Feeding a diurnal rodent at night induces weight loss and body temperature reduction without food-anticipatory activity

Abstract

An important component of behavioral plasticity is the ability of animals to shift the timing of daily or seasonal activities to accommodate changing physical and biotic conditions, a phenomenon known as temporal niche switching. Temporal niche switching in response to experimentally altered food availability in the laboratory is often accompanied by locomotor activity that anticipates the time of daily food access (“food anticipatory activity” = FAA), which reflects the plasticity of the mammalian circadian system. Most FAA studies have been conducted in nocturnal rodents, reporting robust anticipatory increases in locomotor activity and physiological variables such as body temperature before food presentation, but this kind of plasticity has not been tested in strictly diurnal species. In the present study, we show that the diurnal antelope ground squirrel (Ammospermophilus leucurus) fails to exhibit FAA when subjected to temporally restricted food access. We centered 8 hours of food restriction either in the middle of the active/light phase (daytime feeding; DF) or the rest/dark phase (nighttime feeding; NF). In contrast to findings in nocturnal rodents, NF antelope ground squirrels did not show FAA, lost body mass, decreased overall wheel-running activity, and exhibited decreases in body temperature, with two animals becoming hypothermic. DF squirrels ate more than NF squirrels, had greater activity levels, and maintained normal body temperature. These results indicate that exclusive nighttime food availability was not sufficient to support normal energy balance in this strictly diurnal rodent. Thus, whereas negative energy balance can promote shifts toward daytime activity in some nocturnal mammals, AGS did not show the reciprocal shift toward nighttime activity when food was restricted to the dark phase. This suggests that, relative to nocturnal species, strictly diurnal mammals may have a limited capacity to compensate behaviorally and physiologically for food availability during the rest phase of the circadian cycle.

INTRODUCTION

The natural daily rhythms of physiology and behavior expressed by virtually all animals are sustained by endogenous timing mechanisms that allow organisms to anticipate predictable daily environmental cycles. In mammals, the suprachiasmatic nucleus (SCN) of the hypothalamus is the central circadian pacemaker and is synchronized by the natural light-dark (LD) cycle, the dominant cue synchronizing an individual’s physiology and behavior with the environment. Concurrently, animals can shift the timing of activity in response to nonphotic challenges (e.g., food availability, temperature, predation risk), a form of behavioral flexibility often described as temporal niche switching. With respect to feeding time, when access to food is experimentally limited to a fixed daily time window (“time-restricted feeding”, TRF), many species exhibit increased arousal and locomotor activity in anticipation of food presentation (“food anticipatory activity”, FAA), typically one to three hours beforehand. This FAA is considered a manifestation of circadian entrainment to TRF, as the FAA remains present even under conditions of constant illumination and persists under total absence of food. Remarkably, FAA also occurs in animals lacking a functional SCN or molecular clock (1,2), indicating that it is driven by extra-SCN pathways.

Most of the research on the effects of TRF on FAA has been conducted on nocturnal laboratory rodents: rats and mice. FAA emerges consistently in nocturnal rodents, with few exceptions (3,4), within just a few days when feeding is restricted to a daily time window, whether that window falls during the normal activity time (nighttime darkness) or the normal rest time (daytime light) (5,6). Such studies, which were followed up with investigations of the neural mechanisms and peripheral signals involved in FAA (7–10), have provided fundamental insights into synchronization of circadian rhythms by TRF, and have been interpreted as a demonstration of the power of cyclic nonphotic stimuli to synchronize daily rhythms and of the flexibility of the mammalian circadian system in general. However, being based on data from nocturnal mice and rats, these studies leave open the question of whether diurnal mammals can show comparable responses.

To date, no study has explicitly tested food anticipatory activity (FAA) in a strictly diurnal rodent species. However, FAA has been investigated in two rodent species that exhibit at least partially diurnal activity, the Sudanian grass rat (Arvicanthis ansorgei) and the degu (Octodon degus). Both species can shift their typically diurnal locomotor activity to the nighttime (dark) phase when food access is restricted to nighttime hours (11,12). However, under laboratory conditions, these two species often exhibit individually variable activity patterns, ranging between diurnal and nocturnal, or bimodal diurnal rhythms (13–16). For example, approximately 16% of individual Arvicanthis ansorgei display predominantly nocturnal activity in laboratory settings (15), and Octodon degus exhibit abrupt changes between diurnal, nocturnal, and bimodal patterns of activity (16,17). In non-rodents, evidence remains sparse, with diurnal squirrel monkeys (Saimiri sciureus) showing weak responses to TRF under constant light conditions; they have not yet been tested under nocturnal food restriction (18,19), and all other mammalian FAA studies likewise have relied on species that are not strictly diurnal models (20–22). This limitation is particularly significant in the quest for translational research, given that humans are diurnal, and meal timing markedly affects metabolic health, mood, and behavior. Because of these limitations on previous research, we have adopted a strictly diurnal species as a potential model to better address these questions of relevance to human health.

We have selected the antelope ground squirrel (AGS; Ammospermophilus leucurus), a strictly day-active rodent, as observed both naturally and in the laboratory (23–25), for a study of responses to time-restricted feeding. This species demonstrates pronounced, high-amplitude diurnal locomotor activity rhythms and maintains a stable intrinsic circadian period throughout the year (23,26). With an endogenous circadian period of ~24.2 hours, the AGS closely matches that of other diurnal mammals, including humans (27), in contrast to the shorter circadian periods typical of nocturnal rodents (28). We report here the behavioral and physiological responses of AGS exposed to schedules of TRF under a consistent light-dark (LD) cycle and with ad libitum food provision (“normocaloric feeding”) to eliminate the known effects of caloric restriction on behavior and physiology (29,30). We hypothesized that AGS would exhibit FAA under a daily TRF schedule, whether the time of food availability occurred in the middle of the light/daytime phase or the dark/nighttime phase of the daily photoperiod. Specifically, we predicted that AGS would increase locomotor activity and body temperature in the hours preceding the time of food availability and that the animals that received food restricted to the dark phase of the daily cycle would shift their locomotor activity to a time in the night that anticipated the experimentally restricted nocturnal presentation of food.

METHODS

Animal collection and general holding conditions

The 15 animals (4 females, 11 males) used in this study were originally captured in April 2021 in their natural desert habitat in Harney County, Oregon, using Sherman live traps (H. B. Sherman Traps, Inc., Tallahassee, Florida). They were transported to a laboratory facility at the University of Washington and housed in plastic cages (26 wide × 48 long × 15 cm deep) with running wheels (17.5 diameter × 7.5 cm width) and provided with corncob bedding, and maintained under a 12:12h LD cycle (~1500 lux daytime; red light ~1 lux nighttime, measured in middle of cage interior). Room temperature was maintained at 22.4±0.6 °C. Animals received ad libitum food (PicoLab Rodent Diet 20 LabDiet 5053) and water, and were maintained in these general conditions except when exposed to conditions of earlier experiments (Zhang et al. 2025) that concluded in October 2022. Following those experiments, the animals were again maintained under the above general conditions for 15 weeks prior to the beginning of the present study, which was concluded in late June 2023. All studies were approved by the University of Washington Office of Animal Welfare and Institutional Animal Care and Use Committee (protocol #4045–02). Squirrel captures were conducted under a Scientific Taking Permit (032–21) from the Oregon Department of Fish and Wildlife.

Behavioral and physiological measurements

Each cage was equipped with two movement-detection sensors. The first was a passive infrared (IR) motion sensor (Product #189, Adafruit Industries, New York, NY) mounted inside the filtertop lid of the cage, and the second was a magnetic running-wheel sensor that measured each rotation of the running wheel. All sensor data were acquired and stored in 1-minute bins on a computer using Clocklab software and hardware (Actimetrics, Wilmette, IL).

Animals were implanted abdominally in early-May 2023 with core body temperature (T_b) loggers (CubiSens^™ Model TS110, Ann Arbor, MI, USA; mass 0.7g) programmed to record T_b (±0.1°C) at 6 min intervals. A 1–2 cm midline incision was made in the peritoneum beginning approximately 2 cm above the genitals. We closed the incision in 2 layers: the linea alba using absorbable sutures and the skin with synthetic nonabsorbable sutures. Animals were allowed to recover for two weeks prior to beginning the food manipulation experiment.

Experimental design

In order to simplify the experimental logistic process of time-specific (light vs. dark phase) food availability, we housed the two groups of animals in adjacent rooms with light cycles inverted by 12 hours from each other. In the first room, the lights turned on at 0900h and off at 2100h local time, and in the second room, the lights turned on at 2100h and off at 0900h. Animals were initially provided with 24-h ad libitum food and water availability, and their activity and feeding patterns entrained to the respective 12:12 LD cycles of each room (taking approximately two weeks), as observed in running wheel activity and general cage movement activity (IR sensor), and all subsequent baseline recordings were collected after this adjustment period. In addition, body mass (g) and food consumption were assessed at the same time every 2–7 days throughout the duration of the experiment and food consumption reported in g/day.

Because no previous reports exist for protocols of time-restricted feeding in this species, we made a preliminary determination, before our main experiments, of the minimum duration of a daily feeding window that is sufficient for animals to maintain body weight with a normal amount of daily energy intake. To do this, we compared initial 24-h baseline food consumption (mean ± SE, 15.7±0.84 g/day) and body mass (116.5±1.6 g) values to values under different durations of food availability (from a maximum of 24 hours to a minimum of 6 hours) in nine animals housed in the first room with the regular light-dark schedule. We transitioned these animals from 24-h of ad libitum food availability to a minimum 6-h feeding window centered in the middle of the light phase. Within four days, we observed a substantial average decrease in food consumption (to 9.2±0.8 g/day) and a decrease in body mass (to 106.4±1.8 g). We then increased the feeding window to 8 h, and after seven days we observed an increase in average food consumption (to 14.04±1.1 g/day) and body mass (to 115±3.1 g) back to the baseline levels. We thus concluded that an 8-h duration of food availability should be an appropriate window for AGS to maintain normocaloric feeding and body weight. Although FAA may be less robust under longer feeding windows than shorter ones, mice can still express FAA under normocaloric restricted-feeding windows of 8 h (31) and as long as 12 h (32). We therefore used an 8 h window to test whether AGS would express FAA in the absence of an overt energetic deficit. After these preliminary tests, the animals were returned to baseline ad libitum 24-h feeding conditions for 4 weeks prior to the beginning of the experimental conditions described below.

The final set of measurements and observations took place from May 20 to June 15, 2023. Following the pre-experimental stabilization of behavioral rhythms, a new set of baseline behavioral activity recordings was made for seven days. Following this, animals in each room were subjected to one of two patterns of temporally restricted feeding conditions: daytime feeding (DF) (N = 9, 3 females and 6 males) in the first room and nighttime feeding (NF) (N=6, 1 female and 5 males) in the second room. DF animals were given an 8-h interval of food availability centered in the middle of the light phase, and NF animals were given the same treatment in the middle of the dark phase. After eight days of this TRF, two NF animals were showing decreases in T_b and body weight, and we removed those two individuals from the TRF treatment and returned them to ad libitum conditions. All remaining animals were returned to ad libitum conditions after the 20-day TRF manipulation to end the experiment.

Analysis

Measurement and Calculation of T_b

All analyses were conducted with R statistical software (Version 4.4.3; http://www.R-project.org/) unless stated otherwise. Because T_b is influenced by heat generated from locomotor activity, we chose to adjust our measurements of T_b by a correction factor representing heat generated by activity for an interval of 10 min prior to any bout of activity in order to more accurately assess underlying circadian-driven changes in T_b (as described by Cambras et al. 2007, and Lee et al. 2009) (33,34). To accomplish this, we calculated the standardized readings of each animal’s daily T_b and the 85^th percentile of daily maximum locomotor activity under 12:12 LD ad libitum baseline conditions. We then performed a least-squares linear regression analysis, using T_b as the dependent variable and locomotor activity from the preceding 10-min interval as the independent variable. Since this analysis incorporates data across the full 24 h cycle, the slope of the regression line represents the incremental increase in T_b due to locomotor activity, independent of circadian T_b variation. Using the resulting regression equation, we calculated these adjusted (“demasked”) T_b values by subtracting activity-specific T_b increase from the absolute T_b values. These adjusted T_b values were used in all subsequent analyses and are referred to hereafter as T_b.

Daily Locomotor Activity, T_b, Body Mass, and Food Consumption during Time-Restricted Feeding

To evaluate how daily locomotor activity, T_b, body mass, and food consumption changed over the 20-day TRF treatment, we ran separate two-way linear mixed-effects models (LMMs) including average daily IR-detected cage activity, total daily wheel revolutions, minimum daily T_b (T_b,min), body mass, and daily food consumption as response variables. Fixed effects included days since the start of TRF, feeding group (daytime vs. nighttime feeding), and their interactions, while individual identity was included as a random effect. We used Type III ANOVA tests with Satterthwaite’s approximation for degrees of freedom to determine the significance of fixed effects, followed by Tukey-adjusted post-hoc comparisons to identify daily differences between feeding groups. Two animals from the nighttime feeding (NF) group (M1 and M2, Fig. 1) exhibited a significant drop in T_b by day eight of the NF treatment, suggesting insufficient food intake. Consequently, they were removed from the experiment on day nine, and we only included their data through the day of their removal in our analysis. To visualize these longitudinal changes, we plotted mean ± SE values for each variable across the TRF period and indicated significant post-hoc differences where applicable.

Figure 1:

Double-plotted actograms of smoothed 1-h rolling mean activity recorded by infrared detection (IR) and superimposed to core body temperature (T_b) for 15 antelope ground squirrels. Temperature is represented by colors shown on scale; and activity is represented by black overlay. Squirrels were exposed to a 12:12 LD cycle throughout and an 8-h time-restricted feeding (TRF) window centered around either the middle of the dark phase (night; a) or middle of the light phase (day; b) denoted by the white-dotted inset boxes of 8-h duration. Yellow-dotted inset boxes represent the four days of data used in the analyses for Figure 2 and Table 2. Individual animal numbers and sex (F or M) are shown at top of each record.

24 Hour Rhythms of Activity and Temperature

To visualize individual changes in locomotor activity and T_b over the course of the experiment in each individual, we used El Temps software (v.1.228, University of Barcelona, Spain) to generate double-plotted actograms containing raw infrared (IR) activity measures and T_b measures smoothed with a 1-h rolling average. We performed additional LMMs for each sensor type (IR detector, running wheel, and T_b logger) within each feeding group (daytime feeding vs. nighttime feeding). For these models, we used raw 10-min averaged activity counts from the final four days of each experimental stage (baseline vs. time-restricted feeding [TRF]). Fixed effects included time of day (24-h period binned into 10-min intervals), experimental stage, and their interaction. Individual identity and day were included as random effects to account for repeated measures. Significance of fixed effects was assessed using Type III ANOVA tests, followed by Tukey-adjusted post-hoc pairwise comparisons to identify time points with significant differences across experimental stages. To visualize these effects, we plotted 24-h waveforms of behavioral and T_b data, overlaying significant post-hoc contrasts to highlight key differences.

As an additional test for the presence of anticipatory locomotor activity prior to the onset of food availability (FAA), as manifested by significant increases in activity or T_b, we calculated the average cage (IR-detected) locomotor activity, running wheel activity, and T_b during the two hours preceding food availability for the final four days of each experimental stage (baseline vs. TRF) for each animal and conducted paired Wilcoxon signed-rank tests comparing values from the baseline phase to the TRF phase.

RESULTS

Following 20 days of time-restricted feeding (TRF), we observed no evidence of food anticipatory activity (FAA) in 14 of 15 animals (Figs. 1 and 2). All animals under the initial 24-h ad libitum feeding conditions synchronized their daily behavioral (activity) and physiological (body temperature) rhythms with the 12:12 LD cycle as expected and as evidenced by major daily patterns of locomotor activity during the light phase (daytime) of the daily LD cycle and an associated circadian rhythm of increased T_b during the daytime and decreased T_b during the nighttime. Once TRF began, animals in the two treatment groups responded differently over the following days. At the start of TRF, the DF and NF animals had similar levels of body mass and food intake. However, within the first five days of TRF, the NF animals exhibited a significant and rapid decline in both body mass and food consumption, unlike the DF animals, and these differences persisted throughout the first half of the TRF period (Fig. 3a, b; Supplementary Table 1). General locomotor activity in the cage did not differ between groups (Fig. 3c), but NF animals showed a gradual decline in running wheel activity across days, ultimately running significantly less than DF animals by the end of TRF (Fig. 3d). Minimum body temperature (T_b,min) also became lower over time in the NF animals. Although both groups started with similar values of daily T_b,min (DF 33.5±0.36°C; NF 33.9±0.44°C), NF animals experienced a rapid drop of ~2°C by day four (DF 34±0.36°C vs NF 31.8±0.44°C), which remained significantly lower for the remainder of the TRF period (Fig 1e). These patterns were supported by significant interactions of treatment group × time for body mass (F_5,58=12.24, P<0.0001), food consumption (F_5,58=4.38, P=0.002), daily wheel activity (F_20,214=5.4, P<0.0001), and T_b,_min (F_20,230=4.22, P<0.0001), but not daily general cage locomotor (IR-detected) activity (F_20,234=1.32, P=0.17) (Table 1).

Figure 2:

Responses of antelope ground squirrels to an 8-h feeding window centered either around the middle of the light phase (day; a,b,c; N = 9) or middle of the dark phase (night; d,e,f; N = 6). Average 24-h activity waveforms of infrared-detector cage activity (a,d), wheel running activity (b,e), and body temperature (c.f). Baseline measures (green), where squirrels were provided with ad libitum food, are compared to time-restricted feeding (TRF) measures (purple), and significant differences between experimental stages are denoted by asterisks, as determined by post-hoc tests (P<0.05).

Figure 3:

Responses of antelope ground squirrels over the course of the experimental 20-day time-restricted feeding (TRF) interval: (a) body mass, (b) daily food consumption, (c) mean daily infrared (IR) detector activity, (d) sum of daily wheel activity, and (e) daily minimum core body temperature (T_b) (mean ± SE). Blue color represents the feeding group with an 8-h TRF window centered in the middle of the 12-h light phase (day; N = 9) and red represents the 8-h TRF window centered in the middle of the 12-h dark phase (night; N = 6). Days of observed significant differences, as determined by post-hoc tests (P<0.05) between feeding groups, are denoted by asterisks. Note: Animals M1 and M2 (Nighttime Feeding group) experienced substantial drops in T_b by day 8 and were subsequently returned to ad libitum conditions; therefore, their data is only represented in this figure up to day 8.

Table 1:

The influence of time-restricted-feeding (TRF) group (daytime vs nighttime feeding), days since introduction to TRF, and their interaction on behavioral and physiological measures of antelope ground squirrels, as determined by two-way mixed ANOVAs.

Response Variable	Treatment (Day vs Night Fed)		Day Since Time-Restricted Feeding		Interaction
IR Activity	F1,13=0.05	P=0.83	F20,234=1.04	P=0.42	F20,234=1.32	P=0.17
Wheel Activity	F1,12=0.33	P=0.58	F20,214=6.5	P<0.0001 ***	F20,214=5.4	P<0.0001 ***
Core Body Temp	F1,13=12.2	P=0.004 **	F20,234=2.39	P=0.001 **	F20,234=4.6	P<0.0001 ***
Body Mass	F1,13=6.96	P=0.02 *	F5,58=3.25	P=0.01 *	F5,58=12.24	P<0.0001 ***
Food Consumption	F1,13=9.64	P=0.008 **	F5,58=25.0	P<0.0001 ***	F5,58=4.38	P=0.002 **

Examination of actograms and individual T_b records revealed that some animals in the NF group experienced fasting-induced hypothermia (Fig. 1, Supplementary Fig. 1). Two males, M1 and M2, showed substantial drops in T_b within the first eight days of TRF and were removed from the study on day nine of TRF. Both of these animals quickly returned to normal nighttime T_b (T_b,min >33.5°C) after resuming ad libitum 24-h food availability. Among the NF animals, only one individual (M3) successfully shifted its major rhythms of daily activity and associated daily maximal T_b to the nocturnal phase.

Activity and T_b rhythms varied significantly across time of day depending on the experimental stage (baseline vs. TRF), with distinct daily patterns emerging between treatment groups (Table 2). During TRF, DF animals exhibited elevated general cage activity and wheel activity compared to baseline during time points associated with the feeding window, including up to 20–30 min before access to food (Fig. 2a, b; Supplementary Table 2). However, T_b levels of DF animals were also significantly lower during TRF, as compared to baseline, especially in the ~2.5 h before the feeding interval (Fig. 2c). In contrast, NF animals exhibited pronounced differences in behavioral and T_b measures between baseline and TRF conditions (Fig. 2d, e). Notably, NF animals displayed significantly increased general cage activity during the feeding window, while running wheel activity was consistently reduced during the light phase (Fig. 2e). Nighttime food restriction also resulted in lower T_b throughout the light phase and elevated T_b during the initial hours of the feeding interval, as compared to baseline.

Table 2:

The influence of experimental stage (ad libitum baseline vs time-restricted feeding [TRF]), time of day (24-h period binned in 10-min intervals), and their interaction on behavioral and physiological measures of antelope ground squirrels. Animals in the daytime feeding group received an 8-h feeding window centered around the middle of the light phase and nighttime animals received the same treatment in the middle of the dark phase.

Response Variable	Feeding Group	Experimental Phase		Time		Interaction
IR Activity	Dayttime	F1,6=19.3	P=0.005 **	F143,10066=66.1	P<0.0001 ***	F143,10066=2.25	P<0.0001 ***
	Nighttime	F1,7=11.02	P=0.013 *	F143,5411=15.24	P<0.0001 ***	F143,5411=2.38	P<0.0001 ***
Wheel Activity	Dayttime	F1,6=11.14	P=0.016 *	F143,8915=52.63	P<0.0001 ***	F143,8915=2.26	P<0.0001 ***
	Nighttime	F1,6=20.18	P=0.004 **	F143,5411=14.01	P<0.0001 ***	F143,5411=2.81	P<0.0001 ***
Core Body Temp	Dayttime	F1,6=4.37	P=0.082	F143,7796=40.48	P<0.0001 ***	F143,7795=1.39	P=0.001 **
	Nighttime	F1,6=61.58	P=0.0002 ***	F143,4574=3.36	P<0.0001 ***	F143,4573=5.17	P<0.0001 ***

We also tested for signs of food anticipatory activity (FAA) and thermogenesis by examining activity and T_b levels in the two hours preceding food availability. However, neither DF nor NF animals exhibited significant FAA (DF IR activity V=27 P=0.3, DF wheel activity V=21 P=0.4, NF IR activity V=7 P=0.3, NF wheel activity V=4 P=0.7) or FAT (DF V=8 P=0.9, NF V=1 P=0.9).

DISCUSSION

Our findings demonstrate that the antelope ground squirrel (AGS), a naturally strictly diurnal species, does not reliably exhibit food anticipatory activity (FAA) or enhanced thermogenesis when its food supply is temporally restricted to an atypical time of the daily cycle–nighttime. In contrast to nocturnal laboratory rodent species that readily show FAA before an imposed mealtime that is experimentally presented during the rest phase of the daily cycle (6,35), the locomotor activity of AGS in our study remained predominantly diurnal, with no clear anticipatory rises in core body temperature or wheel running and IR-detected activity before the nocturnal time of food availability. This lack of plasticity in the 24-h behavioral and physiological program of AGS was associated with reduced food intake and an overall thermo-energetic imbalance, characterized by lower running wheel activity and body mass and decreased T_b, which in two individuals led to hypothermia. The lack of shifts in activity timing is likely generalized across other possible nonphotic cues, as prior work with AGS has shown that cyclic aversive stimuli (unpredictable footshocks while an animal is foraging and feeding outside of a safe nest) that serve as a potent synchronizing environmental cue in nocturnal rodents (36,37), likewise fail to shift AGS activity to the dark phase of the LD cycle (38). This limited plasticity suggests that diurnal mammals respond differently from nocturnal mammals when photic and nonphotic cues conflict. The exception was one night-fed animal (M3), which over the course of the nighttime-feeding period gradually shifted its activity and Tb rhythms and was nocturnal by the end of the experiment. Although its activity during the middle of the time-restricted feeding (TRF) period may resemble FAA, this is better interpreted as a feature of a transient rhythm during a slow phase shift rather than FAA per se. Notably, by the end of the experiment, there was no clear bout of anticipatory activity preceding food access, underscoring that a behavioral phase shift and FAA do not necessarily co-occur.

Prevailing models of TRF propose that food-sensitive circadian clocks in peripheral tissues (7,8), e.g., liver and adrenal glands, and extra-SCN brain regions (9,10) are entrained by experimentally scheduled feeding and can drive anticipatory outputs even when the SCN central clock remains aligned with the light-dark cycle (39,40). Although the presence of an intact SCN is not necessary for FAA expression, the SCN can inhibit FAA expression under conditions of a light-dark (LD) cycle (2,6,41). For example, rats with SCN lesions show earlier-onset and higher-amplitude FAA than intact controls (6), and mice with molecularly dysfunctional forebrain clocks can show enhanced FAA relative to wild-type mice (42). Computational modeling of competing biological clocks similarly predicts that a strongly LD-locked central clock can dominate a weakly coupled food-driven clock when the cues conflict (43,44). Interestingly, light can also directly control the phase of adrenal gene expression and glucocorticoid release via an SCN-sympathetic pathway (45), and glucocorticoid signaling can slow uncoupling of peripheral clocks from the SCN central clock during restricted feeding in nocturnal mice, suggesting SCN-driven physiological outputs counteract feeding rhythms when their phases conflict (46,47). In AGS, the complete absence of FAA, even in daytime-fed individuals, suggests that feeding-time signals either failed to sufficiently engage the food-entrainable network under our conditions or that the nonphotically entrained clocks were too weak to override photically organized (SCN-driven) rhythms in LD. This latter interpretation is also consistent with the limited response of AGS to other nonphotic schedules under LD conditions (38).

Interactions between energetic state and feeding schedule can also modulate FAA, as caloric restriction has been shown to amplify phase-shifting of activity and anticipatory behaviors in nocturnal rodents (11,48). Accordingly, we first isolated the temporal duration of TRF by determining that an 8-h feeding window in the light phase is sufficient to maintain body mass. However, when we made food available for the same amount of time during the dark phase, animals failed to display FAA, lost weight, and two individuals even became hypothermic, even though they had access to food during the nighttime. Thus, we suggest that the lack of FAA in AGS is associated with an inability to adequately consume and process energy sources during the rest phase. In mice, restricting food to the rest phase shifts circadian rhythms in the liver, kidney, heart, and pancreas while leaving the phase of SCN rhythms largely unchanged (7,9,35). It remains possible that TRF could entrain peripheral clocks in AGS without accompanying behavioral phase shifts of activity. However, the marked reductions of body mass, wheel-running activity and T_b exhibited by AGS under nighttime-restricted feeding suggest that any physiological adjustment was insufficient to support sustained nocturnal food consumption and activity under the prevailing experimental conditions.

It remains possible that a shorter window of food presentation would induce FAA in AGS, but separating timing effects from caloric restriction would be challenging. Critically, caloric restriction in nocturnal rodents (without a fixed mealtime) has been shown to shift nocturnal activity to the daytime and induce nocturnal torpor (30,49), indicating that negative energy balance alone is a potent driver of daily physiological and behavioral organization, independent of the specific effects of TRF. Yet, despite clear evidence of negative energy balance in night-fed individuals (reduced T_b and weight loss), AGS still failed to show robust FAA, suggesting that activity shifts induced by caloric restriction may be constrained in AGS, unlike nocturnal species. From an ecophysiological perspective, the circadian thermoenergetics (CTE) hypothesis proposes that negative energy balance can favor diurnal activity in naturally nocturnal mammals through thermoregulatory advantages associated with foraging during the daytime and resting at night (49,50). The CTE hypothesis also predicts that diurnal animals would have a more rigid temporal niche because their pattern of activity is already optimized thermoenergetically. In other words, the energetic incentive to switch their temporal niche to nighttime in diurnal animals like the AGS is likely very low, which may explain why such behavior is rare in typically diurnal mammals in the wild (51). Consistent with this interpretation, the limited temporal flexibility we observed could also reflect a long evolutionary commitment to the diurnal niche, reflected in specializations for daytime vision such as the pure-cone retina of AGS (52).

Our findings raise new questions regarding the impact of TRF from the perspective of a diurnal mammal. FAA has been most consistently demonstrated in nocturnal laboratory rodent models, and evidence from other mammalian models remains limited (11,12,18,19), and to our knowledge, none has been shown in humans. In humans, a moderate delay in meal timing shifts rhythms of PER2 gene expression in adipose tissue and glucose synthesis without shifting rhythms of melatonin secretion (53), a form of internal desynchrony linked to circadian misalignment and adverse metabolic outcomes (54,55); in addition, recent evidence suggests that humans show physiological anticipation through lower glucose levels before projected mealtimes under constant-routine conditions (56). Regardless of the presence of FAA in diurnal species, including humans, our results raise the possibility that a decreased plasticity in 24-h programs of physiology and behavior is the norm for diurnal species. If this is the case, the adverse metabolic outcomes of nocturnal feeding in diurnal mammals may be much worse than those of diurnal feeding in nocturnal species.

Our findings suggest that strictly diurnal sciurid rodents may be less flexible than the nocturnal murid rodents and other diurnal rodent models in adjusting behavioral and physiological rhythms to altered feeding schedules. Future work should examine FAA and food entrainment under constant conditions of illumination (light or dark) to remove the effect of competition of the LD cycle (43) and should expose animals to shorter (perhaps 2–3-h) feeding windows, which have been shown to elicit stronger effects in other species. However, we also acknowledge that disentangling caloric restriction from the effects of food timing will remain challenging. Parallel measurements of peripheral molecular, e.g., Per2 mRNA expression, and hormonal rhythms, e.g., glucocorticoids, in diurnal species are also lacking, and such studies would help to determine whether food cues synchronize internal physiology even when behavior remains fixed. Overall, we interpret our findings in AGS as the result of strong diurnal organization under the LD cycle and relatively weak coupling of feeding time to overt behavioral and thermoregulatory outputs in a species that is adapted to the daytime activity niche. Given the clinical significance of internal circadian misalignment in humans, as in shift workers, our findings underscore the need for a robust diurnal model to test how feeding schedules influence circadian regulation and, by extension, metabolic and cardiometabolic health.

Data Availability:

Data and code supporting this manuscript are available at the following link.