The correlation between wing interference patterns and body size in Coniceromyia Borgmeier (Diptera: Phoridae) and its implications to the understanding of the former as a sexually selected trait
Abstract
Wing interference patterns (WIPs) are stable structural color reflections of insect transparent wings. The WIP colors are the result of thin-film interference and vary according to wing thickness and other wing characteristics. These patterns have been thought to play a display role during courtship. Recent empirical studies concluded that WIPs affect male drosophilid attractiveness and that WIP evolution is driven by sexual selection. However, these studies did not account for body size variation, a variable that has been demonstrated to be sexually selected and that may be related to wing thickness and WIP color. I consider herein the possibility that body size could be the trait being selected in these studies, and not the WIPs, with the latter being indirectly selected. A first step to consider this alternative hypothesis would be to demonstrate the correlation between WIPs and body size. I analyzed whether such correlation exists through the phylogenetic tree of the genus Coniceromyia (Diptera: Phoridae) by comparing evolutionary models assuming dependent and independent evolution of both traits. I also investigated whether WIPs are correlated to body size within two species of this genus. Strong evidence was found in favor of the correlation between WIPs and body size in the tree analyzed and within one of the two species. If these results are confirmed as a general pattern, the signaling role of WIPs and their direct relation to sexual selection may be questioned by the alternative hypothesis that body size could be the sexually selected trait in recent studies' experiments.
1 INTRODUCTION
Wing interference patterns (WIPs) are stable structural color patterns reflected by transparent wings of small insects (Shevtsova et al., 2011). Only 20% of the light is reflected as WIPs, making them visible only against a dark background when the light reflected by the background does not overpower the WIPs (Shevtsova et al., 2011). The WIPs are the result of thin-film interference effect—the interference between the light transmitted through the wing and reflected by its lower boundary and the light reflected by its upper boundary (Shevtsova et al., 2011). The angle of light incidence and the wing thickness directly influence the distance the light is transmitted through the wing and the color resultant from the interference. As microcorrugations of insect wings cause the angle of light incidence to be relatively constant, the wing thickness is the main determinant of which colors are reflected by the wing following the Newton series scale of two-beam interference colors (Shevtsova et al., 2011).
Wing interference patterns may vary considerably among species or even individuals of a single species, and may allow the recognition of cryptic species (Butterworth et al., 2021; Hosseini et al., 2019; Pielowska-Ceranowska & Szwedo, 2020; Shevtsova & Hansson, 2011). Despite their pronounced variation, WIPs appear to also contain conserved components that could be used as relationship evidence in phylogenetic analyses (Buffington & Sandler, 2011). WIPs vary in color or shape of the patterns (Shevtsova et al., 2011). The color variation is often attributed to differences in wing thickness, with each value of wing thickness between ca. 50 and 1500 nm corresponding to a specific color in the Newton scale (Fusu, 2017; Shevtsova et al., 2011). The variations in shape of the patterns were divided into four categories by Buffington and Sandler (2011): radiform, striatiform, campiform, and galactiform. WIP color and shape are uniform within sexes for some species (Hosseini et al., 2019) but sexually dimorphic for others (Butterworth et al., 2021; Shevtsova et al., 2011).
Wing interference patterns have been thought to function in intra- and interspecific signaling and courtship displays (Shevtsova et al., 2011). Assuming that WIPs could have these aforementioned functions, their diversification could be partly due to sexual selection. Recent studies concluded that WIPs are under sexual selection by demonstrating that populations under stronger sexual selection regimes have higher diversification of WIPs (Hawkes et al., 2019) and that males with specific WIPs are more attractive to females (Katayama et al., 2014). These studies measured WIP diversity as the differences in WIP color components such as hue, color contrast, saturation, and brightness. These components are directly related to which color of the Newton scale the WIP is reflecting and therefore related to wing thickness and possibly to body size. This possible relation between WIPs and body size was not investigated in these studies, though Katayama et al. (2014) mentioned body size as a possible factor influencing their results. Body size has been demonstrated to be a sexually selected trait in several studies (e.g., Crespi, 1989), and, in some cases, it was shown to explain differences in attractiveness that would otherwise be attributed to secondary sexual traits (Baur et al., 2020).
I herein consider the possibility that body size could be the signaling element being selected in the studies associating WIPs with sexual selection, and not the WIPs themselves, in which case the latter would be an indirectly selected by-product. Therefore, a first step to consider this alternative explanation would be to investigate whether WIP color as a continuous variable indicative of wing thickness could have a correlated evolutionary history with body size. I investigated whether there is such a correlation between these traits along with the phylogenetic history of the phorid fly genus Coniceromyia Borgmeier, 1923 (Diptera: Phoridae). I also analyzed whether these traits are correlated within two species of this genus.
Coniceromyia comprises 100 species of tiny flies ranging from 1.4 to 3.7 mm of body length (Ament et al., 2020). Recent studies reviewed the taxonomy of the genus (Ament & Amorim, 2016), described several new species (Ament et al., 2020), and inferred the relationship among its species (Ament et al., 2021). These contributions provided a basis for comparative studies in the genus, which is generally lacking for diverse and small insects. Coniceromyia is also a specially interesting case study given that some species of this genus have male-exclusive features on the wing, such as maculae of different shapes, elongated microtrichia, emarginations, and drastic modifications of wing veins (Ament & Amorim, 2016; Ament et al., 2021). These features were assumed to have a signaling function possibly related to sexual selection (Ament et al., 2021). The effect that some of these wing features have on the WIP display is also illustrated and discussed herein.
2 MATERIALS AND METHODS
Wing interference patterns of males of 54 of the 100 species of the genus Coniceromyia were photographed (Table S1). The specimens had been collected and stored in alcohol and chemically dried using HMDS following the protocol of Brown (1993). All specimens studied were in good condition before and after the drying process, with no deformations or loss of pigmentation (this could be verified in the taxonomic revisions that also studied this material; Ament & Amorim, 2016 and Ament et al., 2020). The only preservation problem observed occurred on some wings, which were partially folded making WIP observation impossible across their entire wing membrane (e.g., Figure 3d,l). Either HMDS drying or critical point drying is the recommended method for the study of WIPs as they preserve features of wing microsculpture that, if altered, could affect the WIPs displayed (Pielowska-Ceranowska & Szwedo, 2020).
Wing interference pattern photography was standardized by positioning all specimen wings horizontally, against the same black background. Photographs were taken with a Zeiss Axiocam 506 color digital camera coupled to a ZEISS SteREO Discovery V12 stereomicroscope with a PlanApo S 1.0x FWD 60 mm objective. The illumination was provided by fluorescent ring light (Leica High Performance Lampe FLRL) with standardized intensity and daylight color temperature: 5000K–6500K. This type of illumination was chosen as it allowed the visualization and image acquisition of WIP to be consistently done, as required for the analyses of this study.
Wing interference patterns were observed in an intermediate line between wing veins M1 and M2 identifying in this line the transitions between the colors of the Newton scale (Figure 2a,b). Afterward, I stretched or shortened the colors of the Newton scale to fit into the color transitions observed, creating a stripe representative of the WIP. I attributed the wing thickness in eleven equally spaced points along the WIP stripe using the Newton scale calibrated to the chitin refractive index (Figure 2b, Shevtsova et al., 2011). The area under these eleven points was calculated as in a line graph through the trapezoidal approximation to estimate the total wing thickness along this line (Tables S2–S7). Body length was measured from the insertion of the antenna to the apex of the sixth abdominal tergite. This distance was chosen to represent body length following studies on Phoridae taxonomy (e.g., Ament et al., 2020; Kung, 2009).
The evolution of wing thickness and body length was simulated based on the phylogenetic tree of Coniceromyia (Ament et al., 2021) under continuous correlated and non-correlated models of evolution in BayesTraits (Pagel, 1999; Pagel et al., 2004). The simulations were MCMC runs with 100 stepping stones and 1000 iterations per stone. The difference in log-likelihood of both models was compared through a Bayes factor test.
For the interspecific analysis, single specimens were studied as representatives of each of the 54 Coniceromyia species herein sampled. Single-specimen sampling, also used in other macroevolutionary studies (e.g., Beerli et al., 2019 Zeuss et al., 2016), was necessary for Coniceromyia given that many of its species are known from a single or few specimens (Ament et al., 2020). This sampling approach is more susceptible to possible differences between the values measured for the specimens and the mean of the species regarding the traits studied (Zeuss et al., 2016). However, the examination of WIPs and body length within two Coniceromyia species with multiple exemplars available for study indicated that the intraspecific variation of these traits is relatively low compared with the total interspecific variation perceived. The standard deviations of body length of Coniceromyia luna Ament et al., 2020, 0.2 mm, and of Coniceromyia piricornis Borgmeier, 1950, 0.1 mm, are relatively low values compared with the total range of interspecific body length variation of 1.66 mm. Similarly, the standard deviation of wing thickness measured through WIPs of C. luna and C. piricornis is 13.20% and 5% of the total interspecific variation found for this trait. Therefore, the deviation of the specimens sampled from the mean of the species is expected to be relatively low and randomly distributed, resulting in a small but non-significant noise to the interspecific correlation analysis (as in Beerli et al., 2019).
As for the intraspecific analyses, I selected C. luna Ament et al., 2020 and C. piricornis Borgmeier, 1950 mostly due to the higher availability of specimens to be compared. The same measurements of WIPs and body length were taken, except that only the apical 64% and 91% of the WIP stripe were measured for C. luna and C. piricornis, respectively, where the colors could be more clearly observed. The correlation between body size and wing thickness within each species was investigated through linear regression.
I used the definitions of striatiform and campiform WIPs of Buffington and Sandler (2011). Striatiform WIPs are approximately parallel striations that differ in colors likely in consequence of differences in wing thickness. Campiform WIPs are a large area of wing membrane with a single WIP color. The term “wing macula” is used to refer to the darkening of the wing membrane caused by differential pigmentation. For Coniceromyia with maculated wings, light photographs of slide-mounted wings are also presented to show the macula intensity and distribution through the wing of each species, but these photographs were not obtained in a standardized way and are not of the same specimens than the photographs for WIP observation.
2.1 WIP color as a direct indicator of wing thickness in Coniceromyia
Besides the wing thickness, the WIPs can be influenced by wing features such as wing pigmentation, chaetotaxy, and venation (Shevtsova et al., 2011). Wing pigmentation clearly influences the WIP color in Culicoides Latreille, 1809, for example, as there is a clear difference of the WIP colors observed on pigmented and on adjacent non-pigmented areas of the same wing (Pielowska-Ceranowska & Szwedo, 2020). In these cases, WIP color could not be a reliable approximation of wing thickness as the colors displayed seem to be influenced by differences in other wing features as well.
The Coniceromyia wings with WIP colors used as indicative of wing thickness in this study do not have considerable differences regarding wing pigmentation, chaetotaxy, and venation. These wings have in general a subtle pale brown color (almost hyaline) and a similar microtrichia pattern. The line selected for the WIPs observation itself was chosen to be in a homologous area as far as possible from the wing veins, avoiding the influence the veins could have on the WIPs. Additionally, all Coniceromyia with wing thickness measured in this study have the striatiform WIP pattern, displaying colors following the sequence of the Newton scale. This type of WIP pattern facilitates the color observation and wing thickness measurement as it follows an already known order of colors. Coniceromyia species with wings differing from these patterns of wing pigmentation, chaetotaxy, and venation were not included in the quantitative analyses of this paper as it was considered that these differences could affect the correspondence between WIP and wing thickness. These other species are discussed in a separate session.
The photographs used for WIP observation herein were obtained from a camera that does not have a linear response to the light intensity of the red, green, and blue (RGB) wave bands of the specimen. Commonly, cameras have this non-linear output to produce images with better quality when observed on monitors that have their own non-linearities (Stevens et al., 2007). The photographs of this study were not calibrated for these non-linearities, and therefore, they should not be used to quantitatively measure the values of the RGB color components.
However, the objective of the observations performed here is not to measure such values, but to identify which WIP colors are present on the wing in comparison with a predefined scale (the Newton scale). This identification was done manually through observation of the photographs on a monitor. It is noteworthy that when non-linearized images are observed on a monitor, the non-linearities of the camera are approximately canceled out with the monitor non-linearities (Stevens et al., 2007). Finally, as the data were gathered for comparative analyses, its most important aspect is the relative value of each specimen in comparison with the other specimens. The comparativeness of these values is maintained by the standardization of the methods for photograph acquisition and WIP observation.
3 RESULTS
Most Coniceromyia WIPs examined are striatiform, transversal to the wing veins (Figure 1). The succession of colors is approximately continuous, with no color in a large space of the wing as in the campiform patterns. The only exceptions are on the maculated wings and wings with other features as discussed ahead. The WIPs varied from dark purple (165 nm) to dull pink (1050 nm).
3.1 WIPs and body size correlation
The disposition of the WIP stripes of Coniceromyia specimens following their order of body size illustrates the possible correlation between these traits—specimens with larger body size in general have greater parts of their stripe with WIPs of larger thickness (Figure 2c,e). The model assuming correlated evolution between both traits on the phylogenetic tree had a logL of −387.934, whereas the model assuming independent evolution between these traits had a logL of −393.687. These values in the Bayes factor test resulted in 11.507, indicating strong evidence in favor of the dependent evolution between these traits.
Coniceromyia luna specimens varied ±0.24 mm in body length, while C. piricornis specimens varied ±0.18 mm. The increase in body length in C. luna is accompanied by the shortening of the WIPs of lower thickness in the WIP stripe (especially the apical WIPs in Figure 2d, yellow, ca. 275 nm, purple, ca. 330 nm, and blue, ca. 375 nm). The WIP color correspondent to the largest wing thickness observed in C. luna, the pale green color (ca. 760 nm), is present exclusively on specimens with 1.96 mm or more of body length, at the base of their WIP stripe (Figure 2d). This apparent correlation between body length and the respective wing thickness of the WIP colors was recovered for C. luna by the linear regression performed (p-value of 0.027).
In C. piricornis, the increase in body length does not seem to be associated with the distribution or presence of colors in the WIP stripe (Figure 2e). For example, there is no considerable difference between the distribution of the WIP colors of the smallest and the largest specimens, C. piricornis N and D, respectively (Figure 2e). Also, the WIP color indicative of the thinnest wing membrane observed in C. piricornis, dark purple (ca. 165 nm), is only present in specimens of intermediate size (C. piricornis L and I; Figure 2e). The linear regression performed confirmed for this species the lack of correlation between body length and respective wing thickness of the WIP colors (p-value of 0.23). Body size–WIP correlation might not be recovered for this last species due to the small sample size and possible imprecisions in WIP observation and measurement, resulting from wings being more frequently folded.
3.2 Coniceromyia male-exclusive wing features and their consequence on WIPs
Species with maculated wings were found to also display striatiform WIPs but with some characteristics not seen in non-maculated wings (Figure 3). The maculated wings examined frequently have a succession of several narrow bands of WIP colors marking the transition between the non-pigmented and pigmented wing membrane (Figure 3, light gray arrows). Also, wing membranes with more subtle macula may display WIPs with either reddish or a complex mixture of red and green colors, possibly a combination of the brown of the macula with the pink hues of the WIP (Figure 3, orange arrows).
No consistent WIP was found for all of the wings with large space between veins M2 and CuA1 (Figure 4). Coniceromyia disparivena and Coniceromyia sanctaetheresae have the only campiform WIP among all species studied, with purple and blue occupying a large space on their wing between wing veins M2 and CuA1 (Figure 4). However, Coniceromyia apechoneura, the sister group of C. disparivena (Ament et al., 2021), has a more usual striatiform WIP. It is noteworthy that the only campiform WIPs in the genus are found in species with the widening of the space between veins M2 and CuA1. The signaling function of both the wide space between veins and its unusual WIPs, however, may still be considered uncertain due to the lack of behavioral observations of these species.
4 DISCUSSION
4.1 The relation between WIPs and body size and its consequences to the understanding of the former as a sexually selected trait
Assortative mating related to body size is widespread across animals (Andersson, 1994; Crespi, 1989). In Drosophila, several works demonstrated the correlation of male size and number of mating events (Partridge et al., 1987), mate choice (Sisodia & Singh, 2004), and effectiveness of courtship songs produced by larger wings (Aspi & Hoikkala, 1995). Such mating success has been demonstrated even in natural conditions (e.g., Markow & Ricker, 1992; Santos et al., 1992). The female preference for larger body size seems to benefit from the fact that in some cases larger individuals may have better fitness components (Cordero, 1995; Johnstone, 1995; Partridge & Fowler, 1993).
The influence of body size on WIPs has been noted in other studies mostly through general observations that larger specimens have WIP colors corresponding to larger wing thickness in the Newton scale (e.g., Butterworth et al., 2021; Fusu, 2017; Shevtsova et al., 2011). The present study goes further by demonstrating that the wing thickness measured through the WIP colors is statistically correlated to body length in two of the three analyses performed. In both of these analyses, body size is an important but not perfect predictor of WIPs and some specimens diverge from the clear statistical signal as outliers (e.g., Figure 2c,f). These divergences could have different causes, including selective forces on wing thickness related to flight aerodynamics or other functions of the wing. In any case, body size should be considered when investigating the effects these factors may have on WIPs given the results of this study and the following discussion.
- Wing interference patterns are the sexually selected signal. Female preference for certain WIPs benefits from more attractive male offspring with better courtship performance (runaway hypothesis, Katayama et al., 2014; Hawkes et al., 2019; e.g., for other groups, Prokop et al., 2012).
- Wing interference patterns are the sexually selected signal and honest indicator of male body size and fitness quality (good-genes hypothesis; Katayama et al., 2014; e.g., for other groups, Puurtinen et al., 2009). Female preference for certain WIPs benefits from offspring with higher fitness.
- Body size is the sexually selected signal, with WIPs being an indirectly selected by-product. This hypothesis is hereafter referred to as “selection for body size.”
4.2 Can selection for body size explain the results of Katayama et al. (2014) and Hawkes et al. (2019)?
This question could be directly assessed whether the analyses of these studies corrected for body size variation. As body size data are not available, I discuss mainly the results of Katayama et al. (2014) in light of the correspondence between the components of WIP colors they evaluated and wing thickness. Katayama et al. (2014) compared the attractiveness of male drosophilid lines with different WIPs in black and white backgrounds. Differences in attractiveness between the backgrounds would suggest that WIPs are influencing the results as they are not visible in white backgrounds. Male drosophilid attractiveness was found to be positively and significantly correlated between black and white backgrounds. Thereafter, Katayama et al. (2014) calculated the residual attractiveness of the backgrounds correlation and investigated the correspondence between this residual attractiveness and the components of WIP colors hue, saturation, and brightness.
Residual attractiveness was significantly correlated to WIP hue and saturation, with females preferring males with intermediary values of hue (magenta among the WIP colors they analyzed) and higher saturation values (their Figure 3). Female highest preference for magenta hue is apparently against selection for body size given that magenta is not the hue representative of the larger wing thickness among the hues sampled. Specific blue WIPs with lower hue values represent larger wing thickness (Figure 5a) and, according to selection for body size, would be preferred. However, lower hue values are also present in colors of lower wing thickness, which would have lower attractiveness (Figure 5b). Therefore, preference for intermediary hue values (magenta) could be explained by selection for body size as WIP colors that would be indirectly favored according to selection for body size (fourth category in the x-axis of Figure 5a) share lower hue values with colors that would be disfavored in relation to magenta hues (first category in the x-axis of Figure 5a).
Selection for body size possibly also explains the higher attractiveness of WIPs with higher saturation as there is an apparent correlation between wing thickness and color saturation for the WIPs sampled (Figure 5b). The thicker the wing, the longer are the intervals of high saturation values (over 90%) and the shorter are the intervals of lower saturation values (Figure 5b). However, the heterogeneity of the saturation data and the frequencies of these saturation values are important factors to consider when evaluating the correlation between saturation and body size.
Hawkes et al. (2019) showed that male drosophilids after successive generations under stronger sexual selection had WIPs with higher average luminance (perceived brightness), luminance contrast, and frequency of colors within the green and blue wavelength regions in comparison with females and males in other sexual selection regimes. It is not possible to analyze their results as done for the results of Katayama et al. (2014) given that they did not specify which WIP color range in the Newton scale they studied. However, their results apparently could also be explained by selection for body size as this selection would favor larger males causing a higher diversification of this trait (and consequently of WIPs) in males of stronger sexual selection regimes.
4.3 What else can be used to evaluate selection for body size?
Sexually dimorphic WIPs differing in the shape of the color patterns, as Shevtsova et al. (2011) indicated in Figure 3a–h may add to the evaluation of selection for body size and the other hypotheses. Pigmented wings displayed in courtships have their maculae frequently varied in shape, possibly in response to sexual selection (Edwards et al., 2007). It can be argued that this sexual selection pressure would also result in WIPs with sexually dimorphic shapes in case they are the sexually selected signal (hypotheses 1 and 2) (Shevtsova et al., 2011). The commonness of sexual dimorphism in WIP shape, however, remains to be better understood.
4.4 The influence of wing maculae on WIPs
The effects of wing maculae on WIPs were recently discussed for Culicoides wings (Pielowska-Ceranowska & Szwedo, 2020). In general, the Culicoides have a complex mosaic macula through the wing that seems to affect the WIPs mostly by causing a WIP color of higher wing thickness in the pigmented areas compared with its surrounding non-pigmented areas (observable in almost all figures of Pielowska-Ceranowska & Szwedo, 2020).
The wing maculae of Coniceromyia differ from the ones of Culicoides by having a more restricted distribution through the wing and in most cases being darker, without any WIP visible in the maculated areas (Figure 3). The wing maculae of Coniceromyia seem to affect the WIPs in two main ways. One way is the formation of a succession of several narrow bands of WIP colors marking the transition between the non-pigmented and pigmented wing membrane. These transitional WIP bands can be seen also on some of the figures of Shevtsova et al. (2011, their figure 4m,o) and on Culicoides wings where there is a single stripe marking the transition (e.g., Figure 2 of Pielowska-Ceranowska & Szwedo, 2020). In all of these cases, the transitional WIP bands follow the transition of colors of the Newton scale from colors of lower thickness on the non-maculated membrane to colors of higher thickness on the maculated membrane. This seems to indicate that the maculae and their pigments could be associated with the thickening of the wing membrane. Another possibility is that the wing pigment is affecting the WIP display. The other way wing pigmentation seems to affect the WIPs in Coniceromyia is by forming a complex mixture of red and green WIPs in the areas where the macula is more subtle. This effect can be also observed in Culicoides (Pielowska-Ceranowska & Szwedo, 2020, figure 6b).
Whether these WIP effects particular of maculated wings have a biological meaning or are only physical consequences of the presence of wing pigmentation remains a question to be investigated. The same question can be applied to the unusual WIPs found for Coniceromyia with other male-exclusive features such as the wide space between veins M2 and CuA1.
5 CONCLUSION
The biological meaning of WIPs remains intriguing. This study demonstrated that WIPs can be correlated to body size and brought the possibility of different explanations for the results of recent experiments of sexual selection. The contrasting views of WIPs as visual signals or by-products of body size may influence even the way they are measured, as colors (as in Katayama et al., 2014) or wing thickness (as in this study). These different views also affect how the WIP variation is perceived as, for example, a different WIP color present in a particular specimen could be more meaningful if this color is understood as part of a visual signal than if it is seen as a slightly thinner wing membrane. More investigation of the correlation between WIPs and body size and copula experiments in different backgrounds correcting for body size variations are essential steps to evaluate whether WIPs are visual signals under sexual selection or not.
ACKNOWLEDGEMENTS
This research was supported by scholarships granted by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brazil (CAPES)—Finance Code 001 and by the São Paulo Research Foundation (FAPESP) via grant #2018/09666-5 to E.A.B. Almeida. I am very grateful to Eduardo A. B. Almeida, Meire Telles, and three anonymous reviewers for a careful revision and important suggestions on the manuscript. I am also greatly thankful to Eduardo Almeida for providing the conditions for the elaboration of this work.
