Introduction

Sleep is an evolutionarily conserved behavior essential for diverse physiological functions, yet natural sleep duration varies enormously across species ([1]; [2]). These interspecific differences often reflect ecological pressures. Under resource limitations, a fundamental trade-off arises: time spent asleep cannot be spent foraging, making sleep and feeding mutually exclusive behaviors. Many insects and mammals forgo sleep when faced with acute starvation to search for food, underscoring a core conflict between sleep and feeding drives. Consistent with this, satiated animals tend to become quiescent post-feeding, whereas hunger strongly disrupts sleep ([3]). Thus, dynamic regulation of sleep and feeding in response to internal state and environmental conditions is thought to maximize fitness. However, the precise strategies by which animals tune these behaviors, particularly under extreme ecological conditions, remain poorly understood.

Environmental and physiological stressors reshape sleep behavior. Thermal stress inputs, such as extreme heat, activate neural circuits that disrupt sleep by triggering high sleep latency and sleep fragmentation ([4]; [5]). Paradoxically, accumulating evidence suggests that sleep itself can improve survival under such harsh conditions by conserving energy and maintaining homeostasis. Reduced locomotor activity and increased sleep duration have been shown to prolong survival during nutrient deprivation ([6]). A similar pattern has also been observed in the desert-adapted species Drosophila mojavensis: these flies maintain high sleep levels during food and water deprivation in contrast to Drosophila melanogaster and survive longer under such conditions, supporting the notion that elevated sleep fosters stress resilience and enhances resistance to resource scarcity ([7]). Related, in the nematode Caenorhabditis elegans, stress-induced sleep emerges after noxious stress exposure, and sleep-active neurons are essential for mediating sleep’s protective role for survival ([8]; [9]). Taken together, sleep functions as a conserved adaptive response for enhancing stress tolerance.

Nearly all studies of sleep and stress have focused on adulthood. However, developmental sleep is hypothesized to fulfill physiological functions that differ from those in maturity ([10]). Across phylogeny, juveniles consistently sleep more than adults, and this ontogenetic transition in sleep timing is accompanied by functional changes to support stage-specific developmental processes. Drosophila larvae exhibit prolonged quiescent periods driven by specific neuromodulatory and neuropeptidergic circuits, facilitating developmental-specific functions such as neurogenesis and neural circuit assembly ([11]; [12]). Similar sleep patterns and functions have been observed in mammals across the lifespan, where neonates spend a greater proportion of time in REM sleep, which gradually declines with age ([15]; [16]). Sleep is thus not a static behavior but a highly plastic state that is restructured throughout development to meet dynamically changing physiological demands. Despite this, it remains unclear how the genetic and neurobiological adaptations that promote sleep-based resilience in adults shape sleep strategies earlier in life. Experimental evolution studies in D.melanogaster demonstrate that selection for adult starvation resistance increases adult sleep and reduces adult feeding, while producing developmentally specified changes such as extended development, elevated larval feeding, with little effect on larval sleep, implying stage-specific regulation of these traits ([17]). Here, we use D.mojavensis to test whether developmental sleep is reorganized within a genetic background optimized for adult survival, and whether it is tuned in anticipation of extreme conditions later in life.

Results

Sleep is reduced in Drosophila mojavensis larvae

Desert-adapted flies exhibit increased sleep in adulthood as part of a survival strategy under harsh conditions. However, during earlier developmental larval stages, these same animals do not encounter the same degree of food scarcity or temperature stress ([18]; [19]). To investigate whether physiological and genetic adaptations in desert flies promote fixed behavioral strategies across the lifespan, we initially assessed D.mojavensis second instar larval (L2) sleep under various rearing and experimental temperatures (Figure S1A). We compared sleep architecture in three geographically segregated D.mojavensis subspecies (Baja, Wrigleyi, and Mojavensis) to that of wild-type strain D.melanogaster (Canton-S). When both species were reared and assayed at 25°C, D.mojavensis L2 slept substantially less than CS with similar locomotor activity, reflecting a reduced sleep phenotype not attributable to health or motor issues. D.mojavensis L2 sleep was also highly fragmented, characterized by markedly shorter sleep bouts (Figure 1A-D). We noticed in these experiments that at 25°C, the standard lab temperature for D.melanogaster, D.mojavensis exhibited delayed development, failing to align with CS. To determine the optimal rearing conditions across species, we measured the time (hours) required for the embryo to reach the L2 stage. We found that raising D.mojavensis’ rearing temperature to 29°C synchronized Baja’s developmental progression with the control (Figure S1B). To examine whether the reduced sleep phenotype observed in D.mojavensis L2 stemmed from species-specific thermal preferences, we reared each species at its optimal developmental temperature and then assessed L2 sleep at 25°C or 29°C. This experimental design allowed us to dissociate developmental and acute temperature shift effects. Under both conditions, D.mojavensis L2 consistently showed a reduction in total sleep and shorter sleep bouts than D.melanogaster, indicating that the short, fragmented sleep phenotype persists regardless of temperature (Figure 1E-L). Thus, desert-adapted fly strains sleep less during larval stages, in contrast to excess sleep observed in adulthood.

Figure 1.

To investigate whether dietary variances might have an impact on sleep, we assessed sleep in L2 raised on an Opuntia cactus-based medium, the natural host substrate of D.mojavensis, instead of standard agar-yeast food. We quantified sleep architecture in both the first (F1) and fifth (F5) generations after transitioning to the cactus diet to minimize potential short-term epigenetic effects resulting from laboratory exposures (Figure S2). In both generations, D.mojavensis L2 exhibited reduced total sleep and fragmented sleep structure relative to D.melanogaster, indicating that the short-sleep phenotype is retained even on the natural diet. In addition to D.mojavensis, comparable sleep phenotypes were also observed in D.arizonae, a closely related desert species in the repleta group that shares similar ecological niches. The similarity in sleep traits points to a conserved short-sleep profile for larvae across desert-adapted Drosophila lineages (Figure S3).

Intact sleep homeostasis and increased sleep efficiency D.mojavensis larvae

To determine whether D.mojavensis L2’s reduced, fragmented sleep reflects impaired homeostatic balance, we tested sleep rebound following sleep deprivation (SD). A high-intensity blue light pulse was intermittently delivered over a 3 hour window to disrupt sleep in D.melanogaster and D.mojavensis (Figure 2A). During the subsequent 3-hour recovery period, both sleep-deprived L2 species exhibited rebound sleep compared to non-deprived controls. In D.melanogaster, rebound was driven by upward trends in both sleep bout number and bout length; in D.mojavensis, rebound was driven by extension in sleep bout length (Figure 2B-C). As expected, waking activity was elevated during deprivation in both species, but during recovery it dipped slightly below baseline in melanogaster, consistent with compensatory suppression of activity to facilitate rebound sleep, whereas in mojavensis it returned to baseline levels (Figure 2D). The normal activity level after deprivation in D.mojavensis supports that observed differences in sleep parameters were not due to poor health. These results indicate that larval D.mojavensis have a normal homeostatic sleep response, analogous to that of D.melanogaster ([11]).

Figure 2.

We next tested whether D.mojavensis L2 compensate for reduced sleep quantity by enhancing sleep quality. We assessed sleep depth via arousal-threshold experiments, delivering 4-second light stimuli of low (1 Amp) and high (5 Amp) intensity every 2 minutes over a 2-hour window. We observed that D.mojavensis L2 were less likely to awaken than D.melanogaster at both stimulus intensities (Figure 2E, F), indicating a deeper sleep state. Activity levels both at baseline and in response the stimuli were similar between species, supporting that differences in arousal were sleep-specific rather than driven by disparities in light sensitivity (Figure 2E, F).

Feeding is a compensatory behavior for sleep across the lifespan in D.mojavensis

Given D.mojavensis L2’s limited sleep yet intact sleep need, we next asked whether their extra wake time is redirected toward an alternative adaptive behavior, specifically feeding. We hypothesized that D.mojavensis larvae may sacrifice sleep to increase nutrient intake and build energy reserves critical for long-term survival in their variable desert environment as adults. Using a dye-based absorbance feeding assay, we first quantified food consumption in L2. When a population of equal numbers of larvae were assessed, D.mojavensis groups consumed significantly more on average than D.melanogaster (Figure 3A), suggesting a shift in behavioral priority. To rule out the possibility of a body size confound, we measured weight across the life cycle and found that D.mojavensis L2 are half the body weight of D.melanogaster at this stage (Figure 3B), suggesting an underestimation of feeding disparities. To account for this scaling difference, we repeated the feeding assay under a weight-matched design by doubling the sample size of D.mojavensis L2 so that the total biomass matched the D.melanogaster group. The increased feeding indeed appeared even greater with this design (Figure 3C), confirming enhanced nutrient intake in D.mojavensis L2. We then asked whether this sleep-feeding tradeoff is limited to the L2 stage or maintained across larval development. In third instar larvae (L3), D.mojavensis likewise exhibited reduced and fragmented sleep compared to D.melanogaster (Figure 3D-G). These altered sleep parameters were again accompanied by an increase in food consumption (Figure 3H).

Figure 3.

In stark contrast to larval stages, D.mojavensis adults exhibit elevated and consolidated sleep compared to D.melanogaster. We thus investigated the sleep-feeding relationship at this stage using a liquid food interaction assay and found adult D.mojavensis engaged in significantly fewer feeding interactions, with reduced licking frequency, shorter feeding durations, and fewer feeding events compared to D.melanogaster adults (Figure 3I-K) ([20]). Together, these results suggest that D.mojavensis larvae may actively extend their wake time to engage in foraging and feeding, thereby conserving and storing energy early in development when metabolic demands are high, resources are abundant, and threats are minimal. In adulthood, scarce resources and unpredictable survival conditions shift D.mojavensis from a feeding-dominant to a sleep-dominant strategy, prioritizing energy conservation. To further explore the physiological basis of this shift, we measured triglycerides (TAG) energy stores across development. D.mojavensis accumulated more lipid reserves than D.melanogaster by the L3 stage and sustained into adulthood (Figure 3L). These findings reveal a life-stage-dependent reorganization of behavioral priorities in D.mojavensis, aligning sleep-feeding strategies with dynamically changing survival demands.

Differential starvation responses and enhanced survival in D.mojavensis L2

We next directly tested the functional consequences of D.mojavensis’s sleep and metabolic profile under acute food stress. In a food-deprived starvation paradigm, both D.mojavensis and D.melanogaster L2 showed the expected response of suppressing sleep relative to fed controls, consistent with hunger-induced foraging behavior. However, closer examination revealed that D.mojavensis L2 suppressed sleep to a lesser extent than D.melanogaster under starvation, including greater sleep duration, more sleep bouts, and longer bout length (Figure 4A-C). Thus, while both species suppress sleep in the absence of food, D.mojavensis L2 do not curtail sleep as drastically as D.melanogaster, perhaps reflecting shared adaptations with desert fly adults that promote starvation resistance.

Figure 4.

How do starvation response behavioral differences affect survival? Adult D.mojavensis are able to survive longer than D.melanogaster in the absence of food and even water ([7]). Under starvation conditions, D.mojavensis L2 also survived longer than D.melanogaster L2 at both 25°C (Figure 4E) and 29°C (Figure 4F), indicating enhanced resilience under stress. Notably, although adult D.mojavensis are starvation resistant, they do not suppress sleep in the absence of food as adult D.melanogaster do. This finding reinforces the concept of distinct, developmental stage-specific survival strategies in D.mojavensis. Larvae actively suppress sleep to maximize foraging opportunities during starvation, whereas adults apparently adopt a different approach, possibly conserving energy by maintaining sleep and relying on pre-accumulated energy reserves (consistent with the elevated larval and adult TAG levels) rather than active foraging.

Discussion

Here we report that in contrast to adulthood, the desert-adapted D.mojavensis reallocate time from sleep to feeding during developmental larval stages. These larvae exhibit reduced and fragmented sleep alongside markedly elevated feeding, a behavioral shift that correlates with enhanced starvation resilience. Adult D.mojavensis not only show excessive sleep but also feed less. Thus, the short-sleeping, food-seeking larvae and the long-sleeping adults represent dissociated phenotypes within the same species. This dissociation between life stages underscores the evolutionary importance of stage-specific energy budgeting. In a desert environment where necrotic cactus substrates are patchy and unpredictable, selection likely favored larvae that suppress sleep to exploit transient food resources and accumulate critical metabolic reserves for later development. The adults, no longer constrained by growth or feeding, can afford to sleep longer as a water- and energy-conserving strategy. This developmental compartmentalization highlights the ecological significance of plastic sleep architecture, with evolution tuning behavioral priorities across ontogeny in response to fluctuating environmental pressures ([15];[20]). Consistent with this ecological framework, our starvation assays provide empirical validation for the adaptive advantage conferred by the larval phenotypes observed in D.mojavensis. The ability to survive for more extended periods without food directly supports the hypothesis that reduced sleep, increased feeding, and heightened arousal threshold are integral components of a robust strategy for resilience in unpredictable environments. The suite of behavioral and metabolic adaptations in larvae (devoting wakefulness for foraging, efficiently utilizing resources, and moderating sleep loss under starvation) directly contributes to their ability to withstand periods of food scarcity.

The persistence of elevated arousal threshold (deeper sleep profile) across life stages suggests a conserved shift toward higher sleep efficiency ([7]). Though the absolute sleep need in larvae versus adults might differ, sleep remains essential for D.mojavensis, even if the underlying physiological functions differ between development and adulthood. At a mechanistic level, we propose that the short and fragmented sleep in D.mojavensis larvae is a genetically encoded adaptation that initiates, rather than responds to, downstream metabolic and developmental consequences. Despite a shared genome between larva and adult stages, larvae’s distinct sleep phenotype suggests that the genes regulating sleep are deployed in a stage-specific manner, likely shaped by developmental context and ecological necessity. If adult sleep elevation were driven by mutations in broadly acting sleep-promoting pathways, we might expect larval sleep to be similarly increased. The reduced and fragmented larval sleep instead implies either that larval and adult sleep are regulated by distinct genetic programs, or that D.mojavensis has evolved independent genetic control across development. This interpretation aligns with previous work showing that selection for adult sleep or starvation resistance in D.melanogaster does not produce parallel changes in larval behavior ([6]; [17]), supporting the idea that sleep is modularly regulated across ontogeny. This framework of stage-specific sleep regulation is reinforced by neurobiological evidence. In D.melanogaster, neuromodulators governing sleep differs across development ([11]). This organization suggests that evolutionary changes can adjust sleep at one developmental stage without necessarily disrupting sleep regulation or homeostasis at other stages. Evolutionary changes in neuromodulator expression and/or sleep-wake circuits across development could give rise to a short-sleeping larval phenotype that drives metabolic investment, and a long-sleeping adult phenotype that enhances stress resilience.

More broadly, our work contributes to a growing body of evidence that sleep is a flexible, ecologically shaped behavior. Across taxa, animals adjust sleep timing and duration to accommodate life-history events such as migration, breeding, and seasonal foraging. White-crowned sparrows reduce sleep dramatically during migration while maintaining cognitive performance, and Arctic sandpipers virtually eliminate sleep during mating season to maximize reproductive success ([21];[22]). In insects, male Drosophila suppress nighttime sleep to court females ([24];[25]). These examples show that animals can sacrifice sleep when ecological demands take priority. D.mojavensis larvae offer a developmental parallel as securing nutrients outweighs the need for sustained sleep at this stage. An important next step will be to identify the neurogenetic mechanisms that govern this reallocation. Given that neuromodulators such as dopamine, octopamine, and NPF are known to regulate sleep and feeding in adults, one hypothesis is that these systems are developmentally re-tuned in D.mojavensis larvae to permit prolonged wakefulness without impairing developmental progression ([26];[27]). Taken together, our findings reveal a developmentally reprogrammed, short-sleep phenotype in desert-adapted larvae, driven by ecological pressures and aligned with survival demands. This work advances the view that sleep is not static but a highly plastic state, reorganized across life stages in response to environmental and physiological priorities.

Methods and Materials

Adult Rearing

Fly stocks were cultured on standard cornmeal molasses media (per 1L H₂O: 12 g agar, 29 g Red Star yeast, 71 g cornmeal, 92 g molasses, 16mL methyl paraben 10% in EtOH, 10mL propionic acid 50% in H₂O) at 25°C with 60% relative humidity and entrained to a daily 12h light, 12h dark schedule. Canton-S (CS) were provided by Dr. Gero Miesenböck (University of Oxford). Primary stocks D.mojavensis.baja (collected March 2020, La Paz, Mexico), D. mojavensis.moja (collected February 2020, North Joshua Tree National Park, CA), D.mojavensis.wrigleyi (collected November 2017, Catalina Island, CA) were a gift from Dr. Luciano Matzkin (University of Arizona). D. arizonae were ordered from the National Drosophila Species Stock Center (SKU: 15081-1271.36).

Larval Rearing

All fly stocks were cultured on a standard molasses medium and raised in incubators on a 12:12 light:dark (LD) cycle. Different optimal raising temperatures were used across species to synchronize their developmental time: CS was maintained at 25℃; baja, moja, wrigleyi, and arizonae at 29℃. To collect synchronized L2, adult flies were placed in an embryo collection cage (Genesee Scientific, cat#: 59-100) and eggs were laid on a petri dish containing 3% sugar, 2% sucrose, and 2.5% apple juice with yeast paste on top.

Larval Body Weight Measurements

For weight, groups of 10 L2, 5 L3, pupae, or adults were washed in Milli-Q water and dried using a Kimwipe. The samples were then weighed as a group on a scale, and the weight in milligrams (mg) was recorded.

Sleep Assays and Starvation Assays

On the day of the sleep assay, molting first instar larvae (L1) were collected and moved to a separate dish to complete the molt to the L2. Freshly molted L2 were placed into individual wells of LarvaLodge containing 100 µl of 3% agar and 2% sucrose media covered with a thin layer of yeast paste. Experiments with Banana-Opuntia media used a recipe from the National Drosophila Species Stock Center (NDSSC; Cornell University): per 1L H2O: 14.16g agar, 27.5 g yeast, 2.23g methyl paraben, 137.5g blended bananas, 95g Karo Syrup, 30g Liquid Malt Extract, 22.33g 100% EtOH, 2.125g powdered opuntia cactus.

For starvation assays, 100 µl of 3% agar-only media was applied in each well. The LarvaLodge was covered with a transparent acrylic sheet brushed with a thin layer of detergent avoiding condensation, and moved into a DigiTherm incubator at either 25℃ or 29℃ for imaging in constant dark.

LarvaLodge Image Acquisition and Processing

Images were captured every 6 sec for 5 hrs with an Imaging Source DMK 23GP031 camera equipped with a Fujinon lens with a Hoya 49mm R72 Infrared Filter. We used IC Capture (The Imaging Source) to acquire time-lapse images. All experiments were carried out in the dark using infrared LED strips positioned below the LarvaLodge.

Images were analyzed using custom-written MATLAB software. Temporally adjacent images were subtracted to generate maps of pixel value intensity change. A binary threshold was set such that individual pixel intensity changes that fell below 40 gray-scale units within each well were set equal to zero (“no change”) to eliminate noise. Pixel changes greater than or equal to the threshold value were set equal to one (“change”). Activity was then calculated by taking the sum of all pixels changed between images. Sleep was defined as an activity value of zero between frames. Total sleep was summed for 3 hrs beginning 1 hr after the molt to L2.

Arousal Threshold, Light Responsiveness, and Sleep Deprivation via Blue Light Stimulation

Blue light stimulation was delivered using two high-power LEDs (Luminus Phatlight PT-121, 460 nm peak wavelength, Sunnyvale, CA) secured to an aluminum heat sink and driven at a constant current of 1 Amp. The blue light stimulus was delivered for 4 sec every 2 min for 1 hr beginning the 2^nd hr after the molt to L2. The number of larvae showing activity changes in response to the stimulus was counted, and the percentage of larvae that moved in response to the stimulus was recorded for each experiment.

For light responsiveness, we derived the arousal threshold data for further processing. We compared larval activity (pixel) 30 sec before each stimulus and 30 sec after each stimulus of all 30 light stimuli for individuals in the sleep state 6 sec before light stimulation and awake during the first light stimulation.

For sleep deprivation, we used a 5-Amp stimulus for 18 sec every 30 sec for 3 hrs beginning 1 hr after the molt to L2. Undisturbed control larvae were placed in a separate incubator under identical conditions without blue light exposure during the deprivation.

Feeding Behavior Analysis

Freshly molted L2 were placed in a petri dish containing blue-dyed 3% agar, 2% sucrose, and 2.5% apple juice with blue-dyed yeast paste on top for 4 hrs at 25°C in constant darkness. After 4 hrs, L2 were washed in water, transferred to microtubes, and frozen at -80°C overnight. 300 µl of distilled water and spun down for 5 min at 13,0000 rpm. The amount of blue dye in the supernatant was then measured using a spectrophotometer (OD₆₂₉). Food intake represents the OD value of each measurement.

Triglyceride (TAG) and Bicinchoninic Acid Protein (BCA) Assays

Triglyceride levels were measured using the Stanbio Triglyceride LiquiColor Kit (EMSCO/Fisher, 2100-430) from Drosophila adult flies, pupae, and larvae. The sample sizes for adults, pupae, and larvae were n = 5, n = 5, and n = 25, respectively. Samples were collected and washed in cold phosphate-buffered saline (PBS) and homogenized 3 x 10 sec each in 125 µL of lysis buffer (140 mM NaCl, 50 mM Tris-HCl [pH 7.4], 20% Triton X-100, and 1X protease inhibitors). The homogenates were centrifuged at 13,000 rpm for 10 min at 4°C, and the supernatants were stored at -20°C until analysis. TAG levels (µg/µL) were recorded according to the kit protocol and measured using a microplate reader at 500 nm. TAG data were normalized to protein levels measured at 562 nm, quantified using the Pierce BCA Protein Assay Kit (Cell Center/Thermo, PI23227).

Survival with Starvation Analysis

Freshly molted L2 were placed into individual wells of LarvaLodge containing 100 µl of 3% agar, transferred into a DigiTherm incubator at either 25℃ or 29℃, and imaged in constant dark for an infinite time until all L2 were observed and determined deceased. To prevent condensation, the transparent acrylic sheet covering the wells was replaced every 24 hours.

FLIC Feeding Assay

Fly Liquid Interaction Counter (FLIC; Sable Systems, North Las Vegas, NV) feeding assay was performed to obtain data on feeding duration and frequency previously described ([20]). 4-8 day old female flies were individually loaded into separate chambers of the Drosophila Feeding Monitors (DFM) to track feeding behavior. DFM feeding wells were loaded with a solution of 5.0% sucrose and 50 mg/L MgCl₂ in distilled water. Individual flies are aspirated into separate DFM chambers between ZT0-2 and behavior data is recorded for 24h. Feeding data includes measurements from two independent experiments. The DFM detects feeding signals from individual flies via a microcontroller circuit board on each well. Raw feeding signal data is coordinated and processed in the Master Control Unit (MCU). Feeding data was analyzed in R Studio using previously described analysis scripts (https://github.com/PletcherLab/FLIC_R_Code) ([28]).

Statistical Analysis

All statistical analysis was done in GraphPad (Prism). For comparisons between 2 conditions, two-tailed unpaired t-tests were used. For comparisons between multiple groups, ordinary one-way ANOVAs followed by Tukey’s multiple comparison tests were used. For comparisons between different groups in the same analysis, ordinary one-way ANOVAs followed by Sidak’s multiple comparisons tests were used.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. All raw data will be made available.

Acknowledgements

We thank members of the Kayser Lab for helpful discussions and input.

Additional information

Funding

NIH R35NS137329 (MSK)

Burroughs Wellcome Career Award for Medical Scientists (MSK)

Author contributions

Conceptualization: All authors

Investigation: SL, MS, CN, SHT

Writing – Original Draft: SL, MSK

Writing – Review and Editing: All authors

Project Supervision and Funding: MSK

Funding

HHS | National Institutes of Health (NIH) (R35NS137329)

• Matthew S Kayser

Burroughs Wellcome Fund (BWF)

• Matthew S Kayser