A Mechanism of Extreme Growth and Reliable Signaling in Sexually Selected Ornaments and Weapons

See allHide authors and affiliations

Science  17 Aug 2012:
Vol. 337, Issue 6096, pp. 860-864
DOI: 10.1126/science.1224286


Many male animals wield ornaments or weapons of exaggerated proportions. We propose that increased cellular sensitivity to signaling through the insulin/insulin-like growth factor (IGF) pathway may be responsible for the extreme growth of these structures. We document how rhinoceros beetle horns, a sexually selected weapon, are more sensitive to nutrition and more responsive to perturbation of the insulin/IGF pathway than other body structures. We then illustrate how enhanced sensitivity to insulin/IGF signaling in a growing ornament or weapon would cause heightened condition sensitivity and increased variability in expression among individuals—critical properties of reliable signals of male quality. The possibility that reliable signaling arises as a by-product of the growth mechanism may explain why trait exaggeration has evolved so many different times in the context of sexual selection.

The most elaborate male ornaments and weapons of sexual selection grow to exaggerated proportions (Fig. 1), especially in the largest and best-conditioned individuals. The size and conspicuousness of these traits make them likely candidates for intraspecific signals, used either by males to assess the size, condition, or status of rival males, or by females to assess the relative genetic quality of potential mates (1, 2). Not only are exaggerated traits easy to observe, they are unusually reliable signals of individual male quality (24), as their growth tends to be more sensitive to the nutritional histories and physiological conditions of individuals than is the growth of other traits (57). Exaggerated structures also tend to be more variable in their expression than other morphological structures (810). Hypervariability in trait size can amplify otherwise subtle differences in the body size or condition of males, further enhancing the utility of these traits as signals. Combined, these structural characteristics—extreme size, heightened condition sensitivity, and hypervariability among individuals—are the foundation for “handicap” and “good genes” models of sexual selection and a central tenet of modern theories of sexual selection and animal communication (24, 1115). We offer a developmental explanation for this phenomenon. We suggest that the evolution of trait exaggeration involves increased sensitivity to insulin/IGF signaling within a growing structure, and we show why such a change in mechanism should also confer both heightened condition sensitivity and hypervariablity to expression of the trait (Fig. 1B).

Fig. 1

(A) Exaggerated growth of weapons and ornaments of sexual selection. Clockwise from top left: Rhinoceros beetle horns (Trypoxylus dichotomus); long-tailed widowbird tail (Euplectes progne); elk antlers (Cervus elaphus); stag beetle mandibles (Lucanus cervus); fiddler crab chela (Uca tetragonon). (B) Proposed mechanism for the evolution of trait exaggeration through increased cellular sensitivity to insulin/IGF signaling (shown for the disc-like appendage primordia of insects). Individual nutritional state and physiological condition are reflected in circulating concentrations of insulin-like peptides and IGFs, which modulate the rate of growth of each of the trait primordia. Traits whose cells are sensitive (17) to these signals [e.g., wings (green)] exhibit greater nutrition-dependent phenotypic plasticity and among-individual variability than other traits whose cells are less sensitive to these signals [e.g., genitalia (red)]. An increase in the sensitivity of cells within a particular trait [e.g., horns (blue); see text] would lead to disproportionately rapid growth of that trait in the largest, best-condition individuals (i.e., exaggerated trait size) and smaller trait sizes in low-condition individuals.

Insulin and IGFs are essential regulators of tissue growth and body size (16). Circulating concentrations of insulin and IGFs are sensitive to nutrition, as well as stress and infection, and the insulin/IGF pathway has emerged as the central mechanism integrating physiological condition with growth in multicellular animal taxa. Insulin and IGF levels within a growing animal reflect the nutritional state and physiological condition of that individual, and circulating concentrations of these signals modulate tissue growth via the insulin receptor pathway in a graded, or dose-dependent, manner. Within an individual, growth will speed up or slow down in response to changes in nutritional or physiological state because of the action of this pathway. Across individuals, growth will differ between high-condition and low-condition individuals, resulting in population-level variation in body and trait sizes. Low-condition individuals have lower levels of these signals than higher-condition individuals, and as a result, they experience slower rates and lower overall amounts of tissue growth.

As long as the various organs and body parts (e.g., legs, eyes, wings) exhibit similar sensitivities to insulin/IGF signaling (17), their sizes will scale proportionally from individual to individual (1821). But some traits deviate in their responsiveness to these signals, profoundly affecting the amount and nature of their growth. Genitalia are insensitive to circulating insulin/IGF signals in Drosophila (20, 21). As a result, their growth is unresponsive to environmental conditions, such as nutrition, and genitalia size is largely invariant among individuals. In contrast, wings exhibit sensitivity to insulin/IGF signaling typical of the rest of the body; wing growth is sensitive to larval nutrition, and wing sizes scale isometrically with among-individual variation in body size (21).

On Science VideoLab

Embedded Image

Related video in the Science Video Portal

We predicted that increased sensitivity to the insulin/IGF pathway might be a mechanism leading to the evolution of extreme growth in showy ornaments and weapons of sexual selection. In our model, individual males differ in their physiological state as a result of differences in their status, nutritional state, competitive ability, and/or health (parasite or pathogen loads), which translate into among-individual variation in circulating concentrations of insulin/IGF signals (Fig. 1B). During their respective periods of growth, the adult structures in these animals would be exposed to insulin/IGF signals, and the sensitivity of cells within each growing structure to these signals would determine both how and by how much each trait grew. Just as wings are more sensitive to insulin/IGF signaling than genitalia in Drosophila (20, 21), so we predicted that exaggerated ornaments or weapons of sexual selection would be even more sensitive to insulin/IGF signaling than wings or other non–sexually selected body parts (Fig. 1B).

Male rhinoceros beetles (Trypoxylus dichotomus) wield a forked horn on their heads. During growth, horns in this species are more sensitive to larval nutrition than other body parts (wings, genitalia), and among adult males, horn size is hypervariable, ranging from tiny bumps to exaggerated structures two-thirds the length of a male’s body (22). We tested whether growing rhinoceros beetle horns were more sensitive to insulin/IGF signaling than wings or genitalia using RNA interference to perturb transcription of the insulin receptor (InR) gene. Developing larvae were injected with a 398–base pair fragment of double-stranded RNA (dsRNA) of T. dichotomus InR as they commenced their transition from larval feeding to gut purge (the onset of the prepupal period and the beginning of metamorphosis). At this time, all growth in overall body size had ceased, but adult structures (including genitalia, wings, and horns) were still growing. Thus, any effects of manipulation of insulin/IGF signaling would be visible as reductions to genitalia, wing, or horn size relative to overall body size. If the evolution of exaggerated horn size resulted in part from an increase in cellular sensitivity to insulin/IGF signaling, then horns should be more sensitive than wings to perturbation of the activity of this pathway. We also predicted that genitalia would be relatively insensitive to pathway perturbation [sensu (20, 21)].

Injections significantly reduced InR transcript abundances for 48 hours near the end of the period of trait growth (i.e., before InR transcript abundance normally drops in these tissues; Fig. 2, A to C). After metamorphosis was completed, we compared morphologies of treated and control animals. Genitalia did not respond to experimental perturbation of InR pathway activity (Wald statistic = 0.1245, df = 1, P = 0.724; Fig. 2D). Wings, which exhibit nutrition-sensitive growth patterns typical of most metric traits (e.g., eyes, legs, elytra), showed a significant reduction in size of ~ 2% (Wald statistic = 8.976, df = 1, P = 0.003; Fig. 2E). In contrast, male horns, the structures most sensitive to nutrition, were reduced by ~16% relative to controls (Wald statistic = 68.37, 1 df, P < 0.0001; Fig. 2, F and G). Using response to InR knockdown as a metric, we found that male horns were eight times as sensitive to insulin/IGF signaling as wings, consistent with our model for the evolution of disproportionate or exaggerated weapon size from enhanced tissue-specific sensitivity to the insulin/IGF pathway.

Fig. 2

Effect of insulin receptor (InR) knockdown on growth of adult structures in rhinoceros beetles. (A to C) Relative transcript abundances for the insulin receptor (InR) gene in genitalia (A), wings (B), and horns (C), measured 24, 48, 72, and 96 hours after the onset of the prepupal period in control (open bars) and dsInR-injected (solid bars) animals. Injection with dsRNA significantly reduced transcript abundances for 48 hours after injection in all three tissues. (D to F) Effects of dsInR knockdown on trait growth. Genitalia were insensitive (D); wings responded significantly but moderately to interrupted insulin/IGF signaling (E) (average reduction in wing length = 2%); and horns responded markedly (F), with an average reduction in horn length of 16%. (G) Head and thorax shown in two orientations (top and bottom) for same-sized control (left) and dsInR-injected (right) males.

A growing body of research now implicates insulin/IGF signaling in the development of extreme animal structures (23). Insulin/IGF signaling is an ancient and conserved physiological pathway that has coupled rates of cell proliferation with available nutrients for at least 500 million years, and we suggest that this pathway has been co-opted repeatedly in lineages experiencing strong sexual selection to yield disproportionate growth in signaling structures. The insulin/IGF pathway would likely have controlled the rate of growth of these structures already; increased cellular sensitivity to these signals would therefore be an easy route to the evolution of accelerated growth if the structure came under directional sexual selection for increased size.

But such a route to exaggeration would only generate exaggerated trait sizes in high-condition individuals because low-condition individuals would have low circulating concentrations of insulin/IGF signals and attenuated rates of tissue and body growth. The same mechanism stimulating increased trait growth in high-quality individuals would also repress trait growth in low-quality individuals (Fig. 1B). This means that whenever exaggerated ornament or weapon size arises due to an increase in trait-specific sensitivity to insulin/IGF signaling, then the exaggerated trait should also show enhanced (or “heightened”) condition-sensitive expression and higher relative variability in trait size between low- and high-condition individuals (as compared to other, nonexaggerated, traits). Signal reliability would be an intrinsic property of these structures because of the developmental mechanism regulating their growth.

Theoretical considerations of sexual selection and animal signaling argue that escalated evolution of signals is most likely when signals are reliable, and it is difficult or impossible for low-quality males to “cheat” by producing full-sized structures (Fig. 3). Signal reliability can be evolutionarily stable under two sets of conditions: Either the signal is sufficiently costly to produce or wield that it is not cost-effective for low-quality individuals to cheat (“handicap” signals), or the signal is intrinsically unfakable (“index” signals, “good genes” signals) (24, 1113, 2433). The largest ornaments and weapons are generally assumed to be handicap signals of male quality, such that the cost of these structures enforces signal reliability (24, 2433). However, for even the largest structures, the process of escalation must have started when these structures were small, and at that early stage, these costs would likely have been minimal. Moreover, several recent studies of exaggerated male ornaments and weapons have failed to find appreciable costs (34, 35), forcing a reconsideration of the question, why don’t low-quality males cheat?

Fig. 3

Sexual selection models whose relevance is affected by the proximate mechanism responsible for trait exaggeration.

We suggest that exaggerated animal structures may be unfakable signals of quality because of the developmental mechanism responsible for their accelerated growth. If true, then our hypothesis of “intrinsic reliability” could help explain why so many different signal traits embark on an evolutionary trajectory of bigger and bigger size. We suggest that whenever receivers responded to variation in insulin/IGF-sensitive structures, they fared relatively well due to the intrinsic reliability of these traits as signals of underlying male quality. As these traits became larger under selection, their utility as signals would have increased, enhancing the benefits to receivers and accelerating the rate of signal evolution still further. Once these structures become large enough to be costly, they may also act as handicap signals, and costs could contribute to signal reliability (Fig. 3). However, as long as the traits exhibit heightened sensitivity to insulin/IGF signals, costs may not be necessary for signal reliability (36). This means that subsequent evolution of compensatory structures alleviating costs to the signaling males (37) need not undermine the reliability of these traits as signals and could explain why some exaggerated sexually selected structures function as reliable signals even when no discernable costs are apparent (34, 35).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S3

Tables S1 and S2

References (3890)

References and Notes

  1. A more complete description of this pathway and references are provided in the supplementary materials.
  2. For this study, we define tissue sensitivity as the extent to which variations in the level of hormone signal influence the rate of cell proliferation via activity of the insulin/IGF pathway. Insensitive tissues grow to roughly the same final size regardless of circulating insulin/IGF concentrations, whereas the amounts of growth of sensitive tissues are strongly regulated by signal levels. Tissue sensitivity is often equated with receptor density. However, in this case, altered expression of any number of downstream genes in the pathway could change the responsiveness of a tissue to insulin/IGF signals. Indeed, in the best-studied example to date, reduced insulin sensitivity in a specific tissue (genitalia) in Drosophila, resulted from lowered levels of expression of a “downstream” element of the insulin-signaling pathway, FOXO, and not from tissue differences in expression of the insulin receptor (21).
  3. Results, as well as all methods for this paper, are in the supplementary materials.
  4. A description of these studies is in the supplementary materials.
  5. In principle, selection on poor-quality males to cheat could lead to evolutionary modifications to the underlying developmental mechanism that buffered expression of the exaggerated trait from the influence of male condition (i.e., that decreased sensitivity to insulin/IGF signals). In this event, the condition sensitivity of trait expression and among-male variability in trait size would decrease (as in male genitalia of these beetles), reducing the reliability of the size of the trait as a signal of male quality. We are aware of no instances in which exaggerated sexually selected signal traits presently display condition insensitivity and/or reduced among-individual variation. This could be because once the traits become exaggerated, their costs reinforce signal honesty and select against cheating males. Or it could reflect the fact that once subsequent insensitivity to insulin/IGF evolves in an exaggerated trait, its reliability as a signal diminishes, favoring receivers who ignore the trait and focus instead on other signals.
  6. Acknowledgments: We thank C. E. Allen, C. Breuner, K. L. Bright, S. T. Emlen, E. McCullough, A. Perkins, A. Shingleton, and three anonymous reviewers for helpful comments on the manuscript; Y. Hongo, H. Gotoh, and N. Kubota for help purchasing beetles; E. Paul (Echo Medical Media) for artwork on Figs. 1B, 2, and 3; and the National Science Foundation (IOS-0919781, IOS-0919730, and IOS-0920142) for funding. Images in Fig. 1A used under license from (rhinoceros beetle, NH; widowbird, Simon_g; elk, W. Aston; stag beetle, H. Larsson; fiddler crab, Manamana). Sequences are deposited in GenBank (accession nos. JX141307 to JX141311).
View Abstract

Navigate This Article