Comparative Morphology of the Wing Base Structure Illuminates Higher-Level Phylogeny of Holometabola

Simple Summary Insects that undergo complete metamorphosis, collectively known as Holometabola, are the most successful lineage of living organism, including about 830,000 species. Understanding the intricate relationships among the major groups of holometabolous insects is a critical task in systematic biology, particularly given their immense diversity. This research analyzed the wing base structure of Holometabola using comparative morphology to further clarify several crucial relationship nodes within Holometabola. Morphological data were selected from both the forewing base and hindwing base, comprising fifty-three discrete characters. Many phylogenetic relationship nodes among Holometabola were recovered exclusively using the wing base structure. Our research further highlights the efficacy of wing base morphology data in understanding insect phylogeny and evolution. Abstract Phylogenetic relationships among Holometabola have been the subject of controversy. The value of the wing base structure in phylogenetic analysis has been demonstrated but remains largely underexplored and scarce in studies of Holometabola. We studied the phylogenetic relationships among Holometabola (excluding Siphonaptera), focusing exclusively on wing base structure. Cladistic assessments were conducted using 53 morphological data points derived from the bases of both the forewing and hindwing. The results of wing base data revealed a sister relationship between Hymenoptera and remaining orders. The sister-group relationships between Strepsiptera and Coleoptera, Mecoptera and Diptera, Trichoptera and Lepidoptera, and Neuropterida and Coleopterida were corroborated. In Neuropterida, our results recovered the sister relationship between Megaloptera and Neuroptera, as well as the monophyly of Megaloptera.


Introduction
Holometabola is the most abundant group in the insect class, so far, with about 830,000 species having been described, comprising more than 80% of all described metazoan species and exhibiting substantial variation in ecology, behavior, and morphology [1,2].This unprecedented diversity provides rich research materials and intricate morphological data for evolutionary biologists and phylogeneticists.Although much research has focused on group phylogeny, the relationships of some important nodes remain unresolved.
Holometabola has long been considered monophyletic and has never been seriously questioned.However, the internal relationships of this group are still a subject of debate.Hennig was the pioneer in offering the first extensive reconstruction of the phylogenetic relationships within Holometabola [3].The basal phylogeny in his proposal is represented as (Coleoptera + Neuropterida) + (Hymenoptera + Mecopterida).Hymenoptera were placed as the sister group of Mecopterida, a relationship that was confirmed by whole mitochondrial genomes and early morphological data [4][5][6].This result, Insects 2024, 15, 199 2 of 23 however, was overturned by numerous studies based on whole nuclear or EST datasets, which overwhelmingly supported Hymenoptera as the sister group of the remaining Holometabola [7][8][9][10][11].The remaining Holometabola are typically divided into two major lineages, as follows: Mecopterida (Amphiesmenoptera + Antliophora (including Strepsiptera?) and Neuropteroidea (Coleoptera (+Strepsiptera?)+ Neuropterida)) [6,[12][13][14][15].In recent years, phylogenetic studies based on nuclear protein coding genes and morphological evidence have consistently confirmed the basal position of Hymenoptera, thereby establishing this concept in a dominant position [16][17][18][19][20][21].However, studies based on nuclear rRNA data usually obtain a wide variety of positions for Hymenoptera, for example, positioning them as sister groups with Neuropterida, which contributes to the persisting uncertainty regarding the placement of Hymenoptera [22][23][24][25].
The position of Strepsiptera in the phylogeny of Holometabola is another highly disputed issue due to their remarkably derived lifestyles and morphological characters.Their peculiar phenotype, characterized by conspicuous sexual dimorphism, shows that females are strongly morphologically simplified, with a complete lack of legs, wings, and external genitalia.This group has shown fascinating endoparasite behavior, with the diminutive primary larvae parasitizing other insects.In the past few decades, extensive studies have placed Strepsiptera in different positions [26].Traditionally, morphological characters uniting Strepsiptera and Coleoptera based on a broad array of morphological features include different body regions and unusual life stages [14,16,17].In contrast, a study integrating molecular and morphological data proposed a clade comprising Strepsiptera and Diptera (Halteria).This was based on ribosomal DNA sequence parsimony analyses and the proposed homology between the hind halteres of flies and the fore halteres of Strepsiptera [13].Alternative placements of Strepsiptera have also been proposed; these include positioning Strepsiptera either as a sister group of Neuropterida or as derived from within the order Coleoptera [19].Recently, the theory proposing a sister-group relationship between Strepsiptera and Coleoptera has gained more credibility from phylogenetic studies using nucleotide sequences obtained from comprehensive genome sequencing projects [27].However, a new study failed to confirm the sister-group relationship between Coleoptera and Strepsiptera [28].In conclusion, the phylogenetic position of Strepsiptera within Holometabola remains ambiguous due to such inconsistent research results.
Mecopterida is the largest lineage within Holometabola, comprising two clades, namely Amphiesmenoptera (Lepidoptera + Trichoptera) and Antliophora (Diptera + Siphonaptera + Mecoptera).The relationships within Antliophora and the monophyly of Mecopterida have been controversial [29].The monophyly of Mecopterida was found by Wiegmann et al. [21], while Kukalová-Peck et al. and Beutel et al. failed to find this concept in their analyses [11,17].Based on molecular evidence, Whiting suggests that Mecoptera and Siphonaptera are paraphyletic, while other orders are monophyletic [30].Wiegmann et al. refuted this hypothesis by analyzing several single-copy protein-coding nuclear genes and recovered the monophyly of Mecoptera [21].Therefore, the monophyly and the relationships within Mecoptera have been a subject of debate.Recently, Meusemann et al. attempted to clarify the phylogeny of Antliophora and verify the monophyly of Mecoptera by analyzing extensive transcriptomic nucleotide sequence data [29].However, their data did not lead to a definite conclusion.Cai et al. attempted to resolve the phylogenetic controversy within the diverse and medically significant group of fleas [31].They used the most extensive molecular dataset to date, comprising over 1400 protein-coding genes.Fleas were consistently identified as nested within scorpionflies (Mecoptera), forming a sister relationship with Nannochoristidae.Zhang et al. reconstructed the phylogeny of fleas among Endopterygota using mitochondrial phylogenomics of two species orders [32].They provided positive support for the hypothesis that Siphonaptera are monophyletic and demonstrated a sister-group relationship between Siphonaptera with orders Diptera + Mecoptera + Megaloptera + Neuroptera.In summary, conflicting phylogenetic relationships among Mecopterida need further clarification.
Insects 2024, 15, 199 3 of 23 Within Neuropteroidea (Neuropterida, Coleoptera, and Strepsiptera), major phylogenetic confusion is focused on the monophyly of this group and the relationships among Neuropterida.Neuropterida have been confirmed as monophyletic by some morphological data [21,33] and several recent molecular studies [1,34,35].This hypothesis, in contrast, was disapproved by Kukalová-Peck [11] and Beutel et al. [17].Longstanding, competing hypotheses have been proposed with respect to Neuropterida (Neuroptera, Raphidioptera, and Megaloptera), archaic holometabolous insects, and the inter-relationships of their three orders, possibly due to enormous disparities in morphology and lifestyle.Numerous studies indicate that Megaloptera are monophyletic [1,15,[35][36][37].However, Beutel et al. [17] proposed a clade comprising Raphidioptera + Corydalidae based on analysis of the morphological character set and molecular data, a finding consistent with the work of Wiegmann et al. [21] and Winterton et al. [38].With respect to the inter-relationships within Neuropterida, which are traditionally based on morphological data, a sister-group relationship between Megaloptera and Raphidioptera has been hypothesized [12,17].However, another cladistic analysis of morphological characters presented a sister-group relationship between Megaloptera and Neuroptera, as proposed by Aspöck [39].Recently, Song et al. [1] used mitogenomic data to identify a clade that includes Neuroptera and a sister group comprising Megaloptera and Raphidioptera.This finding supports previous hypotheses about the close relationship between Megaloptera and Raphidioptera.To reconcile these phylogenetic contradictions, a substantial amount of molecular evidence [35,[40][41][42][43][44][45][46] and morphological data [2,[47][48][49] have been presented, leading to the recovery of the monophyly of Megaloptera and the sister-group relationship between Megaloptera and Neuroptera.
Since the prosperity of molecular systematics, insect morphology has suffered a decline after flourishing in the first two-thirds of the 20th century [50].Innovative molecular techniques offer the outstanding capability to address many phylogenetic questions that are difficult to solve with morphological methods and provide the ability to generate large-scale genomic datasets in a comparatively short time.However, in the latest study of Neuropterida, the authors analyzed a vast amount of molecular data but found that certain nodes in the Neuroptera tree were not robustly resolved.Consequently, they advocate for integrating morphological analyses with sequence-based phylogenomic data [46].Arguably, such controversial hypotheses provided a healthy adjustment and forced morphologists to re-evaluate time-honored hypotheses and generate new morphological data.It is conceivable that the production of abundant morphological records and comprehensive combined datasets is necessary to further advance insect phylogenetics [13,46,51,52].
The evolution of foldable wings marks a seminal milestone in insect diversity, endowing them with superior mobility for migration, feeding, and so on.The intricate systems of elegant insect wings include membranes, veins, folding and flexion lines, and marginal setae.All these elements work together in a way that we understand only partially [53].A cluster of inter-related sclerites at the wing's base, where it connects to the thorax, is essential for various wing actions, including flapping, rotation, and folding.Given their critical functional role, these sclerites exhibit minimal evolutionary change, reflecting their significant mechanical constraints [47,54].In addition, the intricate shapes and articulations of the wing base sclerites make it possible to provide more important morphological characters for phylogenetic estimations [6,47,[54][55][56][57][58].Furthermore, the phylogenetic trees based on wing base morphology align with findings from molecular phylogenetic studies and comprehensive insect phylogenomic research [47].Currently, research on wing base sclerites is predominantly focused on hemimetabolous insects.For instance, the morphology of wing base sclerites of Polyneoptera and Paraneoptera has been extensively explored, which also applies to the studies of phylogenetics and evolution [59].In contrast, research into the wing bases of holometabolous insects remains relatively limited.Here, we provide a morphological data matrix based on wing base sclerites focused specifically on the phylogeny of Holometabola.The main objective of this study is to develop a detailed and thoroughly documented set of morphological characters, focusing on the structures of the forewing and hindwing bases.The character states were analyzed using heuristic parsimony analysis.Our investigations also provide new phylogenetically relevant data for understanding of the higher phylogeny of Holometabola.

Examined Taxa
Ingroup taxa included the species representing three families/subfamilies of Megaloptera, three families of Neuroptera, two families of Raphidioptera, five families of Coleoptera, three families of Hymenoptera, one family of Trichoptera, two families of Lepidoptera, two families of Mecoptera, four families of Diptera, and one family of Strepsiptera (Table S1).For this experiment, research materials were selected from different orders, including Xyelidae, Tenthredinidae, and Diprionidae in Hymenoptera; Nevrorthidae, Osmylidae, and Chrysopidae in Neuroptera; Corydalidae and Sialidae in Megaloptera; Raphidiidae and Inocelliidae in Raphidioptera; Cupedidae, Carabidae, Cicindelidae, Cerambycidae, and Melolonthidae in Coleoptera; Tipulidae, Pyrgotidae, Syrphidae, Tabanidae, and Phryganeidae in Diptera; Phryganeidae in Trichoptera; Corioxenidae in Strepsiptera; Sphingidae and Nymphalidae in Lepidoptera; and Bittacidae and Panorpidae in Mecoptera.The outgroup selection included Gripopterygidae of Plecoptera and Tettigoniidae of Orthoptera, both of which belong to polyneopterous, a group that is a sister clade to the paraneopteran + Holometabola [12].The wing base structure was observed using a ZEISS Stemi 2000-c stereoscope (Carl Zeiss, Jena, Germany).All of the wings were stretched artificially upwards during observation to account for their highly three-dimensional structures.Additionally, all figures were taken from the dorsal view of the wing base.Upon collection, all examined specimens were first stored in 80% ethanol, then moved to long-term storage at −20 • C at the Entomological Museum of China Agricultural University (CAU) in Beijing.

Phylogenetic Analysis
Fifty-three characters from the fore-and hindwing bases were systematically coded for analysis, as detailed in the character description in the Results section.This coding only included quantitative aspects when variations were clearly distinguishable and not part of a continuous range.Although in most groups, the fore-and hindwing base structures usually exhibit analogous modifications [55, 56,59,61], data are generally selected from the forewing or hindwing alone to prevent double counting of any particular character.However, in this study of Holometabola, several features of the hindwing base were markedly different from those of the forewing base.Hence, it was justified to select data from both the foreand hindwings.4Ax and PNWP are regarded as homologous sclerites, as discussed by Zhao et al. [47].4Ax is homologized with the pPNWP [57].
In the phylogenetic analysis, each family was considered as a terminal taxon within both ingroup and outgroup categories.We analyzed 26 forewing and 25 hindwing base characters using TNT ver.1.1 [62] and NONA ver.2.0 [63] and applied heuristic parsimony methods with 100 replications.Bootstrap analysis with 10,000 iterations was conducted, collapsing branches with values ≤ 50%.Bremer's decay indices were calculated using TNT ver.1.1 [62] and WinClada ver.1.00.08 [64].Detailed information about data matrix is shown in Table S2.In order to test, verify, and calculate the CI and RI of each character, the dataset was also analyzed in PAUP*4.0b10[65].This included a heuristic parsimony analysis with 100 random additions of taxa and TBR branch swapping, applying ACC-Insects 2024, 15, 199 5 of 23 TRAN optimization, treating characters as unordered and equally weighted, and using the MulTrees option.

Comparative Morphology of the Wing Base Structures
The general morphology of the wing base structure was previously introduced and summarized by Zhao et al. [47].Remarkably, several elements are particularly important in terms of function and phylogeny, notably the first and third axillaries [6].Obviously, the second axillary plays a crucial supporting role in wing movement, but the actual transmission of the flight muscles' actions from the notum to the wing primarily relies on the first axillary.The first axillary can be divided into the following three parts: the head, the neck, and the basal part (the body).According to Yoshizawa et al. [55], there is no obvious distinction between the head and the neck in Holometabola; however, in our study, we observed recognizable differences.The third axillary plays a critical role in Neoptera and in Holometabola, as it is equipped with a muscle that is considered an autapomorphy of Neoptera [6].To facilitate the understanding of character descriptions, numerous characters are illustrated in schematic diagrams of the axillary sclerites, excluding the relationships among the axillaries.Each sclerite of the wing base is defined by proximal, distal, anterior, and posterior aspects (Figure 1) [66].
Insects 2024, 15, x FOR PEER REVIEW 6 and articulates with 1Ax in the anterodistal part of the forewing (char.39).The DM tegrates with BR into a sclerite in the forewing (char.20).In Trichoptera, there is a p tion in the anterodistal MNWP of the forewing, which might indicate a unique evolu ary characteristic of this order (char.4).3.1.1.Hymenoptera (Figure 2) The wing base structure of Hymenoptera consists of fundamental elements.Articulations and fold and flexion lines also preserve the plesiomorphic condition.However, compared with other orders of Holometabola, the wing base structure of Hymenoptera is different.The first difference refers to the ANWP of the mesothorax.In Hymenoptera, the ANWP is consistently tubular and positioned anteriorly to the Tg in the forewing, whereas in other orders, the ANWP is triangular or stripe-like and located behind the Tg (char.1, 2).In addition, 1Ax in Hymenoptera is distinguished from other groups by strong swelling (char.9).In the anterodistal part of the body of 1Ax in Hymenoptera, there is a small projection that differs from projections of the neck of 1Ax based on the boundary between the neck and body of 1Ax (char.14).Additionally, the 3Ax of Hymenoptera is obviously different from others; it is stripe-like with two lobes rather than plate-like with three lobes (char.17,18).The DMP appears more swollen and distinctly harder than the PMP and is smaller than the 1Ax in Hymenoptera (char.22,23).In other orders, 1Ax is usually smaller than the median plates.The shape of BA is similar to 3Ax, with a stripe-like character (char.25).In the hindwing base, the neck of 1Ax is absent in Hymenoptera (char.35).For 2Ax, its shape is almost rectangular and does not bend (char.41).Meanwhile, in other groups, it is stripe-like and bends.In addition, the BSc and BR are completely fused in   3.1.2.Amphiesmenoptera (Figure 3) Amphiesmenoptera contains Lepidoptera and Trichoptera.Compared to other Holometabola, the body of 1Ax in Amphiesmenoptera is more closely approximated to a rectangle, with the basal and distal lobes being of similar length (char.12).In Lepidoptera, the Tg exhibits a triangular shape in the forewing and is absent in hindwing (char.7, 31).The 1Ax tapers from the neck to the head in the forewing (char.11).The 2Ax is stripe-like and articulates with 1Ax in the anterodistal part of the forewing (char.39).The DMP integrates with BR into a sclerite in the forewing (char.20).In Trichoptera, there is a projection in the anterodistal MNWP of the forewing, which might indicate a unique evolutionary characteristic of this order (char.4).3.1.3.Antliophora (Figure 4) In Antliophora, the variation from the neck to the head of the 1Ax in the forewing is discontinuous, characterized by a narrowing at the neck-head boundary and a broadening at the apex of the head (char.11).Another distinctive characteristic in Antliophora is that the BA is almost twice as large as the 3Ax in the forewing (char.24).In Mecoptera, the structure at the wing base is composed of basic, essential elements.It was noted that the body of the first axillary presents a short and broad structure that is more rectangular than triangular in shape, typically featuring a concave caudal edge in both wings.This 3.1.3.Antliophora (Figure 4) In Antliophora, the variation from the neck to the head of the 1Ax in the forewing is discontinuous, characterized by a narrowing at the neck-head boundary and a broadening at the apex of the head (char.11).Another distinctive characteristic in Antliophora is that the BA is almost twice as large as the 3Ax in the forewing (char.24).In Mecoptera, the structure at the wing base is composed of basic, essential elements.It was noted that the body of the first axillary presents a short and broad structure that is more rectangular than triangular in shape, typically featuring a concave caudal edge in both wings.This observation aligns with previous research, as discussed in the review by Hörnschemeyer [6].We also noted that the Tg in Mecoptera is slimmer and exhibits a more rectangular shape compared to other species, where it tends to be more triangular (char.7).Additionally, the shape of pPNWP is notably distinctive, exhibiting a U shape, while in Hymenoptera, it appears almost stripe-like (char.28).In Diptera, the shape of the body of 1Ax similar to a lady's high heels in the forewing (char.12).In the hindwing, the width of neck is approximately the same as the head of 1Ax (char.34).
Insects 2024, 15, x FOR PEER REVIEW 9 of 23 similar to a lady's high heels in the forewing (char.12).In the hindwing, the width of neck is approximately the same as the head of 1Ax (char.34).3.1.4.Coleopterida (Figure 5) Coleopterida comprises Coleoptera and Strepsiptera.A notable characteristic in Coleopterida is that the DMP partially fuses with the PMP in the hindwing.In Coleoptera, the wing base comprises essential, foundational elements, but lacks the Tg, which is mainly an autapomorphy for this order (char.6).A unique and fascinating feature in the

Coleopterida comprises Coleoptera and Strepsiptera. A notable characteristic in
Coleopterida is that the DMP partially fuses with the PMP in the hindwing.In Coleoptera, the wing base comprises essential, foundational elements, but lacks the Tg, which is mainly an autapomorphy for this order (char.6).A unique and fascinating feature in the wing structure is the presence of a small independent sclerite, located between the first and third axillary in the membrane, which we consider to belong to 3Ax.This sclerite may serve as a muscle attachment point in Coleopterida, differing from other Holometabola, where muscles are directly attached to the third axillary (char.47).The metanotum is longer than the 1Ax, being about 1.4 to 2.4 times the length of the 1Ax (char.33).In our study of Strepsiptera, a notable feature observed in our study is the absence or unrecognizability of the HP in the hindwing (char.32).3.1.5.Neuropterida (Figures 6-8) In Neuropterida, the wing base structure includes essential elements, with its articulations, as well as fold and flexion lines, reflecting ancestral features.In Neuropterida, the pPNWP of the metathorax is sclerotized rather than membranous or absent in other clades of Holometabola (char.29).The shape of the 2Ax is almost stripe-like, but it bends distally at the apex in this clade (char.41).In addition, in the hindwings of Neuropterida, the BSc and BR are partly fused, in contrast to Hymenoptera where they are completely fused in the hindwings (char.46).The 3Ax in Neuropterida consists of three lobes, the anterior, proximal, and distal lobes (char.17).A separate part is present in the PNWP of the hindwing.Further details about the wing base structure in Neuropterida are described by Zhao et al. [47], without additional discussion here.In Neuropterida, the wing base structure includes essential elements, with its articulations, as well as fold and flexion lines, reflecting ancestral features.In Neuropterida, the pPNWP of the metathorax is sclerotized rather than membranous or absent in other clades of Holometabola (char.29).The shape of the 2Ax is almost stripe-like, but it bends distally at the apex in this clade (char.41).In addition, in the hindwings of Neuropterida, the BSc and BR are partly fused, in contrast to Hymenoptera where they are completely fused in the hindwings (char.46).The 3Ax in Neuropterida consists of three lobes, the anterior, proximal, and distal lobes (char.17).A separate part is present in the PNWP of the hindwing.Further details about the wing base structure in Neuropterida are described by Zhao et al. [47], without additional discussion here.
Within the clade of Diptera and Mecoptera, there are two homologous synapomorphies that provide strong support for their sister-group relationship, as follows: (1) the transition in width from neck to head of 1Ax is characterized by a thinned boundary and a widened head apex in the forewing (char.11:3), and (2) BA is twice as large as 3Ax (char.24:1).Furthermore, the monophyly of Mecoptera received support from the following two additional homologous synapomorphies: (1) the shape of the Tg is rectangular (char.7:2), and (2) the pPNWP is U-shaped (char.28:3).
For the clade of Amphiesmenoptera (Lepidoptera, Trichoptera), there is one homologous synapomorphy that supports the monophyly.The body of the 1Ax is nearly rectangular, with a proximal lobe as long as the distal lobe (char.12:1).The monophyly of Lepidoptera also received support from the following four homologous synapomorphies: (1) a triangular shape of Tg (char.7:1), (2) a continuously apically thinned change in width from the neck to the head of 1Ax (char.11:2), (3) BR and DMP fused to a plate (char.20:1), and (4) 2Ax and 1Ax articulation formed by the proximo-cranial and proximo-caudal parts, both separated by a narrow membranous area (char.39:2).As for Trichoptera, the monophyly was supported by one homologous synapomorphy, namely the anterodistal projection of MNWP (char.4:1).

Discussion
Our phylogenetic analysis of forewing and hindwing base structural data supports the monophyly of Holometabola with a 98% bootstrap value and Bremer's decay indices equal to 3. Our investigation is consistent with multiple studies relying on morphology and significant genome data [6,14,17,28].Previous investigations on wing base morphology, largely conducted by Yoshizawa, were primarily centered on groups exhibiting incomplete metamorphosis [56,58,59,67].Regarding Holometabola, wing base structure data are available in Beutel et al. [17] and Hörnschemeyer [6].However, Beutel did not provide a detailed comparative morphological study of the selected wing base structures, while Hörnschemeyer used fewer representative families and species compared to our study.Beutel et al. [17] suggested three synapomorphies based on wing base structure to support the monophyly of Holometabola.In the forewing, the 1Ax articulates with the whole tail of 2Ax (char.253:0).The distal lobe of the 3Ax of the forewing is identifiable (char.257:0).The ANWP of the hindwing is almost triangular (char.257:0).
As shown in Figure 10, Hymenoptera was assigned as the sister group to the remaining orders, a result consistent with the viewpoint proposed by many scholars [7][8][9][10][11].The notion that Mecopterida (Antliophora + Amphiesmenoptera) forms a sister group with Neuropteroidea (Coleopterida + Neuropterida) aligns with the findings of Peters et al. [28], who recently analyzed the transcriptome and morphological data of complete metamorphosis insects.Our analyses are incongruent with the prevalent view that Hymenoptera is a sister group to Mecopterida and seriously challenge the argument about the sister relationship between Hymenoptera and Neuropterida, which is based on 18S rDNA sequence analysis [14,25,30].There were significant differences observed in wing base structure between Hymenoptera and other orders.The apomorphy features supporting the monophyly of Hymenoptera are mostly unique.The most compelling evidence includes the absence of a neck in the 1Ax on the hindwing, the pronounced swelling of the 1Ax, and the distinct shape of the 3Ax, all of which distinguish Hymenoptera from other orders.Additional specific characters are detailed in the Results section.Among Hymenoptera, Xyelidae occupy a basal phylogenetic position, while Tenthredinidae form a sister group with Diprionidae.
There are two homologous synapomorphies supporting the sister relationship of Mecopterida and Neuropteroidea in our study; however, relatively low bootstrap values and Bremer's decay indices indicate that stronger evidence should be found.This sister-group relationship was corroborated by Beutel et al. [17] based on the following homologous synapomorphies: (1) a slender PNWP of the mesothorax (char.244:0), (2) an angle between the distal margin of the metathorax and the 1Ax of the metathorax greater than 50 • (char.275:0).The characteristic of a slender PNWP in the mesothorax was also observed in our experiment.Additionally, the monophyly of Mecopterida has been confirmed by four homologous synapomorphies.Furthermore, Amphiesmenoptera (Lepidoptera + Trichoptera) and Antliophora (Mecoptera + Diptera) are considered a sister group, a view that aligns with traditional perspectives and was corroborated by Peters et al. [28] based on molecular data, although incongruent with the findings of Beutel et al. [17], who suggested Antliophora as a sister group with Coleopterida.The monophyly of Amphiesmenoptera is supported by high bootstrap values and Bremer's decay indices in our results.The sister-group relationship between Mecoptera and Diptera is consistent with the findings of Beutel et al. [17] and Peters et al. [28], supported by two synapomorphies and Bremer's decay indices of 2 in our study.
Coleopterida (Coleoptera + Strepsiptera) was placed as a sister group to Neuropterida (Megaloptera + Neuroptera + Raphidioptera) in our experiment, which is congruent with the traditional perspective based on respective morphological and molecular datasets [1,14,15,26,30,36,68,69].However, the sister-group relationship between Coleopterida and Neuropterida is inconsistent with the finding of Beutel et al. [17], who proposed that Coleopterida and Antliophora have a sister relationship.Our phylogenetic analysis of fore-and hindwing base structural data supports the monophyly of Coleoptera with high credibility, containing the following two synapomorphies: the absence of a Tg (char.6:2) and a 1Ax relatively longer than the metathorax notum (char.33:3).However, many studies based on molecular data suggest that Coleoptera is paraphyletic [30,70,71].Additionally, there is only one synapomorphy supporting the sister-group relationship between Strepsiptera and Coleoptera.The phylogenetic status of Strepsiptera has long been a disputed and unresolved issue.Recently, most molecular phylogenetic studies have advocated for Strepsiptera and Coleoptera as sister taxa [26,28,36,68], but morphological evidence is lacking.The sister relationship between Strepsiptera and Coleoptera has been advocated by Friedrich et al. [16] and Beutel et al. [17], but their morphological support was considered weak or moderate in a subsequent study [69].
Our phylogenetic analysis within Neuropterida confirms its monophyly and positions Raphidioptera as the sister group to (Megaloptera + Neuroptera).This aligns with findings from most prior molecular studies [22,28,[34][35][36][41][42][43].Furthermore, the monophyly of Megaloptera and its sister-group relationship with Neuroptera were corroborated by our wing base data, which are consistent with many results based on morphology, mitochondrial genomics, and transcriptome data [28,35,37,43,47,[70][71][72][73][74][75], as well as a recent phylogenetic analysis utilizing an integrative phylogenomic approach [46].However, this challenges the view that Megaloptera was never recovered as monophyletic and with Raphidioptera in a clade sister to Neuroptera, as proposed by Winterton et al. [38].They conducted a comprehensive phylogenetic study of Neuropterida, utilizing morphology and multilocus DNA sequence data across all extant families of Neuroptera, Megaloptera, and Raphidioptera.Wang et al. [45] primarily investigated the impact of highly informative selected genes or more realistic phylogenetic models on the reconstruction of Neuropterida phylogeny.Their research consistently affirmed the monophyly of Raphidioptera, Megaloptera, and Neuropterida.However, the monophyly of Neuroptera was only obtained when analyzing genes with strong signals or using robust models, which partially aligns with our findings.Furthermore, our study confirmed the higher-level classification of Megaloptera, dividing it into Corydalidae and Sialidae.Within Corydalidae, both Corydalinae and Chauliodinae were included.Furthermore, in Neuropterida, we consider that the pPNWP is a part of the distal part of the PNWP, with an irregular shape.

Conclusions
The structure of the wing base is a valuable tool for reconstructing the phylogeny of Holometabola.Many published works on phylogenetics corroborate the value of wing base structure for resolving higher-level phylogenetic problems [58,59].In our study, several phylogenetic relationships among Holometabola were successfully resolved.The basal location of Hymenoptera, as well as the sister-group relationships of Strepsiptera and Coleoptera, Mecoptera and Diptera, Trichoptera and Lepidoptera, and Neuropterida and Coleopterida were confirmed.However, the relationship between Mecoptera and Siphonaptera remains unresolved due to the deficiencies of the wing base for Siphonaptera.Additionally, our study provided clear insights into the relationships among the three orders of Neuropterida.However, our present study faced a limitation due to the inadequate number of informative characters for resolving phylogenies at the family level.Future comprehensive studies may benefit from the geometric morphometrics of these wing base sclerites, potentially enhancing the resolution of family-level phylogeny.

Figure 1 .
Figure 1.Overall model of the axillary sclerites for Holometabola: 1Ax, 2Ax, 3Ax.This model was modified from Figure 3 of Franielczyk-Pietyra et al. [66].Numbers represent morphological characters.Detailed character descriptions are provided in the character description section.
. 46).In other orders, they are partly fused at most.In Hymenoptera, the BA is strong-larger than 3Ax in the hindwing (char.53).Insects 2024, 15, x FOR PEER REVIEW 7 of 23

Figure 2 .
Figure 2. Wing base of Hymenoptera.The 1Ax is rendered dark blue, the 2Ax in medium blue, and the 3Ax in light blue.Both the PMP and DMP are illustrated in orange.(A) Xyela sp.(Xyelidae), forewing base; (B) same, hindwing base; (C) Tenthredo sp.(Tenthredinidae), forewing base; (D) same, hindwing base; (E) Neodiprion huizeensis Xiao & Zhou, 1984 (Diprionidae), forewing base; (F) same, hindwing base.Number of morphological characters: character state for phylogenetic analysis is indicated by a straight line for relevant position.Detailed character descriptions are provided in the character description section.

Figure 2 .
Figure 2. Wing base of Hymenoptera.The 1Ax is rendered dark blue, the 2Ax in medium blue, and the 3Ax in light blue.Both the PMP and DMP are illustrated in orange.(A) Xyela sp.(Xyelidae), forewing base; (B) same, hindwing base; (C) Tenthredo sp.(Tenthredinidae), forewing base; (D) same, hindwing base; (E) Neodiprion huizeensis Xiao & Zhou, 1984 (Diprionidae), forewing base; (F) same, hindwing base.Number of morphological characters: character state for phylogenetic analysis is indicated by a straight line for relevant position.Detailed character descriptions are provided in the character description section.

Insects 2024 ,
15,  x FOR PEER REVIEW 10 of 23 study of Strepsiptera, a notable feature observed in our study is the absence or unrecognizability of the HP in the hindwing (char.32).

Figure 10 .
Figure 10.Phylogenetic tree of Holometabola based on wing base data.A strict consensus from parsimonious trees is displayed, focusing on forewing and hindwing bases.Features of clear mapping of unambiguous characters: filled circles for homologous traits, open circles for reversals or parallels.Character states are below the circles.Numbers on nodes indicate the bootstrap values and Bremer's decay indices.The sclerites on the right side represent the 1Ax, 2Ax, and 3Ax of the

Figure 10 .
Figure 10.Phylogenetic tree of Holometabola based on wing base data.A strict consensus from parsimonious trees is displayed, focusing on forewing and hindwing bases.Features of clear mapping of unambiguous characters: filled circles for homologous traits, open circles for reversals or parallels.Character states are below the circles.Numbers on nodes indicate the bootstrap values and Bremer's decay indices.The sclerites on the right side represent the 1Ax, 2Ax, and 3Ax of the respective orders.Sclerites numbered 1-12 and 16-30 are from the hindwing base, while sclerites numbered 13-15 are from the forewing base.