Control and Co-Ordination of Ventilation and Circulation in Crustaceans: Responses to Hypoxia and Exercise

The phylum Arthropoda, by virtue of number of species, biomass, range of adaptability, economic importance, apparent ease of sampling and many other criteria has been a popular alternative group to the vertebrates for physiological study. The major class Insecta typically transport respiratory gases directly to and from the tissues in the tracheal system, by-passing the circulatory system. This puts them outside the scope of this review which considers some members of the class Crustacea. Research into the respiratory and circulatory physiology of crustaceans has largely concentrated on the suborder Reptantia of the order Decapoda (the lobsters, crayfish and crabs). The range of environments they inhabit includes the abyssal, sublittoral and littoral marine, estuarine, freshwater and terrestrial and they may walk, run, climb, swim or burrow (Warner, 1977). Individual species show wide ranges of activity levels, are highly adaptable to variables such as temperature (recently reviewed by Taylor, 1981), salinity and oxygen tension and show behavioural modifications including annual and diurnal migrations which affect the range of variables they encounter. This review is restricted to the Reptantia and considers the exchange of respiratory gases in relation to: the functioning and control of the open circulatory system or haemocoel, which typically contains the respiratory pigment haemocyanin in solution conferring on the haemolymph a relatively high affinity for oxygen but a relatively low and variable capacity both for O₂ and the buffering of CO₂ ; ventilation of the respiratory gas exchange surfaces, which are covered by a thin but possibly relatively impervious layer of the chitinous exoskeleton. The physiological adjustments observed when O₂ demand outstrips supply during hypoxia, aerial exposure and exercise are described. Symbols used to denote respiratory variables and indices are chiefly those recommended by Dejours (1975) and are defined in the text when first employed.

The general features of the circulatory system in crustaceans were reviewed by Maynard (1960) and are illustrated in Fig. 1(a). In the Decapoda there is no separation between interstitial fluids and the circulation, which is an open haemocoel. The circulated fluid is termed haemolymph. Haemolymph volume is relatively large and varies with mass from 20–50% of total volume (e.g. Taylor & Wheatly, 1982). The main propulsive organ is the dorsal, single-chambered heart which is composed of striated muscle and operates in a pericardial haemocoelic space from which it draws haemolymph at diastole via valved ostia. The heart is suspended from ligaments and contracts against their elasticity at systole. Diastolic filling is relatively slow and due primarily to elastic recoil in the ligaments returning the heart to its relaxed volume and causing haemolymph to enter from the surrounding pericardial space. Haemolymph is thus aspirated into the heart, which functions as a force-suction pump. Sheets of alary muscles connect the septum, forming the floor of the pericardial space, to the exoskeleton and may by contraction control the volume of the space and the tension on the suspensory ligaments thus influencing venous return and diastolic filling (Wilkens, 1981). Both cardiac stroke volume (SV_H) and heart rate (f_H) increase when the heart is stretched (Izquierdo, 1931) obeying Starling’s Law of the heart.

At systole the heart delivers haemolymph into several elastic walled but non-muscular arteries which are valved to prevent backflow and supply organs such as the C.N.S., appendages, gills, bladder and genitalia. The haemolymph runs into interstitial spaces or lacunae in the tissues then collects in large sinuses bordered by connective tissue. In larger and more active species some arteries terminate in fine vessels 2–10 μm in diameter and resemble capillaries (Sandeman, 1967). These are found in organs such as the C.N.S., the osmoregulatory green gland and the gastrolith fields of the stomach wall which are the sites of calcium storage whose possible significance is considered below. The haemolymph drains from the systemic circulation into several large sinuses which then open into an infrabranchial sinus running ventrally along either side of the body. In each segment of the thorax this sinus receives haemolymph from the locomotory appendage and supplies haemolymph to the afferent branchial sinuses in the gills. From the branchial lacunae the oxygenated haemolymph collects in the branchiopericardial veins which connect back to the pericardial sinus, and from there it enters the heart to be recirculated.

Measurements of the pressures and flows generated in the haemocoel have recently been summarized and compared with other invertebrates and fish by McMahon & Wilkens (1982). In the lobster Panulirus interruptus ventricular systolic pressure was 36 mmHg and diastolic pressure was 16 mmHg (Behnan, 1975, 1976). Elastic compliance kept arterial pressure close to systolic pressure. Mean pressure in the infrabranchial sinus was 19 mmHg and in the pericardial sinus 18 mmHg. Blatchford (1971) measured lower pressures throughout the haemocoel of Carcinus maenas and a pressure differential of approximately 2 mmHg across the branchial vasculature. These low pressures are accompanied by a relatively high cardiac output which implies that peripheral resistance in the circulation is low. The high cardiac output, which is 3–4 times higher than in fish if equivalent mass relates to the relatively low carrying capacity of the haemolymph for oxygen, though the total capacity per unit mass is similar for crustaceans and fish because haemolymph volume is approximately 10 times higher than blood volume and both comprise a similar venous reserve of oxygen for use during periods of oxygen lack or increased demand.

The action of the heart appears sufficient to account for flow through the entire circulatory system (Burger & Smythe, 1953; Blatchford, 1971; Belman, 1976), though perfusion of the gills, which receive haemolymph after its passage through the systemic circulation, may be aided by the aspiration of haemolymph from the branchiopericardial veins into the pericardial sinus, which shows pulsatile reductions in pressure as the volume of the heart decreases at systole (Belman, 1976). Another factor possibly aiding flow through the branchial circulation is the subambient hydrostatic pressure generated in the branchial chambers by the scaphognathites during forward ventilation, which may cause distension of the vasculature and further reduce resistance to flow (Burger & Smythe, 1953 ; Blatchford, 1971). Direct evidence of this effect has recently been obtained (L. E. Burnett – personal communication) and is referred to below. Progressively lower pressures are generated during periods of hyperventilation and if this were reflected in further reductions in branchial vascular resistance it could represent a direct physical mechanism matching ventilatory and perfusion flows to satisfy a demand for increased rates of gas exchange.

Movements of the locomotory appendages by active animals may promote the flow of haemolymph through the branchial circulation by aiding venous return, though the increased pressures inthe limb sinuses during movement may not cause directional flow through the gills if no valves are present (Burger & Smythe, 1953). Recent observations indicate some rectification of flow through the branchial circulation of Carcinus gills, possibly attributable to valve-like structures identified at the openings into the efferent, hypobranchial vessels (H. H. Taylor and E. W. Taylor, unpublished observations).

The anatomical arrangement of the circulatory system renders direct measurement W haemolymph flow extremely difficult. Measurements of flow using a thermodilution technique (Burnett et al. reported by Wilkens, 1981) indicated that cardiac stroke volume (SV_H) varied widely in three species of crabs. The range of variations in heart rate (f_H) observed in response to disturbance, hypoxia and exercise is often compensated or amplified by apparent changes in SV_H which seems to be the more important variable. Wilkens (1981) explored the theoretical reasons, based on the mechanics of heart pumping, for this observed relationship. Clearly, measurements of f_H cannot be a reliable guide to cardiac output.

Estimations of haemolymph flow based on the Fick principle may be suspect as the popular sampling site at the base of a walking leg cannot yield a mixed venous sample and measurements of O_a content should be referred to as C_v, O₂ levels, rather than ⁠,O₂ (Taylor et al. 1977a;,Taylor & Butler, 1978). Another objection to sampling haemolymph from this site is that it often involves removing aquatic crustaceans from water, which rapidly results in internal hypoxia and consequent major distortion of blood gas measurements (Taylor et al. 1973). Use of the Fick principle necessitates continuous monitoring of heart beat as cardiac arrhythmia and arrest, associated with ventilatory changes, are common in both settled (e.g. Butler et al. 1978), and disturbed crustaceans, and this rapidly leads to a decline in haemolymph (e.g. Taylor et al. 1973).

In crustaceans muscle contraction is induced primarily by small, cumulative depolarisations of the muscle fibre membranes called junctional potentials (Fass & Katz, 1953) which arise from a number of motor endplates on each fibre (Hoyle & Wiersma, 1958). Propagated muscle action potentials are not necessary for contraction.

These characteristics rule out the possibility of a functional myogenic pacemaker, as described in mammals. Intracellular recordings of voltage changes in myocardial cells during contraction are superficially similar to those recorded from the mammalian heart but consist of summated junctional potentials which arise from activity in 4 or 5 nerve axons supplying each muscle fibre (Van der Kloot, 1980). The simultaneous contraction of the whole myocardium at systole is achieved by a burst of propagated action potentials travelling from a group of neurones embedded in the dorsal wall of the heart, and called the cardiac ganglion, which constitute the neurogenic pacemaker of the heart (Fig. 1b). Early descriptions of its form and function were reviewed by Maynard (1960), and its characteristics are summarized in Fig. 3. Typically, it consists of 9 neurones ; 4 posteriorly placed, small cells around 50 μm in diameter and 5 anteriorly placed larger cells, 100–200 μm in diameter. Each cell is multipolar with numerous axons, collaterals and complex dendrites, termed dendritic arborizations and speculatively assigned the function of stretch receptors. There is anatomical evidence for reciprocal contact at sites on the various processes and soma of all 9 neurones. Extracellular recordings from the trunk of the cardiac ganglion of the crab Portunus sanguinolentas consisted of spontaneous nervous activity organized into regularly patterned, rhythmic bursts of action potentials from small and large cells (Tazaki & Cooke, 1979). Activity in the processes of small cells started and ended each burst and they fired trains of impulses throughout the bursts which were not synchronized with other cells. The 5 large cells fired trains of impulses which were synchronized, possibly due to electrotonic coupling (Watanabe & Bullock, 196GB and slower than the small cell trains. Intracellular recordings from the soma of small cells showed a slow, ‘pacemaker’ depolarization to a plateau on which were superimposed attenuated action potentials which were passing along the processes of the cell but not invading its soma. Intracellular recordings from large cell soma showed post-synaptic potentials which corresponded with spiking activity in small cells, attenuated spikes and an underlying slow depolarization during bursts with no pacemaker depolarization apparent between bursts. The morphological distinction between small and large cells appeared to reflect a functional difference, with the small cells comprising the primary pacemaker, determining bursting rate (Hagiwara & Bullock, 1957), possibly by acting as endogenous burst oscillators (Otani & Bullock, 1959). The system seems essentially two-layered with electrotonic and synaptic contact from the small pacemaker cells serving to drive the large motor neurones which act as ‘follower cells’ innervating the myocardium (Mayeri, 1973). The subsequent driver potential which arose on the large cell was the source of the depolarizing current which initiated its burst of action potentials and in turn caused the myocardium to contract. Injection of depolarizing and hyperpolarizing currents into large cell soma at various phases of the cycle of bursting activity in the ganglion revealed that the interburst interval related to the previous driver potential duration and that various phase shifts could be induced which indicated that the cardiac ganglion functioned as a ‘relaxation oscillator’ with pacemaker cell function modified by electrotonic feedback from the follower cells (Benson, 1980). When large cell driver potentials were eliminated with octopamine small cell activity increased in frequency suggesting that the small cells were held in a depolarized state by electrotonic spread of the large cell driver potentials and that their duration may be the factor determining the frequency of the cardiac rhythm. This description renders the cardiac ganglion susceptible to entrainment, by virtue of induced phase shifts, in response to externally applied current pulses and comprises a mechanism for the potential synchronization of cardiac activity with other rhythmic behaviours.

Action potentials in the neurones of the cardiac ganglion do not invade the soma so that two separate impulses may travel simultaneously in different processes from the same neurone without interference (Bullock & Terzuolo, 1957). The crustacean heart is known to respond to distension with an increase in rate and amplitude and if this can be demonstrated to be modification of motor output to the myocardium arising from stimulation of the ‘dendritic arborizations’ then they represent the afferent limb of a single-neurone reflex arc (Maynard, 1960).

The crustacean heart receives inhibitor and accelerator nerves from either the suboesophageal or thoracic ganglia which run in separate trunks to the pericardial sinus where some axons supply a pair of neurohaemal, pericardial organs, others supply a plexus on the suspensory ligaments and one fibre from each trunk is combined into a dorsal nerve which enters the heart to innervate the cardiac ganglion as shown in Fig. 1b (Alexandrowicz, 1932; Maynard, 1953, 1960). Peripheral electrical stimulation of the inhibitor nerve trunks causes a marked bradycardia followed by adaptation with no marked after effect. The inhibitory transmitter may be y amino butyric acid (Florey, 1963). The accelerators show no adaptation and a prolonged after effect (Wiersma & Novitski, 1942; Maynard, 1953; Florey, 1960) which is mimicked by acetylcholine though the transmitter is possibly an indol alkylamine (Florey, 1963). Maynard (1960) concluded that the cardioregulatory nerves were probably involved in brief reflex inhibition and acceleration of the heart. They are, however, continuously active in the prepared animal (Taylor, 1970; Field and Larimer, 1975 a; Young, 1978) and may exert a tonic control over heart rate similar to the vagal tone operating on the fish heart (e.g. Taylor, Short & Butler, 1977).

The cardioregulatory effects are imposed directly upon the cardiac ganglion. Intracellular recordings from the large follower cells in Panulirus interruptus included i.p.s.p. resulting from stimulation of inhibitor nerve trunks which were either hyperpolarizing or depolarizing depending upon the membrane potential, and showed both facilitation and summation (Terzuolo & Bullock, reported by Maynard, 1961). The i.p.s.p. could either block or potentiate the development of the generator potential and consequent spiking potentials, though stimulation of the inhibitor nerves typically caused a drop in the frequency of the trains of impulses from the ganglion plus a reduction in the number and frequency of impulses within each train (Maynard, 1961).

Intemeurones, referred to as cardiac command fibres by Field and Larimer (1975b), have been identified in the circumoesophageal connectives which when stimulated electrically cause a tachycardia (the accelerators), cardiac arrest (the strong inhibitors) or a bradycardia (the weak inhibitors) often followed by a post-stimulation rebound (Wiersma & Novitski, 1942; E. W. Taylor, unpublished observations). They are tonically active (Wilkins et al. 1974) and when stimulated either directly or by sensory input may induce reciprocal activity in inhibitor and accelerator outputs (Field & Larimer, 1975a, b).

Substances which affect the beating of the heart have been extracted from a number of potential sites of hormone production (Maynard, 1960). The pericardial organs (P.O.) located in the pericardial space on either side of the heart, which receive axons from the cardioregulatory nerve trunks, appear to be directly involved in hormonal control of the heart. Extracts of P.O. increase the frequency and amplitude of heart beat and may restore steady beating to an arrhythmic preparation (Alexandrowicz & Carlisle, 1953). Cooke (1966) identified the site of action of the extracts as the cardiac ganglion and Cooke & Hartline (1975) pinpointed the effect to trigger zones near the soma of the neurones, particularly the small pacemaker cells, where application caused increases in the rate and duration of the bursts of spiking potentials. Electrical stimulation of the P.O. caused graded cardioacceleration proportional to the number of stimuli given (Cooke, 1964).

The effect of P.O. extract was mimicked by 5-hydroxytryptamine (5HT) (Carlisle, 1956; Maynard & Welsh, 1959) and the P.O. produces a range of biogenic amines including HT, octopamine and dopamine. All three amines cause cardioacceleration in the crayfish Astacus leptodactylus with HT the most effective (Florey & Rathmayer, 1978). Berlind and Cooke (1968) concluded that 5-HT and dopamine levels in the P.O. of Libinia were too low to be effective in cardioregulation and attributed this role to peptides which were released from the P.O. by electrical stimulation at levels causing cardioacceleration. This confirmed the identification of cardioexcitor peptides from extracts of P.O. (Belamarich & Terwilliger, 1966). A substance similar in properties to proctolin, a putative peptide transmitter from insect gut, was isolated from the P.O. of Cardisoma and shown to have an inotropic effect on the heart (Sullivan, 1979). Neurones in the C.N.S. which produce biogenic amines respond to sensory input, are coupled together electrotonically, subject to autoregulation and become spontaneously active above 14 °C (Konishi & Kravitz, 1978), firing bursts of impulses at high temperatures similar to those described in the cardioaccelerator nerve trunks which supply the P.O. (Taylor, 1970).

The site of release of these hormones by the P.O. is directly upstream of the heart and they are aspirated from the pericardial space into the heart at diastole. A similar functional location has been described for the chromaffin tissue which produces biogenic amines in the cardinal sinuses of elasmobranch fishes (Butler & Taylor, 1975). It is possible that tonic control of heart rate and possibly cardiac stroke volume may be exerted by hormones released from the P.O. possibly by the bursts of activity recorded from the accelerator nerve trunks of the crayfish, which corresponded to periodic variations in cardiac ‘tone’ (Taylor, 1970). Hormones produced by the P.O. may coordinate heart and scaphognathite beating as discussed below.

The exchange of respiratory gases between the environment and circulatory system of decapod crustaceans is by passive diffusion and is virtually restricted to the elaborate gills, developed as outpushings of the body wall associated with the bases of the major appendages and enclosed in branchial chambers formed by folds of the body wall termed branchiostegites (Fig. 2a). Consequently it may be adequately described by derivations from Fick’s Law of diffusion (Dejours, 1975). The functional morphology of the gills combines minimum diffusion distance with maximum surface area for exchange thus ensuring a high diffusing capacity favouring effective gas exchange. The gills may possess regularly stacked flattened lamellae as in the phyllobranchiate gills of most brachyuran crabs or consist of filaments which may be single (tricho-branchiate) as in the crayfish or branched (dendrobranchiate) as in some shrimps.

The fine structure of phyllobranchiate gills has been described by Taylor & Butler (1978) and Taylor & Greenaway (1979). As they are extensions of the outer surface, the gills are covered with a thin layer of chitin 2 –5 μm thick. The shortest diffusion distance is 5–8 μm which includes processes of epithelial cells. This is similar to the diffusion barrier described on fish gills (Hughes, 1972) and other aspects of gill structure such as the arrangement of pillar cells and blood spaces are superficially similar to the structure of fish gill secondary lamellae (e.g. Laurent & Dunel, 1980) though the exchange surfaces on crustacean gills are the morphological equivalents of the primary lamellae. Chitin on the gills presents a special problem as its diffusion constant for oxygen is reportedly 10% of the value for tissue (Krogh, 1919) so that the effective diffusion barrier over Carcinus gills may be 5-10 times higher than over fish gills with 80% of it attributable to chitin (Taylor & Butler, 1978). Compensation for this potential problem may be reflected in the large gill areas in aquatic crustaceans relative to their mass and oxygen requirements. Weight specific gill areas for a number of crustacean species (Gray, 1957) fall within the range given for teleost fish by Hughes & Morgan (1973) with the relatively sluggish crustaceans having gill areas similar to those of more active fish species. The chitin layer may be relatively thin over the gills of animals with a high O_a demand such as the swimming crab Callinectes (Booth et al. 1982) or a limited O₂ supply such as Gnathophausia, which inhabits a hypoxic environment (Belman & Childress, 1976) and is 300 nm over the inner wall of the branchiostegite which functions as a Tung’ in the land crab Holthusiana (Taylor & Greenaway, 1979).

Ventilation of the branchial chambers on each side of the animal is accomplished by a baler called the scaphognathite, a specialized appendage which morphologically is a flattened exopodite of the second maxilla, one of the mouthparts (Fig. 2b). The blade of the scaphognathite oscillates dorsoventrally with a complex sinusoidal action (Young, 1975) in the narrow exhalent channel at the front of each branchial chamber (Fig. 2a). In the typical forward mode of ventilation the scaphognathite generates a pulsatile hydrostatic pressure which is below ambient by 1–5 cm H₂O, depending on the mass of the animal and rate of ventilation (e.g. McDonald et al. 1977; Taylor & Wheatly, 1980). Consequently water is sucked into the branchial chambers and over the gills via a series of openings around the bases of the walking legs. In brachyuran crabs such as Carcinus maenas up to 80 % of total ventilatory flow enters through the Milne-Edwards openings at the bases of the chelipeds to supply the anterior 7 gills on each side (Hughes et al. 1969). Pressure recordings from the branchial chamber of a number of aquatic crabs (Hughes et al. 1969; Taylor et al. 1973; McMahon & Wilkins, 1977) lobsters (McMahon & Wilkins, 1975; Butler et al. 1978) and crayfish (McMahon et al. 1974; Taylor & Wheatly, 1980) demonstrate a typical biphasic waveform. Both strokes of the scaphognathite generate forward force (Wilkens & McMahon, 1972; Young, 1975) and are equally involved in movement of water (McDonald et al. 1977; Batterton & Cameron, 1978). Most species ventilate the gills predominantly in the forward direction and a functional counter-current flow of water and blood was described over the lamellae of the phyllobranchiate gills of Carcinus maenas in this mode (Hughes et al. 1969) though values are typically lower than (e.g. Wheatly & Taylor, 1981) and the benefits of counter-current exchange may be masked by a large ventilatory dead-space in the branchial chambers (Johansen et al. 1970; McMahon & Wilkens, 1982).

The beat of the scaphognathite is periodically reversed, in some species, particularly macrurans, for 1 or 2 beats (e.g. Wilkens & McMahon, 1972; Taylor & Wheatly, 1980), in others, typically the brachyuran crabs, for bursts of several beats (e.g. Hughes et al. 1969; Taylor et al. 1973). Several roles have been speculatively attributed to scaphognathite reversals including clearing the gills or apertures of detritus, flushing poorly ventilated areas of the branchial chambers or sampling the water ahead of the animal (Borradaile, 1922; Arudpragasam & Naylor, 1964a; Hughes et al. 1969). Reversals may predominate in buried crabs when forward ventilation is precluded (Arudpragasam & Naylor, 1966; Bridges, 1976). The frequency and duration of reversals increased during progressive hypoxia in Carcinus, and were employed to aerate the branchial chamber (Taylor & Butler, 1973).

Immediately after experimental manipulation crustaceans typically ventilate both branchial chambers continuously. After settling in the experimental regime they often show prolonged respiratory pauses or unilateral ventilation (McDonald et al. 1977; Butler et al. 1978). Pauses may serve to avoid detection by a possible predator (McMahon and Wilkens, 1977) and Carcinus has been observed to show immediate cessation of ventilation, with an associated bradycardia in response to the appearance of the investigator (Taylor et al. 1973). They also decrease the energy cost of ventilation for the resting animal without necessarily depleting O₂ supply to the tissues which may be maintained for short periods due to the relatively large venous reserve of O₂ in the circulating haemolymph (see below).

The flow of water or air through the various apertures and the chambers is relatively complex (e.g. Hughes et al. 1969) and subject to control by movements of accessory organs such as the limb bases, particularly the 1st maxilliped which forms the floor of the exhalent canal (Wilkens, 1981) and the epipodite of the 3rd maxilliped which varies the size of the Milne–Edwards openings and may control the direction of water flow into the branchial chambers in brachyuran crabs (Borradaile, 1922). Movements of these organs may alter the resistance to flow of the apertures and chambers, particularly during reversals (McDonald et al. 1977). The medial, epimeral wall of the branchial chambers is flexible and can be pulled inward, enlarging the chamber and causing subambient pressures which may aid ventilation (Wilkens & McMahon, 1972), and is utilized by a number of air-breathing crabs (see below). Though the work of Batterton & Cameron (1978) established a relationship between f_R and in CaUinectes sapidus there are many examples of changes in independent of a change in f_R which implies that the stroke volume (SV_R) of the scaphognathite is variable (e.g. Arudpragasam & Naylor, 1964b, Wheatly & Taylor, 1981). Beat by beat measurements revealed that SV_R and f_R varied widely and often independently in crayfish and lobsters (Mercier & Wilkens, reported by Wilkens, 1981). Measurements of f_R alone cannot be a useful guide to relative rates of gill ventilation which vary during hypoxia, aerial exposure and exercise. Other factors which render indirect measurements of hazardous are the ventilatory reversals and pauses. Calculation of from measurements of and assuming the Fick relationship is applicable, ignores the possibility that a water sample collected from in front of the exhalent opening may give a fallaciously high value for apparent during a reversal or pause in ventilation (e.g. Butler et al. 1978), and similar sampling problems were encountered in air-breathing crabs during intermittent ventilation (Wood & Randall, 1981a). Many authors have restrained crustaceans in order to collect expired water (e.g. Arudpragasam & Naylor, 1964b; Taylor, 1976); this is undesirable as it is unlikely that they will show settled rates of ventilation. Other studies have used light hoods which separate the inhalent and exhalent streams without restraining the animals (e.g. Butler et al. 1978; Taylor & Wheatly, 1980) or small funnels located in the expiratory stream (Taylor et al. 1977a). These have seemed satisfactory, yielding values for percentage extraction of available O₂ from the inspired water (% Extn_W) of between 30 and 40 %, but have recently been criticized by J.-C. Massabuau (personal communication) who, sampling directly from a cannula inserted in the exhalent canal of the crayfish Astacus leptodactylus, obtained low values and a % Extn_W of approximately 80%.

The scaphognathite is powered by the complex musculature of the second maxilla (Fig. 2b) and studies of its motor control have established the functional role of these muscles in the rhythmic beating of the scaphognathite (e.g. Pasztor, 1968; Pilkington & Simmers, 1973). Young (1975) recognized 10 muscles responsible for scaphognathite movements and classified them into sets according to their functions. These functional divisions were reflected in their innervation and in the organization of bursting activity in the motor nerves from the C.N.S. During the beating cycle the muscles showed sharply defined bursts of excitatory junctional potentials in particular phases of the scaphognathite beat, leading to contraction and alternating with completely silent periods. The peak frequencies of spiking within the bursts were synchronized within muscle sets but coincidental with a trough in the temporally adjacent set, indicating positive synchronizing interaction between motor units within a set, possibly by electrotonic coupling, and negative interaction between units in temporally adjacent groups. Young (1975) related the sequence of burst patterns in the ventilatory muscles with the movements of the scaphognathite in both the forward and reversed modes and the reader is referred to his excellent account and to a review by Wilkens (1976) for an understanding of scaphognathite function and control.

The second maxilla receives a single nerve from the C.N.S. which divides into 3 trunks, one sensory, one mixed, the other predominantly motor (Fig. 2b). Each muscle is innervated by 2 or 3 excitatory neurones with apparently no peripheral inhibitory fibres (Pasztor, 1968; Young, 1975; Pilkington & McFarlane, 1978). Young (1975) concluded that the rhythmicity of the system was determined centrally at premotoneuronal level by sinusoidal input from an oscillator. Mendelson (1971) identified pacemaker neurones in the subeosophageal ganglion of lobsters and hermit crabs and their existence was verified by Simmers (1979) who recorded intracellularly from spiking and non-spiking and non-spiking neurones in the thoracic ganglion of Carcinus which showed oscillatory activity in phase with ventilation. Pilkington (1976) entrained scaphognathite rhythm to a slowly alternating current applied to the ganglion which presumably drove the pacemaker. Young (1975) suggested that the phase constancy of the motor units, which gives rise to the fixed sequential contractions of the ventilatory muscles, was dependent upon the different thresholds for depolarization by the pacemaker of the depressor motoneurones supplying muscles which depressed the scaphognathite and activation of the levator motoneurones by rebound during the relatively repolarized phase of the pacemaker (Mendelson, 1971). He also speculated that there may be intemeurones which modulate the cycling rate of the driver oscillator or pacemaker which could feed forward onto the levator motoneurones and would themselves be subject to feedback from the pacemaker. Young (1975) presented a model for control of the scaphognathite which proposed that its beating is determined centrally by the modulated but fundamentally intrinsic rhythmicity of a pacemaker, though the motoneuronal output is interactive. Reversal of scaphognathite beating may be achieved by a command input which by-passes the pacemaker and changes the relative threshold of the motoneurone sets causing a rephasing of the cycles of depolarization and recruiting some alternative muscles.

Simmers & Bush (1980) located two non-spiking intemeurones in the thoracic ganglion of Carcinus with membrane potentials which oscillated in phase with activity in either the depressor or levator motoneurones and may show reciprocal inhibition. This replaces the proposed existence of a single oscillator and is included in a version of Young’s model as part of Figure 3.

Pasztor (1969) identified and characterized a unique receptor, the oval organ, on the dorsal surface of the scaphognathite (Fig. 2b) which is innervated by large axons discharging in patterned bursts during scaphognathite movements (Pasztor, 1969; Young & Coyer, 1979). Motor output to the scaphognathite can be entrained to movements imposed upon it, suggesting that afferent input can exert a phasic influence on the central pacemaker (Young & Coyer, 1979). The input from the oval organ consists of both impulses and graded receptor potentials which showed a linear relationship with the amplitude of stretch applied to connective tissue strands on the organ (Pasztor & Bush, 1982). These potentials may provide the basis for a tonic influence over scaphognathite motoneurones attributed to the afferent input by Pasztor (1968), Mendelson (1971) and Wilkens & Young (1975).

During active forward ventilation of the branchial chambers there is often significant phase coupling between the ventilatory cycles of the two scaphognathites following either drift and lock or absolute phase-locked co-ordination (Wilkens & Young, 1975; Young & Coyer, 1979). Often the scaphognathites beat in synchrony but other phase relationships also occur. This may imply some advantage from coupling of the rhythms which could result in increased stability and regularity of pumping (Young & Coyer, 1979). Cross-ganglionic motoneuronal processes which could mediate this co-ordination have been described in the lobster (Wilkens & Young, 1975) and crab (Pilkington & MacFarlane, 1978). At other times the scaphognathites beat independently at different intrinsic rates (Batterton & Cameron, 1978) and in settled, inactive animals ventilation may become unilateral for long periods in both water-breathers (McDonald et al. 1977; Butler et al. 1978) and air breathers (Wood & Randall, 1981a). Reversals may occur simultaneously on both sides or independently (Taylor et al. 1973; Batterton & Cameron, 1978).

Berlind (1976) injected extracts of pericardial organs into crabs and crayfish and obtained an acceleration of scaphognathite rate which he attributed to peptides rather than amines. Injection of 5 HT into Carcinus maenas reduced f_R and increased reversal frequency (Berlind, 1977). When 5 HT was applied to the thoracic ganglion it reduced bursting frequency in motoneurones supplying the scaphognathite indicating that the effect was on the central oscillator. Increased reversal frequency, induced by addition of particulate matter to the water was blocked by a 5 HT antagonist. Glutamate and possible acetylcholine increased reversal frequency but dopamine, noradrenaline, octopamine and μ aminobutyric acid were without effect. In later experiments (Wilkens, 1981) 5 HT, dopamine and octopamine each caused a long lasting increase in f_R and f_H in settled Carcinus. Clearly, these results demonstrate hormonal control of scaphognathite function attributable to biogenic amines already implicated in control of the heart.

The control and co-ordination of bilateral ventilation and its relationships to perfusion of the gills with haemolymph delivered by the heart has been the subject of a number of studies. Hypoxic or hypercapnic water and solutions of L-glutamic acid or NaCl applied externally to one side of the crayfish Procambarus simulons produced inhibition of the homolateral scaphognathite accompanied by a simultaneous bradycardia or cardiac arrest (Larimer, 1964b) demonstrating that the scaphognathites can function separately and that changes in their activity are often reflected in responses of the heart. Introduction of sugars such as glucose into the ventilatory stream of the crayfish caused bilateral slowing of the scaphognathite and short-term cardiac arrest Ashby & Larimer, 1965). Settling after a period of disturbance is often accompanied by a progressive reduction in f_R and f_H culuminating in long periods of unilateral ventilation, respiratory pauses and an irregular heart beat (e.g. McDonald et al. 1977; Butler et al. 1978). Exposure to diluted seawater caused a cardioacceleration, typically accompanied by a slowing in scaphognathite beating in Carcinus tnaenas (Hume & Berlind, 1976) whereas hypoxia typically results in a bradycardia and hyperventilation, as described below. Ventilatory pauses and reversals are often accompanied by changes in heart rate (e.g. McMahon & Wilkens, 1972; Taylor et al. 1973). Part of this co-ordination could result from a physical link between haemolymph flow through the gills and ventilation of the branchial chambers. The effective increase in hydrostatic pressure in the branchial chambers during ventilatory pauses or reversals may flush haemolymph from the gills or alternatively temporarily halt flow when noxious stimuli appear at the gills (McMahon & Wilkens, 1977), and could result in selective perfusion of the ventilated gills during unilateral ventilation. In Cancer anthonyi rate of perfusion of an isolated gill with saline at constant pressure was maximum at a subambient external pressure of − 2 cm H₂O, which mimicked forward ventilation. Perfusion was slower at ambient pressure, equivalent to a ventilatory pause and slowest at a superambient external pressure of + 2 cm H₂O, equivalent to a ventilatory reversal (L. E. Burnett, personal communication).

The onset of a respiratory pause is often accompanied by an immediate reduction in heart rate (McMahon & Wilkens, 1972, 1977; Taylor et al. 1973). In Carcinus maenas respiratory pauses are typically accompanied by cardiac arrest, with the heart stopping instantaneously at the onset of the pause (E. W. Taylor, unpublished observation). The closely co-ordinated nature of this response suggests a neural link between the ventilatory oscillators and the cardioregulatory centres in the C.N.S. Wilkens et al. (1974) identified command fibres in the circumoesophageal connectives of Cancer magister which were active during induced changes in cardiac or scaphognathite rhythms. When these fibres were stimulated electrically 69 % produced responses in both the heart and scaphognathites, 28% in scaphognathites alone and 3 % in the heart alone. By varying the frequency of electrical stimulation they established an order of response thresholds for both the heart and scaphognathites in the sequences: bradycardia < cardiac arrest and decreased rate and amplitude of forward ventilation < reversed ventilation < ventilatory pause, which resembled observed behavioural responses to increasing levels of disturbance. Reversals often followed stimulation of one command fibre whilst pauses required high frequency stimulation of several fibres inferring the involvement of both spatial and temporal summation. The sequence of responses and their co-ordination implied an endogenous programme as suggested by McMahon & Wilkens (1977).

Presumably the command fibres stimulated by Wilkens et al. (1974) are the same as those separately identified as affecting the cardioregulatory centres (Wiersma & Novitski, 1942) and the central ventilatory oscillator (Mendelson, 1971). The majority of them are bivalent, innervating two anatomically distinct but functionally correlated systems. Very few command fibres affected the heart alone and 90% of them caused cardioinhibition which is a common response to sudden disturbance in crustaceans (McMahon & Wilkens, 1972 ; Wilkens et al. 1974), though it may indicate that f_H was near maximum due to experimental manipulation (McMahon & Wilkens, 1982).

A prolonged burst of activity in the cardioinhibitory nerve of the lobster, Nephrops norvégiens, caused initial cardiac arrest, followed by adaptation and a postinhibitor) rebound, plus persistent cessation of ventilation (Young, 1978). Continuous recordings revealed an inverse relationship between ventilation and activity in the cardioinhibitory nerves indicating that the cardioregulatory centre interacts with the ventilatory pacemaker. In Portunus bursts of activity in the cardioinhibitor nerve may be associated with ventilatory reversals rather than pauses which agree with the observation that reversals in Carcinus are often accompanied by cardiac arrest (Taylor et al. 1973). Spiking activity in cardioregulator nerves often occurs in preferred phases of the ventilatory cycle (Young, 1978) which is a possible basis for the bimodal and trimodal phase coupling between the heart and scaphognathites described by Young & Coyer (1979).

Most decapod crustaceans possess the respiratory pigment haemocyanin in solution or colloidal suspension in the haemolymph. The concentration is limited, possibly by the effect of colloid osmotic pressure on water fluxes (Mangum & Johansen, 1975) which sets an upper limit on the O₂ carrying capacity of the haemolymph. Bound O₂ varies around 1 vol % (approximately 0·5 μ mol l⁻¹) which is 2–4 times the dissolved level for a whole range of aquatic crustaceans (Fig. 4). Higher values have been reported from land crabs whose gills are not in contact with a ventilatory stream of water (e.g. Wood & Randall, 1981b). Haemocyanin is an important buffer against pH variation in the haemolymph so that similar limits are set on non-HCO₃⁻ buffering capacity, which varies in direct proportion to O₂ capacity (e.g. Wood & Randall, 1981b). Both are low relative to vertebrates which package high concentrations of circulating haemoglobin in erythrocytes.

Crustacean haemocyanins typically have a relatively high affinity for O₂ with P₅₀ varying from 6 mmHg in the crayfish Pacifastacus leniusculus (Rutledge, 1981) to 34 mmHg in the spider crab Maia squinado (W. E. Taylor, unpublished observation). The apparent paradox of an animal such as the lobster living in well aerated sublittoral waters and yet possessing a respiratory pigment with a relatively high affinity for O₂ was attributed to the diffusion barrier presented by chitin at the gas exchange surfaces. Low haemolymph values were reported in animals taken from normoxic water (Redmond, 1955). These results now seem erroneous, due possibly to removal of the animals into air during sampling procedures, and Johansen et al. (1970) reported very high values from Cancer magister, though it is probable that these crabs were hyperventilating due to recent disturbance. Routinely active crustaceans show wide variations in haemolymph levels with varying around the P_SATN level for the haemocyanin. levels are normally around or above the P₅₀ level, leaving a considerable venous reserve of oxygen available for periods when O₂ demand outstrips supply during exposure to hypoxic water, cessation of ventilation, movement into air or bouts of exercise, as described below (Fig. 4).

For accurate assessments of respiratory gas transport it is necessary to measure the important variables directly. It is not possible to infer O₂ content from measurements of O₂ tension because O₂ capacity varies in individuals with nutritive state and the moulting cycle (e.g. Zuckerkandl, 1957) and the relatively high affinity and cooperativity of haemocyanin measured in vitro reveals that content varies widely over a narrow range of tension values at the functional range of the pigment. This range is highly variable as crustacean haemocyanin is characterized by a large Bohr shift with pH variation. Affinity also changes with temperature and salinity (e.g. Truchot, 1973) and is effected by concentration of specific ions such as calcium and some metabolites (e.g. Truchot, 1981). Acid-base balance and CO₂ capacity is similarly complex. Haemocyanin shows a measurable Haldane effect causing pH and CO₂ capacity to vary with degree of oxygenation (Truchot, 1976) and the buffering of a metabolic acidosis by carbonate may generate CO₂ (e.g. Wood and Randall, 19816). Production of free amino acids during hyperosmotic exposure in Pacifastacus leniusculus led to a reduction in haemocyanin levels and consequent reduction in compensated by an increase in affinity and cooperativity which were not reproduced in vitro at measured levels of pH and ionic concentration, implicating other factors in the in vivo situation (Wheatly & McMahon, 1982). Recent studies emphasize the multifactorial and interactive influences effecting gas transport in crustacean haemolymph.

Oxygen levels are variable in many aquatic environments including those inhabited by crustaceans, such as intertidal pools which may be hyperoxic or hypoxic (Taylor & Butler, 1973; Truchot & Duhamel-Jouve, 1980). Periods of hypoxia may be accompanied by increased temperatures and varying salinities such that the animal’s oxygen demand may increase in the face of a reduced supply.

The range of responses to hypoxia shown by invertebrates was recently reviewed (Herreid, 1980). Some species maintain independent of down to low levels whilst in others is dependent upon over the whole range. The distinction between oxygen regulators and conformers is, however, not clear cut (Mangum & Van Winkle, 1973) and the degree of independence within a species can vary with environmental variables such as temperature or salinity (e.g. Weins & Armitage, 1961 ; Taylor et al. 1977b;,Taylor, 1981) and intrinsic variables such as the moulting cycle (Thompson & Pritchard, 1969). The experimental regime is also critical. When recently disturbed the lobster was an oxyconformer (Thomas, 1954) whilst after 48 h of settling in a respirometer was maintained at its normoxic level in moderate levels of hypoxia (Butler et al. 1978) and Spoek (1974) found that restraining lobsters caused them to become oxyconformers. For many species the degree of independence of reduced relates to the level of (Herreid, 1980).

The physiological mechanisms which compensate for reduced during regulation serve to increase the conductance of the gas exchange organs, and circulatory system. They may be typified by describing the responses of the crayfish Austropotamobius pallipes to moderate hypoxia recently studied by Wheatly & Taylor (1981). At 15 °C was maintained independent of down to 40 mmHg (Fig. 5). Progressive hypoxia was accompanied by hyperventilation, with increases in both f_R and SV_R. The convection requirement for water increased markedly. Despite the increased rate of delivery of water to the gills both % Extn_W and effectiveness of removal of O₂ from the water (E_w) rose, due to an improvement in the ability of the respiratory surfaces to transfer O₂ assessed as and termed the transfer factor which more than doubled during moderate hypoxia. This implies recruitment of a greater area for gas exchange, and/or a change in the flow of water or haemolymph at the exchange surface, which served to increase the effective diffusive conductance or eliminate shunts. A similar improvement in during hypoxia was suggested by the results of earlier studies (e.g. Thomas, 1954; Larimer, 1964; Arudpragasam & Naylor, 1964b; McMahon et al. 1974; Taylor et al. 1977b; Batterton & Cameron, 1978) and the phenomenon merits further investigation.

As well as serving to increase the rate of delivery of O₂ to the gills, hyperventilation in the crayfish also reduced haemolymph CO₂ levels, resulting in a respiratory alkalosis (Wheatly & Taylor, 1981). A rise in pH during hypoxia was noted previously (e.g. Truchot, 1975 a; McMahon et al. 1978; Burnett, 1979). This in turn caused an increase in O₂ affinity of the circulating haemocyanin, due to a Bohr shift (Fig. 4a). Despite reductions in pre-and postbranchial O₂ tensions by 50% the arteriovenous O₂ content difference was maintained (Fig. 5), the effectiveness of removal of O₂ into the haemolymph (E_b) remained high and the venous reserve of O₂ was retained. The ‘blow-off’ of CO₂ and resultant alkalosis was a vital component of this response and may be the underlying significance of the hypoxic hyperventilation which is generally considered merely to increase delivery of O₂ to the exchange surfaces, its utility in this regard being limited by the high respiratory cost of ventilation (Jones, 1971). In the settled lobster during long-term hypoxia an initial period of hyperventilation eased after 24 h (Butler et al. 1978) presumably to reduce the respiratory cost of ventilation (McMahon et al. 1974), but the alkalosis in the haemolymph was maintained by elevation of bicarbonate levels and thus oxygen affinity remained above the normoxic level (McMahon et al. 1978).

Hypoxic hyperventilation was accompanied in the crayfish by a bradycardia, though haemolymph flow was maintained by an increase in SV_H (Wheatly & Taylor, 1981). Nevertheless the ventilation perfusion ratio increased. Bradycardia is a common feature of the hypoxic response in aquatic crustaceans (e.g. Larimer, 1962; McMahon & Wilkens, 1975; Taylor et al. 1977b) and in air-breathers (Herreid et al. 1979). In settled lobsters f_H was relatively low and did not change on exposure to moderate hypoxia (Butler et al. 1978). It is possible that the bulk of observations of a marked bradycardia during hypoxia were taken from disturbed animals with elevated f_H (McMahon & Wilkens, 1982).

At levels below 40 mmHg the crayfish became an oxyconformer (Wheatly & Taylor, 1981). Hyperventilation ceased, arteriovenous O₂ content difference decreases the venous reserve of O₂ was depleted and as the haemocyanin was no longer saturated with O₂ during its passage through the gills E_b was reduced. As a consequence fell and was substituted by a switch to anaerobic metabolism with lactic acid accumulating in the haemolymph (Fig. 5). The lactate levels measured in the haemolymph of Carcinus maenas during hypoxia appeared sufficient to replace aerobic energy production and measured activity levels were unchanged (Taylor et al. 1977b). There is a trend towards oxyconformity as temperature increases in many crustaceans and this may be accompanied by enzymic changes favouring lactic acid production in the tissues (Taylor, 1981).

On return to normoxia after hypoxic exposure is often elevated in crustaceans, possibly reflecting the replenishment of the venous reserve of O₂, regeneration of phosphagens and eventual repayment of an O₂ debt as lactic acid is reoxidized, though changes in recovery may not parallel reduction in lactate levels (Bridges & Brand, 1980). Some crustaceans appear to repay an accumulated O₂ debt in full (Teal & Carey, 1967; Taylor et al. 1977b), others repay only part of the debt (Bridges & Brand, 1980) and some, despite accumulating lactic acid during hypoxia, appear not to repay an O₂ debt (Spoek, 1974). Lobster claw muscle accumulated both alanine and lactate during anaerobiosis (Trausch, 1976) indicating the availability of alternative metabolic pathways, which was suggested by the results of Butler et al. (1978). These may provide a greater net yield of ATP (Hochachka & Somero, 1973) which could be available during exercise as lactate constituted only 23% of the metabolic acids accumulated by exercising Gecarcinus (Smatresk et al. 1979).

Instead of reducing activity levels or accumulating an O₂ debt during hypoxia some primarily aquatic crustaceans may become facultative air-breathers thus exploiting a source of O₂ which requires little energy for ventilation. The shore crab Carcinus maenas raised its normally exhalent apertures from shallow hypoxic water and ventilated the water surrounding its gills by prolonged reversed beating of the scaphognathites causing bubbles to stream from the Milne-Edwards openings (Taylor & Butler, 1973). Although the crab remained internally hypoxic this behaviour was accompanied by an immediate recovery tachycardia and increased (Fig. 4b) which restored O₂ supply to the tissues (Taylor et al. 1973). Very little CO₂ accumulated as contact with water was retained, and haemolymph pH stayed at normoxic submerged levels during partial emersion from hypoxic water at 15°C (Wheatly & Taylor, 1979). On exposure to relatively high temperatures Carcinus moved into air (Taylor & Wheatly, 1979) and the freshwater crayfish Austropotamobius pallipes similarly left water to breathe air during progressive hypoxia (Taylor & Wheatly, 1980).

The chitin covering the gas exchange surfaces in crustaceans could provide mechanical support and resistance to dehydration in air. Despite this apparent preadaptation aquatic crustaceans become hypoxic and hypercapnic and accumulate lactic acid when taken into air. In the sublittoral spider crab Maia squinado an accumulating metabolic and respiratory acidosis remained uncompensated and the animal died within 24 b, apparently from asphyxia (E. W. Taylor & P. J. Butler, unpublished observations). The lobster Homarus vulgaris from the sublittoral fringe compensated for the potential acidosis by mobilizing an internal source of HCO₃⁻ with a concomitant increase in haemolymph calcium levels (E. W. Taylor, unpublished observations). Facultative air-breathers such as Carcinus metabolize aerobically and compensate for a respiratory acidosis by elevation of buffer base (Truchot, 1975b) when in air. The crayfish Austropotamobius experienced an initial combined metabolic and respiratory acidosis when first taken into air, which recovered over 24 h due to a reduction in lactate level, which appeared to be sequestered in the tissues as it reappeared on submersion, and elevation of HCO₃⁻ from an internal source of fixed base which may be CaCO₃ as calcium accumulated in the haemolymph during aerial exposure (Taylor & Wheatly, 1982). The recovery in pH served to restore O₂ transport by reversal of a Bohr shift as shown in Fig. 4c (Taylor & Wheatly, 1981). Both the shore crab and crayfish maintain and f_H at submerged normoxic levels whilst in air. Ventilation rate in air was 5 % of normoxic in Austropotamobius, reflecting the increased O₂ content and diffusion rate (Dejours, 1975) and minimizing evaporative water loss which appears to limit their survival in air to about 72 h at 15 °C, due possibly to reduction in circulating haemolymph volume (Taylor & Wheatly, 1982). Land crabs may maintain low resting levels of for similar reasons (Wood & Randall, 1981a). Obligate air-breathers have reduced gill areas and exchange respiratory gases over the well vascularized and thin-walled internal surfaces of the branchio-stegites (Greenaway & Taylor, 1976). Birgus latro survives after removal of its gills though this delays recovery after exercise indicating that they retain a role in acid-base balance (Smatresk & Cameron, 1981). Injection of radioactive microspheres into the haemolymph of Holthusiana indicated that flow to the wall of the branchiostegite may be enhanced after 24 h in air (H. H. Taylor & P. Greenaway, personal communication). The scaphognathite retains its role in ventilation in air in some facultative (e.g. Taylor & Butler, 1978) and obligate (e.g. Cameron & Mecklenberg, 1973) airbreathers. In Holthusiana ventilation is accomplished by alternate expansion and retraction of the epimeral wall into each branchial chamber (Greenaway & Taylor, 1976) and epimeral retractor muscles may cause some air movement in Coenobita (McMahon & Burggren, 1979) and Cardisoma (Wood & Randall, 1981a).

Fully adapted air-breathing crustaceans may have relatively high haemolymph values (e.g. Cameron & Mecklenberg, 1973; Wood & Randall, 1981a) though in others it is routinely low (e.g. McMahon & Burggren, 1979). Haemolymph CO₂ levels are high in many air-breathers and in the absence of compensation this may result in a low pH. In Birgus latro haemolymph pH was 7·54 and [HCO₃⁻] 5·7 m-equiv. I⁻¹, whilst Coenobita has a pH of 7-84 and [HCO₃⁻] of n-o m-equiv. I⁻¹. This specific difference resembles the progressive change in pH and [HCO₃⁻] observed in Austropotamobius during 24 b exposure in air (Taylor & Wheatly, 1981) and confirms the relationship between pH and [HCO₃⁻] in crustaceans, which is also apparent in aquatic species over a range of temperatures (Truchot, 1978). The non-bicarbonate buffering capacity which varies with haemocyanin concentration is also characteristically high in terrestrial crustaceans (Wood & Randall, 1981b). Coenobita may also minimize CO₂ accumulation by retaining a reservoir of water in its adopted shell (McMahon & Burggren, 1979). In the brachyuran land crab Cardisoma camifex the branchial chamber is normally partly filled with water which is renewed every fea minutes when the crab has access to pools of water. This water is flicked over the giM by the flabellum (mastigobranch of the first maxilliped) and C0₂ equilibrates between the gas and water phases so that it provides a significant sink for CO₂ (Wood & Randall, 1981a). Excretion of CO₂ is facilitated by a catalysed HCO₃⁻ dehydration reaction in the gill epithelium which contains carbonic anhydrase (Randall & Wood, 1981).

There is circumstantial evidence for the involvement of O₂ sensitivity at a number of morphological sites in the physiological responses of crustaceans to hypoxia and in particular the initiation of the hypoxic bradycardia. The effect of hypoxia (Stiffler & Pritchard, 1972) or metabolic inhibitors (Livengood & Kusano, 1973) suggested a direct sensitivity of the myocardium or cardiac ganglion to reduced O₂. The latency of cardiac and ventilatory responses to changes in branchial chamber (McMahon & Wilkens, 1972, 1975) and the ventilatory responses to an induced cardiac arrest (Larimer, 1964a) suggested a role for central receptors in the heart and C.N.S. and it seems likely that the cardiac and central ventilatory pacemaker neurones may respond to variations in ⁠.

Evidence for central receptors does not preclude the existence of peripheral receptors monitoring levels in the haemocoel or ventilatory stream. Larimer (1964b) ascribed initiation of the cardiac and ventilatory responses of the crayfish to hypoxic and hypercapnic water to chemoreceptors located somewhere within the branchial chambers. This suggestion is supported by the instantaneous increase in f_H at the onset of air-bubbling, during the emersion response in Carcinus maenas (Taylor & Butler, 1973). Both this recovery tachycardia following emersion and the similarly high f_H level recorded from Carcinus in air (Taylor & Butler, 1978) are associated with internal hypoxia, with values below those normally associated with a bradycardia in submerged, hypoxic animals, indicating that the presence of air in the branchial chambers provides a normoxic stimulus to an externally placed 0₂ receptor. Direct recordings of 0₂ receptor discharge were described from nerves supplying the gills of the chelicerate arthropod Limulus polyphemus (Crabtree & Page, 1974) and similar unpublished information from a crustacean was cited by McMahon & Wilkens (1982). The initial bradycardia and hyperventilation during progressive hypoxia are probably nervously mediated and arise from stimulation of as yet uncharacterized reflexogenic areas. When levels fall below the P_SATN for haemocyanin both f_R and f_H may fall (McMahon & Wilkens, 1975) and is reduced (Wheatly & Taylor, 1981), possibly due to a direct effect of 0₂ lack on the functioning of the two pumps, rather than effective co-ordination of their outputs, as this response may be accompanied by a reduction in and a switch to anaerobiosis.

The ability to breathe air, which offers the advantage of a continuously high level of but the problem of CO₂ elimination, might be expected to result in development of C0₂ sensitivity. This is foreshadowed in aquatic species. Normoxic crayfish show a ventilatory (Massabuau & Dejours, 1981) and associated cardiac (Larimer, 1964b) response to increases in dissolved C0₂. Land crabs respond to hypercapnia with an increase in ventilation. Birgus latro showed a reduction in at low levels and a linear increase as CO₂ was increased (Cameron & Mecklenberg, 1973) which the authors considered signified that they occupied an intermediate position in an evoluationary sequence from water to land. Ventilation increased greatly in response to nypercapnia in Gecarcinus lateralis and Cardisoma guanhami signifying that they were adapted to function as air-breathers (Cameron, 1975). Injection of buffered CO_a rich saline into Birgus was without effect but injection of an acid saline solution caused an immediate increase in f_R and recovery in post exercise f_R followed recovery in haemolymph pH values rather than (Smatresk & Cameron, 1981). This suggests sensitivity to internal pH i.e. [H⁺] or [HCO₃⁻] which are components of strong ion difference (Stewart, 1978) rather than and a parallel may exist in the responses of the vertebrate medulla to the pH of the cerebrospinal fluid (Mitchell & Severinghaus 1965).

During exercise animals typically show a marked increase in over settled levels which may be taken as their aerobic metabolic scope for activity. Shortfall in aerobic scope may be taken up by anaerobic metabolism. Homeothermic air-breathing vertebrates characteristically have a high aerobic scope, adjust rapidly to exercise, including lactate production, and recover rapidly whilst poikilothermic, water-breathing vertebrates typically have a lower aerobic scope and recover less rapidly (Bennett, 1978). Crustaceans may be limited in their aerobic scope by the potential problems of diffusive conductance of O₂ over the chitin on the gills and convective conductance via the open haemocoel. Oxygen uptake and acid base balance during activity in decapod crustaceans were recently reviewed by McMahon (1981).

Changes in O₂ demand and transfer, anaerobic metabolism and acid-base balance following 20 min of exhausting activity enforced by violent disturbance of the relatively sluggish aquatic crab Cancer magister were described by McMahon et al. (1979) and McDonald et al. (1979). The overall responses resembled those described for the land crab Cardisoma carnifex running sideways on a treadmill at close to maximum burst velocity for short periods. Similar short sprints were undertaken by crabs seeking the shelter of their burrows (Wood & Randall, 1981a).

In Cardisoma 10 min of exhausting activity at 25 °C caused a 3 times increase in This degree of aerobic scope resembled that of a relatively sluggish fish ; whereas, in the crayfish, Pacifastacus leniusculus, maximum aerobic scope, attained at 20 °C, was 10 times the resting rate, an increase which approaches that of a moderately active fish such as the bass, Micropterus salmoides (Rutledge & Pritchard, 1981).

The increased was partially attributable to increased ventilation with 5 times resting level and a similar increase in f_R. % Extn_A was low in resting crabs at 12% and further decreased to 3 % during exercise. A smaller proportional increase in f_H of approximately 20% occurred and Herreid et al. (1979) reported a reduction in f_H and cardiac arrhythmia during ‘heavy exercise’ in Cardisoma guanhami. This could imply that during exercise is limited by ventilation and diffusive conductance over the chitin covered exchange surfaces, rather than by perfusion, though estimated by the Fick principle increased during exercise, independently of f_H. This indicates an increased SV_H or possibly improved perfusion of the gills due to movement of the legs improving venous return via the gills from the infrabranchial haemo-coelic spaces.

The maintenance of gas exchange during exercise was partially explained by an increase in over the gills due to reduced levels in the haemolymph. was reduced by the exercising tissues and fell, due possibly to reduced transit time through the gas exchange organs (Fig. 6). This approximately 30 % improvement in was insufficient to explain the increased and diffusive conductance expressed as ⁠, doubled during exercise. The reduced levels had little effect on with saturation of the postbranchial haemolymph (Sa, O₂) decreasing from 87% to 71 %. The reduced levels were, however, indicative of a marked reduction in with reduced from 45 % to 9 % as tissue O₂ consumption shifted the range in the haemolymph onto the functional portion of the haemocyanin O₂ equilibrium curve, thereby utilizing up to 90% of the available venous reserve of O₂. Up to 70% of this reduction in was due to the fall in with the other 30–40% due to a marked Bohr shift induced by an acidosis (Fig. 4d).

The sharp increase in during exercise infers an increase in which rose to 5 times the settled level in Cardisoma (Wood & Randall, 1981b). Despite this increase, values were little affected and remained around 15 mmHg (Fig. 6) which is approximately half that recorded in poikilothermic vertebrates such as turtles which breathe in the intermittent manner characteristic of land crabs (Burggren & Shelton, 1979). Elimination of CO₂ from this air-breathing crab is enhanced by carbonic anhydrase in the gill epithelium (Randall & Wood, 1981) and the presence of a reservoir of water held in the branchial chambers (Wood & Randall, 1981a). The main contributor to acid-base variation during exercise was the accumulation of lactic acid in anaerobically metabolizing tissues. Anaerobic energy production due to lactate fermentation accounted for 80 % of the energy utilized by Cardisoma, and lactic acid accumulated to 7 times the resting level in the haemolymph. Following severe exercise Cardisoma showed a combined respiratory and metabolic acidosis which recovered within 2 h to be followed by an alkalosis. Lactate levels in the haemolymph were still elevated at 6 h and it is probable that the recovery and overcompensation in pH was attributable to the reduced and elevation of bicarbonate buffer by mobilization of calcium carbonate from the exoskeleton or gastrolith fields in the gut which are supplied with arteries and may be involved in calcium regulation. Calcium levels in the haemolymph increased during recovery to a peak 1 h after exercise which was when the apparent discrepancy between lactate levels and relative acidity was greatest (Fig. 6). An internal source of fixed buffer is particularly important in land crabs which have lost external seawater as a source of HCO₃⁻. During periods of inactivity the CaCO₃ may be replenished by retention of respiratory CO₂ which could explain the abnormally low value for gas exchange ratio R measured from Cardisoma (Wood & Randall, 1981b).

The responses of Cardisoma during short sprints may be contrasted with the swimming crab Callinectes sapidus which is active for long periods without fatigue (Booth et al. 1982). Callinectes has a relatively large weight specific gill area and short minimum diffusion distance over the gills of approximately 1 μm (Aldridge & Cameron, 1979) which may account for its ability to maintain and % Extn_W (40–50%) at high levels during prolonged exercise. At the onset of induced swimming it showed rapid increases in f_R, f_H and . Lactic acid accumulated to 14 times the resting level, inducing a large metabolic acidosis. and levels were unchanged but decreased significantly due to a marked Bohr shift which was opposed by a specific effect of lactate ions increasing the O₂ affinity of haemocyanin (Truchot, 1981). The amount of haemocyanin bound O₂ released to the tissues rose but a large proportion of the venous reserve remained intact and the increase in O₂ delivery to the tissues was attributed to a 2·3 times increase in cardiac output (Booth et al. 1982).

There was a virtually instantaneous increase in f_R at the onset of swimming in Callinectes and running in Cardisoma which suggests that neural control, possibly involving proprioceptive feedback from the locomotory appendages, may initiate the respiratory responses which enhance O₂ uptake during exercise.

The functional morphology and physiology of the respiratory and cardiovascular systems in decapod crustaceans differ widely from the relatively well documented vertebrate systems. Nevertheless, there are qualitative similarities in their adaptive responses to problems of O₂ supply or demand. This is a flourishing branch of comparative physiology at present and some aspects which merit further attention are : functioning and control of the circulatory system, including the role of the alary muscles; details of gill ventilation, including the role of the epimeral muscles; localization and characterization of the receptors initiating the responses to hypoxia and exercise; a critical reappraisal of the supposed diffusion barrier presented by chitin; the factors contributing to the improvement in during hypoxia and exercise; the role of active transport over the gills in HCO₃⁻ and NH₄⁺ elimination; the combined roles of haemocyanin in O₂ transport, acid-base balance and osmoregulation, including the associated roles of free amino acids and their contribution to oxidative phosphorylation ; the biochemistry of anaerobic metabolism ; the partitioning of lactate and H+ between the tissues and haemolymph and control of CaCO₃ mobilization during exercise and aerial exposure; changes in any of these variables during moulting.

I am grateful to the following authors for providing me with manuscripts or information prior to publication: Charles Booth, Brian Bush, Brian McMahon, Valerie Pasztor, Dave Randall and Michèle Wheatly. I am also indebted to Jean Hill for typing and Deborah Barrett for proof reading the manuscript.