Towards an integrated understanding of dietary phenotypes

Abstract

Diet and nutrition comprise a complex, multi-faceted interface between animal biology and food environments. With accumulating information on the many facets of this association arises a need for systems-based approaches that integrate dietary components and their links with ecology, feeding, post-ingestive processes and the functional and ecological consequences of these interactions. We briefly show how a modelling approach, nutritional geometry, has used the experimental control afforded in laboratory studies to begin to unravel these links. Laboratory studies, however, have limited ability to establish whether and how the feeding and physiological mechanisms interface with realistic ecological environments. We next provide an overview of observational field studies of free-ranging primates that have examined this, producing largely correlative data suggesting that similar feeding mechanisms operate in the wild as in the laboratory. Significant challenges remain, however, in establishing causal links between feeding, resource variation and physiological processes in the wild. We end with a more detailed account of two studies of temperate primates that have capitalized on the discrete variation provided by seasonal environments to strengthen causal inference in field studies and link patterns of intake to dynamics of nutrient processing.

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

1. Introduction

Food and nutrition are central to many aspects of the ecological, evolutionary, organismal and applied branches of the biological sciences [1]. Not surprisingly, considerable volumes of research directly or indirectly examine the roles of diet and nutrition in biology, from molecular, cellular and systems physiology, to behaviour, morphology, life history and trophic interactions within populations and ecological communities. With the exponential growth in information, and in the technological means for gathering further information, has arisen an increasingly important need for systems approaches that interrelate relevant facets to obtain an integrated understanding of dietary phenotypes, their evolution, and consequences for the social and ecological communities within which they interact. The need for systems theory in biology has long been recognized [2], yet little attention has been paid to systems frameworks in the context of diet. This is a significant gap given the complexity of nutrition and its widespread relevance in biology and ecology [1].

Three decades ago, an integrative framework for nutrition based on Cartesian geometry was developed as a step towards tackling this challenge, called nutritional geometry. Introduced in a study examining the feeding choices and growth responses of locusts in the context of multiple nutrients [3], the implications for integrating across different levels of biological analysis (functional, mechanistic, ontogenetic and evolutionary) [4] were soon thereafter demonstrated [5]. The framework was subsequently applied in numerous laboratory studies involving a wide range of taxa to disentangle the behavioural and physiological responses of animals to diets varying simultaneously in two or more dietary components, mostly using systematically varied artificial diets [6–11].

Such laboratory studies, which offer powerful opportunities for establishing causality through specifically manipulating and controlling variables, have provided many advances in understanding the behavioural and physiological mechanisms used by animals to deal with variation in multiple dietary components and developmental, health and fitness consequences of constraints on those mechanisms. The trade-off, however, is that laboratory studies have limited capacity to establish the relevance of these mechanisms within the more complex ecological contexts in which they evolved and/or normally operate [12–14]. The nutritional geometry approach has thus been extended to observational studies of animals in the wild. These studies have principally involved primates, because they readily habituate to the presence of researchers, mostly have defined home ranges, and form stable social groups, all of which make it possible to collect detailed feeding data over timescales that would be difficult or impossible for many other taxa [15].

Our primary aim is to review some recent field studies on primates that demonstrate the application of nutritional geometry to examine the interactions of nutritional regulatory phenotypes with natural ecological variation in food and nutrient availability. Observational field studies are in many respects the most challenging application of nutritional geometry, partly because of the difficulties of inferring causation from observational data collected in complex ecological settings [16]. We, therefore, emphasize strategies to mitigate this problem and increase certainty around drawing causal inferences about mechanisms from studies of free-ranging animals in environments that are ecologically and evolutionarily relevant. We begin with a brief overview of nutritional complexity and how it is approached using nutritional geometry. Next, we briefly review studies of nutrient balancing in laboratory and field research. In the final section, we examine recent research on two temperate primates, emphasizing how the marked variation in nutrient supply and demand in seasonal ecologies has facilitated the examination of interactions of behavioural and physiological regulatory systems with natural resource variation.

2. Nutritional complexity and nutritional geometry

Nutrition is dauntingly complex [17]. Physiological systems require many nutrients, and their effective functioning can be impaired both by nutrient deficiencies and excesses. They are also impacted in positive and negative ways by non-nutrient components of diets, such as indigestible fibre, antioxidants and toxins. Furthermore, the complex multi-dimensional dietary requirements are not static but change with, for example, activity- or temperature-dependent changes in energy expenditure and through development. A challenge for the organism is to match nutrient supply to these complex dynamic requirements, through selectively eating and processing foods. This task is substantially complicated by the fact that most nutrients are not available independently in the environment but packaged as high-dimensional mixtures in foods (the ‘packaging problem’ [18]). Apart from specifically evolved nutritive mixtures like mammalian milk, very few foods contain the same mixture of components required by an animal, and even fewer have compositions that track the nutrient requirements of their consumers. Most species thus need to combine in the diet several or numerous foods to track their requirements. It is, however, neither biologically feasible nor in many circumstances ecologically possible, to eat foods in combinations that provide the required mix of nutrients [17]. Complex physiological mechanisms have thus evolved to buffer animals from discrepancies between nutrient requirements and the ingested diet, including selective digestion, absorption, interconversion, and storage of nutrients. In addition to such mechanisms for matching supply to demand, in many species behavioural and life cycle specializations have evolved for aligning requirements with nutrient availability, including hibernation and seasonal timing of reproduction.

Unravelling this web of complexity is an important priority. Upon it rests the answers to many unresolved issues in biology and ecology regarding, among other things, the impacts of ecological environments on the animal and the impacts of the animal's foraging and feeding choices on ecological systems [19]. Nutritional systems also underpin urgent applied challenges, for example those that arise when physiological systems are placed in altered ecological environments to which they are not adapted [20]. This is the case for human nutrition in industrialized human food systems [21], and the destruction of the habitats of many other species owing to climate change and other causes associated with human activities [22].

Nutritional geometry is a framework for examining the behavioural and physiological interactions of animals with food environments in graphical models, supported by numerical and statistical analysis. The foundation is a Cartesian space where each axis represents a nutrient (figure 1). Foods are represented as radials that project from the origin at an angle determined by the ratio of the nutrients it contains, termed food rails. A point or region within the graphical space depicts the amounts and ratios of the nutrients that the animal would target if not constrained by the quantity and quality of available foods, termed an intake target. Nutrient ingestion is shown as an intake point along the rail representing the food being eaten at a position determined by the amount of the food eaten. The animal can reach its intake target either by selecting a food that contains the nutrients in the target ratio, or by eating a combination of foods that have complementary proportions of the nutrients. Neither intake targets nor intake points need be considered static but can be modelled as trajectories defined by changing nutritional demand and supply, respectively.

Figure 1.

Geometric model of homeostatic responses to variation in the protein : non-protein energy ratios in resources. The large circle represents the nutrient intake selected by the animal when unconstrained (intake target, IT). Foods are represented as radial lines with angles determined by the balance of the nutrients each contains (nutritional rails). (a) The animal can reach its intake target by eating a balanced food (dashed rail) or composing a diet from two or more imbalanced but nutritionally complementary foods (dashed arrows). (b) Qualitative and quantitative resource constraint. In qualitative constraint, the animal is restricted to a diet that is imbalanced with respect to the intake target (circles) and is thus forced into a trade-off between over-ingesting one nutrient (e.g. NPE + represents surplus ingested non-protein energy) and under-ingesting another (e.g. P− represents a protein shortage). Green, red and blue circles show different responses to the same qualitative constraint. The square is an example of quantitative constraint, in which intake is limited by the amount of the food that can be acquired and processed and the animal thus suffers a shortage of both nutrients (as indicated by the truncated nutritional rail). (c) Patterns of intake diagnostic of the responses in (b). (d) ‘Pseudo constraint’, where changes in the position of the intake target (ΔIT) resemble a response to qualitative constraint, in this case the protein prioritization pattern of regulation (blue circles in (c)). From [23].

Various forms of ecological constraints and animal regulatory responses to constraints can be modelled as the relationship between the intake target, available food rails and intake points [24]. Quantitative constraint—where insufficient amounts of a food are available to meet the animal's needs—is represented as a truncated food rail (figure 1b). Qualitative constraint (i.e. constraint that concerns the quality of available resources) occurs when an animal has access to neither a balanced food nor complementary food combinations and so is forced to ingest a diet containing different proportions of the nutrients than it would normally select (figure 1b).

When qualitatively constrained, the animal is forced into a trade-off between falling short of its target for some nutrients and/or exceeding its target for others (figure 1b). The response adopted by animals in this situation, termed a rule of compromise (figure 1b,c), is an important component of the regulatory biology of animals that evolves for dealing with resource variation. First, it determines the nutritional context against which regulatory physiology is selected, for example, the capacity for interconverting, storing, and excreting different nutrients. Second, it can influence ecological communities through determining the range of environments the animal can inhabit hence its geographical distribution [25] and the pattern of resource use and nutrient recycling within those habitats [26]. Rules of compromise also play important roles in applied problems such as the obesity epidemic [27,28], climate change [29] and species conservation [20].

3. Nutrient balancing in the laboratory and field

Many studies have used the nutritional geometry framework to examine the drivers and consequences of dietary intakes in animals and humans. Most are laboratory studies using synthetic or semi-synthetic diets, but more recently several studies have examined the role of nutrient balance in the foraging choices of free-ranging animals. Here we provide a brief overview of some relevant studies, with the primary aim of setting the background for the following section in which we review research on two temperate primates that take steps towards combining benefits of laboratory and field studies of integrated homeostasis.

(a) . Laboratory experiments

Laboratory studies using nutritional geometry have shown that a principal goal of diet selection by animals is to balance the intake of specific nutrients—i.e. to select an intake target (figure 1a; figure 2a). This is achieved using nutrient-specific appetites, which provide a feedback loop linking current nutritional status (e.g. deficiencies or surpluses in protein nutrition) to the selection of foods that rebalance the diet (e.g. foods high or low in protein, respectively) [28], combined with other mechanisms such as learning [31,32]. Studies also show that the selected target changes to track specific changes in nutrient requirements (e.g. figure 2b). Several laboratory studies have shown that the selected diet maximizes important fitness components relative to alternative diets that vary from the target ratio of macronutrients, such as reproduction [33] and growth [34].

Figure 2.

Regulation of a protein : carbohydrate intake target by locusts. (a) Locust nymphs (Locusta migratoria) were given one of four food pairings that differed in the ratio and concentrations of protein and carbohydrate (% protein : % carbohydrate = 7 : 14 + 14 : 7; 14 : 28 + 28 : 14; 7 : 14 + 28 : 14; 14 : 7 + 14 : 28). Hollow squares indicate expected outcomes if the locusts did not regulate macronutrient intake but ate the same amount of each food in their respective food pairings. That the mean intakes of the four groups (solid circles, ± s.e.) converged indicates locusts ate different amounts of the food pairings to obtain the same macronutrient intake. From [20]. (b) Adult locusts made to fly for 120 min subsequently selected a diet higher in the principal energetic macronutrient carbohydrate than locusts flown for 20 min or unflown controls, whereas protein intakes did not differ significantly. From [30].

Many laboratory studies have confined animals to diets that are nutritionally imbalanced relative to the selected intake target to establish the rule of compromise the nutrient regulatory systems adopt to resolve the trade-off between over- and under-ingesting nutrients in circumstance of qualitative macronutrient constraint. A variety of rules of compromise have been observed, two of which are shown in (figure 3a,b). Studies have shown that the patterns of ingestive regulation are tightly integrated with post-ingestive physiological processes of nutritional homeostasis [1]. For example, excesses of nutrients ingested when confined to imbalanced diets are selectively excreted whereas deficient nutrients are retained and used with high efficiency (figure 3b,c).

Figure 3.

Responses to protein : carbohydrate imbalance by fifth instar nymphs of a dietary specialist (Locusta migratoria) and a generalist (Schistocerca gregaria) locust species. The hollow squares in (a) and (b) show the mean selected intake (as in figure 2) and solid points show mean ± s.e. intakes of one of five imbalanced diets varying systematically in the ratio of protein : carbohydrate. The curved array of intake points for the dietary specialist (a) shows that macronutrient intake was reduced relative to the intake target on nutritionally imbalanced diets, whereas for the generalist (b) intakes aligned along a negative diagonal of slope of −45° showing that total macronutrient intake was constant across diets. Consequently, the green-shaded area in (b) represents the additional macronutrient eaten in the generalist strategy compared with the specialist strategy. (c) Comparison of body composition in the two species fed the same diets as in (a) and (b). The comparison shows that the generalist, but not the specialist, was able to maintain body fat on imbalanced diets with excess protein relative to carbohydrate (% protein : % carbohydrate = 21 : 21, 28 : 14 and 35 : 7), suggesting that the generalist has greater metabolic flexibility for using imbalanced macronutrient mixes than the specialist. That body fat was maintained on an additional dietary treatment that comprised 42% protein with no digestible carbohydrate (not shown in (a) and (b)) demonstrates that the generalist can use excess ingested amino acids in energy metabolism. Modified from [35].

An important priority is to understand the ecological and evolutionary circumstances associated with different patterns of ingestive and post-ingestive regulation, and the roles played by the interactions of ingestive and post-ingestive regulation in adaptation of animals to specific nutritional niches. An example is the illustration in figure 3, regarding the relationship between macronutrient selection, post-ingestive processing of macronutrients and diet breadth. There is reason to believe that there is some generality in these relationships, as several other studies have shown that dietary generalist insect herbivores have a greater capacity to ingest and use nutritionally imbalanced macronutrient mixtures than related dietary specialists [1,35,36].

(b) . Field studies

Such results from laboratory studies, which have been confirmed for many species from slime moulds to mammals [1], give rise to the prediction that in ecological settings animals will likewise employ an integrated suite of nutrient balancing mechanisms to mitigate the effects on fitness of variation in nutrient supply. In recent years, nutritional geometry has been applied to examine this in observational studies of several species of free-ranging primates.

Felton et al. [37] found that the preferred food of Peruvian spider monkeys (Ateles chamek), figs, had macronutrient ratios that closely matched the ratios of the diets composed from combining complementary foods on days when figs were not available, as would be expected if this ratio represented an intake target. The diets of a baboon [38] and golden snub-nosed monkey [23] (figure 4a) maintained consistent ratios of available protein : non-protein energy (henceforth AP : NPE) over 30 and 32 consecutive days, respectively, despite being composed of different food combinations on different days. This situation, where nutrient intakes remain constant across different food combinations, is suggestive of intake target regulation (figure 2a). Similarly, the annual diets of mountain gorillas (Gorilla beringei beringei) in Virunga and Bwindi have similar fibre and macronutrient compositions, despite being composed of different foods [39,40]. Angola colobus monkeys (Colobus angolensis palliates) were found to maintain relatively invariant dietary AP: NPE despite eating different food combinations [41]. Martínez-Mota et al. [42] found that the daily macronutrient intakes of black howler monkeys (Alouatta pigra) fluctuated less in AP : NPE than in AP intake and NPE intake, which they concluded suggests that foods were selected to maintain a relatively constant macronutrient balance. The AP : NPE ratio in the diets of wild diademed sifaka lemurs (Propithecus diadema) remained remarkably constant across seasons and habitats (intact versus disturbed), suggesting macronutrient balancing [43]. Beeby et al. [44] inferred nutrient balancing in another lemur, the black-and-white ruffed lemur (Varecia variegata), whose diet varied in AP : NPE across seasons but, as the authors had predicted, across the year the AP : NPE was maintained more tightly. Johnson et al. [45] reported that guerezas (Colobus guereza) maintained a consistent AP : NPE when feeding across foraging patches, and the ratio was a good indicator of time spent foraging within patches.

Figure 4.

Macronutrient regulation in golden snub-nosed monkeys. (a) Cumulative intakes from feeding events (circles) recorded in full-day follows over 32 consecutive days (colours) of a wild monkey during spring. The green area denotes the range of food macronutrient ratios from which the daily diets were composed. (b) and (c) Daily intakes of captive and wild golden snub-nosed monkeys, respectively, across four seasons. The large variation in fat and carbohydrate compared with protein intake resembles the protein prioritization patterns of intake (figure 1b,c). From [23].

Studies have also used nutritional geometry to examine how non-human primates respond to ecologically enforced variation in the AP : NPE ratio of their diets. Most primates studied to date are reported to show protein prioritization, in which AP intake is maintained within narrower boundaries than NPE (figure 1). These include spider monkeys [37], black howler monkeys [46], Kenyan blue monkeys [47], capuchins [48], golden snub-nose monkeys [23] (figure 4b), orang-outans [49], black-and-white ruffed lemurs [44], small-bodied lemurs [50] and chimpanzees [51]. There are three known exceptions. Mountain gorillas maintain daily NPE intake more constant than AP, rhesus macaques regulate to constant energy intake across diets varying more than two-fold in the AP : NPE ratio (further discussed below), and in circumstances of fat and carbohydrate scarcity sifaka lemurs (Propithecus diadema) reduce intake overall to maintain the target AP : NPE balance [43].

4. Bridging the gap between the laboratory and field: a contribution from temperate primates

The convergence of results between laboratory and field studies suggests that nutrient balancing is an important part of the nutritional biology of animals that is relevant to foraging in natural contexts. There are, however, substantial challenges for field studies in examining integrated nutritional responses as has been achieved in the laboratory (figure 3). One challenge is to disentangle the many possible causes of the patterns of food and nutrient selection strategies observed in free-ranging primates. Another is to non-invasively measure the dynamics of physiological regulatory responses that, unlike behaviour, cannot be directly observed, such as the patterns of nutrient processing.

Temperate primates offer good opportunities for examining these issues. Compared with tropical and sub-tropical species, they live in relatively simple ecologies characterized by discrete and often extreme seasonal transitions in resource availability that can readily be identified and measured [52]. In addition to variation in nutrient supply, temperate primates typically experience marked and discrete variation in nutrient demand owing, for example, to strict seasonal breeding and changes in energetic needs for thermoregulation. Frequently, challenges owing to fluctuations in nutrient demand and supply compound, as is the case during temperate winters when low food availability coincides with high energetic demand for thermoregulation, and foraging time is constrained by the need for energy-conserving behaviours such as huddling [53,54]. Additionally, temperate primates often rely on a relatively low number of foods that provide an indispensable supply of nutrients and are vulnerable to inter-annual fluctuations in their abundance [55].

The discrete, compounding, and acute nature of resource challenges in temperate primates can provide ‘natural experiments’ that help to disentangle the drivers and consequences of nutritional strategies and narrow the gap between the causal strength of laboratory studies and the contextual relevance of field studies. We next explore this drawing on recent studies of two species of temperate primates.

(a) . Golden snub-nosed monkeys

Our first example illustrates a strategy for establishing causality in field studies of free-ranging primates, through selective manipulation of resource availability to disentangle the roles of ecological constraint from regulatory responses and examine their interaction. The study species, golden snub-nosed monkeys (Rhinopithecus roxellana), lives at the edge of the global distribution range of primates, in a temperate habitat characterized by cold autumn and winter temperatures and seasonal resource scarcity [53].

An important challenge in field studies of nutrient balancing is to distinguish homeostatically regulated intake targets from intakes that represent other processes, such as quantitative or qualitative constraint (figure 1) [23,24]. In laboratory studies, this typically is done in one of two ways. First, if equivalent groups of experimental animals provided with unique combinations of complementary foods select the same nutrient intake, this demonstrates that the selected point is a homeostatically defended target (e.g. figure 2a). A second approach is to demonstrate that the selected intake tracks changes in the animal's requirements for specific nutrients, as demonstrated by increased carbohydrate selection by locusts to compensate for energy expenditure during flight (figure 2b).

As discussed above, the first of these situations, in which the balance of nutrients eaten remains constant despite differences in food combinations eaten (e.g. on different foraging days, in different seasons, or across different habitats), has been observed in several studies of wild primates. Few field studies, however, have used the second approach, in which relationships are examined between nutrient selection and specific changes in nutrient requirements. Studies have reported increases in energy intakes and in changes in the proportions of different foods in the diets of free-ranging primates during lactation [41,56–58], but only one of which we are aware has specifically tested for changes in dietary macronutrient ratios during lactation [59] (further discussed below).

A recent study of the golden snub-nosed monkey explicitly tested whether selected nutrient intakes track specific changes in nutrient requirements associated with thermoregulation. Guo et al. [60] predicted that intakes by free-ranging monkeys of the principal energetic macronutrients, fat and carbohydrate, but not protein, would be higher during the cold winter than the mild spring, by an amount that approximated the additional thermoregulatory requirements during winter. The study used full day focal animal follows to measure dietary intakes, and close-range thermal images combined with biophysical equations to compare heat dissipation to the environment in winter and spring. By selecting a study population that in both seasons was able to mix its diet from a range of supplementary foods and natural foods, they were able to establish that any observed seasonal differences in intakes were owing to macronutrient selection, and not a seasonal constraint on nutrient availability. As predicted, no seasonal difference in the intake of protein was observed, and the winter diets were specifically higher in fat and carbohydrate compared with spring diets. Remarkably, the additional fat- and carbohydrate energy eaten very closely matched the additional energy lost through increased heat dissipation during winter (figure 5).

Figure 5.

Selected macronutrient intakes and heat loss by supplementary–fed golden snub-nosed monkeys during mild spring and cold winter seasons. The seasonal difference in fat and carbohydrate calories closely matched the difference in heat loss to the environment. Data from [60], as modified in [61].

While provisioning with supplementary foods in the study of Guo et al. [60] helped ensure that the observed ingestive responses were not due to ecological constraint but reflected an intake target, a consequence is that this study could not establish how the regulatory systems respond to the natural situation of quantitative and/or qualitative seasonal resource constraint. Hou et al. [23] addressed this question by examining a different population of golden snub-nosed monkeys in the same mountain range, in a study that differed in three important respects from Guo et al. [60]. First, the population studied by Hou et al. [23] was only minimally provisioned—sufficient to make observations more comparable to the provisioned population studied by Guo et al. [60], but not nearly sufficient to offset seasonal nutrient shortages. Second, unlike Guo et al. [60] which compared only winter and spring, Hou et al. [23] collected data year-round to understand winter constraint in the context of annual diets. Third, as an internal control Hou et al. [23] used the same observational methods to compare year-round intakes by the wild population with a captive population in an animal rescue and research centre fed domesticated foods and exposed to equivalent winter climatic conditions as the wild population.

The results of that study provided information relevant to both intake target regulation and the rule of compromise in golden snub-nosed monkeys. Considering intake targets, the captive population had higher fat and carbohydrate intakes during the cold seasons (autumn and winter) than the warm seasons (spring and summer), as expected given the demonstration that in this species dietary selection compensates for increased thermoregulatory requirements [60]. In the wild population, too, autumn and winter NPE intakes were higher than spring and summer, but winter NPE and total energy intakes were significantly lower in the wild than in captivity, suggesting that in natural circumstances the monkeys are seasonally constrained in energy availability. This resulted in the captive monkeys having higher annual energy intakes, but interestingly the macronutrient ratio of the annual diet did not differ between the captive and wild population. That the captive and wild populations composed diets with similar macronutrient balance despite being derived from very different food combinations (domesticated versus wild foods, respectively) is, as discussed above, strongly suggestive of intake target regulation. Overall, these results align with Guo et al. [60] in suggesting that macronutrient intake is regulated to track seasonal changes in energetic demand, but in natural food ecologies this capacity is constrained by fluctuations in resource availability.

The study provided interesting results concerning the ways the regulatory systems respond to qualitative resource constraint—i.e. the rule of compromise. In both the captive and wild populations, there was considerable variation in NPE intakes, but protein intakes were maintained within substantially tighter boundaries. Visual inspection and statistical analysis of the data for both captive and wild populations strongly suggested the protein prioritization rule of compromise (figure 4b,c). However, close inspection of the data revealed some nuances with important implications for studies of macronutrient regulation in the wild. Specifically, a substantial component of the variation in NPE intake was owing not to qualitative constraint, but to seasonal changes in the NPE dimension of the intake target as previously shown for this species [60] (figure 5). Indeed, analysis of the within-season and within individual variation in macronutrient intakes provided evidence for protein prioritization only in the captive population during winter and spring, and no evidence for protein prioritization in the wild population. Since rules of compromise are a measure of how regulatory systems respond to qualitative constraint on nutrient availability (figure 1), this highlights the need for vigilance in field studies against the risk of misinterpreting shifts in intake targets as rules of compromise, a situation which Hou et al. [23] referred to as ‘pseudo-constraint’ (figure 1d).

An important question is how the macronutrient mixtures targeted by animals in the wild and rules of compromise interface with physiological nutrient processing. Methods exist for non-invasively measuring physiological processes in free-ranging primates. For example, levels of urinary biomarkers, such as C-peptide [62,63], ketones [64], urea and nitrogen isotopes [65] provide information on energy balance, protein balance and fuel partitioning. Only one study of which we are aware, however, has employed these in the context of patterns of macronutrient regulation [49], but less-direct methods have been used to detect physiological processes in free-ranging primates. For example, as discussed above, Guo et al. [60] used thermal images and biophysical equations to estimate the energetic costs of thermoregulation in a supplementary-fed population of golden snub-nosed monkeys (figure 5). Hou et al. [53] combined this method with non-invasively measured seasonal changes in body weight (figure 6a,b) and equations for estimating daily energy expenditure derived from doubly labelled water studies of a diverse sample of 19 primate species [66] to construct energy budgets for the minimally—supplemented population of golden snub-nosed monkey discussed above. Results showed that in the winter the monkeys experienced a dietary energy deficit compared with calculated needs and lost 14% of body mass consistent with the conclusion from Hou et al. [23] that in the winter these monkeys are subject to quantitative macronutrient constraint (above). Calculations suggested that this deficit was closely offset primarily by consuming excess energy and accumulating fat stores during summer and autumn when energy was not limiting, combined with ancillary measures such as reducing heat loss and activity during winter (figure 6c).

Figure 6.

Seasonal energy budgets of wild golden snub-nosed monkeys constructed using non-invasive techniques. Body mass changes are measured using small amounts of corn to coax monkeys onto a digital balance (a), and thermal energy loss is measured using thermal imaging photography (b). Results (c) showed that daily macronutrient energy intakes (DEI) in summer and autumn exceeded daily energy expenditure (DEE) in those seasons. By contrast, in winter ingested macronutrient energy was less than energy expenditure (baseline daily energy expenditure (DEE) + additional daily energetic cost of thermoregulation (ADECT)), requiring energy drawn from fat metabolism (EDFM) to supplement dietary energy and meet daily energy requirements. From [53].

Overall, this field study shows how golden snub-nosed monkeys coordinate nutrient selection with behavioural and physiological strategies to subsist in an environment characterized by the confluence of resource limitation and low temperatures during harsh temperate winters. Importantly, the data suggested that despite this integrated homeostatic response, these monkeys live close to the threshold of what is energetically viable. This is consistent with the population existing at the northernmost edge of the species range. It also suggests that this population is vulnerable to perturbations such as unusually low winter temperatures and/or disruptions in resource supply [53].

(b) . Taihangshan rhesus macaques

The above programme of research on golden snub-nosed monkeys demonstrates how integrated studies of nutrient-specific regulation can be performed to understand the strategies used by primates to adapt to characteristics of their natural ecology, such as temperate seasonality. However, since both study populations were to some extent supplementary fed, the studies do not establish how regulatory phenotypes respond to a natural unmanipulated resource ecology. We now provide an example of unmanipulated field studies, where predictions based on theory and carefully chosen comparisons of subsets of the data were used to help establish relationships between biological regulatory responses and natural resource variation.

In common with the previous example, the research involved an edge-of-range population inhabiting a highly temperate habitat, Taihangshan rhesus macaques (Macaca mulatta tcheliensis). The ecology of this population is strongly influenced by a keystone resource, seeds of Quercus variabilis. These acorns are an essential source of dietary fat and carbohydrate, but vary substantially in their availability within seasons, across seasons and inter-annually [55].

From the perspective of nutritional ecology, rhesus macaques are a particularly interesting primate. They are the most widespread and ecologically diverse non-human primate, with a highly varied generalist omnivore diet that typically reflects whatever foods are common in their environment [67]. Based on nutrient regulation theory discussed above (figure 3), their varied diet led Cui et al. [59] to predict that rhesus macaques would show the generalist pattern of macronutrient regulation, in which total macronutrient (fat + carbohydrate + protein) intake is maintained constant at the level of the intake target regardless of the macronutrient mix of the available diet. This pattern has previously been found in generalist insects (figure 3), but not hitherto in any primate species. Results showed that this was indeed the case in Taihanghsan macaques, which had constant energy intakes across diets that ranged from approximately 12 to 30% of energy from protein (figure 7a).

Figure 7.

(a) Macronutrient intakes of Taihangshan rhesus macaques in four successive springs across which inter-annual variation in acorn availability drove variation in dietary macronutrient balance between 12% energy from protein (2013) to 30% (2014). Total macronutrient energy intake remained constant across this range but at a substantially higher level for lactating females (hollow circles) than non-lactating adults (solid circles). The pattern of intake by non-lactating animals is not due to quantitative constraint (e.g. shortage of food preventing animals in 2013 and 2015 from reaching the protein intakes shown in 2016 and 2014) because lactating females ate more, demonstrating that resources were available to non-lactating animals (redrawn from [55]). Additionally, consumption was not constrained by upper tolerance limits on non-digestible fibre (b), or tannin (c) intake as lactating females ate significantly more of these than non-lactating adults. From [59].

That this result conformed closely with an a priori theoretical prediction provides confidence that it is true. However, Cui et al. [55,59] derived additional evidence that it represents a response of the regulatory systems to qualitative dietary constraint (i.e. a rule of compromise) and not an artefact of changing nutrient targets (pseudo-constraint). This was achieved by comparing intakes of monkeys not between seasons when nutrient requirements are most likely to vary, as is done in some studies, but within the same seasons across five successive years of study that varied in acorn availability. In both spring (figure 7a) and winter (not shown), the two seasons when the monkeys experienced significant inter-annual variation in the macronutrient balance of their diets, energy intakes were constant across study years, demonstrating that when fat and carbohydrate energy were scarce the monkeys ingested commensurately extra protein to compensate. This contrasts with the protein prioritization response found in most primates studied to date, in which protein intake is maintained constant and fat and carbohydrate varies with the dietary P : NPE ratio (figure 4).

Furthermore, Cui et al. [55,59] were able to rule out other constraints as determinants of the observed pattern of intake, such as constraints on the amounts of food available across study years (i.e. quantitative constraint), or owing to inter-annual variation in non-nutrient substances, such as fibre and/or tannin, which can influence dietary intakes in primates [68,69], including some macaques [70]. This was achieved by establishing that lactating females ate substantially more food and macronutrients than non-lactating adult monkeys, thus showing that the pattern of macronutrient intake shown by non-lactating monkeys was not owing to limited availability of food. Additionally, the lactating monkeys ate significantly more non-digestible fibre and tannins (figure 7b,c), suggesting that diets of non-lactating monkeys were probably not influenced by limited capacity to ingest these non-nutrient substances.

Interestingly, in common with the non-lactating adults, lactating macaques likewise ate to constant energy across years that varied widely in the percentage of energy contributed by protein (figure 7a), but the energy intake was markedly higher during lactation. This is noteworthy in several respects. First, it provides a further example of selected dietary intakes tracking changing nutrient requirements. Second, it is the only known instance of which we are aware where the interaction of changes in macro nutritional requirements (due to lactation status) and ecological constraint (inter-annual variation in the availability of NPE) has been quantified in a free-ranging animal. Third, it provides further confidence that the generalist pattern of macronutrient regulation is a fundamental feature of the biology of these monkeys, occurring in both subsets of the population and persisting under the physiologically very different circumstances of lactation and non-lactation.

It is remarkable that the only known instance in a primate of macronutrient intake strategy that theory and empirical studies on insects predicts should be associated with dietary generalism occurs in an extreme generalist primate. We have not, however, as yet established the physiological component of the prediction, namely that the flexibility shown in terms of the macronutrient composition of ingested energy is associated with flexible fuel use (figure 3), such as an enhanced capacity to use excess ingested amino acids in energy metabolism (gluconeogenesis). Suggestively, however, a recent population genomics study of wild Chinese rhesus macaques found that Taihangshan macaques are distinguished from 16 other Chinese populations studied in having two genes that are associated with the upregulation of gluconeogenesis, Fbp1 and Fbp2, under positive selection.

While the upregulation of Fbp1 and Fbp2 in Taihangshan macaques aligns tightly with the physiological prediction of generalist theory, it raises new questions regarding the ecological and evolutionary circumstances that this theory postulates to be associated with the generalist behavioural (pattern of macronutrient ingestion) and associated physiological (flexible fuel partitioning) phenotype. Specifically, it suggests that the physiological component, at least, is not a species trait, but as yet detected only in a highly specialized temperate population, and therefore cannot easily account for the dietary and ecological diversity of the species as a whole. Furthermore, recent studies demonstrate that the Taihangshan macaques subsist on a relatively small number of foods, and thus do not conform to the expectation that dietary generalists should eat a diverse diet [71]. Rather, the most distinctive feature of their resource ecology is that they face substantial seasonal and inter-annual variation in the combinations of foods, hence macronutrients available [55].

Through integrating behaviour, physiology and ecology to understand nutritional homeostasis, this research programme has answered several questions regarding the nutritional ecology of rhesus macaques and identified priorities for further research. Primary among these priorities is the question of whether the constant energy rule of compromise is restricted to the Taihangshan population, is a species trait in rhesus macaques, or is found in some populations but not others. If the latter, this provides an excellent opportunity to examine within a single species the ecological factors that select for behavioural and physiological nutritional regulatory mechanisms, and how they are integrated to buffer fitness against ecological variation. Answering such questions will ultimately shed light on the theory of dietary generalism and help predict and manage the implications of global change for primates [23].

5. Conclusion

The challenge of understanding how nutritional biology interacts with ecology to generate outcomes that are important for both are substantial. Different facets of these interactions tend to be studied within different disciplines, defined by taxon (e.g. entomology versus primatology) and/or the level of biological organization (e.g. ecology, morphology, behaviour, physiology, evolution). Such research specialization provides benefits, but it also presents the challenge of how to understand the ways the components of organisms are integrated into functioning wholes. An essential requirement for this is an integrative framing that enables the key ecological and biological components and their interactions to be identified and empirically examined. We have discussed an approach for this, nutritional geometry, and shown how it has been applied in laboratory and field studies across taxonomic boundaries spanning insects to primates. It is, however, undoubtedly true that biological and ecological components are best studied in different contexts—laboratory and field, respectively. This presents challenges for studying their interaction, regarding the trade-off between establishing causality in the face of ecological complexity and maintaining realism in simplified laboratory systems. Using studies of temperate primates, we have attempted to demonstrate some steps that could be taken to bridge this gap, through strengthening causal inference in minimally or non-invasive field studies. While, arguably, this is more easily achieved in temperate contexts, we emphasize that by no means need such approaches be restricted to them. Finally, most field studies of primate diets have focussed on foraging and feeding behaviour; future studies should incorporate physiological measures to help understand how the different components of dietary phenotypes interact.

Data accessibility

This article has no additional data.

Declaration of AI use

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

Authors' contributions

D.R.: conceptualization, formal analysis, funding acquisition, methodology, supervision, writing—original draft, writing—review and editing; R.H.: conceptualization, investigation, project administration, supervision, writing—original draft; Y.D.: conceptualization, investigation, project administration, supervision, writing—original draft; C.R.: conceptualization, investigation, project administration, supervision, writing—original draft; Z.C.: conceptualization, investigation, methodology, project administration, supervision, writing—original draft.

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

This study was supported by the National Natural Science Foundation of China (grant nos 32370520; 32170507) and Top-notch Talents of Zhengzhou University (grant no. 32340438). The Society for Experimental Biology and The Company of Biologists provided session funding.