Gas exchange patterns for a small, stored-grain insect pest, Tribolium castaneum

Abstract

Insects breathe using one or a combination of three gas exchange patterns; continuous, cyclic and discontinuous, which vary in their rates of exchange of oxygen, carbon dioxide and water. In general, there is a trade-off between lowering gas exchange using discontinuous exchange that limits water loss at the cost of lower metabolic rate. These patterns and hypotheses for the evolution of discontinuous exchange have been examined for relatively large insects (>20 mg) over relatively short periods (<4 h), but smaller insects and longer time periods have yet to be examined. We measured gas exchange patterns and metabolic rates for adults of a small insect pest of grain, the red flour beetle, Tribolium castaneum (Coleoptera: Tenebrionidae), using flow-through respirometry in dry air for 48 h. All adults survived the desiccating measurement period; initially they used continuous gas exchange, then after 24 h switched to cyclic gas exchange with a 27% decrease in metabolic rate, and then after 48 h switched to discontinuous gas exchange with increased interburst duration and further decrease in metabolic rate. The successful use of the Qubit, a lower cost and so more common gas analyser, to measure respiration in the very small T. castaneum, may prompt more flow-through respirometry studies of small insects. Running such studies over long durations may help to better understand the evolution of respiration physiology and thus suggest new methods of pest management.

Introduction

Insects breathe using one or a combination of three distinct forms of gas exchange based on spiracular control patterns: continuous, cyclic and discontinuous (Marais et al., 2005; Chown et al., 2006; Terblanche and Woods, 2018). Continuous gas exchange (CGE), the ancestral condition (Marais et al., 2005), occurs when the spiracles remain open to support continuous exchange of gases (oxygen into and carbon dioxide out of the body). Cyclic gas exchange consists of two phases, burst (spiracles open) and interburst (spiracles partially closed temporally and spatially), which produce periodic and often erratic cycles of gas exchange, but the spiracles never close completely in the interburst (Nespolo et al., 2007). Discontinuous gas exchange (DGE; Lighton, 1996; Quinlan and Gibbs, 2006; Matthews, 2018) has three distinct phases, with spiracles completely closed (C), fluttering (F) and open (O); the interburst phase includes (C) and (F).

The evolutionary origins of DGE are unclear, and the factors promoting it are highly debated. Eight adaptive and two non-adaptive hypotheses have been proposed to explain the role and evolution of DGE, which vary between taxonomic groups and different habitats. Although none of these hypotheses has unequivocal support (reviewed by Terblanche and Woods, 2018), three adaptive hypotheses have more support than others. The hygric hypothesis suggests that DGE reduces respiratory water loss (during the closed and flutter phase) for insects experiencing water stress, such as in arid habitats (Buck et al., 1953; Lighton, 1996). The chthonic hypothesis proposes that DGE improves O₂ and CO₂ exchange in hypoxic and/or hypercapnic habitats, such as underground (Lighton and Berrigan, 1995). The oxidative damage hypothesis argues that DGE reduces tissue damage during periods of low metabolic demand by hyperoxia in tissues of insects with a respiratory system adapted for high metabolic rates (Hetz and Bradley, 2005).

These gas exchange patterns and metabolic rate have been measured for more than 150 insect species (Marais et al., 2005; Terblanche and Woods, 2018). Most studies of insect gas exchange patterns and metabolic rates use flow-through respirometry (Withers, 2001; Lighton, 2018), with metabolic rate measured as the rate of CO₂ emission ($\dot{V}_{{\rm C}{\rm O}_ 2}$) rather than O₂ consumption ($\dot{V}_{{\rm O}_ 2}$) because it is technically more precise to measure small changes in CO₂ emission compared to O₂ consumption (e.g. Lighton, 1996, 2018). Metabolic rates measured as $\dot{V}_{{\rm C}{\rm O}_ 2}$ and $\dot{V}_{{\rm O}_ 2}$ differ depending on the metabolic substrate, which can be expressed in energy units (e.g. joules per hour) using appropriate conversion coefficients (e.g. Withers, 1992). Generally, these studies have examined the effects of a variety of factors, such as age/instar, sex, activity, absorptive state, temperature, nutrition level and hydration status, on gas exchange patterns and metabolic rate (Contreras and Bradley, 2009, 2010; Schimpf et al., 2012; Rolandi et al., 2014).

There are nevertheless gaps in our knowledge of respiration for insects. First, more than 150 insect species with information are from seven orders, thus there are another 23 insect orders without information (Marais et al., 2005; Terblanche and Woods, 2018). Second, almost all insects measured for gas exchange patterns are large, from 20 to 26,000 mg, and there is little to no information for small insects (Marais et al., 2005; Woodman et al., 2007). Third, the duration of experiments, in general, is short (2–10 h; Terblanche et al., 2010; Huang et al., 2014). There are a few studies where insects have been measured for periods longer than 10 h; all species are from relatively damp habitats. Three are small dipterans: one vinegar fly, Drosophila melanogaster (up to 36 h; Williams et al., 1997; Williams and Bradley, 1998), and two mosquitoes, Anopheles gambiae and A. arabiensis (up to 20 h; Gray and Bradley, 2005). A recent example had Madagascan hissing cockroaches (Gromphadorhina portentosa) being recorded for 23 h in each of two treatments for a total of 46 h (Rowe et al., 2022). Finally, besides metabolic rate measurements of insect grain pest species (Emekci et al., 2002; Pimentel et al., 2007; Arnold et al., 2016), not a single species has been measured for gas exchange patterns by flow-through respirometry. Consequently the current knowledge of insect breathing patterns and metabolic rates does not represent the majority of insect species, and may not be particularly relevant to small grain-pest species.

The objective of our study was to measure patterns of gas exchange and metabolic rate for a small grain-pest insect. Small insects have a high metabolic rate and mass-specific evaporative water loss (Hadley, 1994; Fields and White, 2002) and stored grain is generally a dry environment (Jian and Jayas, 2012), which may constrain their pattern of gas exchange, and so provide evidence for the hygric hypothesis for DGE. We studied adults of the red flour beetle Tribolium castaneum (Coleoptera, Tenebrionidae – darkling beetles) as they are small and inhabit stored grain, thus results should inform possible function of DGE and its implications for CO₂ and H₂O loss. We used flow-through respirometry over an extended duration (48 h) and hypothesised that if the hygric hypothesis was relevant then we would observe increased reliance on DGE and reduction of ($\dot{V}_{{\rm C}{\rm O}_ 2}$) as these beetles desiccated over time. The results of our study will improve our understanding of small insect ecophysiology by adding to previous closed-system studies of metabolism of these pests (Emekci et al., 2002; Pimentel et al., 2007; Arnold et al., 2016), and understanding their patterns of gas exchange may suggest methods for pest management.

Materials and methods

Insects

A ‘wild type’ phosphine-susceptible strain of the red flour beetle (T. castaneum, MUWTC-5000) was obtained from Murdoch University (Professor Yonglin Ren). Beetles were provided with flour and yeast (12:1) in a plastic jar with vents in the lid for gas exchange (following Alnajim et al., 2019). Beetles were maintained in a constant temperature room at standard conditions of 25 ± 1°C, 12L:12D photoperiod, ambient gaseous conditions (21% O₂, 0.03% CO₂) and ambient RH of 60–80% (Emekci et al., 2002; Lu et al., 2009). Adult T. castaneum are a long-lived life stage and survive for months in dry conditions (grain storage), hence they presumably can experience desiccation for extended durations. The adults are easy to handle because they become immobile when touched.

Respirometry

Carbon dioxide emission was measured using standard flow-through respirometry at an ambient temperature of 25°C (following Abbas et al., 2020). Each adult beetle was selected randomly and irrespective of sex for respirometry. Air from a compressed air cylinder (BOC Gases, Canning Vale, WA, Australia) was used as a stable air source, with low and constant CO₂ and H₂O concentration; it was not scrubbed of these gases for experiments. Relative humidity (RH) of the compressed air was verified using a humidity probe (HMP113 Vaisala Corporation, Helsinki, Finland) to be 5–6%. Limitations of respirometric analysis precluded us from making measurements at high RH, equivalent to the maintenance conditions. A glass syringe barrel (1 ml volume) was used as a respirometry chamber for individual beetles. Two respirometry chambers, with stopcocks to switch air flow between them, were used. One contained the beetle (insect system, hereafter) whereas the other (baseline system, hereafter) was empty to periodically record baseline CO₂. Air flow was regulated at 25 ml min⁻¹ Standard temperature and pressure, dry (STPD) using a mass flow controller (AFC 2600 Aalborg, Orangeburg, NY, USA). A bubble flow meter (Gilian Gilibrator 2, Sensidyne, St. Petersburg, FL, USA) was used to calibrate the mass flow controller. A Vaisala HMP113 probe was used to measure the RH and temperature of the air leaving each chamber; the probe was RH calibrated using a DewPoint Generator DG-4 (Sable Systems International, Las Vegas, NV, USA) and temperature calibrated using a traceable mercury in glass thermometer (Australian Calibrating Services, Melbourne, Vic, Australia). A CO₂ analyser (S151 Qubit systems, Kingston, Ontario, CA, USA) measured the CO₂ concentration of the excurrent air. It was calibrated using CO₂ free air (using Sodasorb CO₂ absorbent, W. R. Grace & Co., Chicago, IL, USA) and a certified span gas (0.153% CO₂, BOC Gases). We have used this system to measure two insects of 500–800 mg (the speckled cockroach Nauphoeta cinerea and the darkling beetle Zophobas morio; Abbas et al., 2020); here we attempt to extend the Qubit analyser to insects of 2 mg.

We measured only adult, reproductive flour beetles (completely melanised cuticle) and expressed metabolic rate in mass-specific units (ul CO₂ g⁻¹ h⁻¹), as reproductive status and body mass significantly affect metabolic rate, but sex per se does not (Arnold et al., 2016). A single respirometry trial lasted for 48 h, during which the airflow was primarily through the insect channel. Each beetle was weighed to ±0.01 mg with a digital balance (ER-182A A&D Company Limited, Toshima-Ku, Tokyo, Japan) immediately before and after each respirometry trial. Each individual insect was placed in the respirometry system immediately after the initial weighing; the insect chamber was darkened throughout the respirometry trial to reduce activity. After an acclimation period of 5 h (following Emekci et al., 2002; Pimentel et al., 2007), the airflow was switched to the baseline channel for 40 min to record a start baseline, then to the insect system. Air flow was switched to the baseline chamber for 40 min periodically during the 48 h experiment; the insect's CO₂ emission was recorded in between baselines. Rate of CO₂ release was analysed at 5, 23, 48 h; it was important to measure $\dot{V}_{{\rm C}{\rm O}_ 2}$ close to a baseline period as there was long-term erratic baseline drift of the CO₂ analyser over time. Although we attempted to measure 30 beetles, baseline drift in the Qubit analyser prevented the data from 22 beetles from being useful. Here we present data for the eight adult flour beetles for which baseline drift was not problematic.

Data acquisition and analysis

Digital multimetres (Protek 506, Hung Chang, Seoul, Korea and Thurlby 1905a, Thurlby Electronics Ltd, Huntingdon Cambridgeshire, UK) were used to measure analogue voltage signals from the CO₂ analyser and Vaisala probe. These multimetres were connected via a USB port hub (UC2324, ATEN, North Ryde, Australia) to a desktop PC. Voltage signals were sampled every 0.2 s and converted to ppm for CO₂ concentration and ˚C for temperature using a custom-written Visual Basic program (VB6, written by PC Withers). Data were stored continuously in an Excel file during the respirometry trial. Acquired data were analysed with a laptop using an Excel spreadsheet (written by PC Withers and W Abbas). Raw values of CO₂ and temperature were first calibration corrected, then the rate of CO₂ emission $\dot{V}_{{\rm C}{\rm O}_ 2}$ (ml STPD g⁻¹ h⁻¹) was calculated based on the following equation from Withers (2001).

$$\dot{V}_{{\rm C}{\rm O}_2} = [ ( \dot{V}_{\rm e}\,\times \,F_{\rm e}{\rm C}{\rm O}_ 2) \,\hbox{-} \,( \dot{V}_{\rm i}\,\times \,F_{\rm i}{\rm C}{\rm O}_ 2) ] \,\times \,{\rm 60/}M_{\rm b}$$

where V̇ _i is the incurrent flow rate (STPD ml min⁻¹), V̇ _e is the excurrent flow rate of air (STPD ml min⁻¹), F _iCO₂ is the incurrent fraction of CO₂, F _eCO₂ is the excurrent fraction of CO₂ and M _b is the body mass (g). V̇ _e was calculated from V̇ _i, F _iCO₂ and F _eCO₂ (Withers, 2001); see Abbas et al. (2020) for detailed equations.

Mass loss was determined gravimetrically by difference between the initial and final mass, divided by average mass and the total experimental time (48 h). This mass loss approximates average evaporative water loss ($\dot{V}_{{\rm H}_ 2{\rm O}}$, mg g⁻¹ h⁻¹) over the experimental period (ignoring mass loss as excretion and CO₂). Unfortunately, we were unable to measure total evaporative water loss (EWL) separately for flour beetles as the Vaisala RH probe was not sufficiently sensitive to measure the low EWL (in absolute terms) of these small beetles, so we could not partition total EWL into respiratory and cuticular water loss. However, we could estimate the partitioning of total EWL into respiratory and cuticular water loss from $\dot{V}_{{\rm C}{\rm O}_ 2}$ (see Discussion).

Gas exchange patterns

The CO₂ emission trace was examined for each individual beetle at different time intervals to characterise its respiratory pattern. Each beetle used either CGE, cyclic or DGE, at different time intervals. CGE was clearly identified from CO₂ emission traces, which were always above baseline, and lacked any periodicity, whereas cyclic gas exchange had more periodic CO₂ concentrations that approached baseline. The CGE and cyclic rates of CO₂ emission were calculated as the average $\dot{V}_{{\rm C}{\rm O}_ 2}$over time. A beetle was considered to have DGE if its CO₂ emission trace regularly was at or close to baseline (C) and there was a discernable interburst period (Marais et al., 2005). For calculations of DGE $\dot{V}_{{\rm C}{\rm O}_ 2}$, both closed (C) and flutter (F) were combined as the inter burst phase (IB) as there was no clear distinction between them (Wobschall and Hetz, 2004). We analysed DGE cycles (n = 2–8) for each individual beetle to calculate $\dot{V}_{{\rm C}{\rm O}_ 2}$ and the duration (sec) for each phase, burst (open phase, here after called burst or B for consistency with interburst or IB) and the entire cycle (IB + B).

Statistical analyses

Metabolic rate was compared at different time intervals for beetles by general linear models using the nlme package (v.3.1-140, Pinheiro et al., 2019) in R (v.3.6.1, R Development Core Team, 2020). A simple model (gls function) with time as a fixed factor, and a complex model (lme function) including individual as a random intercept factor (to account for individual variation for repeats) were compared by analysis of variance (ANOVA). For both models, assumptions of normal distributions of residuals and homogeneity of variance were tested using Shapiro–Wilk normality and Levene's tests respectively (car package; Fox and Weisberg, 2019). Individual variability was significant (likelihood ratio χ²₅ = 13.05, df = 5, P ˂ 0.001), so the complex model was used for metabolic rate comparison. The ANOVA model output used the metabolic rate for the middle time interval (24 h, coded as T1) as the reference level for comparison with metabolic rates at initial (5 h, coded as T2) and final time intervals (48 h, coded as T3) to compare cyclic and DGE $\dot{V}_{{\rm C}{\rm O}_ 2}$ at 24 and 48 h with CGE at 5 h. Pattern characteristics of cyclic and DGE (B and IB phase duration and $\dot{V}_{{\rm C}{\rm O}_ 2}$ as well as total cycle duration) were compared at 24 and 48 h time intervals by paired t-test. Means are provided with standard error. For comparison, we quantified the effects of body mass and temperature (10–30˚C) on the IB phase duration for 21 species, using published data and that from this study (Supplementary table S1). We also compared the metabolic rates (CGE) of T. castaneum (at 5 h) with the allometric relationship for other tenebrionid beetles of varying sizes, and other similarly sized small dipterans and hymenopterans (Supplementary table S2) to determine the extent of variation in metabolic rate resulting from body mass.

Results

All beetles survived the 48 h respirometry experiments. They lost an average of 10.8 ± 2.52% of their initial body mass by the end of the respirometry trial, corresponding to an average evaporative water loss of 2.11 mg g⁻¹ h⁻¹ over 48 h.

The gas exchange pattern of beetles changed over time. Initially, they used CGE (at 5 h; fig. 1a) but switched to cyclic release of CO₂ at 24 h (interburst $\dot{V}_{{\rm C}{\rm O}_ 2}$ was low but did not reach zero; fig. 1b). Beetles used DGE at 48 h, with interburst $\dot{V}_{{\rm C}{\rm O}_ 2}$ even lower, for longer periods (fig. 1c).

Figure 1. Example of gas exchange patterns identified from the CO₂ emission of a red flour beetle during its exposure to desiccation/starvation in the flow-through respirometry chamber for 48 h at 25˚C. (a) At 5 h showing continuous gas exchange. (b) At 24 h showing cyclic release of CO₂. (c) At 48 h showing discontinuous gas exchange. The short sections of zero release at the beginning and end of each trace indicate baseline recordings from an empty chamber.

The transition of gas exchange from continuous to cyclic at 24 h was coupled with a 27% reduction of metabolic rate (t ₁₄ = 14.8, P ˂ 0.001, table 1). The DGE pattern at 48 h (fig. 1c) had a small but significant (t ₁₄ = 2.17, P = 0.048) reduction in metabolic rate (5.7%) compared to that at 24 h, and a slightly greater reduction of 31% compared to CGE at 5 h (table 1).

Table 1. Body mass (mg), gas exchange patterns changing from continuous (CGE) to cyclic and discontinuous (DGE) over time, characteristics of interburst (IB) and burst (B) phases, and the cycle duration at 24 and 48 h, and average $\dot{V}_{{\rm C}{\rm O}_ 2}$(metabolic rate, μl g⁻¹ h⁻¹) and evaporative water loss (EWL, mg g⁻¹ h⁻¹) of red flour beetles (n = 8) measured for 48 h at 25˚C

Values are mean ± SEM. Significant differences for interburst (IB) and cycle duration at 24 and 48 h are indicated by a superscript * or ^#, with P value. Average $\dot{V}_{{\rm C}{\rm O}_ 2}$of the beetles at 24 h was significantly different from 5 (*) and 48 h (^#).

The cycle duration increased significantly from 54.5.7 ± 3.06 s at 24 h to 66.1 ± 4.53 s at 48 h (t ₇ = 3.52, P = 0.010; table 1). The IB phase also increased significantly from 31.5 ± 3.21 s at 48 h compared to 21.8 ± 2.28 s at 24 h (t ₇ = 5.35, P = 0.001), being almost 47% of the cycle duration at 48 h compared to 40% at 24 h. The IB phase $\dot{V}_{{\rm C}{\rm O}_ 2}$ was reduced as the IB phase duration increased with time, contributing 17.5% to metabolic rate at 48 h compared to 31% at 24 h. The B phase duration did not change significantly between 24 and 48 h (t ₇ = 2.05, P = 0.080; table 1). The $\dot{V}_{{\rm C}{\rm O}_ 2}$ did not differ significantly between 24 and 48 h for B (t ₇ = 1.79, P = 0.117) or IB (t ₇ = 1.98, P = 0.088).

Allometry of interburst duration and metabolic rate for beetles

The allometric relationship for log (interburst duration; min) with log (mass; g) and log (T _a; °C) was 2.98 (±0.30) + 0.288 (±0.054) log (M) – 0.052 (±0.013) log (T _a), which was highly significant (F _33,2 = 23.7, P < 0.001). Both mass and T _a factors were significant, with higher mass increasing (t ₃₃ = 5.34, P < 0.001) and higher temperature decreasing (t ₃₃ = 4.11, P = 0.002) the IB duration. Q ₁₀-adjusting the temperature for all the data to 25˚C clearly shows the positive allometry of IB duration with mass (fig. 2).

Figure 2. Allometry of DGE interburst duration with body mass for Tribolium castaneum from our study and other species from literature data (species are labelled from 1–21, with Tribolium castaneum from our study labelled as 1 with other Coleoptera from literature as 2–10, Hymenoptera as 11, Blattodea as 12–17 and Orthoptera as 18–21; see Supplementary table S1 for details) after Q ₁₀-adjusting the temperature for all the data to 25˚C.

After adjusting metabolic rate to an ambient temperature of 25˚C (assuming a Q ₁₀ of 2; Lighton, 2018; Supplementary table S2), we found that the scaling relationship between metabolic rate and body mass of the compared species (fig. 3) was significant (log (metabolic rate, MR; μl CO₂ h⁻¹) = 0.79 (±0.026) log (M; g), F _17,1 = 860, P < 0.001). The r ² value of 0.98 indicates that only 2% variation in metabolic rate between species was not explained by differences in body mass.

Figure 3. Allometry of CGE metabolic rate for Tribolium castaneum with tenebrionid beetles (black circles) and small species from several other insect orders (grey circles; see Supplementary table S2).

Discussion

This is the first study to measure metabolic rate concurrently with gas exchange patterns for a small insect that is a pest of stored grain, in a flow-through respirometry system. The only other similar-sized insects measured for breathing patterns, water loss and/or metabolic rates are dipterans. Drosophilia melanogaster adults survived for just 10 h in similar desiccating conditions, with an apparent CGE gas exchange pattern (Williams et al., 1997: fig. 2b). Two species of mosquito, Anopheles arabiensis and A. gambiae, survived 5–20 h (age dependent) in 10% RH still air (fig. 1 of Gray and Bradley, 2005) i.e. not in a respirometry chamber with air flow, which would increase EWL and so likely decrease survival time. These three species live in moist habitats and consume moist food, suggesting that the drier habitat and food of T. castaneum has been a factor in their evolution of a strong desiccation resistance.

Gas exchange patterns

Tribolium beetles transitioned from CGE to cyclic and DGE over time, with a significant reduction in metabolic rate. Beetles experienced both starvation and desiccation during measurement, which makes it difficult to separate the relative roles of each potential stress to the transition of the gas exchange pattern from CGE to DGE. However, Tribolium beetles are resistant to starvation, capable of surviving at least ~11 days when starved even at 34˚C and 60% RH (Scharf et al., 2015), so the effect of desiccation may be stronger. Unfortunately, as we could not weigh beetles throughout the experiment (which would require stopping the respirometry experiment at each time point), we were not able to measure EWL over time to see if there was reduction in EWL with or without DGE, to provide a test of hygric hypothesis (Buck et al., 1953; Lighton, 1996). This approach would be useful for future work.

DGE has been observed for adults of many insect species, ranging in body mass from 2.5 mg for worker fire ants (Solenopsos invicta, Vogt and Appel, 2000) that are around 40% heavier than T. castaneum, to 26 g for the giant burrowing cockroach (Macropanesthia rhinoceros, Woodman et al., 2007) that is four orders of magnitude heavier than T. castaneum (1.78 mg). The IB duration increases with body mass increase, presumably reflecting the lower mass-specific metabolic rate of larger species (fig. 2). For example, the IB duration of M. rhinoceros (10 min) is 2.28-fold higher than for S. invicta (4.4 min) at the same temperature (20°C). Increased ambient temperature has a negative effect on IB duration, reflecting the higher metabolic rate at higher ambient temperature. For example, the IB duration of the Table Mountain cockroach (Aptera fusca; body mass 2.29 g) is 9.5-fold higher at 10˚C than 30˚C (Groenewald et al., 2013).

Resting metabolic rate

The metabolic rate of T. castaneum after the initial 5 h acclimation period (see Emekci et al., 2002; Pimentel et al., 2007) of 1209 μl g⁻¹ h⁻¹ measured in the current study was lower than previous measurements using closed-system respirometry of 1680 μl g⁻¹ h⁻¹ for 200 adults with food at 70% RH averaged over 1 h (at the same temperature of 25˚C; Emekci et al., 2002). Pimentel et al. (2007) measured a slightly higher rate of 2.30 μl h⁻¹ using closed-system respirometry for a group of 20 beetles, than our value of 2.15 μl h⁻¹. Our lower values for T. castaneum are likely attributable to our instantaneous measurement of food-deprived individual adults by flow-through respirometry, which would be expected to result in lower values than measurements of groups of individuals or closed-system measurement (Lighton, 2018). Tribolium castaneum is the smallest of all tenebrionid beetles measured and has the lowest absolute metabolic rate, which agrees well with its allometrically predicted metabolic rate (fig. 3, Supplementary table S2).

Metabolic rate varies in response to a range of factors other than body mass and temperature e.g. starvation and desiccation (Contreras and Bradley, 2009, 2010; Schimpf et al., 2012; Rolandi et al., 2014). The metabolic rate of T. castaneum was downregulated over time, presumably reflecting pattern transition from CGE to DGE, in addition to the combined stress of starvation and/or desiccation in our experiment as explained above. Although DGE might have a beneficial reduction of EWL, it has potentially a disadvantageous metabolic consequence.

Physiological responses to stress such as variation of metabolic rate, water loss and gas exchange patterns of insect pests, other than stored-grain pests, are well documented (see table 1 of Karise and Mänd, 2015). For example, the American cockroach (Periplaneta americana) has a decreased metabolic rate and water loss with the transition from DGE to CGE after exposure to phosphine (800 ppm; Woodman et al., 2008). Less is known for stored-grain pests such as T. castaneum; metabolic rate varies in response to controlled atmosphere treatments i.e. increased CO₂ levels and/or decreased O₂ levels (Emekci et al., 2002, 2004) but it is not known if these variations are linked with gas exchange patterns. However, we have found that T. castaneum beetles switch to DGE and down-regulate $\dot{V}_{CO_2}$ with desiccation/starvation stress.

Evaporative water loss

Red flour beetles survived 48 h at low RH with an average EWL of 2.11 mg g⁻¹ h⁻¹ despite their small mass hence high surface-to-volume, which is associated with a high mass-specific evaporative water loss. Their switch from CGE (5 h) to cyclic (24 h) then DGE (48 h) may provide a fitness benefit of reduced EWL in the low RH environment of dry stored grain i.e. support the hygric hypothesis. Unfortunately, we were unable to measure respiratory EWL (REWL) separately for 5, 24 and 48 h for flour beetles, to determine if it decreased during DGE, but we can estimate REWL from average $\dot{V}_{{\rm C}{\rm O}_ 2}$ (table 1) using the average ratio of 1 mg respiratory H₂O ml O₂⁻¹ (calculated for 30 insect species; Woods and Smith, 2010). Assuming a respiratory exchange ratio of 0.85 (midway between 0.7 for lipids and 1.0 for carbohydrate; calculated from Withers, 2001), REWL would have been 1.42 mg g⁻¹ h⁻¹ at 5 h, 1.04 mg g⁻¹ h⁻¹ at 24 h and 0.99 mg g⁻¹ h⁻¹ at 48 h (DGE); these estimates correspond to a 27% reduction for cyclic gas exchange (24 h) and 31% decrease for DGE (48 h), which by way of our estimation method are necessarily the same as % reductions in $\dot{V}_{{\rm C}{\rm O}_ 2}$. However, haemolymph P _CO2 is relatively constant during CGE but is cyclic during DGE, so the P _CO2 gradient between haemolymph and ambient air during the open phase of DGE is variable. For the hissing cockroach (G. portentosa) the haemolymph P _CO2 during CGE of about 2.0 kPa is similar to the mean P _CO2 during DGE of about 1.9 kPa (varying from about 1.3 to 2.5 kPa; Rowe et al., 2022) so the average EWL/$\dot{V}_{{\rm C}{\rm O}_ 2}$ could be similar for continuous and DGE. The lower REWL (reflecting changes in $\dot{V}_{{\rm C}{\rm O}_ 2}$) has a more modest effect on total EWL of 16% reduction at 24 h (cyclic) and 19% at 48 h (DGE). We can also estimate cutaneous EWL (CEWL) as 1.11 mg g⁻¹ h⁻¹, calculated as the difference between the total mass loss over the experimental duration of 2.26 mg g⁻¹ h⁻¹ and the average REWL estimated over the experimental duration of 1.15 mg g⁻¹ h⁻¹. Actual measurements of TEWL and partitioning of REWL and CEWL are required, to confirm upon these estimates of water loss.

Qubit S151 analyser for very small insects

High precision gas analysers are used for small animals, due to the low volumes of gas fluxes (Withers and Cooper, 2011). Consequently, respirometry studies on insects usually use high-precision gas analysers to characterise gas exchange patterns (Gray and Bradley, 2006). We have now shown that the low-cost Qubit analyser, normally used for larger vertebrates and insects heavier than 1 g (e.g. Robertson et al., 2017; Lailvaux et al., 2018) can be used successfully to characterise gas exchange patterns in T. castaneum with a mass of around 2 mg. There are issues with the system, most importantly baseline drift. Baseline drift can make around the data from 73% of replicates unusable for analysis. Even with this issue, using the lower cost, and more common Qubit analyser may promote more studies of insect respiration.

In conclusion, this is the first study to determine gas exchange patterns in a stored grain insect, with associated changes in metabolic rate. We believe that this information for Tribolium of a switch from CGE to cyclic then DGE with progressive stress from dehydration, and the consequent decrease in metabolic rate, may be of use in applied settings, such as pest management. Pest management of stored grain insects is currently almost entirely achieved by fumigation with toxic gases (e.g. phosphine), but there is considerable research into the potential use of unusual concentration of natural atmospheric gases (i.e. low oxygen, high carbon dioxide, etc.), known as controlled atmospheres (Navarro, 2012). We suggest that knowledge of the gas exchange patterns of pest insects coupled with their effects on metabolic rate and evaporative water loss may help to find new pest management methods.