Coulometric Measurement Of Oxygen Consumption During Development Of Marine Invertebrate Embryos And Larvae

The majority of marine invertebrates have a life cycle that involves a larval stage (Thorson, 1950). Measurements of energy requirements during the larval stage have mostly been carried out using rates of oxygen consumption as a measure of metabolic rate (indirect calorimetry). Zeuthen (1947) used a modified Cartesian diver, and Scholander et al. (1952) used the reference diver technique to make some of the first measurements of the consumption of oxygen by invertebrate embryos and larvae. These techniques, however, depend on small volumes with high numbers of larvae per volume (Sprung, 1984). For example, Scholander et al. (1952) used approximately one individual per microlitre in the micro-respiration chamber, which is equivalent to 1000 individuals per millilitre. More recent studies have used highly sensitive (small cathode) polarographic oxygen sensors (Gnaiger and Forstner, 1983), which have permitted more routine measurement of the metabolic rates of small organisms enclosed in fixed volumes (Gerdes, 1983; Manahan, 1983; McDonald, 1988; Yasumasu et al. 1988; Jaeckle and Manahan, 1989; Shilling and Manahan, 1990). None the less, the concentrations of larvae required to produce a measurable decline in oxygen tensions are still high (e.g. 290 individuals ml^-1 based on 50–150 larvae per 345 μl respiration chamber, Jaeckle and Manahan, 1989). Another disadvantage of relying on decreasing oxygen tensions is the possibility that unknown stresses on the larvae result in atypical metabolic rates. This problem has been solved to some extent by flow-through respirometry systems that pass a continuous stream of fresh sea water through the experimental chamber. Very high concentrations of larvae (e.g. 30 000 in 25 ml, Wang and Widdows, 1991), however, are still required to obtain significant rates of oxygen use and heat dissipation (direct calorimetry) as the sea water flows through the respiration chamber.

Little information is available on the metabolic scaling of the developing stages of marine invertebrates (Zeuthen, 1953; Sprung, 1984; Sprung and Widdows, 1986; Manahan et al. 1989). This is largely because of methodological difficulties in working with small organisms and the limitations of extrapolating short-term measurements, obtained using polarographic oxygen sensors (POS) over the much longer time course of development. Questions concerning (i) the limitations of short-term oxygen consumption measurements, (ii) the possible stress effects of high numbers of animals per respiration chamber, and (iii) the effect of low oxygen tensions can be addressed using coulometric respirometry.

Coulometric respirometers measure oxygen consumption by the quantitative, electrolytic replacement of the oxygen used by an organism during an experiment (Heusner et al. 1982). The amount of oxygen added to the respiration chamber is continually recorded and is used to calculate the metabolic rate of the organism under normoxic conditions. The coulometric respirometer offers several advantages over previous techniques used to measure the metabolic rate of small animals. The coulometric respirometer continually replaces the oxygen used by an organism, allowing metabolic rates to be measured under stable normoxic conditions that can continue for periods of hours to days. In addition, the coulometric respirometer is, under the conditions reported here, about 100-fold more sensitive than the more widely used POS respirometer. Hence, the number of animals used in experiments can be drastically reduced (e.g. to 1–4 larvae ml^-1). Finally, coulometric respirometry allows for the measurement of metabolic rates over developmentally relevant periods (hours to days), thereby providing data on the total amount of energy used by a developing organism. Such long-term measurements are either difficult or impossible to obtain using previous methodologies.

Adult asteroids, Asterina miniata (Brandt), and echinoids, Dendraster excentricus (Eschscholtz), Strongylocentrotus purpuratus (Stimpson) and S. franciscanus (Agassiz), were induced to spawn by intracoelomic injection of either 1 mmol l^-1 1-methyladenine (asteroids) or 0.5 mol l^-1 KCl (echinoids). All culturing and experiments were carried out in sea water that had been passed through a 0.2 μm (pore-size) Nuclepore filter. Following fertilization, zygotes were transferred to 50 l culture vessels which were stirred (approximately 10 revs min^-1) using small rotating paddles (30 cm×30 cm paddle area). Fertilization success ranged between 95 and 100 %. Sea water in the culture vessels was changed every fourth day, and algal food (10⁴ cells ml^-1, Duneliella tertiolecta Butcher) was added every second day starting at 48 h after fertilization. Cultures were maintained at 14–15 °C, except for some cultures of D. excentricus which were maintained at 20 °C. Encysted embryos of Artemia sp. (San Francisco brand) were raised to the larval stage in sea water at 20 °C. Veligers of the Pacific oyster Crassostrea gigas (Thunberg) were supplied by the Coast Oyster Company (Quilcene, Washington). Veligers ranged in size from newly formed D-veligers (valve length=82.0±0.68 μm, S.E.M.) to competent pediveligers (323.0±5.57 μm). Veligers were kept in autoclaved sea water for 24 h (20 °C) after being shipped from Washington, USA, and prior to use in experiments in California, USA. Nauplii and adult copepods (Acartia tonsa) were caught, using plankton nets, from Fish Harbor (San Pedro, California) and were kept in sea water (15 °C) for 24 h prior to any measurements.

Biomass is defined here as ash-free dry organic mass and was measured using a mass-calibrated electrobalance (Cahn, model 29). Eggs, embryos or larvae (200–500 individuals used per determination) were washed three times in isotonic ammonium formate solution (3.4 %, 0.2 μm filtered) and added to small aluminium vessels (pre-ashed at 500 °C for 6 h) prior to being dried at 58 °C to constant mass (which occurred within 10 days). Samples were weighed (total dry mass) and then ashed at 458 °C for 6 h, and weighed again (ash mass). Biomass was calculated as the difference between the total dry mass and the ash mass. For C. gigas, the relationship between valve length (l, μm) and biomass (m, μg dry organic mass) was: logm = 2.82logl - 6.594 (r²=0.99, N=30). Several controls were run to ensure that the aluminium vessels were completely ashed and that complete sublimation of ammonium formate had occurred. Ammonium formate solution (100 μl) was added periodically to a number of ashed aluminium vessels and the vessels treated as above. The mass differences in these control vessels was less than 2 % of typical ash-free dry organic masses seen during mass determinations (data not shown).

The rates of oxygen consumption of embryos and larvae at various stages of development were measured using a coulometric respirometer. All respirometry measurements (coulometric and other techniques) were carried out at the temperature at which the embryos and larvae were cultured.

The design of our system was based on that described by Heusner et al. (1982). Our solid-state current generator supported and collected data from four independent respirometry chambers. Data from each respiration chamber were transmitted from the coulometric device via an RS-232 port to a personal computer, where they were loaded into standard worksheet programs. The thermobarometer (Fig. 1A) and respiration chambers (Fig. 1A,B) were constructed according to Heusner and Tracey (1984). Each respiration chamber was placed within a thermobarometer chamber (Fig. 1A) and, following temperature equilibration (2–3 h), the thermobarometer was sealed using a greased glass stopcock. The animals were placed in the 2.5 ml respirometer chamber (Fig. 1B). As the animals respired, oxygen was removed and the released CO₂ was captured by a CO₂ absorbent. The net drop in partial pressure in the respirometer chamber resulted in an increase in the height of the saturated CuSO₄ solution which, as a result, made contact with a sharpened platinum anode. Oxygen was released upon contact of the saturated CuSO₄ solution with the anode, through which current was passed from a precision capacitor (1.6704 μF in our device). The platinum anodes were sharpened electrolytically as described by Wolbarsht et al. (1960) to reduce the breaking tension between the anode and CuSO₄ solution and thus to ensure a regular rate of capacitor discharge.

The coulometric respirometers were placed vertically within a 40 l insulated acrylic holding tank. Four respiration chambers (each within a separate thermobarometric chamber and each containing 2 ml of sea water) were used simultaneously. The concentration of animals in the respiration chambers ranged from 1 to 2000 ml^-1 depending on the species and the experiment. One chamber in each experiment contained filtered sea water as a control. The temperature of the holding tank had to be kept to within ±0.01 °C (see Fig. 1C) as coulometric respirometers are highly sensitive to changes in partial pressure caused by changes in temperature (Heusner et al. 1982). Temperature control was achieved by placing a circulator with proportional heating control (model N-3, Haake) in the 40 l holding tank. Antagonistic cooling was provided by passing water from an external circulating waterbath (model RDL40, Precision Scientific Group) through a cooling coil (mounted on the Haake circulator). Water in the cooling coil was 5 °C lower than the experimental temperature set-point.

Contact of the anode with the CuSO₄ solution results in the release of one oxygen molecule for every four electrons delivered to the anode (Heusner, 1970). Heusner and colleagues (Heusner, 1970; Heusner et al. 1982; Heusner and Tracey, 1984) calibrated coulometric respirometers by using the ideal gas law. The calculated oxygen release (in these cases) was based on the assumption that all electrons delivered to the anode are used in oxygen production. During calibration, the pressure within the respirometer chamber was decreased and the number of electrons delivered to the anode (as capacitor discharges) recorded. The volume withdrawn from the respirometry chamber (the method used to decrease pressure by Heusner et al. 1982) was measured by monitoring the movement of a bead of mercury moved within a calibrated capillary tube using an optical comparator. In the present study, we also used the ideal gas law in our calibration procedure, but describe a new method that uses a solid-state pressure sensor that permits rapid and sensitive determination of the small volume changes. In addition, we have also tested the assumption that all electrons delivered to the anode produce oxygen. The number of electrons being delivered to the anode per unit time was controlled by the coulometric device and recorded as the number of capacitor discharges. The number of capacitor discharges was converted into the number of electrons from the electrostatic expression: q=CΔV; where q is charge, C is capacitance and ΔV=V₀–V_t (where V₀ is the voltage before discharge and V_t is the voltage after discharge) and Faraday’s Law (to convert charge to moles of electrons). In the present study, the theoretical yield (within 0.05 %) was 4.20 pmol O₂ per capacitor discharge (C=1.6704 μF, ΔV=0.970±0.0005 V).

The calibration device in the present study (Fig. 2A) consisted of a solid-state pressure sensor (NPH82.5D, NovaSensor, California) mounted between two sealed chambers. In this device, the upper chamber served as a thermobarometric reference, whereas the lower chamber enclosed a small air space (less than 500 μl) above a saturated solution of CuSO₄. Both the anode and cathode were in contact with this solution, with the anode passing through the solution within a glass jacket and just coming into electrical contact with the surface. The cathode was in contact with the solution at the base of the lower chamber. The signal from the pressure sensor was amplified by circuitry employing an instrumentation amplifier chip (Fig. 2B; LM363, National Semi-conductor Corporation, California) and the resulting voltage recorded using a 16-bit A-to-D convertor and data acquisition software (Datacan IV, Sable Systems, California). The calibration chamber and part of the electronic circuitry (Fig. 2B, dashed box) were placed within a water-tight container and immersed in the constant-temperature holding tank described above. A calibrated microsyringe (1.0 μl total volume with 1.0 nl intervals, Unimetrics, Illinois) was connected to the lower chamber and used to change the volume, thereby allowing the conversion of voltage output from the pressure sensor to volume changes (Fig. 3A,B). Following calibration of the pressure sensor, current from the coulometric device (as number of capacitor discharges) was delivered to the anode and the resulting change in volume was recorded (Fig. 3C). From these data, the amount of oxygen liberated (assuming all electrons are used at the anode in the production of oxygen) was calculated using the molar volume of gas at the measurement temperature (Fig. 3D). The amount of oxygen (predicted by the pressure sensor technique assuming all the gas produced is oxygen) released by the anode per discharge was 4.19 pmol (see regression equation, Fig. 3D legend). This value was similar to the theoretical yield of 4.20 pmol O₂ per discharge (see calculations above of theoretical yield).

To test the assumption that all electrons delivered to the anode are used in the production of oxygen, the gas produced at the anode was collected in a stream of oxygen-free nitrogen. This was done using a respiration chamber (Fig. 1B) with the following modifications. The anode and the cathode were kept in electrical contact using a wick (Whatman filter paper) soaked in saturated CuSO₄. No CO₂ absorbent was necessary. Nitrogen gas was supplied at a constant rate (using a mass flow meter, model FMA 2DPV, Omega Engineering Inc., Connecticut, USA) at the top end of the manometric glass tube, and the gas stream drawn from the bottom end of the tube at a constant flow rate over the anode (Flow Controller, model R-2, Applied Electrochemistry Inc., California, USA). The oxygen produced by the anode (after a known number of capacitor discharges) was measured using an oxygen sensor and analyzer (model N-37M, Applied Electrochemistry Inc., California, and model S-3A, Ametek Thermox, Pennsylvania, USA). The output signal was recorded using the Datacan data acquisition system. Fig. 3E shows an actual tracing of the pulse of oxygen produced during 7139 capacitor discharges. Similar measurements were made over a range of discharge events from 0 to 60 000 capacitor discharges per pulse (Fig. 3F). The measured amount of oxygen produced per individual capacitor discharge was calculated from the slope of the regression (given in the legend to Fig. 3F) and found to be 4.18 pmol O₂ per discharge (cf. pressure sensor=4.19 pmol O₂ per discharge; theoretical yield=4.20 pmol O₂ per discharge). These data indicated that the original assumption underlying the use of coulometric respirometry in metabolic studies was true (i.e. all electrons used at the anode produce oxygen).

The rates at which oxygen was consumed by larvae of Artemia sp. (2-day-old nauplii), Dendraster excentricus (six-arm plutei) and Asterina miniata (bipinnaria) were measured independently using coulometric respirometry and two other methods. Larvae used with each of the three methods were at the same stage of development and taken from the same culture. Metabolic rates were measured using (i) coulometric respirometry, (ii) Winkler’s titration of dissolved oxygen, and (iii) polarographic oxygen sensors (POS). The latter system consisted of three oxygen meters (model OM 780, Strathkelvin Instruments, Glasgow, UK), each connected to an oxygen sensor (model SI 1302) located either at the base of a 100 μl (model RC 200) or at the top of a 300–1000 μl (model RC 300) micro-respirometer chamber. The outputs from all three meters were recorded simultaneously using a data acquisition system (Datacan). Oxygen consumption rates were calculated from the rate of decline in oxygen concentrations over a 30 min period following a 30–60 min period that allowed the animals to acclimate to the respirometer chamber. Calibration of the POS to an oxygen-free solution was carried out at the beginning of each experiment in sea water flushed with nitrogen. POS were calibrated with saturated sea water, the oxygen content of which was measured using the Winkler’s assay (Parsons et al. 1984). For all cases, the measured amount of oxygen was within 5 % of the value calculated using corrections for temperature, salinity and atmospheric pressure. The concentration of larvae in the chambers ranged from 30 to 400 per 100 μl for most experiments, and from 214 to 809 per 1000 μl for measurements testing the effect of the concentration of larvae on metabolic rate (see Results). Background oxygen consumption rates (filtered sea water only) were interspersed between readings taken in the presence of animals and were less than 0.5 % of a typical reading. Leakage of oxygen into the chambers when they were sealed was also negligible (tested by measuring the rate of change in oxygen-depleted sea water). The temperature of each chamber was controlled to within ±0.02 °C using an external circulation water bath (model RDL20, Precision Scientific Group). For the Winkler’s method (Parsons et al. 1984), seven containers (300 ml, biological oxygen demand bottles) each containing larvae of a given species (per ml: Artemia sp., 21 larvae; D. excentricus, 20 larvae; A. miniata, 16 larvae) were incubated for up to 36 h. At intervals of 4–5 h (the first bottle at time zero), the oxygen content of an individual bottle was measured. The rate at which oxygen was used was calculated per larva per unit time by subtracting the oxygen content of each bottle from the oxygen concentration at time zero.

Rates of oxygen consumption measured with the coulometric respirometer were stable for at least 20 h, after an initial period of 3–4 h to allow for temperature equilibration (Fig. 4A). The sensitivity of the coulometric respirometer is illustrated by the measurement of the metabolic rate of a single copepod swimming freely in 2 ml of sea water (Fig. 4B). The response of the respirometer was proportional to the number of animals per respiration chamber (Fig. 4C,D). The conditions within the respiration chamber allowed adults (copepods) and larvae to survive for several days. Development within the chambers also proceeded normally for days, from newly fertilized eggs to the formation of feeding larvae (e.g. pluteus, Fig. 5A,B,C). Larvae in the respirometry chambers (inspected at the end of each experiment) reached developmental stages identical to those of siblings reared under normal culture conditions, indicating that conditions in the respiration chamber did not affect the rate of development. Development rates and metabolic rates were consistent among chambers when measurements were made simultaneously with three coulometric respirometers (Fig. 5B,C). Immediately after fertilization, metabolic rates increased steadily. At hatching, the rate of oxygen consumption began to increase very rapidly to a peak. Approximately 5 h after the beginning of hatching, metabolic rates decreased and returned to rates similar to those just prior to hatching (Fig. 5A,B,C).

The influence of the concentration of animals per coulometric chamber was investigated for larvae of Crassostrea gigas and Dendraster excentricus. In both cases, there was no significant change (least-squares linear regression, Fig. 6A,B) in the rate of oxygen consumption as a function of increasing concentrations of larvae (e.g. for C. gigas from 4 to 400 individuals ml^-1). Increasing the concentration of animals had no effect on the metabolic rate of C. gigas (241–809 larvae ml^-1) measured using the POS respirometer (Fig. 6C).

Metabolic rates measured with three different techniques were compared (Fig. 7A; given as a percentage of the Winkler titration value; see legend for absolute rates used in statistical analyses). The measured metabolic rate of bipinnaria larvae of Asterina miniata did not differ (analysis of variance, ANOVA) when assessed with the coulometric respirometer or the Winkler’s method [variance ratio, VR=0.52, F_0.05(1,10)=4.96]. The metabolic rate of the same larvae measured using the POS respirometer was, however, significantly lower [VR=14.62, F_0.001(2,15)=11.3] than that measured using the two other methods. The same pattern of comparison held for pluteus larvae of Dendraster excentricus. Again, rates measured with the coulometric respirometer did not differ from those yielded by the Winkler’s method [VR=0.51, F_0.05(1,7)=5.59], whereas rates measured with the POS respirometers differed from the other two [VR=14.62, F_0.001(2,15)=11.3]. There were no significant differences in the metabolic rates of Artemia sp. measured with the three techniques [VR=2.77, F_0.05(2,12)=3.89].

In the case of C. gigas, where metabolic rates were measured with coulometric and POS respirometers, rates measured with the latter technique were lower throughout the entire lifespan of the larvae (Fig. 7B). Both the mass exponents [VR=27.47, F_0.001(1,111)=12.0] and mass coefficients [VR=259.67, F_0.001(1,111)=12.0] were different for the metabolic rates measured with the two techniques.

The metabolic rate of Artemia sp. decreased rapidly as the concentration of oxygen in the 100 μl chambers decreased as the larvae consumed oxygen in the POS respirometer (Fig. 8A). In contrast, there was relatively little influence of decreasing oxygen concentration on the metabolic rate (down to 20 % of full saturation) of the larvae of Dendraster excentricus (Fig. 8B) and Asterina miniata (data not shown). It is also notable that metabolic rates of Artemia sp. were only similar in coulometric and POS respirometers at oxygen saturation (Fig. 8A; the open circle represents coulometric data from Fig. 7A). All measurements of metabolic rates with the coulometric respirometer were carried out at oxygen concentrations near saturation. The production of oxygen by the anode in the coulometric respirometer maintains the partial pressure of oxygen in the air space above the sea water in the respirometer chamber. Under these conditions, the coulometric respirometer is capable of maintaining normoxic conditions. In the data shown in Fig. 8C, a polarographic oxygen sensor was inserted into the bottom of a modified respiration chamber which contained 862 larvae in 2 ml of sea water (C. gigas).

Throughout a 12 h experiment, the oxygen concentration of the sea water in the respiration chamber was maintained at 88.3 % of air saturation despite the large oxygen demand imposed by the high concentration of larvae.

The relationship between the increase in biomass (ash-free dry organic mass) during growth and the corresponding increase in metabolic rate (measured with coulometric respirometry) is given in Fig. 9 for larvae of three species. The mass exponents of the scaling equations describing this relationship were high and close to unity (0.9–1.1). Therefore, metabolic rate increases linearly (slope=1) with increasing biomass for growing larvae of Dendraster excentricus (Fig. 9A), Asterina miniata (Fig. 9B) and Crassostrea gigas (Fig. 9C).

Accurate measurement of the metabolic rate of embryos and larvae is fundamental to studies of the energy requirements of developing marine invertebrates. In this study, coulometric respirometry was used to measure the metabolic rate of embryos and larvae that were growing and differentiating. Coulometric respirometry allows measurements of metabolic rates over long periods (from fertilized egg to feeding larva) and has no effect on the course and timing of development. Assessing the effect of the measurement system on embryos and larvae is an important but often neglected feature of most other techniques used to measure respiration during development. Fertilized eggs placed in the respirometer developed to the larval stage at rates that were similar to those of sibling larvae reared in culture, showing that measurements made with this technique are on normal and actively developing animals.

The need to use high concentrations of larvae to measure metabolic rates has long been a ‘sore point’ (quote from Sprung, 1984; see also Walne, 1966; Millar and Scott, 1967; Crisp, 1976; McDonald, 1988). Reducing the concentration of larvae in these other systems is, however, a problem. Crisp (1976), for example, pointed out the unreliable nature of respiration data obtained for large dilute cultures of larvae. In his re-evaluation of the data of Millar and Scott (1967), Crisp showed that the larvae were not the largest sink for oxygen at low concentrations of larvae and, hence, that the higher oxygen consumption per larva in the most dilute cultures was an artefact. McDonald (1988) included controls to correct for oxygen consumption not attributable to larvae and yet still reported a 35 % reduction in oxygen consumption of bivalve larvae as the concentration of individuals increased. His data, however, show that the relationship between decreasing larval concentrations and increasing metabolic rates is a power function with an asymptote approaching infinity (at zero larvae). The conclusion from his data is that the respiration of a single larva in dilute culture is many times greater than that of larvae at high concentrations.

We investigated the effect of increasing the concentration of larvae on metabolic rates of echinoid and bivalve larvae (Dendraster excentricus and Crassostrea gigas) using both the coulometric and POS respirometers (the latter for C. gigas only). After correcting for the very low background rates of oxygen consumption in the control chamber, metabolic rates (in the coulometric respirometer) remained constant for both species over a 100-fold increase in larval concentration (Fig. 6A,B). A similar conclusion was obtained for metabolic rates measured using the POS respirometer and concentrations of bivalve larvae ranging from 200 to 800 larvae ml^-1 (Fig. 6C). The conclusions of our study differ (for bivalve larvae) from those of McDonald (1988) and Millar and Scott (1967), but are similar (for Artemia sp.) to those of Varo et al. (1993a), and suggest that the effect of larval concentration may depend on conditions associated with the technique used.

The data obtained for the POS respirometer (Fig. 6C) suggest that the lower estimates of metabolic rate obtained using this system were not due to reduced oxygen tension in the respirometer chambers. These data were collected using a 1 ml respiration chamber for the POS respirometer. The metabolic rates of larval C. gigas at the minimum and maximum concentrations did not differ. For example, at 241 larvae ml^-1, the metabolic rate was 131.4 pmol larva^-1 h^-1, similar to the value of 137.3 pmol larva^-1 h^-1 measured for a concentration of 809 larvae ml^-1. These data for C. gigas also allow calculation of the time required for different concentrations of larvae to deplete fully the oxygen in the POS chambers. Given a starting oxygen concentration of 235 μmol l^-1, and bivalve larvae with metabolic rates calculated from the average of the two rates above, it would take between 2.1 and 7.2 h for larvae present at the highest and the lowest concentrations used in this study to use completely the oxygen in a 1 ml chamber. In all these measurements, the larvae were in a sealed 1 ml chamber for 30–90 min (the time required for making a typical measurement). Therefore, at most (highest concentration of larvae, 809 ml^-1), approximately 70 % of the oxygen was depleted during the course of a 90 min measurement. That this extent of depletion had no effect on the metabolic rate of the larvae is evident in the comparison of these rates with rates measured at the lowest concentrations of larvae, where only 21 % of O₂ had been depleted. As has been previously reported (e.g. Widdows et al. 1989), hypoxia does not affect the metabolic rates of bivalve larvae until low oxygen partial pressures are reached. This was also the case for the echinoderm larvae we studied (e.g. Fig. 8B). Larvae of Artemia sp. have been reported to be respiratory regulators (Varo et al. 1993b; De Wachter et al. 1994, and references therein), although embryos do down-regulate metabolism under anoxia (Hand, 1990; Hontoria et al. 1993). Varo et al. (1993b) reported that the ability of Artemia larvae to regulate respiration was reduced with changes in salinity. In our experiments in full-salinity sea water, the larvae of Artemia sp. were not respiratory regulators (Fig. 8A). In sealed POS respiration chambers, the metabolic rate of Artemia decreased with the decrease in oxygen concentration in the chamber caused by the animals’ respiration. We used a different strain of Artemia sp. to those used in the study by Varo et al. (1993b). Perhaps the different abilities of Artemia larvae to regulate respiration in declining oxygen concentrations is straindependent.

The metabolic rate of larvae increased with mass during growth, as has been described for a number of species (Zeuthen, 1953). Mass exponents for echinoderm (Dendraster excentricus, Asterina miniata) and mollusc (Crassostrea gigas) larvae were close to 1.0 and agree with earlier studies (reviewed by Zeuthen, 1953), which concluded that the metabolic rates of embryos and larvae of marine animals scale with a mass exponent 0.95. However, when metabolic rates of C. gigas were measured with polarographic oxygen sensors (POS) in small (100 μl) chambers, the mass exponent was only 0.55 (Fig. 7B) and statistically different from the mass exponent of 0.953 obtained with the coulometric respirometer (Fig. 9C). These data demonstrate that metabolic rates measured with the POS respirometer were proportionally lower for older stages of C. gigas and support the suggestion that conditions associated with the confinement of larvae within the POS chambers (which are presumably exacerbated for larger larvae) may explain the deviation of POS measurements from those using the coulometric technique. These comparative data highlight the importance of the measurement technique in establishing the mass exponent and may help to explain the diversity of mass exponents reported in the literature for larvae and early juvenile stages of marine animals (bivalve, 1.07–1.18, Mann and Gallager, 1985; bivalve, 0.59–0.90, Sprung, 1984; copepods, 0.62–1.14, reviewed by Conover, 1968; fish, 0.65–1.69, reviewed by Giguere et al. 1988).

In his review of the metabolic rates of marine invertebrate larvae, Crisp (1976) concluded that the majority of larvae have metabolic rates in the range 3–10 ml O₂ h^-1 g^-1 dry mass. Crisp also obtained a similar range for metabolic rates calculated indirectly from the rate of loss of energy reserves in starving larvae. Assuming that 50 % of total dry mass is ash, Crisp’s range of values (corrected to the units used in our study) is 268–893 pmol O₂ h^-1μg^-1 dry organic mass. Most of the studies since Crisp’s review have supported this range. For example, Widdows et al. (1989) report that the metabolic rate of bivalve larvae (Crassostrea virginica) is 300 pmol O₂ h^-1μg^-1 dry organic mass using direct and indirect calorimetry. Erickson (1984), using a gradient diver, reported metabolic rates of 701 pmol O₂ h^-1μg^-1 dry organic mass for gastropod larvae. However, there are data that fall outside the lower boundary of this range; e.g. 57 and 196 pmol O₂ h^-1μg^-1 dry organic mass (Wang and Widdows, 1991; Mann and Gallager, 1985, respectively). Our values, despite differences in temperature and species, are all consistent with the range reported by Crisp (1976). In our study, calculations using data obtained with coulometric respirometry predict that a 1 μg larva of Dendraster excentricus, Asterina miniata (both at 15 °C) or Crassostrea gigas (at 20 °C) would consume 330, 344 or 369 pmol O₂μg^-1 h^-1, respectively (values calculated from scaling equations given in Fig. 9A,B,C). As with other organisms (Heusner, 1991), the metabolic explanations for the values of the mass coefficients (metabolic intensities) are unknown.

It is important to note that the values used by Crisp for his comparisons of metabolic rates were obtained using techniques other than polarographic oxygen sensors. If the rate of oxygen consumption in our study is calculated for a 1 μg larva of C. gigas (at 20 °C), using data from the POS respirometer (Fig. 7B), the metabolic rate would be 72 pmol O₂ h^-1, or 20 % of the coulometric value. Rates measured using the POS respirometer were 34 % of the coulometric rate for the two species of echinoderm studied (A. miniata, D. excentricus). Varo et al. (1993a) also found that measurements made with the same POS system as used in our study were lower (by 10 %) than measurements made with other techniques, in their case a Gilson differential respirometer.

What might be the basis of the lower estimates of larval respiration obtained with the POS respirometer? Errors in our use of the POS are unlikely. We followed all of the manufacturer’s specifications, and the standard procedures used for calibration of POS, as in previous studies of larval respiration in this laboratory (e.g. Jaeckle and Manahan, 1989, 1992; Shilling and Manahan, 1990) and others (e.g. Varo et al. 1993a). We found no significant difference in our comparison of methods used to measure respiration on the same group of Artemia larvae (Fig. 7A), which suggests that POS measurements can be accurate for certain animals. We have eliminated the possibilities that declining oxygen or larval concentrations might be causes of the low measurements by POS of larval respiration. We have also tested the possibility that the POS respiration chamber itself (e.g. O-ring, presence of KCl reservoir) might inhibit development and found that it did not, although embryonic development was arrested when current was applied to the POS (data not shown). We can offer no explanation as to why the POS gives low rates of respiration for all species and stages tested other than Artemia. Further studies that use POS for measurements of metabolic rate in larvae of benthic marine invertebrates should proceed with caution on a case-by-case basis, as no single calibration factor can be used to correct between the different methods (e.g. POS and coulometric respirometers: Fig. 7B). We conclude that coulometric respirometry offers greater accuracy and versatility for measuring the metabolic rate of developing larval forms of benthic marine invertebrates.

Our special thanks to Dr A. A. Heusner for his advice and encouragement during the construction of the coulometric respirometer. We are grateful to Drs G. A. Bartholomew and

J. R. B. Lighton for use of their flow-through respirometry system that we used for part of the calibration of our coulometric respirometer. This research was supported by grants from the Office of Naval Research (N00014-90-J-1740) and the National Science Foundation (OPP-8820130).