Carbon dioxide release in Coptotermes ... - Semantic Scholar

Journal of Insect Physiology 47 (2001) 213–224 www.elsevier.com/locate/jinsphys

Carbon dioxide release in Coptotermes formosanus Shiraki and Reticulitermes flavipes (Kollar): effects of caste, mass, and movement Thomas G. Shelton *, Arthur G. Appel Department of Entomology and Plant Pathology, 301 Funchess Hall, Auburn University, Auburn, AL 36849-5413, USA Received 11 October 1999; accepted 20 June 2000

Abstract Movement and carbon dioxide (CO2) release of individual Formosan, Coptotermes formosanus Shiraki and Eastern, Reticulitermes flavipes (Kollar) subterranean termites were recorded simultaneously in real time. Worker, soldier, and pre-alate (nymph) caste termites were recorded over 1-h periods at ambient temperature and normoxia in dry, CO2-free air. No evidence of discontinuous gas exchange cycles (DGCs) was observed in 344 recordings. Intensity of movement was constant in video tape recordings of termites under respirometry conditions. Duration of movement did not have a significant effect on residuals of V˙CO2 regressed on mass. Thus, movement did not effect V˙CO2 for these two species. Overall CO2 release values were calculated for all recordings resulting in mean V˙CO2 (ml CO2 g⫺1 h⫺1), and compared among caste, colony, and species with a nested ANOVA. There was significant interaction (P=0.0161) only for species. Mean CO2 release was significantly greater for R. flavipes (0.507 ml CO2 g⫺1 h⫺1) than C. formosanus (0.310 ml CO2 g⫺1 h⫺1). Mass scaling of termite V˙CO2 was investigated by regressing log10 V˙CO2 on log10 mass. The overall model combining species gave a mass scaling coefficient of 0.861(±0.0791), which approximates a previously published value for the arthropods as a whole (0.825).  2001 Elsevier Science Ltd. All rights reserved. Keywords: Termites; Respiration; Mass scaling; Movement; CO2 release patterns

1. Introduction Subterranean termites (family Rhinotermitidae) nest underground and forage for cellulosic material at and below the soil surface. The Formosan subterranean termite, Coptotermes formosanus Shiraki, and the Eastern subterranean termite, Reticulitermes flavipes (Kollar), are non-mound-building subterranean termites found in many areas of the Southeastern United States (Kofoid, 1934). Their cryptic nature makes in situ behavioral observations impractical. The same is true for estimates of physiological parameters such as CO2 release patterns, and quantitative measurements of respiration (CO2 release and O2 consumption). CO2 release patterns of C. formosanus and R. flavipes have not previously been described.

* Corresponding author. Tel.: +1-334-844-2570; fax: +1-334-8445005. E-mail address: [email protected] (T.G. Shelton).

The discontinuous gas exchange cycle (DGC) is the uncoupling of O2 uptake and CO2 release in tracheate arthropods (Lighton, 1994). The DGC is a specialized subset of cyclic CO2 release gas exchange patterns in insects (Lighton, 1994) and can be separated into open (O), resulting in a burst (B) of CO2 release, closed (C), and flutter (F) phases based on spiracular activity (Schneiderman and Williams, 1955; Buck, 1958; Schneiderman, 1960; Levy and Schneiderman, 1966; Lighton, 1994). The open period occurs in two types, the CFOtype (Schneiderman, 1960) and the CFV-type (using ventilatory abdominal pumping; Kestler, 1971), and has been reviewed by Levy and Schneiderman (1966), Kestler (1985), Lighton (1994, 1998), and Wasserthal (1996) as a possible adaptation to environmental stress. Should DGC occur in termites, it is likely to be of the CFV-type reported from cockroaches and other Hemimetabola (Kestler, 1971). Currently, cockroaches are the closest phylogenetic relatives of termites (Thorne and Carpenter, 1992) for which these patterns have been reported. Two hypotheses have been suggested for the

0022-1910/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 2 - 1 9 1 0 ( 0 0 ) 0 0 1 1 1 - 6

214

T.G. Shelton, A.G. Appel / Journal of Insect Physiology 47 (2001) 213–224

selective force behind the evolution of the DGC by tracheate arthropods (Levy and Schneiderman, 1966; Lighton 1994, 1998). The DGC decreases respiratory water loss by reducing the amount of time spiracles remain open, and one hypothesis is that reduction of water loss is the main selective pressure driving evolution of the DGC (Levy and Schneiderman, 1966; Kestler, 1985; Lighton, 1998). The second hypothesis (Lighton 1994, 1996) assumes that hypercapnic/hypoxic conditions in subterranean burrows has led to the use of the DGC in ants and other insects, and also by convergence in two genera of the Solifugae (Hadley, 1970; Anderson and Ultsch, 1987; Lighton and Garrigan, 1995; Lighton and Fielden, 1996). Hadley and Quinlan (1993) found that in Romalea spp. grasshoppers, water stress resulted in loss of the DGC. The first hypothesis is supported by the work of Williams et al. (1997) with Drosophila melanogaster Meigen, in which strains selected for desiccation resistance exhibited cyclic CO2 release, but not classical DGC. Because subterranean termites live in hypercapnic and hypoxic, as well as humid, environments (Anderson and Ultsch, 1987), it might be expected that C. formosanus and R. flavipes would exhibit the DGC according to the second hypothesis above. Little is known about strategies of energy use in termites. Energy use by ectothermic animals for body maintenance is measured as a standard metabolic rate (SMR), defined as the metabolic rate of an individual that is post-absorptive, quiescent, and unstressed at its thermal preferendum (Withers, 1992). The SMR is often measured as O2 consumption or in units of energy (J, cal, or µW). Whole-colony O2 consumption and CO2 release were examined in several nearctic termites by Wheeler et al. (1996). However, because subcolonies of termites were used (6–150 pseudergates; Wheeler et al., 1996) rather than individuals, no information on the SMR could be derived. Not only was the large number of termites a problem, but movement was not quantified and has certainly confounded their data. Similar problems are found in the recent studies of respiratory quotient (RQ) in termites (Nunes et al., 1997). To determine gas exchange rates accurately, methods of CO2 or O2 measurement of unstressed and quiescent individuals at their thermo- and hygropreferendum is necessary. Allometric mass scaling (the scale at which mass affects physiological rates) of respiratory parameters such as rate of CO2 release (V˙CO2) are important in examining relationships with other animals. Lighton and Fielden (1995) compiled data from numerous studies on mass scaling V˙O2 relationships in arthropods. Their overall regression of log10 V˙O2 on log10 mass for the compiled data resulted in a mass scaling coefficient of 0.825. This estimates the relationship between mass of arthropods (in g) and the metabolic rate (in µW) based on V˙O2. Since O2 consumption is directly related to the release of CO2 through the RQ(=V˙CO2/V˙O2; Withers, 1992), assuming no

change with activity, caste, or age, the mass scaling coefficient of the regression of V˙CO2 on mass is an estimate of the consumption of O2 by these arthropods as well. The coefficient (or slope) is equivalent between the log10 relationships of V˙CO2 and V˙O2 with mass, so only the intercepts differ as RQ differs from 1.0. The purpose of this study was to examine the CO2 release patterns of C. formosanus and R. flavipes individuals. Real-time movement and CO2 release data were obtained simultaneously using flow-through respirometry (see Lighton, 1991a). CO2 release data were used to test the hypothesis that C. formosanus and R. flavipes exhibit DGCs, and whether the DGC varies by species, caste or colony. Due to the mobile nature of these termites, the influence of movement on V˙CO2 was investigated. Specifically, data were used to test the hypothesis that movement altered the magnitude of V˙CO2 of C. formosanus or R. flavipes. There are no data indicating that colony differences (for either species) affect metabolic rate. Since caste members within colonies perform different roles for their colonies, they may also differ in metabolic rates (or costs to the colony). In the light of these questions, we also tested the hypotheses that V˙CO2 varies with caste, colony, or species. Mass scaling of V˙CO2 is reported for each caste, species, and the overall scaling for these termites combined.

2. Materials and methods 2.1. Termites Three colonies of C. formosanus were obtained from aggregation traps (Tamashiro et al., 1973) in Ft. Lauderdale, Broward County, Florida, USA. Reticulitermes flavipes were collected from infested wood (Colony A) obtained from a residential area 6.5 km from the Auburn University campus, Auburn, Lee County, Alabama, USA. Colonies B and C were collected from aggregation traps (Tamashiro et al., 1973) on the campus of Auburn University. Colony B termites were obtained from a building perimeter, while Colony C termites were obtained from a wooded area on campus. Colonies B and C were separated by more than 1 km. Pre-reproductive (nymph) caste termites were not available from Colony C, so nymphs from a fourth colony (D) were collected from infested wood at a residence 苲2 km from the Auburn University campus. For both species, individuals from each colony were placed into plastic boxes (16 cm×27 cm×6 cm), with sand, and pine wood pieces (2.5 cm×10 cm×5 cm), moistened with 100 ml distilled H2O. Boxes were maintained in unlit incubators (苲70% RH) at either 28±1°C for C. formosanus or 22±1°C for R. flavipes. Coptotermes formosanus is a neotropical species surviving best at 28°C, in contrast to R. flavipes, a temperate species,


that survives best at lower temperatures (22°C) (Smythe and Williams, 1972). Although reared at optimal temperatures, these termites are found in nature at a variety of temperatures: C. formosanus ranges from Hawaii to Japan, and the R. flavipes range extends from southern Florida to southern Canada (Kofoid, 1934). Experiments were conducted at ambient laboratory temperatures, well within the temperatures that both termite species encounter in nature. 2.2. Respirometry Individual workers (undifferentiated pseudergates), soldiers, or nymphs (wing-pad-bearing pre-reproductives, when available) were weighed to the nearest 0.01 mg on a digital balance. Body mass, ambient temperature, and barometric pressure were recorded prior to placing an individual termite in the respirometry chamber. All respirometry studies were performed at ambient temperature (23.6±0.08°C; x¯±S.E.). The respirometry chamber consisted of a glass tube (1.8 cm i.d.×6 cm; 15.3 cm3 volume) sealed at both ends with metal stoppers modified to allow air flow through the chamber (Fig. 1). Vacuum grease was used to seal the rubber O-rings on the metal stoppers to the glass tube respirometry chamber. A thin layer of Teflon emulsion (E. I. Du Pont de Nemours & Co., Wilmington, DE) was applied to the interior surface of the metal stoppers to prevent termites from climbing onto the stoppers, thus keeping the termites in the view field of the movement detector (Fig. 1). This relatively large respirometer was used to minimize escape behaviors due to close confinement. Individual termites were sealed within the respirome-

215

try chamber which was placed in a Sable Systems AD-1 infrared movement detector (Sable Systems, Henderson, NV). The respirometry system was plumbed as follows: compressed air was forced at 苲431 kPa through a Whatman Purge-gas Generator (Whatman, Inc., Haverhill, MA) that scrubbed the air of CO2 and H2O, and into a large (170 l) mixing tank. Air flowed into a second, smaller (25 l) pressure manifold, from which air was drawn into the computer-controlled base-lining unit of a Sable Systems TR-3 respirometry system (Sable Systems, Henderson, NV), with tubing to and from the respirometry chamber held in the AD-1 motion detector. Air from the chamber containing the experimental animal was drawn at 50 ml min⫺1 into a CO2 analyzer (Li6262; Li-Cor Inc., Lincoln, NE; in absolute mode), and through a mass-flow controller (Sierra Instruments, Inc., Monterey, CA) with a vacuum pump (Gast Mfg. Corp., Benton Harbor, MI). The system equilibrated for 5 min prior to base-lining and recording. At the start of each recording, the baselining unit was set to automatically by-pass the chamber containing the termite and record an initial base-line measurement. The air was diverted back into the respirometry chamber containing the termite, equilibrated for 5 min, and 1 h of CO2 release and movement recorded. After 1 h, a final baseline reading was taken as described above. For each species, V˙CO2 and movement of a minimum of nine individuals of each caste from each colony were recorded individually. All recordings were recorded and analyzed using Datacan V (version 5.2; Sable Systems, Henderson, NV). Datacan V software was used to convert CO2 release data to standard temperature and pressure prior to analysis. To examine CO2 release patterns in more detail, we

Fig. 1. Schematic of the respirometry chamber used in these experiments. The chamber contained an individual termite and was positioned within the AD-1 motion detector as indicated.

216


reduced the volume of the respirometry chamber from 15.3 to 0.5 cm3 (by using a smaller chamber) and increased the flow rate from 50 to 100 ml min⫺1 (see Williams et al., 1997). This design increased resolution (by reducing the time for CO2 turnover within the respirometry chamber) about 60-fold, but also increased periods of activity. We used soldiers and workers (five each) of both termite species (colony 2 for C. formosanus and colony C for R. flavipes). We also examined the effects of low temperature (15°C) during extended (3 h) recordings to determine if reduced temperature could induce the DGC in workers and soldiers (six each) of both species. 2.3. Analysis of CO2 release patterns The pattern of CO2 release (ml h⫺1) over time was examined for each recording. All traces were examined for evidence of DGC events. Periods when CO2 release dropped to about 0, along with peaks of CO2 at defined intervals (see Lighton, 1994), were our criteria for the presence of the DGC. 2.4. Observation of movement under respirometry chamber conditions To determine type and intensity of activity of termites held within respirometry chambers, individuals (six workers and six soldiers of each species) were recorded on video. Animals were placed into 15.3 cm3 respirometry chambers (see above), through which dry, CO2-free air was drawn at 50 ml min⫺1. Chambers were incubated at 24±1°C, and illuminated with a red 4-W light bulb at 14 cm from the chamber. Behavior of each termite was recorded for 1.5 h using a compact VHS (VHS-C, JVC model GR-AX70, Victor company of Japan, Yokohama, Japan) video camera. Videotapes were examined and transcribed to record behaviors and speed of movement of each termite. Termite speeds were compared using a Mann–Whitney test for each caste by species (Minitab, 1994). 2.5. Analysis of movement effects CO2 release data were converted to ml CO2 h⫺1, and offset (by 28 s) to adjust for the time lag between the activity detector and CO2 analyzer so that the CO2 data corresponded directly to movement. Examination of 3h recordings demonstrated that there is a calmdown period of about 40 min before CO2 release reaches a steady state (Fig. 2). To quantify the presence of a calmdown period, linear regression of log10-transformed CO2 release (in ml CO2 h⫺1) data (on time in min) over the 0–45 min interval of each recording was performed. Percentages of those recordings with a significant negative slope of CO2 release over this interval appear in Table 1.

Fig. 2. CO2 release data recorded from a single R. flavipes worker (mass=0.00279 g; upper trace; left y-axis). CO2 release data recorded from an empty chamber (lower trace; right y-axis). In both traces, chamber volume is 15.3 ml and flow rate is 50 ml min⫺1.

To avoid overestimation of V˙CO2 values, only the mean CO2 release rates (ml CO2 g⫺1 h⫺1) of the last 15 min (i.e. 45–60 min) of each recording were used to estimate CO2 release rate, and in all V˙CO2 comparisons below. Activity scores were calculated as a proportion: number of seconds of movement divided by total number of seconds in the final 15 min section of each recording (as %). Each species was analyzed separately. The effect of movement on V˙CO2 was examined using linear regression. Residuals of the mass scaling relationship (see below) were regressed on activity scores. Using residuals of the mass scaling relationship, the effects of mass on V˙CO2 were removed, resulting in a relationship between V˙CO2 and duration of activity. Paired t-tests were also used to compare activity between the first 45-min and the last 15-min intervals (Minitab, 1994). This analysis tested the hypothesis that the calmdown period is a function of movement duration. 2.6. Analysis of V˙CO2 values Mass-specific V˙CO2 values for all individuals (µl CO2 g⫺1 h⫺1) were analyzed with a nested analysis of variance (ANOVA; SAS Institute, 1985) to test for significance of main effects (caste, colony, species) and all relevant interactions. Comparisons between castes within species were compared via contrasts (SAS Institute, 1985). Values are reported as x¯±S.E. 2.7. Analysis of mass scaling effects For mass scaling effects, log10 V˙CO2 (ml CO2 h⫺1) data were regressed on log10 mass (g). Outliers over 2 standard deviations away from the fitted model were identified using the Studentized residual procedure and removed from the analysis (Glantz and Slinker, 1990; SAS Institute, 1985). Regressions were performed by caste, species, and on both species combined. Data were


217

Table 1 Percentage of recordings showing a calmdown period in CO2 release over 0–45 min. Paired t-test results [Difference (D) and P-values] of 0–45 min activity percentages vs. 45–60 min Species

Colony

Caste

N

Calmdown %

Da

P

R. flavipes

A

Workers Soldiers Nymphs All Workers Soldiers Nymphs All Workers Soldiers All Nymphs Workers Soldiers Nymphs All Workers Soldiers Nymphs All Workers Soldiers Nymphs All

13 17 20 50 16 19 16 51 17 17 34 17 18 15 19 52 19 19 9 47 19 16 18 53

61.5 76.5 70 70 50 78.9 68.8 66.7 64.7 76.5 70.6 64.7 55.6 100 68.4 73.1 89.5 89.5 77.8 87.2 84.2 87.5 66.7 79.2

⫺1.26 ⫺0.01 0.08 ⫺0.35 ⫺2.72 ⫺2.11 0.285 ⫺1.57 0.04 ⫺0.72 ⫺0.29 ⫺0.86 ⫺0.49 0.06 2.22 0.59 ⫺2.27 0.26 3.16 ⫺0.31 ⫺0.29 0.69 2.15 0.86

0.473 0.997 0.673 0.710 0.002 0.036 0.448 0.001 0.945 0.289 0.499 0.178 0.561 0.890 0.101 0.277 0.020 0.754 0.042 0.618 0.646 0.271 0.060 0.085

B

C

C. formosanus

D 1

2

3

a

D=x¯ activity score0−45⫺x¯ activity score45−60

also examined with analysis of covariance (ANCOVA; SAS Institute, 1985) to determine the significance of species on the combined regression. To compare with Lighton and Fielden’s (1995) mass scaling of metabolic rates of the arthropods as a whole, we converted V˙CO2 (ml CO2 h⫺1) to metabolic rate (µW) using their assumptions: Q10 of 2.5 and RQ of 0.72 (Lighton and Fielden, 1995; Lighton, 1991a). Data were analyzed with ANCOVA to estimate the relationship between log10 µW and log10 mass for C. formosanus and R. flavipes, and to compare this relationship with the same relationship for other arthropods (Lighton and Fielden, 1995).

3. Results 3.1. CO2 release patterns There was no evidence of the DGC (as a specialized class of cyclic CO2 release) in any of the 344 recordings examined, regardless of caste, colony or species. An illustration of a typical recording is shown in Fig. 2. The 20 traces using 60-fold greater resolution likewise did not show evidence of DGC events (Fig. 3). CO2 release never dropped to near base-line levels in any of the traces, and showed no evidence of DGC periods (B, F, or C). CO2 release patterns of both workers and soldiers

Fig. 3. Increased resolution CO2 release sample trace recorded from an R. flavipes worker (mass=0.00330 g). Chamber volume is 0.5 ml and flow rate is 100 ml min⫺1.

of both species at 15°C, and in 3 h recordings at ambient temperatures, showed no indication of DGC phases. 3.2. Direct observations of movement Termites recorded on videotape moved at constant rates [C. formosanus workers: 16.4±1.57 (range 12–20) cm min⫺1; soldiers: 17.5±1.12 (range 15–20) cm min⫺1;

218


R. flavipes workers: 9.2±0.65 (range 7.5–12) cm min⫺1; soldiers: 7.0±0.57 (range 5.45–8.6) cm min⫺1] over the course of the recordings. Both workers and soldiers of C. formosanus moved at a faster rate than R. flavipes (workers: W=21.5, P=0.0094; soldiers: W=21.0, P=0.0043). Termites did pause briefly (range 0:05–5:49 min duration), moving only antennae, but resumed movement in all cases. Thus intensity of movement during these video recordings was constant. No abdominal pumping was observed in either caste of either species. 3.3. Effects of movement on CO2 release There was no relationship between V˙CO2 and activity for either species when residuals of the mass scaling relationship were regressed on activity score (C. formosanus: df=1,143; F=1.46; r2=0.01; P=0.229; R. flavipes: df=1,141; F=1.48, r2=0.01, P=0.226). Percentages of recordings showing a significantly negative interaction between time and V˙CO2 over the first 45-min interval ranged from 50 (R. flavipes colony B workers) to 100% (C. formosanus colony 1 soldiers; Table 1). Durations of movement during the 0–45-min and 45–60-min periods were not significantly different for all but one colony, in which the activity duration was greater during the 45–60-min interval (Table 1). 3.4. Caste, colony, and species effects on CO2 release Nested ANOVA of the V˙CO2 data resulted in only one significant main effect. Colony and caste main effects were not significant, but species (df=1, 4.60; F=13.76; P=0.0161) were significantly different. Coptotermes formosanus (all castes combined) V˙CO2 (0.310±0.011 ml CO2 g⫺1 h⫺1) was significantly lower than R. flavipes (0.507±0.014 ml CO2 g⫺1 h⫺1). Interactions were not significant. V˙CO2 ranged from 0.229±0.008 ml CO2 g⫺1 h⫺1 for C. formosanus soldiers to 0.549±0.022 ml CO2 g⫺1 h⫺1 for R. flavipes soldiers (Table 2). Contrast comparisons of caste by species revealed that C. formosanus soldiers released CO2 at a significantly lower rate than C. formosanus workers and nymphs (Table 2). R. flavipes

nymphs released CO2 at a significantly lower rate than R. flavipes workers and soldiers (Table 2). 3.5. Mass loss and termite condition after respirometry Because they are soft bodied, termites are one of the few insects for which dehydration is readily observable. Dehydrated animals have a sunken, flattened abdomen. Termites removed from the respirometer were not noticeably dehydrated. Mean mass lost for each species during the trials, which is assumed to be water loss, was 6.55±0.40% for C. formosanus and 9.56±0.66% for R. flavipes. 3.6. Mass scaling of V˙CO2 in C. formosanus and R. flavipes Regressions of log10 V˙CO2 on log10 mass for each caste within species resulted in four significant (P⬍0.008) mass scaling effect models, workers [⫺1.64(±0.345) M0.537(±0.135) (df=1, 58; F=15.82; r2=0.22; P=0.0002)], and soldiers of C. formosanus [⫺1.29(±0.412) M0.738(±0.165) (df=1, 55; F=20.07; r2=0.27; P=0.0001)], soldiers [⫺0.643(±0.735) M0.848(±0.307) (df=1, 52; F=7.64; r2=0.13; P=0.0079)] and nymphs [⫺1.17(±0.220) M0.650(±0.096) (df=1, 51; F=45.63; r2=0.48; P=0.0001)] of R. flavipes (M is body mass in g). For individual caste models, no mass range was more than 5.87 mg (Table 3). Regressions using each species with combined castes resulted in significant models (P=0.0001 for both species), with mass ranges of 7.26 and 5.78 mg for C. formosanus and R. flavipes, respectively (Table 3). The overall model for combined species was also significant (P=0.0001), resulting in a mass scaling coefficient of 0.832(±0.069) (Table 3). The regression of log10 V˙CO2 on log10 mass combining castes for each species resulted in mass scaling coefficients of 0.759(±0.070) and 0.715 (±0.098) for C. formosanus and R. flavipes, respectively, and is illustrated in Fig. 4. Species did not significantly affect the relationship of log10 V˙CO2 on log10 mass in the combined species model

Table 2 Mean±S.E. of V˙CO2 (ml CO2 g⫺1 h⫺1) and mass (in mg) of all castes by species of R. flavipes and C. formosanus Species

Caste

n

R. flavipes

Workers Soldiers Nymphs All Workers Soldiers Nymphs All

52 53 52 157 59 59 47 165

C. formosanus

a

×¯ ±S.E. V˙CO2a 0.544±0.033a 0.549±0.022a 0.430±0.011b 0.507±0.014 0.373±0.018a 0.229±0.008b 0.332±0.025a 0.310±0.011

×¯ ±S.E. Mass (range) 2.97±0.052 4.06±0.073 5.36±0.181 4.14±0.103 2.95±0.120 3.22±0.076 6.27±0.206 3.99±0.136

(2.04–4.12) (2.75–5.29) (3.17–7.82) (2.04–7.82) (1.32–4.90) (1.91–4.58) (2.71–8.58) (1.32–8.58)

Means followed by the same letter (within species) are not significantly different at the a=0.05 level, as determined by contrasts.

Worker Soldier Nymph All Worker Soldier Nymph All All

R. flavipes

a

47 52 51 150 58 55 45 158 308

n 2.98±0.05 4.06±0.07 5.37±0.18 4.16±0.10 2.95±0.12 3.24±0.08 6.22±0.21 3.99±0.14 4.07±0.09

(2.04–4.12) (2.75–5.29) (3.17–7.82) (2.04–7.82) (1.32–4.90) (1.91–4.58) (2.71–8.58) (1.32–8.58) (1.32–8.58)

×¯ ±S.E. (Range)

P-values are reported for ANOVA test of H0: slope=0 vs. Ha: slope⫽0.

Both species

C. formosanus

Caste

Species

Mass (mg)

0.486±0.580 0.848±0.307 0.650±0.0962 0.715±0.098 0.537±0.135 0.738±0.165 ⫺0.186±0.154 0.759±0.0703 0.832±0.0686

Slope±S.E.

Pa 0.4065 0.0079 0.0001 0.0001 0.0002 0.0001 0.2347 0.0001 0.0001

Intercept±S.E. ⫺1.61±1.47 ⫺0.643±0.735 ⫺1.17±0.220 ⫺1.00±0.235 ⫺1.64±0.345 ⫺1.29±0.412 ⫺3.14±0.343 ⫺1.13±0.172 ⫺0.837±0.166

0.0150 0.1303 0.4771 0.2615 0.2173 0.2710 0.0319 0.4230 0.3207

r2

Table 3 Results of log10 V˙CO2 (ml CO2 h⫺1; from steady-state section only) regression on log10 mass for each species (castes combined), caste within species, and both species combined

T.G. Shelton, A.G. Appel / Journal of Insect Physiology 47 (2001) 213–224 219

220


Fig. 4. Plot of V˙CO2 data (from steady state trace sections only) for all castes of R. flavipes and C. formosanus (combining castes). Individual regression models are drawn for each species as well as both species combined.

(df=1; F⬍0.13; P=0.7236). ANCOVA comparison of our data to those of Lighton and Fielden (1995) demonstrated no significant difference (df=1; F=0.19; P=0.6616) between metabolic rate and termite body mass model and that of the arthropods as a whole. MR⫽3.157(⫾0.192)M0.861(±0.079)

(1)

Eq. (1) models the relationship between the metabolic rate (MR) of R. flavipes and C. formosanus and body mass (M), where MR is metabolic rate in µW at 25°C (Q10=2.5) assuming an RQ of 0.72 (Lighton and Fielden, 1995), and M is mass in grams.

4. Discussion Our data demonstrate the lack of DGC by any caste of either C. formosanus or R. flavipes. Acyclic CO2 release by the termites in this study may be a result of their individual confinement in a dry, CO2-free environment; however, these same experimental conditions have been used to positively identify the DGC in other social (Lighton 1988a, 1990; Lighton and Wehner, 1993; Lighton et al., 1993b; Vogt and Appel, 2000) and solitary arthropods (Lighton, 1988b; Lighton, 1991b; Lighton et al., 1993a; Duncan and Lighton, 1997; Appel, pers. commun.). Direct behavioral observations showed that C. formosanus and R. flavipes do not reach a state of rest, instead they move at a nearly continuous steady pace, pausing only for short periods of time. Thus, movement by these animals varies in duration rather than intensity in respirometry conditions. Active ventilation, i.e. pumping of the abdomen, was not observed in the video recordings. The hypotheses proposed for DGC evolution by arthropods have been, first, to reduce respiratory water loss, and second, to adapt to hypercapnic/hypoxic

environments (Levy and Schneiderman, 1966; Lighton, 1998). The subterranean nests of C. formosanus and R. flavipes are maintained at relatively high humidities (16– 67%; Sponsler and Appel, 1990), as are the foraging tunnels. Grube and Rudolph (1999a,b) have shown that R. santonensis De Feytaud exhibit social humidity control in their subterranean environments. Cuticular permeabilities of these termites range from 15.22 to 42.53 µg cm⫺2 h⫺1 mmHg⫺1 for C. formosanus, and 27.83 to 34.57 µg cm⫺2 h⫺1 mmHg⫺1 for R. flavipes (Sponsler and Appel, 1990). These values are similar to other arthropods living in mesic environments (Hadley, 1994), although the value for C. formosanus soldiers (15.22 µg cm⫺2 h⫺1 mmHg⫺1) is closer to that expected of a xerically adapted arthropod (Sponsler and Appel, 1990). Because these termite species live and forage in high humidity environments, there is probably little selective advantage to the DGC for limiting respiratory water loss. The second hypothesized driving force for DGC evolution is increased tracheal exchange in hypercapnic/hypoxic conditions (Lighton 1994, 1998). Anderson and Ultsch (1987) found that subterranean termites live and forage in hypercapnic (6–8% CO2)/hypoxic (12–14% O2) environments. Therefore, these insects might be expected to exhibit DGC CO2 release patterns. However, our data clearly show that these species do not exhibit DGC at ambient temperatures and cannot be induced to DGC at low temperature or after extended equilibration periods. Lighton (1994) discussed the requirements necessary for an arthropod to be capable of DGC or cyclic CO2 release. Of these requirements, foremost is the ability to open and close spiracles in a controlled manner. It is not known if C. formosanus and R. flavipes workers and nymphs are morphologically capable of spiracular control of ventilation. A number of possibilities exist that may explain the lack of DGC in C. formosanus and R. flavipes. The worker, soldier and nymph castes of subterranean termites are weakly sclerotized, and are not terminally differentiated (i.e. not adults). This low level of sclerotization may allow for cutaneous CO2 release which might mask the presence of DGC. A high rate of CO2 release from the cuticle should show smaller respiratory CO2 peaks above the raised (cuticular) baseline, if DGC were being hidden. However, the traces in Figs. 2 and 3 do not show smaller respiratory peaks. Also, DGC as defined by Kestler (1985) and Lighton (1994) is associated with adult tracheate arthropods. Arguments may be made that individual worker and nymph caste termites might exhibit DGC if they developed into true adults. However, DGC is well known from worker ants (terminally differentiated, but not true adults; Lighton and Feener, 1989; Lighton 1988a, 1990; Lighton and Wehner, 1993; Vogt and Appel, 2000), and the CFO-type was originally described by Schneiderman (1960) from moth pupae.


Coptotermes formosanus and R. flavipes moved almost continuously in the respirometry chamber. Acyclic CO2 release observed here may be the result of nearly continuous activity, as in autoventilation (Kestler 1971, 1991). However, DGC ventilation patterns have been measured in ants exhibiting low levels of activity (Lighton and Feener, 1989; Lighton 1988a, 1990; Lighton and Wehner, 1993). The presence of DGC patterns can vary with circadian patterns and physiological conditions. Hadley and Quinlan (1993) found that Romalea guttata Serville grasshoppers did not exhibit DGC events during a normal diurnal activity cycle, but only exhibited DGC when the animals were engaged in nocturnal ‘roosting’ behavior. The authors also noted that DGC cycling deteriorated in dehydrated R. guttata (Hadley and Quinlan, 1993). Sponsler and Appel (1990) estimated that 50% water loss in these two species resulted in death. While water loss values for our data (6–9%) do not approach a fatal value, it is not known what, if any, effect sublethal dehydration has on gas exchange strategies for these termites. The recordings of both C. formosanus and R. flavipes in this study were conducted between 08:00 and 19:00 hr, and do not address the possibility for periodicity of DGC events. Direct observation of termite behavior in the respirometry chamber demonstrated that intensity of movement does not change, leaving only duration of movement to have a potential influence on V˙CO2 in these species. No interaction was found between the residuals of the mass scaling relationship of V˙CO2 and the activity scores from steady state sections of the recordings in either species. This result suggests that duration of activity does not influence V˙CO2 in C. formosanus and R. flavipes when the effects of mass are removed from the analysis. There were no differences in V˙CO2 between colonies of both C. formosanus and R. flavipes. As a result, colony data (within species) were combined for mass scaling modeling and V˙CO2 results in Table 2. Lack of a colony effect in our V˙CO2 comparisons allows future authors to avoid the complications of using multiple colonies of C. formosanus and R. flavipes in comparisons involving V˙CO2. This is not surprising, as basic physiological parameters should vary on a species, rather than a colony, basis. There was a significant difference in CO2 release between the two species; R. flavipes produced CO2 at a rate about 1.5-times greater than C. formosanus. The differences in metabolic rate allow for calculation of possible differences in energy expenditure between these species. Both of these termites are xylophagous (Kofoid, 1934), and have a nearly infinite food source (Waller and LaFage, 1987), particularly where fallen timber is plentiful. To better understand energy use by termites, differences between solitary and social insects must be considered. For solitary animals, each individual has to

221

budget its time and energy into a variety of activities, including defense, reproduction, nest construction, and foraging. A solitary animal that is more efficient (i.e. using less energy to perform the same amount of work) than another in terms of energy use has a great advantage: it can spend less time foraging (and feeding) than its competitor. This allows the efficient animal more time to spend on the other activities described above and potentially reduces the risk of predation (Krebs and Davies, 1993). Activity of eusocial animals must be examined on a colony rather than an individual basis (Krebs and Davies, 1993; Sudd and Franks, 1987). In termites, most major activities are allocated among the different castes. Castes include reproductives, soldiers (defense), and workers who perform all other duties (nest and tunnel construction, foraging and food transfer, egg care, waste removal; Kofoid, 1934). The advantage that worker termites have over solitary animals is the tremendous number of individuals in a given colony (Sudd and Franks, 1987). While there are a variety of tasks to be completed, there are a number of termites that can complete the tasks simultaneously (Kofoid, 1934; see Sudd and Franks, 1987). Social animals, such as termites, can benefit from being more individually energy efficient (i.e. use less energy for basic maintenance) by allocating more energy into other tasks, such as reproduction, soldier production, and nest construction, than less efficient competitors. Termites that live either in (drywood termites) or near (subterranean termites) their food source (Kofoid, 1934) can forage almost continually (Stuart, 1967; Thorne, 1982). Because food is not a limiting factor (Waller and LaFage, 1987), there appears to be a continuous flow of energy into the colony. If total energy budgets and conversion efficiencies for all species in a comparison are the same, then excess energy should be available to the species with the lower metabolic rate; the only question is how this surplus energy is used. Comparisons between C. formosanus and R. flavipes have been reported in a number of studies. Coptotermes formosanus are far more numerous (per colony; Grace, 1992), maintain much greater soldier proportions (10– 20% for C. formosanus, 8–10% for R. flavipes; Haverty, 1977), have larger foraging areas [per colony; 3571 m2 for C. formosanus (Su and Scheffrahn, 1988) and 1091 m2 for R. flavipes (a ratio of 3.27:1; Grace et al., 1989)], and interestingly have greater foraging rates (per termite; a ratio of 1.65:1, Delaplane and LaFage, 1990; see also Smythe and Carter, 1970; Grace, 1992) than R. flavipes. These are not predictions based on their lower metabolic rates, but known values for the two species. Our data show that C. formosanus has a lower metabolic rate than R. flavipes. Using our data and Lighton and Fielden’s (1995) assumptions (Q10=2.5, RQ=0.72), the mean metabolic rate for C. formosanus is 10.36±0.43 µW, and

222


17.62±0.48 µW for R. flavipes (a ratio of 1.70:1). Our behavioral data show that C. formosanus moves faster than R. flavipes (1.78:1 ratio) under respirometry conditions. The differences in the metabolic rates for C. formosanus and R. flavipes suggest that C. formosanus may be a more energy efficient species, and may be able to convert more energy in the colony into increases in defense, reproduction, etc., as outlined above. Mean metabolic rates for these two species differ by 苲7 µW. A number of possibilities may explain this difference in metabolic rate. Although movement was tracked during these experiments, perhaps R. flavipes individuals were more stressed (Kestler, 1991) and did not increase movement, but increased V˙CO2. However, the direct behavioral observations contradict this idea, as termites continued to move even after steady state for V˙CO2 was reached. Potential differences between species in terms of metabolically active tissue may also affect V˙CO2. Workers of both genera are morphologically similar, while soldier caste individuals of C. formosanus have a defensive secretion reservoir not found in R. flavipes soldiers (Quennedey and Deligne, 1975). This reservoir increases the proportion of non-metabolizing tissue in soldiers of C. formosanus, but not for workers, nor any caste of R. flavipes. Mean V˙CO2 values for C. formosanus soldiers are lowest of all castes and species examined (Table 2). Lipid content, tracheal system and/or alimentary canal sizes may differ between these species, which could also alter the proportion of metabolically active tissue between these species, thus affecting V˙CO2. Models relating V˙CO2 and termite body mass explained between 22.8 and 42.6% of the variation in V˙CO2 (Table 3). Mass ranges among castes were small and never exceeded an order of magnitude (Table 3). Small mass ranges resulted in limited predictions and probably the lack of significant mass scaling coefficients for most regressions (Table 3). The relatively narrow mass ranges may also help to explain the lack of a significant caste by species interaction for V˙CO2 values. For both species, individuals of all castes occur over a nearly continuous mass range (Table 3). If the castes within a species scale V˙CO2 with mass equivalently, then when the mass effect is removed (i.e. conversion of V˙CO2 to ml CO2 g⫺1 h⫺1), there should be no differences among them. This would also indicate that ratios of actively respiring tissue among castes within species were relatively equivalent. Our data suggest that for C. formosanus soldiers (defensive secretion reservoirs) and R. flavipes nymphs (fat tissue for flight and initial nest construction metabolism), there is a reduction in actively respiring tissue ratios compared with the other castes in each species. Lighton and Fielden (1995) presented an equation illustrating the mass scaling of a number of arthropods. Their consensus mass scaling coefficient for the arthropods (not including the Ixodida) was 0.825±0.034. In the

combined species plot of log10 V˙CO2 on log10 mass (Fig. 4), the overall coefficient of mass scaling for C. formosanus and R. flavipes is 0.832±0.0686. Some social insects may pose problems for the determination of the SMR as defined by Withers (1992); one of these problems is movement. Our data indicate that movement of various castes of C. formosanus and R. flavipes does not influence CO2 release. Mass scaling of V˙CO2 for these species does not significantly differ from that of other arthropods (Lighton and Fielden, 1995). It is expected that mass scaling of metabolic rate during periods of inactivity of C. formosanus and R. flavipes would have a lower intercept, but the same slope as our model. Our data indicate that near-continuous movement in these species is the ‘normal’ state of affairs. Speakman et al. (1993) discuss the consideration of the realistic behavior of animals in relation to measuring metabolic rates. They suggest that methods for measuring metabolic rates should reflect the ‘normal’ activity level of the animals studied. Our data have shown that even during steady state CO2 release, these animals do not cease their movement. While this may be the ‘normal’ state of affairs for these species, the metabolic rates reported here can only be considered active metabolic rate (AMR). However, AMR may approximate SMR for species that do not normally cease movement (Speakman et al., 1993). Our data indicate that body mass scaling of subterranean termite metabolic rates can be represented by a model that is not significantly different from that of arthropods as a whole. Additional work is needed in defining the SMR for these termites, as well as investigations into the presence of the DGC within the other major families of the Isoptera. Acknowledgements We thank Lane M. Smith II and Mark West for statistical advice. We are grateful to Brian T. Forschler, Nannan Liu, Mary T. Mendonc¸a, Faith M. Oi, J.T. Vogt, Barbara Joos, and three anonymous reviewers for helpful comments and suggestions on early versions of this manuscript. We also thank Marla J. Tanley for expert technical assistance during this study. The help of NanYao Su and Paul M. Ban in providing C. formosanus for this study is greatly appreciated. References Anderson, J.F., Ultsch, G.R., 1987. Respiratory gas concentrations in the microhabitats of some Florida arthropods. Comparative Biochemistry and Physiology 88A, 585–588. Buck, J., 1958. Cyclic CO2 release in insects. IV. A theory of mechanism. Biological Bulletin of Woods Hole Institute 114, 118–140.


Delaplane, K.S., LaFage, J.P., 1990. Wood excavations of three species of subterranean termites. Entomologia Experimentalis et Applicata 54, 81–87. Duncan, F.D., Lighton, J.R.B., 1997. Discontinuous ventilation and energetics of locomotion in the desert-dwelling female mutillid wasp Dasymutilla gloriosa. Physiological Entomology 22, 310– 315. Glantz, S.A., Slinker, B.K., 1990. Primer of Applied Regression and Analysis of Variance. McGraw-Hill, New York. Grace, J.K., 1992. Termite distribution, colony size, and potential for damage. In: Robinson, W.H. (Ed.), Proceedings of the National Conference on Urban Entomology, College Park, MD, pp. 67–76. Grace, J.K., Abdallay, A., Farr, K.R., 1989. Eastern subterranean termite (Isoptera: Rhinotermitidae) foraging territories and populations in Toronto. Canadian Entomologist 121, 551–556. Grube, S., Rudolph, D., 1999a. Water supply during building activities in the subterranean termite Reticulitermes santonensis De Feytaud (Isoptera Rhinotermitidae). Insectes Sociaux 46, 192–193. Grube, S., Rudolph, D., 1999b. The labial gland reservoirs (water sacs) in Reticulitermes santonensis (Isoptera: Rhinotermitidae): studies of the functional aspects during microclimatic moisture regulation and individual water balance. Sociobiology 33, 307–323. Hadley, N.F., 1970. Micrometeorology and energy exchange in two desert arthropods. Ecology 51, 434–444. Hadley, N.F., 1994. Water Relations of Terrestrial Arthropods. Academic Press, San Diego, CA. Hadley, N.F., Quinlan, M.C., 1993. Discontinuous carbon dioxide release in the Eastern lubber grasshopper Romalea guttata and its effect on respiratory transpiration. Journal of Experimental Biology 177, 169–180. Haverty, M.I., 1977. The proportion of soldiers in termite colonies: a list and bibliography (Isoptera). Sociobiology 2, 199–216. Kestler, P., 1971. Die diskontinuierliche Ventilation bei Periplaneta americana L. und anderen Insekten. Thesis, Julius-MaximiliansUniversita¨t, Wu¨rzburg. Kestler, P., 1985. Respiration and respiratory water loss. In: Hoffman, K.H. (Ed.), Environmental Physiology and Biochemistry of Insects. Springer-Verlag, Berlin, pp. 137–183. Kestler, P., 1991. Cyclic CO2 release as a physiological stress indicator in insects. Comparative Biochemistry and Physiology 100C, 207–211. Kofoid, C.A., 1934. Termites and Termite Control. University of California Press, Berkeley, CA. Krebs, J.R., Davies, N.B., 1993. An Introduction to Behavioral Ecology. Blackwell Scientific Publications, Oxford, 482pp. Levy, R.I., Schneiderman, H.A., 1966. Discontinuous respiration in insects—II. the direct measurement and significance of changes in tracheal gas composition during the life cycle of silkworm pupae. Journal of Insect Physiology 12, 83–104. Lighton, J.R.B., 1988a. Discontinuous CO2 emission in a small insect, the Formicine ant Camponotus vicinus. Journal of Experimental Biology 134, 363–376. Lighton, J.R.B., 1988b. Simultaneous measurement of oxygen uptake and carbon dioxide emission during discontinuous ventilation in the tok-tok beetle, Psammodes striatus. Journal of Insect Physiology 34, 361–367. Lighton, J.R.B., Feener, D.H., 1989. A comparison of energetics and ventilation of desert ants during voluntary and forced locomotion. Nature 342, 174–175. Lighton, J.R.B., 1990. Slow discontinuous ventilation in the Namib dune-sea ant Camponotus detritus (Hymenoptera Formicidae). Journal of Experimental Biology 151, 71–82. Lighton, J.R.B., 1991a. Insects: measurements. In: Payne, P.A. (Ed.), Concise Encyclopedia of Biological and Biomedical Measurement Systems. Pergamon Press, Oxford, pp. 201–208. Lighton, J.R.B., 1991b. Ventilation in Namib desert tenebrionid

223

beetles: mass scaling and evidence of a novel quantized flutterphase. Journal of Experimental Biology 159, 249–268. Lighton, J.R.B., 1994. Discontinuous ventilation in terrestrial insects. Physiological Zoology 67, 142–162. Lighton, J.R.B., 1996. Discontinuous gas exchange in insects. Annual Review of Entomology 41, 309–324. Lighton, J.R.B., 1998. Notes from underground: towards ultimate hypotheses of cyclic, discontinuous gas-exchange in tracheate arthropods. American Zoologist 38, 483–491. Lighton, J.R.B., Wehner, R., 1993. Ventilation and respiratory metabolism in the thermophilic desert ant, Cataglyphis bicolor (Hymenoptera Formicidae). Journal of Comparative Physiology B 163, 11–17. Lighton, J.R.B., Fielden, L.J., Rechav, Y., 1993a. Discontinuous ventilation in a non-insect, the tick Amblyomma marmoreum (Acari, Ixodidae): characterization and metabolic modulation. Journal of Experimental Biology 180, 229–245. Lighton, J.R.B., Fukushi, T., Wehner, R., 1993b. Ventilation in Cataglyphis bicolor: regulation of carbon dioxide release from the thoracic and abdominal spiracles. Journal of Insect Physiology 39, 687–699. Lighton, J.R.B., Fielden, L.J., 1995. Mass scaling of standard metabolism in ticks: a valid case of low metabolic rates in sit-and-wait strategists. Physiological Zoology 68, 43–62. Lighton, J.R.B., Fielden, L.J., 1996. Gas exchange in wind spiders (Arachnida, Solphugidae): independent evolution of convergent control strategies in solphugids and insects. Journal of Insect Physiology 42, 347–357. Lighton, J.R.B., Garrigan, D., 1995. Ant breathing: testing regulation and mechanism hypotheses with hypoxia. Journal of Experimental Biology 198, 1613–1620. Minitab, INC., 1994. Minitab Reference Manual: Release 10 for Windows. Minitab Publications, State College, PA. Nunes, L., Bignell, D.E., Lo, N., Eggleton, P., 1997. On the respiratory quotient (RQ) of termite (Insecta: Isoptera). Journal of Insect Physiology 43, 749–758. Quennedey, A., Deligne, J., 1975. L’arme frontale des soldats de termites. I. Rhinotermitidae. Insectes Sociaux 22, 243–267. SAS Institute, INC., 1985. SAS User’s Guide: Statistics. SAS Institute, Inc, Cary, NC. Schneiderman, H.A., 1960. Discontinuous respiration in insects: role of the spiracles. Biological Bulletin of Woods Hole Institute 119, 494–528. Schneiderman, H.A., Williams, C.M., 1955. An experimental analysis of the discontinuous respiration of the Cecropia silkworm. Biological Bulletin of Woods Hole Institute 109, 123–143. Smythe, R.V., Carter, F.L., 1970. Feeding responses to sound wood by Coptotermes formosanus, Reticulitermes flavipes, and R. virginicus (Isoptera: Rhinotermitidae). Annals of the Entomological Society of America 63, 841–847. Smythe, R.V., Williams, L.H., 1972. Feeding and survival of two subterranean termite species at constant temperatures. Annals of the Entomological Society of America 65, 226–229. Speakman, J.R., McDevitt, R.M., Cole, K.R., 1993. Measurement of basal metabolic rates: don’t lose sight of reality in the quest for comparability. Physiological Zoology 66, 1045–1049. Sponsler, R.C., Appel, A.G., 1990. Aspects of water relations of the Formosan and Eastern subterranean termites (Isoptera: Rhinotermitidae). Environmental Entomology 19, 15–20. Stuart, A.M., 1967. Alarm, defense, and construction behavior relationships in termites (Isoptera). Science 156, 1123–1125. Su, N.-Y., Scheffrahn, R.H., 1988. Foraging population and territory of the Formosan subterranean termite (Isoptera: Rhinotermitidae) in an urban environment. Sociobiology 14, 353–359. Sudd, J.H., Franks, N.R., 1987. The Behavioral Ecology of Ants. Chapman and Hall, New York, 206pp. Tamashiro, M., Fujii, J.K., Lai, P.-Y., 1973. A simple method to

224


observe, trap, and prepare large numbers of subterranean termites for laboratory and field experiments. Environmental Entomology 2, 721–722. Thorne, B.L., 1982. Termite–termite interactions: workers as an agonistic caste. Psyche 89, 133–150. Thorne, B.L., Carpenter, J.M., 1992. Phylogeny of the Dictyoptera. Systematic Entomology 17, 253–268. Vogt, J.T., Appel, A.G., 2000. Discontinuous gas exchange in the fire ant, Solenopsis invicta Buren: caste differences and temperature effects. Journal of Insect Physiology 46, 403–416. Waller, D.A., LaFage, J.P., 1987. Nutritional ecology of termites. In: Slansky, F. Jr., Rodriguez, J.G. (Eds.), Nutritional Ecology of Insects, Mites, Spiders, and Related Invertebrates. Wiley-Interscience, New York, pp. 497–532.

Wasserthal, L.T., 1996. Interaction of circulation and tracheal ventilation in holometabolous insects. Advances in Insect Physiology 26, 297–351. Wheeler, G.S., Tokoro, M., Scheffrahn, R.H., Su, N.-Y., 1996. Comparative respiration and methane production in nearctic termites. Journal of Insect Physiology 42, 799–806. Williams, A.E., Rose, M.R., Bradley, T.J., 1997. CO2 release patterns in Drosophila melanogaster: the effect of selection for desiccation resistance. Journal of Experimental Biology 200, 615–624. Withers, P.C., 1992. Comparative Animal Physiology. Saunders College Publishing, Fort Worth, TX.