Partitioning of Oxygen Uptake between the Gills and Skin in Fish Larvae: A Novel Method for Estimating Cutaneous Oxygen Uptake

The skin is an important site of respiratory gas exchange in fish, particularly during embryonic and larval development. It is the sole site of gas exchange during embryonic development in oviparous species. The relative importance of the skin begins to decline once the gills become functional. Many species, however, do not form gills until fairly late in larval development, and in these species the skin persists as the only site of gas exchange well into the larval stage (Rombough, 1988a). The skin continues to play a significant role in gas exchange even after the gills have formed. Salmonid larvae, which hatch with relatively well-developed gills compared with most species, still obtain approximately 40 % of their oxygen via their skin at the end of the larval stage (Rombough and Ure, 1991; Wells and Pinder, 1996). Adults of water-breathing fish typically obtain 10–20 % of their oxygen across the skin, but values as high as 30 % have been reported (Feder and Burggren, 1985).

Although the importance of cutaneous gas exchange during larval development has long been recognized, little is known about the mechanisms involved. Indeed, it is only recently that attempts have been made even to measure the rate of cutaneous oxygen uptake in fish larvae. Part of the problem in studying cutaneous gas exchange in larvae has been the difficulty in distinguishing cutaneous oxygen uptake from that taking place across the gills. With adult fish, the solution has been to use some kind of membrane (usually a rubber dam) to isolate water surrounding the skin from that flowing over the gills (e.g. Kirsch and Nonotte, 1977). This approach has been used successfully with larvae of chinook salmon Oncorhynchus tshawytscha (Rombough and Ure, 1991) and Atlantic salmon Salmo salar (Wells and Pinder, 1996). However, the larvae of both salmon species are considerably larger (30–100 mg wet tissue mass at hatching) than those of most fish and it is unlikely that the technique can be applied to species such as plaice or zebrafish where larvae weigh less than 1 mg at hatching. The rubber dam method is not well suited for looking at regional differences in oxygen uptake and, because of the way the dam is placed, the pattern of water flow over the skin is not representative of what occurs in nature. Flow pattern may be very important to the efficiency of cutaneous uptake. Many larvae orient themselves so that they face into the current. Liem (1981) reported that Monopterus albus larvae were able to extract a significantly greater fraction of oxygen from the water if the flow was in an anterior-to-posterior direction (countercurrent to the blood flow in the vitelline veins) than if the flow was in a posterior-to-anterior direction. With the rubber dam method, it is virtually impossible to control the direction or velocity of water flow. The rubber dam method has the additional disadvantage that it is relatively stressful. Even when large, robust larvae such as those of salmon are used, not all individuals survive. This raises the possibility that results obtained using this method might not be fully representative of what takes place in unrestrained animals.

The accuracy of the boundary layer method was evaluated by comparing it with the rubber dam method. The rubber dam method only works with fairly large larvae, so rainbow trout Oncorhynchus mykiss was used as a test species.

Rainbow trout Oncorhynchus mykiss (Walbaum) were obtained as eyed embryos from the Rockwood Fish Culture Centre operated by the Canadian Department of Fisheries and Oceans at Gunton, Manitoba. Eggs were transferred to Brandon University, where they were reared in fresh water (pH 7.9; hardness 136 mg l⁻¹ as CaCO₃) in the dark until hatching. Eggs initially were kept at 2–3 °C to slow development and delay hatching. Eggs were transferred to 10 °C to speed up development 5–15 days before they were needed for experiments. This allowed a sequential series of experiments to be carried out using the same group of animals. Once transferred to 10 °C, eggs and larvae were kept at that temperature (±0.1 °C) until experiments were completed.

Two groups of newly hatched larvae (1–5 days posthatch) from the same brood stock were used in this study; one in the rubber dam study and one for the boundary layer experiments. The larvae used in the rubber dam experiments were significantly larger (P>0.05) than those used in the boundary layer experiments (Table 1). In terms of the relative amount of yolk consumed, however, the two groups of larvae were at almost exactly the same stage of development (43.3 % versus 46.7 % tissue; Table 1).

Cutaneous oxygen uptake was estimated for the larger trout larvae (the 50 mg tissue mass group) at 10 °C using the rubber dam method. Equipment and procedures were similar to those employed previously to measure cutaneous oxygen uptake in chinook salmon larvae (Oncorhynchus tshawytscha; see Rombough and Ure, 1991, for details). The major difference was an improvement in respirometer design. The respirometer in the present study consisted of two identical 6.46 ml glass chambers held together by pinch clamps. A 0.25 mm thick rubber dental dam was used to separate the chambers. A miniature (2 mm tip diameter) polarographic oxygen electrode (model MI 730, Microelectrodes Inc., Londonderry, NH, USA) was inserted into a small opening at the far end of each chamber. Stopcock grease and O rings gave a water-tight fit. The electrodes were connected to an OM4 oxygen meter (Microelectrodes Inc., Londonderry, NH, USA), and the output was recorded on a flat-bed chart recorder. Electrodes were calibrated to air and to 100 % N₂. Each chamber contained a small magnetic stirrer to ensure that the water was well mixed. The assembled respirometer was immersed in an insulated 45 l water bath held at 10.0±0.1 °C by a refrigerating circulator.

Larvae were anaesthetized using 50–75 mg l⁻¹ tricaine methanesulphonate (MS222). A small hole was cut in the rubber dam and the larva inserted so that the membrane gripped the body immediately posterior of the pectoral fins. This isolated the gills on one side of the membrane and approximately 85 % of total skin area including the yolk sac on the other side (Table 1). The rubber dam was mounted between the two chambers of the respirometer, and the respirometer was closed. Tests were terminated after approximately 2 h or if oxygen levels in one of the chambers dropped below approximately 50 % of air saturation. Oxygen uptake was calculated from the rate at which in the chambers declined using the longest linear portion of the polarographic record exclusive of the first 15 min of the recording. Uptake rates were adjusted to compensate for bacterial oxygen consumption. In addition, a series of tests was carried out on unrestrained fish to determine whether the stress of being placed in the rubber dam had a significant effect on their overall rate of oxygen consumption. Larvae were removed from the respirometer at the end of the test and placed in a recovery tank without anaesthetic. Test results were not used if the larvae did not recover within 30 min.

Viable larvae from the respirometry study were killed at the end of the recovery period with an overdose of MS222 and preserved in 10 % neutral buffered formalin. Larvae were kept in the preservative for a minimum of 30 days to ensure that shrinkage was complete before any morphometric measurements were made. At the end of the shrinkage period, larvae were weighed to the nearest 0.1 mg. Each larva was dissected into tissue and yolk components, and the two components were weighed separately.

For the purposes of this study, total cutaneous surface area was subdivided into head area, pectoral fin area, yolk sac area and trunk area. Head area was defined by the placement of the rubber dam in the respirometry experiments. Trunk area included the area of the trunk without the yolk sac and the area of the fin folds. Head plus pectoral fin area represented the fraction of total cutaneous surface area in the anterior compartment of the respirometer, while yolk sac plus trunk area was the fraction in the posterior compartment.

Methods for estimating the surface area of the various components were similar to those used by Rombough and Moroz (1990) to measure the cutaneous surface area of chinook salmon larvae. The major difference was that calibrated microscope images of the various components were transferred directly to the computer through a digital camera instead of first being traced onto a digitizing tablet. The microscopic images were obtained from temporary mounts of the various components. Temporary mounts of the yolk sac epithelium were produced by carefully dissecting the yolk sac from the trunk, cutting it into small pieces and flattening the pieces with a coverslip. Iridectomy scissors were used to remove the pectoral fins from the body. The head was then cut from the body immediately posterior of where the pectoral fins had been. A midsagittal cut was made dividing the head into left and right halves. The two halves were mounted medial side down on a glass slide and gently flattened using a coverslip. The trunk was cut transversely into three roughly equal-sized pieces. The two anteriormost pieces were sectioned with a midsagittal cut and mounted medial side down. The posteriormost trunk section consisted mostly of the caudal fin and was thin enough that it did not require midsagittal sectioning. Still images of the various components were taken using a digital camera (model 100C; Pixera Corp.) attached to a Wilde M1 dissecting microscope. The images were stored on computer and analyzed using the SigmaScan for Windows 2.0 computer program (Jandel Scientific).

Oxygen microelectrodes were used to measure profiles in the diffusional boundary layer at seven locations (one on the head, two on the yolk sac, two on the trunk, one at the base of the dorsal fin-fold and one on the proximal region of the caudal fin-fold) on the body of recently hatched (less than 5 days posthatch) rainbow trout (see Fig. 1). Equipment and procedures were similar to those described in Rombough (1992). was measured using Diamond General 737 Clark-type electrodes with tip diameters ranging from 50 to 95 μm mounted in a micromanipulator and connected to a Diamond General Chemical Microsensor II. Current output was recorded on a flat-bed chart recorder. Electrodes were calibrated before and after each experiment to 100 % air saturation and 100 % N₂. Any drift during the course of the experiment was assumed to be linear. Additional calibrations using a 5.16 % O₂ certified gas mixture were carried out periodically to test for linearity of response.

Larvae were anaesthetized using 50 mg l⁻¹ MS222 and placed in fresh water (+MS222) on a cork board mounted on the floor of a double-jacketed flow chamber. Previous studies found that this level of anaesthetic did not affect the rate of O₂ consumption by larvae of chinook salmon but was sufficient to halt opercular and pectoral fin movements (P. J. Rombough, unpublished results). The temperature of the water in the flow chamber was held at 10.0±0.2 °C by circulating chilled water between the walls of the jacket. Larvae were held in place by a thin rubber band (<1 mm wide) passing across the trunk approximately half way between the vent and the tail. The band was held in position with two fine insect pins 1–2 cm from the body surface. Another fine insect pin was pushed into the cork board between the open jaws of the larva (the pin did not pierce tissue) to prevent it from moving forward. Once the larva had been immobilized, the water in the flow chamber was aerated vigorously for at least 20 min to ensure that the ambient of the bulk water was close to 100 % air saturation. Aeration was halted approximately 5 min before the first measurements were taken.

Boundary layer values were measured at positions 2, 3, 5 and 6 (see Fig. 1) for eight individuals and at positions 1, 4 and 7 for seven individuals. The order of position measurements was assigned randomly for each individual. At each position, the electrode was advanced using the micromanipulator until a slight dimpling of the skin could be observed through a dissecting microscope (50× magnification). The electrode was then slowly pulled back until the dimpling just disappeared. This location was assumed to be representative of the at the skin surface. The electrode was kept at this location until the reading stabilized (readings were assumed to have stabilized if there was no change in the level of the trace on the chart recorder for at least 2 min; stabilization typically took 5–10 min). The electrode was pulled back by 250 μm using the micromanipulator and the reading again allowed to stabilize. This procedure was repeated at 500, 750 and 1000 μm from the skin surface. After all the positions had been measured, the larva was removed from the flow chamber and placed in anaesthetic-free water in a recovery tank. After it was verified that the larva was undamaged, the larva was killed with an overdose of MS222 and preserved in neutral buffered formalin.

The boundary layer data were analyzed using regression analysis and analysis of covariance (ANCOVA). The relationships between boundary layer and distance from the skin surface for each individual and body position were tested for linearity of response using the SigmaStat (SPSS Inc.) linear regression program. One-way ANCOVA (Systat; SPSS Inc.) was used to test for significant differences in slopes and elevations due to body position. In this test, was the dependent variable and body position the independent variable. Distance from the skin surface was treated as a covariate. Values for different individuals at the same body position were considered to be replicates. Mean regression lines were calculated using the Sigmastat linear regression program for each of the various body positions and compared pairwise for significant differences in slopes and elevations following the procedure of Snedecor and Cochran (1980). Two-way ANCOVA, in which was assumed to be dependent on the individual as well as on body position (distance remained the covariate), was used to test whether variation due to different individuals masked significant differences with respect to body position. The slope of the combined regression equation averaged over all body positions was used in the Fick equation (⁠⁠, equation 1) to estimate average area-specific rate of O₂ uptake. Values of 1.60×10⁻⁵ cm² s⁻¹ (Wickett, 1975) and 2.2399 μmol l⁻¹ mmHg⁻¹ (2.2399 μmol l⁻¹ kPa⁻¹) (Dejours, 1981), respectively, were used for D and β at 10 °C.

While the rubber dam presumably caused larvae some distress (approximately 10 % of the restrained larvae died before the end of the experiment), it did not appear to have had much of an impact on their overall rate of oxygen consumption. The mean of larvae restrained by the rubber dam during the first couple of hours was not significantly different from that of unrestrained larvae (Table 2). In the restrained larvae, 62.1 % of total O₂ uptake, on average, took place in the posterior compartment of the respirometer where the skin was the only site of gas exchange (Table 2). Assuming that the efficiency of transcutaneous uptake was approximately the same in the anterior compartment as it was in the posterior compartment, this means that the skin as a whole accounted for approximately 73 % of total O₂ uptake in 1-to 5-day-old larvae (Table 2). On the basis of individual rates of O₂ uptake and the corresponding cutaneous surface areas in the posterior compartment of the respirometer, a mean value (±S.D.) of 4.05±1.08 μg O₂ cm⁻² h⁻¹ was calculated for the area-specific rate of uptake of O₂ across the skin (Table 2). A mean mass-specific rate of O₂ uptake of 236 μg O₂ g⁻¹ h. −1 (Table 2) was estimated on the basis of mean total (restrained + unrestrained larvae; Table 2) and mean total tissue mass (Table 1).

In still water, levels at the skin surface (3.08 kPa, Table 3) were, on average, only approximately 14.5 % of the free-stream value. levels in the boundary layer increased fairly slowly with distance and typically did not approach free-stream values until 3–5 mm from the body surface. The decision to use a linear model (equation 1) to estimate O₂ flux was based on the fact that, for the first 1000 μm, the rate of increase was linear or close to linear (Fig. 1). Simple linear regressions based on measurements taken at distance of 1000 μm or less from the skin surface were all highly significant (P<0.001). In most cases, a simple linear regression provided as good a fit, or a better fit, to the data as a hyperbolic relationship (Fig. 1). Both one-way ANCOVA (Table 4) and the paired comparisons indicated that there were no significant differences (P>0.05) in terms of either slope or elevation of the mean regression equations for the various body positions. Two-way ANCOVA confirmed that body position had no significant effect on slope (F=1.11; P=0.358; d.f.=6, 237). There were significant differences in slope among individuals (F=3.94; P<0.001; d.f.=7, 237) but, given that individual variability at the same body position tended to be relatively small (Fig. 1; Table 3), they are probably not important in the context of this study. Because slope did not vary significantly with body position, mean area-specific O₂ uptake could be estimated relatively simply using the value for the combined slope averaged over all body positions. Calculating a mean value for would have been more difficult if the slopes had varied with body position since it would have required estimating the fraction of total surface area to which each of the various slope values applied. On the basis of a combined slope of 10.1 kPa mm⁻¹ (Table 3) and values of 1.60×10⁻⁵ cm² s⁻¹ and 2.2399 μmol l⁻¹ mmHg⁻¹, respectively, for D and β, the mean (±S.D.) area-specific rate of O₂ uptake across the skin at 10 °C was estimated to be 3.13±1.41 μg O₂ cm⁻² h⁻¹ (note that the actual S.D. is probably somewhat greater than ±1.41 μg O2 cm⁻² h⁻¹ since uncertainties associated with the values for D and β were not taken into account).

The two methods of estimating area-specific O₂ uptake gave essentially identical results when the difference in the size of the fish is taken into account. The raw value (mean ± 95 % CI) obtained using the boundary layer method (3.13±0.18 μg O₂ cm⁻² h⁻¹) was slightly less than that estimated using the rubber dam method (4.05±0.42 μg O₂ cm⁻² h⁻¹). The difference, although small, considering the potential sources for error (e.g. different flow regimes), was significant (P<0.05). This appears to be due, at least in part, to the fact that the larvae used in the boundary layer experiments were smaller than those used in the rubber dam tests. Mass-specific O₂ uptake in rainbow trout (Rombough, 1988b) and most other fish (reviewed in Giguere et al. 1988; Rombough, 1988a; Post and Lee, 1996) is independent of body mass during most of larval development. The relative surface area of the skin, in contrast, declines with increasing body mass (reviewed in Rombough and Moroz, 1997). The net result is that smaller larvae effectively have larger surface areas over which to take up the same amount of O₂. This is reflected in lower area-specific uptake rates in small larvae both within and among species (Rombough and Moroz, 1997). On the basis of geometric similarity (i.e. a scaling exponent for skin surface area of 0.67), one would expect for the 32 mg larvae used in the boundary layer experiments to be approximately 17 % less than in the 50 mg larvae used in the rubber dam experiments. If one makes this correction, the mean (±95 % CI) values estimated using the two methods are not significantly different (3.13±0.18 μg O₂ cm⁻² h⁻¹ and 3.36±0.35 μg O₂ cm⁻² h⁻¹ respectively, for the boundary layer and rubber dam methods at a common tissue mass of 32 mg).

The efficiency of cutaneous uptake and the overall importance of the skin in early larval O₂ exchange appear to be fairly similar in the three species of salmonid fishes examined to date. The efficiency of cutaneous gas exchange expressed in terms of area-specific uptake is approximately the same in rainbow trout (3.13 μg O₂ cm⁻² h⁻¹ at 32 mg; this study) as it is in Atlantic salmon Salmo salar (3.2 μg O₂ cm⁻² h⁻¹ at 33 mg tissue mass; Wells and Pinder, 1996). Area-specific uptake rates appear to be considerably higher in chinook Oncorhynchus tshawytscha larvae (7.5 μg O₂ cm⁻² h⁻¹; Rombough and Ure, 1991) but the chinook larvae were somewhat larger (≈100 mg tissue mass at hatching) than the larvae of the other two species so some of the difference is probably due to a smaller surface-to-mass ratio in the chinook. The major cause of the difference, however, seems to be a much higher mass-specific rate of O₂ consumption in chinook (354 μg O₂ cm⁻² h⁻¹; Rombough and Ure, 1991) than in rainbow trout (236 μg O₂ cm⁻² h⁻¹; this study) and Atlantic salmon (245 μg O₂ cm⁻² h⁻¹; Wells and Pinder, 1996). was measured at 10 °C using similar methods in all three species, so why chinook should have such a high metabolic rate is not clear. During the first few days posthatch, approximately the same fraction of total O₂ uptake takes place across the skin in all three species. In rainbow trout, approximately 73 % of total O₂ uptake occurs across the skin (this study). The corresponding mean values for chinook and Atlantic salmon are 74 % (Rombough and Ure, 1991) and 72 % (Wells and Pinder, 1996), respectively.

There do not appear to be any significant regional differences in the efficiency of cutaneous O₂ uptake in still water. The mean slopes of the lines relating and distance from the skin surface (Table 3) used to calculate area-specific uptake rates at the various positions on the skin surface were not significantly different (Table 4). If there were regional differences in efficiency, the slopes should have been significantly different. The extent to which different regions of the skin are vascularized varies considerably (Rombough, 1988a). This has led to speculation that the better-vascularized regions such as the yolk sac may play a particularly important role in larval gas exchange (e.g. Balon, 1975; Lanzing, 1976; Liem, 1981). This does not appear to be the case in rainbow trout. The rate of area-specific O₂ uptake across the yolk sac (positions 2 and 3, Fig. 1) of the trout was not significantly greater than elsewhere on the body surface (Table 3). The situation is similar in Atlantic salmon. Wells and Pinder (1996) were able to modify the rubber dam method so that they could measure O₂ uptake across just the yolk sac epithelium of newly hatched Atlantic salmon. The area-specific uptake rate for the yolk sac (2.7 μg O₂ cm⁻² h⁻¹) turned out to be approximately the same as that for the rest of the body (3.2 μg O₂ cm⁻² h⁻¹). The reason there is so little regional variation in uptake efficiency appears to lie with the diffusive boundary layer. In still water, the boundary layer is several millimetres thick and accounts for approximately 95 % of the diffusive resistance to O₂ flux (Rombough, 1992). According to the ‘pipes in a wall’ model of Malvin (1988), differences in capillary density should have little impact on uptake efficiency if, as is the case for trout larvae in still water, the total diffusion distance is long compared with the distance between capillaries.

The boundary layer method is potentially a very powerful technique for studying exchange processes during development. One of the major problems for those interested in developmental physiology has been the difficulty in measuring flux rates on a microscale. The boundary layer method is a flexible and relatively simple way to do this. Previous studies have shown that the technique works well with cells (e.g. Smith et al. 1994) and isolated tissues (e.g. Collier and O’Donnell, 1997). The present study shows that it also can be used with whole organisms. It is important to recognize that the boundary layer method need not be restricted to the study of O₂ uptake or trout larvae. With appropriate modifications, it can be used with many other organisms (not only fish) to estimate flux rates for virtually any diffusible substance consumed (e.g. Na⁺, Ca²⁺) or excreted (e.g. CO₂, NH₃) during development that can be measured with microelectrodes.

Thanks are extended to the staff of the Rockwood Hatchery for providing the eggs used in the experiments. I also would like to thank Roydon Strank and Vicki McLean for help in measuring boundary layer values. Funds for this project were provided by the Natural Sciences and Engineering Research Council of Canada.