High niche diversity in Mesozoic pollinating lacewings

Niche diversity of pollinating insects plays a vital role in maintaining extant terrestrial ecosystems. A key dimension of pollination niches refers to the insect proboscis length that commonly matches the floral tube length. Here we describe new kalligrammatid lacewings (an iconic Mesozoic pollinating insect lineage) from late Cretaceous Burmese amber and Mesozoic sediments in China. Kalligrammatids display complex configurations of elongate mouthpart elements consisting of well-developed maxillae, labium and their palps. The mouthpart lengths vary among species, from 0.6 to 18.0 mm, suggesting corresponding variability in the floral tube lengths of Mesozoic plants. With the diversification of pollinating habits, the kalligrammatids presented highly divergent traits related to chemical communication and defence mechanisms. Together with other Mesozoic long-proboscid insects, these fossils not only reveal the high niche diversity of Mesozoic pollinating insects but also highlight the diversity of Mesozoic pollinator-dependent plants prior to the rise of angiosperms.

probably artificial. Moreover, the ligula in Fiaponeura is unpaired with bifid apex (see Lu et al. 2 : fig. 4), which conforms to the morphological ground plan of lacewing ligula.
The presence of paired tibial spurs in all species of Cretanallachiinae is herein confirmed. Previous reports on the absence of tibial spurs in Cretanallachius (see Lu et al. 2 ) and Burmopsychops (see Makarkin 1 ) are possibly due to preservation matters.
Association between conspecific males and females was primarily based on the similarity of body size, wing venation, and wing marking pattern. The combination of characters of body-size, wing venations and wing marking patterns is in general stable among conspecific males and females but differs among different species. For example, B. liui and F. penghiani have very characterized wing markings and venations, which can facilitate the association of males and females. Similarly, C. magnificus is the smallest species of Cretanallachiinae known so far and has no wing marking. Therefore, the newly found female of this species was linked to the male by its very small body-size (smaller than females of any other cretanallachiine species) and the immaculate wings.
The male and female genitalia of Cretanallachiinae also show sufficient morphological differences as a distinct family-level group from that of Dilaridae.
Considering the male genitalia of Cretanallachiinae, Makarkin 1 disagreed with the interpretation of the paired large valvate lobes to be the gonocoxites 9 2 , but alternatively considered it to be a part of enlarged tergum 9 as an argument supporting close relationship between this subfamily and Dilarinae. Although in most extant families of Neuroptera (except Nevrorthidae, Sisyridae and some species of Osmylidae) the male gonocoxites are present as small internal sclerites, the condition of this genital sclerite in fossil lacewings, particularly in extinct families/subfamilies is poorly known, and the possibility of the presence of large external male gonocoxites 9 in some extinct lacewing families cannot be excluded. Based on our examination of males of several cretanallachiine species, there is no doubt that these large paired lobes are gonocoxites 9, which are present at the typical position of male gonocoxites 9 (i.e., a position posteroventrad tergum 9) of Neuroptera, while the true male tergum 9 in Cretanallachiinae is a short and arched sclerite posterodorsally connected with paired ectoprocts, which is also a typical feature in Neuroptera. The female genitalia of Cretanallachiinae are highly specialized considering the tergum 9 with a pair of valvate posteroventral lobes that largely envelope the gonocoxites 9.
3 Furthermore, the female ectoprocts are paired and largely associated with the dorsal part of tergum 9, and there is a small hole (putatively homologized with the copulatory fovea in Psychopsidae; see Oswald 5 and Bakkes et al. 6 ) anteriad gonocoxites 8 (discernible in a female of Oligopsychopsis groehni (Makarkin)).
Notably, these remarkable characters, although with some modifications, are present in extant Psychopsidae (see Oswald 5 : figs. 43, 49). However, Makarkin 1 just gave a groundless conclusion that the ovipositor (= gonocoxites+gonostyli 9) of Psychopsidae and Cretanallachiinae were independently evolved and considered an Osmylid-like feature of the ovipositor in Cretanallachiinae. Based on our examination, the female gonostyli 9 are just present on subdistal portion of ventral surface of gonocoxites 9, which was mentioned as the feature of Psychopsidae by Makarkin 1 .
Nevertheless, the most important arguments supporting the closer relationship between Cretanallachiinae and Psychopsidae rather than Dilaridae refer to the aforementioned similar characters of female tergum 9, ectoprocts and gonocoxites 8. Remarks. This genus appears to be closely related to Cretanallachius in having similar immaculate wings, barely branched MA vein, and absence of nygmata, but it can be separated from the latter based on the forewing MA separating from R and the absence of sigmoid stem of hind wing MA. Non-genital characters of female almost identical to male, except for the moniliform antennae.
Female genital segments: Tergum 9 distinctly separated into a pair of broad posterolateral lobes, with dorsal part reduced and membranous; posterolateral lobes strongly protruding posteriad and slightly curved dorsad. Gonocoxites 9 slightly shorter than and largely enveloped by lateral valves of tergum 9, distally tapering; gonostyli 9 invisible due to preservation condition. Ectoprocts paired, subtriangular, with rounded apex; callus cerci absent. Remarks. This genus was originally erected based on a single male specimen from 5 the Burmese amber7. The cretanallachiine affinity of this genus is undoubted in light of the presence of bipectinate male antennae and the siphonate mouthparts. However, after our re-examination of the holotype of the genus-type species, we found that the genus was poorly described with a number of incorrect character descriptions and inaccurate drawings concerning wings in Chang et al. 7 . First, the basal forewing nygma is herein confirmed to be absent, while a median forewing nygma appears to
2. Ratio of maximum width of costal space/subcostal space: (0) no more than 3:1; (1) 4:1-5:1; (2) more than 5:1. Comments: The forewing costal space in lacewings is in general much wider than the subcostal space of the same wing, but in most cases the maximum width of costal space does not exceed 3.0× width of subcostal space.
However, in some broad-winged species, such as the species of Psychopsoidea, the forewing costal space is wider than 4.0× width of subcostal space. Particularly, in Psychopsidae, the forewing costal space is over 5.0 times as wide as subcostal space. Comments: The presence of only one forewing scp-ra crossvein, usually near wing base, is developed in Megaloptera, Raphidioptera and many groups of Neuroptera, and it is considered to be plesiomorphic. However, in a few species, such as the dilarid species of Nallachius, there is no forewing scp-ra crossvein. In Ithonidae and Psychopsoidea, the scp-ra crossveins are well developed. Particularly, in some ithonid species of Rapisma and some large-sized kalligrammatid species with relatively denser crossvenations, there are more than 20 forewing scp-ra crossveins, which is considered to be the most derived condition. forewing RP+MA is fused with R for a short or long distance. The most typical example possessing this character state is Hemerobiidae (brown lacewings). In addition, the dilarid species of Dilar and some species of Cretanallachiinae (e.g. Burmopsychops, Oligopsychopsis, and Cretogramma) also have such fusion between forewing RP+MA and R, but with only one additional branch separated from R. In