The termination of discontinuous gas exchange cycles (DGCs) in severely dehydrated insects casts doubt on the generality of the hygric hypothesis, which posits that DGCs evolved as a water conservation mechanism. We followed DGC characteristics in the two density-dependent phases of the desert locust Schistocerca gregaria throughout exposure to an experimental treatment of combined dehydration and starvation stress, and subsequent rehydration. We hypothesized that, under stressful conditions, the more stressresistant gregarious locusts would maintain DGCs longer than solitary locusts. However, we found no phase-specific variations in body water content, water loss rates (total and respiratory) or timing of stress-induced abolishment of DGCs. Likewise, locusts of both phases re-employed DGCs after ingesting comparable volumes of water when rehydrated. Despite comparable water management performances, the effect of exposure to stressful experimental conditions on DGC characteristics varied significantly between gregarious and solitary locusts. Interburst duration, which is affected by the ability to buffer CO 2 , was significantly reduced in dehydrated solitary locusts compared with gregarious locusts. Moreover, despite similar rehydration levels, only gregarious locusts recovered their initial CO 2 accumulation capacity, indicating that cycle characteristics are affected by factors other than haemolymph volume. Haemolymph protein measurements and calculated respiratory exchange ratios suggest that catabolism of haemolymph proteins may contribute to a reduced haemolymph buffering capacity, and thus a compromised ability for CO 2 accumulation, in solitary locusts. Nevertheless, DGC was lost at similar hydration states in the two phases, suggesting that DGCs are terminated as a result of inadequate oxygen supply to the tissues.