Introduction

Sexual dimorphism, a phenomenon characterized by physiological, behavioral, morphological, and ontogenetic differences between males and females, is a widespread and evolutionarily significant trait observed across both vertebrates and invertebrates1,2,3,4. In insects, including moths and butterflies, sexual dimorphism manifests in traits such as size, shape, and color patterns, which play critical roles in survival and reproduction5,6. Understanding the evolutionary drivers and ecological consequences of these dimorphisms can provide insights into the adaptive strategies employed by species in response to environmental pressures.

Hawkmoths (Lepidoptera: Sphingidae) provide an exemplary system for studying sexual dimorphism due to their diverse morphological adaptations7. This family, encompassing approximately 1500 species across 206 genera and three subfamilies8,9, exhibits significant sexual dimorphism in various anatomical features. These include wing morphology, antennal structure, and epiphyses, which are critical for mating and sensory functions10,11,12,13,14. The epiphysis is a synapomorphy of lepidopterans and, when present15,16, is located on the inner surface of the foretibia17,18,19,20. This specialized spur is a movable structure15,19,21 and it is covered by cuticular protuberances called acanthae19,20,21,22. Given the presence of acanthae, the role of epiphysis as an antenna cleaning structure is indisputable19,21,22. In addition, this structure varies in shape, size and position, such variation is widespread in several families, genera, species and sex17,23.

Sphingidae have prominent and well-developed epiphyses15,17,24 that may vary in length, position17,23,25, and they also may vary between sexes23,26. For example, males of Ambulyx pryeri Distant, 1888, have been reported to possess longer and wider epiphyses than females23. These differences may reflect sex-specific ecological roles, with males requiring increased sensory capabilities for mate detection, while females optimize their flight morphology for oviposition site selection27,28.

Our study addresses a critical gap in the understanding of sexual dimorphism in Sphingidae by examining the allometric variation in epiphyses, particularly within the tribes Dilophonotini and Philampelini. We hypothesized that males would exhibit a larger and wider overall epiphysis than females, driven by their need for more robust grooming structures to maintain longer and broader antennae10,11,29. By empirically testing this hypothesis, we aim to elucidate the evolutionary pressures shaping sexual dimorphism in this family, contributing to broader discussions on the adaptive significance of morphological differences between sexes.

Materials and methods

Specimens

A total of 564 specimens from two tribes of Macroglossinae (Lepidoptera: Sphingidae), Dilophonotini and Philampelini, were analyzed. The dataset included 51 species of Dilophonotini (444 specimens) and eight species of Philampelini (120 specimens), comprising 305 males and 259 females. The specimens were sourced from the Lepidoptera Collection at the Museu de Zoologia da Universidade de São Paulo (MZUSP), São Paulo, Brazil.

Epiphysis preparation and preservation protocols were conducted following the methods outlined by Ancajima & Duarte23. Observations were performed using a Zeiss Discovery V20 stereomicroscope, and images were captured across multiple focal planes using a Zeiss AxioCam MRc5 camera. Image stacking was completed using CombineZP software30 to obtain high-resolution composite photographs.

Measurements

Only specimens with well-preserved structures were included in this study. The number of individuals used for each measurement varied from two to 15, depending on specimen availability, with an average of 10 ± 6.01 (standard deviation). Composite images of the epiphyses and foretibiae were analyzed using ImageJ v1.54 g software31. Antenna measurements were obtained directly using a ZEISS SteREO Discovery V8 stereo microscope, equipped with a ruler attached to the eyepiece. The detailed procedures for measuring the epiphyses, foretibiae, and antennae are illustrated in Fig. 1.

Fig. 1
figure 1

Measurements of the epiphyses, foretibia and antennae of Erinnyis ello (Linnaeus, 1758). The abbreviations used are: (el) epiphysis length, (ew) epiphysis width, (fl) foretibia length, (po) position on the foretibia, (al) antennae length. All the measurements were included as variables in the statistical analyses.

Full size image

Statistical analyses

All statistical analyses were conducted using the R programming language32, with a significance level of 0.05. Generalized linear mixed models (GLMMs) were employed to investigate allometric patterns between variables and to explore sex differences in these relationships. Since classical allometric regression can be performed through linearization, the models were constructed using a log10–log10 linear regression and included sex as a covariate, initially as an interaction term. The GLMMs were constructed and refined using the buildmer package33, with model selection based on the Akaike Information Criterion. Only the final model structures selected through this process were considered, and these models were fitted using the glmmTMB package34. To account for taxonomic interdependence among species, genera were included as random intercepts in the models, following the approach used in similar studies35,36. This approach captures shared variance among species within the same genus37, offering a pragmatic alternative to phylogenetic methods such as PGLS and MCMcglmm38. Conditional and marginal Nakagawa R2 values were calculated using the performance package39, and model performance was assessed through deviance analyses with the Wald chi-square test.

Additionally, ancestral states for the length of epiphyses, treated as continuous characters, were reconstructed and compared between males and females. This reconstruction was performed using maximum likelihood estimation with the fastAnc function and visualized using contour maps generated by the contMap function, both of which are available in the phytools package40. We conducted 100 simulations to estimate ancestral states and determine the 95% confidence intervals for all potential reconstructions. Continuous characters were mapped using modified phylogenetic trees, as proposed by Kawahara et al.8 and Kawahara & Barber41. All graphics generated in R were subsequently edited using Adobe Illustrator 2022.

Results

Among the species of Dilophonotini and Philampelini examined, 93.2% (n = 55) exhibited longer epiphyses in males than in females, with only four species deviating from this pattern. Specifically, Cephonodes hylas (Linnaeus, 1771), Isognathus menechus (Boisduval, 1875), and Pachylia darceta Druce, 1881, presented longer epiphyses in females, whereas Madoryx oiclus (Cramer, 1779) presented no significant sex variation in epiphysis length. Notably, the shortest epiphysis was recorded in the female of Hemaris croatica (1.37 mm), and the longest was recorded in Eumorpha anchemolus (Cramer, 1779), with a mean length of 3.53 ± 0.16 mm (Supplementary Table S1). In contrast, the shortest epiphysis among males was found in Aellopos tantalus (Linnaeus, 1758) at 1.52 mm, whereas the longest, also in Eumorpha anchemolus, measured 3.87 ± 0.12 mm (Supplementary Table S1). A complete list of measurements for all 59 species is available online in Supplementary Table S1 online.

Static allometry in epiphyses

Epiphysis length (el) × position (po)

A statistically significant influence of foretibia position on epiphysis length was observed (χ2 = 300.66; P < 0.001), alongside a notable sexual dimorphism in the allometric relationship (χ2 = 159.70; P < 0.001), with males exhibiting slightly greater values (Fig. 2A). The model accounted for a substantial proportion of the observed variation, as indicated by a Conditional R2 = 0.813, compared with the variation explained solely by the fixed terms (position and sex) (Marginal R2 = 0.389).

Fig. 2
figure 2

Allometric relationships between normalized measurements (using log10–log10 linear regression from 564 specimens of the tribes Dilophonotini and Philampelini). Panel a: Linear allometric relationship between the position of the epiphysis on the fore tibia and its length. Panel b: Linear allometric relationship between epiphysis length and antenna length. Panel c: Linear allometric relationship between the position of the epiphysis on the fore tibia and its width. Panel d: Linear allometric relationship between epiphysis width and antenna length.

Full size image

Epiphysis length (el) × antennae length (al)

A significant allometric relationship was identified between antenna length and epiphysis length (χ2 = 782.357; P < 0.001), with a pronounced sex difference in this relationship (χ2 = 61.861; P < 0.001), where males presented slightly larger values (Fig. 2B). The model explained a significant portion of the total variation, with a Conditional R2 = 0.888, whereas the fixed factors (antenna length and sex) accounted for most of the explained variation (Marginal R2 = 0.799).

Epiphysis width (ew) × position (po)

A significant allometric relationship between epiphysis width and its position on the foretibia was found (χ2 = 106.57; P < 0.001), with clear sexual differentiation in this relationship (χ2 = 23.54; P < 0.001), where males presented slightly greater values (Fig. 2C). The model accounted for a Conditional R2 = 0.627, whereas the variation explained by the fixed terms (position and sex) was lower (Marginal R2 = 0.255).

Epiphysis width (ew) × antennae length (al)

A significant allometric relationship was also observed between epiphysis width and antenna length (χ2 = 201.52; P < 0.001). While the effect of sex alone was not significant (χ2 = 1.78; P = 0.183), the interaction effect between antenna length and sex was significant (χ2 = 5.17; P = 0.023), suggesting that the relationship between these traits may differ slightly between sexes (Fig. 2D). The model accounted for a substantial portion of the variation, with a Conditional R2 = 0.698, and a Marginal R2 = 0.493.

Evolutionary allometry in epiphyses

The character state reconstructions revealed that the common ancestor of the clade encompassing Dilophonotini and Philampelini likely possessed epiphyses of medium size, with average lengths of 2.46 ± 0.587 mm in males and 2.42 ± 0.583 mm in females. This ancestral state suggests that both male and female epiphyses were relatively similar in size at the early stages of the clade’s evolutionary history. However, significant deviations from this ancestral condition were detected across various lineages within the clade.

In the subtribe Hemarina (Fig. 3, node 2), a marked reduction in epiphysis length was observed in Cephonodes and Hemaris. The average size of epiphyses in this group decreased to 1.62 ± 0.118 mm, indicating a significant evolutionary shift toward smaller epiphyses compared with the ancestral state. Conversely, the subtribe Dilophonotina exhibited considerable variability in epiphysis size. Some species within this group retained relatively short epiphyses, as indicated by measurements at specific nodes (Fig. 3, node 7). However, other species, such as Pachylia ficus (Linnaeus, 1758), Oryba kadeni (Schaufuss, 1870), and Pseudosphinx tetrio (Linnaeus, 1771), presented significantly longer epiphyses. These findings highlight the diverse evolutionary pathways within Dilophonotina, where both elongation and reduction of epiphyses have occurred in different lineages.

Fig. 3
figure 3

Ancestral state reconstruction of relative epiphysis length on the basis of phylogeny by Kawahara & Barber (2015)41. The topology is cropped to 25 species represented in our data. The warmer colors indicate shorter epiphyses, whereas the colder colors indicate longer epiphyses. The numbers on the cladogram represent the nodes. The scales represent the trait values.

Full size image

In the tribe Philampelini, notable differences in epiphysis length were observed among phylogenetically closely related species. For example, Eumorpha obliquus (Rothschild & Jordan, 1903) was found within a clade (Fig. 3, node 13) that also includes Enyo lugubris (Linnaeus, 1758), Enyo ocypete (Linnaeus, 1758), Aleuron chloroptera (Perty, 1833), and Unzela japix (Cramer, 1776). Despite their close phylogenetic relationships, these species presented substantial differences in epiphysis length, suggesting that epiphysis size is subject to varying selective pressures even among closely related species.

Additionally, clades characterized by similar epiphysis lengths were identified, this pattern was particularly evident in nodes 2 and 7 (Fig. 3), where species within these clades maintained a continuous character state.

Discussion

The term “static allometry” is commonly used to describe the relationship between body size and a specific organ42,43,44,45. However, it can also refer to the relationship between two organs in individuals at the same developmental stage43,44,46, which is the interpretation adopted in this study. Specifically, we investigated the presence of sexual dimorphism in the allometric variation of the epiphyses in Neotropical hawkmoths from to the tribes Dilophonotini and Philampelini (Macroglossinae). A comprehensive morphometric analysis of 564 museum specimens confirmed significant sexual dimorphism in the epiphyses of these groups.

Sexual dimorphism in grooming structures has been documented in various insect orders, including Hemiptera, Diptera, Hymenoptera, and Lepidoptera23,47,48,49,50. For example, some species of Hemiptera (Pentatomidae) exhibit sexual dimorphism in the number of setae in the foretibial apparatus47, while, Culicoides melleus (Coquillett, 1901) displays sexual variation in the number of spines on the comb50. Similarly, in Hymenoptera, males of Apis mellifera Linnaeus, 1758 and Xylocopa violacea (Linnaeus, 1758)48,49 possess a strigilis with a larger spur and notch compared to females48,49. This pattern aligns with findings in Lepidoptera, where sexual dimorphism has been observed in both morphology (e.g., acanthae shape, comb structure) and linear morphometry (e.g., length). Males typically exhibit longer epiphyses, more robust combs covered by larger more saw-like acanthae compared to females19,23. As noted, sexually dimorphic patterns in other grooming structures across insects may be functionally linked to sexual dimorphism in antennal morphology23,51.

Numerous studies have demonstrated sexual dimorphism in the antennae of Lepidoptera, including differences in type52, size19, the number, the types and distribution of sensilla11,29,53,54, the antennal flagellum11, and OR-encoding genes27. The morphological characteristics of the antennae and grooming structures (epiphyses) are likely correlated, as suggested by Ancajima & Duarte23. Similarly, the sizes of both structures may also be related, a hypothesis supported by our findings.

Species with more pronounced sexual dimorphism in terms of epiphysis size may also exhibit complex mating behaviors19,52. Several studies highlight the importance of grooming the antennae during sexual activity, as observed in species such as Dioryctria abietella (Denis & Schiffermüller, 1775) (Lepidoptera: Pyralidae)55 and Blattella germanica (Linnaeus, 1767) (Blattodea: Blattellidae)56, as well as during oviposition, as in Plodia interpunctella (Hübner, [1813]) (Lepidoptera: Pyralidae)28. These behaviors may also occur in hawkmoths, where sexual dimorphism in their antenna-cleaning structures (epiphyses) is observed.

Furthermore, the evolutionary significance of sexual dimorphism in epiphyses can be examined within the context of Sphingidae mating systems. In hawkmoths, females typically initiate mating through pheromone emission57,58, exposing males to higher concentrations of chemical signals. This suggests that males, with their larger and more developed antennae equipped with more olfactory receptors, may require more frequent grooming than females, who clean their thinner antennae, equipped with fewer olfactory receptors, primarily for oviposition11,27,28. Thus, our findings suggest that the considerable size variation in epiphyses may be linked to sexual dimorphism and could play a role in sexual selection within Sphingidae.

In addition, our study revealed a positive allometric relation between the position of the epiphyses on the foretibiae and the length and width of the epiphysis. Meaning that as epiphyses become longer and wider, they are positioned more distally from the body, a pattern observed in males. In contrast, females tend to have epiphyses located closer to the body. This correlation suggests that mechanical constraints, such as the need to maintain balance and maneuverability, may influence both the size and positioning of these grooming structures14. These findings are consistent with previous observations by Ancajima & Duarte23 in another group of Sphingidae (Smerinthinae: Ambulycini).

Regarding the evolutionary allometry of epiphyses in Dilophonotini and Philampelini hawkmoths, it reflects complex interactions driven by both ecological and sexual selective pressures59. These structures are notably larger than those in other lepidopterans. For example, the smallest epiphyses reported in our study, such as those in the females of Hemaris croatica (Esper, 1800) and the males of Aellopos tantalus (Linnaeus, 1758), exceed the dimensions observed in Helicoverpa spp. (Noctuidae), which range from 0.67 ± 0.04 mm to 0.95 ± 0.11 mm60, and in females of Lymantria dispar (Linnaeus, 1758) (Erebidae: Lymantriinae)19. Moreover, the lengths documented here surpass those recorded in other hawkmoths, such as the females of Akbesia davidi (Oberthür, 1884) (Ambulycini)23. To the best of our knowledge, the epiphysis lengths in our study are the largest reported for any Lepidoptera species19,23,60.

Conversely, the significant reduction in epiphyses observed in the subtribe Hemarina likely represents an adaptation to specific ecological niches or life history traits that diminish reliance on these structures. This reduction may be associated with a shift toward alternative sensory modalities or a reduced need for extensive grooming due to the presence of less elaborate antennae. Such patterns are often seen in species with simpler mating systems or those inhabiting less cluttered environments, where the likelihood of antennae fouling is lower61,62. A similar phenomenon is observed in the reduction of eyespot size in butterflies of the species Bicyclus anynana (Butler, 1879) (Lepidoptera: Nymphalidae), where environmental factors influence the expression of this sexually dimorphic trait, potentially reducing reliance on visual signals in certain ecological contexts63. In contrast, the subtribes Dilophonotina and Philampelina exhibit a wide range of variation in epiphysis sizes among genera and species, reflecting divergent selective pressures acting within these groups. These findings suggest that even among closely related genera and species, diverse ecological and behavioral factors can influence the evolution of these structures. For instance, species such as Eumorpha obliquus (Rothschild & Jordan, 1903) and Pseudosphinx tetrio (Linnaeus, 1771) exhibitlong epiphyses, which may confer advantages in their specific ecological interactions or mating systems. This interspecific variation underscores the role of local environmental conditions and species-specific behaviors in shaping the evolutionary trajectory of morphological traits64. In other words, environmental heterogeneity can impose varying selective pressures across populations, driving local adaptations that result in morphological divergence even among closely related species65. For example, Batocnema coquerelii (Boisduval, 1875), a species endemic to Africa, exhibits a unique inward-bent epiphysis shape, distinct from the forms observed in species from the Neotropical region23.

Patterns of evolutionary allometry in epiphyses are evident and observed across different clades. Micropterigidae, the sister group to all other Lepidoptera66,67, includes the species Sabatinca chrysargyra (Meyrick, 1885), which possesses a small epiphysis located near the middle of the foretibia, with its apex positioned less than a quarter of the way from the foretibial tip17. Despite its size, the shape of this structure is similar to those found in other lepidopterans families, such as Sphingidae and Papilionidae17,20,23. The clade Macroheterocera, which includes Noctuoidea60 and (Bombycoidea23 + Lasiocampoidea17)66,67, further highlights the significance of epiphysis size variation across taxa. Another clade demonstrating morphological variation in grooming structures, including closely related families, is the Papilionoidea clade. Butterflies (Papilionoidea) may or may not have epiphyses. For example, Pieridae lack an epiphysis, whereas Hesperiidae and Papilionidae have a well-developed structure17,20. These variations in epiphyses – ranging from presence or absence to differences in size and shape – are evident even within species of the same superfamily. This reinforces the hypothesis that epiphyses, likely an ancestral trait, have diverged under varying selective pressures, resulting in size and structural differences across clades.

Future research should integrate phylogenetic data with morphological analyses to better elucidate the adaptive significance of these variations and explore the evolutionary dynamics of sexually dimorphic traits.

Conclusions

Our study provides a detailed examination of the intricate relationships among allometric patterns, sexual dimorphism, and associated morphological traits in the epiphyses of the Dilophonotini and Philampelini hawkmoths. The confirmation of significant sexual dimorphism in epiphysis size, with males consistently exhibiting longer and wider epiphyses than females, highlights the potential role of sexual selection in driving these morphological differences.

The correlations we identified, particularly between foretibial length and epiphysis positioning, suggest that these morphological features are not only influenced by mechanical constraints but are also optimized for functional demands, such as grooming efficiency. This nuanced understanding of epiphysis morphology highlights the complex interplay between structural constraints and adaptive evolution, where both physical and ecological factors dictate the development and positioning of these features.

Moreover, our results provide new insights into the potential ecological and behavioral implications of epiphysis morphology. For example, variation in epiphysis size across different clades and species of hawkmoths may reflect adaptations to specific ecological niches or mating strategies. Although sexual dimorphism in epiphysis size is evident, we did not collect data on environmental variables, mating systems, or other selective pressures that would enable a conclusive link between epiphysis size variation sexual selection. Therefore, while these results suggest intriguing possibilities regarding the role of sexual selection, they do not provide direct evidence to support this hypothesis. Future studies should explore the ecological and behavioral correlates of epiphysis size, along with the genetic mechanisms underlying its variation, to clarify the role of these structures in sexual selection and species diversification within Lepidoptera.

Our study not only contributes to the current understanding of sexual dimorphism and allometric patterns in lepidopterans but also opens new avenues for research in evolutionary biology. As we move forward, incorporating a multidisciplinary approach that combines morphological, ecological, genetic, and behavioral perspectives is crucial. This will enable a more comprehensive understanding of the evolutionary dynamics shaping these fascinating traits and their significance in the context of species survival and adaptation in diverse environments.