Warmer ambient temperatures reduce protein intake by a mammalian folivore

Abstract

The interplay between ambient temperature and nutrition in wild herbivores is frequently overlooked, despite the fundamental importance of food. We tested whether different ambient temperatures (10°C, 18°C and 26°C) influenced the intake of protein by a marsupial herbivore, the common brushtail possum (Trichosurus vulpecula). At each temperature, possums were offered a choice of two foods containing different amounts of protein (57% versus 8%) for one week. Animals mixed a diet with a lower proportion of protein to non-protein (P : NP, 0.20) when held at 26°C compared to that at both 10°C and 18°C (0.22). Since detoxification of plant secondary metabolites imposes a protein cost on animals, we then studied whether addition of the monoterpene 1,8-cineole to the food changed the effect of ambient temperature (10°C and 26°C) on food choice. Cineole reduced food intake but also removed the effect of temperature on P : NP ratio and instead animals opted for a diet with higher P : NP (0.19 with cineole versus 0.15 without cineole). These experiments show the proportion of P : NP chosen by animals is influenced by ambient temperature and by plant secondary metabolites. Protein is critical for reproductive success in this species and reduced protein intake caused by high ambient temperatures may limit the viability of some populations in the future.

This article is part of the theme issue ‘Food processing and nutritional assimilation in animals’.

1. Introduction

Ambient temperature is a known determinant of the feeding behaviour and physiology of endothermic herbivores [1–3]. The most conspicuous effect is an increase in food intake accompanying an increase in metabolic rate at cool ambient temperatures [4–6]. However ambient temperature can also influence the digestion, metabolism and excretion of nutrients and plant secondary metabolites (PSMs; [7–10]).

The link between nutrition and elevated ambient temperature has been best investigated in domestic species (e.g. [11]), particularly in intensive production systems such as dairy and poultry. In these production systems, livestock generally defer feeding during the hottest part of the day and decrease intake as temperatures increase [12]. This behaviour comes despite an increase in metabolic rate due to thermoregulatory demands and molecular stressors above the prescriptive zone and is often accompanied by reduced weight and productivity. Note that the prescriptive zone is wider than the thermoneutral zone (which is determined in fasted animals) and includes sustainable increases in metabolic rate and evaporative cooling [5,6]. Heat stress can also negatively affect energy balance and nutrient utilization including reduced lipolysis resulting in fat deposition [13]. Several dietary manipulations have been proposed to ameliorate the consequences of high temperatures including supplementation with lipoic acid to mimic insulin or with chromium to enhance its effects, electrolyte supplementation to counter losses in sweat, and alterations to the macronutrient composition of the diet [14,15]. The primary motivation for altering macronutrient composition in heat-stressed animals is to minimize heat from digestion and metabolism of food, which contributes to the internal heat load but also acts as a satiety signal.

Studies in both domestic and wild species indicate that foraging behaviour and diet composition influence diet-induced thermogenesis (DIT, also known as the specific dynamic action or post-prandial metabolic response). Larger meals generate more heat [3], and macronutrients produce different amounts of heat during digestion and metabolism. DIT is greatest for protein and least for fats, with soluble carbohydrate falling between. DIT from fermentation is variable and depends on particulars of digestive anatomy and physiology, with heat arising from both heat of fermentation and from the less efficient utilization of short-chain fatty acids compared to glucose [14,16,17]. How fermentation of fibre contributes to DIT and feeding in small hindgut fermenters like brushtail possums is unclear, although it is likely a more significant contributor to DIT in large foregut fermenters, in particular ruminants. In cold conditions, DIT from any source can be considered to be compensatory thermogenesis; i.e. it reduces the demand for endogenous thermogenesis. To maximize production under heat stress, diets are formulated for production animals that reduce thermogenic fibre and protein, while maintaining maximal energy intake [15]. The formulated diets used in intensive production systems do not allow animals to choose among foods to balance macronutrient intake or regulate thermogenesis, as wild herbivores would be expected to do.

Unlike domestic animals, wild herbivores can choose from diverse foods in a complex nutritional landscape to satisfy their nutrient and thermoregulatory requirements. As wild animals are also aiming to maximize production in the form of growth and reproduction, we might expect that similar changes in diet composition, including avoiding more thermogenic macronutrients when heat stressed, would be helpful to wild herbivores facing a heat challenge. However, detecting temperature-induced changes to nutrient selection by wild herbivores is complicated by changes in plant composition or nutrient availability that may occur seasonally or with temperature fluctuations and by the presence of plant secondary metabolites (PSMs) which may deter feeding and may have additional thermogenic consequences. Plant secondary metabolites can be as important as nutrients in driving the feeding decisions of herbivores [18,19], and metabolism of PSMs by herbivores can also be influenced by ambient temperature [3,20,21]. Temperature-dependent toxicity is a phenomenon whereby many xenobiotics, including some PSMs, are metabolized more slowly by endothermic vertebrates when they are exposed to them at warmer ambient temperatures, resulting in reduced intake of PSMs and hence of food containing PSMs [1,2,20,22].

Macronutrient and PSM intakes are unavoidably linked and cannot be viewed in isolation [23]. For example, PSMs can reduce the availability of nutrients (e.g. tannins binding to protein), and nutrients can either ameliorate the effects of PSMs or enhance a herbivores' capacity to metabolize PSMs [24–26]. Villalba et al. [27] demonstrated that lambs infused intraruminally with different PSMs varied in their overall food intake and macronutrient preference depending on the PSM. Sheep and goats increased voluntary intake of tannins when diets were also supplemented with protein and supplementary protein also positively influenced intake of shrubs such as sagebrush [28,29]. Increasing dietary protein concentration also allowed increased intake of cineole and benzoate by brushtail possums, in the latter instance by offsetting a protein cost of detoxification [25,30]. Therefore, an increase in ambient temperature is expected to reduce intake of protein which could further reduce tolerance to PSMs.

Here we investigate how a generalist marsupial folivore, the common brushtail possum (Trichosurus vulpecula), alters the macronutrient composition of its diet when exposed to different ambient temperatures with and without 1,8-cineole, a PSM encountered in its natural diet of Eucalyptus leaves. We offered possums a choice between two diets differing in protein and carbohydrate composition. We predicted that possums would not only eat less but that they would assemble a diet with a lower protein concentration under warm conditions than under cool conditions. Although we anticipated that the addition of cineole would reduce intake at all temperatures [31], we tested whether the need to manage DIT through lower protein intake would over-ride the detoxification benefits of higher protein intake. We predicted that when cineole was included in the diet, possums would ingest less food overall, but would mix a diet containing a higher proportion of protein compared to when cineole was not included and that this proportion would be lower at warmer temperatures.

Understanding how temperature drives changes in the selection of macronutrients, and temperature-PSM interactions, will contribute to better predictions of how herbivore populations will respond to hotter climates. More frequent and more severe heat waves are predicted across the entire range of brushtail possums, in addition to warmer average temperatures [32–34]. Furthermore, the nutritional quality of food trees is generally predicted to decline with higher PSM and fibre concentrations resulting from increased carbon dioxide, and heat stress and drought may result in lower leaf moisture [35–40]. Therefore, understanding how herbivores adjust their nutritional decision-making with changes to ambient temperature is imperative for conservation efforts.

2. Methods

(a) . Animals, housing and diets

Twelve adult male common brushtail possums (Trichosurus vulpecula) were caught in wire cage traps baited with apple on the Australian National University campus (Canberra, 35.2809°S, 149.1300°E) in November 2016. The age (by tooth wear, [41,42]) and condition of all possums was checked prior to the feeding experiments to ensure that only healthy adults were included. Possums were housed in cages (90 × 70 × 152 cm) with an open three-sided nest platform (18 × 18 × 34 cm). The cages were placed inside constant temperature rooms (±2°C) with a 12 : 12 h light cycle. Over a period of five weeks, possums were introduced to a compounded wet-mash diet of fruit and cereals (see below). This diet was prepared fresh daily and fresh water was always available ad libitum. Food was offered at 1630 h and refusals collected at 0900 h daily. Possums were weighed weekly.

Throughout both experiments, possums were offered a choice between two diets. Both diets contained 39% finely chopped apple, 29% finely chopped banana, 9% finely chopped carrot, 3% purified wood celluose (‘Just Fibre’, International Fibre Corporation North Tonawanda, NY), 10% ground rice hulls, 2% rolled oats, 2% ground lucerne hay, 1.47% vegetable oil, 0.25% salt, 0.25% dicalcium phosphate and 0.03% vitamin and mineral supplement, all on a wet matter basis. Casein and sucrose were substituted for each other to generate a higher protein diet (3.99% acid casein and 0.01% sucrose) and a lower protein diet (0.01% casein and 3.99% sucrose). The high-protein diet was coloured red with a food dye and always offered on the right-hand side of the enclosure. The low-protein diet was coloured blue and always offered on the left-hand side. The colouring allowed dropped food to be assigned correctly when calculating intake, and preliminary testing showed that the colour did not influence food choice (data not shown). The diets were consistently offered on the same side, as our primary goal was to test the influence of temperature on macronutrient composition rather than the ability of the possums to distinguish between the diets.

(b) . Chemical analysis of diets

The total nitrogen (N), and hence crude protein (N × 6.25), concentration of diets was determined using a LECO TruSpec CN analyser. Accelerated solvent extraction (Thermo Scientific, Sunnyvale CA) in a Dionex ASE350 accelerated solvent extraction machine was used to determine the fat content of diets. Each sample (0.5 g dry matter) was mixed with 2 g preparative diatomaceous Earth and extracted in a 10 ml stainless steel cell in petroleum ether (40–60°C bp). Samples were heated to 120°C for 6 min, followed by a 2 min static hold. Cells were flushed with 60% of cell volume and purged with nitrogen for 60 s before a second cycle. The extracted solution was collected into pre-weighed bottles, which were dried under nitrogen and then dried in a 102°C oven for 2 h prior to weighing. Neutral detergent fibre (NDF) was determined using an ANKOM fibre bag protocol (ANKOM Technology, Method 6). The non-structural carbohydrate content of the diets was calculated by subtraction of fibre, fat and protein from the total dry matter content.

(c) . Macronutrient choice at different ambient temperatures

Possums were randomly allocated into three groups of four, and each group was kept at a constant temperature of either 10, 18 or 26 ± 2°C for 7 days. These temperatures were chosen according to previous respirometry measurements in common ringtail (Pseudocheirus peregrinus) and brushtail possums indicating 10°C is below the thermoneutral zone, 18°C is around the lower critical temperature, and 26°C is around the upper critical temperature of the thermoneutral zone (C Cooper 2015, personal communication, [20,43]). Note that the concept of a thermoneutral zone is strictly applicable only to fasted animals and that for fed animals the ambient temperature range over which body temperature is stable is described as the ‘prescriptive zone’. However, the thermoneutral temperatures were the only relevant data available. Dry matter intake (DMI) of the two diets was measured between 1630 and 0900 h daily by drying a subsample of the food offered at 80°C (to determine the DM; 35% DM on average) and subtracting the dry mass of all food refusals. In addition to total daily DMI, we calculated the daily grams of protein (P) and ‘non-protein’ (NP; sum of fats and non-structural carbohydrates) ingested across 7 days for each possum for each temperature. The possums were then rotated between the temperature treatments for two more rounds of the experiment, using a randomized crossover design so the numbers of observations on each treatment was n = 12.

(d) . The effects of ambient temperature and cineole on macronutrient choice

All possums were housed at 18°C for one week prior to the experiment. Possums were then randomly allocated between two groups. One group was kept at 26 ± 2°C for seven nights, while the other was kept at 10 ± 2°C. Possums were offered the two diets differing in macronutrient composition without addition of any PSM for the first six nights. DMI and diet preferences were measured on the fifth and sixth nights and the mean was taken as an estimate of intake of the basal diets. On the seventh night 1,8-cineole was added to both diets at a concentration of 2.45% WM (approx. 7% DM) and DMI and diet preferences were measured again over two nights. The concentration of cineole was chosen based on a pilot study conducted prior to this study indicating that cineole at that concentration would reduce intake without preventing feeding completely.

(e) . Macronutrient comparison with the natural diet

The diets that were offered to brushtail possums were compared to the nutritional composition of leaves from 25 Eucalyptus melliodora trees within the home ranges of free-living possums at Black Mountain Peninsula, Canberra, Australia, using data collected by Marsh et al. [44]. This dataset included published data on the available nitrogen concentration of young and mature E. melliodora leaves and unpublished data on the soluble fraction lost during the available nitrogen assay as a proxy for non-structural carbohydrates.

(f) . Statistical analyses

Results were analysed in R (packages: lme4, lmerTest, emmeans, pbkrtest) and plotted (packages: ggplot2, wesanderson, gridExtra, plyr). For both experiments, the effect of ambient temperature on the ratio of protein to non-protein (P : NP) eaten was analysed using a generalized linear mixed model. In the first experiment, ratio of P : NP eaten was entered as the dependent variable and temperature was the independent variable. In the second experiment, temperature and presence or absence of cineole, and the interaction of these two factors, were included as the independent variables. Possum identity and experimental round were included as random variables with random intercepts for both models. The models were specified as having a γ distribution with a log-link. An ‘equal intake scenario’ was included in the model as an offset variable to determine if the possums chose randomly or non-randomly between the diets. This ratio of P : NP was calculated as if each possum ate the same amount of food as they did in the experiment, but chose equal amounts of each diet [45]. The model was compared to a model containing only the random effects using an ANOVA. Each pairwise comparison and P-values were gained using the package ‘emmeans’.

3. Results

(a) . Macronutrient choice and food intake at different ambient temperatures

Ambient temperature significantly affected intake of P : NP (${\chi }_2^2=20.57,$ p < 0.001). Possums chose a diet lower in P : NP at 26°C compared to either of the other two temperatures (table 1; figure 1).

Table 1.

Model estimates and each pairwise comparison of protein to non-protein (P : NP) intake by brushtail possums (Trichosurus vulpecula) for temperature each treatment and equal intake scenario in experiment 1. NS, not significant; ***p < 0.001. Of the random effects, possum identity accounted for 0.007 (s.d. = 0.08) and experimental round accounted for 0.0009 (s.d. = 0.03) of variance in the intake ratio, with a residual variance of 0.020 (s.d. = 0.14).

comparison	P : NP	s.e.	10°C	18°C	26°C
10°C	0.22	0.02
18°C	0.22	0.02	NS
26°C	0.20	0.01	***	***
equal intake	0.26	0.003	***	***	***

Figure 1.

Mean ± s.e. total protein versus non-protein (fats and carbohydrate) intake (grams of dry matter; g DM) by brushtail possums (Trichosurus vulpecula, N = 12) exposed to three ambient temperatures (10°C, 18°C and 26°C) for one week. The nutritional rails observed in the experiment are indicated by dotted lines; the dashed line indicates the expected nutritional rail if possums chose equally between the high and low protein diets. The two solid lines indicate the nutritional rail of each diet in isolation.

Temperature also significantly negatively affected DMI such that possums ate less at 26°C compared to either of the other two temperatures (${\chi }_2^2=54.21,$ p < 0.001; table 2; figure 1). Although there was a trend for possums to eat less at 18°C relative to 10°C (table 2; figure 1), the composition of the diet that they selected at these temperatures was the same (table 2; figure 1).

Table 2.

Model estimates and each pairwise comparison for dry matter intake (DMI) by brushtail possums (Trichosurus vulpecula) at each temperature treatment in experiment 1. NS, not significant; ***p < 0.001. Of the random effects possum identity accounted for 155.66 (s.d. = 12.47) and experimental round accounted for 17.60 (s.d. = 4.195) of variance in the intake ratio, with a residual variance of 80.78 (s.d. = 8.98).

comparison	DMI (g)	s.e.	10°C	18°C	26°C
10°C	87.13	4.75
18°C	83.90	4.75	NS (p = 0.08)
26°C	75.52	4.75	***	***

(b) . Macronutrient choice and food intake at different ambient temperatures with cineole

Cineole and temperature significantly influenced macronutrient selection by possums (${\chi }_3^2=43.13,$ p < 0.001). The diet selected in the presence of cineole was higher in protein relative to carbohydrates and fat (P : NP) than the diets selected in the absence of cineole (table 3; figure 2). The intake of macronutrients did not differ significantly between 10°C and 26°C in either the presence or absence of cineole (table 3; figure 2). When the basal diet was modelled in isolation, temperature had a significant effect on the macronutrient selected as in experiment 1 (${\chi }_1^2=7.33,$ p = 0.006). When the cineole diet was modelled in isolation, the effect of temperature on composition was non-significant (${\chi }_1^2=0.13,$ p = 0.91).

Table 3.

Model estimate and each pairwise comparison of protein to non-protein (P : NP) intake by brushtail possums (Trichosurus vulpecula) for each temperature treatment and diets with and without cineole in experiment 2. NS, not significant; P-value given, ***p < 0.001. Of the random effects, possum identity accounted for 0.01 (s.d. = 0.10) and experimental round accounted for 0.0009 (s.d. = 0.03) of variance in the intake ratio, with a residual variance of 0.02 (s.d. = 0.13).

	temp. (°C)	P : NP	s.e.	basal		cineole
				10°C	26°C	10°C	26°C
basal	10	0.15	0.01		NS	***	***
	26	0.14	0.01			***	***
cineole	10	0.19	0.02	***	***		NS
	26	0.18	0.02	***	***

Figure 2.

Mean ± s.e. protein versus non-protein (fats and carbohydrate) intake (grams of dry matter, g DM) by brushtail possums (Trichosurus vulpecula, N = 12) housed at two temperatures (10 and 26°C) for a week and offered diets with or without cineole. The nutritional rails are indicated by dotted lines; the dashed line indicates the expected nutritional rail if possums chose randomly between the high and low protein diets. The two solid lines indicate the nutritional rail of each diet in isolation.

Possums ate less when cineole was included in the diet (table 4; figure 2). However, when analysed in isolation, increasing temperature from 10°C to 26°C significantly reduced intake in both the basal treatment (χ²₁ = 3.86, p = 0.05) and the cineole treatment (${\chi }_1^2=5.18,$ p = 0.02).

Table 4.

Model estimate and each pairwise comparison of dry matter intake (DMI) by brushtail possums (Trichosurus vulpecula) for temperatures and diets with and without cineole in experiment 2. NS, not significant; p-value given, ***p < 0.001. Of the random effects, possum identity accounted for 95.54 (s.d. = 9.774) and experimental round accounted for 3.92 (s.d. = 1.980) of variance in the DMI, with a residual variance of 84.10 (s.d. = 9.171).

	temp. (°C)	DMI	s.e.	basal		cineole
				10°C	26°C	10°C	26°C
basal	10	92.14	3.88		NS	***	***
	26	86.54	3.88			***	***
cineole	10	54.37	4.48	***	***		NS
	26	47.66	4.48	***	***

(c) . Comparison with leaves available in the habitat

The P : NP ratios selected by possums at all three temperatures in our experiments would be achievable for wild possums if they were to select a diet from the young and mature leaves on the available E. melliodora trees within habitat at Black Mountain Peninsula (figure 3). The macronutrient composition (P : NP) of a diet only of unexpanded leaf tips was outside the range that possums could choose by mixing the two diets in our experiment.

Figure 3.

The nutritional composition (in terms of available protein and solubles, which are a proxy for non-structural carbohydrates) of diets chosen by common brushtail possums (Trichosurus vulpecula) after exposure to 10, 18 or 26°C for one week (black crosses) compared to the nutrient space available in leaves in the home range of possums. The dotted line indicates possible compositions given the two captive diets (black points). The polygons indicate the food trees available (Eucalyptus melliodora) within the home ranges of free-living brushtail possums. Polygon colours indicate mature leaves (orange), fully expanded young leaves (blue) and unexpanded leaves (yellow).

4. Discussion

The major finding in this study was that common brushtail possums modified their selection of macronutrients at higher temperatures, but this effect was abolished when a plant secondary metabolite was included in the diet. This result is an important finding because most previous studies have only focused on the effect of global climate change on the nutritional composition of plants [35–40], without recognizing that rising ambient temperatures may also affect the physiology underlying diet choice by herbivores. Available protein is a critical determinant of reproductive success in brushtail possums [46,47] and other herbivorous mammals [48,49]. The finding that even modest changes in temperature led to declines in the voluntary intake of protein demonstrates a mechanism by which a warming climate could negatively affect population recruitment.

The temperatures used in this experiment were based on respirometry data characterizing the thermal tolerance of possums (C Cooper 2015, personal communication, [20,43]). Possums have greater energetic requirements when exposed to 10°C compared to 18°C. At 26°C possum should have similar energetic requirements to 18°C (whether fed or fasted) but are limited by an inability to dissipate excess heat. Our experiment demonstrated that possums respond to the limitations imposed by warmer ambient temperatures by adjusting their macronutrient selection. At 26°C possums consumed less P : NP and less food (and hence energy) overall than at cooler temperatures. This finding is well aligned with results from domestic herbivores and is consistent with our hypothesis that herbivores should aim to minimize DIT by reducing P : NP, because the digestion and hepatic metabolism of protein is more thermogenic than the other macronutrients. In our experiment, possums, both absolutely and proportionately, reduced protein intake in warm conditions. This result is in direct contrast to how possums responded to cooler temperatures. At 10°C compared to 18°C, possums compensated for increased energetic costs only by increasing their total food intake without changing the macronutrient proportions of the diet. This finding reinforces the idea that at warmer temperatures possums choose to eat less protein due to limitations imposed by heat.

Much of what we know about the effects of heat stress comes from domestic animal studies. In these systems, chronic heat exposure shifts the use of energy substrates away from fatty acid oxidation (with the higher DIT this process generates) towards glucose use [8]. Of note, protein synthesis is reduced and protein catabolism is increased under chronic heat stress in cattle, pigs and rabbits [50–53]. Increased protein catabolism likely results in increased gluconeogenesis [8,54]. However, after chronic heat exposure, blood glucose is often reduced regardless of increased intestinal absorptive capacity, hepatic output and renal resorption, indicating heat stress causes negative energy balance [8,55]. Again this finding emphasizes that ambient temperature does not simply change energy requirements, but that nutrient uptake and utilization can also be altered by heat exposure. If animals are eating less protein in the heat to avoid DIT, as indicated by our experiment, increased protein catabolism and reduced protein synthesis are adaptive physiological responses. This finding also points to the main savings in terms of heat production arising from digestion. However, this finding also points to less available protein for growth and reproduction in wild herbivores that are reducing protein intake in response to warm ambient temperature conditions.

It is possible that other physiological factors could contribute to reduced protein intake at warmer ambient temperatures. For example, limits to the excretion of nitrogenous waste could limit protein intake at high ambient temperatures through a tradeoff with water use for excretion and water use for evaporative cooling. Nitrogenous wastes are generated from both dietary protein breakdown and hepatic metabolism of body proteins from skeletal muscle, which also tends to be increased by heat stress [50–53,56]. In herbivores, urea recycling to the gut, which as a source of non-protein N for microbial growth, is a major mitigator of excess water loss from excretion of PSMs and their metabolites. For example, common brushtail possums recycled 59% of endogenously produced urea and excreted a relatively small amount of urea compared to other herbivores on comparable diets [57]. When consuming diets low in protein, this recycled urea can represent a substantial fraction of urea produced [58] and the rate of urea returned to the gut depends on the rate of microbial fermentation in the gut [59]. In ruminants, high ambient temperature leads to no change in renal urea excretion (a relative increase as intake is decreased) but causes a decrease in urea recycling to the gut [60]. However, the effects of ambient temperature on urea recycling have not been investigated in small hindgut fermenters such as the brushtail possum. Further, the amount of DIT produced by fibre fermentation in small hindgut fermenters, and potential influence on feeding, is another area for future research.

Possums consuming a natural diet also contend with an array of PSMs. When the PSM 1,8-cineole was included in the diet, possums chose to eat less food overall but combined the two diets on offer to ingest more protein than the cineole-free treatments. Cineole causes an increase in protein turnover in brushtail possums and is a known feeding deterrent [25,31]. The shift towards a higher protein diet compared to the cineole-free diets can be explained by the fact that possums consume more cineole when supplemented with additional protein [25,30]. The elimination of a temperature effect on diet composition when cineole is added to food, even though a temperature effect on total food intake, remained notable. It is likely our experiment did not have enough power to detect the difference in P : NP with temperature (figure 2) for the basal diet treatments in the second experiment as the effect was relatively small compared to the influence of cineole. A difference in intake of basal diet when animals were housed at the same temperature was also observed between the two experiements. The reason for this difference is unclear although could reflect an impact of the smell of cineole in the temperature-control rooms depressing intake, or potentially an unidentified source of disturbance during the time of the experiment.

Because the addition of cineole to the food reduced intake below the level of the basal diet, the total DIT would also be much lower. In other words, even though possums chose a diet higher in P : NP when cineole was present, the absolute amount of protein ingested was much lower. Notably, 1,8 cineole is an agonist of TRPM8 cold receptors [61], which means possums ingesting 1,8 cineole may feel artificially cool. Treatment with TRPM8 agonists influences thermoregulatory behaviour and facultative thermogenesis in rodents in a similar way as if they were exposed to cool temperatures [61]. However, it remains unknown how this compound influences thermoregulation in marsupials. Therefore, while temperature may have an effect on macronutrient selection, this effect can be altered by PSMs in the diet, potentially including the specific actions of those PSMs.

We have shown that possums reduce P : NP selection with increased ambient temperature, but that PSMs can also influence the impact of temperature on P : NP selection. The diets that possums selected are within the range available in young and mature E. melliodora leaves in their natural habitat (figure 3). Our diets did not allow selection of a diet with similar macronutrient composition to unexpanded leaves; however, these leaves are limited in their availability. This consistency demonstrates that the choices that possums were making are ecologically relevant. Our study, like most captive animal studies, examined the effect of constant temperatures, in this case on nutrition. However, animals in the wild are exposed to complex thermal landscapes [19]. The ambient temperature experienced by an animal fluctuates constantly both temporally and geographically. To capture how animals respond to the complexity of a variable nutrient and thermal environment, we suggest that researchers also need to conduct studies with wild animals (e.g. [62]). Nevertheless, our work has demonstrated how wild herbivores faced with both complex nutrient landscapes and changing thermal environments might adjust their nutritional targets. In turn, this work gives us insight into how these adjustments might impact populations. The importance of protein for reproduction in animal populations is paramount. As stated by Provenza and Villalba [63] ‘Food and sex are basic for life, but of the two, food comes first. Without adequate nutrition, animals do not reproduce’. Reduced protein intake in response to warming climates could have significant detrimental impacts on animal populations. Protein intake affects everything from spermatogenesis and ovulation [64,65], to growth and survival of offspring, to timing of sexual maturation, sexual senescence and first parturition [66,67]. Protein restriction during pregnancy can also impact health later in life in progeny [68,69]. These changes in reproductive performance have flow-on effects to populations, and often what seems like a small difference in individual nutrition can result in dramatic demographic shifts [67]. These ‘amplifier effects’ mean that any change in nutrition due to warmer climates, like those demonstrated in our study, could have significant consequences in wild herbivores.

Ethics

This work was undertaken with approval of the Australian National University Animal Experimentation Ethics Committee. Protocol number A2015/24.

Data accessibility

These data are publicly available from the Dryad Digital Repository: https://datadryad.org/stash/dataset/doi:10.5061/dryad.ttdz08m47 [71].

The dietary analysis data are provided in electronic supplementary material [72].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors' contributions

P.B.: conceptualization, formal analysis, investigation, visualization, writing—original draft, writing—review and editing; W.F.: funding acquisition, project administration, supervision, writing—review and editing; B.M.: investigation, supervision, writing—review and editing; K.M.: conceptualization, investigation, methodology, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

The study was funded through grants from the Australian Research Council to K.J.M. (grant no. DE120101263) and W.J.F. (grant no. DP140100228).