Rosaceae, Brassicaceae and pollen beetles: exploring relationships and evolution in an anthophilous beetle lineage (Nitidulidae, Meligethes-complex of genera) using an integrative approach

Background Meligethes are pollen-beetles associated with flowers of Rosaceae as larvae. This genus currently consists of 63 known species in two subgenera, Meligethes and Odonthogethes, predominantly occurring in the eastern Palaearctic. We analyzed 74 morphological and ecological characters (169 states) of all species, as well as of 11 outgroup species from 7 Meligethinae genera (including Brassicogethes), to investigate their phylogeny. We also conducted a parallel molecular analysis on 9 Meligethes, 9 Odonthogethes, 3 Brassicogethes and 2 Meligethinus species based on DNA sequence data from mitochondrial (COI, 16S) and nuclear (CAD) genes. Results Morphological phylogenetic reconstructions supported the monophyly of the whole genus and clades corresponding to purported subgenera Meligethes s.str. and Odonthogethes. Main species-groups were mostly confirmed, however some unresolved polytomies remained. Molecular data placed members of Brassicogethes (including 42 mostly W Palearctic species associated with Brassicaceae) as sister to Odonthogethes, with this clade being sister to Meligethes s.str. This phylogenetic scenario suggests that monophyletic Meligethes s.str., Odonthogethes and Brassicogethes should be regarded alternatively as three subgenera of a monophyletic Meligethes, or three genera in a monophyletic genus-complex, with mutually monophyletic Brassicogethes and Odonthogethes. Molecular analyses estimated the origin of this lineage at ca. 14–15 Mya from a common stem including Meligethinus. Conclusions We hypothesize that the ancestor of Meligethes specialized on Rosaceae in the Middle Miocene (likely in Langhian Age) and subsequently radiated during Late Miocene and Plio-Pleistocene maintaining a trophic niche on this plant family. This radiation was primarily due to geographic isolation in E Asiatic mountain systems. Combined evidence from morphology, ancestral state parsimony reconstruction of host-plant associations and molecular evidence suggested that Rosoideae (Rosa spp.) represented the ancestral hosts of Meligethes s.str., followed by an independent shift of ancestral Odonthogethes (ca. 9–15 Mya) on Rubus (Rosoideae) and members of Rosaceae Spiraeoideae. Other ancestral Odonthogethes probably shifted again on the unrelated plant family Brassicaceae (maybe 8–14 Mya in S China), allowing a rapid westward radiation of the Brassicogethes clade. Supplementary Information The online version contains supplementary material available at 10.1186/s12983-021-00390-4.

Most species of Meligethes (and several within the related genus Brassicogethes) have been recently analyzed through an integrated approach combining morphological, molecular, and bionomical data on larval ecology [2,5,[14][15][16][17][18][19][20][21][22][23][24][25]. These contributions were based on morphological and molecular data from adults and on larval bionomical data. The thus far available molecular data set for Meligethes and Brassicogethes, including sequences of three mitochondrial and nuclear genes for 23 species (Table 2), allowed depicting a first phylogenetic scenario of Meligethes and some other related genera, and also provided a framework for understanding the origin of this group of Meligethinae, its evolution on different subfamilies of Rosaceae, and, possibly, the shift of ancestral Brassicogethes to Brassicaceae.
Although the genus Meligethes was included in previous phylogenetic studies, those were aimed to resolve suprageneric classification using molecular approaches [24][25][26] and only a few widespread W-Palaearctic species were represented. Herein, we analyzed 69 morphological characters (with 157 character states) of adults ( Fig. 5) for all 63 known Meligethes (s.l.) species (including a thus far undescribed species from S China), as well as 11 outgroup species belonging to 7 different related Meligethinae genera, including 3 Brassicogethes species. Five additional bionomical characters (with 12 character states) were also analyzed combining available (published and unpublished) data on Meligethes larval ecology. The present work, then, constitutes the first comprehensive phylogenetic analysis at the species level of this large group, with a focus on E Palaearctic lineages.

The genus Meligethes Stephens, 1830
As recently summarized [7], the genus Meligethes [ Fig. 1 a (a), (b)] was formally established by Stephens in 1830 [27], based on the type species Nitidula rufipes Marsham, 1802 [present day valid name: Meligethes atratus (Olivier, 1790)]. A new preliminary phylogenetic scenario for Meligethes s.l., previously including a heterogeneous and polyphyletic mixture of taxa, formally comprising more than 500 species worldwide, was recently presented [2,[24][25][26]. In the former paper [2], which was based on morphology of adults and preliminary molecular data, 22 genera were described as new, and 6 previously recognized subgenera of Meligethes were elevated to generic rank, delimiting Meligethes to some thirty Palaearctic species that utilized Rosaceae as larval host-plants. Following the original descriptions of the few European species [27][28][29][30], several new or presumed new species were separately added from the Eastern Palaearctic and northern Oriental Regions by a long series of authors in the time span from 1845 to 1997, including two important (although very preliminary) revisions [31,32] of both purported subgenera. Two Chinese species were added [33,34], before the revision of the whole genus Meligethes [7], as currently delimited [2], and 23 new species (21 from China) have been described. Following this revision [7], a few other new species from China were also added [8][9][10], and other potential new species are presently under scrutiny based on recently collected Chinese material. The genus, comprising a new species still waiting for a formal description, but considered in the present analysis (M. sp. cfr. Pectoralis from S China; Table 1, and Tables 3-4 in Additional files), now includes 63 species in two subgenera (Meligethes s.str. and Odonthogethes Reitter, 1871) [7][8][9][10] (Table 1).
Most Meligethes (s.l.) diversity occurs in the southern countries of the Eastern Palearctic and in the northern portions of the Oriental Region (Middle Asia, China, Japan, N Indian subcontinent); but China certainly   represents the main hot-spot of the genus, ca. 85% of the known species being at least marginally distributed in this country ( Fig. 1 b-c, and Table 1) [7,10]. Species belonging to this supposed monophyletic genus are characterized by the following unequivocal autoapomorphic morphological and bionomical characters: 1) temples behind eyes (posterolateral view): with distinct, deep elliptical pit, positioned inside the posterior terminal portion of the antennal grooves (Fig. 2a); or with distinct, shallow, subcircular pit, placed more dorsad, outside antennal grooves, close to the posteriorlateral edge of the eye ( Several other morphological characters are shared with its supposed sister genus Brassicogethes Audisio & Cline, 2009 ( Fig. 1, a (c)) [2,5,24,26], whose included species all develop as larvae on the unrelated plant family Brassicaceae. These characters include, but are not limited to: 1) terminal tarsomere simple, not toothed at base (in all Meligethes s.str.; slightly to strongly toothed at base in Odonthogethes) ( Fig. 2k-m); 2) terminal maxillary palpomera long and slender, ca. 3× longer than wide (  (Table 1 and Tables 3-4

Trophic relationships of Meligethes with Rosaceae hostplants
Rosaceae is a middle-sized plant family that includes some 3000 species and a little less than 100 genera in 3 recognized subfamilies (Rosoideae, Spiraeoideae, and Dryadoideae) [35,36]. Although exhibiting a worldwide distribution, Rosaceae are particularly diverse in northern Hemisphere temperate forests, where several genera and species of woody shrubs and small trees are important components of local forest communities. The family  [2,37,38], only large shrubs and small trees of the two main subfamilies Rosoideae and Spiraeoideae are utilized [10]. No Meligethes (nor other Meligethinae genera) are, in fact, known to develop as larvae on herbaceous Rosaceae (e.g. the widespread and species-rich Potentilla, Fragaria, and Geum), despite species of these same plant genera being commonly used by adults of several different genera and species of Meligethinae as occasional food-plants [1].
Despite limited or incomplete information on the larval host-plants of some species, in general Meligethes s.l. appear to be associated with Rosaceae belonging to Rosa, Rubus, Malus, Prunus, Crataegus, Pyracantha, Sorbaria, and Photynia [1,7,10]. A couple of these (Rosa, Rubus) belong to the subfamily Rosoideae [35], whereas others (Malus, Prunus, Crataegus, Pyracantha, Sorbaria, and Photynia) to the subfamily Spiraeoideae. The two most commonly used larval hosts of Meligethes s.l. are Rosa L. and Rubus L. Rosa, which represents the only known larval host of species in the subgenus Meligethes s.str., includes some 200 species worldwide; Rubus, which represents the main larval hosts of species in the purported subgenus Odonthogethes, includes at least between 400 and 1000 species [35,[39][40][41]. Both genera exhibit biodiversity hot-spots in China and neighboring  areas, where ca. 100 and more than 200 species respectively are known to occur [42]. A similar pattern occurs in Meligethes s.l., and in its constituent subgenera, species-groups and complexes [1,2,7]. Due to these above mentioned botanical phylogenetic and biogeographic scenarios, we comprehensively explored the evolutionary trajectories within Meligethes s.l. also to provide some hints on the role of the relationships between this group of pollen beetles and their larval host plants.

Molecular vs. morphological analyses
Evidence from our molecular analyses resulted in a rather good agreement with previously established morphology-based systematics of Meligethes and allied genera. Our molecular analyses suggested (Figs. 4-5) that most Odonthogethes species likely differentiated within the last 10 million years. These were mostly included in the O. chinensis species-group, the largest one within this genus, which are associated as larvae with Rubus spp. (Rosoideae) and with several other Spiraeoideae. Although the taxon sampling in our molecular analysis is rather partial, a preliminary conclusions and comparison with morphological-based results, allows confirming the monophyly of the present-day Odonthogethes. Nevertheless, the monophyly of the (sub) genus Meligethes s. str. Remains slightly ambiguous morphologically, suggesting that this taxon might be an assemblage from which Odonthogethes was derived. A future work with additional material of representatives of the M. vulpes species-group ( Fig. 3; Meligethes vulpes and allied species, mostly from hardly accessible countries in Middle Asia) is needed to verify, through molecular data, the phylogenetic position of this isolated group within Meligethes s. str. Additionally, our molecular analysis recovered Brassicogethes as deeply nested within Meligethes pointing to a paraphyletic condition of the genus Meligethes, as recently conceived. This evidence is partially in contrast with morphological analyses, which, on the contrary, suggests a sister-group relationship between Brassicogethes and (a monophyletic) Meligethes [Meligethes s.str. + Odonthogethes] (the tree obtained from combined molecular and morphological dataset is reported in Additional file 7). In presence of this contrasting information, the hypotheses to either downgrade Brassicogethes to the subgenus level, within the genus Meligethes s. l., or, alternatively, to raise Odonthogethes to the genus rank should be taken into account.

Ancestral state parsimony reconstruction of larval-hostplant associations
The ancestral state reconstruction plotted on the topology resulting from the combined morphological and molecular dataset analysis (Fig. 6) suggested that Rosaceae could be regarded as the ancestral host-plant family of Meligethes (s.l.) and Brassicogethes, although the actual succession of family-level larval shifts among the whole clade remains difficult to define.
Noteworthy, Meligethinus species are all associated with monocots in the family Arecaceae [1,2,43]. According to our molecular data, the common ancestor of Meligethes s. l. subsequently colonized a new niche, probably exploiting Rosaceae in the Eastern Palaearctic, while Brassicogethes, maybe also originated in the same areas, later radiated on Brassicaceae, mostly in W Palaearctic subregion. This scenario is supported by the ancestral state reconstruction, which suggested that the diverse genus Rosa L. most likely represented the ancestral host of Meligethes s. str. The larvae of most species of Meligethes s. str. Specialized on species belonging to this plant group, while members of Odonthogethes rather shifted on other Rosoideae, specifically on the genus Rubus L. Subsequently, a species of this subgenus probably colonized Rosaceae belonging to the subfamily Spiraeoideae and this ancestral host shift event may have led to a subsequent diversification of a group whose members radiated by exploiting several plants of this subfamily (e.g. Malus, Crataegus, Pyracantha and Photinia) (Fig. 6). A few isolated species belonging to both subgenera, Meligethes (M. atratus, M. flavimanus, and M. torquatus) and Odonthogethes (O. shirakii) also independently colonized Spiraeoideae of the genus Prunus [1,13]. An even more isolated host shift event was likely experienced by Odonthogethes wagneri, the only species feeding on Spiraeoideae of the genus Sorbaria, and the isolated O. flavicollis, which is the only species apparently feeding on Spiraeoideae of the genus Photinia [10]. Molecular data suggest that a more drastic ecological shift finally involved some ancestral species of Odonthogethes, able to transfer to Brassicaceae, and very likely originating the Brassicogethes clade.

Discussion
Meligethes originated in the SE Palaearctic and NE Asia and adapted to flowers and large inflorescences of Rosaceae for larval feeding. The genus was likely derived from an ancient stem of Meligethinae not distantly related to Meligethinus, Micropria, Kabakovia, Pria, and allied genera, several species of which (except among Pria and Micropria) are associated as larvae with male inflorescences of palms (Arecaceae) [1,2,24,26,43].
As for the timing of lineage splitting, the few fossil data available for Meligethinae [44,45] are based on Baltic amber material (traditionally dated back to some 36-38 Mya). However, although certainly to be referred to genera other than Meligethes as presently circumscribed, the correct classification of these fossils is still pending. The Baltic amber material from Oligocene (ca. 40-50 Mya; "Melipriopsis rasnitzyni") [45], is, for instance, hardly referable to Meligethinae, due to some characters (e.g. the distinctly bordered posterior base of pronotum; the angulose "axyllary line" on metaventrite; the ciliate pronotal sides) which are unknown among true members of this subfamily. Due to the above-mentioned uncertainties, we were not able to rely on distant external calibration points [46]. It is worth remarking, however, that even if distantly related fossils were available; these would have improperly timed the evolution of the focal Fig. 6 The strict consensus tree based on combined morphological and molecular dataset constructed via MrBayes, superimposed on ancestral state parsimony reconstruction of larval-host-plant associations executed in MESQUITE (tribes within Rosaceae according to [35, 36, 39-41]) group due to its intrinsic biological properties [47]. In fact, Meligethinae are known to be affected by rate variation among lineages, incomplete haplotype sorting and a host-plant dependent and fluctuating demography [22,23]. Hence, we decided to tentatively provide a temporal framework for the speciation events of the target taxa by relying on the most accurate substitution rates currently available for beetles and setting a relaxed clock model to overcome some of the recognized limits of this dating method [46,47]. Our molecular analyses suggest that timing of separation between [Meligethes + [Odonthogethes + Brassicogethes]] occurred in a time frame of ca. 11-13 Mya, whereas separation of this clade from the related genus Meligethinus probably dates back to [14][15] Mya. Therefore, the Middle Miocene (Langhian Age) could represent a plausible timing for the first Meligethes lineages to specialize on Rosaceae host plants. Meligethes later continued to evolve and adapt in Eastern Palaearctic and E Asiatic areas on Rosaceae Rosoideae, from which Odonthogethes plausibly radiated during the Late Miocene, Pliocene and Pleistocene on Spiraeoideae (Rosaceae subfamily). Brassicogethes, following their shift to Brassicaceae -likely from a stem of ancestral Odonthogethes -expanded and strongly diversified its constituent lineages mostly in W Palaearctic, paralleling a westward evolution, diversification and expansive radiation of Brassicaceae host-plants. Several closely related species of this (sub) genus likely differentiated only in the last 2 Mys [14,22,23]. These species, in fact, usually exhibit low levels of molecular interspecific differentiation, typical of recently specialized lineages among which genetic differentiation can be markedly slower than diversification among morphological and ecological adaptive traits. This phenomenon, well-known in different groups of recently speciated phytophagous insects [23], is probably due to the retention of common ancestral haplotypes when genetic differentiation has not had the necessary time to be fixed after speciation [23]. Among Odonthogethes, a similar phenomenon certainly involves most members of the Odonthogethes chinensis group, which, although strongly differentiated from one another based on morphological and ecological traits, exhibit low levels of interspecific genetic differentiation (Fig. 5) [Audisio et al. unpublished data].
Only a few Meligethes of the (sub) genera Meligethes and Odonthogethes were able to reach the W Palaearctic, likely during the most recent Pleistocene Glaciation Cycles, maybe due to the moderate number of potential larval host-plants (indigenous Rosaceae) occurring west of Middle Asia. Mountain systems of the SE Palaearctic (in particular those in Central and S China, Nepal, Bhutan, and NE India) and transitional areas between E Palaearctic and N Oriental Regions, seem to represent the most active centers of speciation and evolution of Meligethes and Odonthogethes, reflecting the local strong generic-and species-level diversification of Rosaceae [35,36,[40][41][42].
As mentioned above, Meligethes s. str. Appear to be more strictly specialized on Rosa spp., while Odonthogethes appear to have adapted to a much wider range of Rosaceae genera, including: Rubus to Prunus, Sorbaria and related taxa, in the two main Rosaceae subfamilies, Rosoideae and Spiraeoideae [35,36,48]. The only known more highly polyphagous Meligethes s.str. Are represented by two species (M. atratus and M. flavimanus) that are widespread from N Asia to the Iberian Peninsula. Both species are able to develop, at least in W Palaearctic, on different genera of Rosoideae and Spiraeoideae [1,7]. This could be interpreted as a recent local widening of ecological (host-plant) ranges, following post-glacial colonization of W Europe, in a region where other competing and more specialized Rosaceaedependent species of the same genus are absent.
The two largest genera of Rosaceae Rosoideae, Rubus and Rosa, split more than 40 Mya, well before the here estimated divergence between Odonthogethes (mostly associated with Rubus spp.) and Meligethes s. str. (all likely associated with Rosa spp.) that is tentatively dated ca. 12-13 Mya. This asynchronous timing clearly does not allow supporting the hypothesis of a direct coevolution [49] among these beetles and their host plants. However, a model of 'sequential evolution' -i.e. the shift of insect herbivores onto a pre-existing group of plant species [50] -could rather be raised to explain the origin of the main (deepest) specialized-clades by some ancient common ancestors able to colonize novel plant-groups of Rosoideae (Figs. 5-6). Yet, it is worth noticing that most strictly endemic and locally distributed Meligethes and Odonthogethes species from China and neighboring countries are typically associated as larvae with common and widespread floral hosts, and not to endemic plant species [10]. Therefore, geographic isolation in more or less remote mountain valleys seems to have played a more relevant role than ecological specialization to host plants in driving speciation among members included in both Meligethes and Odonthogethes clades. It is interesting to point out that Meligethes and Odonthogethes probably experienced different selective pressures respect to the W Palaearctic Brassicogethes (associated with the unrelated family Brassicaceae), whose more recently-speciated taxa were able to frequently colonize and adapt to rare and/or endemic plant species [1,5,[14][15][16][17][18][19][20][21][22][23].
In line with the above argument, we observed that morphology allowed to set apart and identify the main lineages of both Meligethes s.str. and Odonthogethes, although the extensive homoplasy between these groups prevented to fully resolve the phylogenetic relationships among species-groups (Fig. 3, Additional file 8). Additionally, several characters of both taxa, although diagnostic at the interspecific level, were widely affected by homoplasy. The sharing of so many morphological traits could suggest that members of several species-group in Meligethes s.str. and Odonthogethes, probably originated recently (e.g., in allopatry) without experiencing the strong selective pressures posed by the adaptation to numerous larval host plants and various ecological contexts, which, on the contrary, rather favored a (relatively) more rapid diversification of Brassicogethes. Consistently, in Brassicogethes, where combined phenomena of speciation triggered by larval host-plant specialization (following repeated larval host-shifts) occurred much more frequently, phylogenetic relationships among the main species-groups appear to be more clearly resolved by morphological data alone, though relationships within some morphologically recognized species groups required the support of molecular data as well [23].
We are confident that further analyses could provide more detailed phylogenetic evidence on the three involved lineages, in a more coherent scenario of their actual evolutionary relationships, despite currently partly contrasting phylogenetic information provided by morphological data and molecular markers. Additionally, molecular data suggests a markedly isolated phylogenetic position for Meligethinus dolosus Grouvelle, 1919 (from NE South Africa and Mozambique). This evidence is confirmed by morphological data [43]. This problem, as well as the internal phylogeny of Meligethinus, is beyond the scope of the present paper, and will be discussed in an upcoming effort that is specifically devoted to a morphological and molecular phylogeny of this genus [43].

Conclusions
The present study analysed 63 species of Meligethes (s.l.), setting up the basis for further research on this taxon. Judging from the relatively high number of new species (> 12) discovered during less than five years of field and Museum research following the recent revision of this group [7], combined with the vast extension of scarcely explored E Palaearctic and N Asian areas (eastern Middle Asia, Northern Indian subcontinent, S China, and northern Indochina), there is a high likelihood that the actual number of species in the (sub) genera Meligethes and Odonthogethes might increase significantly to even more than 70-80 species as a whole. The species of both clades seem to be concentrated in central and southern China, specifically in subtropical evergreen broadleaf forest zones, in eastern portions of the Qinghai-Xizang Plateau alpine vegetation zone, and in southern portions of the warm temperate deciduousbroadleaf forest zone [51] (http://www.chinamaps.org/ china/china-land-cover-map-large-2.html).
However, scarcely explored mountain areas of S China still possess new endemic species. These areas represent true biodiversity hot-spots ( Fig. 1

b-c) for both Meligethes and
Odonthogethes and their Rosaceae host-plants, and also likely include Nepal, Bhutan, Taiwan, and the nearly unexplored surrounding mountain areas of eastern India (Arunachal Pradesh) and northern Myanmar [7][8][9][10].
Despite the incomplete taxonomic and biogeographic coverage, the present morpho-ecological phylogenetic analysis and the thus far available molecular data seem to support at least the following conclusions: 1) Based on molecular evidence, the Meligethes complex of genera univocally represents a monophyletic lineage, including almost certainly three species-rich genera or subgenera (Meligethes s.str., Odonthogethes, and Brassicogethes), which are comparable in the number of inclusive species (32, 31 and 42 species, respectively). Other preliminary molecular data using mitochondrial and nuclear markers (COI, NADH, ITS2, PEPCK) [24,26], suggested that the genus Meligethinus Grouvelle (a small, mostly Oriental and Afrotropical genus including some 20 species associated as larvae with male inflorescences of Arecaceae [1,43]) is placed as a sister taxon of this clade. In fact, the phylogenetic position of Meligethes and Odonthogethes is certainly closer to both Brassicogethes and Meligethinus, than to any other recognized genera inside Meligethinae [43]. Indeed, the positional homologous insert at the apical portion of the ITS2 domain B, which, as previously demonstrated [24], is shared across all Meligethes, Odonthogethes, Brassicogethes and Meligethinus, should be considered a diagnostic sequence insertion. This combined with other slippage-derived signature sequences identified in the same paper for these related Meligethinae genera strongly corroborates a common origin of these four taxa.
2) Based only on morphological and ecological evidence (Fig. 3), the clade [Meligethes s.str. + Odonthogethes] could represent the sister-group of the genus Brassicogethes, with Meligethinus sister of this triplet of taxa. However, the molecular data presented here (Figs. 4-5) strongly suggest a different and likely more compelling scenario, with Brassicogethes and Odonthogethes being sister taxa, Meligethes being sister of the latter clade, and Meligethinus sister of this triplet of taxa.
3) In an ecological-evolutionary perspective, differently from Brassicogethes, geographic isolation in remote mountain valleys of the Eastern Palaearctic seems to have played a more relevant role than ecological specialization in species splitting and diversification within both Meligethes and Odonthogethes genera.
Finally, a short series of very recently published or upcoming papers of our research team is aimed to introduce and formalize an updated taxonomic rank for Meligethes, Odonthogethes, and Brassicogethes, and to describe some recently discovered new Chinese species, as well as the previously unknown males of a few additional taxa [52].

Field research
All species were collected as adults on their known or putative host-plants. Attempts were also made to obtain larvae to confirm the insect-host plant relationships. Larvae were collected alive by hand using entomological forceps inside flower buds, where they mainly develop during the early flowering season of hosts. Adults destined for morphological analyses were killed in small vials containing cork powder and a few drops of ethylacetate. Additional conspecific adults destined to molecular analyses were killed and preserved in absolute ethanol. Larvae destined for morphological analyses were killed and preserved in a mixture of 75% ethanol and 25% filtered pure water, while additional conspecific specimens destined for molecular analyses were killed and preserved in absolute ethanol. A few larvae, when available in number, destined to SEM analyses, were killed in absolute acetone to preserve the larval cuticular surface and corresponding setae, tubercles, and other projections.
In most cases, adult specimens were collected after direct observations on target plants, with some specimens of possible target beetles directly collected using an aspirator or by hand. This technique had to be employed to avoid physical damage to rare and locally protected plants via netting or beating techniques. Netting or beating were used only for locally abundant hostplants, such as widespread or even invasive species of the Rubus or Prunus in shrubby habitats. Pollen beetles on these plants are usually highly dispersed and therefore difficult to find and collect using visual observation only.

Morphological phylogeny and host plant associations
A matrix containing 69 morphological and 5 ecological characters (Tables 3-4) [5,6,18]. Among the 69 selected morphoecological characters, some usually significant in Meligethinae interspecific diagnostics (e.g., some from male and female genitalia) were not included. Those traits, upon cladistics analysis, resulted manifestly homoplasious in different unrelated clades, and therefore were deemed scarcely informative or even confusing in a cladistic framework. The morphological matrix was produced with MESQUITE version 3.51 [52] and subsequently analyzed in TNT version 1.5 [53]. Multistate characters were treated as unordered and zerolength branches were collapsed. Analyses were run as implicit enumeration under implied weights (concavity factor of 1 and higher), under the "traditional search option". The following parameters were applied: general RAM of 1 GB, memory set to hold 1,000,000 trees, setting 1000 replicates with tree bisection-reconnection branch swapping and saving 1000 trees per replicate. The most fitting concavity k-value of the weighting function was found using the TNT script "setk.run" [53], obtaining a k value 10.72. Characters were mapped on one of the most parsimonious trees using Winclada version 1.00.08 [54]. To confirm the deepest relationships inferred by the MP analysis, we also generated a Maximum Likelihood (ML) tree using default settings in IQ-TREE [55].
An ancestral state parsimony reconstruction of larvalhost-plant associations was carried out in MESQUITE version 3.51 [56] using the most recent phylogeny-based general classification schemes of Rosaceae [35,36]. The reconstruction of ancestral state parsimony for larvalhost-plant association (Fig. 6) was carried out and superimposed on our combined morphological and molecular dataset-based cladogram following the methods discussed elsewhere [23].

Insect-host plant relationships and phenology
We collected field information on life history, larval host-plants and phenology on ca. 45 out of 63 species of the genus; for some poorly known and rare species from N Indian subcontinent, Middle Asia, China and Japan no additional information on biology was provided. Among the analyzed species, steno-oligophagy (i.e. dependence of a single beetle species on a short series of closely related plant species belonging to the same genus or to closely related genera) is the dominant condition, while monophagous species are few.
Most species of the genus are active during the breeding season in Summer (i.e., mid-June to late July; 36 species out of 45, i.e., 80%), while only a couple of mostly W Palaearctic species are Spring specialists (breeding from early April to middle May; 2 species out of 45, i.e., 9%: M. atratus and M. flavimanus). The remaining 7 analyzed species more widely extend their annual phenology between May and August.

DNA extraction, amplification and sequencing
A total of 29 adult specimens from 18 different Meligethes s.l. species, as well as 3 species of Brassicogethes and 2 of Meligethinus were collected alive in the field and directly killed and preserved in absolute ethanol. In Table 2, the geographic details for the species are listed. Species identifications were made using morphological characters detailed elsewhere [1,[7][8][9] and in Table 1 and Table 3 in Additional files. Total genomic DNA was extracted from whole specimens, following the salting out protocol [57]. Sequences were obtained from two mitochondrial gene fragments, the Cytochrome Oxidase subunit I (COI) and 16S rRNA (16S); and from one nuclear fragment, a portion of the rudimentary gene (CAD). For PCR amplifications, the following primer pairs were used: COI = LC01490 5′-.
Amplifications of the mitochondrial genes were performed with the following general cycle conditions: initial denaturation at 96°C for three minutes, followed by 35 cycles of denaturation at 94°C for one minute, annealing at 54°-57°C for 40 s, 1-min. Extension at 72°C and a last 7-min. Elongation step at 72°C. Reactions were performed in a 25 μl volume containing (NH 4 ) 2 SO 4 16 mM, Tris-HCl 67 mM (pH 8.8 at 25°C), MgCl 2 3 mM, 1 mM of each dNTP, 0.8 pmol of each primer and 1.25 units of Taq DNA polymerase. A touchdown PCR protocol was used to amplify the CAD marker with the following thermal cycling conditions: 94°C for 3.5 min, followed by 20 cycles of 94°C, 30 s, annealing temperatures stepdowns every cycle of 0.4°C (from 58 to 50°C), 35 s, 72°C, 2.5 min and additional 20 cycles of 94°C, 30 s, 55°C, 35 s, 72°C, 2.5 min. We used an MJ MINI Personal Thermal Cycler (BIO-RAD Laboratories, US) and LifeECO Thermal Cycler to perform PCR amplifications. The PCR products were purified with a GENEAID-Gel/ PCR DNA Fragments Extraction Kit and sent to an external sequencing service (Macrogen Inc.: www. macrogen.com). Sequences were edited and aligned with GENEIUS v9.1.6 [61]. A total of 84 new sequences have been deposited in GenBank (Accession numbers COI: MT949495-MT949523; 16S: MT957149-MT957177; CAD: MT966846-MT966871).

Molecular phylogeny and divergence time estimation
Phylogenetic analyses (BI) were first performed using single-gene alignments and then, Bayesian inference (BI) and Maximum likelihood analysis (ML) were both performed on the concatenated (mtDNA + nucDNA) dataset using, respectively, MRBAYES v3.2.1 [62] and IQ-TREE [55] as implemented in W-IQ-TREE [63]. The best-fit models for the study data sets proved to be a Generalized Time-Reversible model with a proportion of invariable sites and heterogeneous substitution rates following a gamma distribution (GTR + I + G) for the COI and 16S and Generalised Time-Reversible model with a proportion of invariant sites (GTR + I) for CAD. The best fitting model to analyse each partition was selected by JMODELTEST [64] using the Akaike information criterion. The BI analysis was performed by running 5,000,000 generations, with Markov chains sampled every 1000 generations. A 10% burn-in was applied and the remaining trees were used to compute a 50% majority rule consensus tree and posterior probabilities. We assessed convergence of the runs by investigating the average standard deviation of split frequencies, potential scale reduction factor (PSRF) in MRBAYES and effective sample size (ESS) of all parameters in TRACER 1.6 [65]. Average standard deviation below 0.01, value of PSRF close to 1.00 and ESS values over 200 for all parameters were acknowledged as good indicators of convergence. A ML phylogenetic reconstruction was performed running 1000 ultrafast bootstrap replications [66] followed by 1000 replications of assessment of branch supports with single branch tests with SH-like approximate likelihood ratio test. The best fitting model to analyze each partition was selected as for BI.
To estimate the relative age of lineage divergences, an uncorrelated lognormal Bayesian molecular relaxed clock model and a Yule process prior were used on the mtDNA data set using the software BEAST v.1.8.2 [67]. Although some fossils ascribed to Meligethinae have been described and dated from Baltic ambers [44,45], these oldest representatives require a careful revision by other specialists of this group, which consider them either doubtfully attributable to Meligethes or even to the same Meligethinae subfamily. Hence, given the lack of reliable fossil records or useful dated palaeogeographic events to calibrate our trees, we used an average value of the COI substitution rate in a range between 1.5 and 3.54%, which represent the most accurate estimated rate for mitochondrial DNA in coleopterans so far [68][69][70] and hence widely applied in molecular dating of beetle phylogeography and phylogeny (e.g. 6, 23, [71][72][73][74][75]). Therefore, for the molecular clock analysis we applied an average rate of 0.0126. The GTR model was transferred to the HKY [76,77] model because there are low ESS values for some parameters in the analyses when applying the GTR model. The analysis was independently performed three times, with 100 million generations and sampling of trees every 10,000 steps. Effective Sample Size (ESS) was evaluated in Tracer v1.6 [65], considering runs with ESS values above 200. Output trees were generated in Tree Annotator v1.8.2 (BEAST package), using maximum clade credibility (MCC) after a 10% burn-in and median heights.