Slow Discontinuous Ventilation in the Namib Dune-sea Ant Camponotus Detritus (Hymenoptera, Formicidae)

We know little about short-term gas exchange phenomena (discontinuous ventilation) in adult insects while at rest, and nothing about equivalent phenomena in xeric insects. Yet a common thread running through the literature on discontinuous ventilation in insects has been the pivotal importance of water economy in the evolutionary development and ecophysiological significance of this remarkable ventilation system (Buck, 1958; Schneiderman and Schechter, 1966; Brockway and Schneiderman, 1967; Loveridge, 1968; Kestler, 1978, 1980, 1985; Lighten, 1988a; Corbet, 1988). Comparison of discontinuous ventilation phenomena in two congeneric, physiologically similar insects from very different ecological backgrounds, one mesic and the other hyper-arid, should therefore be useful. If considerations of respiratory water economy have indeed affected fitness as a correlate of ventilation strategy, the ventilation strategies of the two organisms can be expected to differ in ways that we can predict with reasonable confidence.

But comparisons need comparative data, and the only quantitative studies of gas exchange during discontinuous ventilation cycles (DVCs) in adult insects are of temperate-zone, mesic species; a tenebrionid beetle Psammodes striatus (Lighton, 1988a) and a formicine ant Camponotus vicinus (Lighton, 1988b). I describe here the first detailed discontinuous ventilation data obtained from an arid-adapted insect: the Namib Desert dune ant Camponotus detritus (Emery).

C. detritus is a large formicine ant (body mass about 45 mg) endemic to the dune-sea of the central Namib Desert, Namibia. The central Namib Desert is hyper-arid, with a mean annual rainfall of about 20 mm per year (Seely and Stuart, 1976) and additional moisture provided by occasional inland migration of advective fogs. High temperatures, hot winds and large water vapor saturation deficits combine to make the central Namib Desert dune-sea a singularly challenging environment for diurnal small-bodied insects.

C. detritus construct their nests beneath perennial vegetation on dune plinths. From their nests, they travel diurnally up to 200 m over bare sand with surface temperatures often exceeding 55°C (Curtis, 1985a), collecting honeydew from scale insects on perennial grasses (chiefly Stipagrostis sabulicola) or foraging opportunistically for dead or moribund insects (Curtis, 1985b). Thus, the environmental context of C. detritus provides a dramatic contrast to the mesic, montane habitat of Camponotus vicinus (San Jacinto mountains, California, pine and oak forests, elevation 1640m; Lighton, 1988b).

The reader is referred to Miller (1981), Kestler (1985), Corbet (1988) and Slama (1988) for excellent reviews of our current understanding of insect ventilation phenomena. The following abbreviations are used in the text. C phase, closed phase: all spiracles are closed and no external respiratory gas exchange takes place; endotracheal falls and hemolymph rises. F phase, flutter phase: triggered by falling ⁠, the spiracle closer muscles are periodically inactivated and limited respiratory gas exchange takes place. B phase, burst phase [also referred to as the O (open) or V (ventilation) phase]: rising hemolymph inactivates the spiracle closer muscles, and CO₂ is expelled from the insect in a large ‘burst’, usually with ventilatory pulsations.

Respiratory water loss is zero in the C phase, very low in the F phase, and high in the B phase (Lighton, 1988a, and references therein; J. R. B. Lighton, in preparation), so one might expect the C phase to be elongated, and the contribution of the F phase to overall gas exchange to be emphasized, in arid- adapted insects. Consequently, one would expect the B phase, with its high rate of respiratory water loss, to occur less frequently; in other words, for the frequency of the DVC to be reduced.

This investigation was carried out in December 1988 at the Namib Desert research station (Gobabeb, Namibia). Camponotus detritus workers were collected from an established laboratory colony (temperature 26±5°C, ambient photoperiod) or directly from two foraging colonies in the dune-sea. Larvae (last instar) and pupae (close to eclosion; eye-spots visibly darkened) were obtained from the laboratory colony only, which was in good condition, with a fertile queen, plentiful larvae and pupae, and free-ranging workers that supplemented the colony diet of sugar-water and insects with extensive foraging around the station.

Air was taken from outside the building, scrubbed of H₂O and CO₂ by a Drierite/Ascarite/Drierite column, and drawn through a respirometer (volume 10 cm³) at a flow rate of 50 ml min⁻¹ regulated by a calibrated mass flow controller. I measured CO₂ concentration in this air stream with an infrared absorbance monitor tuned to respond to CO₂ only, and integrated with a data acquisition engine (Datacan Field, Sable Systems, Los Angeles, CA). Utilizing sample-cell temperature compensation, digital filtration and baseline correction, system resolution was 0·1 p.p.m. CO₂ and long-term drift was less than 0·2p.p.m.h⁻¹. The temperature of the respirometer chamber and 50 cm of temperature equilibration tubing, through which the incoming air stream flowed, was maintained at 30±0·02°C by a Peltier effect device under computer control. 30°C is a reasonable ‘consensus temperature’ for most diurnal insects in the central Namib Desert and corresponds closely to the preferred temperature of C. detritus adults and brood (31·3±2·4°C; Curtis, 1985c).

Prior to each run, I weighed a selected ant, larva or pupa to 0·1 mg and equilibrated it to 30°C within the respirometer for at least 1 h. I then measured the zero CO₂ baseline of the flow-through system by bypassing the respirometer. After reconnecting the respirometer and flushing it of accumulated CO₂ for 5 min, I recorded CO₂ production for 45 min to 1 h, bypassed the respirometer, and recorded the baseline again. In the case of very low and continuous ⁠, I sometimes measured baselines at one or two points within the recording itself to check for non-linear drift.

During analysis, I corrected drift in the CO₂ monitoring system by linear interpolation between beginning and end baseline readings. Any such drift was linear and less than Ip.p.m. over the timescale of the recordings. STP-corrected rates or volumes of CO₂ production could then be determined over any part of the recording.

Means are accompanied by standard deviation and sample size. Regression analysis was performed by least squares, with axis transformation where noted. Regressions were compared by analysis of covariance (ANCOVA), and means by Student’s t-test. The significance level was set at P < 0·05. Most of the statistical tests were performed with SYSTAT 4·0 (Wilkerson, 1988).

Rate of CO₂ production was measured at 30°C in 33 ants; 21 from two colonies in the dune-sea, and 12 from the laboratory colony. Ventilation was always highly discontinuous when the ants were inactive (Fig. 1). Slight activity, such as slow creeping, increased DVC frequency without disrupting its cyclicity. In contrast, vigorous activity such as escape behavior caused apparently chaotic ventilatory patterns (Fig. 2) accompanied by high ⁠. Such data were not analyzed further. Of the ants examined, nine from one colony in the dune-sea and seven from the laboratory colony maintained a low enough level of activity to exhibit sustained, repetitive DVCs for more than 30 min. Ants from the laboratory colony tended to display continuous, low levels of activity; ants from the dune-sea tended to be either vigorously active or inactive. Mean masses did not differ significantly between the dune-sea and laboratory colony subsamples (0·0444±0·0119g, N=9, and 0·0409±0·0098g, N=7, respectively; t=0·6; df=14; P < 0·3).

CO₂ production of C. detritus larvae was continuous (Fig. 3). of the larvae was 0·0069±0·0011mlh⁻¹ (mean mass=0·0541 ±0·0226g, N=7). In spite of the wide mass range investigated (0·0318– 0·0837g), no significant mass scaling of was found (F=0.1, df=l, 5, P < 0.4).

Discontinuous CO₂ emission in C. detritus was very marked in inactive to slightly active adults (Fig. 1). Following the B or V phase, CO₂ emission was insignificantly above baseline levels (C phase). There followed a rise to a measurable, steadily increasing rate of CO₂ emission (F phase), which presumably reflects intermittent partial openings of the spiracles, in DVCs lasting more than 180 s. In DVCs of shorter duration, no F phase was apparent. The C and F phases occupied tightly defined proportions of the complete DVC (Fig. 4). By regression analysis, the C phase occupied 71·4% (±3·3% S.E.) and the F phase 20·3% (±3·2% S.E.) of total DVC duration, with the B phase occupying the remaining 8.3%. Finally, accumulated CO₂ was released in a large burst (B phase). Active ventilation was not visible during this burst; however, ventilation that was not externally visible was presumably still occurring (see Slama, 1988).

The volume of CO₂ released during the F phase, expressed as a percentage of total CO₂ release, was twofold greater in the dune-sea colony [13·9±2·8% vs 7·3±3·6%; t(arcsine of square root transformed data)=4·0; df=14; P < 0·002]. However, this is a consequence of the lower and lower DVC frequency of the dune-sea colony ants, which leads to a longer F phase (Fig. 4; and see Discussion). Neither the absolute volume nor the proportion of CO₂ released during the F phase scaled significantly with mass in either colony (P=0·1).

At low to zero activity levels at a given body mass and temperature, DVC frequency was determined by ⁠, with higher corresponding to higher DVC frequencies (see Schneiderman, 1960; Lighton, 1988b). After the influence of body mass on had been accounted for by multiple regression, the influence of DVC frequency was highly significant (t= 10·1 ; df=13; P < 0·0001).

A common adaptation to aridity is a reduction in metabolic rate (Snyder, 1971; Bartholomew et al. 1985; Peterson, 1990). This ameliorates the effect of scarce and unpredictable food resources, and reduces respiratory water loss rates. However, the of adult C. detritus (mean 0·290 ml CO₂g⁻¹, mean mass 0·0429 g) is typical for ants of their size. For example, the of C. vicinus at 30°C and a body mass of 0·043 g is an equivalent 0·256ml CO₂g⁻¹ (Lighton, 19886); t=0.33, df=15, P < 0·4. of a species very closely related to C. detritus (C. fulvopilosus mean mass 0·043g; Lighton, 1989) is an almost identical (P < 0.4) 0·286 ml CO₂g⁻¹ at 30°C, estimated from assuming an RQ of 0·828 (Lighton, 1988b). Plainly, if C. detritus exhibits respiratory adaptations to an arid environment, it is not in the direction of reduced metabolic rate. This is particularly interesting in view of the fact that C. detritus does not store food in its nests (Curtis, 1985d). If its physiology is similar to that of the very closely related C. fulvopilosus, it may be able to reduce its metabolic rate substantially as a response to starvation (Lighton, 1989).

What, however, of the possibility that rising hemolymph during the discontinuous ventilation cycle may in itself depress in a synergistic reaction, as Barnhart and McMahon (1987) documented in a mol lusc? Plainly, this would allow the C phase and especially the F phase to lengthen significantly as a consequence of internal hypercapnia. However, were this effect to occur in C. detritus, the postulated modulation of would cause a non-linear inflection in the relationship between C and F phase duration and total DVC duration. Since no such inflection is evident (Fig. 4), such a downward modulation of is evidently not significant in C. detritus.

The fact that the of pupae near eclosion did not differ significantly from that of inactive adults is not surprising and has been reported before (e.g. Bartholomew et al. 1988), while the low metabolic rate of larvae has also been noted (e.g. MacKay, 1982). The absence of significant discontinuous ventilation in either larvae or pupae of C. detritus is much more surprising - particularly so because the DVC was originally discovered and described in the pupae of holometabolous insects (though the pupae were in diapause; see review by Miller, 1981). Because practically no comparative data exist in this area, however, it is impossible to say whether C. detritus is unusual in this respect. For example, it is possible that their larvae lack functional spiracular valves. It is worth noting that, unlike immobile immature forms of solitary species, the brood of social insects is subject to stringent environmental control of temperature and humidity. This may relax selective pressures imposed by very low humidity or high temperatures in an uncontrolled setting. The continuous ventilation of C. detritus larvae and pupae may reflect this relaxation of selection for reduced water loss; however, the benefits of their continuous ventilation are problematic.

C. detritus ventilates once per 5 min at 30°C. At that temperature, the DVC frequency of its comparably sized, mesic congener C. vicinus is three- to fourfold faster (Lighton, 19886; P < 0·0005) in spite of the fact that the values of C. detritus and C. vicinus are equivalent. Is the lower DVC frequency of C. detritus adaptive in reducing respiratory water loss? The fact that C. detritus and C. vicinus have similar values is critical, because itself affects ventilation frequency (Schneiderman, 1960; Lighton, 1988b). Respiratory water loss is rapid during the B phase, in which convective ventilation usually occurs (Kestler, 1980; Lighton, 1988a, and references therein; J. R. B. Lighton, in preparation). If the B phase CO₂ emission volume of C. detritus were proportionately larger than that of C. vicinus, this would offset the water conservation benefit derived from reducing DVC frequency. However, the B phase CO₂ emission volumes of C. detritus are not significantly larger that those of C. vicinus at 30°C (Lighton, 1988b; P=0·1). From this it can be inferred that C. detritus emits more CO₂ during its F phase, when water loss rates are low, than does C. vicinus, allowing it to slow its DVC to the observed low rate.

In C. detritus, the C phase was very long (more than 70% of total DVC duration). During this period, no measurable external gas exchange, and hence no respiratory water loss, took place. By contrast, in the mesic beetle P. striatus, the C phase lasted only 6.7 % of total DVC duration (Lighton, 1988a). Unfortunately, DVC data in C. vicinus (Lighton, 1988b) cannot be directly compared at C- and F-phase level with those in C. detritus because extreme sensitivity of CO₂ analysis was not available in the former investigation. However, an accentuated role of the F phase in CO₂ release in C. detritus can be inferred (see above).

The contribution of the F phase to total CO₂ release in C. detritus with low (dune-sea colony) was very similar to that found in P. striatus (Lighton, 1988a; 13·9 % vs 13·0 %). However, the proportional duration of the F phase was much less (20·3 % vs 45·9 % of the DVC; P < 0·001), reflecting the much longer C phase of C. detritus.

The F phase is initiated when tracheal falls below a critical value during the C phase (Schneiderman, 1960; Levy and Schneiderman, 1966). During this time, CO₂ accumulates in the hemolymph, and continues to accumulate during the F phase until the B phase is triggered. At a given within the range characterized by normal discontinuous ventilation, hypoxia and hypercapnia will initiate the F and B phases, respectively, at fixed intervals after the last B phase. The lengths of the C and F phases must therefore remain proportionately constant relative to DVC duration over a wide range of metabolic rates. This has never been documented in adult insects (an analogous situation in saturniid pupae can be inferred from data in Schneiderman, 1960), but is certainly the case in C. detritus (Fig. 4). It is worth noting that the apportionment between ventilation phases is identical in the laboratory and dune-sea colonies, which stresses that differences in their DVC characteristics reflect differing (and hence DVC frequencies) caused by differing activity levels, rather than different physiology.

The C, F and B ventilation phase coefficients may be useful indices for interspecific comparisons. One would expect a xeric species to display larger C and F phase coefficients, and a lower B phase coefficient, than a mesic species. Comparing C. detritus with P. striatus, for example, we find that C. detritus has a far shorter B phase and a far longer C phase than P. striatus. Both differences are in the predicted direction for a comparison of a mesic with a xeric species. The extent to which other factors such as size and phylogeny interact with these proportionality coefficients is speculative, because the base of comparable data is limited to the two species mentioned.

Relative to mesic insects such as P. striatus or C. vicinus, then, the respiratory water conservation strategy of C. detritus is to increase C phase duration at a given and to allow significant quantities of CO₂ to escape during the F phase, thus reducing the volume of the next B phase with its high rate of respiratory water loss. The discontinuous ventilation characteristics of C. detritus therefore exhibit distinct adaptations to reduce respiratory water loss - plainly a positive correlate of overall fitness to a social insect with diurnal foragers in a hyper-arid environment.

I thank Mary Seely of the Desert Ecology Research Unit, Gobabeb, Namibia, for use of facilities and support for local travel; Gideon Louw for introducing me to the Namib Desert; and Barbara Curtis for introducing me to Camponotus detritus. Partial support was provided by an Alexander Hollaender Distinguished Postdoctoral Fellowship, administered for the US Department of Energy by Oak Ridge Associated Universities.