The Transition to Air Breathing in Fishes: II. Effects of Hypoxia Acclimation on the Bimodal Gas Exchange of Ancistrus Chagresi (Loricariidae)

The first paper in this series (Graham & Baird, 1982) reported differences in the levels of aquatic O₂ that elicited facultative air-breathing in two Panamanian loricariid catfish Hypostomus and Ancistrus. It also showed that both these species had significantly reduced air-breathing rates following 14–21 days of air breathing and acclimation to hypoxic (aquatic O₂ partial pressure,⁠) water. The present study investigates the physiological and biochemical bases for hypoxia acclimation in one of these, Ancistrus chagresi Eigenmann and Eigenmann, a species that uses its stomach as an air-breathing organ (ABO) and commences facultative air breathing when drops to or below 33 mmHg (Gee, 1976 ; Graham & Baird, 1982).

Although periodic confinement in hypoxic water may be a regular occurrence in the life histories of most air-breathing fishes (Carter & Beadle, 1931; Kramer, Lindsey Moodie & Stevens, 1978), surprisingly few studies have considered the extent to which, once facultative air-breathing is initiated, a species may become better adapted for aquatic hypoxia, a more proficient air breather, or both of these. Non-air breathing fishes, for example, typically adapt to aquatic hypoxia by increasing either or both their total blood haemoglobin content and blood-O₂ affinity (Wood & Johansen, 1972; Powers, 1980). Would a facultative air breather therefore be expected to make similar adjustments in chronic hypoxia or would access to a rich supply of aerial O₂ obviate the necessity? Weber, Wood & Davis (1979) reported that 4–7 days of air breathing in hypoxia increased the Hb-O₂ affinity of two Brazilian loricariids (Hypostomus and Pterygoplichthys), but had a negative and no effect, respectively, on the total Hb of these species. Moreover, this same treatment did not affect either of these blood parameters in air-breathing Brazilian swamp eels (Synbranchus).

Hypoxia acclimation also appears to act differently on the physiological control of aerial respiration in some species. Neither hypoxia acclimation nor air-breathing history affected the air-breathing thresholds of Ancistrus or Hypostomus (Graham & Baird, 1982) or the mud minnow, Umbra (Gee, 1980). For Brazilian Synbranchus, however, Bicudo & Johansen (1979) reported that fish acclimated to hypoxia for 6 weeks switched to air breathing at a higher than did normoxic control fish.

This paper on A. chagresi examines physiological and biochemical adjustments that occur as a result of hypoxia acclimation and evaluates their combined effects on both the aerial and aquatic gas exchange capabilities and the control of respiration in this species.

Specimens of A. chagresi (2–200 g) were collected in the Rio Frijoles and Rio Frijolitos near Gamboa, Republic of Panama and returned to the Smithsonian Tropical Research Institute, Fort Amador, Panama where studies were carried out from 1974 to 1977. Subsequent work (1979–1982) was done with fish transported from Panama to La Jolla, California. Laboratory fish were maintained in low densities in darkened, aerated aquaria (25–27 °C) on natural photoperiods and were fed primarily on Tetramin flakes and green beans.

All experiments involved comparative testing of fish that had been acclimated to laboratory hypoxia and air breathing for from 14–21 days (25–27 °C), and control fishthat had been maintained in normoxic (130–150mmHg) laboratory aquaria (25–27 °C) and did not breathe air for the same time or longer. In air-breathing tests, comparisons were made between hypoxia-acclimated fish and control fish that had recently (1–36 h) been induced by hypoxia to breathe air. Procedures for hypoxia acclimation were described by Graham & Baird (1982). In the present experiments, tank surfaces were partially covered with plastic sheeting to keep and thus ensure regular air-breathing by all fish.

Control and hypoxia-acclimated fish were compared for the effect of on aquatic oxygen consumption rate without access to an air phase, using a closed respiratory system described by Graham (1973). Fish that had been starved for 24 h were placed in a darkened respirometer chamber and allowed 24 h for adjustment and recovery from handling prior to testing. Depending upon fish size, respirometer volumes ranged from 0·3–1·61. The respirometer was contained in a water bath (25 °C) and filtered, aerated water was continually pumped through it from a reservoir. During tests, the respirometer was closed, and because the same water was recirculated, ambient O2 level steadily declined due to fish respiration.

The rate of O₂ decline (slope), when corrected for respirometer volume, fish mass (volume), and background respiration, indicated fish ⁠. Penicillin was added to the reservoir to reduce background microbial respiration and blank respiration corrections were determined for each fish tested. An O₂ electrode (YSI Clark Type) connected to a millivolt recorder was used to monitor respirometer The electrode was initially calibrated in air and N₂ gas and standardized, before and after each run, against a Radiometer Blood Gas Analyser that was also calibrated at 25 °C.

The effect of on the of both experimental groups was evaluated by comparing relative (corrected slope) over 10 mmHg ranges of (i.e. ambient O₂ in the respirometer). Tests were started in normoxic water and continued until a fish had extracted nearly all O₂ from the system At least two runs, each on successive days, were made on all fish. In the later stages of each run, when was very low, the fish was observed at regular intervals to be sure it did not succumb to hypoxia. Neither nor pH were regulated in the respirometer and during the tests increased (from 1–6 mmHg) while pH usually dropped from 7·8 to 6·5.

Blood haemoglobin (Hb) content, O₂ affinity, and total red cell triphosphate-Hb ratios (P/Hb) were compared in control and hypoxia-acclimated Ancistrus. Blood samples, taken by cardiac puncture, were withdrawn into tuberculin syringes that were flushed with heparin and dried prior to use. Total Hb was estimated for each sample using the cyanmethaemoglobin conversion procedure (Sigma Chemical Co.).

The blood mixing method (Edwards & Martin, 1966) was used to determine O₂ dissociation curves. Pooled blood samples (6–10 ml, 3–5 fish) were equilibrated for 1 h with humidified air or N₂ at 25 °C in 50 ml stoppered flask tonometers. Using a 1 ml glass syringe, volumes of blood were mixed without bubble contamination in proportions that resulted in 95, 75, 40, and 10 % O₂ saturation levels. After the blood had been mixed in the syringe, PO₂ and pH at 25 °C was determined using the Radiometer Analyser. Tonometered blood was used within 2h and replicate determinations at P₄₀ were carried out at the end of all tests to verify that large changes in blood affinity had not taken place in the time course of equilibration.

Previous investigations (Bartlett, 1978, 1980; Johansen, Mangum & Lykkeboe, 1978; Weber, Wood & Davis, 1979) established that loricariid red cells contain both adenosine (ATP) and guanosine (GTP) triphosphate. In the present work, only total red cell triphosphate levels were estimated for control and hypoxia-acclimated fish using the Sigma ATP test kit which is a non-specific indicator for all red cell triphosphates (Bartlett, 1978). Pooled blood samples (4–6 fish) were also used in these tests.

The volume and the O₂ and CO₂ contents of gas expelled by air-breathing Ancistrus were measured in order to determine if fish that had been air breathing for 14–21 days could hold more air, utilize more of the O₂ in a breath, and perhaps release more CO₂ aerially than could control fish that were recently (1–36 h) induced to breathe air. The average duration of air breaths was also compared for the two groups.

Gee (1976) determined the relationship between body length and in vivo stomach volume for Ancistrus. In the present tests, a gas collector similar to that used by Gee (1976) was employed. Individual fish were placed in a hypoxic, partially covered, 101 aquarium (25 °C) that contained an inverted, water-filled funnel attached to a graduated pipette (Fig. 1). By positioning the funnel over the fish, expired air bubbles could be trapped. The water-filled syringe at the top of the pipette was used to withdraw the gas quickly into the pipette and to determine volume. The sample was then pulled up and into the side syringe for analyses of O₂ and CO₂ content. Gases were originally analysed with a Scholander gas analyser, although it was later determined that the Radiometer O₂ and CO₂ electrodes could be used if carefully and regularly calibrated. Previous work (Graham & Baird, 1982) had established that air breathing rate is affected by which was therefore monitored using the YSI O₂ electrode during all gas-release studies. By bubbling small amounts of CO₂ and N₂ into the aquarium it was possible to alter the respiratory gas content of the water and thus determine the relationship between and and expired gas content.

The aquarium and gas collector were placed behind a visual blind and dimly backlighted so that the fish and the sampler could be continually observed. This was necessary to verify that all gas expelled was in fact collected by the funnel. Also, for accuracy, the sample had to be processed immediately upon capture. Calibration experiments were done using small bubbles of air and gas mixtures of known O₂ and N₂ content which were released at the bottom of the aquarium and collected in the funnel. These tests showed that significant contamination of the sample could be avoided if it was processed quickly (15 s). The potential for the diffusion of O₂ in or out of the expelled gas was low because of the usually low diffusion gradient between expired gas and (see Results). CO₂ however, was more difficult to work with because of its higher solubility in water and the larger diffusion gradients that existed in some tests (see Results). The dead space of the side-arm syringe contained hypoxic water and this was expelled through the stopcock as soon as the sample was pulled in and the syringe disconnected. A dark plastic apron attached to the ventral side of the funnel (Fig. 1) increased the effective area of the sampler and made it possible to collect bubbles released near comers of the tank. The apron also darkened the area below the collector and aided in keeping the fish in a fixed position between air breaths. Suspending the sampler by its top syringe made it possible to swing the apparatus in the event the fish changed position slightly (Fig. 1).

Hypoxia-acclimated fish and normoxic fish that had been recently induced to breathe air were compared for the relationship between heart-beat and gill ventilation rates during regular cycles of air-breathing activity. Electrode pairs (hooked, 0 · mm diameter, stainless steel) were implanted around the pericardium and adjacent to the hypobranchial muscle of lightly anaesthetized (MS 222) fish using procedures described by Roberts & Graham (1979). Following recovery from anaesthesia, the fish was placed in a small (41) darkened aquarium (24–26 °C). A Faraday cage covered the tank and the electrode leads were connected to a Gilson recorder. No signal preamplification was made but in some cases an impedance converter was used to obtain clearer signals. Records were made over 12–72 h following electrode implantation. N₂ was gently bubbled into the aquarium to achieve and maintain desired levels of hypoxia and were measured regularly.

Oxygen consumption dependence curves (Fig. 2) compare the effect of on the aquatic of control (n = 7) and hypoxia-acclimated (n = 9) Ancistrus of equal body size. From 140 down to 100 mmHg, the of both groups is similar with a mean of 84·0 ± 4 ml kgh⁻¹ (⁠ ± 95 % conf, int., n = 35, mean wt 26 g, range 3–99 g). Variable respiration rates occurred between 100 and 50 mmHg and, below 40 mmHg the of both groups decreased rapidly with that of the acclimated fish remaining higher relative to the control. Acclimated fish had a low level of respiration at 5 mmHg but the of the control group became nearly zero at 10 mmHg ⁠. In addition, mean rates of the two groups are significantly different (P<0·05, t test) at 10 and 20mmHg. Since neither group of fish had access to air inside the respirometer the measured is assumed to be the maximum attainable by aquatic uptake at each level of hypoxia.

Hypoxia-acclimated Ancistrus have a significantly higher blood Hb content, a significantly lower amount of total red cell phosphate (P < 0·05, t test, both comparisons), and a higher Hb-O₂ affinity (P₅₀, estimated from dissociation curves, Fig. 3) than do control fish (Table 1). Fig. 3 compares O₂ dissociation curves and molar phosphate-Hb ratios of the two groups. To enable visualization of the impact of the Hb increase on the O₂-carrying capacity of hypoxia-acclimated fish blood, the Y axis of Fig. 3 is expressed as total O₂ content rather than percent saturation (Wood & Johansen, 1972). Oxygen content was calculated assuming an O₂-combining power of 1·39 mlO₂ gHb as determined for loricariids and other air-breathing fishes (Johansen et al. 1978). The pH of tonometered blood is similar to other values reported for loricariids (Weber et al. 1979; Wood, Weber & Davis, 1979).

Fig. 4. illustrates the relationship between the volume of gas expelled from the ABO and body weight in Ancistrus and shows that the volume released by hypoxia- acclimated fish is about 25 % greater than that of control fish recently induced to breathe air. The slopes of the two lines in Fig. 4 are not different but covariance analysis revealed that the mean gas release volumes (i.e. line intercepts) of the two groups differ significantly F₂,₃₁ = 12·3, P < 0 ·05). Expired gas volume was nearly constant in all fish tested. It was not affected by and, in most series of repeated determinations, the 95 % confidence intervals of the mean volume estimate were usually within 10 % of the mean. Also, volumes collected from control fish fit perfectly on the fish length—in vivo ABO volume curve of Gee (1976), confirming this organ to be completely emptied upon each expiration.

The O2 partial pressure of an expired breath is inversely proportional to the time it is held (Fig. 5). Both groups of fish usually held an air breath for longer than 2 min during which time, assuming complete expiration of the previous breath (see above), PEO₂ dropped from 155 (air value) to below 75 mmHg. Compared to control fish (Table 2A), hypoxia-acclimated fish had a significantly (P<0·05, t test) lower average (and therefore a larger O2 extraction 80 vs 71 %) and a longer average air-breath duration. Only 27 % of the breaths discharged by control fish had a of or less than 35 mmHg compared to 54 % of the breaths released by acclimated fish (Table 2B). Fig. 5 shows that control fish seldom had a ⁠, as low as 15 mmHg but, in some acclimated fish, was below 10 mmHg. Also, a greater percentage of the air breaths of hypoxia-acclimated fish were held for 10min or longer (32 vs 13%, Table 2B). In neither group was a relationship found between and nor did they differ in regard to mean (Table 2A). The amount of expired CO₂ is very Tow (respiratory exchange ratios <0·1, Table 2A), indicating that most CO₂ is released aquatically. directly affected Blood probably reached an equilibrium state with and the fish used HCO₃⁻ buffering to conserve blood pH (Wood et al. 1979). Tests with mixed gas bubbles released under the funnel showed that CO₂ was also acquired from water when the diffusion gradient was high.

To calculate aerial inspired air volume must be known but, because O₂ removed by the air-breathing organ is not replaced by an equivalent volume of CO₂ (Table 2A), expired gas volume is always less than inspired. If the proportions of the O₂and CO₂ in expired gas are measured and if it can be assumed that the amounts of N₂ and inert gases (79 % of air) in the breath did not change, then inspired volume could be calculated by algebraic substitution. For example, a 25 g hypoxia-acclimated fish would expel about 1·02 ml of gas (Fig. 4) that would on the average (Table 2A) contain about 4·2% O₂ (21 % × 31/155 mmHg) and 0·2% CO₂ (0·03 × 1·49/0·22). Assuming that N₂ and inert gases constitute the balance of the expelled sample, then their volume would be 0·975 ml (1 ·020 − [1·020 X 0·044]). If this volume originally constituted 79% of the inspired volume, then inspired volume was 1-23 ml, 20% greater than the expired volume. For a 25 g control fish with an expired volume of 0·82ml (Fig. 4) and average gas contents of 6·2% O₂ and 0·2% CO₂ (Table 2A), inspired volume is 0·97ml (18% larger). The difference between expired and inspired volume would depend on expired gas O₂ and CO₂ levels. In theory it could not exceed 21 % and would be more difficult to calculate for fish in high owing to the effect of this variable on ⁠.

The impact of hypoxia acclimation on total respiration can be seen in Table 3, which compares the effect of on the bimodal and total respiration rates of control and hypoxia-acclimated Ancistrus. For both groups the combined influences of different air-breathing frequencies (from Graham & Baird, 1982), ABO volumes (Fig. 4) and O₂ extraction percentages (Table 2A) on aerial respiration rate have been computed for a 25 g fish. (Inspired volumes were calculated above. Note that for both groups inspired volume and O₂ extraction do not change with [see above].) Also included in Table 3 are the maximum aquatic respiration rates for a 25 g fish at each (Fig. 2). In both control and acclimated fish, aerial O₂ uptake increases and aquatic respiration decreases as is lowered. At 30 mmHg a control fish has a greater aerial respiration rate than an acclimated fish because of a higher air-breathing frequency (breaths/h × ml/breath = ml/h). But, at 10mmHg its aerial O₂ uptake is below that of an acclimated fish ; it also has about zero aquatic consumption capacity at 10 mmHg (Fig. 2), thus its total potential respiration would fall below routine. From 30–10 mmHg the aerial uptake calculated for a hypoxia-acclimated fish increases more than two-fold and the level of supplementary aquatic respiration needed to sustain routine consumption is very low. This shows that a control Ancistrus may be initially incapable of surviving severe hypoxia but, through gradual hypoxia-acclimation, this fish can utilize aerial respiration more effectively and minimise its use of aquatic respiration, even though this respiratory mode has also become more proficient.

When Ancistrus takes an air breath its heart rate immediately rises (tachycardia) and the frequency and amplitude of its gill ventilation abruptly drops. During the time air is held, heart rate gradually declines and aquatic ventilation slowly climbs to ‘pre-air breath’ levels (Fig. 6). Table 4A compares hypoxia-acclimated and control fish for their mean heart and gill ventilation frequencies and the magnitude of the shifts occurring just after an air breath is taken and immediately prior to the next one. Hypoxia-acclimated Ancistrus have significantly lower (P<0·05, t test) mean ‘pre breath’ heart and opercular-beat frequencies than control fish. Just after an air breath, hypoxia-acclimated and control fish show similar relative drops in opercular frequency (−21 %), but the hypoxia group has a much larger tachycardia (32%).

Fig. 6 compares the pattern of mean heart and gill ventilation-frequency shifts observed for the two groups in the course of one air-breath cycle. An air breath by a hypoxia-acclimated fish results in nearly equivalent shifts in the ratios of heart and gill ventilation frequencies before and after the breath is taken (Table 4B). By contrast, the pre-breath ratio of gill ventilation to heart rate in control fish is 1·32 (Table 4B) but after the air breath is taken, heart rate exceeds gill ventilation by only 1–13 times. These comparisons indicate that hypoxia acclimation enables Ancistrus to reduce both cardiac and gill-ventilatory activity in hypoxic water. Also, complimentary shifts in heart and gill ventilation frequency ratios with an air breath suggest that hypoxia-acclimated fish may be able to integrate the aerial and aquatic phases of bimodal respiration more smoothly.

Hypoxia-acclimation results in a suite of adaptive changes that enhance both the aerial and aquatic respiration capacities of Ancistrus in hypoxic water. An acclimated fish gulps a greater volume of air and, on average, holds it longer than control fish. It also seems able to reduce to a lower level, although this is not regularly done. Moreover, hypoxia-acclimated fish have about a 25 % greater maximum capacity for aquatic respiration over the range (⩽33 mmHg) where facultative air-breathing is required. This discussion examines the way these adaptations may contribute to the separate aquatic and aerial modes of respiration as well as to the total gas exchange of Ancistrus.

In their landmark study, Carter & Beadle (1931) determined that an air-breath expired by Ancistrus anisitsi contained 13·8 % O₂ and no CO₂ (n = 2). The present study with A. chagresi found only slight amounts of CO₂ and discharged-breath O₂ content averaged between 5 and 6% O₂. Carter & Beadle (1931) also described respiratory adaptations in the stomach of Ancistrus. This organ has thin compliant walls that contain layers of smooth muscle which, through contraction, serve to expel its contents. The inner surface of the stomach consists of a single layer of cuboidal epithelial cells which is invested by a uniformly-distributed and dense capillary bed. Blood supply is by the coeliac artery and drainage is through the interrenal vein. The ABO of A. chagresi is morphologically similar to that of A. anisitsi. In vivo inspection revealed no differences in the tributary vessel sizes or in the capillary densities of the stomachs of hypoxia-acclimated and control fish. However, histological comparisons were not made. The increase in ABO volume following hypoxia-acclimation may result from stretching caused by repeated air breathing or growth in the organ.

InAncistrus, as in most other teleostean air breathers, O₂ rich blood that drains the ABO is mixed with other venous blood and must circulate through the gills before reaching systemic circulation (Carter & Beadle, 1931; Johansen, 1970). The increased Hb content and Hb—O₂ affinity of hypoxia-acclimated fish minimize mixing and gill exposure effects. An increased O₂ capacity will offset, to some extent, the mixing of O₂-rich ABO blood with 0₂-poor venous blood and the increased affinity may ensure complete O₂ saturation (Johanson et al. 1978). A higher O₂ affinity would also favour O₂ retention and lessen transbranchial O₂ loss (see below). The lower in some air breaths from hypoxia-acclimated fish (10 mmHg vs 21 mmHg, mean of six lowest values for each group, Fig-5) correlates with their higher Hb—O₂ affinity (Table 1). However, since O₂ depletion from the ABO only rarely proceeded to this point (Fig. 5, Table 2) the primary function of an increased Hb-O₂ affinity is probably not for the capture of aerial O₂.

The significance to aerial O₂ transport of the cyclic shifts in heart beat and gill ventilation during each air breath cannot be quantitatively assessed without information on cardiac output and relative blood flow through gills and the ABO, together with ventilation volume at different times in the air-breath interval. It would also be desirable to know the extent to which branchial blood flow can be controlled autonomically. Measurements of this type would be difficult on fish of the size used in this study but could probably be done on slightly larger specimens (Farrell, 1978). Other work has shown that air-breathing tachycardia occurs in most air-breathing fish as well as other air-breathing vertebrates (Heatwole & Seymour, 1976; Singh, 1976; Randall, Burg-gren, Farrell & Haswell, 1981). It has also been shown that increased cardiac output to the ABO occurs just after inspiration (Singh, 1976; Farrell, 1978). Surprisingly few.studies have examined the coordination of cardiac and opercular activity during air breathing in fishes. Singh & Hughes (1973) found that Anabas completely stopped gill ventilation during periods when it exclusively utilized O₂ in its ABO and when it was exposed in air or to hypercapnic water. During bimodal respiration, gill ventilation was constant and unaffected by air-breathing events even though heart rate cycled. This is unlike the phase shifts in cardiac and gill ventilatory activity that occur with air breathing in Ancistrus.

This study provides insight into the factors controlling air breathing in Ancistrus. The mean air-breathing threshold of this species (33 mmHg) is unaffected by hypoxia-acclimation (Graham & Baird, 1982). The aquatic oxygen consumption dependence curve (Fig. 2) of hypoxia-acclimated fish shows that they can, through increased gill ventilation, maintain routine down to nearly 40 mmHg Below 40 mmHg, however, the of both experimental groups drops precipitously. Thus air breathing is initiated soon after maximum aquatic drops below routine. Even though hypoxia-acclimated fish are more effective air breathers (Table 3) their air-breathing frequency remains inversely related to (Graham & Baird, 1982). This, together with the longer average time an air breath was held by acclimated fish (Fig. 5), suggests that exerts continued indirect control over air breathing through its effect on aquatic and the potential for aerial O₂ loss through the gills.

Even though it can use the O₂-rich aerial medium, Ancistrus makes respiratory adjustments for aquatic hypoxia like those seen in non air-breathing fish (Wood & Johansen, 1972; Powers, 1980). An increased Hb-O₂ affinity improves the effectiveness of aquatic O₂ uptake by acclimated fish and is consistent with the demonstrated left-shift of their oxygen consumption-dependence curves (Fig. 2). In addition to extending the lower limits for aquatic respiration, an increased Hb-O₂ affinity also lowers the effective limit for the transbranchial loss of O₂ ;in control fish, Co saturated blood circulating through the gills would unload O₂ to water at In hypoxia-acclimated fish this would begin at (Fig. 3).

Erythrocytic triphosphates are allosteric modulators of Hb-O₂ affinity and usually decrease in concentration and thus raise O₂ affinity in fish exposed to hypoxic water (Wood & Johansen, 1972). Affinity increases are often accompanied by a rise in total Hb (Powers, 1980) as shown here for Ancistrus. The combined effect of a rise in Hb and a drop in phosphate is to diminish further the molar P-Hb ratio, an index of relative O₂ affinity.

In this work with Ancistrus, the nearly 10 mmHg left-shift of P₅₀ was accompanied by a drop in intracellular P/Hb from 1·32 to 0·65 (Table 1). These ratios and the magnitude of their change are larger than reported for a South American air-breathing honoured catfish (Hypostomus sp.) by Weber et al. (1979). These workers tested fish that had been acclimated to hypoxia for only 4–7 days and correlated a drop in P₅₀ from 17-9 to 13·3 mmHg (pH = 7· 65) with a 50% reduction in red cell phosphate. However, they also found a significant drop in Hb; thus P/Hb was reduced only from 0·47 to 0·28. With another hypoxia-acclimated air-breathing loricariid (Pterygoplichthys), Weber et al. (1979) found a significant drop in total phosphate (34%) but no change in Hb as P/Hb shifted from 0·57 to 0·40. Red cell triphosphate levels can be changed quickly, whereas more time is needed to produce more erythrocytes which raise haematocrit and Hb (Powers, 1980; J. B. Graham, in preparation). With only 4–7 days of hypoxia acclimation these two South American loricariids were able to reduce triphosphate levels but probably did not complete an erythropoietic response like that of Ancistrus.

Only small amounts of respiratory CO₂ are discharged aerially by Ancistrus (Table 2) and because thick scales cover most of the body of this fish it is concluded that CO₂ release must occur aquatically through the gills. Survival in chronic hypoxia would therefore depend upon the ability of Ancistrus to balance the use of its gills for aquatic CO₂ release with the need to prevent the transbranchial loss of O₂ gained from the ABO. Cyclic changes in gill ventilation rate that occur during each air breath period suggest the manner in which this may be done. Some branchial CO₂ exchange probably occurs at all times, but the rate of this process seems likely to be highest during the latter phase of each air breath cycle when, together with the progressive depletion of O₂ from blood and the ABO, gill ventilation increases.

Early stages of this work were supported by the Smithsonian Institution’s Research Foundation and the Smithsonian Tropical Research Institute. Research from 1979 to 1982 was supported partly by NSF Grant DEB 79-12235 and by PHSPRO7011. Appreciation is expressed to D. Abel, G. de Alba, T. Baird, K. Dickson, R. Rubinoff, P. Schumann and N. Smith for advice, comments and assistance. The generous cooperation of the Department of Natural and Renewable Resources (RENARE) of the Republic of Panama is also acknowledged.