Water and Carbon Dioxide Loss From the Cockroach Periplaneta Americana (L.) Measured Using Radioactive Isotopes

Following the classical study of Ramsay (1935a,b), there have been numerous further studies of water loss into air from cockroaches and other insects. Interpretation of some of this work has been confused by failure to understand the basic physics of diffusion processes and by the use of inappropriate and sometimes incorrect units. The data have been collated and revised by Noble-Nesbitt (1991). Some of the problems have been considered by Croghan and Noble-Nesbitt (1989) and Noble-Nesbitt (1991), who point out the advantages of using permeability coefficients to describe cuticular water loss.

Isotopic methods have a number of advantages and these are considered in an accompanying paper (Croghan et al. 1995), which describes a technique using radioisotopes for the measurement and analysis of water and CO₂ loss into air from terrestrial insects. This technique is used here to analyse water loss from the American cockroach Periplaneta americana. In particular, the effects of desiccation on cuticular permeability are considered. In some cases, simultaneous measurements of water and CO₂ loss were made and these are used to analyse the contribution of the cuticular and the tracheal systems to water loss in this species.

Adult male Periplaneta americana (L.) were removed from stock culture (Noble-Nesbitt and Al-Shukur, 1987) and kept without food or water either in humid air for at least 15 h (overnight) or isolated and predesiccated in dry air for 24–96 h before an experiment commenced.

The techniques used to investigate water and carbon dioxide loss are described in Croghan et al. (1995). Tritiated water (THO) and ¹⁴CO₂ were absorbed continuously from the airstream flowing past the insect. The apparatus allowed the investigation to be carried out using relatively unrestrained cockroaches. Before passing over the insect, the airstream was either dried by being drawn through a drying column of 6-16 mesh self-indicating silica gel (Fisons) or humidified by bubbling briskly through two gas bottles in series, both containing saturated NaCl solution. Air in equilibrium with this solution has a relative humidity (RH) of 75 % at ambient temperature (O’Brien, 1948; Winston and Bates, 1960). In practice, the airstream was found to be humidified to 68±2 % RH, as measured with a Vaisala temperature–humidity probe. An air flow rate of 10 ml min^-1 was used in all experiments.

The insects were loaded with THO and in some cases also with ¹⁴CO₃^2-, following overnight THO equilibration, as described in Croghan et al. (1995). Most experiments were run for 8 h, with samples being collected every 10 min. At the end of an experiment, an abdominal sample was taken to obtain residual activity. The THO and ¹⁴CO₂ activities were determined by β-scintillation counting. The data were processed as described in Croghan et al. (1995) and the efflux of THO and ¹⁴CO₂ are expressed as efflux rate constants. In most cases, four replicate experiments were carried out simultaneously. Experiments were carried out at ambient temperatures between 19 and 24 °C.

Some representative examples of experiments carried out in a dry airstream using THO alone are given in Fig. 1. In most cases, the running values of the efflux rate constant were relatively steady, with fluctuations that are taken to represent real fluctuations in water loss. However, it was easy to read off a steady value of the efflux rate constant in order to calculate the permeability coefficient. These permeability coefficients, for individuals that had not been injected, are collated in Fig. 2. The permeability data are plotted against initial mass at removal from the culture, hours of predesiccation and proportion of water lost between removal from the culture and the end of experiment. The mean permeability coefficient for unpredesiccated individuals is 0.54×10⁻⁹±0.062×10⁻⁹ m s^-1 (S.E.M., N=12) and for individuals predesiccated for 72 h the mean permeability coefficient is 0.35×10⁻⁹±0.050× 10⁻⁹ m s^-1 (S.E.M., N=4). This difference is significant (0.02<P<0.05, t-test, variance not assumed to be equal).

As most studies of insect water loss have used gravimetric methods, it was considered useful to compare the gravimetric and isotopic methods directly. In some experiments, using unpredesiccated insects, insects were weighed immediately before being placed in the chamber and then immediately after the end of the experiment. Although some THO loss would occur during the weighing or transfer processes, those times were short compared with the total efflux measurement time. The isotopic loss ratio ΔTHO/THO (where THO is the amount of THO in the insect at the start of the measurement) can be compared directly with mass loss ratio ΔW/W (where W is the mass of water in the insect at the start of the measurement) over the same time period (Fig. 3).

In other experiments, using unpredesiccated insects, the efflux of THO was followed for 2 h in humid air, nominally 75 % (measured 68 %) RH. Then the airstream was changed to dry air and the loss followed for a further 2 h (Fig. 4). A rapid decrease in water loss was seen immediately following the change in relative humidity.

In other experiments, losses of THO and ¹⁴CO₂ were studied simultaneously. Fig. 5 is an example of experiments carried out over 3 h in dry air with samples collected at 10 min intervals. The ¹⁴CO₂ data of Fig. 5 are the same as those used by Croghan et al. (1995) in developing a multicompartment model for ¹⁴CO₂ efflux. A running value of the corrected efflux rate constant can be calculated (Croghan et al. 1995) and this is given in Fig. 5B. The efflux rate constant for ¹⁴CO₂ is always much greater than that for THO (Fig. 5C). In order to improve time resolution, a series of experiments was carried out using the 2 min collection procedure. An example is given in Fig. 6. Again, the efflux rate constant for ¹⁴CO₂ is much greater than that for THO, but now the loss of CO₂ can be seen to be pulsatile. During these pulses, the loss of THO seems to be relatively unaffected. Another example using the 2 min collection procedure is given in Fig. 7. In this example, an increase of water loss during the ventilation bursts is clearly apparent.

As will be shown below, integumentary water loss in Periplaneta is predominantly cuticular and thus the loss can be expressed as a cuticular permeability coefficient (P_d) using equation 1. Although the permeability values for a given individual are consistent (Fig. 1), there is considerable variation between individuals (Fig. 2). The values lie around 0.5×10⁻⁹ m s^-1, with the lowest values below 0.3×10⁻⁹ m s^-1. These isotopic results are consistent with published values for Periplaneta obtained using gravimetric methods (Noble-Nesbitt, 1991). A lower permeability for Periplaneta (0.049×10⁻⁹ m s^-1) has been reported by Machin et al. (1992), using a ventilated capsule attached to the pronotum. As has been previously pointed out (Croghan and Noble-Nesbitt, 1989; Noble-Nesbitt, 1991), values as low as these are difficult to explain in terms of the water diffusion properties of liquid oil or solid wax.

There is no obvious relationship between water permeability and initial mass of the insect (Fig. 2A). However, there is an interesting relationship between P_d and time of predesiccation (Fig. 2B). Permeability is reduced after predesiccation for 72 h. This reduced permeability may be related to the reduction of body water during the predesiccation period (Fig. 2C). A reduction in water loss during desiccation has been found previously in Periplaneta (Noble-Nesbitt and Al-Shukur, 1987, 1988a,b). Permeability is not further reduced after predesiccation for 96 h and the values at this time are more variable. This could be interpreted as the onset of physiological deterioration, since these individuals had lost more than 30 % of their body water (Fig. 2C). This is likely to be close to the lethal limit for water loss and may adversely affect the mechanisms controlling cuticular permeability. Owing to the considerable individual variation, it is clear that further investigations of long-term permeability adaptation should be carried out on the same individual over a long period.

A careful simultaneous comparison was also made of loss of tritiated water and loss of mass (Fig. 3). If the two techniques are measuring the same processes, ΔTHO/THO should be identical to ΔW/W. There are discrepancies, especially at the extremities of the ranges. The nature of these is uncertain. However, the accuracy of the gravimetric method with active insects is much reduced at low loss rates and unrecorded THO losses occur during the setting up and weighing procedures. Compartmentalisation can also be a source of error in the isotopic method (Croghan et al. 1995) and this error could be greater at high rates of water loss. The differences found, using living intact insects, are much less than those given by Machin et al. (1992) for excised pronotal discs. However, the differences found in the present paper are comparable with those between isotopic and balance methods in other animals listed by Nagy and Costa (1980).

The isotopic method allows water loss to be studied in airstreams of different relative humidities. The influence of RH on water loss was studied (Fig. 4). A rapid decrease in isotopic water loss was seen on switching from an airstream of 68 % RH to dry air. This decrease is particularly significant since water absorption on surfaces (insect, chamber or tubing downstream from chamber) during the period of high humidity would be expected to result in a transient increase in THO collection after the switch to dry air. The sudden fall in the rate of water loss obtained, when a dry airstream was substituted for a humid airstream, reinforces earlier conclusions that cuticular water loss is under physiological control and responds rapidly when the insect is suddenly exposed to dry air (Noble-Nesbitt and Al-Shukhur, 1987, 1988a,b). This rapid reduction of permeability must have occurred before measurements commenced in all the experiments in which a dry airstream was used. It must be a different physiological process from the much slower gradual reduction in permeability seen during prolonged desiccation (Fig. 2B) and water stress (Noble-Nesbitt and Al-Shukur, 1987). A suggested model of cuticular permeability control has been developed by Croghan and Noble-Nesbitt (1990) and future application of this isotopic technique should provide further information on the nature of the control of cuticular permeability.

In a number of cases, simultaneous measurements of the loss of THO and ¹⁴CO₂ were made (Figs 5, 6, 7). Expressing the losses as the same physical quantity (rate constant) enables direct comparison of the losses to be made, particularly when the ratio is plotted. The efflux rate constant for ¹⁴CO₂ can be used to calculate the CO₂ output from the insect and, if it is assumed that CO₂ is lost from the insect only by ventilation of the tracheal system, the ventilation rate can also be calculated (Croghan et al. 1995). For the eight animals studied, the estimated CO₂ output ranged from 87 to 623 μl g^-1 h^-1 with an average value of 391 μl g^-1 h^-1. This estimated CO₂ output is comparable with the value for Periplaneta determined by Kestler (1971). The calculated ventilation rate ranged from 66 to 197 μl g^-1 min^-1, with an average value of 124 μl g^-1 min^-1.

The data relating k_w,_v/k_w and k_w are plotted in Fig. 8. This indicates that in Periplaneta the tracheal water loss is a relatively small proportion of total water loss. However, as the loss rate decreases, the proportion of total water loss that is due to ventilation increases. If k_w,_v were constant, a hyperbolic relationship might be expected and this fits the data of Fig. 8 well. A trivial contribution of tracheal water loss was, of course, the assumption made in calculating cuticular water permeability from the efflux rate constant for THO. The experiments were carried out with relatively quiescent insects. If the insects were more active, hyperventilation and an increase in tracheal water loss would be expected. The assumption would also not necessarily be correct for a more xeric, presumably less permeable, species, where the total water loss might be very low and the contribution of tracheal water loss to total water loss would be expected to be greater. In some cases, simultaneous loss of THO and ¹⁴CO₂ was studied using 2 min collection periods (Figs 6, 7). With this time resolution, the CO₂ output can clearly be seen to be periodic. Periodic CO₂ output was initially observed in insects by Punt (1950), although unfortunately he did not illustrate his Periplaneta results. However, Wilkins (1960), using an infrared technique, very clearly demonstrated a periodic release of CO₂ in Periplaneta. In the example illustrated in Fig. 6, the contribution of tracheal water loss to total water loss was only 0.64 % and increased loss of water during the CO₂ ventilation burst was hardly perceptible. In another example (Fig. 7), where the contribution of tracheal water loss to total water loss was 5.9 %, an increase of water loss during the ventilation burst is clearly seen. The difference between these two particular insects was mainly due to the differences in ventilation. Cuticular water loss, however, remains much the greater component of overall water loss.

A result similar to that illustrated in Fig. 7 was obtained, using a quite different technique, in a mesic grasshopper with similar rates of water loss to those found here for the cockroach, by Hadley and Quinlan (1993). An apparently similar result, suggesting an even higher proportion of water loss in the ventilation burst, has been reported for xeric ants (Lighton, 1992), although in that work the CO₂ output and water loss were measured successively rather than simultaneously as they were in the experiments reported here.

The results reported in this paper provide a more detailed picture of the way water is lost from the mesic insect Periplaneta americana. The cuticular permeability has been estimated and some factors that influence the permeability, such as predesiccation and ambient humidity, have been studied. The techniques described in this paper and in the accompanying paper (Croghan et al. 1995) provide a basis for obtaining comparable data for a range of mesic and xeric insect types that will be relevant to further studies of the physiological processes involved in adaptations to control water loss.

We are grateful for the assistance of Dr M. Williams and Mr. G. Thomas in producing computer-generated figures and equations.