The crucian carp (Carassius carassius L.) is one of the most anoxia-tolerant fishes. An important strategy used by the crucian carp to survive anoxia is to lower its rate of energy consumption. Anoxia-tolerant fish are known to utilize simultaneously two different strategies for reducing energy consumption during anoxia, the first being a reduction in locomotor activity and the second being a depression of cellular energy demands. Nevertheless, the reduction in locomotor activity during anoxia has never been measured quantitatively. This lack of information is apparently because technical problems have prevented the measurement of spontaneous locomotor activity in fish. It is now possible to use computerized video-imaging techniques to record the movement of an animal continuously. By the use of such a technique, we show that crucian carp respond to anoxia (330min at 9°C) by rapidly decreasing their locomotor activity (spontaneous swimming distance) to about 50% of that displayed during normoxia. Frequency diagrams of spontaneous swimming speed showed no bimodality and indicated a general decrease in swimming speed from a median value of 1.82 m min−1 during normoxia to 0.82 m min−1 during anoxia. It is tentatively estimated that the anoxic depression of locomotor activity corresponds to a 35–40% reduction in total energy consumption. The role of locomotor activity in fish energy budgets is discussed.
The most anoxia-tolerant fishes appear to be the two closely related species in the Palaearctic cyprinid genus Carassius, the crucian carp (C. carassius L.) and the goldfish (C. auratus L.). Both species have the capacity to survive days of anoxia at room temperature and, at temperatures close to 0°C, the crucian carp can survive without O2 for several months (Blazka, 1958; Piironen and Holopainen, 1986). In northern Europe, the unsurpassed anoxia-tolerance of the crucian carp makes it the sole piscine inhabitant of many small lakes and ponds that become ice-covered and anoxic during the winter (Holopainen and Hyvärinen, 1985).
Energetic adaptations to anoxic survival have been studied in Carassius for more than three decades. These studies have revealed that both Carassius species have the unique ability to produce ethanol as their major anaerobic end product, thereby avoiding a build-up of toxic lactate levels (Shoubridge and Hochachka, 1980; Van Waarde, 1991). Glycogen is the main source of fuel during anoxia, and the glycolytic pathways of Carassius have been found to be supplied by the largest liver glycogen stores of any vertebrates (Hochachka and Somero, 1984; Hyvärinen et al. 1985). A total depletion of this glycogen store appears to be the factor than finally limits anoxic survival in crucian carp (Nilsson, 1990). Consequently, anoxic survival time in Carassius will be directly related to the degree to which the rate of energy consumption can be reduced.
Two cooperating strategies for saving energy, one physiological and one behavioural, are utilized by anoxia-tolerant animals (Ultsch, 1989). The first strategy is a decrease in cellular energy metabolism and it is usually referred to as metabolic depression. In goldfish, this strategy is reflected by a 70% reduction in the standard metabolic rate measured as heat production during anoxia (Van Waversveld et al. 1989). The second strategy is a reduction in locomotor activity. This reduction does not contribute to the 70% reduction in the standard metabolic rate, since standard metabolic rate is measured in physically inactive animals. Although it has been noted by several authors that anoxia-tolerant animals respond to anoxia by decreasing locomotor activity (Blazka, 1958; Ultsch, 1989; Nilsson, 1990), quantification of this phenomenon has been lacking. Quantitative activity studies in anoxic fish have been hampered by experimental difficulties in monitoring spontaneous swimming in fish, and previous attempts to quantify the lowered locomotor activity of anoxic Carassius have failed (Kutty, 1968; Ruff and Zippel, 1968). However, because of the recent rapid development of computerized video-imaging techniques, it is now possible to make direct quantitative measurements of spontaneous locomotor activity in fish.
In the present study, we have used a computerized image-analysis system to measure and compare spontaneous locomotor activity continuously (swimming distance) in anoxic and normoxic crucian carp.
MATERIALS AND METHODS
The crucian carp used in the experiments weighed 17.7±5.2g (mean ± S.D.) and had a body length of 92–116mm. They were caught in September in a small pond near Uppsala. The fish were kept indoors in a holding tank continuously supplied with aerated Uppsala tapwater (8–10°C). The automatic light/dark regime reflected conditions at a latitude of 51°N, with the length of the light/dark periods being continuously and automatically adjusted to seasonal variations. The fish were fed daily with commercial fish pellets (EWOS ST40, Astra-EWOS Sweden) at 1–2% of their body weight. The experiments were conducted in April after 6 months of acclimation to the indoor conditions.
The experimental arrangement is shown schematically in Fig. 1. All activity measurements were made on one fish at a time. Locomotor activity (horizontal movement) was measured in a polyvinylchloride box called the activity chamber. The activity chamber consisted of a test area, measuring 450mm×320mm×70mm (length×width×depth), which, on each short side, was limited by a plastic net (10mm mesh). Airstones for N2 bubbling or aeration were placed outside each of these plastic nets and the whole chamber was covered by an acrylic sheet that was curved upwards at each end to stop air or N2 bubbles (that would obscure the video image) from reaching the test area. A fish was introduced into the test area by removing the removable wall indicated in Fig. 1 and the nearby plastic net. In order to maintain a constant temperature in the test area, the long sides and bottom of the activity chamber consisted of double walls (not shown in Fig. 1) between which water at 8°C was circulated. The water in this system was separated from the water inside the activity chamber. This arrangement gave a temperature in the test area of 9±1°C. An O2 electrode (Oximeter OXI 96 from Wissenschaftlich Technische ätten, Weilheim, Germany) was placed half way along the test area. In order to make water circulate over the electrode membrane, the electrode was equipped with a small magnetic propeller (supplied by the manufacturer) driven by a magnetic stirrer placed outside the activity chamber. The electrode was calibrated before each experiment.
A black and white CCD video camera (Panasonic WV-BL200, 4.8mm lens, 0.5lx light sensitivity) from which the infrared filter had been removed was placed 35cm above the test area. The camera was connected to a video monitor and a computer in an adjacent room (see below). Two 20W halogen lamps (L1 and L2, Fig. 1) equipped with a red glass filter (RG 780, Melles Griot; F1, Fig. 1), allowing no transmission of wavelengths shorter than 750nm, were used as the light source for the video camera. This infrared light was reflected by a glass mirror and a sheet of aluminium foil through a plate of opalescent white glass situated immediately under the test area. On the video monitor this arrangement gave a light background, without reflections, against which the fish was readily detectable. The visible light for the fish was provided by a 60W tungsten lamp directed against a white wall 100cm from the test area, thereby giving indirect light in the test area. Disturbance of the video picture by reflections of visible light was prevented by a Kodak WRATTEN filter 89B (which allows no transmission of wavelengths shorter than 670nm) mounted in front of the video camera lens (F2, Fig. 1).
Computer hardware and software
The hardware consisted of a PV VISION PLUS video digitizer board (Imaging Tech. Inc.) connected to an IBM-compatible computer (CPU: Intel 386 DX, 20MHz). TEA image manager TIM 3.30 (Difa Measuring Systems, Breda, The Netherlands) was used as the driver program for the card.
The motion analysis program MOTION 0.12β (written by Jacob Rousseau, Institute of Molecular Biology and Medical Biotechnology, University of Utrecht, The Netherlands) has previously been described by Spruijt et al. (1992) for behavioural studies on rats. In this system, the data acquisition procedure is based on the following principle. A picture of the empty test area is digitized and subtracted from a digitized picture of the test area containing the animal, leaving data for the animal and the noise. The noise is removed by a threshold operation and the remaining signal is transformed to a binary image. Before each experiment, the computer automatically determines the appropriate threshold; irrelevant objects such as faeces are ignored because objects were tested against a criterion of size.
The computer recorded the coordinates of the centre point of the fish at a sampling rate of 1Hz. Before each test, the computer was calibrated by placing a ruler in the test area. In the present experiment, we let the computer calculate the distance travelled for periods of 33min and 1min. The resolution was set at 3 pixels, which in the present experimental set-up meant that the smallest movement that could be detected between two subsequent samples was 10mm. Consequently, movement was not registered if the fish was standing still or moved back and forth within a radius of 10mm around its centre point. However, if the fish moved 5mm between records in a steady direction, this movement was correctly recorded as 10mm for every second record.
All experiments were performed during the light period between 9:00 and 17:00h.
Before the experiments, the activity chamber was filled with aerated Uppsala tapwater (100% air saturation). Individual fish were transferred to the anoxia chamber 15min before the activity measurement started, and spontaneous locomotor activity was measured over 429min (13 periods of 33min). For fish exposed to anoxia, N2 bubbling was started at the same time as the activity measurements began. After 330min (10 periods), the anoxic fish were exposed to normoxia again by switching from N2 bubbling to air bubbling. For normoxic controls, the water was bubbled with air during the whole experiment. Care was taken not to disturb the fish during the experiments.
Values are given as means ± 95% confidence limits. Statistical significance was tested using a two-tailed Wilcoxon rank sum test.
The computer recorded the position of the fish once every second. The general smoothness of the swimming patterns registered by the computer (Fig. 2) suggested that the sampling rate gave a satisfactory measure of locomotor activity and that the swimming distances measured closely corresponded to the actual distances travelled. Accurate measurement of distance travelled was facilitated by the low spontaneous swimming speed of the crucian carp, generally about 2 m min−1 (see Fig. 4A), corresponding to an average movement of 3.3cm (1/3 body length) between each positional record. The swimming speed very rarely exceeded 4 m min−1 (see Fig. 4A). One possible source of error was that only two-dimensional (horizontal) movement could be measured. However, vertical movements of the fish were minimized by keeping the height of the chamber low (70mm) compared to the height of the fish (approximately 40mm).
Effects of anoxia on locomotor activity
When N2 bubbling was started, the concentration of O2 in the water in the centre of the test area decreased rapidly (Fig. 3A). Within 60min, it fell from 11.5mg l−1 (=100% air saturation) to 0.1mg l−1 (the lower detection limit of the O2 electrode), where it remained during the rest of the N2 bubbling period. The period from 66 to 330min was denoted the anoxic period and included eight 33min activity measurement periods.
As the O2 concentration fell, there was a marked reduction in the activity of the fish (Fig. 3B). Similarly, when ambient O2 concentration was increased, the activity of the fish rose. In contrast, spontaneous activity of normoxic controls showed no significant change, although there was a tendency towards a slow progressive decline. During the 264min anoxic period, the total distance swum by anoxic fish (247±86m, N=5) was significantly lower than that of normoxic controls (512±87m, N=6, P=0.004). Anoxia caused a 52% decrease in locomotor activity, corresponding to a fall in average swimming speed from 1.94 to 0.94 m min−1. During the last 66min of the experiment, there was no significant difference between the activity of the normoxic controls and that of the individuals that had just been exposed to anoxia, although the activity of previously anoxic fish tended to be higher than that of normoxic controls, especially during the last 33min period ( P=0.08).
The difference in swimming activity between the anoxic group and controls is presented in Fig. 3C, which shows that the depression of swimming activity remained relatively constant throughout the anoxic period.
When the distance travelled by normoxic and anoxic crucian carp during the 264min anoxic period was summed for each minute, frequency diagrams (Fig. 4) showed that the fish generally moved much slower during anoxia. In order to simplify the comparison of the frequency distribution of swimming distance in anoxic and normoxic fish, one normoxic individual was randomly deleted from the analysis (giving N=5 in both groups). The median distances travelled by individual normoxic fish during 1min periods varied between 0.98 and 2.23 m min−1 (grand median=1.82 m min−1), while those of anoxic fish were significantly lower (P=0.016, Mann–Whitney U-test), varying between 0.56 and 1.02 m min−1 (grand median=0.82 m min−1). Thus, the results showed that there was a 55% decrease in median distance travelled during anoxia.
The results of the present study show that the crucian carp reacts to anoxia by decreasing its locomotor activity. The frequency distribution of swimming distance during 1min periods (Fig. 4) clearly suggested that the fish responded to anoxia with a general decrease in swimming activity. There was, for example, no obvious bimodality in the frequency distribution of swimming distance in anoxic fish, which appears to exclude the possibility that they respond to anoxia by standing still most of the time but, when moving, swimming at a similar speed as during normoxia. A general decrease in activity during anoxia was also indicated by the similarity between the decrease in median distance travelled during 1min periods (a 55% fall) and the decrease in total distance travelled during the whole anoxic period (a 52% fall).
The glycogen stores of even an anoxia-tolerant vertebrate, such as the crucian carp, are limited, so the rate of energy consumption (=glycogen use) during anoxia determines the survival time of anoxic crucian carp (Nilsson, 1990). Thus, the decrease in locomotor activity displayed by anoxic crucian carp is probably a strategy for saving energy, thereby extending the anoxic survival time. In response to anoxia, the closely related goldfish has previously been shown to reduce its standard metabolic rate (measured as heat production) by 70% (Van Waversveld et al. 1989), a reduction in energy requirement that does not rely on decreased locomotor activity since standard metabolic rate is measured in inactive fish. The energy saved by the decrease in physical activity must, therefore, be in addition to the savings made by the depression of standard metabolic rate. Thus, in Carassius, the decrease in energy use during anoxia appears to be accomplished by two separate mechanisms, the first being a reduction in the muscular energy consumption underlying locomotor activity, and the second being a reduction in the cellular metabolic work that is not related to locomotion, probably involving tissues other than muscle.
From an ecological perspective, it is interesting to know how much energy is saved by the 52% reduction in locomotor activity displayed by anoxic crucian carp. Although available data do not allow this to be calculated exactly, estimates can be made. The difference between standard metabolic rate (inactive fish) and average routine metabolic rate (spontaneously swimming fish) appears to be about threefold in several species, including goldfish (Beamish and Mookherjii, 1964), common carp (Cyprinus carpio) (Lomholt and Johansen, 1979) and rainbow trout (Oncorhynchus mykiss) (Steffensen, 1989). This suggests that locomotor activity is responsible for about 67% of the energy used by spontaneously active fish. Assuming that this is also true for crucian carp, the 52% reduction in locomotor activity displayed by anoxic crucian carp would correspond to a 35% decrease in energy use. This approximation also relies on the assumption that there is a fairly linear relationship between distance travelled and the energetic cost of swimming at these slow speeds. Such a linear relationship between spontaneous locomotor activity and energy consumption is supported by the results of Smit (1965) as well as by an empirical model recently developed by Boisclair and Tang (1993). Indeed, the data obtained by Smit (1965) from goldfish at 20°C suggest that the reduction of average swimming speed seen in the present experiments (from 1.94 m min−1 to 0.94 m min−1) would correspond to a 40% fall in energy consumption, a value similar to the 35% calculated above.
There is an interesting parallel between the reduction in spontaneous locomotor activity during anoxia and that displayed by fish during gonadal development. Koch and Wieser (1983) showed that in the roach (Rutilus rutilus L.), another North Palaearctic cyprinid, there is a considerable reduction in spontaneous locomotor activity, while energy consumption is maintained, during the annual period of gonadal development. Thereby, energy is redirected from locomotor activity to gonadal synthesis. Interestingly, as in anoxic crucian carp, the activity reduction seen during gonadal development in roach was about 50%.
Thus, it may be a general tendency for fish to use their locomotor activity as a buffer in their energy budget. By periodically down-regulating their locomotor activity, they can save energy during temporary energy shortage (anoxia) or redirect energy to meet special demands (gonadal development). It is also possible that other vertebrates could be included in this generalization. For example, rats have been shown to reduce their locomotor activity in response to food deprivation (Westerterp, 1977).
Although a reduction in spontaneous locomotor activity in anoxic Carassius has been noted by several investigators, the present experiments are the first accurate measurements of this phenomenon. Most physiological studies of anoxic Carassius have focused on the effects of a few hours of anoxia. Similarly, a relatively short anoxic period was examined in the present experiments. However, because Carassius has the ability to survive several days of anoxia at 9°C, it should be kept in mind that changes in locomotor activity during anoxia in this species could be multiphasic and that long-term anoxia might involve a further reduction in physical activity.
The study was financially supported by the Swedish Council for Forestry and Agricultural Research (grants nos 0890/89 V 88 and 40.0608/92) and the Helge Ax:son Johnson Foundation.