Metabolic Rates at Different Oxygen Levels Determined by Direct and Indirect Calorimetry in the Oxyconformer Sipunculus Nudus

Most intertidal invertebrates, in particular those of the infauna, are temporarily exposed to hypoxic conditions when the supply of oxygenated water is interrupted during low tide. Prosser (1973) described two different patterns of reaction to situations of reduced oxygen availability: oxyregulators keep their oxygen consumption independent of the ambient oxygen tension down to a certain ⁠, below which regulation ceases. Oxyconformers, however, steadily reduce their oxygen uptake with decreasing ⁠.

The occurrence of various patterns of oxygen consumption in relation to the ambient ⁠, found even in different individuals of the same species (Mangum and van Winkle, 1973; Bayne, 1971), makes it necessary to define clearly the transition phases in oxygen consumption: the traditional approach is to select the critical (P_CI), below which oxygen consumption declines. In addition, a second critical (P_CII) may be defined, below which anaerobic metabolism starts to compensate for the decline in aerobic energy production (Portner et al. 1985).

For the lugworm Arenicola marina, the available data suggest that P_cI and P_cII differ. P_cl, indicating the transition to oxyconformity, lies near air saturation at 16 kPa (Toulmond and Tchernigovzeff, 1984), whereas the onset of anaerobic processes (P_CII) was observed only at oxygen tensions below 6 kPa (Schöttler et al. 1983). In oxyconforming Mytilus edulis, however, calorimetric experiments revealed that anaerobic metabolism is switched on as soon as oxygen uptake declines (Famme et al. 1981). Consequently, P_CI and the P_CII seem to be identical for this species. Anaerobic ATP generation could presumably compensate for the decrease in aerobic energy provision, thus keeping the total ATP supply constant down to a certain level, below which both aerobic and anaerobic metabolic rates decline rapidly.

For the marine worm Sipunculus nudus L., Pörtner et al. (1985) suggested that aerobic oxyconformity prevails over the entire range of ambient partial pressures from normoxia to 6.7 kPa. At lower oxygen tensions oxygen consumption decreases more rapidly. The accelerated decline of O₂ uptake coincides with the onset of anaerobic metabolism, as indicated by the formation of anaerobic end products.

Low anaerobic metabolic rates, however, may not be detectable by biochemical analyses. Small amounts of accumulated end products may be masked by large individual variations in metabolite concentrations, thus requiring additional evidence in the characterization of metabolic transition phases. In addition, it is still uncertain whether all the anaerobic pathways for energy production and their end products are known. Indeed, calorimetric measurements and biochemical analyses for Lumbriculus variegatus and Mytilus edulis suggest that a significant fraction of anoxic heat flux cannot be explained on the basis known anaerobic pathways (Gnaiger, 1980a; Shick et al. 1983). In contrast, Famme and Knudsen (1984) postulated an agreement between biochemical and calorimetric analyses in Tubifex tubifex. Re-analysing Famme and Knudsen’s (1984) results, Gnaiger and Staudigl (1987) claimed that the caloric equivalent employed for excreted volatile fatty acids was too high. Recalculation of the data resulted in an unexplained heat fraction of 34 %.

In the present study the non-invasive techniques of simultaneous direct calorimetry and oxygen consumption measurements were applied to determine the oxygen uptake and the prevailing energy metabolism in the marine worm Sipunculus nudus. Heat dissipation measured during extreme hypoxia was compared with enthalpy changes calculated from changes in metabolite levels. Using this approach we investigated whether the known linear decline in oxygen consumption was reflected by a similar reduction in heat production. In addition, we estimated the relative contributions of aerobic and anaerobic energy production to the total energy expenditure below the critical (P_CII). We tried to reevaluate the existence of additional, hitherto unknown, hydrogen acceptors possibly necessary to maintain redox balance during extreme hypoxia.

Specimens of Sipunculus nudus with a mean live mass of 40±8g were collected in March 1989 along the coast of Brittany, France. The animals were kept for several weeks in aquaria containing sand from the site of collection and artificial sea water (34 ‰ salinity) at a temperature of 15±1°C.

The calorimetric experiments were carried out at the Gorlaeus Laboratoria, University of Leiden, The Netherlands. Heat production and oxygen consumption of groups of 4–5 animals were measured simultaneously. Direct calorimetry was performed with a twin-flow microcalorimeter Setaram GF 108, equipped with a 1 dm³ animal chamber, as described by van Waversveld et al. (1988). At a flow rate of 16cm³min^-1 a sensitivity of approximately 94±0.5μVmW^-1 was achieved. The system was calibrated with a 10 mW pulse for 6h before and after each experiment with a reproducibility of ±0.05 mW. Baseline drift was less than 0.1 mW during calibration.

For measurements of oxygen consumption, samples of the water flowing out of both measurement and reference vessels were alternately led through an electrode cell with a flow rate of 2 cm³ min^-1. The oxygen sensor (E5046 Radiometer, Copenhagen) was calibrated with air-saturated water and a solution of 10% Na₂SO₃ (0 % O₂). Oxygen consumption rates were determined by integrating the differential readings between measurement and reference vessel in order to correct for blank rates of oxygen consumption. The solubility of oxygen in sea water (a) was calculated from the values of Carpenter (1966): α=12.4 μmol I^-1 kPa^-1 at 15 °C and 34 %o salinity.

Measurements started after the equilibration of the animal chamber at each oxygen tension. Corrections for instrumental lag were not considered necessary, since heat dissipation and oxygen consumption were integrated over extended periods (2.8 h) compared to the time constants for calorimetric and respirometric measurements (τ=54min and 62min, respectively).

The experiments were carried out at 15°C. Prior to the measurements, weighed animals were kept for 24h under experimental conditions, i.e. in a calorimeter vessel with a flow-through system without sediment, in order to reduce stress. After acclimation, animals were put into the calorimeter. Stability of the calorimeter signal was achieved within 24 h.

After 48 h of normoxia, the was successively decreased to 8.66, 2.66 and OkPa and kept at each level for 24 h. Different levels were obtained by equilibrating the inflowing water with a mixture of air, N₂ and CO₂, delivered by gas-mixing pumps (M30F and M303a-F; Wösthoff, Bochum, FRG). Actual oxygen concentrations to which the animals were exposed were calculated as the mean levels in inflowing and outflowing water. was kept constant between 0.03 and 0.04 kPa. Oxygen diffusion into the flow-through system was prevented by the use of stainless-steel tubes. During extreme hypoxia (⁠ nominally zero) the was kept below the detection limit of the oxygen electrode (=66 Pa). After 24 h of extreme hypoxia animals were allowed to recover for 48 h under normoxic conditions.

Samples of the outflowing water were collected at each oxygen level for the determination of the mean rate of acetate and propionate release. Evaporation of the volatile fatty acids was prevented by adding 0.4 % (v/v) 1 mol dm^-3 NaOH to each sample.

For biochemical determinations the same experiments were performed outside the calorimeter under identical conditions, using an identical calorimeter cell. After the incubation, animals were dissected rapidly and the coelomic fluid was collected. Body wall musculature was freeze-clamped in liquid nitrogen (Wollen-berger et al. 1960) and both coelomic fluid and muscle samples were kept frozen at —80°C. For the body wall musculature the mass fraction coefficient, which is the mass of the body wall musculature divided by the total wet mass, was found to be 0.34±0.03 (mean±s.D., N=20).

After perchloric acid (PCA) extraction of muscle tissue (Beis and Newsholme, 1975) and of coelomic fluid (Portner et al. 1984b) the following metabolites were determined enzymatically: succinate (Michal et al. 1976), phospho-L-arginine, L-arginine and octopine (Grieshaber et al. 1978) and aspartate, ATP, D-and L-alanine (Bergmeyer, 1974). Alanopine and strombine were estimated by high pressure liquid chromatography (Siegmund and Grieshaber, 1983).

Acetate and propionate were determined in diluted PCA extracts with a Dionex Bio LC ionchromatograph (Idstein, FRG). Separation was performed with an ion exclusion column (Dionex HPIC E-AS1), using 0.25 – 0.5 mmol dm^-3 octane sulphonic acid as an eluent with a flow rate of 1cm³ min^-1. The column was thermostatted at 32°C. Peaks were monitored with a conductivity detector with a sensitivity of 3 μ V for the full scale. Background conductivity was decreased using a micro membrane supressor (Dionex AMMS-ICE) regenerated with 5 mmol dm^-3 tetrabutylammonium hydroxide. Calibration curves were linear between 10 and 100,umoldm^-3 acetate and propionate with a reproducibility of ±4 %. Water samples were extracted and concentrated by water vapour distillation prior to analysis. The recovery efficiency for internal standards was 86±5 %.

The significance of differences was tested using Student’s t-test at a significance level of 5 %.

A typical time course of heat flux of Sipunculus nudus incubated at different oxygen levels is shown in Fig. 1. During the first 24 h after inserting the animals into the calorimeter, peaks of heat dissipation appeared periodically, probably reflecting enhanced activity by the animals. After 24 h these activity peaks occurred less frequently. With decreasing oxygen tensions, heat dissipation dropped steadily from 110.1 ±10.8 kJ h^-1 g^-1 fresh mass (mean±s.D., N=4) under normoxic conditions to 66.7±11.8 and 36.5±6.1 kJ h^-1g^-1 fresh mass at 8.66 and 2.66 kPa, respectively. During extreme hypoxia a value of 21.5±3.5 kJ h^-1g^-1 fresh mass was observed, which amounts to only 20% of the normoxic value. Each change in ambient was accompanied by a short and sudden increase in heat flux, the size of which decreased with decreasing ⁠. Except for these peaks, the activity of the animals was reduced under moderate (⁠and 2.66kPa) and under extreme hypoxia as shown by the general decrease in the number of peaks. After 112 h, reoxygenation coincided with an immediate rise in heat production, which reached a slightly higher level than the control value and decreased steadily thereafter. However, control values were not reached even after 48 h of recovery. During recovery, the animals did not seem to be active, since peaks of heat dissipation were absent.

Rates of heat dissipation determined by calorimetry and calculated from oxygen consumption are compared in Fig. 2. The diagram shows gaps during transitions between oxygen tensions, because oxygen consumption could not be measured under non-steady-state conditions. To calculate the aerobic heat flux an oxycaloric equivalent for mixed substrates of 450 kJ mol^-1O₂ (Gnaiger, 1983a) was used. An excess of measured heat flux over the values predicted from oxygen consumption must be due to anaerobic metabolism, whereas a rise in the caloric equivalent of oxygen uptake above measured heat production, the so-called aerobic overshoot, may be caused by endothermic processes or by oxygenation of haemerythrin and the physical solution of oxygen in the coelomic fluid.

Under normoxic conditions the energy demand was largely met by aerobic metabolism. Only during activity bursts, when heat dissipation reached values almost twice as high as the steady rate, did anaerobic metabolism occur in addition to enhanced aerobic metabolic rates. The resulting oxygen debt was discharged during the resting periods, as reflected by the aerobic overshoot. The overall calorimetric/respirometric ratio was determined in four experiments of 24 h of normoxia to be 423.4±27.1 kJ mol^-1 O₂.

At an oxygen tension of 8.66 kPa, aerobic metabolism, albeit decreased to 55 % of the normoxic value, was sufficient to maintain the animal’s energy demand. At a of 2.66 kPa the aerobic metabolic rate was reduced to 18 % of control values. At this oxygen tension, however, the anaerobic component increased to 42% of the total heat flux. During extreme hypoxia heat production was even further reduced and derived exclusively from anaerobic metabolism.

During the first hours of recovery the animals consumed more oxygen than under control conditions. This increase was not quantitatively reflected by a concomitant rise in heat dissipation. Later, anaerobic heat production began and continued for the next 30 h of recovery.

The relationship between total heat flux and aerobic (calculated) heat production at different oxygen levels is shown in Fig. 3. Aerobic heat production fell almost linearly with decreasing ⁠. At a of 8.66 kPa, oxygen consumption amounted to 55 % of the normoxic control value. Total heat production matched the aerobic (calculated) values above 8.66 kPa, but at lower oxygen levels a significant discrepancy between total and aerobic heat production was observed, indicating the onset of anaerobic metabolism. The anaerobic heat flux increased with decreasing from 15.3 mJ h^-1g^-1 fresh mass at 2.66kPa to 21.5 mJ h¹ g^-1 fresh mass under extreme hypoxia.

To interpret the metabolite concentrations given in Tables 1 and 2 it is necessary to bear in mind that the animals were subjected to a stepwise reduction in ⁠. This means that animals were kept at 8.66kPa, then at 2.66kPa, before finally being exposed to extreme hypoxia.

Metabolite concentrations (Tables 1, 2) did not change significantly between normoxic and hypoxic (8.66kPa) animals except for acetate, which decreased in the body wall musculature, propionate, which increased in the coelomic fluid, and succinate, which increased slightly in both compartments. Aspartate was degraded to some extent in the muscle tissue. A significant accumulation of anaerobic end products was found at a of 2.66kPa. Strombine concentrations rose from 0.78±0.30 to 3.69±1.80μmolg^-1 body wall muscle. The increases in octopine, L-alanine, succinate and propionate contents were less pronounced. Because of high individual variations, the rise in D-alanine concentration was not significant. Acetate was not produced at a of 2.66kPa. A significant drop in phosphagen levels from 30.43±0.25 to 19.83±5.10μmol g^-1 body wall muscle corresponded to an increase in arginine concentration from 7.87±1.96 to 15.40±6.62μmolg^-1body wall muscle. ATP levels remained constant.

After the transition to extreme hypoxia most anaerobic end products accumulated at similar rates to those under exposure to 2.66 kPa (Table 1). The formation of propionate was enhanced and, additionally, acetate production started. Tissue concentrations remained relatively low (acetate 0.17±0.07 and propionate 0.59±0.22μmolg^-1bodywallmuscle), whereas most of the fatty acids were excreted into the coelomic fluid (acetate 1.26±0.54 and propionate 1.29±0.40μmolml^-1) and the ambient water (acetate 21.3±4.5 and propionate 5.7±5.3 μmolg^-1 fresh mass). Phosphagen levels continued to decrease during extreme hypoxia but the decrease was less pronounced than under 2.66 kPa (from 19.83±5.10 to 13.50±l.43μmolg^-1bodywallmuscle). ATP levels fell from 2.20±0.52 to 1.71±0.19μmolg^-1body wallmuscle during severe hypoxia. In all samples, concentrations of alanopine were below the detection limit. Previous measurements have shown that after 24 h of anoxia no D-or L-lactate is detectable in the body wall musculature of S. nudus. The increase in malate level under these conditions can be neglected (Pörtner, 1982).

To calculate the expected metabolic heat dissipation from the changes in concentrations of anaerobic metabolites and to compare it with the measured heat production, anoxic rates of metabolite formation have to be multiplied by the molar enthalpy change for the respective end products (Gnaiger, 1983b; see Table 3) :

Table 3 shows the rates of metabolite formation during 24 h of severe hypoxia, based on the mean values given in Tables 1 and 2, and the associated heat and proton production.

where _bQ̇ is the catabolic heat dissipation due to buffer reactions (mJh^-1g^-1), is the metabolic proton production, derived from the stoichiometric equations given in Table 4, (μmolh^-1g^-1) and Δ_bH_H+ is the enthalpy of neutralization (kJ mol^-1).

The main buffer substances occurring in the body fluids of S. nudus and their respective molar enthalpy changes are listed in Table 5. Metabolic proton production amounts to 123.9 nmol h^-1g^-1 fresh mass (Table 3). The proton fractions neutralized by each buffer component were calculated as follows. (1) The change in phosphate protonation with a pH drop from 7.2 to 7.1, which was found after 24h of anoxia in S. nudus (Portner et al. 1984a), was determined. Based on phosphagen depletion, the increase in phosphate concentration was calculated as 7.2 μmolg^-1 fresh mass. This results in a proton buffering of 14.7 nmol h^-1g^-l fresh mass by phosphate protonation. (2) Proton consumption due to NH₃ formation expected from ATP depletion was calculated as 6.9 nmol h^-1g^-1 fresh mass. (3) Proton buffering by titration of bicarbonate was calculated as 24.6 nmol h^-1g^-1 fresh mass by the drop in HCO₃^- content after 24 h of anoxia (based on Portner, 1987b), and as 30.2 nmol h^-1g^-1 fresh mass by bicarbonate formation during propionate synthesis (see Tables 3, 4). Neutralization of the remaining 47.5 nmol H⁺ h^-1 g^-1 fresh mass was attributed to buffering by histidine residues. For volatile fatty acids, released into the ambient water, a A_bH_H+ of -11 kJ mol^-1 was employed for bicarbonate buffer.

The sum of the enthalpy changes of end product formation (15.0±4.5 mJ h^-1g^-1 fresh mass, N=5) and of proton buffering (2.9±l.2mJh^-lg^-1 freshmass, N=5) amounted to 17.9±4.7 mJ h^-1g^-1 fresh mass, which is not significantly different from the calorimetrically measured heat production of 21.5±3.5 mJh^-1 g^-1 fresh mass.

As an inhabitant of the intertidal zone, the marine worm Sipunculus nudus is regularly exposed to reduced water, and therefore oxygen, availability. The animal responds to hypoxic conditions by reducing its oxygen consumption correspondingly (see Fig. 3). This oxyconformic behaviour of S. nudus had already been demonstrated by Henze (1910) and was recently confirmed by Pörtner et al. (1985).

The comparison of heat production and oxygen consumption demonstrates that energy production remains fully aerobic down to values between 8.66 and 2.66 kPa. These results differ from the observations made by Famme et al. (1981) for Mytilus edulis and by Hammen (1983) for Crassostrea virginica. These authors found that the two bivalve molluscs compensated for the decline in aerobic energy production down to very low oxygen tensions by increasing anaerobic metabolism, such that the overall heat dissipation was kept constant. This pattern of energy expenditure is associated with a high rate of substrate consumption. S. nudus, however, adjusts its rates of energy expenditure and substrate depletion to the availability of oxygen and is thus able to use its body stores economically.

At oxygen tensions below 8.7 kPa aerobic metabolism alone cannot satisfy the energy demands of S. nudus and additional anaerobic energy production becomes necessary (Fig. 3). This is consistent with the accumulation of anaerobic end products observed at 2.66 kPa (Tables 1, 2). From these results we conclude that the critical (the so-called P_cII) that indicates the onset of anaerobic metabolism must lie between 2.66 and 8.66 kPa. These data correspond well with those of Portner et al. (1985), who determined the critical to be close to 6.7 kPa in large specimens of S. nudus.

The overall heat production rate is further reduced with decreasing until it amounts to only 20 % of the normoxic value under extreme hypoxia. This value may have a small aerobic component because very low rates of oxygen diffusion into the experimental system cannot be ruled out. In these experiments the at extreme hypoxia was kept below the detection limit of the oxygen electrode (i.e. 66Pa=0.5 mmHg). At a flow rate of 0.1 ml min^-1 g^-1 fresh mass, complete extraction of the residual oxygen by the animals would yield an aerobic heat production of 2.2 mJ h^-1 g^-1 fresh mass, which is only about 10% of the measured total heat production of 21.5 mJ h^-1 g^-1 fresh mass. Thus, the anaerobic metabolic rate certainly amounts to 18–20 % of the normoxic value, which agrees well with values measured in other worms, such as the oligocheates Lumbriculus variegatus (23%. Gnaiger and Staudigl, 1987) and Tubifex tubifex (10–15 % ; Famme and Knudsen, 1984) and the polychaete Nereis diversicolor (10 % ; A. D. F. Addink, unpublished results).

Unfortunately, the turnover of ATP cannot be deduced from the total heat dissipation, because there is no constant relationship between these variables in aerobic and anaerobic pathways. The dissipated heat per unit of ATP turnover is approximately -80kJmol^-1 for fully coupled aerobic metabolism, but less than —40 kJ mol^-1 for the acetate-propionate pathway (Gnaiger, 1983b). Therefore, ATP turnover rates must be calculated from biochemical data, using stoichiometric coefficients of ATP production for aerobic and anaerobic glycogen degradation, as published by Gnaiger (1980b, 1983b) and Portner (1987a). For the glycolytic end products D- and L-alanine, octopine and strombine, 1.5 mol ATP per unit of product are derived from the Embden-Meyerhof-Parnas pathway. For succinate and propionate, formation via fumarate reductase has to be assumed for the sake of redox balance maintenance, which yields 2.5 and 3.5 mol ATP, respectively, per unit of product. Finally, acetate production is accompanied by the formation of 2.5 mol ATP per unit of product.

The metabolite flux through the Embden-Meyerhof-Parnas pathway can be calculated from the oxygen consumption and from the accumulation of anaerobic end products, assuming that glycogen is the only substrate used by S. nudus.

During normoxia the turnover rate of ATP is 1.6 umol h^-1g^-1 fresh mass and the metabolite flux through the Embden-Meyerhof-Parnas pathway amounts to 43.3 nmol glycosyl units h^-1g^-1 fresh mass, as calculated from oxygen consumption (see Table 6). The contribution of anaerobic metabolism can be excluded, because a caloric/respirometric ratio of 423.4±27.1 kJmol^-1O₂, found during normoxia, indicates fully aerobic metabolism.

When the animals are exposed to moderate hypoxia at a of 8.66 kPa, the ATP demand and the rate of glycogen depletion are reduced to 55 % of the normoxic values. From this result it is evident that under steady-state conditions of moderate hypoxia S. nudus reduces its energy expenditure. Except for the small amount of aspartate degradation (Table 1) and a small, but significant, accumulation of succinate, no marked anaerobic energy contribution can be seen. The slight anaerobiosis might be due to an anaerobic burst occurring at the beginning of moderate hypoxia (Figs 1, 2). It can therefore be concluded that, during moderate hypoxia, ATP is still provided by aerobic metabolism, albeit at a depressed rate.

At a lower steady state of oxygen provision the ATP turnover is further decreased to 37.5 % of the normoxic rate, but now 52 % of the total ATP yield is supplied by anaerobic processes (Table 6). Despite the pronounced depression of energy expenditure, the metabolite flux through the Embden-Meyerhof-Parnas pathway is increased compared to normoxic values. During acute hypoxia (⁠ nominally zero) a further depression of heat dissipation to 19.5 % of the normoxic rate occurs. Although the ATP turnover yields only 27.5 % of the normoxic value, the ATP requirement can only be met by an enhanced degradation of glycogen, which amounts to 178% of the normoxic flux. Thus, at least in S. nudus, there is no suppression of the Embden-Meyerhof-Parnas pathway during extreme hypoxia; instead, it is activated to almost twice its normoxic rate.

From these results one must conclude that under moderate hypoxia metabolic depression leads to a lower basal metabolic rate which is sustained by an aerobic energy metabolism. As a consequence, the animal will save substrate. When the critical oxygen tension (P_c) is passed, the animal can lower its energy expenditure still further, but it cannot save stored substrate, because the low ATP yield of anaerobic energy provision requires an increased degradation of glycogen. As a corollary, S. nudus may save substrate when the tide is receding and it may thus compensate for the enhanced glycogen degradation during acute hypoxia at the end of low tide. Overall, the degradation of glycogen during ebb may not be very different from that during normoxic conditions.

During the first hours of recovery a marked increase in oxygen uptake above control values was observed, whereas heat production was only slightly enhanced. This leads to a calorimetric/respirometric (CR) ratio of 335 kJ moF¹O₂, which is lower than the value for fully dissipative aerobic metabolism (450 kJ mol^-1O₂, Gnaiger, 1983a). In Mytilus edulis the CR ratio was 360 kJ mol^-1O₂ during the first hour of recovery from 72 h of anoxia (Shick et al. 1988). For recovering oyster larvae a value as low as 180 kJ mol^-10₂ has been observed (Widdows et al. 1989) CR ratios below 450kJ mol^-1 O₂ may indicate non-metabolic oxygen uptake, i.e. oxygenation of the haemerythrin store and saturation of the coelomic fluid. For S. nudus, an additional O₂ uptake of 0.47 _JumolO₂g^-1 fresh mass is necessary to saturate the coelomic fluid and reoxygenate the haemerythrin after extreme hypoxia (calculated using data from Portner et al. 1985). The occurrence of endothermic anabolic reactions, such as gluconeogenesis or phosphagen repletion, also results in significantly decreased CR ratios, down to 200 kJ mol^-1O₂ for succinate clearance for instance (Shick et al. 1988). In S. nudus, accumulated succinate is degraded within 6 h and the phosphagen is restored within 3 h after the end of anaerobiosis (Portner et al. 1986).

During prolonged recovery, anaerobic processes become involved, a phenomenon that has also been observed in M. edulis by Shick et al. (1986). The authors found a correlation between the frequency of contractions of the adductor muscle and anaerobic metabolic rate during recovery, which led to the conclusion that additional anaerobic energy is used for shell movements. In case of S. nudus increased respiratory peristaltic movements could require anaerobic energy provision.

Famme and Knudsen (1984) assumed that the results obtained by direct and indirect calorimetry were in agreement for Tubifex tubifex. Their calculations were criticized, however, by Gnaiger and Staudigl (1987), who stated that the caloric equivalent for excreted acids of -146 kJ mol^-1 acid used by Famme and Knudsen (1984) was too high. Using Gnaiger’s (1983b) values, the biochemically explained heat amounts to only 66% of the calorimetrically measured heat production. Gnaiger (1980a) found a similar discrepancy for the values in Lumbriculus variegatus. He explained this ‘anoxic gap’ by postulating unknown sources of anaerobic heat production. Gnaiger’s studies should only be considered as rough estimations, since he compared the measured heat production of L. variegatus with the release of fatty acids by Tubifex spp. (actually a mixture of different species) determined by Schottler and Schroff (1976). Even if the metabolism of the animals used is comparable, as assumed by Gnaiger (1980a), the differences in experimental conditions and seasonal variations might have caused the observed discrepancy. A more detailed study was carried out by Shick et al. (1983). They determined heat production and end product accumulation simultaneously in Mytilus edulis and found an unexplained heat fraction of 63 % between 3 and 48 h of anoxia. The authors explained the observed discrepancy by postulating additional enthalpy changes caused by protonation of CO₃^2- from the shell, which is observed under prolonged anoxia. But since the caloric coefficients for anaerobic end products employed by Shick et al. (1983) incorporated the enthalpy changes due to physiological proton buffering (Δ _bH_H+ = — 30kJmol^-1), this explanation has to be rejected (J. M. Shick, A. de Zwaan and A. M. Th. de Bont, personal communication).

A close agreement between the results obtained by direct and indirect heat production, as we found in the present study, implies that all essential processes in anaerobic metabolism have been considered. This is also confirmed by the recent finding that changes in the acid-base status in anaerobic Sipunculus nudus could be completely explained by the formation of anaerobic end products and the associated protons (Pörtner, 1987 b).

In conclusion, we have demonstrated that changes in oxygen consumption above P_cII reflect changes in total metabolism of the animals. We have thus verified our earlier study (Pörtner et al. 1985), which, based on biochemical analyses, led us to define P_c as the critical below which anaerobiosis starts. The use of non-invasive calorimetry also allowed quantification of the metabolic rate below P_c. Heat production under acute hypoxia could be attributed to the formation of known anaerobic end products. In accordance with calculations based on anaerobic metabolic changes, ATP expenditure decreased to 27 % of normoxic values and reflected the energy-saving strategies of the animals during anaerobiosis.

We thank Erich Gnaiger for critically reading the manuscript. Supported by the Deutsche Forschungsgemeinschaft (Gr 456/11-1) and Fonds der Chemischen Industrie (M.K.G.).