A Note on Interactions Between Temperature, Viscosity, Body Size and Swimming Energetics in Fish Larvae

The effects of temperature on swimming in aquatic animals have been studied by recording the kinematics and/or energy expenditure of the animals. Since, in water, changes in temperature are obligatorily linked to changes in viscosity, attempts have been made to separate the effects of these two variables. This can be achieved by altering the viscosity of water independent of temperature by the addition of an osmotically inactive substance such as methyl cellulose, dextran or polyvinyl pyrrolidone (Linley, 1986; Podolsky, 1994; Podolsky and Emlet, 1993). Recently, Fuiman and Batty (1997) applied this method to small (9.6 mm) and large (18.2 mm) herring larvae Clupea harengus. They recorded kinematic variables from spontaneously swimming larvae at 6, 9 and 13 °C, and artificially altered viscosity from 1.2×10⁻⁶ to 2.0×10⁻⁶ m² s⁻¹. Their major findings were (1) that voluntary swimming speeds of small larvae are strongly affected by viscosity but relatively little by temperature at equal viscosities, and (2) that the viscous hydrodynamic regime for larval herring extends further up the scale of Reynolds number (to approximately 450) than is usually assumed (Weihs, 1980; Webb and Weihs, 1986; Osse, 1990). The effect of the viscosity of water on the swimming behaviour of small herring larvae is demonstrated best by the observation (Fig. 2 in Fuiman and Batty, 1997) that voluntary swimming speed declined linearly from approximately 15 to 5 mm s⁻¹ as viscosity increased from 1.2×10⁻⁶ to 2.0×10⁻⁶ m² s⁻¹ and that this relationship was totally unaffected by water temperature. These experiments with spontaneously swimming fish raise questions regarding the energetic consequences of such a behaviour. If, as the results of Fuiman and Batty (1997) show, small herring larvae select an optimal swimming speed at a constant level of energy expenditure, how would energy expenditure be affected if the fish were forced to swim outside their selected (optimal?) range of speeds? The major question of physiological interest here is the extent to which it is possible to separate direct from indirect effects of temperature on the swimming energetics of the larvae.

On the basis of a fairly complete set of data from a previous study on the swimming energetics of the larvae of a species of freshwater fish, the Danube bleak Chalcalburnus chalcoides (Kaufmann, 1990; Kaufmann and Wieser, 1992), we are able to show that temperature exerts an indirect influence on the swimming energetics of fish larvae not so much via its effect on the viscosity of water but via its effect on the contractile properties of swimming muscles.

Thus, the effects of temperature on the active and standard metabolic rates can be derived separately. Note, however, that it is assumed that R_s is constant. If, during swimming, maintenance functions included in R_s were temporarily suppressed – and there are indications that this may sometimes be the case (Kaufmann and Wieser, 1992; Wieser, 1995) – results regarding the metabolic ‘efficiency’ of swimming would have to be reconsidered. From equation 2, we calculated the hydromechanical power for the three size classes of larvae swimming at 15 °C and 20 °C (Fig. 1B). The excess hydromechanical power required at 15 °C compared with that at 20 °C for larvae of exactly the same length would be proportional to the change in dynamic viscosity, i.e. 13.7 % (see footnote to Table 1). This corresponds to a Q₁₀ of 0.77 under the viscous regime, irrespective of swimming speed (Fig. 1C). In consequence, this would be the value expected for the active metabolic rate R_a if the difference in energy expended by our fish at the two temperatures were due solely to hydromechanical constraints. If the fish are assumed to have swum in the inertial regime, Q₁₀ would be close to 1.0 (Fig. 1C).

Our measurements of active metabolic rate (Fig. 1A) show (exponent of 2.0 in equation 2). The difference between the metabolic and hydromechanical temperature effects is expressed by the Q₁₀ values summarized in Table 1 and plotted in Fig. 1C. At U_crit (15 °C), the Q₁₀ of R_a ranged from 0.72 to 0.59 over the three size classes. These values are lower than the Q₁₀ values calculated for the hydromechanical power under viscous (0.77) or inertial (1.0) regimes.

The transition from slow to fast swimming at the two temperatures is strikingly illustrated by the trajectories of the Q₁₀ values of total metabolic rate (R_t) in relation to swimming speed (Fig. 1C). Calculated Q₁₀ values for the standard metabolic rate R_s ranged from 1.7 (large larvae) to approximately 2.5 (medium and small larvae) (Table 1). The increasing total metabolic cost of swimming R_t at the lower temperature (due to increases in R_a) is reflected by the negative slope of the Q₁₀ curves in Fig. 1C. that at higher speeds the fish larvae expended considerably more energy at 15 °C than can be explained by hydromechanical forces (compare Fig. 1A and B). Moreover, R_a at 20 °C increased less steeply with swimming speed (values of b in equation 3 were less than 2.0) than did the corresponding curves representing hydromechanical power

Our data demonstrate that, in fish larvae swimming at 15 °C, or 5 °C below their acclimation temperature, their net energy expenditure R_a, and thus their net cost of transport (COT), is higher than can be explained by increased hydrodynamic forces, even under the unrealistic assumption that they are swimming in the viscous regime at Reynolds numbers as high as 2000. This is in contrast to the situation for large fish, in which COT remains constant or decreases with temperature (e.g. Brett, 1964; Beamish, 1970, 1981; Smit et al. 1971). A possible explanation for this difference in swimming energetics between larvae and adult fish can be derived from recent findings on the mechanics and energetics of isolated muscle fibres at different temperatures (Rome, 1990; Rome et al. 1990).

Since the work of Hill (1964), it has been known that the efficiency of muscle (the ratio of mechanical power produced to metabolic energy utilized) depends greatly on the ratio V/V_max (shortening velocity as a proportion of maximum shortening velocity), maximum efficiency being reached at values between 0.2 and 0.4. In consequence, animals moving at different speeds would benefit from the use of different muscle fibre types which reach optimal V/V_max ratios at different speeds. It is now well known (Rome et al. 1988) that fish achieve this goal with a two-gear system of red and white fibres, the former being optimized for slow speeds, the latter for fast speeds (Rome, 1990; Videler, 1993). There is a striking effect of temperature on the energetics of muscle fibres. At shortening velocities below the efficiency maximum (V/V_max approximately 0.3), efficiency is relatively unaffected by temperature, whereas at higher velocities efficiency decreases much more rapidly in cold-exposed than in warm-exposed muscle fibres. In consequence, in fish swimming at low temperatures, the fast white fibre type is activated at lower swimming speeds than in fish swimming at high temperatures. This has been called ‘compression of recruitment order theory’ (Rome, 1990). In fish larvae, the arrangement of muscle fibres differs from that in adult fish in several respects. Red fibres are represented by a unicellular red layer which envelops the whole body, receiving oxygen largely by diffusion across the skin. The central muscle mass is composed of fibres much richer in mitochondria than are the white fibres of adult fish. The result of this arrangement is that, in fish larvae, swimming is almost entirely aerobic up to the highest speeds (Kaufmann, 1990; Wieser, 1995). Thus, we suspect that the ‘two-gear system’ is not yet functional in fish larvae, the consequence being that when these fish are swimming at high speeds in cold water the muscle fibres have to operate over an increasingly inefficient range of shortening velocities. The intricate relationships between swimming speed, temperature, muscle mechanics and energy metabolism explain the shape of the R_a curves in Fig. 1A, the metabolic cost of swimming at 15 °C increasing more rapidly with swimming speed than that of swimming at 20 °C.

Fuiman and Batty (1997) concluded that when studying the effects of temperature on the active metabolic rate of fish larvae the physical effect of viscosity has to be taken into account. We extend this with the caveat that in these animals any hydromechanical effects may be overwhelmed both by developmental constraints, particularly with respect to differentiation of the swimming muscles, and by the effects of temperature on the mechanics of muscle function.

This work was supported by the ‘Fonds zur Förderung der wissenschaftlichen Forschung’ of Austria, project no S-35/04. We thank Lee Fuiman for helpful comments on an earlier version.