State‐dependent risky choices in primates: Variation in energy budget does not affect tufted capuchin monkeys' (Sapajus spp.) risky choices

Economic models predict that rational decision makers' choices between a constant, “safe” option and a variable, “risky” option leading, on average, to the same payoff, should be random. However, a wealth of research has revealed that, when faced with risky decisions, both human and nonhuman animals deviate from economic rationality. According to the risk‐sensitivity theory, individuals should prefer a safe option when they are in a positive energy state and a risky option when they are in a negative energy state. The abundance/risk hypothesis proposes that individuals should prefer risky options when diet quality exceeds their nutritional requirements. We tested how energy budget affects decision making under risk by presenting 22 capuchins belonging to two colonies (IT: N = 12, US: N = 10) with a risky choice task. Capuchins had to choose between a constant option (always four food items) and a variable option (one or seven food items with a 50% probability) in two conditions. In the Low‐energy condition capuchins were tested before their main meal, whereas in the High‐energy condition they were tested following a high‐caloric meal. In neither colony did we find a significant difference between conditions, suggesting that the energy budget did not affect risk preferences. However, we found differences between colonies in their general response to risky choices: US capuchins were more risk‐prone after selecting a safe option than a risky option and after selecting a bad (one food item) than a good (seven food items) risky option, whereas this did not hold true in IT capuchins. Furthermore, in the IT colony, males were more risk‐prone under the High‐energy condition compared to the Low‐energy condition. Subtle differences in individual characteristics, management conditions, or stochastic founder effects may be implied, with relevant consequences for the outcomes of research on risky decision‐making across laboratories.


| INTRODUCTION
Decision-making under risk is ubiquitous in both human and nonhuman animals' (hereafter animals) everyday lives. Individuals often face situations in which their choices lead to a variety of possible outcomes. When deciding between foraging patches, for example, animals must choose between options that differ probabilistically in their payoff. As such, how animals decide in these situations has been a central issue for biology, psychology, and economics.
Although classical economic models predict that rational decision makers should maximize their expected utility by taking into account the expected value of each option (Baron, 2000;von Neumann & Morgenstern, 1947), research has repeatedly revealed that, when faced with variable, "risky" decisions, both humans and animals may depart from economic axioms of rationality and show a marked interspecific variability in risk preferences. For instance, chimpanzees, capuchin monkeys, and macaques (Macaca mulatta) preferred risky options (e.g., Haun et al., 2011;Hayden & Platt, 2007;Heilbronner et al., 2008;De Petrillo, Ventricelli, et al. 2015), whereas bonobos (Pan paniscus) and several lemur species were risk averse (Haun et al., 2011;Heilbronner et al., 2008;MacLean et al., 2012).
According to the foraging ecology hypothesis, foraging patterns (such as diet and extractive techniques) are the main factors that shaped species-specific decision strategies (Rosati, 2017;De Petrillo & Rosati, 2019; see also Gigerenzer et al., 1999). A comparative study on chimpanzees and bonobos supports this hypothesis: risk-prone chimpanzees feed mainly on fruits, a highly variable food source, whereas risk-averse bonobos rely largely on terrestrial herbaceous vegetation, a stable food source (Heilbronner et al., 2008;Wrangham & Peterson, 1996).
At the individual level, many factors may account for the violation of normative models, including cognitive biases (Rosati, 2017), contextual factors (Heilbronner & Hayden, 2013), social influences (Rosati & Hare, 2012;Zoratto et al., 2018), and individuals' energy budgets (Stephens, 1981). The energy budget rule (Stephens, 1981) states that individuals should favor a constant, safe option when they are in a positive energetic state. Conversely, they should favor a variable, risky option when they are in a negative energetic state, as it may be the last chance to meet their energetic requirements (Caraco, 1981). This rule has been empirically demonstrated in a small bird species, the yellow-eyed junco (Junco phaenotus). When choosing between safe and risky feeding stations, yellow-eyed juncos preferred the risky option after undergoing a longer period of food restriction (i.e., when in a negative energetic state), whereas they opted for the safe option after undergoing a shorter period of food restriction (i.e., when in a positive energetic state) (Caraco et al., 1980). A similar finding emerged in bees (Apis mellifera), in which changes in risk preferences are linked to shifts in energy budget rather than to the individual's absolute energetic state.
However, other previous work found only weak support for the energy budget rule (Brito e Abreu & Kacelnik, 1999;Kacelnik & El Mouden, 2013); even though it seems to be verified in small-bodied species (Caraco et al., 1990;Cartar & Dill, 1990;Croy & Hughes, 1991;see Kacelnik & Bateson, 1996 for a review), it is still unclear if the same holds true in larger-bodied species. Whereas small-bodied animals may be sensitive to energy budget manipulation because they are physiologically constrained by high metabolic rates that can rapidly lead to the risk of starvation (Caraco et al., 1990), larger-bodied animals with lower metabolic rates (and larger energy stores) may more rarely experience such risk (Kacelnik & Bateson, 1996;Kacelnik & El Mouden, 2013). Thus, they may be less or not at all sensitive to a decrease in their energy budget.
The abundance/risk hypothesis suggests that larger animals may engage more in risky activities when food sources are abundant, such that the costs associated with a negative outcome are easier to overcome. This hypothesis is supported by data coming from 14 years of observations on chimpanzees' behavior in Kibale National Park, Uganda. Wild chimpanzees hunt colobus monkeys-a risky activity of which the result is unpredictable-more frequently when high-quality food is abundant (Gilby & Wrangham, 2007). The energy budget rule and the abundance/risk hypothesis, however, are not mutually exclusive. The first refers to an "energetic threshold" that must be reached to ensure the individual's survival, whereas the latter highlights the significance of diet quality and abundance of food resources required for engaging in other activities.
One significant limitation of previous work is that most of the studies carried out so far involved a drastic modification of the experimental subject's energy budget (e.g., Caraco et al., 1980), whereas it has been poorly investigated how risk preferences vary under less extreme parameters like daily fluctuations in hunger and circadian rhythms. This is particularly important when considering animals that live in relatively enriched environments or captive animals, for which extreme shifts in energy budgets do not occur.
Moreover, up to now, only a few studies have investigated the role of variations in energy budget in medium-sized and large animals. Thus, it is important to gather data to further test both the energy budget rule and the abundance/risk hypothesis.
Here, we examined the effect of energy budget manipulations on risky decision-making in tufted capuchin monkeys and evaluated whether our data are consistent with the expectations of the energy budget rule or the abundance/risk hypothesis. Capuchins are a particularly good model species to test for this because they are medium-sized animals that in the wild generally live in relatively enriched environments and, consequently, experience low starvation risk. In addition, our colonies of capuchin monkeys are never food restricted, increasing the ecological validity of our results. We tested 22 adult tufted capuchin monkeys living in one of two colonies in a risky choice task (using the same method as De Petrillo, Ventricelli et al., 2015) in two experimental conditions in which their energy level was manipulated (Low-energy and High-energy). In each of these conditions, we presented capuchins with choices between a risky option and a safe option.
If capuchins behave according to the energy budget rule, we expected them to choose the risky option more in the Low-energy condition than in the High-energy condition. Conversely, if they behave according to the abundance/risk hypothesis, we expected to observe the opposite pattern. Additionally, if variation in energy requirements or food abundance are the main factors driving capuchins' risk preferences, there should be no differences between the two colonies in how capuchins respond to the energy manipulation. Conversely, population-specific histories of participation in cognitive testing, or husbandry and rearing, may influence individuals' behavior (Stevens, 2017). Indeed, previous comparisons of delay choices between capuchins housed in different laboratories have shown population-level differences (Addessi et al., 2013) even when general parameters (i.e., socially housed, never food deprived) were the same. Thus, the comparisons between colonies will allow us to also assess the impact of capuchins' experiences in their current captive environment on their risky choices, shedding light on the unsolved question of whether differences in decision strategies stem from individuals' experiences, or are rather features typical of each species (De Petrillo & Rosati, 2021 In the IT colony, each capuchin group was housed in a separate indoor-outdoor enclosure. Depending on group size, the outdoor enclosures range in size from 53.2 to 374.0 m 3 , whereas, for each group, the two indoor enclosures measure a total of 25.4 m 3 . Wooden perches, tree trunks, and branches are present in each enclosure. Subjects underwent testing in one of the two indoor enclosures. To achieve separation for individual testing, the group was divided into smaller units using sliding doors and then the subject was allowed to enter the indoor enclosure. Testing took place between 09:30 and 13:30 h. The main meal (made of fresh fruits, vegetables, and monkey chow) was provided to capuchins in the afternoon. There was unlimited access to water. This study was carried out in accordance with protocols approved by the Italian Health Ministry (n. 337/2017-PR to E. Addessi), following the Directive 2010/63/EU on the protection of animals used for scientific purposes.
In the US colony, each capuchin group was housed in their own enclosures that included both indoor (range 13.07-52.18 m 3 ) and outdoor space (range 33.09-444.02 m 3 ); subjects had access to the outdoor space except during inclement weather and testing. During testing, the monkeys could enter a testing chamber attached to their group's indoor enclosure to participate in the task. Subjects chose to participate voluntarily and were never compelled to do so through restriction of food, treats, water, or access to the outdoor space or social partners; if subjects chose not to enter the testing chamber, their only consequence was not participating in the day's testing.
Testing occurred between 08:00 and 10:00 h, before the monkeys received their main meal. The capuchins were offered primate chow at mid-day, unless they chose to participate in the test, in which case they got food rewards during testing and fruit after testing. All

| Apparatus
In the IT colony, two options were presented on a platform (65 × 40 × 16 cm) with two transparent boxes (9.5 × 20 × 15 cm each), spaced 28 cm apart (see De Petrillo, Ventricelli, et al., 2015). Subjects could choose their preferred option by putting their finger in a hole (diameter: 2 cm) of one of the two boxes through one of two openings in the wire mesh, which were 8.5 cm × 3.8 cm each, and aligned with the boxes. After choosing, the subject received the food underneath the chosen option (see below) from the experimenter.
In the US colony, the two transparent boxes measured 16.36 × 16.36 × 13.89 cm each, and the platform measured 24.13 × 34.29 × 0.64 cm. Subjects chose their preferred option by touching one of the transparent boxes through openings in the wire mesh wall of the testing chamber.

| Procedure
To familiarize capuchins with the contingencies of the task (i.e., to make sure they learned the association between each option and the corresponding bowl), in both colonies, before starting the experiment, for each subject we carried out three familiarization sessions (on three different days) consisting of 16 forced-choice trials (i.e., trials in which a single option was presented). In eight trials we CIACCI ET AL. | 3 of 9 presented the safe option (i.e., four food items) and in eight trials we presented the risky option (i.e., either one or seven food items-four trials each). The order and position (left/right) of the forced-choices were pseudo-randomized within each session.
The main experiment involved two conditions: (1) Low-energy and (2) High-energy, of which the presentation order was counterbalanced across subjects. In the Low-energy condition, subjects were tested prior their daily main meal, whereas in the High-energy condition they were tested following a high-caloric meal.

| Caloric manipulation
In the High-energy condition, IT capuchins received a high-caloric meal composed of (i) 100 g of mashed banana (89 kcal reported to consume daily, on average (total energy expenditure: 342 ± 108 kcal; Edwards et al., 2017). The subject was considered satiated if (i) the mixture was totally consumed or (ii) 3 min had passed since the last mouthful was ingested (the foods in the High-energy condition were all preferred, so motivation to consume them was high).
In each session of the High-energy condition, we scored the amount of food eaten (resulting from the difference between the amount of food offered and the leftovers, if any, after the subject stopped eating). In both conditions, we also scored the number of calories consumed by the subjects in other experiments, if any, carried out before the risky choice task. In the Low-energy condition, the IT colony did occasionally receive some food (e.g., carrots) during another experiment occurring before the present one; the US colony systematically received 100 g of carrot (89 kcal) before testing, although they did not always consume all of the carrot, which is a less preferred food (we recorded what was left uneaten and calculated their caloric intake). Overall, capuchins in the IT colony consumed on average 168.92 ± 13.76 kcal before the Highenergy condition and 6.49 ± 2.82 kcal before the Low-energy condition.
Capuchins in the US colony instead consumed on average 209.97 ± 7.76 kcal before the High-energy condition and 23.28 ± 1.67 kcal before the Low-energy condition.

| Risky choice task
The risky choice task (originally developed by Heilbronner et al., 2008 in chimpanzees and bonobos) was previously used with eight out of the 12 individuals belonging to the IT colony (De Petrillo, Ventricelli, et al., 2015), whereas none of the capuchins belonging to the US colony had ever been tested in this task (although some had participated in other tasks measuring risk). We chose to use this task as it has already been well-validated in primates.
As described in De Petrillo, Ventricelli, et al. 2015, each capuchin could choose between a safe option (always four food items) and a risky option (yielding one or seven food items with a 50% probability). Food items consisted of tiny pieces of peanuts (each corresponding to 1/8 of a peanut seed). The two options were represented by a pair of upsidedown bowls that were visibly different from one another, and the assignment of each bowl to the safe and risky option was counterbalanced across subjects. In each condition, capuchins were given 10 sessions, each with 10 free-choice trials and six forced-choice trials. In In the IT colony, the choice apparatus was positioned in the area in front of the indoor compartment. The subjects were tested by two experimenters: experimenter 1 (E1) sat behind the apparatus, in front of the subject's indoor compartment, whereas experimenter 2 (E2) sat next to E1. In each trial, E1 baited the apparatus, while E2 prevented the subject's visual access to the apparatus by means of an opaque screen.
After baiting, E2 raised the opaque screen and E1 moved the apparatus in the direction of the wire mesh, to allow the subject to choose. To avoid giving the subject any verbal or visual cues, both experimenters remained silent and avoided looking at either of the two boxes. After the subject's choice, the experimenter moved the chosen bowl in the direction of the subject, allowing the subject to take the food. As soon as the subject finished eating, E2 replaced the opaque screen and, after a 30-s intertrial interval, the subsequent trial began. E2 scored data live, and all sessions were videotaped. The trial was repeated up to three times if the subject did not select either option within 30 s; in cases when the subject did not choose three consecutive times, the session was repeated on the following day.
In the US colony, the choice apparatus remained attached to the side of the testing chamber. One experimenter tested the subjects.
Before each trial, the experimenter positioned an opaque barrier between the apparatus and the chamber, so subjects were unable to see or touch the apparatus as it was baited. The experimenter then lifted the barrier, allowing the subject access to view the boxes and make a choice. All the other methodological details were the same as for the IT colony.

| Statistical analyses
In the US sample, we could not obtain capuchins' data for 34 trials because of experimenters' errors or camera failure. We analyzed the data in R v4.0.2 (R Core Team, 2021). For our main analysis we implemented Generalized Linear Mixed Models (GLMM; Baayen, 2008) with binomial error structure using the glmer function from the lme4 package (Bates et al., 2015). The dependent variable was whether capuchins opted for the risky option on a given trial, while the test predictors were the energy condition (High-energy or Low-energy) and the colony (IT and US), as well as their interaction.
We included sex (male or female), trial number (within session), and session number as control predictors and subject as a random effect to account for repeated measurements (Baayen, 2008). We also included session and trial as random effects to account for the fact that our data were not independent due to the repeated measurements for each subject across multiple sessions and trials. To assess the overall significance of our predictors of interest, we constructed two models: a full model, including our test predictors, control predictors, and random effects, and a null model, retaining only control predictors and random effects. We then used a likelihood ratio test to compare these models (controlling for Type I error rate, Forstmeier & Schielzeth, 2011).
In our exploratory analyzes, we ran the same full model but additionally examined the effect of the outcome capuchins obtained in their previous trial, and whether capuchins' risky decisions were affected by their body mass or number of calories they consumed before the experiment (see Section 3 for details). We used the emmeans package with a Tukey correction (Lenth & Lenth, 2018) for post hoc comparisons. We also compared males' and females' body mass and caloric intake within and between colonies by means of the Mann-Whitney U test.
The data set generated and analyzed during the current study is available in the Supporting Information and at the following OSF link: https://osf.io/nmqgf/?view_only=d9b4a8045921470c8d78fad4e46 4347d.

| RESULTS
Overall, capuchins of neither colony showed any preference for either the risky or the safe option in either condition (one-sample  Table S2).
We then conducted some exploratory analyzes to investigate whether capuchins in the two colonies used different choice strategies and whether other factors, such as sex, body mass, and caloric intake, affected their choice. First, we added the outcome of capuchins' previous choice (1, 4, or 7) and the interactions with our main predictors to the full model. This comparison revealed a significant effect of the interaction between colony and previous outcome (estimate = 0.827, standard error [SE] = 0.173), χ 2 (9) = 76.399, p < 0.001). Post hoc comparisons showed that while capuchins in the IT colony were not affected by the previous outcome of their choice, capuchins in the US colony chose the risky option more after receiving four items than one item (p < 0.001), after receiving one item than seven items (p = 0.02) and after receiving four than seven items (p < 0.001, see Figure 1). Since this model also revealed a significant effect of sex, with males preferring the risky option more than females (estimate = 0.931, SE = 0.471, p = 0.05), we further explored the interaction between colony, condition, and sex.
Adding this interaction to the previous model significantly improved model fit, χ 2 (3) = 9.074, p = 0.028; see Supporting Information: Table S3. Post hoc comparisons showed that, only in the IT colony, male capuchins chose the risky option more in the High-energy condition than in the Low-energy condition (p = 0.04, see Figure 2).

| DISCUSSION
In the current study, we assessed the impact of a manipulation of the energy budget on risk preferences in two colonies of tufted capuchin monkeys. We aimed to test two hypotheses: (i) the energy budget CIACCI ET AL. | 5 of 9 rule and (ii) the abundance/risk hypothesis. The energy budget rule (Stephens, 1981) states that individuals should prefer a constant option when in a positive energy state, whereas they should prefer a risky option when in a negative energy state. This hypothesis works well for small-bodied animals, but apparently not for larger-bodied species (Kacelnik & Bateson, 1996). The abundance/risk hypothesis (Gilby & Wrangham, 2007) states that larger animals should prefer risky options when the quality of their diet exceeds their nutritional requirements. Furthermore, there is evidence that metabolic state also influences decisions under risk in humans, with mixed evidence supporting ecological models, suggesting a multifactorial and complex relationship between decisional processes and internal state in our species (Levy et al., 2013;Symmonds et al., 2010).
We found that capuchins had no preference for either the safe or the risky option. Additionally, there was no significant effect of the energy state on capuchin monkeys' risk preference, as the frequency of their risky choices was similar between the two experimental conditions (High-energy vs. Low-energy). These findings mirror previous results obtained in two studies testing individuals of the same species in the accumulation task (Parrish et al., 2016;De Petrillo, Micucci, et al., 2015), a self-control task in which food items are progressively accumulated in view of the subject, but the accumulation process stops once the participant takes at least one item. In both studies, no significant relationship between self-control performance and food intake (De Petrillo, Micucci, et al., 2015) or glucose intake (Parrish et al., 2016) was found. Similarly, a recent study on northern bottlenose whales (Hyperoodon ampullatus), a large marine mammal, showed that lipidic stores-which were assumed to be an indicator of an individual's body condition and energy statedid not predict risk behavior (i.e., prioritizing foraging as compared to predator avoidance). This suggests that individual differences could provide alternative explanations for the state-behavior relationships that were observed (Siegal et al., 2022). Indeed, previous research has shown that body size is a crucial factor to take into account when considering how species respond to energy budget manipulations in decision-making tasks. For instance, small-bodied species like bees (Apis mellifera) tested in a self-control task exhibited an increase in impulsive behavior when their energy budget decreased (Mayack & Naug, 2015). Similarly, when presented with a risky decision-making task, bees showed an increase in risk proneness as their energy budget decreased, while an increase in their energy budget was associated with a preference for the constant option (Mayack & Naug, 2011).
In the present study, regardless of the energy budget manipulation, male capuchins were more risk-prone than females. At a proximate level, this may be due to their significantly higher body mass as compared to females. According to the abundance/risk hypothesis, individuals may show an increase in risk preference when diet quality exceeds their nutritional requirements (Gilby & Wrangham, 2007). This hypothesis may be tentatively extended to differences in body mass across individuals, if considering heavier individuals as being, by definition, in a more positive energy state than lighter ones. Also, from an ecological rationality perspective, the difference in feeding ecologies between male and female capuchins may explain their different risk preferences, as it was proposed to explain sex differences in a self-control task (Addessi et al., 2011).
F I G U R E 1 Capuchins' proportion of choices of the risky option in each condition for each colony and in response to the outcome they received in the previous trial.
F I G U R E 2 Capuchins' proportion of choices of the risky option in males and females for each colony and in response to the energy budget manipulations. IT colony: six females, six males; US colony: seven females; three males.
Capuchin males in the wild, in fact, show a more opportunistic foraging style than females. Males forage more often on the ground, preferring small vertebrates and large invertebrates without external protection, whereas females forage more frequently a few meters from the ground, preferring small invertebrates with external protection (Sapajus libidinosus, previously Cebus libidinosus, Verderane, 2010; similar observations have been made for C. olivaceus, Fragaszy, 1986, andC. capucinus, Rose, 1994). We also found that capuchin males belonging to the IT colony were more risk prone in the High-energy condition than in the Low-energy condition.
It is unlikely that this result is accounted for by a difference in body mass or in caloric intake, as both were similar between males belonging to the two colonies, suggesting that other factors may be at play. However, the above findings should be taken with caution as they are based on limited male/female samples. In the IT colony six males and six females participated in the study, while in the US only three males were tested, probably explaining why we found the above effect only in the IT colony. Overall, this indicates the need to use larger samples to understand sex effects in decision-making.
Finally, although the two colonies did not differ in their average risky choices regardless of condition, they did differ in how the outcome of the previous choice affected the following choice. US capuchins were more risk-prone after choosing a safe option than a risky option and after selecting a risky option with a bad outcome than a risky option with a good outcome, whereas the IT capuchins' risky choices were not influenced by the previous choice outcome.
These discrepancies between the two colonies were probably due to subtle differences in methodologies or previous testing experience, whereas we can exclude an effect of age, as all the tested individuals but one were adults. Though the two colonies were tested with the same protocol, there were differences in the housing environment that could have impacted performance. In particular, the two colonies differed in their feeding schedule (see Section 2.1), a variable that has been recently identified as an important factor affecting replicability in animal studies (Madhusoodanan, 2022). Indeed, a recent study aimed to shed light on factors affecting inequity aversion in capuchin monkeys across laboratories suggested that the feeding regime is an important element to consider in cross-laboratory comparisons (Schweinfurth & Call, 2021). Furthermore, a previous study involving the IT and US capuchin colonies in two self-control tasks showed some differences between colonies: specifically, US capuchins overall showed better self-control abilities than IT capuchins. Three not mutually exclusive explanations were proposed for the observed difference (Addessi et al., 2013). First, US capuchins could see and hear other capuchins during testing, whereas IT capuchins did not; this may have provided US capuchins with more opportunities for distractions than IT capuchins, therefore enhancing their self-control performance. Second, although individuals belonging to both colonies had previously participated in numerous cognitive studies, only the US capuchins had long-term experience with computerized test systems employing joysticks, which necessarily implies a certain degree of behavioral inhibition. Third, IT capuchins had previously taken part in another self-control task and this earlier experience may have affected their performance. Although these specific factors may not be as directly relevant in the current study, experiential effects undoubtedly influence how capuchins make decisions in cognitive tasks (Leinwand & Brosnan, 2019). Also semi-random processes, like founder effects or genetic drift, may account for the above results, as well as the fact that animals previously classified as Cebus apella in laboratory colonies may belong to different species (e.g., S. apella, S. libidinosus, S. nigritus), which had been previously recognized as separate subspecies of C. apella (Fragaszy et al., 2004;Groves, 2001).
There is increasing evidence of differences in the feeding ecology of wild populations belonging to different Sapajus species, which may reflect their decision-making attitude in captivity. For instance,

S. libidinosus populations living in a semi-arid habitat in Notheastern
Brazil mainly feed on clumped and high-quality fruits all year round, whereas S. nigritus populations living in the Atlantic forest can switch to regularly distributed, low-quality leaves during periods of fruit shortage (Izar et al., 2012). As it has been demonstrated that, in captive settings, species mainly relying on clumped, seasonal fruitsas chimpanzees-are more risk-prone than species mainly relying on evenly distributed vegetable food sources-as bonobos (Heilbronner et al., 2008)-this may hold true also when comparing different Sapajus species. Future investigation should assess whether this is indeed the case.
In conclusion, our study provides insights on the complex relationships between body mass, energy state, and risk preferences.
Even though only a few experiments have been carried out so far and a wider number of species needs to be tested, species' body mass remains an important factor to take into account when considering whether and how individuals respond to energy budget manipulations (Kacelnik & Bateson, 1996). Furthermore, our results indicate how subtle differences in individuals' daily life experiences can play a role in shaping their decision-making strategies. Considering these aspects when comparing studies across laboratories, as well as a better standardization of housing conditions and experimental protocols, is crucial to obtain more reliable comparisons between populations (Farrar et al., 2021).