The Kinematics of Swimming in Larvae of the Clawed Frog, Xenopus Laevis

Recent studies on the locomotion of the typical anuran larvae, Rana and Bufo (Hoff & Wassersug, 1985; Wassersug & Hoff, 1985) have emphasized that the kinematics of their locomotion is similar to that of many fishes, despite gross differences in morphology. Indeed, although anuran larvae characteristically have globose bodies, no pectoral or pelvic fins and little axial muscle, they move through water as efficiently as most other aquatic vertebrates of similar size.

The present study investigates whether differences in the ecology and morphology of tadpoles can affect swimming performance. Data are obtained for the larva of the African clawed frog, Xenopus laevis (family Pipidae), which is distinguishable from the tadpoles of Rana and Bufo in that it is an obligate, midwater, microphagous suspension feeder. It is positively buoyant and hovers midwater in large schools in a head-down position (Fig. 1) by sculling with the terminal portion of a long filamentous tail (Van Dijk, 1972; Passmore & Carruthers, 1979). Both Rana and Bufo are negatively buoyant and are frequently found resting on the bottom. They do not hover midwater and do not have long filamentous tails.

We examine swimming performance by quantifying standard kinematic parameters using high-speed cinematography and computer-assisted frame-by-frame analysis. We use the kinematic parameters most commonly used in studies of aquatic undulatory locomotion in other vertebrates. These include the amplitude, frequency and length of the propulsive wave, as well as Froude efficiency -a derived measure of kinematic efficiency based on swimming speed of the animal and the speed of the propulsive wave.

We show that the swimming of Xenopus larvae differs markedly from Rana and Bufo in several ways that relate to differences in life history. We also suggest that the distinctive morphologies and modes of swimming reflect major differences in the neuronal mechanisms for regulating locomotion.

Xenopus laevis larvae were raised in the laboratory and maintained at room temperature (21±1°C). Four medium-sized larvae (total body length, L, 3·5–4·6 cm) were used to film 21 sequences of normal swimming.

Axial muscle mass was determined by gross dissection of 20 tadpoles ranging from 10 to 90 mm in total length. The masses recorded include notochord and connective tissue. For calculations of centre of mass, tadpoles were cut into 10% L sections and the mass of each section was assumed to be at its geometric centre. All animals used in this study were at free-swimming, pre-metamorphic developmental stages (Nieuwkoop & Faber, 1956).

All other materials and methods were as described by Wassersug & Hoff (1985).

All results are presented in comparison with published data for more typical tadpoles of the families Ranidae and Bufonidae (Wassersug & Hoff, 1985). To facilitate those comparisons, data from Rana and Bufo are most often pooled. However, where there are major differences among the previously studied species they are treated individually.

In Xenopus there is no correlation between swimming speed (U) and tail beat frequency (f) at velocities below 6 Ls·⁻¹ (r = 0·20, P> 0·5; see Fig. 2). Below 6 Ls⁻¹ the mean tail beat frequency is 9·75 s⁻¹ (S.E. = 1·37). At higher velocities, however, the tail beat frequencies of Xenopus tadpoles fall on the regression line for Rana and Bufo larvae. The lowest swimming speed recorded for Xenopus was 2·1 cm s⁻¹ or 0·46Ls⁻¹ for a 4·6-cm individual. The highest speed was 49·7 cm s⁻¹ or 14·2Ls⁻¹ for a 3·5-cm individual. Tail beat frequency ranged from a low of 7s⁻¹ to a high of 22s⁻¹.

The specific maximum amplitude (A) for Xenopus is generally lower than for Rana and Bufo throughout the entire range of swimming speed (Fig. 3). Linear regressions of the log-transformed data show a positive correlation between U and A for both Xenopus (r=0·40, P<0·05) and Rana and Bufo (r=0·50, P<0·05). These two lines differ in both slope (F = 2302, P<0·05) and elevation (F = 594, P<0·05). The non-Xenopus data form two distinct clusters on either side of the common Rana/Bufo line. The line, in fact, divides the Rana/Bufo group into two by relative tail length; species in the upper group (Rana catesbeiana and Bufo americanas) have relatively shorter tails (60% L), while those in the lower group (Rana clamitans and Rana septentrionalis) have relatively longer tails (67% L). Xenopus has the same relative tail length as the lower group and clusters with those animals at least at moderate speeds of 3·9 Ls⁻¹. A comparison of the regression for Xenopus with one for just the longer-tailed ranids (r=0-77, P<0·01) shows that the lines still differ in slope (F = 445, P<0·05), but not in elevation (F = 53, 0·l>P>0·05). This difference reflects the fact that Xenopus can swim more slowly and with much lower tail beat amplitude than any of the other tadpoles. [See Wassersug & Hoff (1985) for further discussion of the kinematic consequences of tail length for the Rana/Bufo group.]

The amplitude of the travelling wave all along the body in Xenopus is similar to that for Rana at high swimming speeds, but at lower velocitiesXenopus shows lower amplitudes at all points along the body than any of the Rana and Bufo we have examined (Fig. 4). At all speeds there is considerable yaw at the snout inTfana and Bufo. Xenopus, however, shows no measurable lateral movement of the anterior two-thirds of its body at low speed. At higher swimming speeds Xenopus, much like Rana, swims with higher amplitudes all along the body and considerable yaw at the snout.

In Xenopus, as well as in Rana and Bufo, there is approximately one full wave on the body at all swimming speeds. For Rana and Bufo there is no correlation between wavelength and specific swimming speed, but Xenopus shows a strong positive correlation between these parameters at lower speeds (Fig. 5). At higher speeds (>6Ls⁻¹) wavelength is relatively constant (⁠⁠, S.E. = 0·02), following the pattern of the other tadpoles, although wavelength in the other larvae is considerably greater (⁠⁠, S.E. = 0·09).

Froude efficiency [η = 1−0·5(1−U/V)] (where V is the speed of the propulsive wave) values for Xenopus follow the line plotted for the other tadpoles (Fig. 6). However, Xenopus will swim readily both slower and faster than other tadpoles. At swimming velocities below those recorded for other tadpoles Froude efficiency is lower and at the highest velocities Froude efficiency is higher for Xenopus than for the other larvae.

Xenopus shows a dramatic increase in the relative amount of axial muscle with size (r = 0·96, P< 0·01 ; slope = 9·7; see Fig. 7). The smallest Xenopus have less than 10 % of their mass as axial muscle, while the largest are 45 % axial muscle. Rana has more axial muscle than does Bufo (19–26% compared to 8–14%), but both genera show a fairly constant percentage of axial muscle throughout the entire size range. For these two genera only Rana catesbeiana shows a significant increase of muscle mass with size, but the slope is very low (r = 0·64, P < 0·05 ; slope = 0·38) compared with Xenopus.

Total body mass is also arranged differently in Xenopus than in other tadpoles. The centre of mass m Xenopus is more anterior (at 17–20 % L; N = 8) than in Rana catesbeiana (at 30–35 % L; N = 6).

Xenopus tadpoles differ from Rana and Bufo tadpoles in that Xenopus swim continuously. This swimming is normally downward against their own buoyancy with little apparent forward movement. At these very low speeds Xenopus move only the posterior portion of the tail. This portion of the tail undulates at a near constant frequency (approximately 10 Hz) independent of swimming velocity in the range of slow to moderately fast cruising speeds (<6Ls⁻¹). At these speeds Xenopus has a relatively novel way of regulating velocity. Instead of increasing tail beat frequency as it increases speed (like most fishes and other tadpoles), Xenopus recruits more of the tail, thereby increasing wavelength and propulsive surface area. In contrast, Rana and Bufo tadpoles swimming at comparable low to moderate speeds use the whole tail. It should be pointed out that although the whole tail undulates in these forms, bends in the posterior portion of the tail may be strictly passive (Wassersug & Hoff, 1985).

These kinematic differences alone suggest that neuromuscular regulation of locomotion is fundamentally different in these two genera. This suggestion is supported by the observation that the spinal cord of Rana larvae is dramatically reduced posteriorly compared to that of Xenopus larvae of the same size and developmental stage (Nishikawa, Hoff & Wassersug, 1985). Spinal nerves in the posterior portion of the tail of Xenopus may thus arise from the spinal cord more caudally than in Rana. Furthermore, Xenopus myotomes in the posterior portion of the tail abut closely and may be electrically coupled, whereas in Rana posterior myotomes are separated by large bands of connective tissue and electrical coupling is not possible (Nishikawa et al. 1985).

These structural differences in the nervous system and musculature lead us to suggest that in Xenopus reflex loops in the posterior portion of the tail may be much shorter and tail movements may be initiated more caudally than in Rana. Also a wave of muscle contraction initiated in the posterior portion of the tail could be transmitted myogenically in Xenopus but not in Rana. The caudal initiation of movement in Xenopus is documented in this study, but myogenic transmission of movement has yet to be demonstrated.

Body morphology and swimming styles have recently been put in the context of feeding strategies for fishes (Webb, 1984) and the same type of analysis is possible for anuran larvae. Although the tadpoles of the three genera studied to date are all suspension feeders, Rana and Bufo are not obligated to midwater microphagy; the tadpoles of these genera have hard mouth parts that they use to graze on vegetation or other substrates. They are negatively buoyant and are most often found resting on the bottom or in vegetation. They swim only sporadically, either to avoid predation (Huey, 1980; Feder, 1983) or to move between patches of food. When they do swim, they are kinematically similar to fishes like cod and trout (see Hoff & Wassersug, 1985, and references to fish swimming cited therein). However, they usually do not maintain constant velocity for more than a few seconds and even when they are forced to swim in a constant velocity flow chamber their swimming at all flow speeds is often sporadic (Wassersug & Feder, 1983). They swim in short bursts and then allow themselves to drift backwards with the current.

In contrast, the Xenopus tadpole has no hard mouth parts and cannot graze on substrates. Rather, it hovers continuously in midwater. In a flow chamber, Xenopus maintains its position in the current as the water velocity increases. The anterior portion of the body shows no lateral displacement during slow swimming, but at higher speed there is considerable yaw, much as in Rana and Bufo. We have little data in the range 2–5 Ls⁻¹ so we cannot yet determine whether the shift in swimming mode is achieved by a gradual transition from tail tip to whole body propulsion or whether there is a discrete switch from the slow to fast mode. In Figs 2 and 5 we have tentatively indicated that this shift in swimming mode takes place abruptly at about 6Ls⁻¹, but it may begin at speeds as low as 2Ls⁻¹.

Anterior stability in Xenopus at low swimming speeds may be important in maintaining feeding currents for tadpoles in schools (Katz, Potel & Wassersug, 1981). Xenopus larvae tend to orient parallel to neighbours in a school (Wassersug, Lum & Potel, 1981), and it is clear from observations of undisturbed tadpoles in schools in large aquaria that the individual tail filament entrains water currents around the body of each tadpole and its neighbours. Large tail amplitude during schooling and lateral displacement of the anterior portion of the body would add turbulence to these otherwise relatively laminar currents and could disrupt feeding. Increased turbulence would also increase the rate of mixing of filtered water with unfiltered water and could lower individual feeding efficiencies (cf. Seale, Hoff & Wassersug, 1982). It may be noteworthy that at higher swimming speeds, when the snout of a Xenopus larva wobbles back and forth like that of Rana and Bufo, Xenopus reduces its buccal pumping rates (Wassersug & Feder, 1983). This necessarily reduces buccal irrigation for both respiration and feeding (Seale & Wassersug, 1979; Feder et al. 1984).

The mechanism for reducing lateral displacement at the snout is not fully understood. We would expect that any lateral displacement at one end of the body would be accompanied by recoil at the other, but in Xenopus no recoil is observed. Because Xenopus maintains a full wave in the tail at all swimming speeds, lateral forces would be balanced on either side of the body, and yaw would be reduced. The fact that the propulsive wave has low amplitude (Fig. 3) and is confined to the posterior portion of the tail (Fig. 1) may further reduce lateral displacement at the snout. In addition, Lighthill (1970) suggested that anterior yaw can be reduced in aquatic vertebrates by expanding the forward portion of the body dorsoventrally and moving the centre of mass forward. Indeed, Xenopus has a ventral fin that extends forwards onto the body and may serve as a keel to reduce anterior yaw (Fig. 1). Such a fin fold is unknown in bottom-dwelling tadpoles. The centre of mass is also significantly further forward in Xenopus than in Rana.

The lowest swimming speeds recorded for Xenopus are half the lowest speeds recorded for Rana and Bufo (see region near origin in Fig. 6). We believe that this difference is not due to biased sampling; after many weeks of observing tadpoles swimming in laboratory and field situations we conclude that Rana and Bufo larvae simply are not capable of sustained swimming at speeds much below l·0Ls⁻¹.

At these low swimming speeds Xenopus swims with a mechanical efficiency more than 50 % below the lowest efficiency recorded for Rana or Bufo (Fig. 6). However, because Xenopus uses only the posterior filamentous portion of the tail, little muscle mass is involved in very slow swimming or hovering. Thus, the low mechanical efficiency in Xenopus does not reflect low metabolic efficiencies (see Feder & Wassersug, 1984, for metabolic data on hovering Xenopus larvae and references cited therein for comparative data on other species).

While Xenopus shows the lowest swimming speed and the lowest mechanical efficiency of any tadpole studied to date (Fig. 6), it also has the highest mechanical efficiency and the highest swimming speed. Xenopus adults are aquatic and inhabit large, more or less permanent bodies of water. In these large bodies of water the pelagic Xenopus larvae occur in open water (Passmore & Carruthers, 1979) and presumably far from cover compared to bottom-feeding tadpoles living in smaller ponds or pools. The greater speed and greater efficiency at high speed of Xenopus larvae may reflect the need for longer flights to escape predation.

The positive allometry of muscle mass with body size shown strongly in Xenopus, but much less so in Rana and not in Bufo (Fig. 7), is consistent with this difference in maximum velocity and efficiency at high speed. Both Rana and Bufo maintain approximately the same proportion of their body mass as muscle throughout their larval lives, while in Xenopus relative muscle mass increases dramatically with size.

Gray (1953) states that because the rate of output of energy per weight of muscle does not vary with the size of an animal one would expect maximum velocity to vary as the cube root of body length (if the relative amount of muscle remains the same). Indeed, Fry & Cox (1970) and Wardle (1977) report decreases in maximum specific velocity with increases in length from approximately 25 Ls^{− 1} for a 20-cm fish to 6Ls^{− 1} for a 100-cm fish. In some fishes, muscle mass (as a percentage of total body mass), also decreases with increasing size (47·4% at 20cm to 37·8% at 100cm for cod; Bainbridge, 1960), and this may also contribute to the decline in maximum specific velocity. Because Xenopus shows the opposite trend (namely the larger individuals have relatively more muscle), we would expect Xenopus to show less decline in maximum specific velocity with increasing size than tadpoles of other species.

One reason why other anurans do not increase relative axial muscle mass with increasing size may be that they breed in small bodies of water where the most common predators are relatively small invertebrates. Neither high mechanical efficiency during sustained swimming nor maximum swimming speed may be the critical factor in escaping such predators in these confined habitats. Rather, as Brodie & Formanowicz (1983) and Crump (1984) have suggested, in these situations tadpoles may escape predation simply by growing rapidly and becoming too large to be attacked successfully.

A further reason may be that the overwhelming majority of frogs have terrestrial adults whereas Xenopus is totally aquatic. Wassersug & Sperry (1977) have pointed out that during the transition from an aquatic to a terrestrial way of life the anuran is more vulnerable to predation. Therefore, it is reasonable to invest minimally in the ephemeral tadpole locomotor morphology; principally axial musculature and associated caudal structures. Both.Rana andBa/o have relatively little axial musculature at the onset of metamorphosis and tail resorption is fast. In Xenopus laevis, the metamorphic transition is comparatively slow (26 % of larval life compared to a mean of 14% for 11 other species studied; from data in Wassersug & Sperry, 1977) and metamorphosing Xenopus swim relatively effectively using both a long muscular tail and well-developed hind limbs. Thus Xenopus does not face the trade-off between high performance in the aquatic environment and the transience of the tadpole body form to the same extent as terrestrial frogs.

Based on adult characteristics, Xenopus and other members of its family, the Pipidae, are usually considered to be generalized frogs compared to the ranids and bufonids (e.g. Dowling & Duellman, 1978; Laurent, 1979). The more protracted metamorphosis in pipids, which is here associated with an increase in axial muscle mass in Xenopus during larval development, has been independently interpreted as evidence for the more generalized or archaic nature of pipid frogs (Wassersug & Hoff, 1982). Alternatively, it may only reflect continuous occupation of the aquatic environment.

As a final point, our locomotor analysis to date deals strictly with constant velocity swimming in a straight path. It remains to be determined what effects the unusual morphology of the Xenopus larva has on swimming performance during acceleration and turning.

We thank R. L. Kirby, C. A. Putnam and J. J. Videler for the use of a high speed motion picture camera, image analysis equipment and software for the kinematic analysis of undulatory locomotion, respectively, in their care. The photographs in Fig. 1 were taken by J. Hanken. Help with animal care and maintenance was provided by V. A. King. Assistance with production of the manuscript was provided by C. Anjowski and F. Sasinek. Critical comments on manuscript drafts were provided by R. Marlow, K. Nishikawa and P. W. Webb. We are most grateful for the continued support of the Natural Sciences and Engineering Research Council of Canada.