Do developmental constraints in ant colonies limit gas exchange strategies to those displayed by female alates (presumptive queens)? In the xeric harvester ant genus Messor, we found that M. pergandei and M. julianus female alates ventilated highly discontinuously, as predicted, but M. julianus workers ventilated less discontinuously and M. pergandei workers (which occur in more xeric habitats) ventilated continuously. We present the salient characteristics of the discontinuous ventilation cycles of the species and the manner in which they are modulated by CO2 emission rates at a single temperature (24 °C). We demonstrate that, in M. julianus workers, open-spiracle phase CO2 emission rate only slightly exceeds overall CO2 emission rate, making discontinuous ventilation marginal, a state extrapolated in M. pergandei to continuous ventilation. However, workers are plainly capable of fargreater rates of CO2 emission than when inactive at 24 °C, so the lack of discontinuous ventilation in M. pergandei under normoxic conditions is not likely to be imposed by physiological constraints and may, in fact, be a response to its xeric environment. We hypothesize – aside from phylogenetic effects – that discontinuous ventilation occurs primarily in insects that may experience hypoxic and hypercapnic conditions, such as ant queens during claustral colony foundation and perhaps workers within the nest environment; that discontinuous ventilation is not necessarily essential to reduce respiratory water loss; and that it will not necessarily occur in castes or species routinely exposed to xeric but normoxic conditions.

Discontinuous ventilation, in the sense of discontinuous external gas exchange, is widely distributed in adult insects (see reviews by Miller, 1981; Kestler, 1985; Slama, 1988; Lighton, 1994). Both the existence and some aspects of the control mechanisms of the discontinuous ventilation cycle or DVC have been studied in cockroaches (Kestler, 1980, 1985; Machin et al. 1991), grasshoppers (Hadley and Quinlan, 1993), beetles (Lighton, 1988a, 1991a) and ants (Lighton, 1988b, 1990, 1992; Lighton and Wehner, 1993; Lighton et al. 1993a,b). These studies may, however, give the erroneous impression that discontinuous ventilation is universal. Researchers naturally tend to study a phenomenon in its more pronounced and quantifiable forms, so exceptions that test the implied rule tend to be under-represented in the literature. For example, it is quite reasonably assumed that mealworm larvae (Tenebrio molitor) ventilate discontinuously (Snyder et al. 1995; Loudon, 1989), and it may be equally reasonably concluded in the light of evidence from African xeric tenebrionid beetles (Lighton, 1991a) that American xeric tenebrionid beetles such as Eleodes spp do so as well. Neither is in fact the case (Lighton, 1994; M. C. Quinlan and J. R. B. Lighton, in preparation). Far from being universal, discontinuous ventilation is, in fact, patchily distributed among insects (see also Hadley and Quinlan, 1993). Ants are unusual in this respect; adults of every species so far studied have ventilated discontinuously.

Why do adult ants ventilate discontinuously? The unusual life-history of most ant species is reflected in their respiratory physiology. During the claustral phase of colony establishment, the founding queen excavates a burrow, seals the entrance, lays eggs, tends and feeds larvae, and feeds the new workers prior to the initiation of foraging, all from her internal reserves (Hölldobler and Wilson, 1990). If – as is presumably the case – the claustral chamber becomes significantly hypercapnic and hypoxic, discontinuous gas exchange will be the most practical gas exchange strategy because it maximizes partial pressure gradients and minimizes durations of spiracular opening (see Discussion).

Given the indisputable evidence for discontinuous ventilation in workers of primitive ant genera (Pogonomyrmex, Lighton et al. 1993b; M. C. Quinlan and J. R. B. Lighton, in preparation; Leptogenys, Duncan and Crewe, 1994) and somewhat more derived genera (Camponotus, Lighton, 1988b, 1990, 1992; Cataglyphis, Lighton and Wehner, 1993; Lighton and given the necessity for discontinuous ventilation in the claustral stage and thus in the queen, discontinuous ventilation in the workers may be a correlated developmental character rather than a necessary characteristic. Accordingly, we hypothesized that, if a selective advantage accrued to abandoning discontinuous ventilation (cf. Tenebrio and Eleodes), then loss of discontinuous ventilation would be found in workers rather than queens. Further, such a loss would most probably arise in highly derived species with an extreme tolerance for xeric environments because, in such species, cuticular permeability is low enough to elevate the selective significance of respiratory water loss (Lighton et al. 1993b; Lighton, 1994) and decoupling of correlated characters in the worker caste is more likely to have occurred than in more primitive and conservative genera, such as Pogonomyrmex.

An excellent example of a highly derived, xeric ant species is Messor pergandei. This member of the myrmicine subfamily is among the most specialized xeric granivores of all ants. In the words of Hölldobler and Wilson (1990), they ‘flourish in the deserts of southwestern Arizona, southern California and Baja California. In Death Valley, one of the driest and hottest places in North America, the Messor pergandei are the most abundant ants; they have a biomass approximately equal to that of the total rodent population in the same area (Went et al. 1972). The toughness of the species in the face of harsh conditions is legendary… [In one study] colonies survived even after twelve years of severe drought (Tevis, 1958).’ We therefore selected workers and alates of M. pergandei as our primary study animal in this investigation. To control, to at least some extent, for phylogenetic effects, we also studied worker and alate ventilation in a congener, Messor julianus, that thrives in the more mesic and productive environment of the Baja peninsula in Mexico.

Animals

Alates and workers from M. pergandei (Mayr) and M. julianus (Pergande) colonies were collected during the February mating flight season from a site 15 km south of Phoenix, Arizona, USA, and from Baja California, Mexico, respectively, and transported by air to our laboratory at the University of Utah shortly after collection. We maintained the ants in small plastic containers at an ambient temperature of 24±0.5 °C and supplied them with water and oat flakes ad libitum. Our experiments were performed within 2 weeks of capture.

Respirometry

Our respirometric techniques are described in full elsewhere (Lighton, 1991b; Lighton et al. 1993b). Briefly, we utilized a Sable Systems TR-3 high-resolution, high-sensitivity flow-through respirometry system (Sable Systems, 476 E. South Temple, Salt Lake City, UT 84112) to detect, with a temporal resolution of 1 s, the CO2 added by a single worker or alate ant to a CO2-free airstream. From CO2 concentration and flow rate, controlled at 50 ml min 1 by a mass flow controller, we calculated the rate of CO2 output according to Lighton (1991b). The three phases of the discontinuous ventilation cycle (DVC) are easily distinguishable in high-resolution records of and the interpretations used in this paper have been verified in other species of ants by direct visual observation of spiracular activity (Lighton et al. 1993a). All measurements took place at ambient temperature in an air-conditioned laboratory regulated at 24±0.5 °C. Computerized data acquisition and analysis were performed using DATACAN V high-resolution A/D hardware and software (Sable Systems).

Statistics

Means are accompanied by standard deviations and sample sizes. Means are compared with Student’s t-test, utilizing the square root of arcsine-transformed data in the case of ratios and percentages. Least-squares regressions, with axis transformation where appropriate, were used. The significance level of regressions was evaluated with analysis of variance (ANOVA), and regressions were compared with analysis of covariance (ANCOVA). The significance level was set at P<0.05.

Standard metabolic rate

The standard metabolic rates (SMRs) of the workers and alates of M. pergandei and M. julianus are shown in Tables 1 and 2 with sample sizes. All castes and species displayed SMRs typical for ants of their body masses (Fig. 1). We emphasize this point here because some of our ventilation results (below) differ significantly from those previously reported in ants. These differences are not caused by significantly deviant SMRs.

Table 1.

Metabolic and ventilatory parameters of Messor pergandei workers, male and female alates at an ambient temperature of 24°C

Metabolic and ventilatory parameters of Messor pergandei workers, male and female alates at an ambient temperature of 24°C
Metabolic and ventilatory parameters of Messor pergandei workers, male and female alates at an ambient temperature of 24°C
Table 2.

Metabolic and ventilatory parameters of Messor julianus workers and female alates at an ambient temperature of 24°C

Metabolic and ventilatory parameters of Messor julianus workers and female alates at an ambient temperature of 24°C
Metabolic and ventilatory parameters of Messor julianus workers and female alates at an ambient temperature of 24°C
Fig. 1.

The relationship between live body mass and metabolic rate in 23 species of ants (open circles), including Messor pergandei female alates (filled circle), male alates (open square) and workers (filled square), and M. julianus queens (+) and workers (X). All data are corrected from measurement temperature (23–30 °C) to 25 °C assuming a Q10 of 2.5 (Withers, 1992); references to the other ant data are in Lighton and Fielden (1994). The equation relating metabolic rate (MR) in μW to live body mass (M) in g is MR=743.0M0.812 [r2=0.87; F(1,21)=140; P<0.0001].

Fig. 1.

The relationship between live body mass and metabolic rate in 23 species of ants (open circles), including Messor pergandei female alates (filled circle), male alates (open square) and workers (filled square), and M. julianus queens (+) and workers (X). All data are corrected from measurement temperature (23–30 °C) to 25 °C assuming a Q10 of 2.5 (Withers, 1992); references to the other ant data are in Lighton and Fielden (1994). The equation relating metabolic rate (MR) in μW to live body mass (M) in g is MR=743.0M0.812 [r2=0.87; F(1,21)=140; P<0.0001].

The discontinuous ventialtion cycle

Occurrence

When inactive, all alates of both species ventilated discontinuously (Figs 2, 3). In M. julianus, all inactive workers also ventilated discontinuously (Fig. 3). However, we observed no such tendency in M. pergandei workers (Fig. 2). In a total of 38 M. pergandei workers examined, we found no evidence of consistent or unambiguous discontinuous ventilation. Most displayed a constant with very little variation from its mean value (Fig. 2). This was not caused by an activity-induced disruption of discontinuous ventilation (Lighton, 1988b). We observed the ants standing motionless in the respirometer vials for long periods while ventilating continuously and the SMR measured was characteristic of motionless ants of their body mass (Fig. 1). In another investigation, carried out in Zzyzx, California, on Messor pergandei locomotion energetics (J. R. B. Lighton and F. D. Duncan, in preparation), no evidence of ventilatory discontinuity was found in any of over 100 recordings of motionless workers at a wide range of temperatures. In one unusually large M. pergandei worker, in the present investigation, we did observe a brief episode of discontinuous ventilation on one occasion (data not shown) so we cannot conclusively state that M. pergandei workers never, under any circumstances, ventilate discontinuously. We are confident, however, that these events are exceedingly rare, at least under the normoxic and normocapnic (or acapnic) conditions of our experiments.

Fig. 2.

Ventilation, as externally measured rates of CO2 emission (V·CO2 in mlh−1) of a typical individual Messor pergandei female alate (A), male alate (B) and worker (C). Note scale differences on the y-axes. Live body masses were 41.7, 18.6 and 6.6 mg respectively.

Fig. 2.

Ventilation, as externally measured rates of CO2 emission (V·CO2 in mlh−1) of a typical individual Messor pergandei female alate (A), male alate (B) and worker (C). Note scale differences on the y-axes. Live body masses were 41.7, 18.6 and 6.6 mg respectively.

Fig. 3.

Ventilation, as externally measured rates of CO2 emission (V·CO2 in mlh−1) of a typical individual Messor julianus female alate (A) and worker (B). Variation in V·CO2 between open (O) phases in the worker is caused by instrument noise (note scale differences on the y-axes). Live body masses were 22.6 and 5.5 mg respectively.

Fig. 3.

Ventilation, as externally measured rates of CO2 emission (V·CO2 in mlh−1) of a typical individual Messor julianus female alate (A) and worker (B). Variation in V·CO2 between open (O) phases in the worker is caused by instrument noise (note scale differences on the y-axes). Live body masses were 22.6 and 5.5 mg respectively.

Salient values of the DVC in M. pergandei and M. julianus are listed in Tables 1 and 2.

The distinction between discontinuous and non-discontinuous ventilation is qualitatively obvious (Figs 2, 3), but can also be quantified by using the coefficient of variation (CV; standard deviation/mean; Tables 1, 2). Utilizing the arcsine of square-root-transformed CVs for testing purposes (to normalize CV distribution), M. pergandei workers were highly significantly less variable in their ventilation than male or female alates of their own species or the workers or female alates of M. julianus (t>5.8; P<0.00001). Interestingly, the workers of M. julianus, although they were still able to ventilate discontinuously, had significantly lower CVs than the female alates (Table 2; P<0.001).

Phases and timing

The ventilation phase coefficient is a measure of the incremental increase in ventilatory phase duration as overall DVC duration increases (Lighton, 1990, 1991a). The durations of the closed (C), flutter (F) and burst (O) phases as a function of total DVC duration in both species are shown in Figs 4, 5 and 6, respectively. In M. pergandei female alates, the C ventilation phase coefficient was 14.1±4.4% (S.E.), while the F ventilation phase coefficient was 61.7±5.4% and the O ventilation phase coefficient was 19.1±5.4%. Male alates differed chiefly in not displaying a consistent F phase, which was found in only three of our sample of 26 ants. Their C ventilation phase coefficient was therefore larger, at 86.7±5.1% (Fig. 4). In M. julianus female alates, the C ventilation phase coefficient was 16.1±2.2%, the F ventilation phase coefficient was 63.3±3.6% and the O ventilation phase coefficient was 16.9±2.6%. No workers displayed an unambiguous F phase; consequently, their C ventilation phase coefficient was greater, at 76.4±5.0%. Their O ventilation phase coefficient was 16.9±4.1% (Fig. 6).

Fig. 4.

The relationship between closed (C) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. Male alates seldom, and workers never, displayed a flutter (F) phase (three male alates that did display an F phase are excluded from this graph); their C phase therefore occupies a larger proportion of their total DVC duration than in female alates of either species (ANCOVA: shared slope 70.5%; P>0.4). The female alates shared a common slope or C ventilation phase coefficient of 18.5% (ANCOVA: P>0.4).

Fig. 4.

The relationship between closed (C) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. Male alates seldom, and workers never, displayed a flutter (F) phase (three male alates that did display an F phase are excluded from this graph); their C phase therefore occupies a larger proportion of their total DVC duration than in female alates of either species (ANCOVA: shared slope 70.5%; P>0.4). The female alates shared a common slope or C ventilation phase coefficient of 18.5% (ANCOVA: P>0.4).

Fig. 5.

The relationship between flutter (F) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. Male alates seldom, and workers never, displayed an F phase (three male alates that did display an F phase are excluded from this graph); their F phase durations therefore lie along the horizontal line y=0. The female alates shared a common slope or F ventilation phase coefficient of 64.0% (ANCOVA: P>0.2).

Fig. 5.

The relationship between flutter (F) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. Male alates seldom, and workers never, displayed an F phase (three male alates that did display an F phase are excluded from this graph); their F phase durations therefore lie along the horizontal line y=0. The female alates shared a common slope or F ventilation phase coefficient of 64.0% (ANCOVA: P>0.2).

Fig. 6.

The relationship between burst (O) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. The line of equality between O phase and total DVC durations is also shown (see Discussion). All species and castes shared a common slope or O ventilation phase coefficient of 17.3% (ANCOVA: P>0.4).

Fig. 6.

The relationship between burst (O) phase duration and total DVC duration in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. The line of equality between O phase and total DVC durations is also shown (see Discussion). All species and castes shared a common slope or O ventilation phase coefficient of 17.3% (ANCOVA: P>0.4).

In the case of the O phase, ventilation phase coefficients can be easily misinterpreted owing to their incremental nature. If O phase duration is expressed instead as a percentage of total DVC duration, a more intuitive measure of the relative duration of the O phase emerges. In M. pergandei, O phase duration is 47.0±11.8% of the DVC in female alates and a statistically equivalent 47.4±14.0% in male alates. In female alates of M. julianus, relative O phase duration was significantly less at 35.9±8.6% (P<0.01), reflecting the longer DVCs in this species (Tables 1, 2). In workers of M. julianus, however, O phase duration was long indeed, occupying over half of each DVC (57.1±17.0%), significantly more than in male or female alates of either species (P<0.001).

Modulation

In M. pergandei alates, ventilation frequency increased with (Fig. 7), but the effect was only modestly significant, with explaining less than 15% of DVC frequency variation. Males displayed a slight, but significant, trend towards higher DVC frequencies at high (analysis of covariance, ANCOVA: P<0.05). Burst volume, or the volume of CO2 emitted during the O or burst phase, also increased with (Fig. 8), but less than 40% of burst volume variance was explained by and male and female alates did not differ significantly in the modulation of burst volume by (ANCOVA: P>0.1). Under these circumstances, it is reasonable to hypothesize that alates displaying high burst frequencies will also display low burst volumes, and vice versa. This hypothesis can be tested by examining the relationship between unexplained variances in DVC frequency and burst volume versus If this is done by analysis of the residuals of data in Figs 7 and 8, this hypothesis cannot be disproved (Fig. 9). In spite of slightly higher DVC frequencies in males, the inverse co-modulation of DVC frequency and burst volume in male and female alates is identical [ANCOVA on residuals in Fig. 9: F(1,47)=0.02; P(same slope)=0.4: F(1,48)=0; P(same intercept)=0.4]. Equivalently, multiple regression analysis, using as the dependent variable and DVC frequency and burst volume as the independent variables, yields a coefficient of multiple correlation of 0.964. Thus, 93% of the variance of CO2 is explained by the equation:
Fig. 7.

The relationship between overall V·CO2 and DVC frequency in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. They do not share common slopes (ANCOVA: P<0.001; see text). Where y is DVC frequency in mHz and x is V·CO2, in M. pergandei female alates, y=3.43+202.9x (r2=0.06, P>0.2) and in male alates, y=0.61+938.0x (r2=0.52, P<0.001). In M. julianus female alates, y=0.24+862.2x (r2=0.26, P<0.001) and in workers, y=5.43+4354.5x (r2=0.26, P=0.003).

Fig. 7.

The relationship between overall V·CO2 and DVC frequency in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. They do not share common slopes (ANCOVA: P<0.001; see text). Where y is DVC frequency in mHz and x is V·CO2, in M. pergandei female alates, y=3.43+202.9x (r2=0.06, P>0.2) and in male alates, y=0.61+938.0x (r2=0.52, P<0.001). In M. julianus female alates, y=0.24+862.2x (r2=0.26, P<0.001) and in workers, y=5.43+4354.5x (r2=0.26, P=0.003).

Fig. 8.

The relationship between overall V·CO2 and open (O phase) CO2 emission volume in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. They share a common slope of 26.79 [ANCOVA: F(3,132)=2.54; P=0.06; see text]. The intercepts of the alates are statistically equivalent, but the worker intercept is significantly smaller (see text for further details).

Fig. 8.

The relationship between overall V·CO2 and open (O phase) CO2 emission volume in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown. They share a common slope of 26.79 [ANCOVA: F(3,132)=2.54; P=0.06; see text]. The intercepts of the alates are statistically equivalent, but the worker intercept is significantly smaller (see text for further details).

Fig. 9.

The relationship between unexplained variance in the relationship between DVC frequency and V·CO2 O-phase (Fig. 7) and the unexplained variance in the relationship between burst volume and V·CO2 (Fig. 8) in M. pergandei female alates (open circles) and male alates (filled circles), showing that a strong inverse co-modulation of DVC frequency and O-phase volume explains 90% of the previously unexplained variance of these parameters as a function of V·CO2. Where x is burst volume residuals and y is DVC frequency residuals, y= 18.56x (r2=0.90, P<0.001). See text for further details.

Fig. 9.

The relationship between unexplained variance in the relationship between DVC frequency and V·CO2 O-phase (Fig. 7) and the unexplained variance in the relationship between burst volume and V·CO2 (Fig. 8) in M. pergandei female alates (open circles) and male alates (filled circles), showing that a strong inverse co-modulation of DVC frequency and O-phase volume explains 90% of the previously unexplained variance of these parameters as a function of V·CO2. Where x is burst volume residuals and y is DVC frequency residuals, y= 18.56x (r2=0.90, P<0.001). See text for further details.

where is in ml h1, DVCF is discontinuous ventilation frequency in mHz, and BV is burst volume in μl (combined male and female alate data).

Changes in burst volumes as a function of were not mediated by changes in O phase duration (r2=0.03; P>0.3). Rather, an increase in the rate of CO2 emission during the O phase was responsible for the increase in burst volume. This increase in O phase was strongly correlated with overall (r2=0.53; P<0.001), with O phase increasing 33±18% (S.E.M.) faster than overall in both male and female alates (Fig. 10).

Fig. 10.

The relationship between overall V·CO2 and open (O phase) CO2 emission rate in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown; note that the lower line is not a regression line but is drawn at equality between O phase V·CO2 and overall V·CO2 (see text). The fitted lines share a common slope of 1.21 (ANCOVA; P=0.3; see text). The intercepts of the alates are statistically equivalent, but the worker intercept is significantly (tenfold) smaller (see text for further details).

Fig. 10.

The relationship between overall V·CO2 and open (O phase) CO2 emission rate in discontinuously ventilating M. pergandei and M. julianus. M. pergandei workers did not ventilate discontinuously and are not shown. M. pergandei female alates are open circles, male alates are filled circles; M. julianus female alates are filled squares and workers are crosses. Individual regression lines are shown; note that the lower line is not a regression line but is drawn at equality between O phase V·CO2 and overall V·CO2 (see text). The fitted lines share a common slope of 1.21 (ANCOVA; P=0.3; see text). The intercepts of the alates are statistically equivalent, but the worker intercept is significantly (tenfold) smaller (see text for further details).

The situation is broadly equivalent in M. julianus female alates and workers, but alates differ from workers in the details of DVC frequency and burst volume modulation. In both castes, DVC frequency increased with increasing (Fig. 7), but significantly more rapidly in the case of the workers (ANCOVA: P<0.01). The power of in predicting DVC frequency was, however, poor (<26%; Fig. 7). Female alates did not modulate burst volume in accordance with ; burst volume varied over a threefold range, apparently at random (Fig. 8; r2=0.003; P=0.4). Workers, in contrast, increased burst volumes significantly with increased , although the relationship was still fairly weak (r2=0.33; Fig. 8). Analysis of the residuals for M. julianus in Figs 7 and 8 revealed, however, inverse co-modulation of DVC frequency and burst volume broadly equivalent to that of M. pergandei (Fig. 11). However, workers and female alates differed highly significantly in the details of the modulation. In terms of variance unexplained by , workers modulated DVC frequency versus burst volume some 30 times more steeply than female alates (Fig. 11).

Fig. 11.

The relationship between unexplained variance in the relationship between DVC frequency and V·CO2 (Fig. 7) and the unexplained variance in the relationship between burst volume and V·CO2 (Fig. 8) in M. julianus female alates (filled squares) and workers (crosses). In both cases, a strong inverse co-modulation of DVC frequency and burst volume explains more than 90% of the previously unexplained variance of these parameters as a function of V·CO2 Where x is burst volume residuals and y is DVC frequency residuals, in female alates y= 10.94x (r2=0.92, P<0.001), and in workers, y= 349.7x (r2=0.90, P<0.001). The slope is significantly greater in the workers. This means that workers modulate DVC frequency approximately 30-fold more than burst volume as a function of V·CO2 (see text and equation 3 for further details).

Fig. 11.

The relationship between unexplained variance in the relationship between DVC frequency and V·CO2 (Fig. 7) and the unexplained variance in the relationship between burst volume and V·CO2 (Fig. 8) in M. julianus female alates (filled squares) and workers (crosses). In both cases, a strong inverse co-modulation of DVC frequency and burst volume explains more than 90% of the previously unexplained variance of these parameters as a function of V·CO2 Where x is burst volume residuals and y is DVC frequency residuals, in female alates y= 10.94x (r2=0.92, P<0.001), and in workers, y= 349.7x (r2=0.90, P<0.001). The slope is significantly greater in the workers. This means that workers modulate DVC frequency approximately 30-fold more than burst volume as a function of V·CO2 (see text and equation 3 for further details).

By multiple regression analysis, in female alates of M. julianus:
(coefficient of multiple correlation 0.965, 93% of variance in explained).while in workers of M. julianus:

As with M. pergandei, an increase in O phase was responsible for increasing burst volume as increased; O phase duration remained unchanged as a function of (P>0.2). By ANCOVA, the relationship between O phase and overall in all discontinuously ventilating species and castes shared a common slope of 1.2 (ANCOVA; P<0.001; Fig. 10) but the O phase of the workers was significantly lower, corresponding to a significantly lower intercept (10% of the female alate value; ANCOVA: P<0.001; Fig. 10).

Discontinuous ventilation

Modulation

In all discontinuously ventilating insects studied so far, changes in are accommodated by orderly, linear changes in DVC frequency (Lighton, 1988b; Lighton and Wehner, 1993). In xeric ants, such as the Saharan thief ant Cataglyphis bicolor, burst volumes are fixed at a maximal level that is independent of (Lighton and Wehner, 1993), while in more mesic ants, such as Camponotus. vicinus, burst volumes decline linearly with increasing and increasing DVC frequency (Lighton, 1988b). Thus, in Cataglyphis, only DVC frequency is modulated by , while in Camponotus both DVC frequency and burst volume are modulated, but in a linear, predictable fashion. The alates of M. pergandei and M. julianus are therefore atypical. They modulate burst frequency and burst volume in an apparently capricious fashion, quite unlike the more stereotypical modulation observed in the other species examined to date. Neither burst frequency nor burst volume change with in ways that suggest tight regulation of one parameter or the other.

Direct comparisons of our ventilation modulation data with those presented in the papers cited above are complicated, however, by the relatively small range of shown by our ants, which were investigated at a single temperature. In the other investigations, was changed over a much wider range by altering the ambient temperature. It is possible (but speculative as yet) that DVC modulation by may differ if it is measured at single temperatures and/or over narrow ranges versus a wide range of temperatures and values.

Nevertheless, some informative principles emerge from our data. The modulation characteristics of the alates of both species were generally similar (Figs 711). However, as a function of overall , the workers of M. julianus modulated DVC frequency tenfold more, and burst volumes 3.4-fold less, than did M. julianus alates (Fig. 11; equation 3). The low O phase of the M. julianus workers explains their reliance on the modulation of burst frequency, rather than burst volume, with changing this is very clearly shown in Fig. 10. This aspect of their respiratory physiology is particularly informative when considered in the light of the apparent inability of workers in the congeneric species M. pergandei to ventilate discontinuously under the conditions of our experiments.

Caste effects

We have demonstrated that the alates of both Messor species examined here ventilate discontinuously, while the workers of M. julianus do so rather poorly – with over 50% of each DVC spent in the O phase – and the workers of M. pergandei do so extremely rarely. What is the basis of this difference between alates and workers? The picture becomes clearer if we consider a reductio ad absurdum, namely diffusive gas exchange by an insect that emits CO2 during the O phase at a rate equal to mitochondrial formation of CO2 or ‘overall ’. Such an insect is incapable of discontinuous ventilation. Its ‘O phase’ is a continuous process of diffusion. At the other extreme, an insect that emits CO2 very rapidly during its O phase, and therefore minimizes the relative duration of its O phase, must ventilate highly discontinuously. Between these two extremes is a broad continuum in which O phase rates of CO2 release exceed steady-state, mitochondrial-level by variable margins. Obviously, the closer an insect approaches the line of equality between O phase and overall , the more marginal is its capacity for discontinuous ventilation. In this respect, Fig. 10 is highly informative. Because the O phase of the M. julianus workers, as a function of their overall is tenfold lower than that of the female alates, the workers are barely above the line of equality between O phase and overall at which the DVC must break down. In addition, M. julianus workers devote a large enough proportion of their DVC to the O phase to draw them close to the line of equality at which the DVC becomes a continuous ‘O phase’ (Fig. 6). Note that burst CO2 emission volumes scale as body mass to the power 1.67 in M. julianus (data from Table 2). Thus, in the O phase, female alates of M. julianus release 11 times more CO2, or 2.6 times more CO2per unit body mass, than do workers of that species.

It follows that M. julianus workers barely utilize the strategy of discontinuous ventilation, although the alates of both species do so to a very marked extent. Further, M. pergandei workers do not utilize discontinuous ventilation at all, but are situated on the lines of equality in Figs 6 and 10, a position in no way predictive of poor adaptation to xeric environments, as they themselves (quite apart from the examples of Eleodes and Tenebrio) prove. Is this lack of discontinuous ventilation dictated by physiological necessity, i.e. limited maximal gas exchange rates?

In this context, it is important to realize that, in the case of M. julianus workers at least, the O phase data in Fig. 10 are by no means indicative of maximal We infer this from the observation that M. julianus workers (like M. pergandei workers) are active at temperatures more than 15 °C above those utilized in the present investigation, corresponding to a threefold increase in without even taking the effects of activity into account. Unless three-to tenfold changes in trans-spiracular diffusion gradients occur, which is unlikely, any increase in overall must be accompanied by decreased occlusion of the spiracles and/or by active ventilation.

Thus, the rather mediocre performance of M. julianus workers as discontinuous ventilators and the failure of M. pergandei workers to ventilate discontinuously do not reflect physiological constraints imposed by low maximal rates of CO2 output. In addition, their ventilation strategy does not appear to exact a water loss penalty. Certainly, in the case of M. pergandei workers, failure to ventilate discontinuously has not elevated their overall water loss rates to unusual levels; on the contrary, the species thrives in arid areas (see Introduction) and the water loss rates of workers are low and typical of xeric arthropods (Edney, 1977; Lighton et al. 1995).

Given the appropriate stimulus, it is therefore likely that more pronounced discontinuous ventilation can be elicited from M. julianus workers and that M. pergandei workers can be induced to ventilate discontinuously. This stimulus is likely to be hypoxia and/or hypercapnia of the degree (currently unknown) to which workers are exposed within the nest itself. This hypothesis is readily testable. The advantage (if any) that accrues to maintaining continuous ventilation in normoxic, dry air remains problematic.

Is discontinuous ventilation necessary?

A cautiously comparative approach yields a working hypothesis concerning the puzzlingly patchy distribution of discontinuous ventilation in adult insects. We note that discontinuous ventilation is common in insects likely to encounter high concentrations of CO2 (or low concentrations of O2) in their natural environments. For example, subterranean ant nests are generally assumed to be hypoxic and hypercapnic, and in the case of cockroaches, the burrows beneath stones in which they generally live in nature are poorly ventilated. Many of the Namib Desert tenebrionids known to ventilate discontinuously spend most of their time buried in sand, where because of slow diffusion their immediate environment is likewise likely to be hypoxic and hypercapnic. In contrast, their confamiliar American representatives generally do not burrow or sand-swim, but seek shelter by creeping under vegetation where conditions are close to normoxic and normocapnic, as is also presumably the case with the natural haunts of Tenebrio.

The trend is suggestive, although as yet we lack the broad base of comparative data required to progress beyond speculation. And, as always, phylogeny may introduce unwelcome if potentially heuristic complications, such as normoxic insects that ventilate discontinuously on a somewhat capricious basis (see, for example, Hadley and Quinlan, 1993). A clear mechanistic determinant of the trend towards continuous ventilation in normoxic environments is, however, obvious. It is impossible to initiate a net flux of CO2 or O2 from areas of low to high partial pressures. The endotracheal volume of an insect must, of necessity, be hypoxic and hypercapnic relative to its environment or metabolically useful diffusion-driven flux cannot occur. An insect can solve the conundrum of diffusive gas exchange in significantly hypoxic and hypercapnic environments in two distinct ways. First, it may open its spiracles maximally for long periods, thus increasing net diffusive flux and compensating to some extent for a shallow partial pressure gradient. However, unless the insect’s environment is perfectly water-saturated, this strategy may impose an unacceptable water loss penalty; overall water loss rates can increase several-fold while the spiracles are open (Lighton et al. 1993b). The second alternative is to sequester CO2 and deplete O2 within the tracheal system for as long as physiologically feasible. In this case, the partial pressure gradient will increase until a brief period of spiracular opening can effect the required net flux. Hence, discontinuous ventilation.

In normoxic and normocapnic environments, the DVC may be considered a somewhat intemperate strategy because of the far steeper partial pressure gradients driving net fluxes of O2 into, and CO2 out of, the insect’s tracheal system, and it may not, in fact, be necessary. We infer this from the observed absence of discontinuous ventilation in some successful denizens of xeric habitats (Tenebrio and Eleodes). The advantages, if any, accruing to this more sedate ventilatory strategy are problematic, particularly because the partial pressure gradient driving the diffusive uptake of O2 is approximately 18 kPa, while that driving the diffusive release of CO2 is only about 4 kPa (Levy and Schneiderman, 1966). This explains the short-term respiratory exchange ratio of approximately 0.2 during the steady-state, diffusive F phase of those insects in which it has been measured (Lighton, 1988a; J. R. B. Lighton, in preparation). Obviously, the F phase is not an indefinitely sustainable gas-exchange strategy. For continuous diffusion-driven gas exchange to be practical, O2 and CO2 partial pressure gradients need to be similar, which can be accomplished by lowering O2 partial pressure gradients (at the expense of increasing spiracular diffusion rates and presumably entailing a water loss penalty) or by elevating endotracheal CO2 partial pressures approximately fivefold (with problematic, if not calamitous, effects on pH regulation). From this we deduce that some form of continuous active, convective ventilation is likely to accompany any shift away from discontinuous gas exchange. If the correct balance between active ventilation and passive diffusion is found, however, one potential benefit of continuous gas exchange may be a more constant internal environment (see Bridges and Scheid, 1982). The detailed nature of this ventilatory strategy awaits investigation, but it is not unduly Panglossian to infer that some selective advantage must accompany the trend away from discontinuous gas exchange in some highly xeric insects. It is, in any event, clear that the respiratory physiology of adult insects may be more subtle, sophisticated and varied than is generally supposed.

We gratefully acknowledge support from the US NSF (grants BSR 9006265 and IBN-9306537 to J.R.B.L.) and from the David and Lucille Packard Foundation (Fellowship to J.R.B.L.). We thank Drs Laura Fielden and Michael Quinlan for critically reading the manuscript, two anonymous reviewers for helpful suggestions, Drs Gary Snyder and David Scholnick for allowing us access to an unpublished manuscript, and Dr Robert Johnson for excavating our ants and arranging for their transport; this investigation would not have been possible without his generous and valued assistance.

Bridges
,
C. R.
and
Scheid
,
P.
(
1982
).
Buffering and carbon dioxide dissociation of body fluids in the pupa of the silkworm moth Hyalophora cecropia
.
Respir. Physiol.
48
,
183
198
.
Duncan
,
F. D.
and
Crewe
,
R. M.
(
1994
).
A comparison of the energetics of foraging of three species of Leptogenys (Hymenoptera, Formicidae)
.
Physiol. Ent. (in press)
.
Edney
,
E. B.
(
1977
).
Water Balance in Land Arthropods.
Berlin
:
Springer Verlag
.
Hadley
,
N. F.
,
Machin
,
J.
and
Quinlan
,
M. C.
(
1986
).
Cricket cuticle water relations: permeability and passive determinants of cuticular water content
.
Physiol. Zool
.
59
,
84
94
.
Hadley
,
N. F.
and
Quinlan
,
M. C.
(
1993
).
Discontinuous carbon dioxide release in the Eastern lubber grasshopper Romalea guttata and its effect on respiratory transpiration
.
J. exp. Biol.
177
,
169
180
.
Hölldobler
,
B.
and
Wilson
,
E. O.
(
1990
).
The Ants.
Cambridge, MA
:
Harvard University Press
.
Kestler
,
A.
(
1985
).
Respiration and respiratory water loss
. In
Environmental Physiology and Biochemistry of Insects
(ed.
K. H.
Hoffmann
), pp.
137
183
.
Berlin
:
Springer Verlag
.
Kestler
,
P.
(
1980
).
Saugventilation verhindert bei Insekten die Wasserabgabe aus dem Tracheensystem
.
Verh. dt. zool. Ges.
1980
,
306
.
Krogh
,
A.
(
1941
).
The Comparative Physiology of Respiratory Mechanisms, chapter
8
, pp.
114
143
. Philadelphia, PA: University of Pennsylvania Press.
Levy
,
R. I.
and
Schneiderman
,
H. A.
(
1966
).
Discontinuous respiration in insects. II. The direct measurement and significance of changes in tracheal gas composition during the respiratory cycle of silkworm pupae
.
J. Insect Physiol.
12
,
83
104
.
Lighton
,
J. R. B.
(
1988a
).
Simultaneous measurement of oxygen uptake and carbon dioxide emission during discontinuous ventilation in the tok-tok beetle, Psammodes striatus
.
J. Insect Physiol.
34
,
361
367
.
Lighton
,
J. R. B.
(
1988b
).
Discontinuous carbon dioxide emission in a small insect, the formicine ant Camponotus vicinus
.
J. exp. Biol.
134
,
363
376
.
Lighton
,
J. R. B.
(
1990
).
Slow discontinuous ventilation in the Namib dune-sea ant Camponotus detritus (Hymenoptera, Formicidae)
.
J. exp. Biol.
151
,
71
82
.
Lighton
,
J. R. B.
(
1991a
).
Ventilation in Namib Desert tenebrionid beetles: mass scaling and evidence of a novel quantized flutter phase
.
J. exp. Biol.
159
,
249
268
.
Lighton
,
J. R. B.
(
1991b
).
Measurements on Insects
. In
Concise Encyclopedia on Biological and Biomedical Measurement Systems
(ed.
P. A.
Payne
), pp.
201
208
.
Oxford
:
Pergamon Press
.
Lighton
,
J. R. B.
(
1992
).
Simultaneous measurement of CO2 emission and mass loss in two species of ants
.
J. exp. Biol.
173
,
289
293
.
Lighton
,
J. R. B.
(
1994
).
Discontinuous ventilation in terrestrial insects
.
Physiol. Zool.
67
,
142
162
.
Lighton
,
J. R. B.
and
Fielden
,
L. J.
(
1994
).
Mass scaling of standard metabolism in ticks: A valid case of low metabolic rates in sit-and-wait strategists
.
Physiol. Zool. (in press)
.
Lighton
,
J. R. B.
,
Fukushi
,
T.
and
Wehner
,
R.
(
1993a
).
Ventilation in Cataglyphis bicolor: regulation of CO2 release from the thoracic and abdominal spiracles
.
J. Insect Physiol.
39
,
687
699
.
Lighton
,
J. R. B.
,
Garrigan
,
D. A.
,
Duncan
,
F. D.
and
Johnson
,
R. A.
(
1993b
).
Spiracular control of respiratory water loss in female alates of the harvester ant Pogonomyrmex rugosus
.
J. exp. Biol.
179
,
233
244
.
Lighton
,
J. R. B.
,
Quinlan
,
M. C.
and
Feener
,
D. H.
(
1995
).
Is bigger better? Water balance in the polymorphic desert harvester ant Messor pergandei
.
Physiol. Ent. (in press)
.
Lighton
,
J. R. B.
and
Wehner
,
R.
(
1993
).
Ventilation and respiratory metabolism in the thermophilic desert ant, Cataglyphis bicolor (Hymenoptera, Formicidae)
.
J. comp. Physiol.
163
,
11
17
.
Loudon
,
C.
(
1989
).
Tracheal hypertrophy in mealworms: design and plasticity in oxygen supply systems
.
J. exp. Biol.
147
,
217
236
.
Machin
,
J.
,
Kestler
,
P.
and
Lampert
,
G. J.
(
1991
).
Simultaneous measurements of spiracular and cuticular mass losses in Periplaneta americana: implications for whole-animal mass loss studies
.
J. exp. Biol.
161
,
439
453
.
Miller
,
P. L.
(
1981
).
Ventilation in active and in inactive insects
. In
Locomotion and Energetics in Arthropods
(ed.
C. F.
Herreid
II
), pp.
367
390
.
New York
:
Plenum
.
Slama
,
K.
(
1988
).
A new look at insect respiration
.
Biol. Bull. mar. biol. Lab., Woods Hole
175
,
289
300
.
Snyder
,
G. K.
,
Sheafor
,
B.
,
Scholnick
,
D.
and
Farrelly
,
C.
(
1995
).
Gas exchange in the insect tracheal system
.
J. theor. Biol. (in press)
.
Tevis
,
L.
(
1958
).
Interrelations between the harvester ant Veromessor pergandei (Mayr) and some desert ephemerals
.
Ecology
39
,
695
704
.
Went
,
F. W.
,
Wheeler
,
J.
and
Wheeler
,
G. C.
(
1972
).
Feeding and digestion in some ants (Veromessor and Manica)
.
Bioscience
22
,
82
88
.
Withers
,
P. C.
(
1992
).
Comparative Animal Physiology. Fort Worth: Saunders College Publishing
.