Introduction

Identifying the evolutionary factors promoting diversification and co-existence of closely-related species in sympatry is a major challenge. Since the evolution of traits within species in sympatry can affect ecological interactions between individuals, it either fuels or impairs the speciation process. Closely-related species often share similar suites of traits because of their shared evolutionary history. Furthermore, in sympatry, habitat filtering excludes species with suites of traits mal-adapted to local conditions, therefore enhancing trait similarity (Chazot et al., 2014). The shared selective pressures encountered by individuals living in sympatry also often promote the evolution of locally adapted traits within species and even trait convergence among species (Muschick et al., 2012). That said, closely-related sympatric species often display divergent suites of traits because they are usually partitioned in divergent ecological niches (Berlocher & Feder, 2002). The co-existence in sympatry may also promote divergent evolution of traits in incipient species, because of competitive interactions between individuals or as a result of reinforcement (Butlin & Smadja, 2018). The evolution of traits in incipient species living in sympatry thus depends on the balance between convergence and divergence, resulting either in (i) evolutionary trade-offs of traits jointly influenced by antagonistic natural and sexual selective pressures (Ellers & Boggs, 2003; Maan & Seehausen, 2011) or (ii) convergence in some traits and divergence in others (Le Roy, Roux, et al., 2021).

Theoretical and empirical studies have shown that sexual selection may favor the evolution of preferences for locally-adapted traits within species (Servedio, 2004; Servedio & Boughman, 2017; van Doorn et al., 2009). For instance, the predator-deterrent coloration of poison frogs is also detected and used as mating cues by females (Reynolds & Fitzpatrick, 2007). Similarly, habitat- dependent coloration of sympatric cichlid fish is also used as a visual cue for mate recognition (Seehausen et al., 2008). Yet, sexual interactions are likely to occur between individuals from closely-related species when they live in sympatry, and similar preferences for adaptive traits may thus result in substantial reproductive interference (Gröning & Hochkirch, 2008). For example, because sympatric amphibians can sometimes share the same mating cues and preferences, they often engage in misdirected mating attempts towards heterospecifics, a potentially deleterious behavior (Soni et al., 2024). In addition, hybrids, when produced, can be unfit, thus favoring the evolution of sexual preferences for species-specific cues, rather than locally-adapted traits (Maisonneuve et al., 2024). Comparing variations in adaptive traits in sympatric vs. allopatric populations of recently-diverged species, and testing the sexual preference for those traits can shed light on the selective processes targeting traits modulating reproductive isolation and co-existence in sympatry.

In butterflies, the evolution of wing color patterns can be influenced by both natural and sexual selection. The visual discrimination of wing color pattern has been shown to enable intraspecific recognition during courtship in many species (Costanzo & Monteiro, 2007; Li et al., 2017). However, the evolution of wing color patterns is also strongly influenced by the visual effect perceived by predators (Finkbeiner et al., 2014; Oliver et al., 2009). Whether these selective pressures promote trait convergence or divergence in sympatric species might depend on their relatedness: for instance, a recent study in Papilionidae showed multiple color pattern convergences between distantly-related species living in sympatry, while divergent colorations are found in closely-related species (Puissant et al., 2023). Positive selection acting on divergent traits involved in species recognition could be increased because of higher reproductive interference in closely related species than in distantly related taxa (Pfennig & Pfennig, 2009). In sympatric species with chemical-defenses, such as Heliconinii butterflies, local predation pressures tend to promote the convergence of similar conspicuous warning wing patterns (i.e. Müllerian mimicry, (Joron et al., 1999; Merrill et al., 2014)). But the costs associated with hybrid production in turn favor the evolution of alternative divergent mating cues in mimetic butterflies (Estrada & Jiggins, 2008), and divergence in male pheromone bouquets and female attraction has been found among mimetic sister species (González-Rojas et al., 2020). Similarly, the evolution of specific visual mate recognition signals, limiting reproductive interference but indistinguishable by predators, can also be promoted on the wings of mimetic butterflies (Llaurens et al., 2014). Thus, ecological interactions among recently-diverged species living in sympatry can impact the evolution of traits involved in both local adaptation and mate recognition and may trigger the evolution of alternative traits. Investigating variations in traits used as mating cues in allopatric vs. sympatric populations of sister-species can help ascertain how trait evolution might favor speciation and co-existence in sympatry.

Here we focus on the evolution of mating cues in the neotropical butterfly genus Morpho, where multiple closely-related species co-exist in sympatry (Blandin & Purser, 2013). In some Morpho species, striking iridescent blue coloration is displayed on the dorsal side of the wings, due to nanostructures on wing scales (Siddique et al., 2013). Variations in the reflected wavelength depending on the angle of observation and/or illumination (i.e. iridescence) have been documented in Morpho species (Giraldo et al., 2016). The light signal reflected by iridescent surfaces can be very directional, as hue and brightness of iridescent objects can drastically change depending on the light environment or the observer’s position (Doucet & Meadows, 2009). Directional iridescent signals generated in animals could then enhance recognition by mates while remaining poorly detected by predators (Endler, 1992). For instance, in some hummingbird species, when males tilt their head in a specific direction, flashes of bright coloration can be observed in these otherwise darkly-colored birds (Simpson & McGraw, 2019). The specific directional signal produced by the iridescent feathers is a cue during courtship, suggesting that the antagonistic sexual and natural selective pressures finely shape the evolution of iridescence. Similarly, male Hypolymnas butterflies display abrupt shifts in UV brightness on their wings and their courtship behavior enhances the production of flashes perceived by females (White et al., 2015). How much sexual selection shapes the evolution of iridescence in sympatric Morpho species is currently unknown, but behavioral experiments carried out in the field in Amazonian Peru highlighted strong visually- based territorial interactions among males from sympatric species and limited species discrimination based on female coloration in males (Le Roy, Roux, et al., 2021).

This raises questions on the key visual cues involved in mate choice, given that the iridescent blue coloration shared by closely-related species encountered within the understory is likely selectively constrained by predators. The iridescent bright blue dorsal coloration of Morpho wings contrasts with the brown and matte ventral side, and generates a peculiar visual effect during flight. The combination of the blue flashes produced by the alternate exposure of the bright blue vs. brown sides of the wings during flapping flight, in addition to erratic flight trajectories makes these Morpho very difficult to follow by predators (Young, 1971), potentially enhancing their evasive capabilities (Murali & Kodandaramaiah, 2020; Vieira-Silva et al., 2024). Experimental trials with evasive prey have shown that predators learn to avoid prey they repeatedly fail to catch (Páez et al., 2021). The display of iridescent wings could thus be associated with a higher survival rate in nature because of both (i) direct effects, through successful escape of predator attacks, and (ii) indirect effects, by limiting predation attempts by birds recognizing the blue signals and refraining from attacking (Pinheiro et al., 2016). This indirect effect could promote the evolution of convergent blue patterns in sympatric species, similar to the mimicry observed in species with chemical-defenses. In line with this hypothesis, repeated local convergence in the proportion of iridescent blue vs. black areas on the dorsal side of the wings have been documented in the sister- species Morpho helenor and Morpho achilles living in sympatry throughout the Amazonian basin: indeed, parallel geographic variations in the width of the dorsal blue band was detected in these sympatric species (Llaurens et al., 2021). Precise quantification of variations in iridescence is now needed to assess the respective effects of selection by predators and mates, that may drive convergent vs. divergent evolution of iridescence in sympatric species sharing the same micro- habitats. Investigating this convergent vs. divergent evolution of visual cues that may act as pre- zygotic barriers to gene flow between these two species will shed light on the mechanisms allowing their genetic divergence and coexistence in sympatry.

To do so, we quantified variations of iridescence in the two sister species M. helenor and M. achilles in different geographic localities to investigate the antagonistic effects of selective pressures generated by mate recognition and shared predation. We used behavioral experiments to test for the effect of this color variation in visual signals in mate preferences. Finally, we studied variations in the volatile compounds produced by wild males and females from both species to explore the evolution of potentially alternative traits, such as chemical cues, possibly acting as a reproductive barrier and favoring the coexistence in sympatry of closely related Morpho species.

Results

We investigated differences in wing iridescence, color perception and color pattern-based mate choice of two sister species of Morpho, M. helenor and M. achilles, which diverged ∼ 3.6 Mya (Chazot et al., 2021). We relied on geographic variations of wing color patterns within M. helenor (Fig 1) to compare the iridescence of three different populations of M. helenor (subspecies M. helenor bristowi, M. helenor theodorus and M. helenor helenor) and test the associated visual effect using visual models.

Fig 1:

The convergent iridescent signals of sympatric Morpho species are not differentiable

We measured the reflectance of Morpho wings at different angles of observation and illumination, in two different planes (referred to as proximo-distal and antero-posterior planes) following two complementary protocols (referred to as “specular” and “tilt”; see Fig 2.A and Supporting Information Text 1 for extended methods). We then extracted for each spectrum three commonly- used colorimetric variables: Hue, Brightness and Chroma (analyzed in S4 Fig).

Fig 2:

We found that hue and brightness always vary depending on the angle of incidence of the light, as confirmed by PERMANOVAs detecting a significant effect of the “Angle” for all methods of measurements (Fig 2 and statistical analyzes in S1 Table and S5 Fig) confirming that in both species, both sexes are iridescent. Second, the brightness of Morpho wings is highly sexually dimorphic, as the effect of sex is always significant. The brightness measured with the “tilt” protocol (Fig 2.F and 2.G) shows that males tend to be brighter than females in both Ecuador and French Guiana.

To finely appraise the complete spectrum of variation in iridescence within and among Morpho species, we then used a multivariate approach allowing to analyze the full reflectance spectra obtained from both proximo-distal and anteroposterior measurements. At the intraspecific level, significant divergence of iridescence between populations was detected (Fig 3.B). In contrast, iridescence did not differ between species living in sympatry (PERMANOVA, p-value = 0.084), suggesting that iridescence converges in sympatry (Fig 3.A). In both cases, the interaction term taxon*sex was non-significant, suggesting that the sexual dimorphism of iridescence was similar for all taxa.

Fig 3:

We then modeled the perception of color contrast in the visual systems of M. helenor and UV- sensitive birds respectively, and tested whether the differences of reflectance observed between sexes and species at different angles of illumination could potentially be perceived by a Morpho observer or a putative predator. As sexual dimorphism on wing reflectance was detected, we measured the chromatic distances for males and females separately (Fig 3.C). As the results were similar along the two directions of measurements (antero-posterior and proximo-distal), only proximo-distal results are shown below (antero-posterior results in S6 Fig). Morpho butterflies and predators can visually perceive the difference in the blue coloration between the allopatric populations of M. helenor: for both Morpho and bird visual systems the mean chromatic contrasts are always higher than 1 JND, for both sexes and for all angles. In contrast, the mean chromatic contrasts measured between sympatric species were below 1 JND: the minute differences observed between sympatric species are thus unlikely to be discriminated either by Morpho or bird predators.

Evidence of mate preferences based on visual cues within species

We then performed a series of mate choice experiments to identify the visual cues used during mate choice within M. helenor. We specifically tested for preferences based on the differences in color measured between the two previously studied populations of M. helenor (M. h. bristowi and M. h. theodorus), which significantly differs in both wing pattern and iridescence.

We first conducted a tetrad experiment using the allopatric subspecies M. h. bristowi and M. h. theodorus. The highest number of mating events occurred between the two assortative pairs (see Table 1). More than half of the matings occurred between a female and a male of the M. h. bristowi population. Indeed, a Chi-square test performed on the contingency table showed a significant departure from random mating (X-squared = 4.98, p-value = 0.04), confirmed by a Fisher exact test (p-value = 0.04). The time until an assortative mating (mean = 48 min, sd = 38 min) or until a disassortative mating (mean = 48 min, sd = 38 min) however did not significantly differ (T-test; stat= -0.27, p-value = 0.79).

Table 1:

Different pairs of female dummies were presented to the M. helenor males to investigate assortative preference between (sub)species (experiment 1), discrimination of the wing pattern (experiment 2) and discrimination of the iridescent coloration (experiment 3) (see the experimental details in the M&M section).

Experiment 1 tested the male assortative visual preference using two dummies each harboring the wings of a female of a different M. helenor subspecies. M. h. theodorus males did not show any preference for any dummy female, as shown by the probability of approaches and touches that were not different from the random 0.5 expectation (approaches: Wilcoxon test, p-value=0.764, Fig 4.A; touches: Wilcoxon test, p-value=0.287, Fig 4.B). Morpho helenor bristowi males, in contrast, approached and touched a con-subspecific M. h. bristowi dummy female significantly more often than expected by chance alone (approach: Wilcoxon test, p-value = 0.0448, Fig 4.A; touch: Wilcoxon test, p-value < 0.001, Fig 4.B), suggesting a preference toward the M. h. bristowi females based on the recognition of visual cues alone.

Fig 4:

Experiment 2 tested male pattern-based discrimination, using two dummies harboring the wings of M. h. bristowi females (i.e. same iridescence), but one of which was modified so as to present the thin blue bands found in M. h. theodorus (i.e. a different pattern). M. h. bristowi males tended to approach and touch significantly more often the wild-type M. h. bristowi female model as compared to the modified M. h. bristowi female model exhibiting a theodorus pattern (Wilcoxon tests on approaches; p-value = 0.0188, on touches; p-value < 0.001, see S7 Fig). As both female dummies shared the same blue iridescence, the choice of M. h. bristowi males was likely influenced by the difference in the pattern displayed by the dummy females, suggesting that the color pattern is used by males during mate choice.

Experiment 3 tested male color-based mate discrimination using two dummies harboring the wings of females from the two subspecies (i.e. different iridescence), but those of M. h. bristowi were modified so as to resemble M. h. theodorus (i.e. same pattern). Morpho helenor theodorus males tended to approach and touch significantly more often the M. h. theodorus female model compared to the artificially modified, theodorus-like M. h. bristowi female dummy (Wilcoxon test on approaches; p-value = 0.0201, on touches; p-value = 0.0343, see S8 Fig). As both dummy females shared the same color pattern, the preference expressed by M. h. theodorus males likely stems from the differences in the iridescent blue color suggesting color-based discrimination in males. Nevertheless, in experiment 2 and 3, modified dummies were strongly rejected by males. We cannot entirely rule out that the artificial modifications performed on our dummies could generate a specific visual rejection by the males.

Visual cues are not pre-zygotic reproductive barriers between the mimetic species M. h. helenor and M. a. achilles In a final experiment (experiment 4), we tested whether males from the two sympatric species from French Guiana (M. achilles and M. helenor) have a visual preference toward their conspecific females: two female dummies harboring the wings of females from the two species were presented to each male. While none of the males (from either M. helenor or M. achilles species) touched the female models, they did approach them in the cages. The approach rate and the number of approaches did not significantly differ between M. helenor males (rate = 61%; mean = 4.21, sd = 5.34) and M. achilles males (rate = 66%; mean = 7.18, sd = 8.29; binomial GLM: p-value = 0.626), revealing similar levels of male courtship motivation. The independent analysis of the mean preference per species shows that neither M. helenor (Wilcoxon test, p-value = 0.503) nor M. achilles (Wilcoxon test, p-value = 0.406) males approached significantly more con-specific females as compared to the hetero-specific females in the cage (Fig 4.C), suggesting that the difference among the convergent color patterns and blue iridescence of female M. helenor and M. achilles is not perceived by males.

Divergent chemical profiles observed in males between the mimetic species M. helenor and M. achilles

We finally investigated the volatile chemical compounds extracted from the genitalia of males and females from the mimetic sympatric species M. helenor and M. achilles. The nMDS analyses performed (Fig 5) suggest a strong divergence in the chemical profiles between males of the two species, especially for C16 to C30 compounds. This result contrasts with the strong similarity of chemical profiles found between the females of these two species in these longer-chain compounds (S9 Fig).

Fig 5:

The PERMANOVA of the chemical spectra of M. h. helenor and M. a. achilles genitalia revealed a significant effect of sex, species (p-value < 0.001 on both the C8 to C16 and the C16 to C30 datasets, see Fig 5) and a significant interaction between the two variables. When analyzing the results for each sex separately (post-hoc), we found a significant effect of the species variable (PERMANOVA, p-value=0.001 for males and p-value =0.049 for females in the C8 to C16 compounds, PERMANOVA, p-value=0.001 for males and p-value = 0.048 for females in the C16 to C30 compounds), suggesting a substantial divergence between species in chemical bouquets, especially marked in males.

Next, we performed an Indicator Value Analysis to identify the different compounds significantly associated with each species, separating males and females (Table S2). Among all the annotated compounds, we found traces of beta-ocimene, previously identified as a butterfly pheromone. Interestingly, different proportions of this compound were found on the genitalia of M. helenor and M. achilles males.

Discussion

Convergent iridescence between sympatric Morpho butterflies suggests a prominent role of predation on the evolution of iridescence

Our study revealed a striking similarity in the iridescent coloration of the blue patches in the sympatric species of M. helenor and M. achilles in French Guiana, in sharp contrast with the divergent iridescence observed between allopatric populations of M. helenor sampled in different geographic areas. Together with the convergent evolution of the blue band width previously detected in these sympatric species (Llaurens et al., 2021), this observation is consistent with there being a primary role of visual predation on the evolution of wing coloration in Morpho (Chouteau et al., 2016; Stuckert et al., 2014). The results of the visual models showing that these convergent iridescent signals are likely undistinguishable by avian predators further supports there being positive selection generated by predators promoting mimicry in these sympatric species. These results suggest that predation pressure can strongly impact the evolution of iridescence, and are further supported by the lack of directionality for the iridescent signal generated by Morpho wings. Directionality in iridescent signals is expected when sexually selected traits incur a predation cost due to their conspicuousness: directionality might enhance signaling to mates while remaining poorly detected by predators (Endler, 1992). Our study shows that M. helenor and M. achilles wings display a gradual variation of hue and brightness when changing the illumination/observation angles contrary to some other iridescent butterfly species (e.g. Hypolimnas bolina; (White et al., 2015)). The lack of signal directionality observed in these Morpho species thus suggests that the iridescent signal is likely on display for both potential mates and predators to see. These gradual changes in hue and brightness could also enhance the confusing effect generated by the alternate exposure of the iridescent and brown side of their wings during erratic flapping flight (Le Roy, Amadori, et al., 2021; Young, 1971).

Overall, the lack of directionality and the gradually changing aspect of Morpho iridescent signals do not point at any private iridescent signal. Instead, these visual effects are consistent with the emission of a confusing visual signal that increases the chances of escaping predators. The convergence of visual signals could have been favored because of the indirect advantage that results in a reduction in attacks from predators that have experienced and learned to avoid the escape capabilities of these prey, consistent with the hypothesis of escape mimicry (Pinheiro & Freitas, 2014; Ruxton et al., 2004).

Interestingly, our reflectance measurements showed that males are generally brighter than females in both species, implying that the flashes generated by males during flight could be more conspicuous. Field experiments have highlighted the territorial behavior of males found patrolling along rivers and displaying competitive interactions with other males (Le Roy, Roux, et al., 2021). This flight behavior strongly differs from female behavior, which are much less frequently observed in the wild and typically spend more time hidden in the understory (Young, 1973). Considering that these ecological niche differences between males and females could result in different predation pressures in contrasted light conditions (open river banks vs. cluttered understory), iridescent signals could have evolved differently in the two sexes, consistent with a similar effect of divergent micro-habitats of males and females in both species (Allen et al., 2011; Reimchen & Nosil, 2004).

Altogether, our results show that natural selection exerted by predators may have favored the evolution of mimetic iridescent color patterns, as well as sexual dimorphism in sympatric Morpho species.

Evolution of visual and olfactory cues in mimetic sister-species living in sympatry

Our tetrad and male mate choice experiments in M. helenor butterflies clearly show assortative mating based on wing coloration. This result is consistent with the high attraction generated by blue coloration in males, as well as the preference of males for local vs. exotic wing color patterns found in previous field behavioral experiments carried out in Amazonian Peru (Le Roy, Roux, et al., 2021). Our male mate choice experiments suggest that both the wing patterning and the iridescent properties of the blue patch could have an effect on male preference. Furthermore, visual modeling also suggests that the differences of iridescent coloration between the populations of the same species can be perceived by mates. Coloration and color perception are key in mate recognition in many taxa (marine invertebrates (Baldwin & Johnsen, 2009), birds (Caro et al., 2021), mammals (Waitt et al., 2003)) but patterns can also inform mate choice (Houde, 1987; Pérez-Rodríguez et al., 2017). In jumping spiders, both pattern and coloration are important cues used during mate choice (Zhou et al., 2021). Similarly, in Heliconius butterflies, both pigmentary coloration and wing pattern are used in mate choice, although the preference is hierarchical and coloration is the most important cue (Finkbeiner et al., 2014). Further experiments are needed to disentangle the relative effects of iridescent coloration and wing pattern in mate preference in Morpho butterflies, but our results show that iridescent color patterns can be used as mate recognition cues in M. helenor.

Interestingly, intraspecific variations in courtship motivation were detected in our behavioral experiments within the species M. helenor. Individuals from the western areas of the Andes displayed high courtship motivation while butterflies from the Amazonian area, where M. helenor coexists with M. achilles, showed a strong decrease in responses to our visual dummies. The joint convergence of the color pattern and blue iridescence between M. helenor and M. achilles in Amazonia probably makes visual cues poor species recognition signals: visual modeling even suggests that M. helenor cannot perceive the color difference between the two. This is further supported by our male choice experiment showing that although males are attracted by blue iridescent colors and approach each female, they do not preferentially interact with their conspecific female dummies. Selection on visual preference might be relaxed in Amazonian areas where the two sister species are found in sympatry: the shared predator communities may strongly favor convergent evolution of dorsal iridescent color pattern, making visual signals poor species recognition cue.

The convergent evolution of iridescence in sympatric species might thus promote divergent evolution of alternative mating cues. By quantifying chemical compounds found on the genitalia of males and females sampled in Amazonian French Guiana, we did indeed find a strong divergence in the chemical profiles of sympatric M. helenor and M. achilles, especially in males. High male chemical divergence in sympatric species suggests the evolution of female mate choice based on olfactory discrimination (Mérot et al., 2015). Although all compounds were not identified, some of those that were have previously been identified as pheromone compounds used during male/female interactions in butterflies. In particular, beta-ocimene was found on the genitalia of males and is known to contribute to mate choice in courtship (Li et al., 2017) or as an anti-aphrodisiac used by Heliconius males to repel other males after mating with a female (Schulz et al., 2008). Additionally, we found strong divergence between males for heavier chemical compounds (C16-C30). Because cuticular hydrocarbons can be involved in mate discrimination during courtship (Ômura et al., 2020), diverging non-volatile compounds could also be used as a discriminating cue. Although we did not perform mate choices to test for the discrimination of olfactory cue by females, divergence of odorant cues could be due to reproductive interference between sympatric species (Bacquet et al., 2015; Dyer et al., 2014). Altogether, our results therefore strongly suggest that the convergent evolution of iridescent wing pattern between the sympatric M. helenor and M. achilles in Amazonia, likely promoted by shared predation pressures, have relaxed selection on visually-based preferences and may have favored the evolution of divergent olfactory cues.

Conclusions

Overall, our study shows that local ecological interactions could result in iridescent color patterns and mate preferences evolving separately in closely-related sister species. We find evidence of converging iridescent patterns in sympatry suggesting that predation could play a major role in the evolution of iridescence, by favoring the emergence of evasive mimicry in a similar manner as the Müllerian mimicry of chemically-defended species (Pinheiro et al., 2016). There are nevertheless striking differences between the two types of mimicry: while a wide diversity of conspicuous color patterns are involved in Müllerian mimicry (MALLET & GILBERT, 1995), the variation of iridescent patterns involved in evasive mimicry might be more tightly constrained, because iridescence itself directly participates in survival via visual confusion (Murali & Kodandaramaiah, 2020). Moreover, switches in color pattern is generally considered as a magic trait facilitating speciation in chemically defended species (Merrill et al., 2012), and Müllerian mimicry frequently occurs among distantly-related species (Puissant et al., 2023). Here we observe the co-existence of Morpho sister-species displaying convergent color patterns, suggesting that speciation can be promoted by the evolution of traits other than visual such as olfactory chemical cues, or by the specialization in other ecological temporal niches (Le Roy, Roux, et al., 2021).

Altogether, this study shows how convergent trait evolution might lead to mutualistic interactions enhancing species coexistence (Boussens-Dumon & Llaurens, 2021), and alter how selection acts on other traits. It illustrates how the feedback between trait evolution and ecological interactions can fuel speciation and co-existence in sympatry.

Materials and methods

I. Study system

We investigated trait variations in the sister-species M. achilles and M. helenor, that occur in sympatry in the Amazonian basin (Fig 1), with two main sampling schemes, focusing on intraspecific and interspecific variations respectively.

M. helenor populations of Ecuador: Pupae from the Ecuadorian populations of M. h. bristowi and M. h. theodorus were purchased from a breeding farm located in Ecuador (Quinta De Goulaine, https://quintadegoulaine.com/es/papillons.php). These two subspecies of M. helenor are found in western vs. eastern areas of the ecuadorian Andes respectively. These commercially-bought pupae were then raised in insectaries at STRI in Gamboa, Panama, between January and March 2023.

Sympatric species of French Guiana: M. h. helenor and M. a achilles individuals were wild-caught and raised in insectaries between July and September 2023 in the Amazonian forest of the Kaw Mountains (GPS coordinates: 4.57, -52.21), French Guiana, France.

II. Quantification and characterization of coloration and iridescence of Morpho butterfly wings

A. Measures of wing reflectance

We investigated the variations in iridescence in M. h. bristowi, M. h. theodorus (from Ecuador), M. h. helenor and M. a. achilles (from French Guiana), using 10 females and 10 males of each. We estimated the iridescence of the blue patches of the dorsal side of the wings by performing a series of reflectance measurements on wings positioned on a flat surface, following a template to ensure that the wing orientation remained the same across all measurements. Reflectance spectra captured at different angles were recorded on the right forewing, at a specific point located in the wing zone defined by the M3 and Cu1 veins according to the nymphalid ground plan (see (Martin & Reed, 2010)). Wing reflectance was measured using a spectrophotometer (AvaSpec-ULS2048CL-EVO- RS 200-1100nm, Avantes) coupled with a deuterium halogen light source (AvaLight-DH-S-BAL, Avantes) for 21 combinations of illumination and observation angles, covering the proximo-distal and anteroposterior planes of the wing (Fig 2.A, see Supporting Information Text 1 for extended methods).

These measurements were repeated 3 times for all individuals to estimate the repeatability of the reflectance measurements at each angle using the R package rptR 0.9.22 (Stoffel et al., 2017). Since no difference was detected between replicates for any of the tested angles (S12 Fig), the mean of these 3 measurements at each angle for each individual was used in the subsequent analyses.

B. Statistical analysis of iridescence

Each illumination/observation position produced a reflectance spectrum whose characteristics (i.e. distribution of reflectance) define the optical properties of the illuminated surface. We focused our analyses on the wavelength range [300, 700 nm], relevant for butterfly and bird observers. All reflectance spectra were analyzed using the R package pavo2 2.9.0 (Maia et al., 2019).

a. Extraction and analysis of colorimetric variables

To identify the main optical effect produced by the wing at each angle, we extracted three commonly-used colorimetric variables from each spectrum, using the summary() function in pavo2: namely Hue (color in the common meaning of the term - measured as the wavelength at the maximum reflectance), Brightness (average proportion of reflected light - measured by the mean reflectance over each spectra) and Chroma (also referred to as saturation, it characterizes the color’s purity - measured as the ratio of the reflectance range of each spectra by the brightness). Because the normality and heteroscedasticity assumptions were not met for the distributions of values in these three parameters, we relied on permutation-based ANOVAs (vegan 2.6.4 package, (Oksanen et al., 2001)) to test the effect of sex, angle of measurement, locality and species on these three colorimetric variables. To specifically test whether differences in iridescence patterns were observed among species or localities, we estimated the interaction between the effects of Morpho populations and of the angle of illumination.

b. Global iridescence analysis

As iridescence broadly refers to any change in reflectance in relation to illumination and observation angle (Doucet & Meadows, 2009), we characterized the iridescence of each individual using all complete spectra obtained at all angular positions (400 wavelengths x 21 angular positions). We applied a Principal Component Analysis (PCA) to this dataset, where each dot corresponds to the iridescence of an individual wing and thereby showing the similarities and differences in iridescence among groups. The samples from Ecuador and French Guiana were analyzed separately to test for more precise differences in iridescence between allopatric and sympatric Morpho butterfly populations. The coordinates of the 10 first PCs (representing 98% of the explained variance) were then extracted and a PERMANOVA was performed to test for the effect of sex, locality and species on iridescence patterns.

III. Modeling how Morpho butterflies and avian predators perceive M. helenor

To investigate whether the variations observed among the different spectra could be perceived by different observers, we then used the R package pavo2 2.9.0 (Maia et al., 2019) to model both avian and butterfly visual capabilities. We used the avian UV-sensitive vision model implemented in pavo (“avg.uv”) as a proxy for predator vision and specifically built a model of M. helenor vision, using the spectral sensitivity measured in (Pirih et al., 2022). We built an hexachromatic visual model using the sensmodel function, using 345, 446, 495, 508, 570 and 602 nm as the maximum of absorbance in our six simulated cones respectively (S10 Fig). The vismodel function was used to compute the quantum catch of each receptor in the Morpho vision model under standard daylight illumination for each spectrum. Finally, assuming that the discrimination of colors is limited by photoreceptor noise (Receptor-Noise Limited model, (Vorobyev et al., 2001)), we estimated the distance between colored stimuli as perceived by Morphos using the bootcoldist function. We used the default options (weber = 0.1, weber.achro = 0.1) and set the photoreceptor densities for each cones used in the visual model to 1. The resulting chromatic distances (dS) and confidence interval were then used to test how much the visual capacities of Morpho butterflies could allow the discrimination of wing reflectance depending on sex, localities and species, at each angle. Similarly, we used the avian UV sensitive visual model implemented in pavo to test the discrimination of Morpho wing reflectance by predators. We used dS ≥1 Just Noticeable Difference as the threshold of discrimination.

IV. Mate choice experiments

All mate choice experiments took place in cages of similar dimensions (3m x 3m x 2m). All Morpho butterflies were fed 30 minutes before the beginning of the experiments taking place between 8am and 12am.

a. Testing for mate discrimination within M. helenor

To study the discrimination of mating partners within the species M. helenor, tetrad experiments were performed between M. h. bristowi and M. h. theodorus. For each experiment, one virgin female and one mature male of both M. h. bristowi and M. h. theodorus were placed within the same cage. Their starting positions in each corner of the cage were randomized. The first mating occurring out of the four possibilities was recorded and the experiment was stopped. Experiments could also be ended when no mating occurred within 5 hours of trial. We carried out N = 33 trials to estimate whether the first mating observed preferentially involved assortative pairs, due to either male and/or female mate preference, of which N=30 trials actually resulted in assortative or disassortative matings. We statistically tested for departure from the random mating expectation using a chi-square test of independence, confirmed with a Fisher’s exact test.

Note that M. achilles is not commercially bred and wild females are usually all mated, preventing experiments on virgin females in this species, and thus restricting experiments on virgin females in the commercially-bred species M. helenor.

b. Identifying specific visual cues used during mate choice by M. helenor males

To investigate the visual preference of M. helenor males for the dorsal color pattern of females, we performed several choice experiments using dummy butterflies built with actual female wings. To ensure that preferences were not triggered by olfactory cues, all female wings used to build the artificial dummies were washed with hexane before any experiment. The details of the experimental design and specific visual cues tested are described in Figure 6.

Figure 6:

First, we tested male preference for the visual appearance of M. h. bristowi and M. h. theodorus females (Experiment 1). Since M. h. bristowi and M. h. theodorus females differ both in terms of color patterns (width of blue band) and iridescence, we then conducted experiments to isolate these two components.

We specifically tested the effect of the proportion of blue to black areas (Experiment 2) by presenting males M. h. bristowi to female dummies harboring M. h. bristowi wings, except one was modified with black ink to resemble the narrow blue band on M. h. theodorus female wings.

We finally tested the effect of the blue iridescent band within the same black and blue pattern (Experiment 3) by presenting M. h. theodorus males to a M. h. theodorus female with a narrow blue band, and to the modified M. h. bristowi female dummy with a narrow band. The wing pattern modifications were performed using a black permanent marker.

To ensure that ink odor would not bias the behavior of Morpho males, all dummies used in experiments 2 and 3 contained traces of black marker (either on the modified bands or on the ventral side of the wings of natural females).

In all experiments, two female dummies were hanging from the ceiling of the cage, 1.5 m apart, and presented to a live male. The male was released in the experimental cage in front of the two female dummies and was free to interact with them for 10 minutes. The positions of the female models were swapped 5 minutes into each trial to avoid potential position-related bias. Two types of male behaviors were recorded. Each time a male flew within a 40 cm radius from a female model, an “approach” was recorded. Each time a male touched a female dummy, a “touch” was recorded. The number of “approaches” and “touches” toward each female dummy per trial was used to determine male preference. Each male performed this experiment 3 times on 3 different days to account for potential individual variations in mate preference.

c. Testing whether visual cues can act as a pre-zygotic barrier between M. h. helenor and M. a. achilles

In order to investigate the importance of color vision in species recognition, we then performed a male choice experiment with M. h. helenor and M. a. achilles (experiment 4 in Figure 6) in French Guiana, using natural wings of females M. h. helenor and M. a. achilles as artificial butterflies following the exact same protocol.

d. Statistical analyses

Comparing male mating motivation

Because we noticed behavioral differences between the species and subspecies of Morpho during the course of the experiments 1-4, we attempted to account for variations in male mating motivation. The approach and touch rates of males, i.e. their willingness to approach and touch (respectively) any female dummies during each trial, was defined as the proportion of trials where the males approached and touched (respectively) either female dummies at least once. The approach rate and touch rate of M. h. bristowi, M. h. theodorus, M. h. helenor and M. a. achilles males were calculated, and a binomial generalized linear model (GLM) was used to test whether the approach or touch rate was significantly different between males from different populations.

All statistical analyses were performed with the R package glmmTMB (Brooks et al., 2017). The normality, homogeneity of variance and overdispersion tests for the GLMs were performed using the R package DHARMa (Hartig, 2022). We controlled for potential individual variations by using the male ID and trial number as a random effect in all the tests.

Visually-based male mate preferences

We then tested for male preferential attraction for different dummy females. The mean proportion of approaches and touches toward each female dummy was calculated over the four types of trials performed by each male. Non-parametric Wilcoxon tests were used to test whether the mean probability of approaching or touching a dummy 1/dummy 2 was significantly different from the 0.5 proportion expected under random mating. A significant male preference towards either dummy was detected when the mean probability was significantly above 0.5.

V. Pheromone analysis for sympatric species

a. Chemical analyses

We investigated variations in chemical compounds between the two sympatric species M. helenor and M. achilles sampled in French Guiana. The male and female genitalia of wild caught M. h. helenor and and M. a. achilles individuals (from 9 up to 12 per sex per species) were dissected and extracted in 20µL of dichloromethane, containing 5µL of octadecane as an internal standard. Two sets of chromatography were performed in order to extract either the light compounds (all volatile carbonate compounds below C16) or the heavier ones (all carbonate compounds above C16) found on the genitalia of those Morphos.

The chemical compounds of those samples were analyzed by gas chromatography and time-of- flight mass spectrophotometry (GC-ToF-MS) using a 8890 GC system (Agilent, Santa Clara, USA) coupled to a PEGASUS BT mass spectrophotometer (LECO, Saint Joseph, USA). The connected capillary column was a Rxi-5ms (30 m x 0.25 mm, df = 0.25 µm, Restek, Bellefonte USA) and helium was used as carrier gas at 1 mL min^-1 constant flow rate. 1 µL of each sample was injected in a GC split/splitless injector at 300 °C. Two temperature programs were used: (i) the analysis of the C8-C16 compounds started at 40 °C maintained for 1 min, then increased at 210°C at 6 °C min^{- 1} rate, finally increased to 340 °C at a 10 °C min^-1, and (ii) the analysis of the C16-C30 compounds started at 70 °C maintained for 1 min, then increased to 150 °C at 30 °C min^-1 rate and finally increased to 340 °C at a 6 °C min^-1 rate. Mass spectra were recorded with a 70eV electron impact energy. Mixtures of alkanes C8 to C19 and C16 to C44 (ASTM® D5442 C16-C44 Qualitative Retention Time Mix, Supelco, Bellefonte, USA) were injected under the same conditions to be used as external standards.

The chromatograms were converted in netCDF format and operated using MZmine 2.43 for compound detection and area integration. To this aim, chromatograms were cropped before 4 min and after 27 min, and baseline-corrected using the Rolling ball algorithm (Kneen & Annegarn, 1996). Mass detection was performed using the centroid mass detector, and chromatograms of each mass were constructed using the ADAP algorithm (Myers et al., 2017; Pluskal et al., 2010). Chromatograms were deconvoluted into individual peaks using the local minimum search algorithm and aligned on a master peak list with the RANSAC algorithm. Finally, the Peak Finder gap filling algorithm was used to search for missing peaks, and a matrix compiling the area under each peak detected for each butterfly sample was extracted. This matrix was purged from siloxane like compounds as well as compounds present in less than 10% of the samples. The areas of the blanks were subtracted to the samples, and the final data were used for statistical analysis. Mass spectra were then visualized using the ChromaTOF® software, compared to MZmine-selected ones, in order to manually annotate the chemical compounds. Annotation were based on scan comparison with the NIST2017 database, and linear retention indexes (LRI) calculated from the alkane mixtures with free database (https://pubchem.ncbi.nlm.nih.gov/) and literature (Adams, 2017).

b. Statistical analyses of the chemical compounds

In order to quantify variations in chemical compounds detected in the different sample, we performed a nonmetric multidimensional scaling (NMDS) ordination using the Bray-Curtis dissimilarity calculation. By considering the amount of each compound found in our samples as a variable, we performed a permutation MANOVA with the adonis2() function found in the vegan 2 package on R. We tested whether the sex or the species had a significant effect on the chemical bouquet produced by their genitalia.

We also performed an Indicator Value Analysis to identify the chemical compounds significantly associated to our different Morpho samples with the mutlipatt function from the indicspecies package in R using the IndVal index from (Dufrêne & Legendre, 1997).

Acknowledgements

The authors would like to thank Aurélie Tournié and Doris Gomez for the advice provided on iridescence measurements. We are also grateful to Owen McMillan from the Smithsonian Tropical Research Institute (Panama) and Mathieu Chouteau from LEEISA in French Guiana (France) for providing facilities to perform the behavioral experiments. We exported the wings of the butterflies from Panama using exportation permit number PA-01-ARB-028-2023, and declared to the French authorities the exportation of butterfly wings from French Guiana to the French metropole. J.L. PhD was funded by an IBEES grant from Sorbonne Université. This study was funded by the European Union (ERC-2022-COG - OUTOFTHEBLUE - 101088089). Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them.

Additional files

Supplementary files