals reaching speeds of 1.61 m s^–1 or 34.8 body lengths s^–2. When compared to shallow-water decapods at similar temperatures these values are low for a lightly calcified shrimp such as A. eximia despite a maximum muscle mass specific power output of 90.0 W kg^–1. A preliminary oxygen consumption measurement indicated similar rates to those of oceanic crustacean scavengers and shallower-living Mediterranean crustaceans once size and temperature had been taken into account. These animals appear to have high routine swimming speeds but low burst muscle performances. This suite of traits can be accounted for by high competition for limited resources in the eastern Mediterranean, but low selective pressure for burst swimming due to reductions in predator pressure.Communicated by J.P. Thorpe, Port Erin     

Relationship of oxygen consumption to swimming speed in Euphausia pacifica

Swimming efficiency (the ratio of thrust power required to overcome hydrodynamic drag to net metabolic energy expenditure) was calculated for the vertically migrating euphausiid Euphausia pacifica swimming at speeds of 1–20 cm s^–1 and at temperatures of 8° and 12°C. Efficiencies ranged from 0.014 to 2.8% at 8°C and 0.009 to 1.69% at 12°C. A comparison with efficiency in fishes 2–3 orders of magnitude larger in weight (efficiency range 10–25%) indicates that locomotion in E. pacifica is far less efficient, a probable result of the organism's small size (x=33.5 mg WW) and multiple-paddle mode of propulsion. Net cost of transport of E. pacifica is three to six times the cost of a hypothetical value for sockeye salmon. Low swimming efficiencies in zooplankton such as E. pacifica are responsible for the underestimation of zooplankton swimming costs. Multiple-paddle propulsion is less efficient than the undulatory mode of fishes.     

Mating behavior of the marine copepodOithona davisae

Oithona davisae Ferrari and Orsi were collected from the innermost region of Tokyo Bay, Japan, in 1980 and 1981. The mating behavior of this marine cyclopoid copepod consists of several steps, starting with the paddling of the male in random search of a mate. This behavior is followed by a spiraling movement in pursuance of a mate. Next is the copulatory grasp during which the male grasps the fourth swimming legs of the female partner by means of his first antennae. Whilst in this position, the male's urosome vibrates to allow the spermatophores to extrude from his genital openings. The mating behavior then culminates in the spermatophore transfer. Males do not grasp the uro-some or caudal setae of the mating partner before proper copulatory grasp. The specialized setae of the female's fourth legs may help the male to grasp her legs firmly. Spiraling occurs when the male approaches or traverses the trail of a female that is ready to copulate and that presumably emits a sex-attractant pheromone. The turning radius reduces gradually from more than 1 mm to ca. 0.25 mm as the male approaches the mate female. Females may register spiraling as a mate (male)-approaching signal. Spiraling may lead the male to locate a pheromone source more accurately, and to promote diffusion of the pheromone to prevent other males from pursuing the source. This swimming strategy can increase the copulatory chance of mature virgin females.     

Respiration and excretion by oceanic salps

Rates of oxygen consumption and ammonium nitrogen excretion were measured on the solitary and/or aggregate generations of ten species of oceanic salps collected by SCUBA divers during cruises in the Atlantic Ocean (1982–1985). Species that were visibly more active had higher metabolic rates than did less active species. Rates were 1.5 to 2 times lower and O:N ratios were lower when salps were held before incubation than when incubation began at the time of collection. Respiration rate showed a better relationship to length than to weight, suggesting that metabolic activity may be connected mainly with swimming. O:N ratios were between 13 and 28 for most species and generations, but higher and more variable in Pegea spp. Exretion of urea was low or undetectable. Rates of metabolic demand (turnover) ranged from 9.7 to 99% body carbon d^-1 and 6.4 to 55.6% body nitrogen d^-1.Contribution No. 5988 from the Woods Hole Oceanographic Institution and No. 412 from the Allan Hancock Foundation     

Using time-depth-light recorders to measure light levels experienced by a diving marine mammal

Acoustic telemetry was used to track vertical and horizontal movement patterns and to monitor the stomach temperatures of seven juvenile shortfin mako sharks (Isurus oxyrinchus Rafinesque) in the Southern California Bight from July to November 2002. Makos (80–145 cm fork length, FL) were attracted to the tracking vessel, where they were fed a mackerel containing an acoustic transmitter that reported temperature and pressure. Tracks ranged from 6.8–45.4 h. Collectively, the mako sharks spent 80% of the track record at 0–12 m, 15% at 12–24 m, and 5% at depths >24 m. The average horizontal swimming speed was 2.3 km h^–1 or 0.55 FLs s^–1, and the greatest distance traveled was 145 km in 45.4 h. For the six tracks >21 h, there was a positive correlation between body size and maximum depth. Makos used more of the water column during daylight hours. Mean stomach temperature was 3.8±1.5°C above ambient, and body size was positively correlated with both maximum and average stomach temperature. Stomach content analyses of four makos captured at the end of tracking verified the occurrence of feeding events as indicated by changes in stomach temperature.Electronic Supplementary Material Supplementary material is available in the online version of this article at Communicated by J.P. Grassle, New Brunswick     

Energetics of underwater swimming in the great cormorant (Phalacrocorax carbo sinensis)

Resting metabolic rate (RMR), energy requirements and body core temperature were measured during underwater swimming in great cormorants (Phalacrocorax carbo sinensis) at the zoological garden in Neumünster, Germany, using gas respirometry and stomach temperature loggers. We used a 13 m long still water canal equipped with a respiration chamber at each end. Birds swam voluntarily in the canal at a mean speed of 1.51 ms^-1. Power input during underwater swimming averaged 31.4 W kg^-1. Minimal costs of transport of 19.1 J kg^-1 m^-1 were observed at a speed of 1.92 m s^-1. Body core temperature was stable in all birds within the first 60 min spent in the canal. After that, body temperature dropped at a rate of 0.14°C min^-1 until the birds voluntarily left the water. Our data indicate that great cormorants spend 2.7 times more energy than Adélie penguins (Pygoscelis adeliae) during underwater swimming. This can be essentially attributed to their poor insulation, their mode of locomotion underwater and differences in streamlining. RMR on land was related to body mass via VO₂=0.691 M^0.755 (where VO₂ is O₂-consumption in litre h^-1 and M is body mass in kg). In order to quantify the effects of external devices on energy consumption during underwater swimming, we tested a dummy data logger attached to the back of the cormorants as well as a ring on the leg. The ring had no apparent influence on the swimming energetics of the cormorants. In birds equipped with dummy loggers, swimming speed was not significantly influenced, but both power input and costs of transport increased by a mean of 19% for swimming speeds between 1.4 and 1.8 m s^-1.     

Heavy metals in the Sea-Skater Halobates robustus from the Galápagos Islands: Concentrations in nature and uptake experiments,with special reference to cadmium

Samples of Halobates robustus Barber (Heteroptera: Gerridae) from the Galápagos Islands were analysed by optical emission spectrometry. The levels (in g g^-1 dry weight) of Zn (134), Cu (155), Pb (< 1), Cd (7), and Cr (3) were not significantly different among insects of different sexes or developmental stages. The low natural levels of Cd in H. robustus from the relatively unpolluted environment of the Galápagos Islands are compared to the high concentrations of Cd in Halobates spp. from relatively polluted regions. Since the measured levels of Cd in their natural zooplankton food rarely exceed 10 g g^-1, and very little of the Cd is found in the soft tissues, the high Cd concentrations (100 to 200 g g^–1) in some seaskater species have evidently been derived by drinking from the surface microlayer of the seawater.     

Ultrasonic tracking of Greenland sharks,<Emphasis Type="Italic"> Somniosus microcephalus</Emphasis>, under Arctic ice

Six Greenland sharks, Somniosus microcephalus (Bloch and Schneider, 1801), 190–355 cm fork length, were tracked under land-fast sea ice off northern Baffin Island (73.2°N; 85.3°W) between 16 and 28 May 1999, using ultrasonic telemetry. The sharks were tracked continuously for periods of 5.5–13.0 h, with the tracks of two individuals lasting 31.4 and 42.8 h, respectively, each with an interval when the track was lost. Several sharks dove after release and moved along the ocean bottom for the duration of the tracking period, while others varied their movements regarding course and depth. Two sharks made repeated visits to within 11 m of the ice–water interface from deeper water. The tracked sharks exhibited no apparent depth or temperature preferences, and pooled data indicated that sharks remained deep during the morning and gradually moved into shallower depths through the afternoon and night. Rates of descent (average=0.099 m s^–1) were significantly greater than rates of ascent (average=0.058 m s^–1) for all sharks, and the average rate of horizontal movement over ground was estimated as 0.215 m s^–1. Based on the movements of tracked sharks and information contained in the literature, S. microcephalus may prey on seals in areas covered by land-fast sea ice.Communicated by J.P. Grassle, New Brunswick     

Effect of temperature on the escape responses of larval herring,Clupea harengus

Herring (Clupea harengus L.) larvae from spring and autumn spawning stocks were reared at different constant temperatures from 5° to 17 °C. At equivalent developmental stages, the spring larvae were longer than the autumn larvae and the larvae reared at low temperatures were longer than those reared at high temperatures. At hatching and at the end of the yolk-sac stage, the larvae were induced, by a probe, to make C-start escape responses, which were recorded and analysed using a high-speed video recording at 400 frames s^-1. The response was rapid and of short duration. The tailbeat frequency and swimming speed were measured during the burst of swimming following the C-start at different test temperatures and in larvae with different temperature histories. The tail-beat frequency was strongly temperature-dependent, rising from 19 Hz at 5 °C to 37 Hz at 17 °C with no effect of temperature history, season or developmental stage. The burst-swimming speed ranged at hatching from 75 to 90 mm s^-1 at 5 °C to 110 to 160 mm s^-1 at 17 °C and at yolk resorption from 90–115 mm s^-1 at 5 °C to 175–190 mm s^-1 at 17 °C. The longer, spring-spawned larvae swam faster than the shorter autumn-spawned larvae. When the swimming speeds were expressed as body lengths (L) s^-1, these differences disappeared. Larvae swam from 7–9 L s^-1 at 5 °C to 15–20 L s^-1 at 17 °C at hatching, and from 8–9 L s^-1 at 5 °C to 15–17 L s^-1 at 17 °C at yolk resorption. There was, however, a significantly faster specific swimming speed by the larvae reared at 12 °C in spring 1991.Honorary Research Fellow of the Scottish Association for Marine ScienceUnfortunately, Karen Fretwell was drowned in an accident on 9 January 1993     

Otolith microstructure and daily increment validation of marbled sole (<Emphasis Type="Italic">Pseudopleuronectes yokohamae</Emphasis>)

We examined the daily deposition of otolith increments of marbled sole (Pseudopleuronectes yokohamae) larvae and juveniles by rearing experiments, and estimated the growth pattern of wild larvae and juveniles in Hakodate Bay (Hokkaido Island, Japan). At 16°C, prominent checks (inner checks; ca. 19.8 µm in diameter) were observed on the centers of sagittae and lapilli extracted from 5-day-old larvae. On both otoliths, distinctive and regular increments were observed outside of the inner checks, and the slopes of regression lines between age and the number of increments (n_i) (for sagittae: n_i=0.98×Day–5.90; for lapillus: n_i=0.96×Day–5.70) did not significantly differ from 1. Inner check formations were delayed at lower temperature, and the inner checks formed 13 days after hatching at 8°C. Over 80% of larvae, just after their yolk-sac has been absorbed completely (stage C), had inner checks on both their otoliths. On the lapilli, other checks (outer check) formed at the beginning of eye migration (stage G). To validate the daily deposition of increments during the juvenile stage, wild captured P. yokohamae juveniles were immersed in alizarin complexone (ALC)-seawater solutions and reared in cages set in their natural habitat. After 6 days, the mean number of rings deposited after the ALC mark was 5.7. The age–body length relationship of wild P. yokohamae larvae and juveniles caught in Hakodate Bay was divided into three phases. In the larval period, the relationship was represented by a quadratic equation (notochord length=–0.010×Age²+0.682×Age–2.480, r²=0.82, P<0.001), and the estimated instantaneous growth was 0.38 mm day^–1 at 15 days, 0 mm day^–1 at 34 days and –0.12 mm day^–1 at 40 days. The age–body length relationship in the early juvenile stage (<50 days) and the late juvenile stage (>50 days) were represented by linear equations (standard length=0.055×Age+5.722 and standard length=0.345×Age–9.908, respectively). These results showed that the growth rates in the late larval periods and the early juvenile stage were lower than those in the early larval stage and late juvenile stage; during the slow growth period, energy appears to be directed towards metamorphosis rather than body growth. This study provided the information needed to use otolith microstructure analysis for wild marbled sole larvae and juveniles.Communicated by T. Ikeda, Hakodate     

Swimming performance of three southwest Pacific fishes

The logarithm of stamina for each of Sardinops sagax (4 to 6 600 s), Scomber japonicus peruanus (16 to 27 000 s) and Odontestes regia (7 to 9 900 s), adjusted to a length of 10 cm, decreased linearly over swimming speeds of 31 to 82, 25 to 78 and 24 to 75 cm s^-1, respectively (19°C). The regression coefficient was -0.064 for both S. j. peruanus and O. regia and -0.049 for S. sagax. Critical swimming speed (60 min, 5 cm s^-1) for S. sagax (10cm), 32 cm s^-1, is within the range found for other species of similar length. The suggestion of a change in regression coefficient as swimming speed increased from prolonged to burst (Brett, 1964) was not supported by the results of this study.     

Respiration rate and cost of swimming for Antarctic krill,<Emphasis Type="Italic"> Euphausia superba</Emphasis>, in large groups in the laboratory

Constructing realistic energy budgets for Antarctic krill, Euphausia superba, is hampered by the lack of data on the metabolic costs associated with swimming. In this study respiration rates and pleopod beating rates were measured at six current speeds. Pleopod beating rates increased linearly with current speed, reaching a maximum of 6 beats s^–1 at 17 cm s^–1. There was a concomitant linear increase in respiration rate, from 1.8 mg O₂ g_D^–1 h^–1 at 3 cm s^–1 to 8.0 mg O₂ g_D^–1 h^–1 at 17 cm s^–1. The size of the group tested (50, 100 and 300 krill) did not have a significant effect on pleopod beating rates or oxygen consumption (ANCOVA, F=0.264; P>0.05). The cost of transport reached a maximum of 75 J g^–1 km^–1 at 5 cm s^–1, and then decreased with increasing current speed to 29 J g^–1 km^–1. When considered in light of energy budgets for E. superba, these data indicate that the cost of swimming could account for up to 73% of total daily metabolic expenditure during early summer.Communicated by G.F. Humphrey, Sydney     

Feeding and swimming of lysianassid amphipods in a shallow cold-water bay

The potential for dispersal by lysianassid amphipods and their localization to carrion in a shallow cold-water bay in the Middle Saint Lawrence Estuary were assessed by means of endobenthic sampling, SCUBA observations, measures of swimming speeds, and by exposure of bait (50–100 g of fish) in traps. Seventy-five to 99.9% of animals attracted to traps were lysianassid amphipods belonging to five species. Lysianassid species were spatially segregated in the Bay at low tide but all were more or less dispersed at high tide. Second cohortAnonyx sarsi Steele and Brunel,Boeckosimus edwardsi andOnisimus littoralis (Krøyer) were more dispersed than the small first cohort individuals. Second cohortA. sarsi were crawlers or low (0–0.5 m off the bottom) suprabenthic swimmers in the day, but upper (0.5–2 m) suprabenthic swimmers at night. In contrast, first cohortA. sarsi were crawlers or low suprabenthic swimmers day-and-night, whileOrchomenella pinguis (Boeck) followed this swimming pattern at night but were generally akinetic in the day. Mean swimming speeds ofA. sarsi (13.6 cm s^-1) andOn. littoralis (12.1 cm s^-1) were 2 to 3 times greater than those ofOr. pinguis (7.4 cm s^-1) andPsammonyx nobilis (Stimpson) (4.4 cm s^-1). Catchability coefficients (i.e. ratio number of individuals per trap:endobenthic abundance) were 74 (A. sarsi), 8 (On. littoralis), 7 (Or. pinguis), and 0.7 (P. nobolis) m² of bottom. Gut content analysis indicated thatA. sarsi fed mostly on large carrion, whileOn. littoralis were markedly opportunistic, andOr. pinguis andP. nobilis relied on detritus, algae, and small crustaceans.     

Whirligig beetle aggregations: what are the costs and the benefits?

Laboratory experiments investigated the benefits and costs of aggregation formation in the whirligig beetles, Gyrinus marinus and G. substriatus (Coleoptera: Gyrinidae). Different sized groups of whirligigs were exposed to fish predators, and capture rate per group and per individual were estimated. Attack rate per group increased with group size, suggesting that these aggregations behave as selfish herds. In another series of experiments in which whirligigs were exposed to top- and side-attacking predators, large groups were found to detect predators when they were further away than small groups. Video analysis of these groups showed that beetles tended to increase their swimming speed after exposure to a predator but did not show an increased tendency to circle. In natural aggregations, filmed in the field, swimming speed was found to be related to group size. Body condition of field collected beetles, as estimated from the regression of body mass on body length, was not found to be related to group size, implying that large groups do not suffer reduced condition. Received: 31 January 1997 / Accepted after revision: 24 November 1997     

Fish school density and volume

All the fish in a school occupy a volume estimated as N·BL³, where N is the number of fish and BL is their mean body length. We present extensive data from our experiments on cruising schools of saithe (Pollachius virens), herring (Clupea harengus) and cod (Gadus morhua) to validate this formula. Two methods of calculating the volumes of schools are described. One method is aggregative and depends on measuring the envelope of free space around a schooling fish, whereas the other is based on the dimensions of the school as a whole. The whole-school method is more reliable since it includes lacunae between the sub-units which exist in schools. For this method, we derive a computation which eliminates bias from outliers. The most realistic theoretical aggregative packing model predicts a volume per fish of 0.6 BL³. In saithe, the envelope of free space is approximately an ellipsoid, which, although it becomes more compressed at higher swimming speeds, yields a volume close to 0.7 BL³. From the whole-school method we calculate average volumes of 1.4 BL³ for saithe and 0.7 BL³ for herring. Increase in swimming speed produces more compact schools in saithe, but changes in arousal level can generate equally large differences. Changes in volume were not adequately explained by changes in nearest neighbour distance, giving support to the whole-school method.     

Critical swimming speeds of late-stage coral reef fish larvae: variation within species,among species and between locations

The swimming abilities of larval fishes are important for their survival, potentially affecting their ability to avoid predators, obtain food and control dispersal patterns. Near settlement swimming abilities may also influence spatial and temporal patterns of recruitment. We examined Critical speed (U-crit) swimming ability in late stage larvae of 89 species of coral reef fishes from the Great Barrier Reef and the Caribbean. Coefficients of variation in U-crit calculated at the individual level were high (28.4%), and this was not explained by differences in size or condition factor of these same larvae. Among species U-crit ranged from 5.5 cm s⁻¹ to 100.8 cm s⁻¹ (mean=37.3 cm s⁻¹), with 95% of species able to swim faster than the average current speed around Lizard Island, suggesting that most species should be capable of influencing their spatial and temporal patterns of settlement. Inter-specific differences in swimming ability (at both the family and species levels) were significantly correlated with size and larval morphology. Correlations were found between swimming performance and propulsive area, fineness ratio and aspect ratio, and these morphological parameters may prove useful for predicting swimming ability in other taxa. Overall, the swimming speeds of larvae from the same families at the two locations were relatively similar, although the Lutjanidae and Acanthuridae from the Caribbean were significantly slower than those from the great barrier reef. Differences in swimming speed and body form among late stage larvae suggests that they will respond differently to factors influencing survival and transport during their pelagic phase, as well as habitat use following settlement.     

Wave exposure,swimming performance,and the structure of tropical and temperate reef fish assemblages

We examined the relationship between swimming performance, wave exposure, and the distribution patterns of labrids on temperate rocky reefs, in comparison with previous functional analyses of a tropical assemblage. Visual censuses of the distribution and abundance of labrids across two major gradients of wave exposure (depth and aspect to prevailing winds) were made at two offshore islands near Port Stephens, New South Wales, Australia. Distinct shifts in species composition and abundance were evident between high and low wave exposure habitats on temperate rocky reefs, particularly between deep and shallow habitats on exposed reef fronts. The swimming performances of temperate labrids were assessed through examination of pectoral fin shape (aspect ratio) and in situ swimming speeds. A diversity of pectoral fin morphologies was exhibited within this temperate assemblage, ranging from rounded to tapered fins (aspect ratios of 0.52 and 1.43, respectively). Fin shape was strongly correlated (Pearsons correlation 0.884, P<0.001) with swimming speed (ranging from 1.05 and 3.06 body lengths s^–1), in a relationship comparable to that observed in tropical labrids. Inter-specific differences in swimming ability provided some explanation for differences in the distribution and abundance of temperate labrids in relation to wave exposure. However, our findings suggest that although coral reef labrids appear to predominantly use high aspect-ratio fins to successfully occupy wave-exposed habitats, temperate labrids appear to be using an enhanced swimming ability through increased body size.Communicated by G.F. Humphrey, Sydney     

In situ ontogeny of behaviour in pelagic larvae of three temperate, marine, demersal fishes

The ontogeny of behaviour relevant to dispersal was studied in situ with reared pelagic larvae of three warm temperate, marine, demersal fishes: Argyrosomus japonicus (Sciaenidae), Acanthopagrus australis and Pagrus auratus (both Sparidae). Larvae of 5–14 mm SL were released in the sea, and their swimming speed, depth and direction were observed by divers. Behaviour differed among species, and to some extent, among locations. Swimming speed increased linearly at 0.4–2.0 cm s⁻¹ per mm size, depending on species. The sciaenid was slower than the sparids by 2–6 cm s⁻¹ at any size, but uniquely, it swam faster in a sheltered bay than in the ocean. Mean speeds were 4–10 body lengths s⁻¹. At settlement size, mean speed was 5–10 cm s⁻¹, and the best performing individuals swam up to twice the mean speed. In situ swimming speed was linearly correlated (R ²=0.72) with a laboratory measure of swimming speed (critical speed): the slope of the relationship was 0.32, but due to a non-zero intercept, overall, in situ speed was 25% of critical speed. Ontogenetic vertical migrations of several metres were found in all three species: the sciaenid and one sparid descended, whereas the other sparid ascended to the surface. Overall, 74–84% of individual larvae swam in a non-random way, and the frequency of directional individuals did not change ontogenetically. Indications of ontogenetic change in orientated swimming (i.e. the direction of non-random swimming) were found in all three species, with orientated swimming having developed in the sparids by about 8 mm. One sparid swam W (towards shore) when <10 mm, and changed direction towards NE (parallel to shore) when >10 mm. These results are consistent with limited in situ observations of settlement-stage wild larvae of the two sparids. In situ, larvae of these three species have swimming, depth determination and orientation behaviour sufficiently well developed to substantially influence dispersal trajectories for most of their pelagic period.