A Combined Molecular and Morphological Approach to Explore the Higher Phylogeny of Entimine Weevils ( Coleoptera : Curculionidae ) , with Special Reference to South American Taxa

The Entiminae are broad-nosed weevils constituting the most diverse subfamily of Curculionidae, with over 50 tribes. We performed Bayesian and Maximum Parsimony combined phylogenetic analyses with the main objective of testing higher-level relationships and the naturalness of the major Neotropical and Southern South American (Patagonia and Andes) tribes, including some members from other regions. We compiled a data matrix of 67 terminal units with 63 Entiminae species, as well as four outgroup taxa from Cyclominae, by 3522 molecular (from nuclear 18S rDNA and 28S rDNA, and mitochondrial 16S rDNA and COI gene sequences) and 70 morphological characters. The resulting trees recover a clade Entiminae with a monophyletic Cylydrorhinini and Premnotrypes branching off early. The tree resulting from parsimony analysis shows a clade of Leptopiini from the Australian region and another clade including taxa mainly distributed in the Palaearctic and Neotropical regions, but in the Bayesian tree the South American and Australian Leptopiini are grouped together. The mainly Palaearctic Entiminae (e.g., Brachyderini, Laparocerini, Otiorhynchini, Peritelini, Polydrusini, Phyllobiini and Sciaphylini) form a subclade separated from Southern Hemisphere taxa. Among the latter, the well-supported Naupactini are the sister group of the South American Tanymecini, excluding Platyaspistes, herein transferred to Leptopiini (new placement). Another well-justified clade is Eustylini–Geonemini, which also includes the enigmatic Galapagonotus, and the genus Artipus, thus corroborating its recent exclusion from Naupactini.


Introduction
Entiminae constitute the most species-rich subfamily of Curculionoidea, with about 12,000 described species worldwide, classified into ca.1370 genera and 54 tribes [1][2][3].This subfamily includes mainly taxa that were grouped in the section Adelognatha of older Curculionidae classifications.With few exceptions, most entimine species have soil-dwelling larvae, many of them polyphagous.Monophyly of Entiminae is suggested by some synapomorphic characters in the adult (e.g., mandibles with deciduous mandibular processes or their corresponding scar) [1] as well as the larva (e.g., cushion-like antennal sensorium) [4].Molecular analyses consistently show the Entiminae as being part of a larger clade of broad-nosed weevils that also includes taxa classified in the Cyclominae and Hyperinae [5][6][7][8][9][10].Such molecular studies, however, do not provide evidence supporting a clade Entiminae distinct from Cyclominae, except for a shared gene order rearrangement in the mitogenome [7,11].The entimines and allied taxa have in common a relatively short, broad rostrum that is not used in oviposition site preparation; and their larvae are mostly ectophytic, feeding on roots in the soil or on aerial plant parts (some species with endophytic or semi-endophytic larvae, especially in Cyclominae) [3,12].Among the structural features that would support the monophyly of "broad-nosed weevils" are the bilobed, largely sclerotized basal part of male sternite IX, the meso-and metatibiae apically either unarmed or mucronate (not uncinate) and the occurrence of iridescent scales with a particular nanostructure [3].
For historical reasons, weevil generic and suprageneric classifications remain essentially based on the Holarctic fauna.The relatively less studied entimines from other regions have been often classified in already well-known genera and tribes from the Holarctic region, but based on similar observed features which do not necessarily reflect unique common ancestry.A more detailed examination of the morphological features, however, suggests a deep divergence of the austral faunas relative to the Holarctic fauna, and a probable closer relationship between some Australian Entiminae and Cyclominae with those from southern Argentina and Chile [48][49][50][51][52].
In this contribution, we present results of a phylogenetic study focused on South American tribes of Entiminae.The study was based on nuclear and mitochondrial molecular markers combined with morphological data for a sample of species representing the Neotropical and Patagonian-Andean faunas as well as entimines from other regions in both southern and northern hemispheres, including some South American Cyclominae used as outgroups.
The main hypotheses to be tested are as follows: that the Neotropical and Patagonian-Andean members of the tribes Cylydrorhinini, Entimini, Eudiagogini, Eustylini, Geonemini, Leptopiini, Naupactini, Premnotrypini and Tanymecini form a clade separated from the mainly Palaearctic tribes (e.g., Brachyderini, Laparocerini, Otiorhynchini, Peritelini, Polydrusini, Phyllobiini and Sciaphylini) and Oriental Entiminae (e.g., Celeuthetini); and that the South American species of Leptopiini are related to genera of this tribe occurring in the Australian region.In addition, we are interested to address the phylogenetic positions of Platyaspistes Schoenherr and Galapagonotus Anderson & Lanteri, two South American genera with unclear affinities but currently classified in Tanymecini and Eustylini, respectively.

Specimens and Taxon Sampling
Table 1 lists the specimens and taxa used in the phylogenetic analyses, corresponding to 67 terminal species of Entiminae and Cyclominae, including valid names, acronyms, geographic data and GenBank accession numbers of DNA sequences for the four markers used.
The tribal classification used in this paper is based primarily on the world generic catalogue by Alonso-Zarazaga and Lyal [2], but incorporating some changes made subsequent to this publication.Artipus Sahlberg was classified in Geonemini and not in Naupactini [28].The names of the Palaearctic species, genera and subgenera, and their assignment to tribes were updated according to Alonso-Zarazaga et al. [53].Since the concept of Tropiphorini is very controversial (e.g., [53,54]), for the moment, we prefer to classify the genera from the southern hemisphere in the tribe Leptopiini, as suggested by Pullen et al. [55].
The classification of geographic regions follows Morrone [56] except for the Andean region in which case we recognized four biogeographic units: High Andes, Central Chile, Patagonia and Subantarctic.
We selected 63 species of Entiminae in 42 genera representing 17 tribes, as well as four species of Cyclominae in three genera from two tribes used as outgroups.The taxon sampling includes the tribes of Entiminae best representing the Neotropical and Patagonian-Andean faunas (Cylydrorhinini, Entimini, Eudiagogini, Eustylini, Geonemini, Leptopiini, Naupactini, Premonotrypini and Tanymecini), as well as some tribes typical of the Palaearctic fauna (Brachyderini, Laparacerini, Otiorhynchini, Peritelini, Phyllobiini, Polydrusini, and Sciaphilini) and some representatives of the dominant element of the Australian fauna in tribe Leptopiini [55] and Celeuthetini, corresponding to the Australian and Oriental regions.In Southern Chile and Argentina three species of Polydrusus Germar are associated with Nothofagus spp.trees [57,58].In our study, the Polydrusus specimens from this area were not preserved for DNA extraction, and consequently we only analyzed Polydrusini species from Europe.
Samples of adult specimens assayed for molecular analyses were collected in different areas of Argentina and Brazil, using a beating sheet or a sweep net, or were provided by colleagues (see Acknowledgements).The material was stored in 96-100% ethanol at −20 • C freezer until DNA extraction.Voucher specimens were deposited at the entomological collection of the Museo de La Plata, Argentina.The experiments of DNA extraction and PCR amplification of genetic material were performed at the molecular laboratories of IADIZA-CONICET (Mendoza, Argentina) and IMBIV-CONICET-UNC (Córdoba, Argentina).
Total genomic DNA was extracted from adult voucher specimens using an adapted "salting out" protocol [59] or, alternatively, the DNeasy Blood and Tissue Kit (QIAGEN, MD, USA.).Tissue was processed from 1-2 legs or part of the thorax.Extracted DNA was stored at −20 • C. Four molecular markers (two nuclear and two mitochondrial) were used in this study: 18S rDNA (entire), 28S rDNA (regions D2, D3), 16S rDNA (regions IV, V) and COI ("barcode" or 5´region).
Amplification and sequencing of the targeted loci was performed using primers listed in Table 2. Polymerase chain reaction (PCR) was performed in a 50 µL volume: 10 pmol for each primer, 0.8 mM dNTPs (Genbiotech SRL, Buenos Aires, Argentina), MgCl 2 50 mM to a final concentration of 2-4 mM, 5 µL 10 × Buffer, and 1.25 units of Taq DNA Polymerase (Invitrogen SA, Buenos Aires, Argentina).The 18S (~2000 bp) was amplified in two parts with the primer pairs E and F1094, and R1138 and P, with PCR conditions as in McKenna et al. (Seul, South Korea).Electropherograms were edited and contig assembled using ProSeq v.2.91 [62] or Sequencher v.5 (GeneCodes Corp.).All sequences were deposited in GenBank under accession numbers provided in Table 1.

Alignment of Sequences
The sequences of the ribosomal markers (nuclear 18S and 28S and mitochondrial 16S) were aligned using information on secondary structure of the rRNA genes to identify homologous positions [63].The alignment was constructed manually using as reference the structural model of arthropod rRNA [64,65].The program PAUP [66] was used for editing the annotated alignment as text file and for exporting the data in other formats for analyses.Regions identified as being of ambiguous alignment were bracketed to be excluded from phylogenetic analyses.

Morphological Characters
External and internal (dissected) adult structures were observed with a Nikon SMZ1000 stereomicroscope (Tokyo, Japan), Leica S6D and MZ16 stereomicroscopes (Wetzlar, Germany); line drawings were done with a camera lucida attached to the former scope.Digital photographs were taken with cameras attached to the S6D and MZ16 steromicroscopes.For dissections, we used standard entomological techniques [73].The terminology used for morphological characters follows Marvaldi et al. [3], and Lanteri and del Rio [28].Some characters have been illustrated to facilitate recognition of different character states (Figures 1-4).
For each terminal, we scored 70 discrete morphological characters, of which 62 correspond to the morphology of adults and 8 to larvae.The list of morphological characters is given in Table 3 and the data matrix is shown in Table S1.When larval specimen or genitalia from adult could not be examined, character states were scored with "?" and treated as missing data.For some taxa (e.g., Aegorhinus Erichson, Listroderes Schoenherr, Cylydrorhinus Guérin-Méneville, and Entimus Germar), larval information was available from specimens belonging to different, but clearly co-generic species of those included in the analysis (see [4,74,75]).Rostrum, length: more than 1.25 × as long as wide (minimum width, measured across apex without scrobes) (0) (Figure 1a); about as long as wide (slightly longer to slightly shorter) (1) (Figure 1b).

Phylogenetic Analyses
The molecular and combined data sets were analyzed using Bayesian Inference (BI) and Maximum Parsimony (MP) approaches.
The Bayesian analysis was performed using BEAST 2 v2.4.8 [80] on the Cipres Science Gateway (http://www.phylo.org)[81] with random starting trees without constraints.Each individual gene was treated as a separate partition in the analyses, resulting in a dataset comprising 3522 aligned nucleotide positions.The optimal substitution model for each partition was selected using the jModeltest software v.2.0 [82], on the basis of the corrected Akaike Information Criterion, as suggested by Burnham and Anderson [83].We applied the following substitution models: TVM + I + G (16S, 28S), SYM + I + G (18S), TIM2 + I + G (COI) and Lewis MK for morphological data.We assumed a Yule speciation model and strict molecular clock.Clock and tree parameters were linked across partitions.All priors were left as the default values in BEAUti [80].The analyses were run for a total of 30 million generations with sampling every 20,000 generations.The convergence of the runs was evaluated by accessing log files in TRACER v1.6 [84].We generated a maximum clade credibility tree in TreeAnnotator v2.4.8 [80], using a burn-in of 10% (1500 trees) and visualized in FigTree v1.4.3 [85].
Parsimony analysis was conducted using TNT v1.5 [86], excluding third codon positions of COI and considering all other characters as un-weighted and non-additive.Gaps were treated as a fifth state (default in TNT).A heuristic search with TBR branch swapping was applied to a series of 500 random addition sequences, retaining 30 trees per replicate.A strict consensus tree was calculated when more than one most parsimonious tree (MPT) was obtained.Clade stability was evaluated by 1000 parsimony bootstrap replications [87].For the MPTs we provided the basic parameters as total length (L), consistency index (CI) [88] and retention index (RI) [89].The program WinClada [90] was used for character mapping (under unambiguous, fast and slow optimization options) and to prepare MP tree figures.

Phylogenetic Analyses
The molecular and combined data sets were analyzed using Bayesian Inference (BI) and Maximum Parsimony (MP) approaches.
The Bayesian analysis was performed using BEAST 2 v2.4.8 [80] on the Cipres Science Gateway (http://www.phylo.org)[81] with random starting trees without constraints.Each individual gene was treated as a separate partition in the analyses, resulting in a dataset comprising 3522 aligned nucleotide positions.The optimal substitution model for each partition was selected using the jModeltest software v.2.0 [82], on the basis of the corrected Akaike Information Criterion, as suggested by Burnham and Anderson [83].We applied the following substitution models: TVM + I + G (16S, 28S), SYM + I + G (18S), TIM2 + I + G (COI) and Lewis MK for morphological data.We assumed a Yule speciation model and strict molecular clock.Clock and tree parameters were linked across partitions.All priors were left as the default values in BEAUti [80].The analyses were run for a total of 30 million generations with sampling every 20,000 generations.The convergence of the runs was evaluated by accessing log files in TRACER v1.6 [84].We generated a maximum clade credibility tree in TreeAnnotator v2.4.8 [80], using a burn-in of 10% (1500 trees) and visualized in FigTree v1.4.3 [85].
Parsimony analysis was conducted using TNT v1.5 [86], excluding third codon positions of COI and considering all other characters as un-weighted and non-additive.Gaps were treated as a fifth state (default in TNT).A heuristic search with TBR branch swapping was applied to a series of 500 random addition sequences, retaining 30 trees per replicate.A strict consensus tree was calculated when more than one most parsimonious tree (MPT) was obtained.Clade stability was evaluated by 1000 parsimony bootstrap replications [87].For the MPTs we provided the basic parameters as total length (L), consistency index (CI) [88] and retention index (RI) [89].The program WinClada [90] was used for character mapping (under unambiguous, fast and slow optimization options) and to prepare MP tree figures.
The outgroup Aegorhinus silvicola Kuschel (Cyclominae, Aterpini) was used to root the trees in both analyses.Posterior probability values (under BI) were considered as follows in the Results and Discussion: 0.95-1, strong to maximal clade support; 0.70-0.94,low to moderate clade support; and <0.70, not supported.

Annotated Structural Alignment
The annotated structural alignments of the entire 18S rRNA gene, the D2 and D3 expansion segments of the 28S rRNA gene and domains IV and V of the 16S rRNA gene performed for 63 entimine taxa and four cyclomine outgroups, are available online as Supplementary Materials (Table S2).

Bayesian Analysis
The tree obtained from the Bayesian analysis is shown in Figure 5. Entiminae are monophyletic (PP 0.92), with the tribes Cylydrorhynini-Premnotrypini (Andes, Patagonia) (Clade I) retrieved as a sister group of the remaining members of this subfamily, which form a quite well-supported clade (Clade II).The latter contains four main groups (A-D) of which the first three include taxa from the Southern Hemisphere and the fourth from the Palaearctic and Oriental regions.Group A shows three subclades: the pair Entimus-Eudiagogini, the sampled Leptopiini from South America (Strangaliodes Schoenherr and Vossius Kuschel) with Platyaspistes (Tanymecini) as a sister group of Vossius (PP = 1), and the sampled Leptopiini from the Australian region (Catasarcus Schoenherr, Cecyropa Pascoe and Irenimus Pascoe); Group B includes the sister tribes Naupactini-Tanymecini; Group C includes the Eustylini-Geonemini tribes, even though they may not be reciprocally monophyletic; and Group D consists of entimines mainly from the Palaearctic and Oriental regions.Within Group D, there are three subgroups, one corresponding to Bachyderini, Phyllobiini, Sciaphilini and Polydrusini; another to Otiorhynchini; and a third to Laparocerini with Simo Dejean (Peritelini), sister of the strongly supported pair Drouetius Méquignon (Peritelini) and Celeuthetes Schoenherr.

Combined Maximum-Parsimony Analysis and Synapomorphies
The Maximum Parsimony (MP) analysis yielded four most parsimonious trees (L = 3325 steps; CI = 0.35; RI = 0.53), the strict consensus of which is shown in Figure 6.Bootstrap values over 40% are indicated above the branches of the tree.The main differences from the Bayesian (BI) tree are as follows: (1) The Leptopiini from the Australian region (Catasarcus, Cecyropa and Irenimus) are separated from those of South America (Vossius and Strangaliodes) (Group A of the BI tree is broken up).( 2

Combined Maximum-Parsimony Analysis and Synapomorphies
The Maximum Parsimony (MP) analysis yielded four most parsimonious trees (L = 3325 steps; CI = 0.35; RI = 0.53), the strict consensus of which is shown in Figure 6.Bootstrap values over 40% are indicated above the branches of the tree.The main differences from the Bayesian (BI) tree are as follows: (1) The Leptopiini from the Australian region (Catasarcus, Cecyropa and Irenimus) are separated from those of South America (Vossius and Strangaliodes) (Group A of the BI tree is broken up).( 2  The selected MP tree (Figure 7) shows unambiguous optimization of morphological characters (see Figures S1 and S2 for character changes under fast and slow optimization options), with unique and homoplastic changes in black and white, respectively.It suggests that the monophyly of Entiminae is mainly justified by the presence of a scar or lasting appendage in the mandibles (18.1)The selected MP tree (Figure 7) shows unambiguous optimization of morphological characters (see Figures S1 and S2 for character changes under fast and slow optimization options), with unique and homoplastic changes in black and white, respectively.It suggests that the monophyly of Entiminae is mainly justified by the presence of a scar or lasting appendage in the mandibles (18.1) and several larval synapomorphies: antennal sensorium wider than long, cushion-like (63.1) and elliptical in apical view (64.1), maxillary mala with four setae (65.1), mandibles without accessory teeth on the intermediate portion of the cutting edge (67.1) and slightly sclerotized mandibular scrobes, which are paler than the rest of the mandible (68.1).The monophyly of entimines is also supported by two other adult mouthpart characters using the fast optimization (Figure S1): absence of two or more teeth on the inner margin of mandibles (17.1) and prementum completely concealing the maxillae (19.1); however, in Cylydrorhinini, these characters appear to have reversed to the primitive condition and the mandibular scar is sometimes markedly reduced.
Most Entiminae, except for Cylydrorhinini and Premnotrypini, share the synapomorphies of tibiae lacking spurs (49.1) and moderately separated tarsal claws (53.1), although this character shows further change.Some European groups such as Otiorhynchini have tibial spurs but these are usually more slender and shorter than those of Cyclominae (Aterpini and Listroderini) and Cylydrorhinini.Other characters supporting the monophyly of this group under fast optimization (Figure S1) are: epistome not projecting beyond the anterior margin of the rostrum (2.0) and rostrum not separated from the forehead by a deep impression (21.1).However, rostrum and forehead are separated by a deep impression in several Leptopiini from Australia, mainly Catasarcus.
Except for the Australian Leptopiini, the remaining entimines form a clade supported by characters of the rostrum, which is usually less than 1.25 × as long as wide at the apex (1.1), has an impressed or flat dorsum (6.0) and is neither unicarinate nor tricarinate (8.0); eyes usually rounded (27.0); absence of postocular lobes (36.0); and a truncate-conical or subcylindrical pronotum (38.0), in contrast to the subhexagonal pronotum of most South American Listroderini, Cylydrorhinini, and Premnotrypini.This clade split into two major subclades, one including the Palaearctic and Oriental tribes and the other the mainly Neotropical tribes.The Palaearctic and Oriental subclade is justified by the following characters: slightly defined scrobes widening backwards (13.1), occipital sutures transversal relative to the longitudinal axis of the rostrum and usually not connected with eyes (16.1), setae behind eyes directed posteriad (23.1), reduced to indistinct mucro of protibiae (47.0), narrow metepisternum and complete metepisteral-metasternal suture (54.1), larvae with labral rods diverging in their distal half (69.1) and larvae with a single seta on the alar area (70.1).The Neotropical subclade is justified by a usually present rostral groove (7.1), scrobes partially visible in dorsal view (10.1) and strongly curved towards the ventral side of the head or passing below the eyes in lateral view (12.0-1), and a scape passing below the eyes (32.0).
The intertribal relationships within both Palaearctic and Oriental and Neotropical subclades are weakly supported by synapomorphies.Within the Palaearctic and Oriental subclade, the group that includes most of the tribes (i.e., all except Laparocerini, Peritelini and Celeuthetini) is mainly justified by the presence of teeth on the three pairs of femora (45.1).Within the Neotropical subclade, the relationship between Tanymecini and Naupactini is the best supported based on the following characters: rostrum with subparallel or convergent sides toward the apex (5.1), and usually bicarinate (9.1); occipital sutures reduced to a small fovea (16.2), a usually slender and capitate scape (30.1), inner margin of protibiae with a line of denticles (48.1), presence of dorsal comb at metatibial apex (52.1) and presence of baculi along the ovipositor (62.1).Most of these characters are also present in other tribes, e.g., Eustylini-Geonemini.Although baculi are also found along the ovipositor of Australian and South American Cylydrorhinini and Leptopiini, in these groups, they are more laterally placed and usually curved at the distal end.

Separate Molecular Analyses (Bayesian and Parsimony)
The monophyly of Entiminae was supported by the BI and MP trees derived from combined molecular and morphological data, but not by those obtained from molecular data only (Figures S3  and S4), because Listroderes and Rhigopsidius Heller were included in the ingroup.In addition, several suprageneric groups (e.g., tribes) were resolved as monophyletic independently of morphological evidence (e.g., Cylydrorhinini, Naupactini, Tanymecini (except for Platyaspistes), Eudiagogini, Brachyderini, Otiorhynchini (except for Otiorhynchus subgenus Metopiorrhynchus) and Laparocerini).Moreover, both the BI and MP molecular trees justify the relationships Geonemini-Euslylini (including Artipus and Galapagonotus) and Celeuthetes-Drouetius, while the relationship Entimus-Strangaliodes was supported by the BI molecular tree as well as the combined and molecular MP trees.

Southern South American Entiminae and Cyclominae
The Cyclominae are widely distributed in the Southern Hemisphere [12,91,92], with some fossils being found in Antarctica [93], suggesting a Gondwanan ancestry [52].The tribe Aterpini, used as outgroup in our analyses, is distributed in the Australian region (Australia, Tasmania, New Guinea, New Caledonia and New Zealand) and southern Argentina and Chile [94], along the southern Andes from 37 • S to the Cape Horn, including the archipelago of southern Chile, Tierra del Fuego, South Georgia, and the Falkland Islands, an area mainly corresponding to the Subantarctic subregion [51,95,96].
Listroderini, Cylydrorhinini and Leptopiini are other tribes of Cyclominae and Entiminae with South American elements that would have close relatives in the Australian region [49,58,[97][98][99][100][101][102].However, they are not mainly associated to the Subantarctic and have diversified in northern and eastern areas, mainly Central Chile, Patagonia and the Andes.In comparison with Aterpini, these tribes are less clearly related to the Australian groups, probably because the latter areas have been subjected to more drastic geological and environmental changes after the separation of the southern continents [99].It is worth mentioning that Rhigopsidius Heller and Listroderes Schoenherr are sister taxa in the Bayesian tree, consistent with the hypothesis of Oberprieler [91] that Rhythirrinini are confined to the African region and Rhigopsidius is assumed to belong to the tribe Listroderini.Morrone [52], who recovered the relationship of Rhythirrinus Schoenherr with Rhigopsidius based on a morphological cladistic analysis of Listroderini using other cyclomines as outgroups, considered that this inference required further confirmation.From a biogeographical point of view, it seems more plausible that Rhigopsidius is more closely related to other Andean genera of Listroderini than to any African Rhytirrhinini.The morphology of Rhygopsidius, including that of the female genitalia [91,103], suggests that it is a Listroderini which has acquired new characters in response to a new environment.Indeed, the species of this genus inhabit the Andean region from southern Peru to Argentina at higher altitudes than do most other listroderines (mainly highland plateau of Puna), and their larvae bore into potato tubers [103], unlike the aerial and ectophytic larvae of most members of this tribe.
Several authors have classified Listroderini and Cylydrorhinini in the same subfamily Cylydrorhininae [104][105][106][107]. Their distribution ranges overlap but Cylydrorhinini mostly occur from 36 • to 48 • S, while Listroderini are mainly distributed from 48 • S southward [48,52], including the Tristan da Cunha-Gough Islands [49].Both tribes are phanerognathous, with a subcylindrical rostrum which is frequently unicarinate or tricarinate and has a transversal impression separating it from the forehead.In addition, they show well-developed postocular lobes and oval, dorso-ventrally oriented eyes; a typical subhexagonal pronotum which is transversally impressed behind the anterior margin; tibiae with spurs and widely separated tarsal claws (Figure 4d).The latter two characters are most likely plesiomorphic and have not been found in other South American Entiminae.These tribes mainly differ in larval morphology [49,74], and the adults show the following differences: Listroderini lack mandibular scars and their ovipositor is reduced to a pair of distal gonocoxites, whereas Cylydrorhinini have mandibular appendages or scar (though sometimes extremely reduced) and a long ovipositor with baculi.
The monophyly of Entiminae was recovered by our analyses based on molecular and morphological data (including Cylydrorhinini and excluding Listroderini), but not by the trees obtained from molecular data only.A more accurate analysis of the relationship between Cyclominae and Entiminae-with a particular focus on Listroderini and Cylydrorhinini-would be achieved by a modern taxonomic revision involving diverse, yet poorly studied, genera such as Cylydrorhinus.Moreover, such phylogenetic analysis should include increased taxon and character sampling of both groups and allied taxa.
The Clade I, Cylydrorhinini-Premnotrypini in trees from BI and MP analyses based on combined evidence, is rather weakly supported.Under slow optimization (Figure S2), this relationship is justified by having the anterior part of the epistome usually raised and projecting beyond the anterior margin of the rostrum (2.1).The Premnotrypini occur in the high Central Andes, from Colombia to northern Chile, within 2800-4500 masl.The adults show some particular features (e.g., presence of false corbels) and external similarities with Rhigopsidius (see Figure 3d) probably because adults and larvae share habits associated with development in potato tubers [108,109].Interestingly, the geographic distributions of the South American groups that attack potatoes in the Andes (Rhigopsidius and Premnotrypes Pierce) are partially sympatric with those of their putative related taxa (Listroderini and Cylydrorhinini, respectively).
The Leptopiini (type genus Leptopius Oke endemic to Australia) sensu Kuschel [58,97] basically correspond to the entimines from the Southern Hemisphere, which are listed in Tropiphorini by Alonso-Zarazaga and Lyal [2].They share some characters with Cylydrorhinini, particularly those of the rostrum and the hexagonal pronotum.However, they are adelognathous, as are most Entiminae, their tibiae lack spurs, their tarsal claws are less separated than in Cylydrorhinini, and their metatibial apex usually has true corbels (also referred to as close corbels).The leptopiines are particularly well-represented in extreme environments such as deserts, mountain areas and Subantarctic islands [58,97,101,102] although some also occur in temperate forests [110].The external expansion of the protibiae in Leptopiini from deserts may represent an adaptation to this environment [45].
Kuschel [97] described several genera and species of leptopiines from arid and semiarid environments in Chile, Argentina and Peru [58,107].More recently, new synonymies, new species and a new genus have been established [45,110,111].Thompson [43] revised the Australian genus Catasarcus and Brown [46] studied some genera endemic to New Zealand, such as Irenimus and Cecyropa.
The South American Leptopiini, Strangaliodes and Vossius form a monophyletic group with those from the Australian region in the Bayesian tree (Figure 5), but not in the MP tree (Figures 6 and 7).The former result could be interpreted as a phylogenetic signal of an ancient southern connection, whereas the MP tree suggests that Strangaliodes is related to some typical Neotropical entimines.In fact, Strangaliodes is distributed in the Transition Zone sensu Morrone [56] between the Brazilian and Patagonian biotas, the latter being more related to the Southern Hemisphere fauna.The relationship Strangaliodes-Entimus is mostly justified by molecular evidence, recovered in both combined and molecular MP trees and in the BI molecular tree.According to previous phylogenetic analyses based on morphology, Strangaliodes would be closer to other southern South American letopiines, e.g., Geniocremnus Kuschel and Megalometis Schoenherr, and less related to Leptopius and some genera from the Australian region [44].Conversely, Entimini would be related to the Neotropical Lordopini [14].Unfortunately, no samples of these groups were available in our study.
Although it is beyond the scope of our analyses, we consider that the wide concept of Tropiphorini (type genus Tropiphorus Schoenherr from the Palaearctic) as in Alonso-Zarazaga and Lyal [2] is not appropriate for the South American fauna, because the Palaearctic Tropiphorini exhibit a different combination of characters, such as mandibles without an obvious scar, absence of postocular lobes and connate tarsal claws.Therefore, we prefer to use the name Leptopiini as originally used by Kuschel [97] and more recently by Pullen et al. [55], at least until a more comprehensive study is carried out.According to molecular analyses based on mitogenome data, the Palaearctic Tropiphorini would be related to Otiorhynchini [7,8].Moreover, we disagree with the placement of Strangaliodes in the Palaearctic tribe Alophini (as a synonym of Tropiphorini) as proposed by Alonso-Zarazaga et al. [112], because of the synonymy of the monotypic genus Ctenolobus Debrochers based on a single specimen from Morocco, which was probably mislabeled.

Neotropical Entiminae
Entimini, Eudiagogini, Naupactini and Eustylini have been based on Neotropical genera and are highly diversified in this region, particularly in tropical and subtropical forests, although some derived lineages have colonized other areas and a few genera or species occur in the Nearctic [28,113].
Entimini include 13 genera [14], most of which have been revised taxonomically [13,15,114,115].They are composed of large species (about 20 mm long) with iridescent scaly vestiture and well-developed humeri and hind wings; their rostrum usually shows a median sulcus and the postocular lobes are always present.Females have a subtriangular sternite VIII, with a slightly longer than plate apodeme-as in most entimines-and a short ovipositor (about one-third the length of the abdomen) bearing styli and sclerotized proximal and distal gonocoxites [15].The phylogenetic position of Entimini is doubtful because our analyses included a single species of Entimus and lacked representatives of other allied taxa, such as Lordopini [14].
Eudiagogini are probably close to Entimini (in the BI, Figure 5), as suggested by shared larval features [74], but their adults are differentiated by having several apomorphic characters: a very short rostrum with broad pterigia; very pronounced postocular lobes covering more than half of the eyes; laterally expanded mandibular lasting appendages in some species (see Figure 1b); protibiae bearing a hook-like mucro, which forms a forceps with a tuft of stiff setae; metafemora being usually wider than the profemora; and metatibial apex showing a very broad, squamose corbel.The type genus Eudiagogus Schoenherr was taxonomically revised by Warner [116] and studied by O'Brien and Kovarik [117].
The results of both combined analyses indicated that Naupactini are monophyletic and closely related to Tanymecini, but this relationship was not recovered in the separate molecular analyses.Most species in our sampling belong to the Pantomorus-Naupactus complex [29], but other genera outside this complex (e.g., Cyrtomon Schoenherr, Stenocyphus Marshall, Hadropus Schoenherr, Ericydeus Pascoe, Briarius Fischer de Waldheim) are more similar to some Eustylini (e.g., Compsus Schoenherr, Exoderces Schoenherr) [28], probably by convergence.
The Eustylini sensu Franz [18] are an exclusively Neotropical assemblage with 20 genera and approximately 325 species, Compsus being the most diverse genus (104 species).This tribe is the subject of ongoing studies by Franz and colleagues [20,118,119], who suggested that Geonemini are paraphyletic relative to Eustylini [18].According to our analyses, Eustylini become monophyletic when the Neotropical Geonemini are included.Moreover, we corroborated that Artipus belongs to the clade Eustylini-Geonemini, and is not a Naupactini (see [28]), while Galapagonotus Anderson & Lanteri and probably also Coconotus Anderson & Lanteri (not included) belong to Eustylini [120].
On the other hand, Geonemus Schoenherr as well as other genera currently assigned to Geonemini occur in the Palaearctic region (e.g., Barynotus) and we are doubtful about their close relationship with Neotropical Geonemini.This uncertainty is supported by results based on mitogenome data [8].We think that Geonemini are probably valid for some genera outside the Neotropics, but this issue must be addressed by specialists.
Something similar occurs with Tanymecini, whose type genus Tanymecus Germar is from the Holarctic region.The Neotropical tanymecines have been extensively studied by Howden [21][22][23][24][25], particularly Pandeleteius Schoenherr, Hadromeropsis Pierce and Airosimus Howden, which were included in our analyses and constitute a well-supported monophyletic group.Conversely, Platyaspistes Schoenherr, currently classified in the Tanymecini subtribe Piazomiina (type genus Piazomias Schoenherr from Africa and Asia) has never been found to be related to the remaining Tanymecini but rather to Vossius (Leptopiini).The larval characters of Platyaspistes studied by Marvaldi [74] concur with those of the leptopiine larvae (e.g., two alar setae) but not with those of the larvae of Piazomiina, as in Piazomias and Leptomias Faust, and other Tanymecini as in Pachnaeus Schoenherr, described by van Emden [121].
Platyaspistes includes five Chilean species distributed from the Atacama desert to Valparaiso (with one present in Argentina), and differs from the sampled Tanymecini and other Neotropical Entiminae in its tarsal claws, which are connate -as in most of the Holarctic entimines studied herein-but unequal in length.Moreover, the distal coxites of the ovipositor are strongly sclerotized and bifurcated and females show a particular type of oviposition, as observed in Platyaspistes glaucus Farhaeus [122] and in P. argentinensis Kuschel [123].The eggs are laid between, and glued to, two adjacent surfaces of a host plant leaf which was previously folded by the female, in the same way as does Cyphometopus marmoratus (Blanchard) (currently classified in Leptopiini).The larvae emerge through a hole made by gnawing the meristem and complete their development in the soil as do other entimines [122].
Although there was not complete agreement between the results of the Bayesian and the Parsimony analyses regarding the Leptopiini, we propose to transfer Platyaspistes from Tanymecini to Leptopiini.

Palaearctic, and Australian and Oriental Entiminae
Both analyses recovered the entimine tribes mainly distributed in the Palaearctic region as a monophyletic group.In our study, every tribe represented by more than one species/genus is monophyletic, although no definite conclusions can be drawn due to insufficient taxon sampling.In general terms, we conclude that Brachyderini, Phyllobiini, Polydrusini and Sciaphylini might be closer to each other than to Otiorhynchini, Laparocerini and Peritelini.
The phylogenetic position of Drouetius is controversial.It was formerly treated as a subgenus of Laparocerus [35] and later as a separate genus [33] and was assigned to Peritelini [33,53].Laparocerus includes about 240 flightless species and subspecies endemic to the Atlantic islands of Macaronesia (Madeira, Selvagens, and Canary Islands), except for one species from northern Africa, Morocco [36], whereas Drouetius is endemic to the Azores Islands (northern Macaronesia), thus more distant geographically.
In a Bayesian analysis based on 16SrRNA, Drouetius was closer to Peritelus Germar than to Laparocerus [33] and probably for this reason it was transferred to Peritelini.In our analyses, Drouetius is the most closely related to Celeuthetes sp. and the pair Droetius-Celeuthetes is within the same group as Laparocerus (Figures 6 and 7) or as Laparocerus-Simo (Peritelini) (Figure 5).The Celeuthetini are distributed in the Oriental or Indo-Australian region, mainly diversified across islands of the Pacific (New Guinea, Molucas, Sulawesi and Lesser Sunda Islands) [124].According to a recent phylogenetic analysis, this tribe is most likely related to other Indo-Australian entimines, e.g., Pachyrhynchini [125] and, therefore, we believe that the close relationship Droetius-Celeuthetes may not be recovered in further studies expanding the taxon sampling.
Our results support the hypothesis that the genus Drouetius is independent from Laparocerus, but reject the hypothesis that it belongs to Peritelini.Indeed, the BI tree (Figure 5) shows that the only Peritelini included in our analysis (Simo) is close to Laparocerus and not to Drouetius, and in the MP trees Simo is in an unstable position (Figure 6).A more complete taxon sampling is necessary to elucidate the correct tribal placement of Drouetius.

General Remarks
Our study provides evidence for the monophyly of several Neotropical tribes, however, the relationships among the large clades are in general weakly supported by both analyses, preventing us from drawing firm conclusions.Notwithstanding this, results suggest that the Palaearctic entimines evolved independently from the typical Neotropical ones, e.g., Entimini, Eudiagogini, Eustylini, Naupactini, Tanymecini and other tribes not included in our analysis, such as Lordopini and Anypotactini.Future research should be aimed at expanding the geographic representation of [6]: 3 min at 95 • C; then 30 cycles of 1 min at 94 • C, 1 min at 60 • C, 1 min at 72 • C; final extension of 5 min at 72 • C; kept at 4 • C. In few instances, another combination of primers was used to obtain two smaller 18S fragments of ~500 bp each that contain the variable regions V4 and V7-V9 of 18S [60]: 18S a07 and 18S b25, and V7V9up and V7V9dwn, with PCR conditions: 3 min at 95 • C; then 10 cycles of 30 s at 95 • C, 30 s at 46 • C, 30 s at 72 • C; then 30 cycles of 30 s at 95 • C, 40 s at 48 • C, 40 s at 72 • C; final extension of 10 min at 72 • C. The fragment used of 28S (~700 bp spanning regions D2 and D3) was amplified with primers S3660 and A335, with PCR reactions as in Brown et al. [61]: 2 min at 94 • C; 40 cycles of 15 s at 94 • C, 30 s at 54 • C, 75 s at 72 • C; final extension of 7 min at 72 • C. The 16S (~600 bp spanning regions IV and V) was amplified with primers N13398 and J12887, with PCR conditions as in Hundsdoerfer et al. [5]: 4 min at 94 • C; 5 cycles of 45 s at 93 • C, 90 s at 40 • C, 90 s at 72 • C; then 26 cycles of 45 s 93 • C, 90 s at 50 • C, 90 s at 72 • C; final extension of 10 min at 72 • C. The COI gene was amplified with the primer pair LCO and HCO that produce the standard barcode fragment of 658 bp, but sometimes with the primers LCO and A3014 that allow a longer fragment of 1000-1300 bp; using in both cases the PCR conditions as in McKenna et al. [6]: 3 min at 94 • C; then 5 cycles of 30 s at 94 • C, 30 s at 42 • C, 90 s at 72 • C; then 34 cycles of 1 min at 94 • C, 30 s at 45 • C, 90 s at 72 • C; final extension of 5 min at 72 • C. The PCR products were purified and bi-directionally sequenced with the Sanger method, by means of the Sequencing Service of "Unidad de Genómica de INTA-Castelar" (Buenos Aires, Argentina) or by Macrogen Inc.

Figure 5 .
Figure 5. Phylogenetic tree of Entiminae resulting from partitioned Bayesian analysis (MCC) of DNA sequences from four markers and morphological data (see Section 2Material and Methods).Posterior probability values indicated above branches.Clades discussed in the text are indicated in red (I, II, and A-D).
) The remaining entimines are grouped into two main clades, one including tribes mainly distributed in the Palaearctic and Oriental regions, and the other including those from the Neotropics, Andes and Patagonia.(3) The latter clade includes Groups B and C and the Neotropical members of Group A of the BI, Strangaliodes-Entimus, the pair Platyaspistes-Vossius, and the Eudiagogini.(4) Within the Palaearctic and Oriental Group D, the consensus tree shows that the basal relationships are not resolved, resulting in a polytomy of Simo (Peritelini), Otiorhynchini, Laparocerus Schoenherr

Figure 5 .
Figure 5. Phylogenetic tree of Entiminae resulting from partitioned Bayesian analysis (MCC) of DNA sequences from four markers and morphological data (see Section 2 Material and Methods).Posterior probability values indicated above branches.Clades discussed in the text are indicated in red (I, II, and A-D).
) The remaining entimines are grouped into two main clades, one including tribes mainly distributed in the Palaearctic and Oriental regions, and the other including those from the Neotropics, Andes and Patagonia.(3) The latter clade includes Groups B and C and the Neotropical members of Group A of the BI, Strangaliodes-Entimus, the pair Platyaspistes-Vossius, and the Eudiagogini.(4) Within the Palaearctic and Oriental Group D, the consensus tree shows that the basal relationships are not resolved, resulting in a polytomy of Simo (Peritelini), Otiorhynchini, Laparocerus Schoenherr and Celeuthetes-Droetius, and a clade of the remaining tribes, within which Polydrusini are closer to Brachyderini and Phyllobiini than Sciaphilini.and Celeuthetes-Droetius, and a clade of the remaining tribes, within which Polydrusini are closer to BrachyderiniPhyllobiini than Sciaphilini.

Figure 6 .
Figure 6.Strict consensus tree of four most parsimonious trees of Entiminae resulting from Maximum Parsimony analysis of DNA sequences from four markers and morphological data (see Material and Methods).Bootstrap values from the combined molecular and morphological data are indicated above branches and Bootstrap values from the separate molecular data are indicated below branches.Tribes are indicated in different colors.Clades recovered in the Bayesian tree are indicated in red.

Figure 6 .
Figure 6.Strict consensus tree of four most parsimonious trees of Entiminae resulting from Maximum Parsimony analysis of DNA sequences from four markers and morphological data (see Material and Methods).Bootstrap values from the combined molecular and morphological data are indicated above branches and Bootstrap values from the separate molecular data are indicated below branches.Tribes are indicated in different colors.Clades recovered in the Bayesian tree are indicated in red.

Figure 7 .
Figure 7.Most parsimonious tree of Entiminae, selected from the four MPTs obtained after combined parsimony analysis.Morphological unambiguous character changes are indicated at branches, with numbers above and below corresponding to character and state, respectively; unique changes are in black and homoplastic changes in white.The most important clades are illustrated and highlighted with boxes and their main biogeographic areas are indicated.

Figure 7 .
Figure 7.Most parsimonious tree of Entiminae, selected from the four MPTs obtained after combined parsimony analysis.Morphological unambiguous character changes are indicated at branches, with numbers above and below corresponding to character and state, respectively; unique changes are in black and homoplastic changes in white.The most important clades are illustrated and highlighted with boxes and their main biogeographic areas are indicated.

Table 1 .
Studied taxa, geographical sources and GenBank accession numbers.Codes in boldface indicate the new sequences produced for this study.* Tribes of Outgroups, in Cyclominae.

Table 2 .
List of primers used for PCR amplification and sequencing of the markers used in the phylogenetic analyses of Entiminae.

Table 3 .
List of the 70 morphological characters, character states and codes.