Control of Resting Ventilation Rate in Grasshoppers

The abdominal pumping or ventilatory rate of grasshoppers and other insects is known to increase with increasing ambient and decreasing ambient (Miller, 1960a; Arieli and Lehrer, 1988). However, it is not known whether sensitivity to internal and is involved in the normal regulation of resting ventilation rate or whether insects regulate their tracheal or ⁠. In this study, we tested the hypothesis that grasshoppers regulate their tracheal and levels by varying ventilation rate. It has also been suggested that the ventilatory system participates directly in short-term extracellular acid–base regulation in grasshoppers as it does in terrestrial vertebrates (Harrison, 1989a). To test this hypothesis, we examined the effect of extracellular acid–base status on ventilation rate.

In grasshoppers, abdominal pumping is accomplished by contraction of inspiratory and expiratory intersegmental abdominal muscles, which drive convective ventilation (Lewis et al. 1973). The coordination of abdominal expansion with opening of the thoracic spiracles, and abdominal compression with opening of the abdominal spiracles, produces a unidirectional thorax-to-abdomen airflow (McCutcheon, 1940; Weis-Fogh, 1967). Abdominal pumping constitutes the most important component of convective ventilation in resting locusts (Miller, 1960a; Weis-Fogh, 1967).

The respiratory rhythm of locusts is driven by pacemakers in the metathoracic and abdominal ganglia (Miller, 1960a; Lewis et al. 1973), with afferent feedback from abdominal stretch receptors (Hughes, 1952; Farley and Case, 1968). The effects of perfusion of local regions of dissected locusts with CO₂ suggest that CO₂ affects the head and thoracic ganglia directly and that locusts lack peripheral afferent nerves sensitive to CO₂ or pH (Miller, 1960a). The enhancement of ventilation by hypoxia or hypercapnia may indicate that ventilation rate is controlled by internal or under normal physiological conditions in insects. Alternatively, abdominal pumping rate may be controlled by an endogenous rhythm insensitive to the changes in internal or that occur in vivo. Supporting this possibility, internal (tracheal or haemolymph) values of resting grasshoppers have been reported to be 2–5 kPa (Krogh, 1913; Harrison, 1988, 1989a), and ventilation rate was not significantly stimulated by ambient values of 3 kPa (Hustert, 1975; Harrison, 1989b). Thus, it is possible that the responsiveness of the ventilatory system to ambient CO₂ and O₂ levels may have evolved in order to help control ventilation in insects living in burrows or other microenvironments characterized by hypoxia or hypercapnia. We tested for an effect of tracheal or on resting ventilation rate by independently manipulating tracheal and values above and below normal tracheal levels using perfusions of the tracheal system via the metathoracic spiracle. There is a great deal of correlational data suggesting that the ventilatory system may participate in extracellular pH regulation in insects as it does in vertebrates. In grasshoppers, changes in ventilation rate during hypercapnia or after locomotion are correlated with changes in both haemolymph and pH (Harrison, 1989a; Harrison et al. 1991). Case (1961) demonstrated that the fictive ventilatory rhythm of cockroach ganglia in vitro is sensitive to bathing solution pH. In dissected Nauphoeta cinerea cockroaches, changes in the pH or of fluids irrigating the ventral nerve cord elicited changes in ventilation rate (Snyder et al. 1980). However, grasshoppers injected with HCl recover from acidosis without changes in haemolymph ⁠, suggesting that decreased haemolymph pH does not stimulate ventilation in grasshoppers (Harrison et al. 1992). We tested directly for an effect of extracellular acid–base status on ventilation rate using injections of various acid–base solutions into the haemocoel.

For the experiments examining the effects of haemolymph pH on ventilation rate, we used eastern lubber grasshoppers, Romalea guttata Serville, collected from field sites in central Florida, USA. Grasshoppers were maintained in the laboratory as previously described (Harrison and Kennedy, 1994). This large species (mean body mass 4.2 g in males and 8.4 g in females) is particularly useful for such studies, owing to its large haemolymph volume and its passive nature. Since it relies on the secretion of phenol compounds and quinones from the exocrine defensive glands rather than escape tactics to deter predators, it remains calm and relatively immobile under observation (Whitman, 1990). Romalea guttata do not have spiracles suitable for cannulating as their spiracles bleed when cannulation is attempted. Therefore, for the experiments involving the manipulation of tracheal gases, we used the desert locust Schistocerca americana Drury (body mass range 1.4–2.3 g). All experiments were carried out at room temperature (22–25 °C).

In these experiments, haemolymph acid–base status was manipulated with injections of acid–base solutions into the haemocoel, and we tested for an effect on ventilation rate. Male and female R. guttata were weighed, and the haemocoel was cannulated using a 20 cm length of polyethylene tubing (PE 20) filled with 0.5 mol l^-1 NaCl. Using a Dremel drill (bit diameter 0.4 mm), we drilled a hole into the animal’s left side anterior to the metathoracic spiracle. The cannula was inserted into the hole and secured with glue from a hot glue gun. Animals were then placed in clear plastic containers with access to water but not food. Each container was wrapped in Mylar, and a fibre-optic light was shone into it from the top, allowing the Mylar to function as a one-way mirror to avoid disturbance of the grasshoppers. Small wire ladders were glued onto the side of each container for the grasshoppers to climb and be observed in a more natural position. The cannula extended from the animal to the outside of the container, where it was connected to a 50 μl Hamilton gas-tight syringe. The next day (14–20 h post-cannulation), the ventilation rate of each grasshopper was counted visually for 6 min. Each abdominal pumping event, regardless of volume, was counted as a ventilation event. Each animal was then injected through the cannula with one of four 0.5 mol l^-1 solutions: NaCl, NaHCO₃, HCl or NaOH. Females were injected with 50 μl of each solution (to give, on average, 7.2 μl g^-1 body mass), and males were injected with 30 μl (to give, on average, 5.9 μl g^-1 body mass). Injection volume was corrected for the dead space of the cannulae to ensure that the correct volume was injected. The effects of acid–base injections into the haemocoel of grasshoppers persist for hours but are most pronounced during the first 10–15 min (Harrison et al. 1992). Therefore, we counted ventilation rate for 10 min after each injection, and compared the pre- and post-injection ventilation rates for each animal.

The effect of the injections on haemolymph acid–base status was tested by collecting haemolymph samples from an incision in the ventral cervix (throat) using a glass microcapillary pipette and 10 μl gas-tight Hamilton syringes (Harrison, 1988). Haemolymph samples were collected 6 min after the injections (midway through the period during which ventilation rate was analyzed). Haemolymph sampled from the ventral cervix area should be representative of that bathing the ventral nerve cord. The pH and (total [CO₂], mmol l^-1 ) of the haemolymph were analyzed at the body temperature of the animal using a glass capillary pH electrode and gas chromatography as previously described (Harrison and Kennedy, 1994). Haemolymph and [HCO₃^- ] were calculated using CO₂ solubility coefficients and carbonic acid dissociation constants for locust haemolymph (Harrison, 1988). The NaCl injection was used to control for stress associated with the injection, since a similar injection has been shown to have no effect on haemolymph acid–base status in Schistocerca gregaria (Harrison et al. 1991).

In these experiments, tracheal and were manipulated with gas perfusions via a cannula placed into the metathoracic spiracle. Schistocerca americana were restrained by rubber bands across the thorax against a metal screen and had cotton packed around their heads to limit their vision. The restraint system did not confine the abdomen or physically impede abdominal pumping. After a few minutes, the animals stopped struggling and appeared to ventilate normally. We allowed the animals 20–40 min to acclimate to the restraint because this reduced the likelihood that animals would struggle during the cannulation procedure. A heat-stretched piece of polyethylene tubing (approximate tip diameter 0.2 mm) was slid into the spiracle, where it was held in place by the external spiracular valve while it was sealed to the animal using a low-melting-point glue gun. In all cases, we waited 20–40 min after installation of the cannula before performing perfusions to allow animals to recover from the handling stress imposed by the cannulation procedure and because preliminary experiments suggested that ventilatory rates return to resting values during this period. Test gases could be perfused into the trachea via the tracheal cannula, and the effect on ventilation rate measured. We perfused the tracheal system for short periods (3–5 s) with a large volume (3–5 ml) relative to normal tracheal volume (less than 1 ml; Harrison, 1989b) and minute ventilation rate (about 1 ml min^-1 ; Weis-Fogh, 1967) in an attempt to flush the tracheal system with the test gas. The perfusion did inflate the abdomen, but tracheal pressures returned to normal levels within a few seconds after cessation of the perfusion (Krolikowski and Harrison, 1996). We then measured ventilation rate and tracheal gas levels during the 1 min after cessation of the perfusion (since ventilation rates and tracheal gas levels returned to normal values in approximately 1 min, see Results).

Tracheal gas samples were taken via the methoracic spiracular cannulae, using a 50 μl Hamilton syringe at a rate of 6–10μl s^-1. In a preliminary experiment, we tested the effect of the size of the tracheal gas samples taken on the CO₂ and O₂ levels measured. Tracheal samples of 10–40μl had similar levels of CO₂ and O₂, while larger samples had lower CO₂ levels and higher O₂ levels. These data suggest that ambient air may enter the animal through other spiracles when samples larger than 40 μl are taken, hence diluting the sample. Therefore, we used 30 μl samples of tracheal air for all experiments.

The cannulae used had an average dead space of 1.95±0.48 μl (S.E.M., N=8), which probably contained a mixture of ambient and tracheal air. The maximum error in tracheal gas measurement due to dilution with dead space air (if the dead space was completely filled with room air) is 6 %.

Each animal received all of either the O₂-manipulating or CO₂-manipulating perfusions (see below), allowing 5 min between perfusions. We injected one of the gas mixtures and then drew out 30 μl of tracheal air 15, 30, 45 and 60 s after the tracheal perfusion. CO₂, N₂ and O₂ fractions of the air samples were analyzed with a Varian 3400 gas chromatograph and gas-chrom MP-1 column (Varian Analytical Instruments, Walnut Creek, California, USA). Voltage output peak heights from the thermocouple detector were digitized and recorded (Sable Systems, Salt Lake City, Utah, USA). Peak heights were converted to gas composition by calibrating the system with injections of gases of known composition (mixed on a Brooks 5878 mass-flow controller and meters). The coefficient of variation of CO₂ and O₂ values measured on 30 μl samples of calibration gases was approximately 5 %.

Ventilation rate was counted for 1 min before and 1 min after perfusion of the trachea with the test gas, since preliminary experiments indicated that ventilation rates returned to normal values within 1 min after the perfusion (see Results). In the first series of experiments, tracheal was varied (18, 10, 4.4, 2.3 or 0.4 kPa) at constant (15 kPa, balance N₂). In the second series of experiments, we varied tracheal (0.5, 4.7, 14, 30 or 41 kPa) at constant (2.2 kPa, balance N₂). Within a series, each locust received all injections. Gases were mixed using a Brooks 5878 mass-flow controller and Brooks mass-flow meters, with the gas composition of each mix confirmed using the Varian gas chromatograph. Injections were given in random order. All protocols were the same in both series except that, for the CO₂ series, we injected 5 ml of test gas; in the later O₂ series we reduced the volume of test gas perfused to 3 ml, which reduced the number of animals experiencing burst abdomens during perfusions from about 10 % to zero.

The restraint and cannulation procedures used in experiment 2 might affect ventilation rate as a result of stress or by changing tracheal gas levels because of spiracle occlusion. Grasshoppers inhale via six thoracic spiracles, three on each side. The cannula occluded the left metathoracic spiracle, potentially interfering with gas exchange. Therefore, we tested for effects of restraint and cannulation on resting ventilation rate. Male S. americana were placed into separate clear containers inside an opaque box and starved overnight. The front of the box was covered with Mylar, and the grasshoppers were back-lit with a fluorescent bulb to allow the Mylar to function as a one-way mirror as previously described. The next day, the light was turned on 1 h prior to observations. We counted resting ventilation rate visually for 4 min for these completely undisturbed grasshoppers. The animals were then removed from the cups and restrained against the metal screens as described previously. They were left undisturbed for 20–40 min, and a restrained, uncannulated ventilation rate was then counted over 4 min. We then cannulated the animals via the metathoracic spiracle as described above, left them undisturbed for 20–40 min, and then counted ventilation rate over a period of 4 min. Finally, the cannula was removed, animals were left undisturbed for 20–40 min, and ventilation rate was counted again. Based on the results from these experiments (see below), a 20–40 min recovery period was allowed between cannulation and measurement of ventilation rates throughout experiment 2.

Male S. americana were collected and cannulated via the metathoracic spiracle as described previously. After 20–40 min, we sampled 30 μl of air via the metathoracic cannula and analyzed its O₂ and CO₂ fraction using gas chromatography. We then injected 30 μl of either 0.5 mol l^-1 NaCl or NaHCO₃ into the haemocoel. We sampled tracheal gases via the cannula every 2 min over the 10 min following the injection.

Means ± S.E.M. are presented. Data were analyzed using paired t-tests (comparing pre-manipulation with post-manipulation values) when individuals received two treatments, or by repeated-measures Huynh–Feldt-corrected analysis of variance (ANOVA) when individuals received more than two treatments. When different individuals received different treatments, statistical analysis was performed using univariate ANOVA or t-tests as appropriate. All statistical analyses were performed using SYSTAT (Wilkinson, 1989), with our within-experiment type I error controlled at 5 %.

The injections had very similar effects on haemolymph acid–base status and ventilation rate in males and females, so data from the two sexes were pooled. The injections significantly affected haemolymph pH (ANOVA, F_3,45=6.83, P<0.01) and haemolymph [HCO₃^- ] (ANOVA, F_3,42=4.15, P<0.05), but not haemolymph (ANOVA, F_3,42=0.19, P>0.6). We tested whether each injection produced a significant effect on haemolymph pH, [HCO₃^- ] or relative to the NaCl-injected control group using a two-tailed Dunnett’s test. Relative to the NaCl-injected group, injection of NaHCO₃ significantly raised haemolymph pH and [HCO₃^- ], while injection of HCl significantly reduced haemolymph pH and [HCO₃^- ] (Fig. 1). Injection of NaOH had no significant effect on haemolymph acid–base status because of the small sample size (N=2, Fig. 1).

Injections of HCl, NaOH and NaCl had no effect on the ventilation rate averaged over the 10 min after the injection (Fig. 2). Injection of NaHCO₃, however, caused a significant 60 % increase in the ventilation rate (Fig. 2). Although the pre-injection ventilation rate for the NaHCO₃-injected group appeared low relative to the other treatment groups (Fig. 2), this did not explain the significant effect of the NaHCO₃ injection on ventilation rate. The resting ventilation rates of the four treatment groups did not differ significantly (ANOVA, F_3,58=0.97, P=0.41). The increase in ventilation rate after the NaHCO₃ injection occurred in 15 out of 17 animals tested and was significant at each minute during the 10 min after the injection (Fig. 3, post-hoc comparisons, repeated-measures ANOVA).

Normal resting tracheal for restrained, cannulated S. americana was 18.8±0.62 kPa (N=15). When gases of varying were perfused, in all cases there were transient changes in tracheal ⁠, followed by recovery of tracheal towards normal levels (repeated-measures ANOVA, P<0.01). Data for two of the perfusion mixtures are shown in Fig. 4. During the first 15 s after perfusion, tracheal increased with increasing perfusate (repeated-measures ANOVA, F_4,36=52.1, P<0.001, Fig. 5), although tracheal did not match the of the perfusate. Perfusing the trachea with gases of varying O₂ content (⁠ constant at 2.2 kPa) generally had little effect on tracheal (Table 1). However, in three cases, there were small but significant increases in tracheal (Table 1).

Normal tracheal for restrained, cannulated S. americana was 1.5±0.37 kPa (N=15). When gases of varying were perfused through the trachea, we measured transient changes in tracheal ⁠, followed by a return towards normal levels (repeated-measures ANOVA, P<0.01). Data for two of the gases are shown in Fig. 6. During the first 15 s after perfusion, tracheal increased with perfusate (repeated-measures ANOVA, F_4,36=43.6, P<0.001, Fig. 7), although the change in tracheal was much smaller than the change in of the perfusing gas. Perfusing the trachea with gases of varying CO₂ content (⁠constant at 15 kPa) did not affect tracheal (Table 2).

The perfusions caused transient changes in ventilation rates. We show data for the changes in ventilation rate with time after perfusion for four of the gas mixtures (Figs 8, 9); similar patterns were observed for the other gas mixtures. Ventilation rate during the 15 s after perfusion was significantly affected by the of the perfusate (repeated-measures ANOVA, F_4,28=10.3, P<0.001), and mean ventilation rate was negatively correlated with the mean tracheal (Spearman’s rank correlation coefficient = 0.9, P<0.05, Fig. 10). Ventilation rate during the 15 s after perfusion was also significantly affected by the of the perfusate (repeated-measures ANOVA, F_4,24=22.8, P<0.001), and mean ventilation rate was positively correlated with the mean tracheal (Spearman’s rank correlation coefficient=1.0, P<0.05, Fig. 11).

Ventilation rate did vary with treatment during the analysis of the effects of restraint and cannulation on ventilation rate (repeated-measures ANOVA, F_3,24=5.69, P<0.01), as ventilation rate decreased with each successive treatment (Table 3). However, ventilation rate did not differ significantly between unrestrained and restrained locusts (Table 3, group 1 versus group 2, post-hoc comparisons, repeated-measures ANOVA). The ventilation rate of the cannulated animals did not differ from the mean ventilation rate of the pooled pre- and post-cannulation animals (group 3 versus groups 2 and 4 pooled, post-hoc comparisons, repeated-measures ANOVA).

We conclude that neither restraint nor tracheal cannulation had large effects on the ventilation rate of our grasshoppers.

Time after injection (up to 10 min) did not significantly affect the tracheal measured for either the NaHCO₃-or NaCl-injected animals (repeated-measures ANOVA). Animals injected with NaHCO₃ showed a significant 23 % increase in tracheal levels averaged over the 10 min following injection (N=11, pre-injection 1.6±0.17 kPa, post-injection 2.0±0.14 kPa; paired t-test, t=3.06, P=0.014).

Injection of NaCl did not significantly affect tracheal (N=10, pre-injection 1.9±0.32 kPa, post-injection 2.1±0.29 kPa; t=0.94, P>0.05).

In grasshoppers, haemolymph acidification does not increase resting ventilation rate. Resting ventilation rate is, however, strongly affected by tracheal and ⁠. Our data suggest that grasshoppers homeostatically regulate the levels of both tracheal and by varying abdominal pumping rate. Ventilation rate is markedly reduced when the tracheal rises above or falls below normal values (Figs 10, 11). These results strongly suggest that the levels of and in the trachea stimulate ventilation rate in normal, resting grasshoppers.

Ventilation rate is unaffected by a reduction in haemolymph pH in R. guttata. We manipulated haemolymph pH within normal physiological ranges without affecting ventilation rate. In jumping grasshoppers, pH changes by 0.1–0.2 units; while with temperature changes, haemolymph pH varies by 0.17 units 10 °C^-1 (Harrison, 1988, 1989b). In the present study, we reduced haemolymph pH by up to 0.37 units without any effect on ventilation rate. These results strongly suggest that locusts lack chemoreceptors for pH in the haemolymph. It is possible that ventilation volume, but not rate, is affected by haemolymph pH. However, if ventilation volume is increased relative to metabolic rate in response to a decrease in extracellular pH, then haemolymph acidosis should be accompanied by a temporary decrease in haemolymph ⁠. This is not the case in grasshoppers (Fig. 1, Harrison et al. 1992), suggesting that neither ventilation volume nor ventilation rate is affected by haemolymph acidosis.

The relationship between haemolymph [HCO₃^- ] and ventilation rate is less clear. NaHCO₃ injection increased ventilation rate and haemolymph [HCO₃^- ]; however, decreases in haemolymph [HCO₃^- ] (HCl injection) had no effect on ventilation rate. NaHCO₃ injection also increased tracheal by 0.4 kPa, presumably as a result of the actions of intracellular carbonic anhydrase, perhaps explaining the stimulatory effect of this injection on ventilation rate. An increase in tracheal of 0.4 kPa elevated ventilation rate by approximately 15 % in the tracheal perfusion experiments (Fig. 11), while NaHCO₃ injection increased ventilation rate by approximately 50 % (Fig. 2). The disparity in these figures may be due to (1) the more long-lasting effects of the NaHCO₃ injections than the perfusions on tracheal and ventilation rate, (2) a more effective increase in the tracheal near the ganglia in response to the NaHCO₃ injection than in response to the tracheal perfusion, or (3) a direct effect of high haemolymph [HCO₃^- ] on neuronal cells which generate or modulate the ventilatory rhythm. We were unable to manipulate haemolymph successfully with our injections (Fig. 1), presumably because CO₂ can move rapidly between the tracheal, intra- and extracellular compartments as a result of the actions of intracellular carbonic anhydrase. Therefore, we cannot rule out the presence of chemoreceptors which sense haemolymph ⁠.

Our results differ from those presented by Snyder et al. (1980) for the cockroach Periplaneta americana in which the nerve cord of a dissected cockroach was irrigated with fluids varying in pH, and ⁠. In their study, changes in saline pH had strong effects on ventilation rate. It is not clear whether the differences between these results and those reported here are due to the use of different animals or different methods. It is possible that, in P. americana, haemolymph pH does affect ventilation rate. Alternatively, the low pH of the saline could have stimulated CO₂ formation from tissue HCO₃^-, causing an elevation of tracheal which stimulated ventilation rate. Also, because the cockroaches were dissected open and pinned, with potential disruption of the tracheal system, results from this preparation may differ from those using intact organisms.

In summary, our data suggest that the widely held assumption that ventilatory rate is sensitive to extracellular pH in insects, as in terrestrial vertebrates, should be re-examined. However, this question clearly needs to be addressed using a wider variety of insect species and with further study on the direct responses of the chemosensory cells to their microenvironment.

While our tracheal perfusions were successful in manipulating tracheal and ⁠, we were unable to manipulate tracheal gas levels to values identical to those of the perfused gases, despite our rapid perfusion rate (Figs 5, 7). Tracheal values were 15–20 % lower than perfusate when hyperoxic gases were perfused, and tracheal was 14 kPa 15 s following the perfusion of 0 kPa (Fig. 5). These data suggest that our tracheal perfusions were accompanied by (1) incomplete flushing of the tracheal system (particularly the contralateral side to the cannulae and the tracheoles), followed by mixing of these unflushed gases with the perfusion gases, and (2) rapid removal of the perfusate gas via the spiracles. The rapid recovery of tracheal gases to normal levels (Figs 4, 6) suggests that grasshoppers completely renew their tracheal gases within 1 min under these conditions.

Tracheal values were only 25 % of the perfusate when hypercapnic gases were perfused and were 1.1 kPa 15 s after perfusion with 0.5 kPa (Fig. 7). The lower tracheal-to-perfusate ratio for CO₂ than for O₂ can be explained by the much higher solubility of CO₂ than O₂ in body fluids. Changes in in the trachea cause rapid changes in of the body fluids (Krolikowski and Harrison, 1996), damping any variation in tracheal ⁠.

Ventilation rate was negatively correlated with tracheal and positively correlated with tracheal (Figs 10, 11).

Although there were small changes in tracheal with some of the tracheal manipulations (Table 1), these were inconsistently related to the changes observed in ventilation rate. Elevation of tracheal above normal levels or depression of below normal levels decreases ventilation rate (Figs 10, 11), suggesting that the levels of O₂ and CO₂ in the trachea stimulate ventilation rate in normal, resting grasshoppers. Approximately 30 % of the animals perfused with 0.5 kPa CO₂ completely ceased abdominal pumping during the first 15 s after the injection. These results strongly suggest that resting locusts regulate tracheal gas levels through variation in abdominal pumping rate. An effect of tracheal and on ventilation rate is also supported by the similar time courses of the changes in tracheal and and ventilation rate after perfusions.

In terrestrial vertebrates, extracellular acid–base status has strong effects on ventilation which are mediated by extracellular chemoreceptors, and this chemosensitivity allows the respiratory system to be an important component of short-term pH regulation (Truchot, 1987). It might be expected that terrestrial insects would have similar chemoreceptors and ventilatory responses, since the high oxygen capacity of the respiratory medium (air) and the efficiency of tracheal gas exchange should allow flexibility in ventilation requirements. Our study suggests that the ventilatory systems of terrestrial insects and vertebrates differ in their sensitivity to extracellular pH. Why do terrestrial insects and vertebrates differ in this way?

One possibility is that extracellular acid–base chemoreceptors evolved in vertebrates to ensure a proper acid–base environment for haemoglobin. In vertebrates, blood pH affects haemoglobin oxygen-affinity and therefore gas exchange. In insects, haemolymph pH should have relatively small effects on gas exchange, since O₂ and CO₂ move primarily via the tracheae. Since gases exchange through the tracheae rather than via the blood as in vertebrates, there may not be a need for the ventilatory system to receive direct sensory information on haemolymph pH in insects.

In terrestrial vertebrates, both the ventilatory and renal systems respond to extracellular pH and are involved in blood pH regulation (Truchot, 1987). Our data suggest a more definite division between the systems involved in acid–base homeostasis in grasshoppers, with the ventilatory system responsible for regulating tracheal and the renal system regulating non-volatile acid–base equivalents and responding to haemolymph pH (Harrison, 1994). The grasshopper alimentary canal and renal system have a high capacity for acid–base transport, and excretion rates of both titratable acid and ammonium vary with haemolymph acid–base status (Harrison and Phillips, 1992; Harrison and Kennedy, 1994; Phillips et al. 1994). In grasshoppers, changes in blood pH with temperature are also primarily driven by non-respiratory mechanisms, consistent with the hypothesis that ionic rather than respiratory mechanisms predominate in the regulation of extracellular acid–base status in locusts (Harrison, 1988, 1989b).

The ventilatory system participates indirectly in extracellular acid–base balance by maintaining a relatively constant tracheal ⁠. Owing to the extensive diffusing capacity of the tracheoles and the action of intracellular carbonic anhydrase (Buck and Friedman, 1958), tracheal strongly influences, but is not identical to, haemolymph (Krolikowski and Harrison, 1996). In the case of an acute extracellular acid load, the respiratory and bicarbonate buffer systems participate in ameliorating extracellular pH changes (Harrison et al. 1990, 1992). Extracellular protons combine with bicarbonate, producing CO₂ which diffuses from the haemolymph into the tracheae. This elevates tracheal ⁠, stimulating ventilation and hence the convective removal of CO₂.

Since our data and that of Krolikowski and Harrison (1996) suggest that CO₂ moves rapidly between the tracheae and the haemolymph, we cannot determine whether or levels are detected in the trachea, haemolymph or body fluids. Resolution of this question will require a study of the microenvironment and responses of identified sensory cells.

Similarly, the cellular mechanism by which tracheal controls ventilation rate remains to be determined. Case (1961) demonstrated that the discharge rhythm of ganglia of the cockroach is sensitive to CO₂ levels and pH changes in vitro. He suggested that rises in external affected this rhythm by reducing the pH within the ganglia. Case (1961) also showed that a variety of weak acids could increase the ventilatory discharge rate of cockroach ganglia in vitro, indicating that undissociated weak acids penetrate the nerve cell membrane and change intracellular pH. While we have shown that decreases in haemolymph pH have no effect on ventilation rate in grasshoppers, it is possible that the actual transduction mechanisms involve changes in the pH of nerve cytoplasm or perineural fluid.

A rise in ventilation rate of grasshoppers after locomotion is not affected by tracheal or ⁠, suggesting that the sensitivity of ventilation rate to tracheal O₂ and CO₂ levels does not function to allow the respiratory system to respond to the increased demands for gas exchange associated with exercise (Krolikowski and Harrison, 1996). Regulating tracheal at a higher level or at a lower level would increase both evaporative water loss and the energy costs of ventilation. However, it is less clear why the trans-spiracular gradients for O₂ and CO₂ are so low. There is no evidence that maintenance of tracheal at 18 kPa (the value we report for unmanipulated grasshoppers) is necessary for adequate O₂ delivery. Metabolic rate in resting grasshoppers does not decrease until ambient falls below 3.4 kPa (Arieli and Lehrer, 1988). However, the large store of O₂ in the tracheal air sacs (created by regulating a high tracheal ⁠) may provide an important oxygen source during locomotion. In Melanoplus bivittatus, 40 % of the O₂ used during a few minutes of jumping comes directly from within the tracheae rather than from freshly inspired air (Harrison et al. 1991). It is also possible that a large O₂ gradient enhances O₂ supply to the leg muscles during burst locomotion.

Ventilation in large insects is regulated jointly by control of spiracular opening and abdominal pumping, with the two systems tightly coordinated (McCutcheon, 1940; Miller, 1960b). Control of spiracular opening appears to be primarily regulated by tracheal gas levels, rather than by extracellular pH (Case, 1957; Hoyle, 1960). During discontinuous ventilation in pupae of the Cecropia silkworm, ‘within the insect the normal stimulus for cyclical valve activity is a combination of decreasing internal tension of O₂ and increasing tension of CO₂ during the interburst period’ (Schneiderman, 1956). A similar mechanism appears to regulate abdominal pumping in quiescent locusts. An accumulation of CO₂ or a lack of O₂ in the grasshopper tracheal system causes the animal to pump its abdomen and open its spiracles, increasing the delivery of fresh air and restoring tracheal O₂ and CO₂ to their regulated levels.

Support for this research was provided by NSF IBN 9317784 to J.F.H., and a grant to S.L.G. from the Howard Hughes Medical Institute through the Undergraduate Biological Science Education Program. Special thanks is given to K. Krolikowski for help with the experiments and G. Walsberg, A. Cooper, and two anonymous reviewers for helpful comments on the manuscript.