Austromonticola, a new genus of broad-nosed weevil (Coleoptera, Curculionidae, Entiminae) from montane areas of New Zealand

Abstract Austromonticola gen. n. is proposed for a group of eight New Zealand alpine broad-nosed weevil species, all of which are here described: A. atriarius sp. n. (type locality: Umbrella Mountains, Central Otago), A. caelibatus sp. n. (type locality: Ohau Range, Mackenzie), A. furcatus sp. n. (type locality: Old Man Range, Central Otago), A. inflatus sp. n. (type locality: Hawkdun Range, Central Otago), A. planulatus sp. n. (type locality: St Marys Range, Central Otago), A. postinventus sp. n. (type locality: Kirkliston Range, South Canterbury), A. mataura sp. n. (type locality: Mt Dick, Otago Lakes) and A. rotundus sp. n. (type locality: Old Man Range, Central Otago). All species occur exclusively above 1000 m elevation in the mountains of Central Otago and South Canterbury in the South Island. A phylogeny of the genus, including six outgroups, was inferred from 33 morphological characters. It resolved the genus as monophyletic, and revealed two strongly supported clades within Austromonticola. DNA sequences of four gene regions were obtained from five species. Of these, the 3' end of COI proved to be the most suitable for the identification of specimens. Females of all species have diagnostic secondary sexual structures on the elytra and ventrites. These structures are hypothesised to have evolved to assist with oviposition in and beside cushion plants or by selection for structures to mitigate the costs to females of prolonged mating.


Introduction
The indigenous entimine weevil fauna of New Zealand currently consists of 28 described genera, containing 247 species. Taxonomic research on these weevils, especially at the genus level, has been dominated by the works of Francis Polkinghorne Pascoe (1875, 1877, 1876a, 1876b), Thomas Broun (1880, 1881, 1886, 1903, 1909a, 1909b, 1911, 1913, 1915 and David Sharp (1886) in the late 19th and early 20th centuries. Since then, few additional species have been described (Marshall 1926(Marshall , 1931(Marshall , 1937Barratt and Kuschel 1996), and-with the exception of several generic synonyms proposed by Kuschel (1964Kuschel ( , 1969Kuschel ( , 1972Kuschel ( , 1982-the composition of most New Zealand entimine weevil genera has remained largely unmodified since Broun's (1921) last work on the group. Recent research, however, indicates that understanding of the genus diversity of broad-nosed weevils in New Zealand has been obscured by imprecise and polyphyletic generic concepts (Brown 2017), and many species and genera remain undescribed. This paper describes a new genus of entimine weevils that is restricted to high-alpine vegetation types and whose females exhibit exaggerated ornamentation on the abdominal ventrites.
The mountains of New Zealand are some of the most dramatic and recognisable landscapes of the country. Areas above 1000 m in elevation form a significant proportion of the available land area in the South Island. Geological evidence reveals that these landscapes have been formed relatively recently, with most ranges only appearing in the past five million years (Youngson et al. 1998;Craw et al. 2012). Despite this youth, these alpine regions harbour a rich flora and fauna, which are both endemic to New Zealand and restricted to alpine areas (Mark 2012). The alpine endemic biota include plants (McGlone et al. 2001), birds (Michelsen-Heath and Gaze 2007), lizards (Whitaker 1984;Bell and Patterson 2008), beetles (Leschen and Buckley 2015;Seago et al. 2015), moths (Gaskin 1975;Hoare 2012), cicadas (Buckley and Simon 2007;Dugdale and Fleming 1978), cockroaches (Chinn and Gemmell 2004) and Orthoptera (Trewick et al. 2000;Trewick 2008). Resolving this paradox of distinctive and highly endemic biota in a recent landscape has been a research priority in recent decades (Heenan and McGlone 2013;Buckley and Simon 2007;Winkworth et al. 2005).

Materials and methods
Field collected specimens were killed in 100% ethanol or placed directly into a freezer at -20°C. Ethanol-preserved specimens were used preferentially for DNA extraction and sequencing.
Genitalia were examined by softening specimens for a short time in warm water, before removing the abdomen by inserting fine forceps between the metaventrite and ventrite 1. The abdomen was digested in porcine pancreatin enzyme solution for c. 36 h (Álvarez Padilla and Hormiga 2008), the lysate of which was subsequently used for DNA extraction. If specimens had not cleared satisfactorily at the end of this time, or were unsuitable for DNA extraction, abdomens were digested in room-temperature 10g/l KOH for up to two hours.
After clearing, the abdomen was flayed by cutting down the right side of the abdomen with spring scissors. Male genitalia were removed by severing the strong ligaments connecting sternite 8 to tergite 8, then cutting through the pretegminal membrane between the phallobase and the anus. Female genitalia were stained briefly by immersion in a 1g/l solution of Chlorazol Black in 70% ethanol, then removed by cutting through the membranes connecting tergites 7 and 8. Sternite 8 and tergite 8 were separated from the gonocoxites by cutting through their connecting membranes. Genitalia were photographed, then mounted on a card using dimethyl hydantoin formaldehyde (DMHF) (Liberti 2005), which was then pinned below the specimen.
Genitalia illustrations were prepared from photographs, using the program Inkscape (v. 0.91, Inkscape Team 2004-2017. Other line drawings were made with a Zeiss Stemi SV6 stereo microscope fitted with a camera lucida. These drawings were scanned and inked digitally in Inkscape. Habitus photographs were taken using a Nikon DS-Ri1 microscope fitted with a digital camera and a mechanical z-stepper. The program Nikon NIS Elements v. 4.10 was used to prepare the image stack and to produce the final montaged image. Terminology follows Oberprieler et al. (2014), Lawrence et al. (2010) and Wanat (2007). Body length was measured in lateral view, from the anterior margin of the eyes to the apex of the elytra. Rostrum width was measured across the antennal insertions in dorsal view. Legs are described in their idealised laterally extended position, thereby having dorsal, ventral, anterior and posterior surfaces. Everted ovipositors were measured from the centre of the ovipositor level with the apices of sternite 8 and tergite 8, to the apex of the gonocoxites. Pappolepidia (Brown 2017) are multiply finely divided scales (Fig. 114, "multifid hairs" of Kuschel 1969), found in abundance on the abdominal and thoracic ventrites of some species. The term 'dolabriform' is used to describe relatively short, broad scales that have a similar shape to an adze blade (Torre-Bueno 1979).
Descriptions of colour follow the terminology provided by the National Bureau of Standards (Kelly and Judd 1976). The NBS centroid colours are a comprehensive dictionary of colours, with natural-language descriptions. Digital representations of these colours have been provided by Jaffer (2011). The difference in colour contrast between elongate setiform scales ('setae') and their surrounding appressed scales is given using the rough descriptors `pale', `concolorous' and `dark'.
Specimens were prepared for scanning electron microscopy (SEM) by separating the abdomen from the specimen, removing the tergites and genitalia and brushing down the sternites. Specimens were then air-dried before being mounted with doublesided carbon tape onto aluminium SEM stubs (11 mm high, 12 mm diameter). Specimens were coated with gold using a Emitech K975X sputter coater. Photographs were taken using a JEOL JSM-7000F field emission scanning electron microscope (JEOL, Tokyo, Japan), with an accelerating voltage of 3 kV.

DNA sequencing and analysis
Only freshly collected specimens were used for sequencing. Genomic DNA was extracted from the pancreatin lysate (see above) using the Zymo Quick g-DNA Miniprep Kit (Zymo Research Corporation, Irvine, CA, U. S. A.), following the manufacturer's instructions for a proteinase k extraction. Four gene regions were sequenced: the cytochrome c oxidase subunit I (COI) mitochondrial gene, the D2-D3 region of the 28S ribosomal RNA gene, the nuclear protein-coding gene arginine kinase (ArgK) and the nuclear protein-coding carbamoyl-phosphate synthetase 2-aspartate transcarbamylase-dihydroorotase (CAD) gene.
DNA was amplified using a 25 μl polymerase chain reaction (PCR) consisting of 1.25 U iStar Taq (iNtRON Biotechnology, Seongnam, South Korea), 0.4 mM dNTP, 1.5 mM MgCl 2 and 0.2 μM of forward and reverse primers ( Table 1). The COI primer combination LCO1490-JJ/TL2-N-3014 was used preferentially in order to amplify the whole gene, which was then sequenced using all four primers. If amplification using this combination was unsuccessful, C1-J-2183/TL2-N-3014 was used. Reactions were run on a C1000 Touch thermal cycler (Bio-Rad Laboratories Inc., Hercules, CA, USA) or a MJ Mini thermal cycler (Bio-Rad Laboratories Inc.) with an initial denature at 94 °C for 2 min, followed by 40 cycles at 94 °C (20 s), variable annealing temperature (20 s) and 72 °C (60 s), and with a final extension at 72 °C for 5 min. Annealing temperatures were 45 °C for COI, and 52 °C for 28S reactions. ArgK and CAD reactions were amplified using a touchdown protocol, with annealing temperatures starting at 50 °C, decreasing by 1 °C per cycle for 5 cycles, followed by 35 cycles at 45 °C. Purified PCR products were sequenced by Macrogen (Seoul, Korea) using ABI BigDye 3.1 technology on an ABI3730XL platform (Applied Biosystems).

Morphological phylogenetic analysis
A total of 33 morphological characters were scored for 14 species (  Broun, 1915, "Inophloeus" sternalis Broun, 1904, and an undescribed genus and undescribed species represented by specimens collected from Chancellor Hut, Fox Glacier, Westland Te Poutini National Park. Specimens of this last taxon have been deposited in NZAC with specimen numbers IRE7143, IRE7144, IRE7145 and IRE7147. The phylogenetic matrix was prepared using Mesquite (version 3.10) (Maddison and Maddison 2016). Parsimonious cladograms were inferred using the parsimony ratchet (Nixon 1999), as implemented in Phangorn (version 2.0.4) (Schliep 2011), using Fitch parsimony with a random starting tree and subtree pruning and regrafting (SPR) rearrangements. The ratchet was run 100 times to ensure thorough sampling of the treespace. Bootstrap and jackknife (delete-half method, Felsenstein 2004) support values were calculated using Phangorn with 100 replicates each. Due to A. planulatus, A. caelibatus and A. postinventus not having suitable specimens available for DNA sequencing, morphological and sequence data were not combined.
Diagnosis. Integument densely covered with small, grey appressed scales, elongate setiform scales ('setae') conspicuous along elytral interstriae. Rostrum stout, in dorsal Description. Body length ranging from 3.4 mm to 8.9 mm. Densely covered with appressed scales on all surfaces, interspersed with elongate setiform scales ('setae'); appressed scales on dorsum oval, 35-55 μm long, ridges visible at 30 × magnification, generally coloured bluish grey, brownish grey or blackish grey, easily abraded. Rostrum. Subparallel proximally in dorsal view, widened at antennal insertions. Epistome punctate, plurisetose, slightly raised above frons but separation indistinct. Epifrons with longitudinal median carina, lacking sulci; continuous with occiput, without distinct dorsal separation between head capsule and rostrum. Antennae. Sockets dorsolateral, situated in apical 1/3 of rostrum. Scapes clavate, reaching posterior margin of eye in repose. Funicular segments clavate, subspherical or oblately spheroid, moderately to loosely articulated, segments 7 almost as wide as club. Clubs two times longer than wide, tapering apicad. Head capsule. Interocular width in dorsal view greater than width of rostrum at base. Eyes large, lateral, flat, ovate to subcircular with long axis vertical, parallel with sagittal axis. Ventral curvature of head capsule and rostrum in lateral view angulate, approximately 90°. Pronotum. Disc in dorsal view smooth, evenly convex. Postocular lobes poorly to well developed; fringed with numerous short vibrissae attaining a maximum length of 1/3 times anterior-posterior length of eye. Elytra. In dorsal view approximately parallel-sided in anterior 2/3. Setae arising from interstriae. Elytral declivity in lateral view rounded in males, but sutural margin at top of declivity developed into tubercles in females of several species. Interstriae 3 above the declivity slightly swollen in both sexes of most species, interstriae 5 above the declivity rarely swollen. Ventral margin in lateral view sinuous, highest point near level of metacoxae. Thorax. Procoxae contiguous. Prosternum visible behind procoxae as a raised tubercle ("prosternellum"). Metaventrite with median suture visible only as a small, circular fovea posteriorly. Metanepisternal sutures complete. Abdomen. Ventrites 1 and 2 fused, subequal in length in middle; ventrites 3 and 4 subequal in length, approximately 0.5 times shorter than 1 or 2; ventrite 5 approximately equal in length to 1 or 2. Suture separating ventrites 1 and 2 curved anteriad in middle, other sutures straight. Wings. Absent. Legs. Uniformly and densely covered with appressed scales and setae, except for the posterior surface of the metafemora. Femora unarmed, maximum girth at about distal quarter. Pro-and mesotibiae with indistinct denticles along ventral margin and mucrones at apex; protibiae wider in distal 1/3 than proximal 1/3, incurved at apex. Metatibiae with dorsal and ventral margins subparallel; apical setal comb arcuate, pale; mucrones small, inconspicuous; without corbel. Tarsi with long, coarse setae on dorsal surface, without appressed scales; underside of segments 1 to 3 with short, dense setae forming pads medially divided by an inconspicuous glabrous line. Claws simple, separate, diverging. Male genitalia. Hemisternites 8 fully separate, with a forked membranous sclerite on the anterior margin of the membrane connecting them ('spiculum relictum', Thompson 1992;Wanat 2007;Franz and Cardona-Duque 2013). Penis with pedon tubular, strongly curved, lateral lobes meeting or narrowly separated dorsally; temones shorter than pedon. Endophallus moderate in length, usually reaching anterior 1/3 of temones when in repose; armed with a variably-shaped sclerite surrounding the primary gonopore ('gonoporial sclerite'), other sclerites variably present. Tegmen with ring complete; parameroid lobes moderately developed, 0.35 times length of manubrium (Figs 85,86); manubrium shorter than temones. Female genitalia. Sternite 8 with spiculum ventrale more than twice as long as blade. Gonocoxites divided into two parts; proximal part about 2.3 times longer than distal part, largely unsclerotised except for a strongly sclerotised rod; rods ventrally situated, broadening at proximal end; distal gonocoxite moderately sclerotised. Bursa copulatrix with a single sclerite.
Distribution. Restricted to alpine regions in Otago and South Canterbury, New Zealand.
Etymology. Derived from the Latin australis, meaning 'southern' and monticola, meaning 'mountain dweller', alluding to the habitat of the species of this genus, being confined to the mountains of the southern part of the South Island. Gender masculine.
Most specimens have been collected by hand collecting, though some have been captured in pitfall traps or by heat extraction from litter and turf samples.
Etymology. From the Latin noun atriarius, 'porter, doorkeeper', in reference to the armature surrounding the female genital opening and alluding to a possible function of preventing unwanted mating attempts. The name is a noun in apposition.

Austromonticola caelibatus
Etymology. From the Latin noun caelibatus, 'celibacy', an allusion to the fact that the species is thus far known only from the male sex; the species name is a noun.
Biology. No plant associations recorded.
Biology. Specimens have been collected in association with Phyllachne cushions and Celmisia daisies. In particular, the largest series was associated with C. prorepens Petrie, 1887, but specimens have also been found with C. haastii Hook.f., 1864, and C. sessiliflora Hook.f., 1864.
Mesoventral process narrowly rounded. Mesanepisterna, mesepimera and metanepisterna covered with small pappolepidia, contrasting with metaventrite densely covered with appressed scales. Abdomen. Ventrites clothed almost exclusively with appressed scales; ventrites 1 and 2 densely clothed in females, scales dense laterally and sparser medially in males; ventrites 3 to 5 increasingly sparse. Males with ventrite 1 depressed medially; ventrite 5 flat. Females with ventrite 1 flat; ventrite 4 posterior margin produced into a lamina, usually with a deep median emargination (Fig. 104, 105) but variably shallower to entire; ventrite 5 disc with median furrow, deeply emarginate with a broad horn on either side of emargination. Apex narrowly rounded. Male genitalia. Fig. 85-88. Hemisternites with spiculum relictum slender. Penis with apex acute, upturned; ostial region thickened, forming a crest. Endophallus with gonoporial sclerite small, anterior and posterior lobes reduced. Temones 0.71 times as long as pedon. Female genitalia. Fig. 89  Etymology. This species is named after its distribution in the headwaters of the Mataura River; the name is a noun in apposition. The word mataura is Māori, of obscure meaning. Mataura was an ancestor of Ngatoro-i-rangi, the priest of the Arawa canoe. It may mean `glowing face', which is appropriate for its collection localities thus far have been on the eastern slopes of the Eyre Mountains.

Molecular diagnostics
Specimens of five species of Austromonticola were available for DNA sequencing. No fresh specimens of A. planulatus, A. caelibatus and A. postinventus were collected. Multiple specimens were available only of A. inflatus and A. mataura, and only the latter yielded multiple sequences for all gene regions. Due to these low sample numbers, conclusions regarding intra-specific variability are necessarily limited. The three protein-coding genes could all be unambiguously aligned, 28S being the only locus that required alignment gaps. The COI alignment was divided into two regions. The first represented the 5' region, corresponding to the region favoured for DNA barcoding (Hebert et al. 2003), and consisted of 669 bp, beginning at position 1239 of the Tribolium castaneum (Herbst, 1797) mitochondrial genome KM244661.1. This region was only sequenced for A. mataura and A. rotundus due to difficulties in amplifying it in other species. The second region, at the 3' end of the gene, consisted of 799 bp beginning at position 1909 of the same T. castaneum mitochondrial genome sequence. The 28S alignment was 756 bp long, beginning at position 1121 of the Tenebrio sp. reference sequence AY210843.1 (Gillespie et al. 2004). The ArgK alignment was 681 bp long, beginning at position 419 of the T. castaneum reference sequence XM_966707.4. The CAD alignment was 460 bp long, beginning at position 2082 of the T. castaneum reference sequence XM_967097.3.
Genetic variation existed in all gene regions, COI showing the greatest amount of variation, followed by CAD and ArgK, and 28S displaying the least (Figs 122-124). Of the genes sampled here, COI exhibited the greatest amount of genetic variation, as expected (Lin and Danforth 2004).
COI proved to be the most suitable gene for identifying specimens of Austromonticola. The 3' end of COI allowed unambiguous differentiation of all species with available data. This region has the greatest taxon coverage, though indications are that the 5' 'barcoding' end of the gene has higher levels of variation if amplification is successful (Fig. 122). ArgK is also a possible candidate for identification purposes, as all species displayed differences between them, however the level of variation was substantially lower than that of COI (Fig. 123). 28S and CAD are both unsuitable for specimen identification, due to there being some zero distances between species (Figs 124, 121).
The same pattern of variation in each gene region was observed when the number of diagnostic nucleotides was calculated (Figs 125-129). All species can be diagnosed using COI, with the number of nucleotides ranging between 7 and 20, and a median of 15. However, due to the lack of intra-specific sampling, these diagnostic sites should be used with caution.
Across all three protein-coding genes, A. atriarius, A. furcatus and A. mataura showed the smallest interspecific distances. In COI, A. rotundus was nearest to A. inflatus (Fig. 120), while in CAD and ArgK it was nearest to A. atriarius (Fig. 121, 123) and in 28S it was nearest to A. mataura (Fig. 124). There were no differences in the 28S sequences of A. atriarius, A. furcatus and A. inflatus (Fig. 124).

Phylogenetic analysis
Analysis of the character matrix (Table 2) resulted in a single most parsimonious cladogram (Fig. 118), with a length of 68 steps, a consistency index of 57 and a retention index of 70. Collapsing unsupported nodes (Fig. 119) increased the tree length to 78 steps.
In the discussion of characters below, the significance of synapomorphies is only discussed in relation to Austromonticola, due to limited representation of outgroup taxa.
The species of Austromonticola are united by three unambiguous synapomorphies, scale structure (character 2), the obsolete striae at the elytral apex (character 19), and the proximally widening gonocoxal rods (character 27).
In Austromonticola there are two strongly supported clades, one consisting of the larger species A. caelibatus, A. postinventus and A. inflatus (the A. inflatus clade) and the other consisting of the smaller species possessing metanepisterna with three rows of pappolepidia, a penis with an ostial crest and ventrite 5 with a strongly emarginate apex in the females, A. atriarius, A. furcatus and A. mataura (the A. mataura clade). Grouped with the latter clade is A. rotundus; however, the support for this relationship is weak, and the sagittate apex of the penis shared with A. planulatus hints at a possible relationship. Together, A. planulatus and A. rotundus provide a transitional series between the distinctly different A. inflatus clade and the A. mataura clade.

Development of ventrites in females
The highly modified ventrites in many species of Austromonticola are a particularly fascinating feature of the genus. There is a range of developments of the posterior margin of ventrite 4 of the females. No laminae are found in A. inflatus and A. postinventus, rather the posterior margin of ventrite 4 is recurved anteriad. A short lamina with a wide emargination is present in A. planulatus. Long bifurcate laminae are found in A. mataura and A. atriarius, while A. furcatus has a broader lamina with a deep emargination. Finally, A. rotundus has a long, broad lamina with a shallow emargination. The unknown female of A. caelibatus is predicted in the phylogenetic tree inferred above to be lacking a lamina, but it is equally parsimonious to infer a lamina being present in A. caelibatus, given the basal position of the species in the clade. The form of the lamina in this species would be of interest. This range of development make Austromonticola a suitable system for investigating the function of these laminae. Two hypotheses are presented in detail here.

Preparation of oviposition sites
The first hypothesis is that these ventral structures assist in the preparation of oviposition sites in cushion plants. The cushion vegetational form is a distinctive feature of New Zealand alpine plants, such as Raoulia and Phyllachne. This growth habit presents densely packed foliage underlain by a peaty layer formed by decaying leaves still attached to the plant (Cockayne 1921). In this hypothesis, the long abdominal laminae and apical horns of A. mataura and related species are used to force aside the foliage of cushion plants to allow the ovipositor to extend into the underlying layer to deposit the eggs (Fig. 130). This model of function may explain the emargination present in the lamina, its correspondence with the horns around the genital orifice, the correlation of these structures with a long, flexible ovipositor that can be extended to 3/4 of the weevil's body length, and the damaged apex of the lamina in the specimen of A. atriarius (Fig. 100). Predictions made by this model are that oviposition occurs while the female is sitting on top of the cushion plant, that females thus exposed will exhibit disruptive coloration and that eggs and early-instar larvae will be found in the centre of cushions, feeding on the peaty material beneath the foliage.
This hypothesis also provides an explanation for the recurved margin of ventrite 4 of A. inflatus and A. postinventus. In these species, it is hypothesised that the form of ventrite 4 allows maximum flexion of ventrite 5, which assists in ovipositing under the side of the cushion plants where the plant meets the surrounding substrate (Fig. 131). This model explains the much stouter ovipositor possessed by A. inflatus and A. postinventus. Predictions of this model include that eggs and early-instar larvae will be found towards the edges of the cushions, feeding on roots in the soil, that adult specimens will be more frequently found beside cushions rather than on top of them, and will be coloured like the surrounding substrate.
The rather different laminae of A. planulatus and A. rotundus suggest different oviposition behaviours or host plants from those of the two scenarios postulated above.

Mate hindrance
The second hypothesis is that these structures are mate hindrance devices. Mating pairs of entimine weevils are frequently encountered in copula in the field, and studies of their mating behaviour in captivity show that males will remain mounted on females for extended periods of time (D Watkin and SDJ Brown, unpub. data). The costs imposed by extended mounting include the energy expended in carrying males (Watson et al. 1998), potentially increased predation risk (Magnhagen 1991) and the losses involved in reduced foraging time (Stone 1995). Structures developed in Austromonticola females, such as elytral sutural tubercles, abdominal laminae and armature around the genital orifice, may enable prevention of unwanted mating attempts or assist in dislodging males if mounting becomes excessively prolonged. Evidence for this mechanism of sexual selection have been found in studies of water striders, which show similar exaggerated structures in females. Females of a number of species of Gerris Fabricius, 1794 (Heteroptera: Gerridae) possess elongate abdominal spines that decrease the duration of premating struggles, thereby decreasing energetic costs to the females (Arnqvist and Rowe 1995).
The two hypotheses presented above are not necessarily mutually exclusive. These two selection pressures may act synergistically, which may explain the rapid evolution of these structures. Further observations of oviposition and mating behaviour of Austromonticola, combined with experiments manipulating the form of the laminae, will be required to evaluate these hypotheses.
Other, alternate hypotheses for these ventral structures could include male stimulation during mating, pre-copulatory species recognition signals to prevent hybridisation, assisting the retraction of the ovipositor after oviposition has been completed and providing an area for sensory organs to determine optimum oviposition sites.
The cushion growth form is a feature of alpine vegetation worldwide, and is prevalent in the Himalayas (Chen et al. 2015, Dolezal et al. 2016, where several of the weevil species discussed above are found. Additionally, Syzygops is associated with ferns, which frequently present a dense rhizome mat. These observations lend support to the first hypothesis detailed above, which posits that abdominal laminae assist with preparation of oviposition sites in close-packed vegetational structures. Further investigation of the oviposition behaviour of these weevils will be necessary to accurately evaluate this hypothesis. It also predicts that weevils displaying abdominal laminae will be found in other regions where cushion vegetation is present, such as Tasmania (Gibson and Kirkpatrick 1985), the Andes (Molina- Montenegro et al. 2006) and Siberia (Volkov and Volkova 2015).

Relationships
The morphological phylogeny is largely consistent with the molecular data, in that both indicate a close relationship between A. furcatus, A. atriarius and A. mataura.
However, the position of A. rotundus in the morphology-based tree, placed as the sister taxon of the A. mataura clade, is not supported by the molecular data. The overall signal from the molecular data is that A. rotundus is the most distant of all the species for which DNA sequences were obtained, however no consensus was gained regarding its nearest relative. Results from COI and 28S are surprising. In the analysis of COI, A. inflatus was nearest to A. rotundus, whereas A. inflatus has the same 28S sequence as A. atriarius and A. furcatus. These results serve to bolster confidence that A. inflatus, A. postinventus and A. caelibatus are congeneric with the other members of the genus. Obtaining DNA sequences from the other species in the genus, especially the morphologically distinct A. planulatus, will be important for further insights into the relationships of species in the genus.
This diverse community is at apparent odds with the young geological age of the environment. The Southern Alps began rising around 5 million years ago (Sutherland 1996). The Old Man Range is older, with uplift estimated to have begun in the middle Miocene (c. 15 mya) (Craw et al. 2012). The Hawkdun and Kirkliston Ranges are estimated to have begun rising in the late Miocene (Youngson et al. 1998;Forsyth 2001). Prior to this time, the landscape of the Central Otago region is inferred to have been a low-relief basin, dominated by the large, freshwater Lake Manuherikia (Mildenhall 1989).
A geobiological model of the origin of the New Zealand alpine flora posited by Heenan and McGlone (2013) infers that a sizable component of the modern flora is derived from the community of plants that inhabited infertile and boggy lowland environments. It is noteworthy that one of the plant genera mentioned explicitly by Heenan and McGlone (2013) as providing evidence for their model is Phyllachne, upon which several species of alpine Entiminae have been collected (SDJ Brown pers. obs.). This model predicts that closely related species or genera may be found in lowland bogs. Unfortunately, these habitats have been greatly diminished as a result of agricultural intensification. However, there remain relatively intact remnant wetland systems in Southland that have a similar vegetation community to alpine bogs (McGlone 2009) and which may harbour sister taxa of alpine specialists. An example of this scenario is the crambid moth Orocrambus thymiastes Meyrick, 1901, which is found in alpine boggy regions, but has a population in the Awarua-Waituna Wetlands (Gaskin 1975).
An alternative possibility for the origin of the New Zealand alpine biota is dispersal from alpine regions in Australia, South America or the Northern Hemisphere. These areas have been the main sources for the majority of New Zealand alpine plant radiations (Winkworth et al. 2005). Dispersal from alpine areas elsewhere is an unlikely scenario for New Zealand's alpine weevils, as most of the genera are New Zealand endemics, with no close relatives elsewhere. However, little work has been done on the relationships of New Zealand weevils to other world faunas, and until these studies have been done, the dispersal hypothesis will remain untested.
Research into the mechanisms by which these weevil lineages have adapted to alpine environments, as has been investigated in other New Zealand alpine insects (Wharton 2011;Dunning et al. 2014), will be useful to inform further hypotheses of the origin of the New Zealand alpine weevil fauna.

Conservation significance
The localised distribution of most of the species of Austromonticola place them within the Naturally Uncommon (Range Restricted) threat classification category (Townsend et al. 2008). All species already have significant portions of their range administered by DOC as Stewardship Areas. The main threats to these species are likely to be predation by introduced mammals (O'Donnell et al. 2017) and encroachment of weeds. Non-target parasitism by adventive wasps is also a potential threat (Barlow et al. 2004;Barratt et al. 2007). All of these threats are expected to increase due to climate change (Halloy and Mark 2003). The much larger distribution of A. rotundus results in it being given the classification of Not Threatened.

Conclusion
Additional research into the biology, behaviour and physiology of the species of Austromonticola described here will offer insight into the function of the exaggerated abdominal structures of the females, and into processes by which sexual selection accelerates speciation. Further exploration and collecting, especially in areas such as the Mt Teviot/Manorburn region in Central Otago, the Pisa Range, Dunstan Mountains and Mt Aspiring National Park, will be vital for discovering additional species in the genus, which will provide further data for evaluating hypotheses of the role of historical contingency and environmental pressures on the evolution of alpine insects.          Figures 120-124. Heatmaps of the uncorrected pairwise genetic distances between Austromonticola specimens sampled. Lighter colours indicate greater distances. 120 the 3 ' region of the COI mitochondrial protein-coding gene 121 CAD nuclear protein-coding gene 122 the 5 ' ("barcoding") region of COI 123 ArgK nuclear protein-coding gene 124 28S nuclear ribosomal RNA gene.