Insects are the most successful group of organisms on earth, constituting more than 75% of all described animal species (Gullan and Cranston, 2014). Along with other explanations, one potentially important aspect of insects that can explain their success is their remarkable ability to exploit a vast variety of food resources (Chown and Nicolson, 2004;Harrison et al., 2012). Insects are particularly well known for their rapid adaptations to novel food resources, which could be greatly facilitated by their simple but highly effective digestive system (Chapman, 2013). In this regard, insects have always lain at the heart of nutritional ecology which mainly explores the nutritional nexus between an organism and the environment at physiological, ecological, and evolutionary levels (Mattson, 1980;Scriber and Slansky, 1981;Slansky, 1993;Raubenheimer et al., 2009;Simpson et al., 2015a). Over the last three decades, we have witnessed an unprecedented expansion of our knowledge on how nutrient affects biological processes, organismal phenotype and more ultimately Darwinian fitness in insects. This advancement was greatly indebted to the development of a novel and integrative paradigm in nutritional biology that is called the “Nutritional Geometry” (henceforth, NG; Simpson and Raubenheimer, 2012). In this review, we aim to present an overview of the recent developments made in the field of insect nutrition, with a particular emphasis on those recent discoveries made by the application of this new approach. This review consists of three main parts. We first start by introducing the background of why and how this new modelling approach was conceived and developed, as well as its basic concepts. Then, we show some recent examples of how NG has revolutionized insect nutrition and integrated nutritional physiology, physiological ecology, and evolutionary ecology into a common framework. Finally, we discuss how this novel approach can further broaden our understanding of the role play by nutrition in orchestrating biological interactions and also how the knowledge gained from NG has implications for more applied research areas in the future.

Nutritional Complexity

Nutrition is the one of the most important determinants of individual fitness, population dynamics, and geographic distributions in all organisms. It also mediates the trophic relationships in the food web, linking different individuals, species, and communities in the ecosystem (Simpson and Raubenheimer, 2012). Despite its fundamental importance, nutrition is notoriously difficult to tackle because of its complex nature (Simpson and Raubenheimer, 2012;Simpson et al., 2015b). There are three key aspects of nutrition that make it highly intractable. First, nutrition is multidimensional in nature. Foods are complex mixtures of many different macro- and micronutrients that have very different biological functions. For instance, carbohydrates are the major substrate for metabolic energy, whereas proteins serve as the major building blocks for somatic tissues and enzymes, and also can be used to fuel metabolism. Through food consumption, animals thus have to handle different mixtures of multiple nutritional factors simultaneously, not one by one. These features suggest that the process and outcome of food consumption should be viewed as multivariate dataset. Second, the effects of multiple nutrients are not independent but often interact with each other in an extremely complicated manner. There is mounting evidence suggesting that the phenotype of an organism is influenced not simply by the quantity of a single nutritional factor but more profoundly by the balance among more than two of these factors (protein, carbohydrate, or lipid). Hence, the nature and strength of the effect of a single nutrient is often contingent upon the presence and quantity of other components in the same food. Third, nutrition is stochastic and dynamic in terms of both availability and demand. In the food environment, the distribution and abundance of nutrients are highly unpredictable and variable in time and space, indicating that securing sufficient nutritive sources for maintaining fitness is always challenging for animals in nature. Further adding to this complexity is that the amount and balance of multiple nutrients required by an animal are not fixed but change over physiological, developmental, and evolutionary timescales (Simpson and Raubenheimer, 1993;Raubenheimer and Simpson, 1999;Raubenheimer et al., 2009). Taken all these points into account, it is apparent that the traditional approaches in nutritional sciences, which are largely based on a unitary nutritional currency, cannot appropriately analyze the multivariate and interactive nutritional dataset (Simpson et al., 2015b). This calls for the need to develop a unifying and integrative framework to address nutritional complexity.

Nutritional Geometry: Central Concepts

Traditionally, the investigation on the impact of nutrition on performance and fitness has been carried out by taking “one variable at a time” (OVAT) approach where the effect of each nutritional component (e.g., proteins, carbohydrates, lipids, sterols, minerals, vitamins, etc.) was examined separately and independently of each other (Simpson et al., 2015b). Although it has been successful in providing some useful insights, this simplistic approach has major limitations in accommodating the multivariate nutritional dataset and disentangling the complicated interactions among multiple nutrients. The NG was first developed by Stephen Simpson and David Raubenheimer to overcome these obstacles (Simpson and Raubenheimer, 1993; reviewed by Simpson and Raubenheimer, 2012;Simpson et al., 2015b). The main key to understand this NG framework is that it is a state-space model that allows the visual representation of the multivariate nutritional variables, such as the current nutritional status of an organism and the nutritional composition of the food, as points and lines in multidimensional Cartesian coordinates. Unlike traditional approaches in nutritional biology, NG made a major innovation in the field by highlighting the balance between nutritional components as the most important and influential dietary determinant of nearly all aspects of biology in animals.

In this model, there are three key concepts: 1) nutrient space; 2) nutrient rail; and 3) intake target (Simpson and Raubenheimer, 1993). The nutrient space is a two-dimensional space in which each axis represents the gradient of each of two nutritional components of interest, such as protein and carbohydrate (Fig. 1). In this space, the amount of each nutritional component ingested by an animal can be described as a point. The nutrient rail is a line radiating from the origin (Fig. 1) and represents the relative proportion of nutrients in the food consumed by an animal. Accordingly, the slope of this projecting line corresponds to the ratio of two nutritional components in the food. Finally, the intake target is also a specific point in a nutrient space, representing the optimal amount and balance of different nutritional components that must be ingested by an animal to achieve the maximum performance or fitness. The intake target of an animal can be reached either by consuming a food that comprises a balance of nutrients that perfectly matches the animal’s ideal requirement (Fig. 1A) or by selectively choosing between two nutritionally imbalanced but complementary diets (Fig. 1B). However, when restricted to feed on a nutritionally imbalanced diet, an animal is constrained from reaching its intake target and so has to choose one of following options (Fig. 1C). If an animal prioritizes one particular nutrient (protein in case 1 and carbohydrate in case 2 in Fig. 1C) over the other, it will regulate its intake to reach the target level of the nutrient that is prioritized, but will incidentally over-eat (carbohydrate in case 1) or under-eat (protein in case 2) the other nutrient. Alternatively, an animal may make a compromise by regulating its intake to a position somewhere between the two points that are reached if either protein or carbohydrate is prioritized (case 3 in Fig. 1C).

Nutrient Regulation and Balancing

Since its development in the early 1990’s, the NG has been proven as a powerful tool for identifying the nutrient regulatory responses in a wide variety of insects from different feeding guilds. The fact that insects selectively forage on food based on their nutritional requirements has been well described (Simpson and Simpson, 1990;Waldbauer and Friedman, 1991). Through the lens of NG, however, it has become even more clear that, regardless of their trophic level in the food web, most insects have strong capacity to achieve their optimal balance of multiple macronutrients (protein, carbohydrate, lipid) through actively and independently regulating their macronutrient intake (Maytnz et al., 2005;Behmer, 2009;Simpson and Raubenheimer, 2012). Over the last three decades, there have been a growing number of studies reporting the self-regulated intake targets of taxonomically diverse insect species belonging to the order Blattodea, Orthoptera, Coleoptera, Lepidoptera, Hymenoptera, and Diptera (Fig. 2). The self-regulated intake target ratio of protein to nonprotein energy source (carbohydrate, lipid) varies considerably among species, but it is generally viewed that most of these interspecific variations in the intake target are largely attributed to differences in feeding ecology and life-histories among species (Simpson and Raubenheimer, 1993). For example, the self-regulated ratio of protein-to-carbohydrate (P:C) has been shown to be higher in fast-growing and sedentary larvae of holometabolous insects (e.g., caterpillars) than in slow-growing and highly mobile nymphs of hemimetabolous insects (e.g., grasshoppers). Another notable example is that the insects with endosymbiotic microorganisms providing essential amino acids (e.g., aphids) generally prefer a diet with extremely low P:C ratios compared to those without (Abisgold et al., 1994).

The position of the intake target is not static but can move around the nutrient space depending on insect’s developmental stage and age (Simpson and Raubenheimer, 1993;Paoli et al., 2014), physical activity (Raubenheimer and Simpson, 1999), sex and mating status (Lee, 2010;Lee et al., 2013;Bowman and Tatar, 2016;Camus et al., 2018), body or ambient temperature (Lee et al., 2015;Schmitz et al., 2016;Rho and Lee, 2017), and feeding history (Mayntz et al., 2005;Raubenheimer and Jones, 2006;Lee et al., 2012). For example, Lee et al. (2013) observed that female Drosophila melanogaster flies significantly increased their self-selected P:C intake ratio from 1:4 to 1:1.5 after they were stimulated to produce eggs by mating. This shift in the intake target reflects the increased demand for protein during egg production in reproductively active female flies. Even more intriguingly, some recent studies have begun to unveil the fact that insects can adaptively adjust their nutrient preference in response to the risks caused by pathogen/parasites (Lee et al., 2006b;Povey et al., 2009;Shikano and Cory, 2016) and predators (Hawlena and Schmitz, 2010). When exposed to spider predators, for example, Melanoplus femurrubrum grasshoppers significantly increased their carbohydrate intake but without altering their protein intake. This had led grasshoppers to shift their self-regulated P:C intake ratio from 1:1 to 1:1.6 in response to predation risk. This change in nutrient preference is explicable as the grasshopper’s compensatory response to meet the increased requirements for dietary carbohydrates to fuel elevated metabolic rates induced by physiological stress of predation (Hawlena and Schmitz, 2010).

The regulation of nutrient intake or nutrient balancing occurs even when an insect is constrained from reaching their intake target. As we have briefly mentioned earlier, when confined to a single food that is nutritionally imbalanced in terms of protein and carbohydrate content, an insect has to make tradeoffs between the over-ingestion of one of these nutrients and the under-ingestion of the other, relative to what it would have selected. The global pattern of nutrient balancing has been examined by visually inspecting how the regulated protein- carbohydrate intake points on an array of nutritionally imbalanced diets are aligned across the nutrient space. Numerous studies have reported that the manner in which the intake points are aligned strongly correlates with insect’s feeding ecology, such as the probability of encountering nutritionally complementary foods in the food environment (see Lee et al., 2002, 2003, 2006a;Raubenheimer and Simpson, 2003). The most striking example of how insects differ in their nutrient balancing comes from the comparison made between herbivorous and carnivorous species (Simpson and Raubenheimer, 2012). When restricted to single nutritionally imbalanced diets, many herbivores have demonstrated a propensity to prioritize the intake of protein over that of non-protein energy source (carbohydrate or lipid) and this reaction norm is graphically illustrated by a near vertical alignment of intake points as described in Fig. 3A. This regulatory tendency of herbivores to secure their target protein intake above non-protein energy is likely to have evolved as adaptions to food environments where protein is limited. Under this protein-prioritization rule, the insects that are confined to a carbohydrate-biased diet are predicted to consume large excess of carbohydrate before reaching their target protein intake. In a manner opposite to herbivores, no matter whether they are insects or not, many carnivores tend to regulate the intake of non-protein energy source more tightly than protein, leading them to be more prone to over-eat protein when constrained on protein-biased diets. This ingestive response generally described for carnivores can be visually represented as a near horizontal configuration of intake points seen in Fig. 3B (Raubenheimer et al., 2007;Ruohonen et al., 2007;Mayntz et al., 2009;Jensen et al., 2011, 2012;Al Shareefi and Cotter, 2019). This nutrient balancing strategy is likely to have evolved in carnivores as a result of their adaptions to foods abundant in protein but deficient in non-protein energy source and reflects that carnivores are probably more adept at processing protein surpluses than herbivores.

Nutritional Fitness Landscapes

Nutritional ecology mainly concerns how animals respond to the quantity and quality of variable nutrients found in their environments within and across generations (Raubenheimer et al., 2012). There have been a multitude of studies documenting the effect of single nutritional component on various metabolic, physiological, behavioural, and life-history traits. However, no conventional methods in the field of nutritional ecology have enabled the integrated analysis of the combined and interactive effects of more than two nutritional currencies. The application of NG has overcome this limitation by envisaging the performance outcome of the intake of multiple interacting nutrients as a topographical map superimposed on the nutrient space, which is termed as “the nutritional fitness landscape (nutritional landscape, henceforth)”. In this landscape, the phenotype of a focal trait expressed across a wide range of dietary macronutrients, say protein and carbohydrate, can be mapped onto the nutrient space where the gradients of protein and carbohydrate are represented on the x- and y- axis, respectively (Fig. 4). The region in the nutritional fitness landscape where the trait phenotype is expressed at its maximum is identified as its nutritional optimum (Simpson et al., 2004). In theory, the self-regulated intake target is anticipated to be located on the summit of the fitness mountain, and this has been experimentally verified in a number of insects (Simpson et al., 2004;Lee et al., 2008;Jensen et al., 2012). Only after through mapping nutritional landscapes, we were finally able to have a strong analytical power to disentangle the complicated interactions between multiple nutritional components (see example in Lee et al., 2008).

The application of multidimensional nutritional landscapes has greatly deepened our understanding of the complex effects of multiple nutritional components and is now successfully embraced as a standard methodology by many different disciplines beyond entomology, which include human health and medicine (Raubenheimer and Simpson, 2016;Simpson et al., 2017), collective behaviour and social interactions (Lihoreau et al., 2014;2015), microbial ecology (Dussutour et al., 2010;Shik et al., 2021), mathematical and statistical modelling (Rapkin et al., 2018;Morimoto and Lihoreau, 2019), etc. There is also a vigorous discussion on the possibility of expanding the use of this nutritional landscape approach to model the multidimensional fundamental nutritional niche (Behmer and Joern, 2008;Machovsky-Capuska et al., 2016).

However, one of the most significant scientific contributions made by the application of nutritional landscape is the way in which it has resolved the long-standing debate over the lifeextending effects of dietary restriction. Moderate reduction in caloric intake has long been believed to be mainly responsible for extending lifespan (Masoro, 2005;Sinclair, 2005). However, there has been also growing argument coming from D. melanogaster gerontological studies, which contends that this well-known dietary intervention in aging is actually driven by the amount and mixture of specific nutrients consumed rather than the caloric intake itself (Mair et al., 2005;Min et al., 2006). To resolve this debate, Lee et al. (2008) designed a large-scale feeding experiment where 1,008 female D. melanogaster flies were restricted to feed on one of 28 semi-synthetic diets that systematically varied in P:C ratio and in total caloric content throughout their adult lives. When the lifespan of 1,008 flies was fitted as nutritional landscape mapped over protein-carbohydrate intake (Fig. 5A), it was clearly revealed that the pattern of fly’s lifespan was predominantly explained by the dietary balance of protein and carbohydrate, but not by total amount of calories consumed. This result was later replicated by a number of other studies in insects (e.g., Maklakov et al., 2008;Fanson et al., 2009;Jensen et al., 2015;Jang and Lee, 2018) and also in vertebrates (Solon-Biet et al., 2014, 2015;Moatt et al., 2019), providing strong evidence that the profound effect of macronutrient intake on lifespan is evolutionarily conserved (Simpson and Raubenheimer, 2009;Piper et al., 2011;Nakagawa et al., 2012;Tatar et al., 2014;Le Couteur et al., 2016;Simpson et al., 2017;Moatt et al., 2020).

Nutrient-mediated Life-history Trade-off

Life-history traits are a suite of characteristics that define the life table of an organism, such as growth rate, fecundity, body size at sexual maturity, longevity, etc. The importance of life-history traits in evolutionary biology lies in the fact that they are the key components of Darwinian fitness (Stearns, 1989, 1992;Roff, 2002;Flatt and Heyland, 2011). A central tenet in evolutionary biology is that there is a fundamental trade-off between these life-history traits, which is indicated by a negative covariance between any of these fitness components. The most well-known example of this trade-off occurs between lifespan and reproduction. For instance, it has been repeatedly demonstrated that, no matter whether it was genetically selected or environmentally mediated, an increase in early-life reproduction is associated with shortened lifespan in a wide range of organisms (Partridge et al., 2005;Flatt, 2011). According to the conventional resource allocation model, this trade-off between lifespan and reproduction is likely to arise because these fitness components are competing for a limited pool of internal resources (Stearns, 1992;Roff, 2002). As such, an increased investment of resource into reproduction inevitably results in a decreased investment of resources into somatic maintenance, leading to shortened lifespan (Kirkwood, 2005).

This classical view has been however challenged by a series of recent experimental studies which raise the possibility that the appearance of a trade-off between lifespan and reproduction may also arise because these traits require completely different composition of macronutrients for their maximum expression (Lee et al., 2008;Simpson and Raubenheimer, 2012). In this case, nutritional landscape has been instrumental in identifying the occurrence of this so called “nutrient-mediated life-history trade-off (Morimoto and Lihoreau, 2019)” or “nutrient-space based life-history trade-off (Rapkin et al., 2018)”. The very first study that leveraged the nutritional landscapes to tackle this issue was done by Lee et al. (2008) as described earlier. Lee et al. (2008) found that lifespan and the rate of egg production were maximized at different dietary P:C ratio. While lifespan was the longest at the carbohydrate-biased P:C ratio of 1:16 (Fig. 5A), egg production rate peaked at the P:C ratio of 1:2 (Fig. 5B). Furthermore, there were also opposing effects of dietary P:C ratio on these two traits. For example, an increase in dietary P:C ratio resulted in increased egg production in females but at the expense of decreased lifespan. Increased egg production in female flies upon consuming high protein diet reflects substantial protein investment into oogenesis (Wheeler, 1996;Mirth et al., 2019), but the mechanisms underlying the life-shortening effect of high protein intake still remain elusive. It is suspected that the lifespan-shortening effect of high protein intake is potentially mediated through increased production of mitochondrial reactive oxygen species, toxic effects of elevated nitrogenous waste products, changes in immune function, and/or altered nutrient-signaling pathways (e.g., insulin/insulinlike growth factor-1, target of rapamycin signaling pathways) (Sanz et al., 2004;Kapahi et al., 2010;Le Couteur et al., 2016). Whatever the mechanism behind this nutrient-mediated life-history trade-off, similar pattern of divergence in nutritional optima between lifespan and the measure of female reproduction (i.e., egg laying) was also observed in a number of studies using other D. melanogaster flies (Skorupa et al., 2008;Bruce et al., 2013;Reddiex et al., 2013;Jensen et al., 2015;Lee, 2015;Jang and Lee, 2018;Semaniuk et al., 2018;Kim et al., 2020), Bactrocera fruit flies (Fanson et al., 2009;Fanson and Taylor, 2012), crickets (Maklakov et al., 2008;Harrison et al., 2014), and rodents (Solon-Biet et al., 2014;2015), with all consistently showing that the optimal P:C ratio for lifespan is much lower than that for female reproductive performance. Collectively, the message derived from these results is clear that there is no single macronutrient composition that simultaneously satisfies the maximum expression of both traits.

It is also important to note that this nutrient-mediated lifehistory trade-offs is not restricted to the relationship between lifespan and reproduction, but also can apply to those among other components of fitness (e.g., body size at maturity, development time, immune function, etc.) because the shape of their nutritional landscape is known to differ significantly from each other. For example, numerous studies have indicated the possible nutrient-mediated life-history trade-offs between various fitness-related traits expressed during the juvenile stage (e.g., preadult survivorship, development rate, body size) (Matavelli et al., 2015;Rodrigues et al., 2015;Shingleton et al., 2017;Silva-Soares et al., 2017;Gray et al., 2018;Jang and Lee, 2018). Interestingly, different components of immunocompetence (e.g., phenoloxidase activity, lysozyme activity, hemolymph protein) were shown to respond to dietary protein and carbohydrate qualitatively differently in Spodoptera littoralis caterpillars (Cotter et al., 2011), suggesting that macronutrient intake also mediates the trade-offs between different components of insect immune function.

The presence of nutrient-mediated trade-offs between two or more components of fitness can be verified by visually inspecting whether or not the nutritional optima for these traits are located in different regions in the nutrient space, but NG has also offered us an opportunity to infer how strongly the negative covariance between these traits is nutritionally determined. In theory, the strength of the nutrient-mediated life-history trade-offs between two traits of interests can be quantified by calculating the Euclidean distance (d) and angle (θ) between the two position vectors projected from the origin to the nutritional optima for each of two traits (Fig. 6). As exemplified in Fig. 6A, a strong trade-off is expected if the nutritional optima for two traits are placed far away from each other in terms of the distance and the internal angle. On the other hand, if the two optima are located close enough to each other so that they are not statistically distinguishable (Fig. 6B), the trade-off is predicted to be absent or weak at best. Currently, there is an ongoing endeavour to develop a robust mathematical tool for making statistical inference about the occurrence and strength of nutrient-mediated life-history trade-offs (Rapkin et al., 2018;Morimoto and Lihoreau, 2019).

Conclusions and Future Directions

Our review demonstrates how the novel nutritional framework of NG has developed over the last decades and has revolutionized the field of insect nutritional physiology through fostering its integration with evolutionary ecology and life-history theory. Over the recent years, we have clearly seen that the utility of NG has rapidly expanded across taxonomic boundaries and different scientific disciplines, including public health, medicine, aquaculture, and so forth. In particular, NG has played a key role in providing a new perspective on the dietary interventions in aging (Simpson and Raubenheimer, 2009) and the origin of obesity and other metabolic disorders in humans (Simpson and Raubenheimer, 2005;Gosby et al., 2014). From a fundamental biological perspective, there are four promising research agendas that can benefit by applying NG: 1) the effect of the quality of macronutrients (amino acids, carbohydrate) on healthy aging (Piper et al., 2017;Solon-Biet et al., 2019;Wali et al., 2021); 2) the nutritional mediation of host-microbiome interactions (Holmes et al., 2017;Leulier et al., 2017;Simpson et al., 2017); 3) the gene-by-nutrient interactions/personalized nutrition (Camus et al., 2017;Nagarajan-Radha et al., 2019); and 4) the transgenerational epigenetic effects of nutrient intake (Bonduriansky et al., 2016;Polak et al., 2017;Koemel et al., 2022). We also predict that NG will help us in addressing some emerging issues in ecological and evolutionary research: 1) the impact of climate change on trophic interactions (Rosenblatt and Schmitz, 2016); 2) the interactions between diet and other environmental factors (Kutz et al., 2019;Alton et al., 2020); and 3) the long-term adaptations of animals to changing nutritional environments (Warbrick-Smith et al., 2006;Cavigliasso et al., 2020).

From an application point of view, a fertile area of research where NG can be practically implemented is the industrial production of beneficial or economically important insects. In recent years, there has been a rising demand for alternative protein sources for animal feed and also for human consumption (van Huis, 2013;2020). Insects have emerged as a promising solution to this challenge because they are not only highly nutritious but also cost-effective and environmentally-sustainable to produce (van Huis, 2013;Oonincx, 2017). In order to establish an efficient and healthy insect mass-production system, we need to have an extensive knowledge of the optimal diet conditions required for breeding and rearing the insects that can be used as feed and food (Jensen et al., 2017). We have a strong belief that NG can play an indispensable role in identifying species- and stage-specific nutritional requirements of these beneficial insects. Using this nutritional framework, we have recently launched a series of investigations on the nutritional requirements and outcomes of two most popular feed insects, the yellow mealworm (Tenebrio molitor; Rho and Lee, 2014;2015;2016;2017;2022) and the black soldier fly (Hermetia illucens; Cheon et al., 2022). Finally, we conclude that this powerful conceptual and experimental framework of NG will continue to help us in enhancing our fundamental understanding of insect nutrition and its application.