Weevils, weevils, weevils everywhere * Author Oberprieler, Rolf G. Author Marvaldi, Adriana E. Author Anderson, Robert S. text Zootaxa 2007 1668 491 520 journal article 10.5281/zenodo.274039 2d13c1a2-0d8f-4031-986a-3900e879582c 1175-5326 274039 Curculionidae With about 4 600 genera and 51 0 0 0 described species, the family Curculionidae is an order of magnitude larger than any other in weevils and comprises in excess of 80% of all weevil species. Its stupendous species richness is a principal factor in the large size of the Phytophaga and in fact of all Coleoptera , thus in Haldane’s Inordinate Fondness for beetles. Curculionidae occur all over the world, from the arctic zone in the north to the subantarctic islands in the south, from beaches to mountain tops, from deserts to rainforests. They feed on virtually all plants, mainly angiosperms but also gymnosperms, pteridophytes, bryophytes and lichens and occasionally they even browse on algae and cyanobacteria. Unlike all other weevil families, curculionids also make extensive use of monocotyledons as hosts, the basal subfamilies Dryophthorinae and Brachycerinae being predominantly associated with them and several taxa of other subfamilies as well. It is therefore likely that monocotyledons constitute the ancestral hosts of Curculionidae and that they may have played a pivotal role in the diversification of the family ( Marvaldi et al . 2002 , Oberprieler 2004b ). Curculionid larvae predominantly live an endophytic life inside all parts of plants, from underground roots to buds, flowers and seeds high in the canopy. However, several groups have also adopted a more ectophytic life, the larvae feeding exposed on leaves or in the soil on roots, and a few have evolved specialised life styles such as coprophagy (dung-feeding), myrmecophily and even predation. The classification of Curculionidae into natural subfamilies and tribes probably remains the largest outstanding problem in the higher classification of Coleoptera even half a century after Crowson (1955) wrote these words. A significant first step in addressing this problem was the identification of those taxa with plesiomorphic, pedotectal3 ( Alonso-Zarazaga 2007 ) male genitalia by, mainly, Morimoto (1962) , Kuschel (1971) , Thompson (1992) and Zimmerman (1993 , 1994a ). Thompson (1992) excluded all of these groups from Curculionidae , as separate families Dryophthoridae (as Rhynchophoridae ), Brachyceridae , Raymondionymidae , Cryptolaryngidae , Erirhinidae and also Platypodidae , leaving in Curculionidae only those taxa with the derived, pedal type of genitalia. In phylogenetic analyses ( Kuschel 1995 , Marvaldi & Morrone 2000 , Marvaldi et al . 2002 , Morimoto & Kojima 2006 ), however, they are included as basal lineages in Curculionidae , and Kuschel’s concept of the subfamilies Brachycerinae and Curculioninae includes both types of genitalia. Indeed, it remains to be demonstrated that the derived, pedal type of male genitalia, with no tectum remaining, has evolved only once in Curculionidae . In the expanded phylogenetic concept, the family Curculionidae is delimited by the autapomorphic characters of, in the adult, geniculate antennae (apparently independently evolved also in Brentidae : Nanophyinae ) and compact antennal club and, in the larva, 3–4 dorsal folds in the abdominal segments, a prothoracic position of the thoracic spiracle and the frontal sutures of the head blocked by a frontoepicranial bracon (bridge). A condition similar to the last also occurs in the brentid Ithycerus and is the principal reason why this genus is often placed in Curculionidae ; however, it differs from that in Curculionidae in that the frons almost completely encloses the antenna and the frontal sutures are not properly separated from the mandibular membranes (see also Sanborne 1981 ) and it is probably not homologous. Other features typical of Curculionidae (although perhaps not autapomorphic) are, in the larva, the postoccipital condyles and, in the adult, the dorsoventral flexion of the head (retractable onto the prosternum), a metendosternite with well separated anterior tendons and a spermatheca with well separated duct and gland. The number and concepts of subfamilies in Curculionidae remain chaotic and controversial, many of the 16 4 recognised in the recent generic catalogue ( Alonso-Zarazaga & Lyal 1999 ) suffering from poor definitions, amalgamation of not evidently closely related genera and an over-inflated status in the broader picture of curculionid diversity. It will take considerable efforts of painstaking character analysis, both morphological and molecular, to identify and properly delimit the main lineages of particularly the higher Curculionidae (with the pedal type of male genitalia). The smaller number of subfamilies here outlined is meant as an indication of such possible major lineages, as yet mostly without firm concepts but grouping together traditional subfamilies that are difficult to separate as smaller units. The Dryophthorinae (fig. 15) are one of the few well-defined and evidently monophyletic curculionid subfamilies, identified by several clear synapomorphies ( Kuschel 1995 , Marvaldi & Morrone 2000 ). It occurs throughout the world and includes the largest weevils known, and the majority of its approximately 1 200 species is associated with woody monocotyledons, especially palms and pandans. It is nowadays usually divided into five tribes, Dryophthorini , Stromboscerini , Orthognathini, Cryptodermatini and Rhynchophorini , and their phylogenetic relationships are under study ( Kojima & Lyal 2000 ). The Platypodinae , counting about 1 500 species world-wide, are specialised woodborers whose larvae develop on ambrosia fungi cultivated in galleries in the wood. Usually related to bark beetles, Scolytinae , and often treated as a distinct family, their phylogenetic position is still unclear but recent evidence from larval characters ( Marvaldi 1997 ) and, more tentatively, also from molecular sequences places them at the base of Curculionidae close to Dryophthorinae . 3. the aedeagus featuring a distinct dorsal tectum and ventral pedon, also referred to as “orthocerous type ” of genitalia as they typically occur in the families with orthocerous (non-geniculate) antennae; in the pedal (“gonatocerous”) type the tectum is absent 4. exluding those placed in the separate families Brachyceridae , Dryophthoridae , Erirhinidae , Raymondionymidae and Cryptolaryngidae FIGURES 17–24. 17—an undescribed species of the southern-African endemic Cryptolarynx ( Curculionidae : Brachycerinae ) on leaf of its hostplant Moraea (Iridaceae) . 18—the localized South African Somatodes misumenus ( Curculionidae : Cyclominae ) on a spathe of its larval hostplant, Bobartia (Iridaceae) . 19 – the Australian aterpine Aesiotes leucurus ( Curculionidae : Cyclominae ) on a branch of its hostplant, Callitris (Cupressaceae) . 20 – Afroleptops coetzeei ( Curculionidae : Entiminae ), a relict South African relative of the Australian leptopiines, feeding on a leaf of its Protea hostplant ( Proteaceae ). 21—the Australian Tranes lyterioides ( Curculionidae : Molytinae ) on a male cone of the cycad Macrozamia communis (Zamiaceae) , which it pollinates. 22—the long-snouted Ludovix fasciatus ( Curculionidae : Curculioninae ) from South America, which lays its eggs inside the stems of the water hyacinth Eichhornia (Pontederiaceae) , where its larvae feed on Cornops grasshopper eggs. 23—an undescribed Tychius species ( Curculionidae : Curculioninae ) from South Africa feeding on a shoot of Rooibos Tea ( Aspalathus linearis , Fabaceae ), in whose flower buds its larvae develop. 24— the South African hyperine Frontodes brevicornis ( Curculionidae : Curculioninae ) on its host, Chaetacme aristata (Ulmaceae) , on whose leaves its exposed larvae feed and spin mesh-like silken cocoons for pupation The Brachycerinae , here treated in a wider sense that includes Erirhinini , Ocladiini, Cryptolaryngini (fig. 17) and Raymondionymini ( Oberprieler 2004b ), combine more typical curculionid taxa with pedotectal male genitalia and number about 1 200 species. The difficulty of separating the flightless African Brachycerini (fig. 16) from traditional Erirhinini is exemplified foremost by the Ocladiini ( Ocladius , Tetracyphus , Desmidophorus ), which agree with the former group in genital and larval characters ( Thompson 1992 , Kuschel 1995 , Marvaldi 2000, Morimoto & Kojima 2006 ) but with many genera of traditional Erirhininae (e.g., Afghanocryptus , Aonychus , Arthrostenus , Hypselus , Tadius ) in having a setose groove below the antennal insertion and a swollen or raised, basal elytral locking mechanism. In fact, a preliminary phylogenetic analysis of Brachycerinae in this sense on morphological characters ( Oberprieler 2004b ) suggests that the deepest split may lie between the New-World Tanysphyrini (= Stenopelmini ) and Erirhinini + Ocladiini + Cryptolaryngini + Byrsopini + Brachycerini (Raymondionymini not included). Brachycerinae are also predominantly associated with monocotyledons, the Tanysphyrini with aquatic Alismatales and Commelinales (some also with aquatic grasses and ferns), the Erirhinini and Ocladiini with Poales (a few with dicotyledons or mosses) and the Cryptolaryngini, Byrsopini and Brachycerini with geophytic Asparagales and Liliales . Further studies, especially of molecular characters, are required to elucidate the precise composition of Brachycerinae and the relationships among its subgroups. The subfamily may be paraphyletic in that some groups of higher Curculionidae (with a pedal type of genitalia) may have derived from it, such as Bagoini and also Entiminae . The latter were included in Kuschel’s (1995) concept of Brachycerinae and share conspicuous characters such as deciduous mandibular cusps and metatibial corbels, although other phylogenetic studies ( Marvaldi 1997 , Marvaldi et al . 2002 ) place them in a clade with Cyclominae . The Cyclominae are a ‘subfamily of convenience’ for now, sharing no obvious synapomorphic characters. It combines several southern-hemisphere taxa of mostly large weevils, namely the African Cyclomini (fig. 18), Hipporhinini and Gronopini, the Australian Amycterini and Aterpini and the more widespread Rhythirrinini and Listroderini , together numbering more than 1 0 0 0 species. The Diabathrariini5 and Gonipterini, included in Cyclominae by Morrone (1997a , b ) and the latter placed close to some of these groups in the phylogenetic analyses of Marvaldi (1997) and Marvaldi et al . (2002) , probably do not belong here but rather in Curculioninae , judging by the endophytic habits of the former and the long rostrum (and also likely endophytic larvae) of several gonipterines. Features of the ovipositor group Hipporhinini, Gronopini and Amycterini together, and possibly Cyclomini, Rhythirrinini and Listroderini are also related, while Aterpini in their current composition are a mixed group and may include elements of other subfamilies, such as Molytinae . Host associations are poorly known in Cyclominae , but the larvae generally develop in the soil feeding on or in roots and underground stems. Amycterini and Cyclomini appear to be largely associated with monocotyledons but the other groups with dicotyledons. The larvae of Aterpini mostly tunnel in roots, trunks, stems, shoots and inflorescences of Myrtaceae and Proteaceae but some ( Aesiotes , fig. 19) also in conifers, and, unlike the other tribes, the females of at least some genera use their rostrum for preparing oviposition sites. Detailed studies, especially of the larvae, are required to properly define both the subfamily and its constituent tribes and to establish monophyletic groupings. The Entiminae (fig. 20), with more than 12 0 0 0 species described, are the largest group of weevils and spectacularly successful not only in diversity but also in distribution and abundance, including many serious agricultural pests. They are characterised by a short, broad rostrum with adelognathous mouthparts (the prementum closing the buccal cavity from beneath), mandibles bearing deciduous cusps that assist the teneral weevil to escape form its earthen pupal cell but then break off, and, in the larva, a cushion-like antennal sensorium. The last character appears to constitute a good synapomorphy for Entiminae , whereas the other two also 5. limited to the African genera Aphanonyx , Diabathrarius and Onychogymnus (which develop in seed pods of legumes such as Schotia ), and to exclude the unrelated, wood-boring Australian Aromagis , Atelicus , Kershawcis and Strongylorhinus , whose affinities are unclear (see also Zimmerman 1994a : 677) occur in other groups. The similar deciduous mandibular cusps in Brachycerinae are interpreted as a parallel development by Thompson (1992) and Marvaldi (1997) , but their homology or not with those of Entiminae remains to be convincingly established. Many genera also bear corbels on the hind tibiae like they occur in a number of Brachycerinae ; again the homology of this feature remains to be elucidated. Adult Entiminae mostly feed on leaves and young shoots and their larvae on roots in the soil, but host associations tend to be very broad and loose. The classification of the Entiminae into natural tribes and subtribes is chaotic, as many as 55 recognised in the generic catalogue of Alonso-Zarazaga & Lyal (1999) and many based on Palaearctic genera with no clear relationships to southern-hemisphere forms. The recognition of a smaller number of more distinct groups as tribes by Thompson (1992) and Marvaldi (1997 , 1998 ) — Alophini , Pachyrhynchini , Ectemnorhinini , Sitonini and Entimini — seems a more meaningful approach but needs to be expanded to cover the entire diversity of the subfamily. The subfamily Molytinae (fig. 21) as treated here combines the majority of the wood-boring taxa in Curculionidae . Its traditional, narrow concept was first expanded by Kuschel (1987) , who also commented on the difficulty of separating the group from especially the traditional Cryptorhynchinae and Cossoninae . Indeed, the “ Cryptorhynchinae ” are a mixture of taxa that have in common only a pronounced prosternal canal into which the rostrum folds in repose, but such a canal also occurs in other groups (e.g., Brachycerinae ) and is in fact quite differently constructed in different groups of “ Cryptorhynchinae ”. Kuschel (1987) moved the Ithyporini from Cryptorhynchinae to Molytinae , and Zimmerman (1994a) afforded subfamily rank to Camptorhinini and Aedemonini (as Mechistocerinae), pointing out the affinity of the latter with Molytinae , and transferred several other genera from Cryptorhynchinae to Molytinae . Other features, such as sclerolepidia (Lyal et al . 2006), are also useful in distinguishing the disparate elements of “ Cryptorhynchinae ”. Cryptorhynchini in the narrow sense may be identified by a particular type of rostral canal (ending in a mesosternal receptacle) and, like the others, appear to be just another specialised group of Molytinae . Also Psepholacini are distinct from Cryptorhynchini and approach Scolytinae in their adaptations to boring into wood with the entire body, hence reducing the rostrum and developing stout denticles on the tibiae. The boundary is also hazy between Molytinae and traditional Lixinae ( Marshall 1932 , Aslam 1963 ), and additional problems are posed by a large and poorly studied fauna of microphthalmic or eyeless species in leaf litter that are difficult to assign to any of the traditional subfamilies (e.g., Nesiobiini). Also Mesoptiliini appear to belong here, although their larvae display significant agreements with those of Scolytinae ( Lekander 1967 , May 1993 ). In the widened sense as used here, Molytinae comprise around 10 0 0 0 species worldwide. Their larvae predominantly develop in dying wood, but several also attack living stems, trunks and roots. The adults nearly always have a stout uncus (spine) at the end of their tibiae, evidently an adaptation for clinging onto wood and also aiding the female in drilling oviposition holes with her rostrum. Host associations are mainly with dicotyledons, but several taxa are also associated with monocotyledons (especially palms) and gymnosperms (conifers and cycads). The Cossoninae , here treated as a separate subfamily, are nonetheless difficult to distinguish from Molytinae . Groups such as Trypetidini , Amorphocerini, Phoenicobatini and Nesiobiini , traditionally included in Cossoninae , combine features of both and are more suitably treated as belonging in Molytinae , leaving the concept of Cossoninae more or less restricted to forms with deep mandibular sockets (limited below by a prominent hypostomal tooth), large proventricular grinding plates, a short, broad aedeagus and a strongly asymmetrical male sternite 9. Kuschel (1995) and Kuschel et al . (2000) listed some other apomorphic characters for the group, but their phylogenetic significance is not clear and many appear to be adaptations to a life in tight spaces under bark or deep in moist wood. The precise composition of Cossoninae is therefore still unclear. Alonso-Zarazaga & Lyal (1999) listed 18 tribes but Kuschel (1992) recognised only five, Onycholipini , Rhyncolini , Cossonini , Dryotribini (as Cotasterini) and Pentarthrini ; however, their concepts remain to be confirmed and clarified. Others are even more obscure, such as the Araucariini, which are often regarded as close relatives of Scolytinae or the “link” between Cossoninae and Scolytinae but most likely an artificial concept. The six included genera ( Kuschel 1966 ) are held together only by the feature of stout, socketed tibial spines but display numerous differences and also similarities with other genera (e.g., Amorphocerus with Porthetes , together placed in Molytinae as Amorphocerini ), and the Australian / New Zealand genera ( Coptocorynus , Inosomus , Mastersinella and Xenocnema , ) cannot be regarded as close relatives of the South American Araucarius nor exclusively of each other. Araucarius has a more molytine type of tibial apex rather than the typically cossonine condition, and the other genera differ from each other again in mainly the tibial apex (typically cossonine in Inosomus but shovel-like expanded in the others) but also i.a. in the number of funicular segments ( 5 in Coptocorynus , 7 in all others), the shape of the rostrum (very short and broad in Xenocnema , long in Mastersinella ) and the shape of the antennal club (various). The proper cossonines generally develop in dying or dead wood, the adults tunnelling into stems and trunks in advanced stages of decay. Most develop in dicotyledons but without any great degree of host specificity; however, some have more or less strict associations with monocotyledons, conifers and ferns. About 1 700 species are classified in Cossoninae , many of them widely distributed or even cosmopolitan. The Scolytinae or bark beetles are a large (ca. 6 0 0 0 species) and highly adapted group of weevils of considerable economic significance because of their impact on trees and the forestry industry. Their classificatory position remains controversial, specialists and sylviculturists usually treating them as a distinct family but recent morphological and phylogenetic studies of weevils generally including them as a subfamily in Curculionidae (but see Morimoto & Kojima 2003 , 2004 ). In Curculionidae they are usually closely affiliated with Platypodinae and/or Cossoninae (e.g., Kuschel et al . 2000 ), but a close relationship to Platypodinae is unlikely (see above) and in regards to Cossoninae it is unclear whether they might be adelphic to all Cossoninae or to just a subset of genera (for the link to “Araucariini” see above). Considerable character agreements also exist with the Psepholacini , to the extent that Psepholax latirostris was redescribed as Protohylastes annosus and interpreted as the most primitive scolytine ( Wood 1973 , for details see Zimmerman 1994a ), and the genus Dactylipalpus , embedded in the tribe Hylesinini of the subfamily Hylesininae in Wood’s (1986) classification, is in fact also a psepholacine, taking the reduction of the rostrum in this group to a similar extent as it occurs in Scolytinae . Further, Lekander (1967) and May (1993) identified several similarities between scolytine larvae (especially Scolytus ) and those of Mesoptiliini . In view of these extreme character convergences and misinterpretations of genera, it remains to be demonstrated that Scolytinae as currently composed are in fact a monophyletic group, and doubts have to be cast on the reliability of many of the morphological characters used in phylogenetic analyses to establish the relationships of the group. It appears that comprehensive molecular analyses, spanning a large selection of curculionid groups, are required to properly elucidate the relationships and taxonomic status of this group. Within Scolytinae , two groups are traditionally recognised, subfamilies Hylesininae and Scolytinae in Wood’s (1986) classification, but the former has already been shown to be paraphyletic ( Sequeira et al . 2000 ), as have some tribes in either group ( Tomicini , Sequeira & Farrell 2001 ; Dryocoetini , Jordal 2002 ). The current system of 26 tribes is therefore likely to undergo significant changes in future. The Baridinae are here used in the sense of Zherikhin & Egorov (1990) and Zherikhin & Gratshev (1995) , who included Conoderinae , Ceutorhynchinae , Trigonocolinae and Orobitinae in it, the former study as tribes in a subfamily but the latter as subfamilies in a separate family, Barididae. Several characters are given in support of this grouping: a transverse carina at the hind margin of the pronotum, a strongly curved submarginal fold at the interior surface of the elytra, a total fusion of metepisternum and metepimeron, a strong median carina on the inside of the metathorax (an apparently unique feature in Curculionoidea) and a number of agreements in wing venation. This relationship and the characters supporting it have not been tested and confirmed, and while there are at least some species in this grouping that have a separate metepisternum and metepimeron, the taxa in this group also share some other notable features, such as large ascending mesepimera and a similar pygidium, so that there appears to be some merit in pulling them together. A more detailed evaluation of this indicated relationship is, however, direly needed. Baridinae in this concept comprise about 8 0 0 0 species, over half of them in Baridini , which are especially diverse in the American tropics. Their larvae develop mostly in fruits but also in stems of various angiosperms, but those of Conoderini are often woodborers. Associations with monocotyledons and gymnosperms (including Gnetales) also occur. Ceutorhynchini are predominantly associated with Brassicaceae and Polygonaceae and include important biological control agents of invasive weeds. The Curculioninae include the remaining tribes of the family, such as Acalyptini , Anthonomini , Cionini , Cryptoplini , Curculionini , Derelomini , Erodiscini (fig. 22), Eugnomini , Mecinini , Ochyromerini, Otidocephalini , Rhamphini , Smicronychini , Storeini , Tychiini (fig. 23), Viticiini and likely also Diabathrariini, Gonipterini, Omophorini and others. Often referred to as “flower weevils”, their larvae develop predominantly in reproductive plant organs such as flowers, fruits and seeds. Hyperini appear to belong here as well but are rather different in their biology, their larvae feeding ectophytically on leaves and spinning net-like silken cocoons (fig. 24). The concepts of many of these tribes are yet unclear, and phylogenetic relationships among them even more so, but a few promising advances have recently been made, e.g. in the delimitation of Acalyptini and Derelomini ( Kojima & Morimoto 2005 , Franz 2005 ). Whether Curculioninae in this or a similar concept constitute a natural group remains to be seen; Kuschel (1995) advocated an even large one that includes also the Molytinae and Baridinae as outlined here as well as Erirhinini and Raymondionymini.