The Apionidae group is only weakly (NJ) or insufficiently (ML < 50%) supported in our analyses (Fig. 1, Supp. 1) but, on the contrary, the monophyly of the genera (i. e. Aizobius, Hemitrichapion, and Osellaeus) found strong support in both the ML and NJ tree.

The genus Osellaeus is represented with three strictly subalpine-alpine taxa in the western alpine arch – O. bonvouloirii baldensis (Bellò, Meregalli & Osella, 1980) on Monte Baldo, O. bonvouloirii s. str. (Ch. Brisout, 1880) in the central and western Alps and O. bonvouloirii occidentalis Germann, 2010 in the Vercors (Germann & Szallies, 2011). We included three Swiss populations of the nominal subspecies, but the third one from the Valais did not produce a positive PCR. The one from Uri (Brisen) and the other from Fribourg (Kaiseregg) are from localities just 91 km distant from each other. As O. bonvouloirii is a flightless, and restricted to its alpine habitat and thus a very low mobile species, the detected differences (K2 distance: 0.059; Table 1) are well explainable.

Three other Apionidae were included, of which Mesotrichapion punctirostre (Gyllenhal, 1839) did not give a result. The species with the widest distribution reaching from Central Asia to France is Aizobius sedi (Germar, 1818). However, the species is restricted to xerothermic places and unable to fly, this may explain for the rather large intraspecific genetic distance (0.027) between the two samples taken 300 km from each other. The third species sampled is Hemitrichapion waltoni (Stephens, 1839), recorded from Hungaria to France. The samples taken at localities separated by a distance of 340 km, a species which has normally developed hind wings and is the most mobile of all species included and may therefore show the lowest genetic distance of all Apionidae included (0.011).

These results underline once more the importance of the need for conservation of isolated populations of flightless, ecologically highly specialised and thus low mobile endemic species.

The samples of the genus Anthonomus form a strongly supported monophyletic group in both our analyses (Fig. 1, Supp. 1). The Swiss populations of the species-pair Anthonomus rubi (Herbst, 1795) / brunnipennis Curtis, 1840 were investigated. There is some ambiguity about the status of A. brunnipennis in the Alps. The species shows a supposedly boreoalpine distribution (Germann, 2010b) and lives on Dryas octopetala, a boreoalpine cushion plant, and in northern Europe it lives also on Filipendula ulmaria L., Potentilla palustris L. and P. erecta L. Anthonomus rubi on the other side is a widespread species living on different Rosaceae, but also Cistaceae. Both species are very difficult to separate based on morphological traits, which overlap largely. The finds of brunnipennis from Switzerland were preliminarily termed as somewhat doubtful and a molecular re-investigation was suggested (Germann, 2010b, 2011a).

We here included a heterogeneous set of samples collected from the northern Alps, from Grisons and Ticino, and collected from either Dryas octopetala (sample 150 from Grisons; sample 144 northern Alps) being small and brownish and thus corresponding to A. brunnipennis, and from Helianthemum and Potentilla (sample 157 from nearby Italy and sample 152 from the northern Alps) being bigger and black and corresponding to typical A. rubi. However, the investigated COI sequences do not support the hypothesis that the specimens collected from Dryas octopetala are a sister-clade to the remaining supposedly true“ Anthonomus rubi (highest intraspecific variability of 0.046; range 0.002-0.046). This might indicate that A. brunnipennis does not occur in Switzerland, however this should be corroborated with specimens of typical A. brunnipennis from northern Europe. On the other hand, an incomplete lineage sorting and/or a too short speciation time being detected by our COI barcode marker might explain our outcome (see also the discussion about the Hypera nigrirostris-group below).

The monophyly of the genus Dichotrachelus is strongly supported in both our analyses (Fig. 1, Supp. 1). Within this genus, there is definitely more hidden diversity in these relatively immobile typically alpine living species distributed from the Rif Mountains of Morocco to the Carpathians in the east, with a speciation centre in the arc of the Alps. The species are ecologically bound either to mosses (“old” lineages) or Saxifragaceae (“derived” lineages) (Meregalli et al., 2015). Based on COI sequences, we found at least in three species considerable differences among the samples, promoting the hypotheses of existing cryptic species.

Data from D. koziorowiczi Desbrochers des Loges, 1873 from two localities on Corsica (one in the North at Col de Verghio; the other in the South on Monte Calva) show that two taxa (K2 distance: 0.067; Table 2) are likely to occur on this island, instead of one at present described species. Only the examination of the type specimen(s) will help to resolve this issue, as no precise type locality on the island has been given by Desbrochers (1873).

Similarly, with the D. maculosus Fairmaire, 1869 -species group, where specimens of D. maculosus from rather isolated populations in the Vercors, at the western border of the main distribution area, differ from those from the Swiss Prealps (K2 distance: 0.026).

Also in the D. augusti F. Solari, 1946 -species complex, more morphological diversity was discovered (see Germann, 2011b), here corroborated partly by the detected genetic diversity. The rather isolated population from the Saas Valley (sample 89) differs genetically considerably (K2 distance: 0.115!) from those of samples from the Grand St. Bernard and Col de Balme regions at the Swiss-Italian and Swiss-French border, which is indeed surprising, as it is surprisingly not reflected in their morphology, whereas D. sondereggeri Germann, 2011 shows differences, but solely results in a genetic distance of 0.016 compared with the western populations of D. augusti. Furthermore, the different forms of the penis (Germann, 2011b) detected in the western populations of D. augusti in turn are not supported by relevant differences in the COI (0.002). However, to definitely delimit and show more solid insights into the systematics of the D. augusti-species complex we would still have to include samples from the type locality of D. augusti from around Champoluc in Valle d'Aosta. Additionally, the highly specialised habitat demands of the D. augustispecies complex might explain for the genetic differences between geographically close populations: all species of this complex live in mosses growing in alpine scree slopes, an unusual and certainly underestimated habitat, less in Carabidae (where exciting discoveries have been reported e.g. Molenda, 1996; Molenda & Gude, 2003; Huber & Molenda, 2004), or Staphylinidae (Molenda, 1999), but more in weevils where hardly any research has been done, and a promising field for investigations lies idle (Nikolai Yunakov, personal comm.). The alpine scree slopes thereafter can be seen as islands for the populations of the D. augusti-complex, where gene exchange via migrating individuals across alpine grasslands and glaciers might be very limited. This specific case once more shows that samples from populations of a species, at least if we deal with low mobile species, should be chosen very carefully.

The samples of species assigned to the D. rudeni-species group, based on a similar external morphology and male genitalia with a prolonged, laterally flattened tip of penis, also clustered together (D. rudeni Stierlin, 1853, D. imhoffi Stierlin, 1857 and D. variegatus Daniel & Daniel, 1898) and therefore support the outcomes from previous morphological investigations (Table 2). The samples of D. rudeni cluster all together with high bootstrap support (ML 99%, respectively NJ 98%), although there is some herogeneity in it with sample 109 from the eastern border of the distribution near Disentis (sample 109) differing most from the others (0.010 to 0.016).

The genera Brachypera, Donus and Hypera form a strongly supported monophyletic group in both our analyses (Fig. 1, Supp. 1), with both individuals of Brachypera vidua (Gené, 1837) placed within Donus samples, even if with insufficient bootstrap support (<50%).

Despite of recent efforts to unravel the relationships at genus-level based on morphology within Hyperini (Skuhrovec, 2013), we recovered an alternative hypothesis regarding Donus and Brachypera ; where the latter at best represents a subgenus within Donus. Although, in our dataset Brachypera is solely represented by Brachypera vidua. However, these results are supported by those of Stüben et al. (2015), who included Brachypera grandini (Capiomont, 1868), B. dauci (Olivier, 1807) and B. lunata Wollaston, 1854, which clustered also paraphyletically in different clades within Donus. In our analyses, the bootstrap support for two separate clades (Donus s. l. vs Hypera) is surprisingly low and their monophyly could not be therefore definitively established based on our sequence data.

Even at the species-level, we found no support for a monophyly of all the Hypera species investigated here based on our COI data (Fig. 1, Supp. 1). In particular, the recorded genetic distances (Table 3) were relatively low (from 0.003 to 0.012) for any of the three species of the H. nigrirostris group [nigrirostris (Fabricius, 1775), ononidis (Chevrolat, 1863) and melarynchus (Olivier, 1807)]. The morphologically weakly supported hypothesis of the species status for Hypera ononidis was already questioned by Stüben et al. (2015) in their barcode approach. Although, obvious ecological differences are evident (H. ononidis lives on Ononis spp. and occurs in a sub-area of H. nigrirostris, which accepts a wider range of Fabaceae). Therefore, a more recent speciation process (not yet detectable with the possibly too conservative COI-marker), and thus the evolution of eco-species at an early stage of differentiation might be an explanation for this circumstance. Interestingly also the morphologically close H. melarynchus – living on the Fabaceae Ononis ramosissima – clustered together with H. ononidis + nigrirostris. However, H. melarynchus shows several morphological characters (biggest species of all three > 5 mm; rostrum long and slender, at least as long as pronotum; 7th article about as wide as club; elongate elytra parallel along middle; penis S-shaped in lateral view, tip elongate, tongue-like) that allow an unambiguous separation from H. nigrirostris and H. ononidis, and therefore the species status has never been questioned. This provides further evidence that the nigrirostris-species group might indeed represent a younger group where speciation is at an early stage with an incomplete lineage sorting and highlighting therefore the limited resolution power of the used barcoding marker (see Germann et al., 2010 for a similar case in Diptera).

A large number of the specimens coped with this study belongs to the genus Otiorhynchus (annex 1, Fig. 1, Supp. 1). This genus is one of the most specious genera – if not the most specious – in weevils. More than 1500 species are presently assigned to this genus and the systematics is midly expressed rather chaotic. Based on our data, the monophyly of Otiorhynchus is supported by insufficient bootstrap values in both the NJ and ML analyses. However, some new insights into alpine subgenera could be gained even if the overall relationships among all the proposed subgenera within this large genus are not always strongly supported in our analyses based on a relatively short fragment of the COI gene.

It was Yunakov (2006) who proposed subgenus Metopiorrhynchus as synonym to Nihus Reitter, 1912, which was reinstated by Magnano & Alonso-Zarazaga (2013). [The type species of Metopiorrhynchus is O. singularis (Linné, 1767) – included in our samples, and O. carinatopunctatus in Nihus, a sample that remained negative] We here provide support to the former synonymy, where species of Nihus cluster together within the subgenus Metopiorrhynchus. Interestingly, the only representative (O. grischunensis Germann, 2010) of Eunihus, a subgenus which has temporarily been placed in synonymy with Nihus, but is actually accepted as proper subgenus (Magnano & Alonso-Zarazaga, 2013), does not cluster together with Nihus. Even if the position of Eunihus remains unclear since not supported by enough bootstrap values, our result underlines its self-standing position in relation to the typical Nihus representatives. On the other side, the Corsican endemic species (O. corsicus Fairmaire, 1859) at present assigned to the subgenus Phalantorrhynchus Reitter, 1912 results in the clade Metopiorrhynchus + Nihus (bootstrap support ML: 58; NJ: 61).

Schütte et al. (2013) and Stüben et al. (2015) already provided molecular evidence to a common clade of Nihus + Metopiorrhynchus + Aranihus Reitter, 1912 [represented by the species O. parvicollis Gyllenhal, 1834 and O. ligneus (Olivier, 1807)] + Edelengus Reitter, 1912 (O. atlasicus Escalera, 1914; O. allardi Stierlin, 1872).

In all our analyses (Fig. 1, Supp. 1), all the samples belonging to Otiorhynchus pupillatus Gyllenhal, 1834 clustered together with high bootstrap support. Otiorhynchus pupillatus is a highly polymorphic species. It varies in many characters as size, proportions (rostrum, pronotum and elytra), vestiture (e.g. form of scales, density), size of teeth on femora, and (female) genital organs. It reproduces almost strictly parthenogenetically; males are only known from the junior synonym teretirostris Stierlin, 1866 in the Seealps (mentioned by Stierlin in the description, but never revised since). The validity of several of the synonymous names is highly debated, part of them were recently resurrected in Magnano & Alonso-Zarazaga (2013). Such synonyms are subdentatus Bach, 1854 (described from Thuringia, Germany), frigidus Mulsant & Rey, 1859 (from the western Alps), cyclopterus F. Solari, 1946 (Tirol, Italy/ Austria, Bayern, Germany) and the before mentioned teretirostris.

Describing every single population as a separate species cannot be the goal of studying biodiversity [in the cases of parthenogenetically reproducing populations (unfertilized eggs producing only females, and apomixis, where no meiosis is involved) we have mostly nearly identical genotypes (but see also last section of this part)]. We therefore included 15 samples of O. pupillatus, which resulted in four roughly separable genetic lineages, where three of them differ in few substitutions, and a single specimen from Grisons (sample 085) differs substantially from all others (K2 distances: 0.073-0.086). There is no morphological match with any of the before mentioned debatable species or morphotypes. The first clade comprises samples from the Central and Eastern Swiss Alps (samples 028, 054, 055, 065, 075, 084, 097, 124 from Valais, bordering Italy and Grisons), the second one a specimen from the Val Mustair (sample 106), the third specimens from the Bernese Alps and Lower Engadine (samples 007, 012, 042, 074, 088), and the fourth one (the most differing, as already mentioned), a single specimen from Central Grisons (sample 085). Well supported sister to all samples of O. pupillatus is O. difficilis, an amphigonic, also morpholologically close standing species from northern Italy, Ticino up to the Valais in the Simplon region.

The same discrepancy between morphology and genetic lineages (the retrieved clades do not include specimens sharing the same set of characters) was observed in Otiorhynchus nodosus (O. F. Müller, 1764) belonging to the subgenus Postaremus (K2 distances 0.068-0.07, Table 4). Not less than 12 synonymous names belong to this highly variable, boreo-alpine species (colour of legs from black to red, shape of body, vestiture). As already mentioned for O. pupillatus, O. nodosus is also parthenogenetic in most of its area, and throughout the Swiss Alps.

In both species mentioned, the observed well separated clades may more likely mirror several post glacial immigration lineages. An alternative explanation would be that these asexually reproducing species represent complexes of species in statu nascendi in the sense of Dobzhansky & Spassky (1959). A phenomenon reported just recently from an identically parthenogenetically reproducing entimine weevil: Naupactus cervinus Boheman, 1840 in South America (Rodriguero et al., 2013). Thereby the presence of different evolutionary units correlating with faint morphological and ecological differences could be shown, driven by many well-known evolutionary forces as mutation, selection, drift going along with geographic isolation. Whatsoever, naming these purely genetically recognisable evolutionary units/ populations will not (yet?) make sense, and unnecessarily blow up the taxonomy of Otiorhynchus. To gain a more complete insight into these complexes, definitely more samples from a broader geographical range and additional nuclear markers are needed.

All species samples from this subgenus clustered together, although with insufficient bootstrap support (Fig. 1, Supp. 1), including O. morio Fabricius, 1781, type species (!) of the subgenus Phalantorrhynchus Reitter, 1912, but morphologically hardly separable from Otiorhynchus s. str. This might show, as already suspected by the span of morphological differing members, and species only weakly differing from Otiorhynchus s. str. (as e.g. O. tenebricosus versus O. putoni Stierlin, 1891), that Phalantorrhynchus is a polyphyletic construct which needs to be thoroughly re-analysed in future.

In the case of the two samples of Otiorhynchus (s. str.) meridionalis Gyllenhal, 1834 included, one comes from Switzerland, Bern (sample 126), the other from southern France, Var (sample 129) and corresponds to the junior synonym O. civis Stierlin, 1861. This result uncovers a synonymy proposed by the first author in Pelletier (2005: 111) and later implemented in Magnano & Alonso-Zarazaga (2013). The type specimens of O. civis in the Gustav Stierlin collection (conserved in the Deutsches Entomologisches Institut, Müncheberg, Senckenberg) were examined in 2005, and one male specimen with the following label data “Gall. mer.” [Gallia meridionale = southern France] is selected, and is here designated as lectotype, labelled with a red label: “LECTOTYPUS Otiorhynchus civis Stierlin 1861 des. C. Germann 2016″. The selection of the lectotype is of special importance, as Stierlin (1861) erroneously mentioned “Griechenland” [Greece] as type locality of O. civis. In his collection there was, among other specimens from southern France, also a female specimen from Greece determined as “O. civis Stl.”. However, O. meridionalis is not (yet probably, the species is currently spreading across Europe) known from Greece, and as already stated by Reitter (1913), the specimen from Greece is most likely mislabelled. Furthermore, it is a female specimen, whereas Stierlin (1861) clearly portrayed a male specimen in his description.

The examination of the penis, including the internal sac, did surprisingly not reveal any relevant differences in the two species (the main reason for the proposed synonymy in 2005), but the external morphology, supported here by the molecular data, allows a differentiation between the two species. Therefore Otiorhynchus civis Stierlin, 1861 stat. rev. is removed from the synonymy with O. meridionalis Gyllenhal, 1834. Figure 3 shows both species, the broad elytra and the rugose surface and the denser grey hairs on elytra of O. civis (Fig. 3A) allows a differentiation from O. meridionalis, where the elytra are more elongate oval and shiny (Fig. 3B; a differentiation already given by Reitter, 1913: 44). O. civis is – after present knowledge and specimens examined – still restricted to southern France, whereas O. meridionalis is recorded more and more from surrounding countries (details in Magnano & Alonso-Zarazaga, 2013 under meridionalis).

The third species of the O. meridionalis-species group in our data set is O. aurifer Boheman, 1842, is also included in our dataset and it is well separated (Table 5).

From subgenus Nehrodistus the four species O. armatus Boheman, 1842, O. turca Boheman, 1842, O. obesus Stierlin, 1861, and O. pesarinii Diotti, 2008 are included. These species did not form a monophyletic clade, and species of other subgenera e.g. Otiolehus cluster within (Fig. 1, Supp. 1). This may show that a natural group of relatives including species of Nehrodistus – mainly characterised by the teethed femora, the rugose pronotum, the spotty distributed scales on elytra, these deprived of hairs and the slender antennae with second article almost twice as long as first – may include species of other subgenera as well. However, the detailed relationships among these species are not supported by sufficient bootstrap values and remain therefore questionable with our COI sequence data.

In the case of O. armatus the sample from the Ligurian coast differed substantially (K2 distance: 0.077) from the one taken on Ischia island. Just recently Diotti (2008) revised the species close to O. armatus and described with O. pesarinii a new species from Basilicata. The subsequent comparison with a paratype specimen provided by the author, the con-specificity of the sample specimen from Ischia Island with O. pesarinii could be confirmed.

Interesting and surprising from the morphological point of view are Otiorhynchus lugens (Germar, 1817) and O. ovalipennis Boheman, 1842 as highly supported sister taxa (ML and NJ both 99). Where a species with a single tooth on the femora, a slender rostrum, eyes laterally standing, elytra dull and deprived of hairs, and robust legs (subgenus Misenatus) is sister to O. ovalipennis (Melasemnus) with several additional small rasp like teeth on fore femora, a short rostrum, dorsally oriented eyes, shiny elytra with hairs, and gracile slender legs may represent unreliable characters for morphological estimates on phylogenetic relationships. The differences regarding teeth on femora is also present in the – although in both our analyses moderately supported – clade of O. magnicollis Stierlin, 1888 + O. thaliarchus Reitter, 1914 (Choilisanus Reitter, 1912 with unarmed femora, versus Melasemnus with teeth, often even several on fore femora). Another example for the absence and/or presence of teeth is the clade Metopiorrhynchus (teeth present) + Aranihus (teeth absent, or minute and often overlooked as in O. ligneus!) + pars Phalantorrhynchus (teeth absent) + Nihus (teeth absent), however with lower support (ML: 58; NJ: 61).

We included five samples of this genus belonging to 3 species out of 2 subgenera (Piezocnemus Chevrolat, 1869 and Chlorodrosus K. Daniel & J. Daniel, 1898). The species were not retrieved in a monophyletic clade (Fig. 1, Supp. 1), suggesting that the species concept of Polydrusus is also polyphyletic, which is not really a surprise, regarding the span of morphological variability. In the case of the sibling alpine species Polydrusus paradoxus Stierlin, 1859 / chaerodrysius Gredler, 1866 the differences in the COI support the very subtle morphological characters; both species can be distinguished mainly by the form of the scales on their femora (Germann, 2012). Thus it can be stated that small morphological differences are mirrored by a considerable genetic distance (K2 distance: 0.048). Furthermore, although from apparently very isolated populations, the samples of P. chaerodrysius collected in Valchava GR (sample 103) and Schwarenbach BE (208 km from each other; sample 143 / sample 063) differ in solely 0.2 % (Table 6). More localities were not discovered at present, despite of several specific excursions in-between. An explanation could be their parthenogenetical reproduction, where no gene-exchange as in sexual reproduction occurs.

The species status of Phyllobius vespertinus (Fabricius, 1792) was (and still is) highly debated (e.g. Dieckmann, 1979; Germann, 2011a; Alonso-Zarazaga, 2013) and recently regarded as synonym to P. pyri (Linné, 1758) (e.g. Colonnelli, 2003; Yunakov, 2013). While Phyllobius pyri lives mostly on arboreous Rosaceae and shows a more elongate body and a regularly coloured vestiture, P. vespertinus is more xerothermophilous, lives on various herbaceous plants, its body is more stout, the elytra often with a striped vestiture. We here included further specimens from the southern side of the Alps, where the characters of P. vespertinus are well pronounced [and from there (Monte Rosa, Val d'Entremont, St. Bernhard) once described as separate taxon artemisiae Desbrochers, 1873, junior synonym of P. vespertinus]. However, we provide further support that the taxa are not separable based on analyses of COI sequences (Fig. 1, Supp. 1), as already shown by Schütte et al. (2013). Similar to the Hypera nigrirostris-group, COI might be not sensitive enough to show differences, due to recent (ecological) separation of the taxa (i. e. incomplete lineage sorting), and/or genetical interchange (hybridisation) might still occur.

Both Nebria and Oreonebria are monophyletic and cluster together with good (ML) to strong (NJ) bootstrap support (Fig. 2, Supp. 2). In the case of the high-alpine Nebria cordicollis Chaudoir, 1837 -group, we here included three taxa: N. heeri K. Daniel, 1903, recently raised to species level from a subspecies of cordicollis by Szallies & Huber (2013), N. cordicollis escheri Heer, 1837 from southeastern Switzerland, and N. cordicollis tenuissima Bänninger, 1925, the westernmost populations in the Swiss Alps. All species of the cordicollis-group, as well as N. fontinalis rhaetica K. & J. Daniel, 1890 show conspicuously low interspecific K2 distances (0.002–0.018; Table 7).

As already mentioned by the authors (Szallies & Huber, 2014) in their very recent description of O. buemlisalpicola, the included samples are clearly separate (K2 distances: intraspecific = 0.0-0.015; interspecific: 0.04-0.043; Table 8) and belong to the eastern distributed Oreonebria bremii, whereas the western ones belong to O. bluemlisalpicola.