Acid-Base and ion Balance, Metabolism, and Their Interactions, After Exhaustive Exercise in Fish

An experimental exercise protocol variously described as ‘intensive’, ‘strenuous’ or ‘exhaustive activity’, or more accurately as ‘exhaustive exercise stress’, has been intensively exploited over the past 15 years. In its least invasive form, the procedure involves swimming the fish at progressively higher speeds in a swim tunnel until the maximum sustainable velocity (U_crit) is reached, followed by burst swimming at higher velocity to ensure exhaustion (Butler et al. 1986; Primmet et al. 1986; Mommsen and Hochachka, 1988). More commonly, the fish is manually chased to exhaustion in a small tank or trough (Turner et al. 1983a; Milligan and Wood, 1986a; McDonald et al. 1989a), a technique first used extensively 30–40 years ago in the pioneering work of E. C. Black (Black, 1958; Black et al. 1962). In another variation, electric shocks, either alone or in combination with chasing, are employed to ensure complete exhaustion (Holeton et al. 1983; Wieser et al. 1985; Pearson et al. 1990). The endpoint is clearly manifested as an inability to perform further burst exercise (slow swimming via aerobic red muscle generally persists), and accompanying physiological changes are remarkably uniform within a particular batch of fish.

The validity of this approach in duplicating ‘natural’ exhaustive exercise has been much debated, especially by referees, but it is equally debatable whether truly natural exercise is ever exhaustive in fish. However, through man’s intervention in the natural environment, the phenomenon is now a common occurrence. Examples include angling (especially catch and release sportfishing; Bouck and Ball, 1966; Wydoski et al. 1976; Beggs et al. 1980; Schwalme and Mackay, 1985b; Wells et al. 1986), commercial fishing ‘throwback’ (Parker and Black, 1959; Neilsen et al. 1989), netting and electrofishing for biological surveys (Schreck et al. 1976; Mesa and Schreck, 1989), and transportation, confinement and handling stresses resulting from aquacultural and fish-stocking practices (Fraser and Beamish, 1969; Barton et al. 1986; Woodward and Strange, 1987). The physiological disturbances resulting from exhaustive exercise stress may be so severe as to cause the death of the animal several hours later (Black, 1958; Wood et al. 1983).

The applied relevance notwithstanding, studies on exhaustive exercise stress in fish have yielded a wealth of basic physiological knowledge. The process of exhaustion itself is rapid, the physiological condition changing from rest or steadystate aerobic exercise to glycolytic exhaustion within a few minutes. As yet, there is little information on the sequence of events occurring during this brief period of dynamic change. The original study of Black et al. (1962) is still the most detailed on time course, while Parkhouse et al. (1987, 1988) have provided some modern biochemical information. However, most of our new knowledge is of the consequences prevailing once exhaustion has occurred, and the changes that take place during post-exercise recovery. The slow time course (4–24 h) of this recovery has greatly facilitated dissection of the mechanisms involved. The present review presents an overview of these responses following exhaustive exercise stress, and focuses on some areas of particular interest or controversy. Randall and Brauner (1991, this symposium) review aerobic exercise in fish.

Several physiological responses distinguish exhaustive exercise from steady rate aerobic exercise (Wood and Perry, 1985). Selected aspects are illustrated in Fig. 1.

There is a significant depletion of glycogen reserves and an associated large build-up of lactate in white muscle caused by the anaerobic glycolysis used for burst performance (e.g. Black et al. 1962; Milligan and Wood, 1986b; Dobson and Hochachka, 1987). Blood lactate levels may rise substantially (Fig. 1B), though never to the levels seen in white muscle (see Fig. 4). There is also a substantial degradation of muscle ATP and creatine phosphate stores, to levels far lower than those seen at exhaustion in many higher vertebrates (Driedzic and Hochachka, 1976; Dobson et al. 1987; Mommsen and Hochachka, 1988). While some glycolytic generation of lactate occurs during steady-state swimming, production and oxidation rates appear to reach equilibrium so that elevations of blood and muscle lactate levels are small (Johnston and Goldspink, 1973; Driedzic and Kiceniuk, 1976; Wokoma and Johnston, 1981; Duthie, 1982; Boutilier et al. 1984; Butler et al. 1986). Muscle ATP and creatine phosphate levels are not depressed in aerobically swum fish (Parkhouse et al. 1987; Dobson and Hochachka, 1987).

A marked depression of blood pH (Fig. 1A) occurs in all fish after exhaustive exercise stress (Wood et al. 1977; Jensen et al. 1983; Turner et al. 1983a,b; Holeton and Heisler, 1983; Schwalme and Mackay, 1985a; Walsh, 1989). This results from a combination of metabolic (⁠ elevation; Fig. 1B) and respiratory acidoses (⁠ elevation; Fig. 1C). While the cause of elevation has been controversial (see below), metabolic acidosis has generally been attributed to protons generated in association with lactate production and ATP breakdown in white muscle (Hochachka and Mommsen, 1983; Wood and Perry, 1985). It is now clear that proton extrusion from red blood cells may also play an important role in some species (see below). In contrast, during sustained aerobic exercise, blood acid-base status exhibits only minimal changes from control levels (Kiceniuk and Jones, 1977; Boutilier et al. 1984; Butler et al. 1986; Thomas et al. 1987).

Exhaustive exercise stress causes a profound disturbance of ionic, osmotic and fluid volume homeostasis. A large net fluid shift from extracellular to intracellular compartments results from the osmotic attraction offered by greatly increased intracellular lactate levels (Milligan and Wood, 1986a,b, 1987b,c). Haemoconcentration occurs, and concentrations of all major plasma electrolytes generally increase (Fig. 1D) because of this decrease in plasma water (Turner et al. 1983a,b; Holeton et al. 1983; Thomas et al. 1987; van Dijk and Wood, 1988). Plasma electrolyte levels rise, despite net losses through the gills and kidney in freshwater fish (Wood and Randall 1973a; Wood, 1988). In contrast, the limited observations available on fish undergoing steady-state exercise indicate minimal changes in levels of plasma electrolytes (Rao, 1969; Farmer and Beamish, 1969; Parkhouse et al. 1987; Thomas et al. 1987) and branchial ion flux rates (Wood and Randall, 1973b). However, it is interesting that both exercise regimes cause increased branchial water entry and compensating renal excretion (Wood and Randall, 1973c; Hofmann and Butler, 1979; Wood, 1988). Increased water flux is presumably an unavoidable correlate of increased O₂ diffusing capacity at the gills.

An important difference between exhaustive and sustained exercise, and one that may play a key role in explaining other differences, is the extent of catecholamine mobilization. It is now clear that fish performing sustained aerobic exercise exhibit little or no increase in circulating catecholamine levels (Ristori and Laurent, 1985; Butler et al. 1986; Primmet et al. 1986; Axelsson and Nilsson, 1986; Hughes et al. 1988). However, once U_crit is surpassed and burst performance starts, there is a rapid release of adrenaline and noradrenaline into the bloodstream (Primmett et al. 1986; Butler et al. 1986, 1989). This mobilization seems to be particularly pronounced when the stress component is high, as in chasing to exhaustion (Fig. 1E; Milligan and Wood, 1987a; Tang and Boutilier, 1988; van Dijk and Wood, 1988; Milligan et al. 1989; Walsh, 1989; Wood et al. 1990b). Levels of both catecholamines (around 10⁻⁹ mol l⁻¹ at rest) may increase by 1–2 orders of magnitude; Milligan and Wood (1987a), Tang and Boutilier (1988) and Perry et al. (1989) provide useful tabulations of the recent literature.

The proximate stimulus for catecholamine release remains unclear. Some workers have related the phenomenon to acidosis (Boutilier et al. 1986; Tang and Boutilier, 1988), others to a decfine in blood O₂ content (Perry et al. 1989; Fievet et al. 1990). It is quite possible that both stimuli are involved. The discharge of catecholamines from chromaffin tissue appears to be directly responsive to the chemical characteristics of the perfusing venous blood (Perry et al. 1991a). Venous blood becomes both acidic and severely hypoxaemic once U_crit is exceeded (Kiceniuk and Jones, 1977; Butler et al. 1989). A psychological component (‘panic’), acting through the sympathetic innervation of the chromaffin cells, is also quite possibly involved. Section of the sympathetic spinal nerves to the head kidney reduces catecholamine mobilization during stress (Nilsson et al. 1976; Butler et al. 1989; Perry et al. 1991a).

From the above discussion, it is clear that the physiological state of the fish after exhaustion is very different from that prior to exercise. The task of recovery is to restore the original condition with as little additional metabolic cost as possible. This cost can be quantified as the ‘excess post-exercise oxygen consumption’ or EPOC. The measurements of Brett (1964) on salmon are still the most detailed; these indicate that EPOC lasts 4–6 h after exhaustion and would be large enough to support about Ih of aerobic swimming at U_crit. The traditional view of mammalian physiology, formulated by A. V. Hill, R. Margaria and their associates in the 1920s (Hill and Lupton, 1923; Margaria et al. 1933), and still found in many textbooks today, has been that most of the EPOC is accounted for by the metabolism of lactate back to glycogen. However, modern re-assessments have concluded that this ‘oxygen debt hypothesis’ is overly simplistic. A host of other metabolic disturbances besides lactate clearance are now thought to contribute to EPOC in mammals; the most important of these is the post-exercise elevation in body temperature (see Gaesser and Brooks, 1984, for a review). In poikilothermic fish, temperature elevation is clearly not a factor; it is surprising, therefore, that no detailed assessments of the oxygen debt hypothesis in fish have been published.

Recently, Scarabello et al. (1991a) have performed such an assessment by making frequent measurements of the rate of oxygen consumption and metabolite levels in the whole body of juvenile trout during recovery from exhaustive exercise stress. As in mammals (Margaria et al. 1933), it was possible to resolve total EPOC into two components by exponential analysis. A ‘fast’ component (half-time=0.23 h) constituted about one-fifth of the total, and a ‘slow’ component (half-time=2.1 h) accounted for the other four-fifths (Fig. 2). Most (83%) of the fast component could be accounted for by the same factors as in mammals - measured restoration of creatine phosphate and ATP stores and reasonable estimates of the costs of increased cardioventilatory work and the recharging of body O₂ stores. However, only about 25% of the larger SLOW component could be explained by measured rates of lactate clearance an glycogen resynthesis and a metabolic scenario based on the classical ‘oxygen debt hypothesis’, i.e. glycogen resynthesis as the primary fate of lactate. A scenario assuming oxidation as the primary fate of lactate produced a similarly unreasonable result. EPOC was far larger than could be explained by lactate metabolism and glycogen resynthesis.

To confirm this conclusion, tests were performed to see whether EPOC could be experimentally dissociated from lactate clearance. In fish depleted of glycogen by starvation prior to exercise, both lactate clearance and glycogen resynthesis were reduced by 30%, but EPOC was unaltered (Scarabello et al. 1991b). Conversely, in fish subjected to a second exercise bout 6 h after the first, EPOC was reduced by 40%, but lactate clearance and glycogen resynthesis were unchanged (M. Scarabello, G. F. Heigenhauser, and C. M. Wood, unpublished results). Clearly, EPOC and post-exercise lactate metabolism are not quantitatively related. Several other studies have also discounted a tight relationship between EPOC and lactate metabolism in fish, for different reasons (Wieser et al. 1985; Milligan and McDonald, 1988). Even though temperature elevation cannot be involved, other factors must be sought to explain the major portion of EPOC in fish.

Important contributors to EPOC may be the costs of altered membrane transport processes associated with re-establishing ion, acid–base and fluid volume homeostasis. The major focus of research has been on the gills, the closest point of contact of the extracellular fluid with an osmotically hostile evironment. However, as illustrated by Fig. 3, the kidney and the ECF/ICF interface actually exhibit much larger net fluxes, a point that has been generally overlooked.

At the gills, net losses of major electrolytes and gains of water in freshwater fish (and the opposite in seawater fish) may continue for some time during recovery from exhaustive exercise (Fig. 3); these deficits must eventually be corrected by the increased expenditure of metabolic energy. Elevated water and O₂ fluxes are probably linked by simple permeability effects associated with increased branchial surface area, reduced diffusion distance and catecholamine mobilization (see Wood and Perry, 1985). The same factors will influence ion fluxes, but the situation is complicated by the dynamic modulation of branchial ion exchanges for the purpose of acid-base correction. Metabolic acid released from white muscle is transiently shuttled to the external environment across the gills (Fig. 3), and then later retrieved for final metabolic correction (Holeton and Heisler, 1983; Milligan and Wood, 1986a, 1987b; Milligan et al. 1991). At least in freshwater fish, the branchial movements of are linked to movements of strong ions (Holeton et al. 1983; Wood, 1988; McDonald et al. 1989a,b). During early recovery, active Na⁺ influx (Na⁺/acid exchange) is stimulated while Cl⁻ influx (Cl⁻/base exchange) is inhibited, thereby achieving net excretion. During later recovery, these trends are reversed, thereby achieving net uptake. Modulation of branchial ion exchange may be a direct response to internal acid-base status (Milligan et al. 1991; Goss and Wood, 1991) and/or catecholamines may play a controlling role (Wood and Perry, 1985; Vermette and Perry, 1987; McDonald et al. 1989a,b). This temporary ‘storage’ of in the water represents only a small fraction of the whole-body load, but is important in preventing intolerable acidosis in the poorly buffered extracellular compartment. Almost no lactate anion is lost across the gills (Holeton and Heisler, 1983; Milligan and Wood, 1986a).

The kidney makes only a very small contribution to acid-base regulation after exercise, and urinary lactate losses are also minimal (Fig. 3; Holeton and Heisler, 1983; Holeton et al. 1983; Wood, 1988). In freshwater fish, urine flow rate increases substantially in response to elevated glomerular filtration rate, thereby compensating for increased water uptake at the gills (Wood and Randall, 1973c; Hofmann and Butler, 1979; Wood, 1988). However, renal ion losses are only slightly elevated (Fig. 3). This conservation of ions in the face of increased filtered load is achieved by large elevations in the tubular reabsorption of Na⁺, Cl⁻, lactate and other electrolytes (Fig. 3). The changes in net tubular fluxes are actually larger than those at the gills during the first 6h after exercise in trout. Most of the metabolic expenditure of the kidney is probably incurred in active tubular reabsorption, and so would increase by about 60% during this period.

By far the greatest ion and fluid disturbances occur at the ECF/ICF interface. In freshwater trout, a shift of water out of the extracellular compartment occurs which is about sixfold greater than the elevated water entry at the gills (Fig. 3). This shift is largely entrained by the osmotic effect of the high muscle lactate load (Fig. 4; Milligan and Wood, 1986a,b). Swelling of red blood cells (RBCs) in response to acidosis and elevated plasma catecholamine levels also occurs (Nikinmaa, 1982; Primmett et al. 1986; Milligan and Wood, 1987a; Milligan et al. 1989; Wood et al. 1990b), but can account for less than 10% of the observed response.

There are also large net shifts of Na⁺ and Cl⁻ into the intracellular compartment which are 5- to 20-fold greater than their net losses through the gills (Fig. 3). In contrast, a smaller net flux of K⁺ into the extracellular compartment occurs, similar in quantity to the branchial loss of K⁺. The site(s) of these fluxes are unknown. However, again it is probable that white muscle plays an important role. Part of the Cl⁻ influx could be coupled with lactate efflux, and part of the Na⁺ influx with efflux. In mammals, exhausted acidotic muscle becomes more permeable and slightly depolarized, gaining Na⁺, Cl⁻ and water and losing K⁺ (Sjogaard, 1986; Kowalchuk et al. 1988). The post-exercise catecholamine surge may also be involved; catecholamines are known to generally increase cellular permeability to Na⁺ and K⁺ (Horwitz, 1979). Catecholamines also cause large net entries of Na⁺ and Cl⁻ into fish RBCs, fluxes that are associated with intracellular pH (pHi) regulation (Ferguson and Boutilier, 1989; Nikinmaa et al. 1990). At most, this RBC response can account for only 10% of the observed net ECF/ICF Na⁺ and Cl⁻ shifts, but a similar phenomenon occurring to only a very slight extent in other tissues throughout the body would explain the observed data. The metabolic costs of restoring these massive shifts at the ECF/ICF boundary could well be greater than those incurred at the gills and kidney.

is produced in excess of lactate in white muscle after exhaustive exercise stress, probably because of degradation of ATP stores (Milligan and Wood, 1986b, 1987c). The movements of lactate and out of white muscle into the extracellular compartment appear to be dissociated processes. Active, pelagic fish (e.g. trout, dogfish, tuna) may release appreciable quantities of lactate such that the lactate load considerably exceeds the metabolic acid load in the blood (Figs IB, 4A; Black et al. 1962; Piiper et al. 1972; Turner et al. 1983a; Holeton and Heisler, 1983; Perry et al. 1985; Milligan and Wood, 1986a,b; Dobson and Hochachka, 1987; Mommsen and Hochachka, 1988). In salmonids and some other species, this discrepancy occurs despite the fact that part of the extracellular results from β-adrenergically stimulated H⁺ extrusion from RBCs (e.g. Nikinmaa, 1982; Nikinmaa et al. 1984; McDonald et al. 1989a; Wood et al. 1990b). More sluggish, benthic species (e.g. flatfish, sea raven) release almost no lactate; blood lactate levels remain extremely low, generally less than 2 mequiv l⁻¹, considerably below (Fig. 4B; Wardle, 1978; Turner et al. 1983b; Milligan and Farrell, 1986; Milligan and Wood, 1987a,b,c; Milligan et al. 1991). These differences probably reflect strategies for recovery that are adaptive for pelagic versus benthic lifestyles, as discussed by Wood and Perry (1985). However, the key point is that the overall release is small; in both strategies and at all times after exhaustive exercise, most of the lactate and is retained in the white muscle (Fig. 4) for apparent metabolism in situ (see below).

Even at rest, lactate concentrations in white muscle ICF are far greater than in extracellular fluid (Fig. 4). Intracellular [H⁺] is also greater than extracellular [H⁺]: pHi is 0.23 (flounder; Milligan and Wood, 1987c) to 0.56units (trout; Milligan and Wood, 1986a) lower than extracellular pH (pHe). Assuming a typical membrane potential of −83 mV for fish muscle, both lactate and H⁺ distributions are maintained well out of electrochemical equilibrium (Wright et al. 1988). Intracellular [lactate] is too high and intracellular [H⁺] too low to occur as a result of a passive distribution; the situation must be supported by the expenditure of metabolic energy. Despite the large measured changes in intracellular and extracellular [lactate] and in pHe and pHi caused by exhaustive exercise, calculations based on these data suggest that these disequilibria either do not change appreciably (flounder) or decline slightly (trout) relative to the resting state. However, the metabolic cost of maintaining them will increase approximately in proportion to the passive rates of H⁺ and lactate leakage. The labelled lactate turnover studies of Milligan and McDonald (1988) indicate that these leakage rates increase threefold (flounder) to ninefold (salmon) during the 0–2 h period of post-exercise recovery.

As yet, we know little about the mechanisms of, or the controls for, transmembrane and lactate flux. Blood flow to white muscle increases substantially after exhaustive exercise (Wardle, 1978; Neumann et al. 1983), so perfusion limitations are probably not involved in the retention phenomena. However, in light of the preceding discussion it seems likely that lactate retention involves an active inward transport from ECF to muscle ICF. This idea is supported by the experiments of Batty and Wardle (1979) on plaice after exercise, using labelled lactate. An apparent net uptake of lactate into white muscle occurred against the concentration and electrical gradients. Similarly Turner and Wood (1983) demonstrated that treatment with SITS, an anion exchange blocked increased net lactate release from a perfused post-exercise trunk preparation of the trout. In plaice, Wardle (1978) provided pharmacological evidence that catecholamine mobilization, acting via β-adrenoreceptors, played an important role in preventing lactate release from white muscle. However, repeated attempts to confirm this phenomenon, in both flatfish (Wood and Milligan, 1987) and trout (van Dijk and Wood, 1988; McDonald et al. 1989a; Tang et al. 1989), have proved unsuccessful, β-Adrenergic stimulation does appear to lower blood after exercise, despite a stimulation of proton extrusion from RBCs, but this can be explained by a stimulation of efflux across the gills, rather by than an effect on release from white muscle.

The model calculations of Holeton and Heisler (1983) for dogfish suggest that the net movement of from muscle ICF to ECF is subject to ‘equilibrium limitation’ such that efflux falls as pHe falls. This idea is supported by the observation that net release from the perfused trout trunk is inhibited by reduced pHe, while lactate release is unaffected (Turner and Wood, 1983). It is debatable whether the process is passive, because in both cases the electrochemical gradient still favours the net inward movement of H⁺ from ECF to ICF. However, an analysis based solely on H⁺ concentration gradients may be overly simplistic; relative buffer capacities and the balancing movements of strong electrolytes are probably also important. This whole area clearly deserves further investigation.

The classical mammalian view of lactic acid disposal and glycogen resynthesis is described by the Cori cycle. Muscle lactate is released into the bloodstream and taken up by the liver. Here the oxidation of a small fraction of the lactate provides the necessary ATP to resynthesize the major portion into glucose, which, in turn, is transported back to muscle to recharge glycogen reserves. This scenario has now been seriously questioned in mammals, and considerable evidence for oxidation or glyconeogenesis in situ, or in other muscles, has been now been gathered. The whole area remains controversial (see Gaesser and Brooks, 1984; Peters-Futre et al. 1987). In fish, a substantial body of evidence has now accumulated suggesting that the Cori cycle does not play an important role in the disposal of the postexercise lactate burden. Whatever the exact route of metabolism or its contribution to EPOC, most of the lactate and appears to stay in the white muscle for metabolism in situ.

In brief, this evidence consists of the following points, (i) The bulk of the lactate and load is found in the white muscle at all times after exercise (see above), (ii) The ability of teleost hepatocytes to perform gluconeogenesis from lactate in vitro is minimal (Walsh et al. 1985, 1988; Mommsen et al. 1988). For example, in the toadfish, the measured rate of gluconeogenesis by hepatocytes could account for only 2% of post-exercise lactate clearance (Walsh, 1989). (iii) In the only in vivo study of Cori cycle activity in a fish after exercise, Weber et al. (1986) reported that less than 5% of labelled lactate was converted to glucose in skipjack tuna. In resting eels, this conversion was 35% (Cornish and Moon, 1985). (iv) The turnover rate of the blood lactate pool, measured by the use of labelled lactate in tuna (Weber et al. 1986), salmon and flounder (Milligan and McDonald, 1988), was elevated after exhaustive exercise, but was still far too low to explain the observed rates of clearance of muscle lactate. This last piece of evidence would tend to rule against not only the Cori cycle, but any mechanism of lactate disposal based primarily on release to the bloodstream, (v) The turnover rate of the blood glucose pool after exhaustive exercise, measured by the use of labelled glucose, was far too low to account for the observed rate of muscle glycogen restoration in the plaice (Batty and Wardle, 1979). Very recently, A. Pagnotta and C. L. Milligan (unpublished results) found that incorporation of labelled blood glucose accounted for less than 1% of resynthesized muscle glycogen in both trout and flounder, (vi) Finally, the original study of Batty and Wardle (1979) also indicated that labelled lactate, injected into the muscle of exhausted plaice, was directly resynthesized into glycogen. Exact quantification was problematical; this critical observation should be replicated.

The logical conclusion to be drawn from this evidence is that the bulk of lactate and are metabolized within the white muscle, either by oxidation or by glyconeogenesis. The small amounts released into the bloodstream are probably oxidized by other tissues such as liver (Mommsen et al. 1988), heart (Milligan and Farrell, 1991), red muscle (Bilinski and Jonas, 1972), gill (Mommsen, 1984) and RBCs (Wood et al. 1990a). The ability of teleost white muscle to oxidize lactate has also been documented (Bilinski and Jonas, 1972), though proof that a full complement of gluconeogenic enzymes exists in this tissue is still lacking.

The relative proportions cleared by the oxidative and glyconeogenic pathways in situ, and the proportion of glycogen resynthesized from lactate, may vary with species and with physiological condition. For example, in trout, the amount of lactate accumulated per unit of glycogen depleted varies greatly, depending on factors such as feeding condition (Scarabello et al. 1991a,b) and training (Pearson et al. 1990). Clearly, the proportion of glycogen that can be resynthesized from lactate will be limited by the size of the available lactate pool. The situation is further complicated by the fact that glycolysis may continue to produce lactate for some time after the end of exercise, even while lactate disposal is proceeding (Dobson and Hochachka, 1987; Lackner et al. 1988; Scarabello et al. 1991a,b; A. Pagnotta and C. L. Milligan, unpublished results). Acid–base status may be an important regulator of the fate of lactate. In trout, Milligan and Wood (1986b) demonstrated that the first 4 h of post-exercise recovery was a period of intense intracellular acidosis, during which there was considerable lactate and clearance from white muscle (Fig. 4A), apparently by oxidation. Restoration of glycogen levels started only once muscle pHi had been partially corrected. Catecholamines may be another important control; adrenergic stimulation of gluconeogenesis has been demonstrated in teleost hepatocytes (Mommsen et al. 1988; Danulat and Mommsen, 1990) and reptilian skeletal muscle (Gleeson and Kolok, 1990).

The preceding discussion has repeatedly implicated the post-exercise catecholamine surge as a modulator of responses during recovery. In mammals, general stimulatory effects of catecholamines on metabolism are thought to be another Significant contributor to EPOC (Barnard and Foss, 1969; Gaesser and Brooks, 1984; Bahr et al. 1987). These actions may be either direct or indirect; that is, vs increased Na⁺ and K⁺ permeabilities (Horwitz, 1979) or by increased ‘futile’ substrate cycling (Newsholme, 1978). This raises the question of whether metabolic effects of catecholamines are a significant contributor to EPOC in fish.

In addressing this question, it must be remembered that the catecholamine surge occurs at a time of intense acidosis and internal disturbance, not under resting conditions. Both in vivo and in vitro, many earlier studies have shown that high levels of exogenous catecholamines will stimulate various aspects of aerobic metabolism (Mazeaud and Mazeaud, 1981). However, only recently has there been an emphasis on duplicating the in vivo conditions in terms of realistic postexercise catecholamine levels, substrate levels and acid-base status. In general, these studies have demonstrated that altered acid–base status (decreased pHe and [HCO₃⁻]e, increased ⁠) and substrate levels (elevated blood glucose and lactate concentrations) are at least as important as catecholamines, if not more so, in accounting for observed changes in post-exercise metabolism.

For example, trout RBCs substantially increase their rate of lactate oxidation (and decrease their rate of glucose oxidation) during recovery from exhaustive exercise stress (Fig. 5A; Wood et al. 1990b; Walsh et al. 1990). In vitro factorial experiments demonstrated that more than half of this response was due to elevated blood [lactate], with additional smaller stimulations by elevated and elevated catecholamine levels. Interestingly, catecholamines exerted this stimulatory effect only in the presence of acidosis, and were without effect at normal blood pH. Total RBC stayed more or less constant during recovery (Fig. 5B). However, would have been markedly depressed by post-exercise acidosis, were it not for the presence of the elevated catecholamine levels, which restored it to normal levels. Note that RBC was significantly depressed only at 2 h postexercise, when the catecholamine surge had virtually disappeared, but acidosis persisted (Fig. 5B). Again, catecholamines had no effect at normal blood pH. Studies with other in vitro systems, such as the trout heart (Farrell and Milligan, 1986; Farrell et al. 1988), trout hepatocytes (Mommsen et al. 1988; Walsh et al. 1988; Perry et al. 1988) and toadfish hepatocytes (Walsh, 1989), have identified a variety of stimulatory or inhibitory effects of acid–base variables or substrate levels. In general, catecholamines appear to have little influence under control circumstances, when they would not be mobilized anyway. However, during postexercise acidosis, they play an extremely important adaptive role in stimulating, protecting or modifying metabolism, but not in elevating tissue above resting level.

In vivo, there have been no tests of the effects on whole-body of catecholamine infusion or adrenergic blockade during post-exercise recovery. However, Butler et al. (1989) reported that infusion of physiological doses of adrenaline and noradrenaline slightly improved the aerobic swimming performance of cod in which endogenous catecholamine mobilization had been largely blocked by denervation of the head kidney. In resting trout, administration of exogenous adrenaline and noradrenaline, at doses either duplicating or exceeding typical post-exercise levels, caused a negligible elevation of whole-body (Steffensen et al. 1987; Playle et al. 1990). While there is clearly a need for more work both in vivo and in vitro, a tentative conclusion is that catecholamine effects on are not an important contributor to EPOC in fish.

Whatever factors are responsible for EPOC, it is essential that the fish maintain an increased ventilation for some time during post-exercise recovery to facilitate this elevated The control of ventilation in fish has recently been reviewed elsewhere (Perry and Wood, 1989), so only points relevant to the postexercise situation will be recapped here. The primary ventilatory drive is set by O₂, probably through detection of arterial blood O₂ content (Smith and Jones, 1982). However, during recovery from exhaustive exercise stress, arterial O₂ content is close to normal (Primmett et al. 1986; Milligan and Wood, 1987a) and the fish are generally motionless, so neither hypoxaemia nor proprioceptive input can be important stimuli for increased ⁠. Catecholamines have been implicated in direct ventilatory stimulation (Peyraud-Waitzenegger 1979; Aota et al. 1990). However, recent studies in which physiological doses of adrenaline or noradrenaline were infused into resting or acidotic fish have shown negligible or hypoventilatory effects (Playle et al. 1990; Kinkead and Perry, 1990, 1991). In two elasmobranchs, the dogfish and the skate, blood acid-base status appears to play an important role in ventilatory control, probably acting through arterial pHa (Heisler, 1989; Wood et al. 1990a). If the same is true in all fish, then the large post-exercise acidosis of both metabolic and respiratory origin (Fig. 1A,B,C) may be the primary stimulus driving post-exercise hyperventilation, and catecholamines may play an important indirect role.

Both the cause and the functional significance of the respiratory component of post-exercise acidosis have been controversial (Wood and Perry, 1985; Perry, 1986; Perry and Vermette, 1987; Steffensen et al. 1987; Tufts et al. 1988; Playle et al. 1990). However, recently, Wood and Perry (1991) and Perry et al. (1991b) have confirmed that catecholamines inhibit RBC CO₂ excretion in trout in vitro by a β-adrenergic mechanism linked to Na⁺/H⁺ exchange on the erythrocyte membrane. The same inhibition is seen in vivo during post-exercise recovery (C. M. Wood and R. S. Munger, unpublished results). In effect, the mechanism that protects RBC pHi and blood O₂ transport (Nikinmaa, 1982; Nikinma et al. 1984, 1990; Primmett et al. 1986) simultaneously reduces the rate of HCO₃⁻ dehydration by RBCs passing through the gills, causing CO₂ to back up in the animal. A marked increase in and decrease in pHa result (Fig. 1A,C). The phenomenon appears to be coupled to the decrease or reversal of the extracellular–intracellular pH gradient across the RBC membrane (Wood and Perry, 1991). All fish probably exhibit a decrease in this pH gradient after exhaustive exercise stress, simply as a result of the different buffer capacities of the two compartments. However, it remains to be seen whether this effect alone is sufficient to explain the post-exercise increase in exhibited by species where the adrenergically regulated Na⁺/H⁺ is weak or lacking (e.g. elasmobranchs; Holeton and Heisler, 1983).

In any event, the respiratory component exacerbates the metabolic acidosis (due to movement out of muscle and RBCs) in all fish that have been examined (Wood and Perry, 1985). This large compound acidosis probably serves to stimulate after exercise in the absence of other appropriate stimuli. Recent experiments on trout support this interpretation (C. M. Wood and R. S. Munger, unpublished results). Relative during recovery from exhaustive stress is related to the decline in pHa in a simple exponential fashion (Fig. 6). Intra-arterial injection of sufficient bovine carbonic anhydrase to double the net HCO₃⁻- dehydration activity of the blood does not affect exercise performance, but reduces the post-exercise elevation in by about 50%. As a result, the decline in pHa is greatly reduced; this, in turn, is correlated with a substantial reduction in the extent of post-exercise hyperventilation (Fig. 6) and therefore in ⁠. In this context, post-exercise acidosis appears to be adaptive in sustaining elevated to meet the demands of EPOC.

In an earlier review (Wood and Perry, 1985), we pointed to the need for direct determinations of plasma catecholamine levels to support the suspected involvement of these stress hormones in a variety of responses to exercise. The intervening 6 years have seen a substantial increase in such measurements, as well as in tests of realistic doses of catecholamines under realistic conditions. From the preceding discussion, it is now clear that circulating catecholamines are critically involved in many facets of post-exercise physiology. Overall, elevations in circulating adrenaline and noradrenaline levels are probably more important in recovery from exhaustive exercise stress than they are in steady-state aerobic swimming. The next stage is to probe the cellular bases of these effects, and then to integrate them back into the response of the whole animal. In this regard, recent reports that adrenoreceptor density can be rapidly adjusted in stress situations are particularly intriguing (Marttila and Nikinmaa, 1988; Reid and Perry, 1991).

I thank Drs D. G. McDonald, C. L. Milligan, S. F. Perry, and R. Gonzalez for helpful discussions. Original research reported here was supported by NSERC grants to C.M.W.