Analyses of all data recovered the monophyly of the three clades relevant to this analysis, each consisting of two to five closely related Eniclases species (Figs. 4, 5). We identified the recent origin of most Eniclases species and mtDNA introgression, indicated by individuals with different nuclear but identical or highly similar mtDNA sequences (Figs. 4, 5). Our species delimitations are based on RAD analyses, as this extensive, genome-wide data can better resolve clusters of individuals with shared ancestry, particularly among closely related species and recently separated populations . Results from sNMF and PCA analyses identified some genomes even with the different genetic structure within the RAD phylogeny-based species-rank entities (Additional file 1: Figures S10–S14) and collectively, the analyses suggested ongoing gene flow and that most species represent recently split lineages, typically within the last million years (Fig. 4b).
Mimetic patterns in Eniclases
The foundational premise of Müllerian mimicry holds that unprofitable prey benefit from convergence in warning signals . The phylogeny of Eniclases showed that resemblance among various Eniclases is the product of parallel evolution rather than common ancestry (Fig. 4a). Our data support the Müllerian mimicry hypothesis by the noted sympatric occurrence of distantly related but similarly colored Metriorrhynchina with Eniclases (Figs. 1g, h–l, 2). Additionally, the detailed delimitation of species confirms the presence of intraspecific colour polymorphism in these Müllerian mimics with intraspecific forms resembling unrelated Eniclases and Microtrichalus species.
Aggregations of net-winged beetles in the field indicate that multiple species truly coexist in a single community and do not have a microhabitat-based mosaic distribution. Up to six Eniclases aposematic patterns were recorded within single localities in our small study area (Table 1; Fig. 3b) and additional patterns were displayed by closely related Metriorrhynchina (altogether up to 10 patterns in a single locality; Table 1). Unlike findings from other Müllerian mimetic systems , these observations suggest that a dominance effect on mimetic polymorphism stemming from microhabitat preference is improbable in Eniclases.
Coloration of Eniclases is limited to pale to bright yellow, yellowish-orange, and black, but these beetles display multiple distinct color forms with a putative signaling function (Figs. 4a, Additional file 1: Figures S1–S4). Dominant color patterns combine a bright-colored pronotum and elytral humeri with apically dark elytra. Higher contrast is represented by a steeper transition between bright and black parts of the elytra or pronotum. Three species in the study area are black, two of which are polymorphic with some individuals having a yellow pronotum (Fig. 4a). These patterns dominate in all communities containing high representation of trichaline net-winged beetles, generally those < 1500 m in elevation. Additional patterns among high-elevation species are uncommon and include pale yellow or creamy white in combination with a black pronotum and elytral apex (Fig. 4a). Eniclases are never red, green, or metallic blue (Figs. 1l-o, 2j–l) and never display three colors, unlike numerous sympatric Metriorrhynchina. It is possible that the absence of red pigment in Eniclases is genetically constrained. Earlier studies have shown that easily remembered, high-contrast patterns, e.g., a red-black combination, provide higher protection than lower contrast ones (e.g., [32, 33]. Relative to other net-winged beetles in the same communities, we consider color patterns in Eniclases as low contrast, and their effectiveness is likely further reduced by relatively high intraspecific variability, given that 5 of 12 species from Central New Guinea are polymorphic  (Fig. 4a, Additional file 1: Figures S1–S4).
Aposematic color patterns are controlled by selection [1, 2] and their distinctness depends on the ability of predators to discriminate among them and on the intensity of predation upon intermediate forms . In contrast to the observed high signal variability, we recorded a distinct, fine-tuned warning signal in E. divaricatus and its co-mimics (E. similis, Trichalus sp., and Cautiromimus sp.) represented by characteristic bright-colored humeral patches (Fig. 4a). The specificity of brightly colored humeral patches versus whole humeri color might support the potential role of predators  but the present study design cannot provide any experimental evidence. Nevertheless, the structure of Eniclases communities shows that under certain conditions, both polymorphism and high pattern similarity can persist within a multi-pattern Müllerian community.
Behavior and signal perception
Given that trichalines commonly occur in multi-species and multi-pattern aggregations (Table 1), predators encounter a spectrum of aposematic signals related to components such as body shape and colour [12, 36,37,38]. Further complexity is added to aposematic signaling by different perception of the signal in space and time.
A body shape and size signal are predominantly perceived when an individual sits on the underside of a translucent leaf (Fig. 1a–h). This perception depends on light intensity, microhabitat conditions, and season . We observed that Eniclases prefer the undersides of leaves, as do other trichaline genera. We assume that convergent evolution has led to the observed morphological uniformity in body shape and size of > 100 species of trichalines in New Guinea (Additional file 1: Figures S1–S4, Table S4).
Color-based component of signals include hue, uniform or bicolored upper bodies, color patch shape, and the level of contrast between body parts (Figs. 1, 2, 4a). Color is best detected under diffuse-light conditions under a dense forest canopy or when an individual is on upper leaf surfaces, but Eniclases and its co-mimics usually avoid upper leaf surfaces, unlike brightly colored lycids such as Porrostoma, Metriorrhynchus, and Cladophorus (, field observation). When color pattern is not readily observable, we can assume that the shape component of the signal is relatively stronger. Therefore, the further research is needed to clarify if selection for a shared color pattern might be relaxed. Penny et al.  showed the importance of relaxed selection in the origin of imperfect mimicry; we suggest that it may also increase the persistence of a greater number of color patterns in a community (Fig. 4a, Table 1).
Overall, we documented apparent convergence and/or advergence to shared color patterns in distantly related species within Eniclases and in unrelated Microtrichalus (Figs.2, 4a; compare phenotypically similar and phylogenetically unrelated individuals as described above). Such similarity is usually caused by selection [1, 2, 4,5,6], which was not studied in this project by experiments. Therefore, excluding similarity due to the common origin, we can speculate that observed similarity of these unpalatable and aposematically coloured beetles is a result of natural selection. In contrast with the Müllerian model which predicts strong selection against rare phenotypes and poor mimics [1,2,3,4], we identified multiple examples of intraspecific polymorphism and a high number of different aposematic patterns in a single community. Further detailed research should be focused on the role of factors that can relax or slow the effects of purifying selection, e.g., the generalization of multiple patterns by predators, evolutionary constraints such as the ability, or inability, to rapidly adopt the dominant local aposematic pattern [3, 7, 16, 39] and here suggested behavioral adaptation and variable perception of the body shape and colour (compare Fig. 1a, c, e, g with b, d, f, h showing perception of insect individuals against clear sky and the same individuals illuminated by an artificial light source).
Eniclases communities in space and time
The model of Müllerian mimicry is simple and earlier experiments have not considered fluctuating conditions during the process of convergence or advergence. Based on our observations, we must assume that the communities in which Eniclases occur dynamically change in space due to local conditions and across the altitudinal gradient (Table 1). As a result migrating Eniclases commonly enter different mimetic complexes. While we found that similar communities can be expected at similar altitudes in New Guinea, we observed a distinct community composition in the Dombomi locality, which does not contain brightly colored trichalines despite being located at a low elevation. Dombomi is situated on the windward slope of a high mountain range and differs from Elelim in having high levels of precipitation and fog, which supports dense, marshy, medium-height forests with a different fauna. It means that communities with different species and pattern composition might not necessarily be geographically distant.
Additionally, we identified substantial altitude dependent differences in species composition. Despite a distance of less than 2 km, Bokondini communities sampled from 1250 and 1850 m in elevation shared no common species (richness of 26 and 58 species, respectively). Although net-winged beetles are poor dispersers, even low levels of continuous migration could increase the number of aposematic patterns in a community, delay pattern convergence, and increase the number of color-polymorphic species simultaneously adverging to different models . Further research should consider that the communities we analyze today may be products of very different histories, particularly concerning migration.