Wing interference patterns (WIPs) are color patterns of insect wings caused by thin film interference. Thin film interference is the same phenomenon responsible for the refracted spectral colors sometimes visible on soap bubbles. Insect WIPs are static patterns due to the variable thickness of wing membranes and the colors produced depend on the thicknesses of wing membranes. While WIPs have been studied in several taxa of small insects, they have not been broadly adopted by insect taxonomists. We surveyed WIPs in one moderate-sized genus of parasitoid wasps, Parapanteles (Braconidae: Microgastrinae). Using an inexpensive microscope camera set-up and free imaging and analysis software, we detected consistent WIP differences between Parapanteles species. In some cases, WIPs can be used to diagnose sibling species that would otherwise require SEM images to differentiate or DNA barcodes. Wing interference patters are an underemployed character that may be similarly useful in many other taxa of small clear-winged insects.

Braconidae, color patterns, Microgastrinae, Parapanteles, WIP, Wing interference patterns

Wing interference patterns (WIPs), the rainbow colors that can appear on clear insect wings against dark background, have not been broadly adopted by insect taxonomists as morphological characters. Shevtsova et al. (2011) comprehensively investigated and called attention to these patterns, discovering that they are stable non-iridescent color patterns produced by thin film interference, where light that is reflected off of the upper or lower surface of a clear membrane constructively or destructively interferes with light approaching the membrane. The perceived color pattern is primarily caused by the varying thickness of the wing itself, and are, unlike iridescent colors from butterfly wing scales, static at a range of viewing angles (Shevtsova et al. 2011).

Wing interference patterns are under-used in species descriptions and as a tool for species-level identification of small clear-winged insects. Since their discovery as stable color patterns, they have rarely been reported in taxonomic works and even less frequently been used in species diagnoses or identification keys. In addition to discovering them, Shevtsova et al. (2011) comprehensively described the physical phenomenon that causes them and documented examples of WIPs in several Diptera and Hymenoptera taxa. Since then, WIPs have been documented in just 20 taxonomic or descriptive works (Hansson 2011; Shevtsova and Hansson 2011; Buffington and Sandler 2012; Hansson 2012; Hansson and Shevtsova 2012; Hernández-López et al. 2012; Simon 2012; Stigenberg 2012; Buffington and Condon 2013; Mitroiu 2013; Buffington and Forshage 2014; Drohojowska and Szwedo 2015; Zhang et al. 2014a, 2014b, 2016; Hosseini et al. 2019, 2020, 2021 2021; Pielowska-Ceranowska and Szwedo 2020; Butterworth et al. 2021; Conrow and Gelhaus 2022) and five experimental studies (Katayama et al. 2014; Brydegaard et al. 2018; Hawkes et al. 2019; Dong et al. 2020; White et al. 2021). Most of these studies focus on Hymenoptera (161 species), followed by Diptera (58 species), Hemiptera (8 species), and Odonata (1 species) (Suppl. material 1).

Because WIPs are a function of the varying thickness of wings, some authors have speculated that color may vary intraspecifically because overall wing thickness may be correlated to individual size (Shevtsova and Hansson 2011; Hernández-López et al. 2012). Therefore, they concluded that the colors of WIPs are less important than the patterns they form. Despite this, the majority of taxonomic works that document WIPs describe them in terms of qualitative colors and the relative portion of the wing those colors occupy (e.g., distal 1/3 magenta). Wing interference patterns have been used as characters in species diagnoses in only three publications to date (Hansson 2011; Shevtsova and Hansson 2011; Hansson and Shevtsova 2012), and have been used in a taxonomic key only three times (Mitroiu 2013; Zhang et al. 2014b; Hosseini et al. 2021). Efforts to quantify and compare WIPs have generally found them to be species-specific but rarely sexually dimorphic (Hawkes et al. 2019; Hosseini et al. 2019; Hosseini et al. 2020; Butterworth et al. 2021; White et al. 2021). To-date, WIPs have not been broadly adopted by taxonomists of small insects.

Microgastrinae (Hymenoptera: Braconidae) is a hyper-diverse subfamily of small parasitoid wasps that attack Lepidoptera (Mardulyn and Whitfield 1999). Microgastrinae currently has 2,999 described species (Fernández-Triana et al. 2020), representing roughly 5–10% of the estimated worldwide diversity of this group (Rodriguez et al. 2012; Fernández-Triana and Ward 2015). Their diminutive adult size and small number of morphological characters have made the generic-level taxonomy of this group difficult, and species-level diagnoses, absent DNA barcoding, often rely on subtly variable or minute characters that often require SEM imaging to observe (e.g., Valerio et al. 2009). Wing interference patterns have never been reported for microgastrines but are readily visible in living wasps in a container (DHJ, WH, personal communication).

Parapanteles Ashmead is a small genus of Microgastrinae with several species that are morphologically very similar to other genera (Dolichogenidea and Glyptapanteles) and frequently misdiagnosed (Valerio et al. 2009; Parks et al. 2020). Here, we document the WIPs of 7 described and 12 putative undescribed Parapanteles species from Costa Rica and Ecuador and present a simple and inexpensive method for quantifying and comparing WIPs that can contribute to identification keys and rapid species diagnosis.

Here we present the first study of WIPs in Microgastrinae (Hymenoptera: Braconidae), and an attempt to quantitatively compare the WIPs of closely related species using materials and methods already common in or freely available to most taxonomic laboratories that focus on small clear-winged insects.

Adult wasps used in this study were collected by two long-term Lepidoptera/parasitoid rearing projects: Área de Conservación Guanacaste (ACG) in Costa Rica (Janzen and Hallwachs 2009, 2016) and Yanayacu Biological Station in Ecuador (Dyer et al. 2017). A list of specimens used in this study is available in Table 1.

One set of fore and hind wings were removed from each adult wasp from samples stored in ethanol. Where available, wings from one male and one female per brood were removed and slide mounted on temporary slides. All species sampled are gregarious (i.e. the female lays multiple eggs in a single host) except Parapanteles sp. J and Parapanteles sp. K, which are solitary (i.e. females lay a single egg per host). We assume that all wasps eclosing from the cocoons from one caterpillar are siblings. Wings were sandwiched between two microscope slides which were taped together at the ends. This flattens wings more reliably than using a standard slide cover. As in Shevtsova and Hansson 2011, a drop of black India ink was spread on one slide to create a uniform black background behind the wings.

Wings were photographed at 50× magnification using a Cannon Rebel Xsi camera and an Amscope LED-144A-YK 144 LED ring light at maximum brightness. Wing images were not visually adjusted. Materials examined and qualitative descriptions of WIPs are available in Suppl. material 2. Images used in our analyses are available in Suppl. material 3.

The average RGB (red, green, and blue) values of pixels in each fore wing image were measured using the “RGB Measure” feature in ImageJ v1.49 (Schneider et al. 2012). The value for each color component was divided by the average of all three average color values to calculate the relative “redness,” “greenness,” and “blueness” of each fore wing image (e.g., redness=R/((R+G+B)/3)). This averages out the contribution of black (R/G/B=0/0/0), white (R/G/B=255/255/255), and grey (R/G/B are all equal) pixels.

Arrays of relative redness, greenness, and blueness for each species were tested for normality in R v4.2.2 (R Core Team 2017) using the ‘agricolae’ and ‘nortest’ packages (Gross and Ligges 2015; de Mendiburu and Yaseen 2020) via the Shapiro-Wilk test and for skewness, and then compared across species via ANOVA and Tukey’s HSD test and visualized with ggplot2 (Beck 2017). Species with sample size lower than 3 were excluded from our statistical analysis. Data files and R code are available in Suppl. material 4.

Several metrics of fore wing size were measured to test whether they correlated with WIP patterns, because if they do then species-level differences in WIPs may simply be caused by some species being larger than others. Fore wing length (measured from the junction of C+Sc+R and M+Cu to the distal end of 3/M) and area were compared to each color array. In addition, overall fore wing shape was measured by dividing length by width (measured from the junction of r-rs and the stigma to the distal end of the anal lobe) to test if wing narrowness has any effect on wing thickness. Measurements were done in ImageJ v1.49 (Schneider et al. 2012) and tested for correlation via the Pearson Correlation test in R v4.2.2 (R Core Team 2017) using the ‘hmisc’ package (Harrell and Dupont 2019).

Linear discriminate function analyses were used to test how useful our quantification of microgastrine WIPs were by themselves for identifying species. Linear discrimination analyses were done in R v4.2.2 (R Core Team 2017). Several subsets of models were tested, and variables included the relative redness/greenness/blueness values for both fore wing and hind wing for all species, fore wing only for all species, hind wing only for all species, fore wing and hind wing data for each subclade containing two or more taxa, fore wing and hind wing data for species collected in the same country (Costa Rica or Ecuador), and fore wing and hind wing data for species that attack the same host family (Erebidae, Geometridae, Notodontidae, or Saturniidae). In each case 50% of the dataset was used to train the model and 50% of the dataset was used for validation. R code and data files are available in Suppl. material 4.

The wing interference patterns of Parapanteles are consistent within species and distinct between species, often enough to be diagnostic by themselves. Among the species surveyed, the WIPs of Parapanteles tessares, P. continua, P. sicpolus, P. sp. H, and P. sp. C were the most distinct. These species tended to have more green and purple in their WIPs, while the remaining species’ WIPs were predominantly red and/or yellow.

Wing interference patterns are directly related to the thickness of wing membranes, and previous publications have speculated that WIP colors should change as individuals get larger because cuticle thickness may increase with body size (Shevtsova and Hansson 2011; Hernández-López et al. 2012). We are not aware of any studies investigating the allometry of body or wing cuticle thickness. Among the species we surveyed, some relative redness, greenness, and/or blueness arrays were significantly correlated with wing size and/or area in some species, but in each of these cases the slope of the corresponding linear regression was very slight (Table 3). Correlation with wing size (as a proxy for body size) alone does not account for the differences between the WIPs of closely related Parapanteles species. We were not able to use WIPs alone to reliably predict the identity of an unknown specimen from a large number of species, but were able to discriminate between species in some subclades or subsets of species defined by location or host use (Suppl. material 4). Wing interference patterns are not likely to be useful for automated species identification for many taxa, but are useful as an additional and generally overlooked morphological character to be used in conjunction with other characters for species diagnosis, as any morphological character traditionally would be. When viewed this way they are often one of the most conspicuous and accessible morphological characters of the physically small taxa on which they appear.

Wing interference patterns are directly related to the wavelength of the light passing through the wing membrane, which is a major weakness for using any measurement derived from RGB values for diagnostic purposes. The relative RGB values we measured in this study were not consistent if the wing was illuminated with a different light source. This limitation can be solved by using a consistent light source, and the light source which we used for all WIP photographs in this study, an Amscope LED-144A-YK 144 LED ring light, is widely available and relatively inexpensive. Using one or more lasers of specific wavelengths to illuminate WIPs could offer a more replicable and standardizable method for documenting WIPs, although using one or a few wavelengths would result in less data than full spectrum white light. Wing interference patterns can be observed in situ on pinned specimens, but these are of little use compared to WIPs observed on slide-mounted wings. Including WIP slides (wing slides with India Ink painted on the back) of at least a few paratype individuals with the type series of small winged insects would ameliorate most of the problem posed by variations between light sources, and expand the usefulness of WIPs for future studies.

Experiments in Drosophila have repeatedly shown WIPs to be subject to sexual selection (Katayama et al. 2014; Hawkes et al. 2019). While this has not been experimentally tested in other taxa, this and other studies have found that WIPs are frequently species-specific (Shevtsova et al. 2011; Buffington and Sandler 2012; Zhang et al. 2014b, 2016; Hosseini et al. 2019; Butterworth et al. 2021; Hosseini et al. 2021). Similarly to the Drosophila species used in the sexual selection experiments, microgastrinae males also display their wings to females during courtship (Bredlau and Kester 2019). The colors of WIPs are visible in situ and in natural settings whenever insect wings are displayed in front of a dark background (e.g. green leaves), and the colors that compose them occur in spectra visible to most insects (Shevtsova et al. 2011; Brydegaard et al. 2018; Butterworth et al. 2021). Anecdotally, we found that closely related sympatric species tended to be more subjectively different (i.e. ((Parapanteles tessares, P. continua), P. sicpolus) and (P. em, P. valerio05) from Costa Rica and (P. sp. B, P. sp. C) from Ecuador), while closely related allopatric species tended to be less distinct (i.e. (P. paradoxus, P. sp. I) and (P. sp. E, P. tlinea) (Fig. 2). This suggests that WIPs may be used by microgastrines for conspecific recognition, but this is entirely speculative and would require a broader survey of microgastrine WIPs to test. We only included two solitary species (i.e. females oviposit a single egg into each host, P. sp. J and P. sp. K) in our study. These two species had the most variable WIPs and wing sizes. The relationship to host quality and adult wasp size may be more direct in solitary species that use small host caterpillars than gregarious species attacking larger caterpillars. In such solitary species, poor quality hosts may have less resources available for parasitoids and result in smaller adults, while gregarious species can oviposit fewer eggs to account for poor quality hosts which may result in more consistent adult wasp sizes. Even so, P. sp. J fore wings are significantly redder than P. sp. K (Figs 1, 2).

Figure 2. 

Right wings of three different individuals from seven Parapanteles species showing wing interference patterns. A shows three gregarious sympatric sister species ((P. tessares, P. continua), P. sicpolus) from Area de Conservación Guanacaste (ACG) in Costa Rica. B shows two gregarious allopatric sister species, one from AVG (P. paradoxus) and one from Yanayacu Biological Station in Ecuador (P. sp. I). C shows two solitary sister species from Yanayacu Biological Station.

In general, WIPs can be observed and documented with very little additional effort for most taxonomists who work on small winged insects. We predict that they can be a large source of new morphological characters for the taxonomy and systematics of these tiny animals. The only materials required are a dissecting microscope with a camera attachment, a ring light, glass slides, and India Ink. Wing interference patterns are often species-specific and useful for Parapanteles wasps, and will likely be for most other microgastrine wasps.

This work was supported in part by the National Science Foundation grant DEB 1146119 and DEB 1442103. We thank the National Institute of Biodiversity – Ecuador (INABIO) for support and the Ministry of the Environment of Ecuador for providing permits under the genetic access contract MAE-DNB-CM-2016-0045 and the project “Interacciones entre plantas, orugas, y parasitoides de los Andes del Ecuador.” All specimens were collected, exported and to be DNA barcoded under Costa Rican government permits issued to BioAlfa (Janzen and Hallwachs 2019) (R-054-2022-OT-CONAGEBIO; R-019-2019-CONAGEBIO; National Published Decree #41767), JICA-SAPI #0328497 (2014) and DHJ and WH (ACG-PI-036-2013; R-SINAC-ACG-PI-061-2021; Resolución Nº001-2004 SINAC; PI-028-2021). We thank the GDFCF/ACG parataxonomist team of Area de Conservacion Guanacaste, northwestern Costa Rica for finding and rearing the caterpillar hosts of these parasites, which are documented individually by their DHJPAR..... voucher codes at http://janzen.sas.upenn.edu/caterpillars/database.lasso.