The age and phylogeny of wood boring weevils and the origin of subsociality

Molecular Phylogenetics and Evolution 59 (2011) 708–724

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/locate /ympev

The age and phylogeny of wood boring weevils and the origin of subsociality

Bjarte H. Jordal a,⇑, Andrea S. Sequeira b, Anthony I. Cognato c

a Natural History Collections, Bergen Museum, University of Bergen, NO-5020 Bergen, Norwayb Department of Biological Sciences, Wellesley College, 106 Central Street Wellesley, MA 02481, USAc Department of Entomology, Michigan State University, 243 Natural Science Bldg., East Lansing, MI 48824, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 October 2010Revised 20 January 2011Accepted 12 March 2011Available online 22 March 2011

Keywords:CurculionidaePlatypodinaeScolytinaeCossoninaeNode age estimationArginine kinaseCAD

1055-7903/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.ympev.2011.03.016

⇑ Corresponding author.E-mail addresses: [email protected] (B.H. Jo

(A.S. Sequeira), [email protected] (A.I. Cognato).

A large proportion of the hyperdiverse weevils are wood boring and many of these taxa have subsocialfamily structures. The origin and relationship between certain wood boring weevil taxa has been prob-lematic to solve and hypotheses on their phylogenies change substantially between different studies.We aimed at testing the phylogenetic position and monophyly of the most prominent wood boring taxaScolytinae, Platypodinae and Cossoninae, including a range of weevil outgroups with either the herbivo-rous or wood boring habit. Many putatively intergrading taxa were included in a broad phylogeneticanalysis for the first time in this study, such as Schedlarius, Mecopelmus, Coptonotus, Dactylipalpus, Coptoc-orynus and allied Araucariini taxa, Dobionus, Psepholax, Amorphocerus–Porthetes, and some peculiar woodboring Conoderini with bark beetle behaviour. Data analyses were based on 128 morphological charac-ters, rDNA nucleotides from the D2–D3 segment of 28S, and nucleotides and amino acids from the proteinencoding gene fragments of CAD, ArgK, EF-1a and COI. Although the results varied for some of the groupsbetween various data sets and analyses, one may conclude the following from this study: Scolytinae andPlatypodinae are likely sister lineages most closely related to Coptonotus; Cossoninae is monophyletic(including Araucariini) and more distantly related to Scolytinae; Amorphocerini is not part of Cossoninaeand Psepholax may belong to Cryptorhynchini. Likelihood estimation of ancestral state reconstruction ofsubsociality indicated five or six origins as a conservative estimate. Overall the phylogenetic results werequite dependent on morphological data and we conclude that more genetic loci must be sampled toimprove phylogenetic resolution. However, some results such as the derived position of Scolytinae wereconsistent between morphological and molecular data. A revised time estimation of the origin of Curcu-lionidae and various subfamily groups were made using the recently updated fossil age of Scolytinae(100 Ma), which had a significant influence on node age estimates.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

Weevils are the most diversified group of organisms on this pla-net with more than 62,000 species known to science (Oberprieleret al., 2007). These beetles are found in a broad range of habitats,from deserts to tropical forests, feeding on fungus and dead woodto seeds and green leaves, and found in all kind of decompositionalstages of plant tissues. The greatest diversification occurred withinthe so-called advanced weevils (Curculionidae), a radiation fre-quently ascribed to a tight association with the evolution of flow-ering plants (Farrell, 1998; Marvaldi et al., 2002; McKenna et al.,2009). In this perspective it is particularly interesting that severalspecies rich lineages have reversed the phytophagous life style to-wards being primary decomposers of dead wood, dominating the

ll rights reserved.

rdal), [email protected]

forest decomposition guild in forests world wide. Prominent repre-sentatives for such wood boring groups include the infamous barkand ambrosia beetles in the subfamilies Scolytinae and Platypodi-nae where all species construct tunnels for mating and nesting inmainly dead, woody tissue.

Beetle engravings in bark and wood are frequently associatedwith the inoculation of fungus spores and mycelial growth (Beaver,1989; Farrell et al., 2001), which often becomes a serious problemfor forest health and timber trade (Mathew, 1987; Zwolinski andGeldenhuys, 1988). Considerable scientific efforts have thereforebeen dedicated to the study of wood boring ecology and pest man-agement. Much less attention has been given to other interestingaspects of wood boring weevil biology such as their highly variableand fascinating patterns of subsocial family structure (Kirkendallet al., 1997). Elaborate care for offspring is a common trait in barkand ambrosia beetles where males and females provide variablesupport during larval development. Subsociality of this kind(Zablotny, 2003) is found sporadically in several insect orders,

http://dx.doi.org/10.1016/j.ympev.2011.03.016

mailto:[email protected]




http://www.sciencedirect.com/science/journal/10557903

http://www.elsevier.com/locate/ympev

B.H. Jordal et al. / Molecular Phylogenetics and Evolution 59 (2011) 708–724 709

but is only known from a few weevil groups beyond Scolytinae andPlatypodinae, such as members of the cossonine tribes Araucariini(Kuschel, 1966; May, 1993) and Onycholipini (e.g. Mecke, 2002),and from the baridine subtribe Campyloscelina (Conoderini)(Thompson, 1996). Despite considerable efforts in reconstructingthe phylogeny and age of weevils (Kuschel, 1995; Kuschel et al.,2000; Marvaldi et al., 2002; McKenna et al., 2009), the origin andevolution of the wood boring habit is still controversial. A morestable phylogeny with predictable relationships between at leastsome of the most contentious taxa would therefore improve ourunderstanding of weevil classification and the evolution of woodboring and subsociality.

Resolving mega-diverse clades of Cretaceous age (see e.g.Grimaldi and Engel, 2005; McKenna et al., 2009) is frequentlyimpeded by the lack of sufficient molecular and morphologicaldata. Even more problematic is the typically low number ofsampled taxa relative to the total diversity in a clade (Franz andEngel, 2010), particularly so when missing taxa are transitionalbetween distantly related lineages and thus may reduce theproblems associated with long branches (see e.g. Pick et al.,2010). The phylogenetic resolution in weevils seems particularlylimited by the apparently short time frame in which the maincurculionid lineages originated (McKenna et al., 2009). Not onlyare diagnostic molecular changes hard to detect at the base of suchancient rapid radiations, but a correspondingly near continuousvariation in morphological characters is equally problematic(Oberprieler et al., 2007). With apparently very little extinctionof main lineages since the origin of Curculionidae, the classificationof subfamilies and tribes and their relationships is therefore anintricate task to solve.

To foster greater progress in resolving the evolutionary conun-drum of weevil relationships, a better integration of all possible datafrom molecular as well as morphological sources is needed to im-prove resolution and stability. An important strategy in this respectincludes the sampling of critical transitional taxa which have beenshown by e.g. Kuschel et al. (2000) to be crucial in establishing char-acter homology between more distantly related taxa. This studytherefore aims at combining the strengths of broad taxon and char-acter sampling, by including DNA sequences from four nuclear andone mitochondrial gene, constructing a large and unbiased matrixof 128 morphological characters coded for all three main develop-mental stages, and by including for the first time in a combinedphylogenetic analysis a range of putatively transitional taxa (seeSection 2). Perhaps the most enigmatic weevil group is Platypodinae,which has shown higher than average substitution rates and aphylogenetic position which is particularly ephemeral (Farrellet al., 2001; Jordal, 2007; Jordal et al., 2008; McKenna et al., 2009).We will therefore focus more strongly on this group by testing theirphylogenetic position with the inclusion or exclusion of taxa such asSchedlarius, Mecopelmus and Coptonotus which are all potentiallytransitional between Platypodinae and other weevils.

All recent and comprehensive phylogenetic analyses haveunambiguously placed Platypodinae and Scolytinae within Curcu-lionidae, with Brentidae as the sister group to all Curculionidae(Kuschel, 1995; Marvaldi, 1997; Marvaldi et al., 2002; McKennaet al., 2009). This seems apparently at odds with the fossil recordwhich indicate that Scolytinae is slightly older than other definedlineages within Curculionidae (Cognato and Grimaldi, 2009;Kirejtshuk et al., 2009). Until lately, the oldest reliable fossil usedto date the origin of Scolytinae was from early Tertiary time (Lon-don Clay 55 Ma, see also McKenna et al., 2009). However, mucholder cretaceous scolytine fossils were recently discovered and de-scribed from 100 Ma Lebanese and Burmese amber (Cognato andGrimaldi, 2009; Kirejtshuk et al., 2009). These recently discoveredfossils are older than all other known Curculionidae fossils (upperTuronian about 90 Ma, see Oberprieler et al., 2007 for detailed re-

view), which raise not only important questions about the age ofScolytinae and other weevils, but also the phylogenetic positionof Scolytinae. Nearly doubling the age of one such important cali-bration point likely involves substantial changes in node age alsofor other weevil lineages and we will provide new estimates usingan updated topology comprehensively sampled for putatively tran-sitional taxa. We will furthermore explore the effect on dating lin-eages using radically different fossil ages on a single calibrationpoint, to illuminate potential errors associated with biased fossilrecords.

2. Materials and methods

2.1. Taxon sampling

The classification system of Oberprieler and co-workers (2007)has been followed in this study (Table 1), in which the authors re-duce several subfamilies to tribal level, particularly so the manygroups now included in Baridinae (e.g. Conoderinae, Ceutorrhyn-chinae), Molytinae (e.g. Cryptorhynchinae) and use a broader con-cept of Brachycerinae (includes e.g. Erirhininae). We aimed atsampling the majority of taxa that have been proposed transitionalbetween Scolytinae and Platypodinae, and between Scolytinae andCossoninae, Molytinae (and Cryptorhynchini/-inae). In relation tothe enigmatic status of Platypodinae, we included Notoplatypus,Periommatus, Schedlarius and Mecopelmus. While the first two taxahave some minor morphological features deviating from the typi-cal Platypodinae, the last two show a mixture of platypodine andnon-platypodine features, which has led several authors to proposethat these taxa are transitional between Platypodinae and Scolyti-nae (Wood, 1973, 1993), with affinities to Cossoninae (Kuschelet al., 2000). With respect to the possible transition between Cos-soninae and Scolytinae (see Kuschel, 1966; Kuschel et al., 2000;Marvaldi, 1997; May, 1993), we included several genera of the barkweevils in Araucariini, which have lateral socketed spines on theirtibiae similar to many Scolytinae (Kuschel, 1966; Mecke, 2005).Coptonotus was retained in Platypodinae (Platypodidae: Coptonot-inae) by Wood (1993) irrespective of its previous transfer byThompson (1992) from Platypodinae to Scolytinae based on theshape of the male sternite VIII and the tarsus shape. This taxon alsoshows some affinities with Cossoninae and the genera in the scol-ytine tribe Scolytini in tibial structures; its phylogenetic position istherefore highly uncertain (Jordal and Oberprieler, 2011). Potentialaffinities between the cryptorhynch Psepholax and the cryptor-hynch-like scolytine genus Dactylipalpus were also tested due totheir similarly vestigial rostral channel and the lack of sclerolepidia(Zimmermann, 1994; Lyal et al., 2006). Similarly, the alternativepositions of Amorphocerini (Amorphocerus, Porthetes) were testedwith respect to their current placement in Molytinae (AlonsoZarazaga and Lyal, 1999), and previously in Cossoninae (Kuschel,1966). Furthermore did we include some recently discovered woodboring and subsocial Conoderini to assess their phylogeneticrelationship to other weevils.

Sample sizes were distributed relatively equally among thethree major wood boring groups Scolytinae (33 spp.), Platypodinae(20 spp.) and Cossoninae (20 spp.) and the other weevil ingroupspecies combined (32 spp.). The inclusion of more Scolytinaespecies relates to the uncertain classification of some taxa inScolytinae, e.g. Dactylipalpus, Phloeoborus, Scolytini (Scolytus,Camptocerus, Cnemonyx), Scolytoplatypus, and the oldest docu-mented fossil of Scolytinae – Microborus (see Cognato andGrimaldi, 2009). Outgroups were selected among Brentidae,Anthribidae and Attelabidae. Brentidae is unequivocally estab-lished as the sister group of Curculionidae (Kuschel, 1995;Marvaldi et al., 2002; McKenna et al., 2009).

Table 1Samples of 112 weevil taxa used in this study and their GenBank accession numbers. The classification scheme follows Oberprieler et al. (2007). Code refers to DNA specimenvoucher at Bergen Museum; if missing, data were taken from GenBank. Taxa with observed subcortical egg laying behaviour and parental care are marked with

pin left margin.

Family Tribe Species Code Country Locality CO1 EF-1A 28S CAD ARGK

AnthribidaeTribe? Indet. Anthrib01 South

AfricaEast Cape: Stutterheim, KologhaForest

HQ883607 – HQ883527 HQ883764 –

Tribe? Indet. Anthrib02 Cameroon Limbe, Ekonjo HQ883608 HQ883696 – HQ883765 HQ883841

AttelabidaeApoderini Apoderus coryli AtApo01 Russia Primorsky Krai, 60 km E.

VladivostokHQ883609 – HQ883528 – HQ883842

Apoderini Apoderus jekeli AtApo02 Russia Primorsky Krai, Anisimovka HQ883610 HQ883697 HQ883529 – HQ883843

BrentidaeApioninae Apion sp. A BrApi01 Norway N. Trøndelag: Sør-Gjeslingan HQ883612 HQ883698 HQ883531 HQ883767 –Apioninae Apion sp. B AY131096 AY131125 AY131067 – –pBrentinae Indet. BrBre01 Sarawak Bako NP HQ883613 HQ883699 HQ883532 – HQ883845pBrentinae Indet. BrBre02 Cameroon Limbe, Bonadikombe HQ883614 HQ883700 HQ883533 HQ883768 HQ883846

CurculionidaeBaridinae

Baridini Indet. BaXxx02 Russia Primorsky Krai, 20 km WAndreev

HQ883611 – HQ883530 HQ883766 HQ883844

pConoderini Homoeometamelus

sp.CsXxA01 Uganda Kibale National Forest, S. Fort

PortalHQ883643 HQ883723 HQ883558 HQ883785 HQ883872

pConoderini Campyloscelina –

Keibaris? sp. ACdCod01 South

AfricaEast Cape: Ecca Pass, N.Grahamstown

HQ883615 HQ883701 HQ883534 – HQ883847

pConoderini Campyloscelina,

indet., sp. BCdCod02 Cameroon Limbe, Ekonjo HQ883616 HQ883702 HQ883535 HQ883769 HQ883848

pConoderini Scolytoproctus sp. CdSpr01 South

AfricaEast Cape: Tsitsikamma NP HQ883618 HQ883704 HQ883537 HQ883770 HQ883850

Conoderini Indet. CdZyg01 Russia Primorsky Krai, Zonadvaroka HQ883619 HQ883705 HQ883538 – HQ883851Conoderini Metialma sp. CdMet01 South

AfricaEast Cape: Ecca Pass, N.Grahamstown

HQ883617 HQ883703 HQ883536 – HQ883849

Ceutorhynchini Rhinoncuspericarpius

CeRhi01 Norway Sogn&Fjordane: Gulen HQ883620 – HQ883539 HQ883771 HQ883852

Ceutorhynchini Zacladus affinis CeZac01 Norway NTr, Øksningøy HQ883621 HQ883706 HQ883540 HQ883772 HQ883853Lixini Larinus sp. ClLar01 Russia Primorsky Krai, 20 km W

AndreevHQ883622 HQ883707 HQ883541 HQ883773 HQ883854

Lixini Lixus sp. ClLix01 Russia Primorsky Krai, 20 km WAndreev

HQ883623 HQ883708 HQ883562 – HQ883855

BrachycerinaeErirhinini Penestes sp. AY131097 AY131126 AY131068 – –Erirhinini Tanysphyrus

lemnaeAY131098 AY131127 AY131069 – –

Erirhinini Himasthlophallusflagellifer

ErHim01 Russia Primorsky Krai, Anisimovka HQ883654 HQ883730 HQ883569 – –

CoptonotinaepCoptonotini Coptonotus

cyclopusCpCop01 Costa Rica Herida prov. HQ883624 – – HQ883774 HQ883856

Cossoninae HQ883625pAraucariini Araucarius major CsAru02 Argentina Neuquen AY040285 HQ883711 AF308350 – HQ883860pAraucariini Araucarius minor CsAru01 Argentina Neuquen AF375307 AF308346 AF308351 – HQ883859Araucariini Coptocorynus? sp.

ACsCpt01 Papua N.

GuineaBulolo HQ883630 HQ883713 HQ883545 – –

Araucariini Coptocorynus? sp.B

CsCpt02 Papua N.Guinea

Bulolo HQ883631 HQ883714 HQ883546 HQ883776 HQ883862

Araucariini Xenocnema? CsXen01 Australia Conondale NP, Gympie HQ883642 HQ883722 HQ883557 – –Cossonini Cossonus linearis CsCos02 USA HQ883629 HQ883712 HQ883544 – HQ883861Cossonini Mesites fusiformis CsMes01 Spain La Gomera EU191838 EU191870 HQ883549 HQ883778 HQ883865’Dobionus group’ Dobionus

araucarinusCsDob01 New

CaledoniaMt. Do HQ883632 – – – –

Onycholipini Pselactus sp. CsPse01 Madeira Madeira, Achadas las Cruz EU191839 EU191871 HQ883552 HQ883780 HQ883868Onycholipini indet. CsPse02 South

AfricaEC, Grahamstown HQ883637 HQ883718 HQ883553 HQ883781 HQ883869

Onycholipini indet. CsPse03 SouthAfrica

EC: Ecca Pass, N. Grahamstown HQ883638 HQ883719 HQ883554 HQ883782 HQ883870

pOnycholipini Pseudostenocelis

sp.CsPsc01 Papua N.

GuineaBulolo HQ883636 HQ883717 HQ883551 HQ883779 HQ883867

Pentarthrini Macroscytalus sp. CsMac01 Papua N.Guinea

Wau HQ883634 – HQ883548 HQ883777 HQ883864

Pentarthrini Microcossonus sp.A

CsMic01 Australia Good Nights Shrub, Gympie HQ883635 HQ883716 HQ883550 – HQ883866

Pentarthrini Microcossonus sp.B

CsRhy01 Australia Yarraman Forest, Gympie HQ883640 HQ883721 HQ883556 HQ883783 HQ883871

pRhyncolini Rhyncolus sp. A CsEla01 Papua N.

GuineaEddie Creek, Wau HQ883633 HQ883715 HQ883547 – HQ883863

710 B.H. Jordal et al. / Molecular Phylogenetics and Evolution 59 (2011) 708–724

Table 1 (continued)

Family Tribe Species Code Country Locality CO1 EF-1A 28S CAD ARGKp

Rhyncolini Rhyncolus sp. B CsRhi01 Papua N.Guinea

Mt. Kaindi, Wau HQ883639 HQ883720 HQ883555 – –

pRhyncolini Stenancylus sp. CsSte02 Argentina Salta, 7 km SW Gen. Enrique

MosconiHQ883641 AF375264 FJ867715 HQ883784 –

CurculioninaeTychiini Sibinia sp. CuSib01 South

AfricaWest Cape: Cederberg,Burgerskloof

HQ883649 HQ883725 HQ883563 – HQ883878

Indet Indet. CuXxx03 Uganda Budongo Forest HQ883650 HQ883726 HQ883564 – HQ883879

DryophthorinaeRhynchophorini Cactophagus

spinolaeAY131112 AY131141 AY131083 – –

Rhynchophorini Rhynchophoruspalmarum

AY131121 AY131150 AY131092 – –

Rhynchophorini Dynamis borassi AY131105 AY131134 AY131076 – –Sphenophorini Sphenophorus

venatusAY131117 AY131146 AY131088 – –

Litosomini Cosmopolitessorditus

AY131111 AY131140 AY131082 – –

Litosomini Sitophilus oryzae AY131099 AY131128 AY131070 – –

EntiminaeOtiorhynchini Otiorhynchus

auropunctatus?EnOti01 Norway Nordland: Øksningøy HQ883652 HQ883728 HQ883567 – HQ883883

Polydrusini Polydrususcervinus

EnPol01 Norway Telemark, Gvarv HQ883653 HQ883729 HQ883568 HQ883793 HQ883884

Sitonini Chlorophanussibiricus

EnChl01 Russia Primorsky Krai, Zonadvaroka HQ883651 HQ883727 HQ883566 – HQ883882

MolytinaeIndet Indet. MoXxx01 Cameroon Ekonjo, N. Limbe HQ883667 HQ883738 HQ883578 HQ883806 HQ883896Amorphocerini Amorphocerus

rufipesMoAmo01 South

AfricaEast Cape: Cathcart HQ883664 HQ883736 HQ883575 HQ883803 HQ883893

Amorphocerini Portheteshispidulus

MoPor01 SouthAfrica

Eeast Cape: Kokstad HQ883666 HQ883737 HQ883577 HQ883805 HQ883895

Cryptorhynchini Indet. Crh_sp02 Cameroon Limbe, Ekonjo HQ883628 HQ883710 HQ883543 – HQ883858Cryptorhynchini Indet. Crh_sp01 South

AfricaWest Cape: Nature’s Valley HQ883627 – – – –

Hylobiini Hylobius piceus MoHyl01 Norway Hordaland: Utåker HQ883665 – HQ883576 HQ883804 HQ883894Psepholacini Psepholax sp. Crh_Psx01 Papua N.

GuineaBulolo HQ883626 HQ883709 HQ883542 HQ883775 HQ883857

PlatypodinaepMecopelmini Mecopelmus zeteki MeMec01 Panama Colon: San Lorenzo Forest HQ883663 HQ883735 HQ883574 HQ883802 HQ883892pSchedlarini Schedlarius

mexicanusCpSch01 Mexico St. Catharina HQ883625 – – – –

pPlatypodini Crossotarsus

minusculusPlCro02 Papua N.

GuineaBeitabag, Madang HQ883669 HQ883739 HQ883579 HQ883809 HQ883899

pPlatypodini Doliopygus

rhodesianusPlDol01 Uganda Kibale, 10 km S. Fort Portal HQ883670 HQ883740 HQ883580 HQ883810 HQ883900

pPlatypodini Mesoplatypus sp. PlMes01 Ghana Ankasa HQ883671 HQ883741 HQ883581 HQ883811 HQ883901pPlatypodini Platypus

incompertus(PLP05)

PlPla05 Australia New South Wales: CumberlandState Forest, West Pennant Hills

HQ883673 AF375266 AF375298 HQ883813 HQ883903

pPlatypodini Platypus jansoni PlPla02 Papua N.

GuineaMadang, Beitata HQ883672 HQ883742 HQ883582 HQ883812 HQ883902

pPlatypodini Teloplatypus sp B PlTel02 Costa Rica HQ883674 HQ883743 HQ883583 HQ883814 HQ883904pPlatypodini Trachyostus

schaufussiPlTra01 Ghana Bokuro-Abaa HQ883675 HQ883744 HQ883584 HQ883815 –

pTesserocerini Cenocephalus sp. TsCen01 Costa Rica HQ883682 HQ883751 HQ883593 HQ883826 HQ883912pTesserocerini Chaetastus

montanusTsCha01 Uganda Kibale HQ883683 HQ883752 HQ883594 HQ883827 HQ883913

pTesserocerini Chaetastus

tuberculatusTsCha02 Cameroon Limbe, Ekonjo HQ883684 HQ883753 HQ883595 HQ883828 HQ883914

pTesserocerini Diapus unispineus TsDia02 Papua N.

GuineaMt. Kaindi, Wau HQ883685 HQ883754 HQ883596 HQ883829 HQ883915

pTesserocerini Genyocerus exilis

(TSD01)TsGen02 Sarawak Sarawak, Bako NP HQ883686 HQ883755 HQ883597 HQ883830 HQ883916

pTesserocerini Genyocerus

serratus (TSD02)TsGen03 Sarawak Sarawak, Bako NP HQ883687 HQ883756 HQ883598 HQ883831 HQ883917

pTesserocerini Notoplatypus

elongatusTsNot01 Australia NSW: Coffs Harbour HQ883688 HQ883757 HQ883599 HQ883832 HQ883918

pTesserocerini Periommatus sp. TsPer01 Ghana Kakum HQ883689 HQ883758 HQ883600 HQ883833 –pTesserocerini Tesserocerus

dewalqueiTsTes02 Costa Rica Heredia: La Selva Biol. St. HQ883692 HQ883761 HQ883603 HQ883835 HQ883921

pTesserocerini Tesserocerus

ericerusTsTes01 Costa Rica Heredia: La Selva Biol. St. HQ883691 HQ883760 HQ883602 HQ883834 HQ883920

(continued on next page)


Table 1 (continued)

Family Tribe Species Code Country Locality CO1 EF-1A 28S CAD ARGKp

Tesserocerini Tesserocranulusnevermanni

TsTcr01 Peru Loreto Prov., 20 km NW Iquito HQ883690 HQ883759 HQ883601 – HQ883919

ScolytinaepCtenophorini Gymnochilus

reitteriCtGym01 Costa Rica Castilla HQ883644 – EU090353 HQ883786 HQ883873

pCtenophorini Micoborus boops CtMic03 Cameroon Mt. Cameroon, S lope HQ883645 – HQ883560 HQ883788 HQ883874pCtenophorini Microborus boops? CtMic01B Madagascar HQ883646 HQ883724 HQ883559 HQ883787 –pCtenophorini Pycnarthrum

hispidumCtPyc01 Costa Rica Castilla HQ883647 – EU090352 HQ883789 HQ883875

pCtenophorini Pycnarthrum sp. CtPyc01_2 Mexico HQ883648 – HQ883561 – HQ883876pCtenophorini Scolytodes

acuminatusCtSct01 Costa Rica EU191844 EU191876 EU090351 HQ883790 HQ883877

pDryocoetini Dryocoetes

autographusDrDry01 Russia St. Petersburg AF444054 AF259830 HQ883565 HQ883791 HQ883880

pDryocoetini Xylocleptes

bispinusDrXyl01 Ukraine Crimea EU191848 EU191880 EU090347 HQ883792 HQ883881

pHylastini Hylastes opacus HtHyt05 Sweden Gotland HQ883660 HQ883732 HQ883927 HQ883799 –pHylastini Hylurgops

rugipennisHtHyg09 USA NC, Pisgah Nat. Forest HQ883659 AF308408 AF308364 HQ883798 HQ883889

pHylesinini Alniphagus

aspericollisHlAln01 USA Washington State EU191849 EU191881 AF308367 HQ883794 HQ883885

pHylesinini Dactylipalpus

grouvelleiHlDac01 Ghana Bokuro-Abaa HQ883656 HQ883731 HQ883570 HQ883795 HQ883886

pHylesinini Hylesinus varius HlHyl02 Sweden Gotland, E. Visby HQ883657 AF308409 AF308365 HQ883796 HQ883887pHylesinini Phloeoborus sp. HlPhb02 Guyana Iwokrama HQ883658 – HQ883571 HQ883797 HQ883888pIpini Ips acuminatus IpIps02 Norway Hjuksebø, Notodden HQ883661 HQ883733 HQ883572 HQ883800 HQ883890pIpini Pityogenes

bistridentatusIpPit01 Ukraine Crimea, Yalta HQ883662 HQ883734 HQ883573 HQ883801 HQ883891

pPhloeosinini Phloeosinus

punctatusPhPhl03 USA Washington State HQ883668 AF308405 AF308361 HQ883808 HQ883898

pPhloeosinini Pseudochramesus

acuteclavatusPhPch01 Argentina Salta, 7 km SW Gen. Enrique

MosconiAF375328 AF308404 AF308360 HQ883807 HQ883897

pPhloeotribini Phloeotribus

spinulosusPtPht01 Norway N. Trøndelag,: Lierne EU191862 EU191894 HQ883585 HQ883816 HQ883905

pPremnobiini Premnobius

cavipennisXyPre01 South

AfricaEeast Cape: Grahamstown HQ883694 HQ883762 HQ883605 HQ883839 HQ883925

pScolytini Camptocerus

aenipennisScCam02 Guyana Iwokrama NP HQ883676 HQ883745 HQ883587 HQ883818 HQ883907

pScolytini Camptocerus

auricomisScCam01 Costa Rica EU191864 EU191896 HQ883586 HQ883817 HQ883906

pScolytini Cnemonyx

vismiacolensScCne01 Guyana Iwokrama NP EU191865 EU191897 HQ883588 HQ883819 HQ883908

pScolytini Scolytus intricatus ScScl02 Sweden Smaaland: Oskarshamn HQ883677 HQ883746 HQ883589 HQ883820 HQ883909pScolytini Scolytus scolytus ScScl06 Denmark NEJ, Tofte Skov HQ883678 HQ883747 HQ883590 HQ883821 HQ883910pScolytoplatypodini Scolytoplatypus

africanusSpScp01 Uganda Kibale Forest (Fort Portal) EU191866 EU191898 AF308391 HQ883822 –

pScolytoplatypodini Scolytoplatypus

entomoidesSpScp04 Papua N.

GuineaMadang HQ883679 HQ883748 EU090345 HQ883823 –

pTomicini Dendroctonus

micansToDen01 Estonia Rohukula, Haapsalu HQ883680 HQ883749 HQ883591 HQ883824 –

pTomicini Tomicus piniperda ToTom01 Norway Oppland: Lom, Elveseter HQ883681 HQ883750 HQ883592 HQ883825 HQ883911pXyleborini (-a) Anisandrus dispar XyXyl02 Norway Oklungen, Porsgrunn HQ883695 HQ883763 HQ883606 HQ883840 HQ883926pXyloterini Indocryphalus

pubipennisXtInd01 Japan Kyushu, Mt. Tachibana HQ883693 AF375276 HQ883604 HQ883836 HQ883922

pXyloterini Trypodendron

lineatumXtTry01 Norway Trondheim AF187132 AF186682 AF308394 HQ883837 HQ883923

pXyloterini Xyloterinus politus XtXyl01 USA NH: Mt. Monadnock AF187133 AF186683 EU090334 HQ883838 HQ883924


2.2. Morphological character matrix

One hundred and twenty eight morphological characters wereincluded, with 29 larval and pupal characters and 99 adult char-acters, with 26 of these based on head features and 57 additionalexternal characters, and 16 internal features (Supplementaryinformation: Tables S1 and S2). DNA was extracted mainly fromspecimens used to code morphological features, with only somewell described internal characters based on published work (e.g.Calder, 1989; Morimoto, 1962; Thompson, 1992), particularly inthe literature on larval morphology (Browne, 1972; Lekander,1968; Marvaldi, 1997; May, 1993, 1997; Muskus-Arrieta andAtkinson, 1992).

2.3. PCR amplification and sequencing

DNA was extracted from legs or heads of the larger weevils, andfrom whole specimens of smaller individuals, keeping the ex-tracted specimen or part of body as voucher for further morpholog-ical studies. We amplified and sequenced fragments of themitochondrial gene cytochrome oxidase 1 (COI), the D2–D3 seg-ment of the nuclear large ribosomal subunit (28S), and the threenuclear protein encoding genes elongation factor 1a (EF-1a),arginine kinase (ArgK) and carbamoyl-phosphate synthase 2 –aspartate transcarbamylase – dihydroorotase (CAD).

PCR reactions were performed in a 25 ll volume containing0.2 lM of each primer (Table 2), 0.25 mM of each dNTP, 0.65 unit

Table 2Gene fragments targeted for PCR and the primers used. Sequencing primers were identical to those used in PCR.

Gene Primer name Annealing Primer sequence First cited

COI S1718 46 50-GGAGGATTTGGAAATTGATTAGTTCC-30 Simon et al. (1994)A2411 50-GCTAATCATCTAAAAACTTTAATTCCWGTWG-30 Simon et al. (1994)A2237 50-CCGAATGCTTCTTTTTTACCTCTTTCTTG-30 Normark et al. (1999)

28S S3690 55 50-GAGAGTTMAASAGTACGTGAAAC-30 Dowton and Austin (1998)A4394 50-TCGGAAGGAACCAGCTACTA-30 Whiting et al. (1997)A4285 50-CCTGACTTCGTCCTGACCAGGC-30 Sequeira et al. (2000)

EF-1a S149 58–44a 50-ATCGAGAAGTTCGAGAAGGAGGCYCARGAAATGGG-30 Normark et al. (1999)A1043 50-GTATATCCATTGGAAATTTGACCNGGRTGRTT-30 Normark et al. (1999)A754 50-CCACCAATTTTGTAGACATC-30 Normark et al. (1999)

ArgK ArgK forB2 52 50-GAYTCCGGWATYGGWATCTAYGCTCC-30 Dole et al. (2010)ArgK revB1 50-TCNGTRAGRCCCATWCGTCTC-30 This studyArgK LTrev2 50-GATKCCATCRTDCATYTCCTTSACRGC-30 Danforth et al. (2004)

CAD 581F2 50 50-GGWGGWCAAACWGCWYTMAAYTGYGG-30 Moulton and Wiegman (2004)CAD forB2 50-GARAARGTNGCNCCNAGTATGGC-30 Dole et al. (2010)CAD for4 50-TGGAARGARGTBGARTACGARGTGGTYCG-30 Danforth et al. (2004)CAD rev1mod 50-GCCATYRCYTCBCCYACRCTYTTCAT-30 Danforth et al. (2004)

a Touch down cycling (see Section 2).


of Qiagen HotStar taq� DNA polymerase, 1� buffer with MgCl2 to afinal concentration of 2.0 mM. PCR of COI and 28S used 1 ll DNAand the other PCR reactions 2 ll DNA. Typical PCR cycles consistedof 15 min initial heating to release the encapsulated enzyme,thereafter 40 cycles of denaturing at 94 �C, followed by 44–55 �Cannealing for 45–60 s, and 72 �C extension for 60 s, followed by afinal extension for 10 min. For EF-1a we used touch down cycling,starting with two cycles of 58 �C annealing temperature, steppingdown two degrees per two cycles reaching 44 �C with a final 26 cy-cles at this temperature. Sequencing reactions used PCR primers(Table 2), following Applied Biosystems recommended protocols.

2.4. Alignment and phylogenetic analyses

Sequences were blasted in GenBank and our local databases ofunpublished sequences to examine possibilities for paralogouscopies and other potential PCR errors (contamination, nematode/fungus DNA, pseudogenes). All protein encoded gene sequenceswere readily aligned by eye, with introns identified by insertionsdemarcated by the initiating GT and terminating AG motif. Thesewere excised before further analyses. Alignments of rDNA se-quences from the D2–D3 domains of the large ribosomal subunit28S were initially aligned by the MUSCLE software (Edgar, 2004)using default parameters. The resulting alignment was thereafteradjusted according to a secondary structure model for Scolytinae(Jordal et al., 2008) where indel-rich unalignable regions werepruned from the matrix. Two alignments were used in the phylo-genetic analyses: one that included indels with likely homology as-sessed according to the secondary structure model (778 basepairs); and one most conservative alignment with all ambiguouspositions removed (407 base pairs). Each of these two alignmentswas used in combination with the remaining data in data sets I–III (see below).

Phylogenetic trees were reconstructed in a Bayesian frameworkusing MrBayes 3.1.2 (Huelsenbeck and Ronquist, 2001) and by par-simony searches in PAUP⁄ (Swofford, 2002). Three different datasets were used: data set I included all nucleotides and morpholog-ical data; data set II excluded 3rd positions from the protein encod-ing genes; data set III had four protein encoding genes translatedinto amino acid data. Each data set was used in combination withthe longer and shorter version of the 28S alignment. For the Bayes-ian analyses we selected the best model for each partition usingMrModeltest (Nylander, 2004). We partitioned the nucleotide databy genome (mtDNA vs. nDNA) and nucleotide position. Studies

ranging from vertebrates (e.g. Li et al., 2008) to insects (e.g. Milleret al., 2009) have shown that applying a common model for eachcodon position across different loci in the same genome have themost optimal fit to the data. Data set I was therefore divided intoeight partitions: morphology, COI 1st, 2nd, and 3rd positions, allprotein encoding genes 1st position, 2nd position and 3rd position,and 28S. Most partitions had a GTR + I + C model selected by AIC,with COI 1st and 2nd positions each selected a GTR + C model.To investigate the influence of data partitioning we also analysedthe data with each gene and morphology as partitions, with aGTR + I + C model selected by MrModeltest for each molecular par-tition (six partitions). Data set II which excluded 3rd positions of allprotein encoding genes thus had six partitions (COI 1st, COI 2nd,nDNA 1st, nDNA 2nd, rDNA, morphology). Data set III, which usedamino acid translation of the protein coding genes, used a mixedevolutionary model estimated for each gene during the run. COIestimated the MtRev model (posterior probability (p.p.) = 1.0),EF-1a and CAD estimated the Jones model (p.p. = 1.0 and 0.91,respectively), and the ArgK data estimated a Dayhoff model(p.p. = 1.0), as the most optimal to explain amino acid variation.Morphological data were analysed with a gamma-distributed rateof character changes in all analyses. Up to 10 million generationswere run with sampling every 1000 generation. The level of con-vergence from two parallel runs were analysed by the sump com-mand in MrBayes and each analysis was repeated once to verifyconvergence across different searches. Burn-in level was basedon visual stationarity in likelihoods for parallel runs. The lengthof runs was based on the standard deviation of split frequenciesreaching below 0.05, and the average potential scale reduction fac-tor (PSRF) value was not higher than 1.01 in any analysis.

Parsimony analyses consisted of 2000 heuristic searches with30 random additions and TBR swapping for each search. Node sup-port was estimated by 200 pseudoreplicates of 20 random additionreplicates each. In the bootstrap analysis of the morphological data,the number of stored trees was set to 10.000. Alternative topolo-gies were reconstructed with all wood boring and potentially sub-social behaviour constrained (Scolytinae, Platypodinae, Coptonotus,Campyloscelina, Cossoninae). Differences in tree length weretested by the Kishino–Hasegawa test and by Templeton’s test asimplemented in PAUP⁄ (Swofford, 2002).

2.4.1. Pruning putatively transitional taxaTo explore the effect of taxon sampling with particular rele-

vance to the enigmatic position of Platypodinae, several putatively


transitional taxa (see Section 2.1) were removed in subsequentBayesian analyses of data sets I–III. A total of 36 analyses weremade, 6 for each of the data sets using the longer and shorter28S alignment. These analyses were carried out on a TITAN clusterof 300 computer cores freely available online at the Bioportal(www.bioportal.uio.no). Topological changes between analyseswith different combinations of taxa indicate that particular taxamay be critical to the phylogenetic results, given the data. Eachof Schedlarius mexicanus, Mecopelmus zeteki and Coptonotus cycl-opus was removed in various combinations from the analyses, withor without the six species of Dryophthorinae, a weevil subfamilythat has been proposed by some authors as sister lineage to Platy-podinae (Marvaldi, 1997; McKenna et al., 2009).

2.4.2. Ancestral state reconstructionWood boring and the number of origins of subsocial behaviour

was traced in Mesquite v. 2.74 (Maddison and Maddison, 2010),using the likelihood criterion. A simple Markov k-1 model (Mk1)was used which implies that any particular change between statesis equally probable. Hence, we do not assume that lineages consist-ing of wood boring species with parental care necessarily origi-nated from an ancestor with larval development in wood. Thetaxa included were coded in three categories: 0, larvae and adultsfeed on plants and seeds (true herbivores, no parental care); 1, lar-vae develop in dead wood (adults oviposit externally, no parentalcare); 2, complete entire lifecycle in dead wood (parental care).Only taxa placed in the last category are truly subsocial withparental care.

2.5. Estimating divergence times

We estimated divergence times in the software BEAST(Drummond and Rambaut, 2007) using a constraint tree from ourBayesian phylogenetic analysis of data set III. This tree was pre-ferred over those from data sets I and II as the most conservativeestimate of phylogeny based on the data with less homoplasy(higher CI and RI values). BEAST integrates phylogeny and topolog-ical uncertainty in their estimates, but cannot handle morphologi-cal data which had substantial influence on the resulting topologyfrom our data. Input files were generated in BEAUti (part of theBEAST package) with a user-defined starting tree and topologykept constant by removing all operators that act on the ‘treeModel’(arrowExchange, wideExchange, wilsonBalding, subtreeSlide).Each analysis was run for 50 million generations, deleting the first5 millions as burnin.

Eight priors were used based on a well documented fossil re-cord, indicating minimum ages for these nodes. Calibration 1,Brentidae: the oldest known fossils of Brentidae are from themid-Cretaceous Santana formation, about 112 Ma old (Gratshevand Zherikhin, 2003). Calibration 2, Scolytinae: Previous fossils usedto date the origin of Scolytinae is a fossil from London clay, 55 Ma(Britton, 1960). Two recently discovered fossils were both indepen-dently dated to be 100 Ma (Cognato and Grimaldi, 2009; Kirejtshuket al., 2009), which provide a reliable fossil record for this group.We included additional analyses using the 55 myr old fossil toexamine how large effect a much older fossil may have on a singlecalibration point. Calibration 3, Dryophthorinae: the oldest knownfossil from this group is from the Florrisant bed, 34 Ma (Scudder,1893). Calibration 4, Platypodini: the oldest known fossil of Platypo-dinae is a typical Platypus and thus the Appenninian age (33 Ma) ofthis fossil should apply to the tribe Platypodini only, especially be-cause Platypus and related genera are recently derived within thisgroup (see Figs. 1–4). Calibration 5, Polydrusus: The Entiminaegenus Polydrusus was well established 35 Ma ago and the split be-tween this genus and Chlorophanus is set to this minimum. Calibra-tion 6, Cenocephalus: this Neotropical genus was well established

with many species in Dominican amber some 30 Ma (Bright andPoinar jr, 1994) and indicate the minimum age for a split withthe closely related African genus Chaetastus. Calibration 7, Dryocoe-tes: the presence of close relatives of Dryocoetes and the absence ofXyleborus in Dominican amber (Bright and Poinar jr, 1994) placesthe age of 30 Ma for the Dryocoetes-Xyleborus split. Calibration 8,Hylastes and Hylurgops: both genera were present in Baltic amber,45 Ma (reviewed in Larsson (1978)), indicating a minimum age fortheir divergence.

To assess how the fossil record may deviate from our molecularassessment of node age, we re-estimated age of calibration pointsby excluding one calibration point each time, allowing the nodeage to be estimated freely based on the seven remaining fossil cal-ibration points. Additionally, we performed one last analysis whereall previously constrained nodes were allowed to estimate withoutpriors but keeping only calibration 1 (Brentidae).

3. Results

3.1. Sequence properties and character performance

Among the four protein encoding genes, CAD and EF-1a con-tained a single intron in the sequenced region in most taxa. InCAD, 24 of 50 sequences contained an intron beginning at base96 downstream of primer cadBfor2 (Table 2). The intron variedin length from 51 to 143 base pairs, but was generally absent inAnthribidae, Brentidae and in Platypodinae (including Mecopel-mus). It was furthermore absent in the scolytine tribe Scolytini(Camptocerus and Cnemonyx) and in Scolytodes acuminatus, and intwo cossonine taxa (Macroscytalus and an Onycholipini genus closeto Pselactus). For EF-1a, most species contained an intron at the753/754 boundary as described in Jordal (2002). This intron wasabsent in only 3 out of the 93 successfully amplified sequences:two species of Scolytoplatypus and one Camptocerus (all inScolytinae).

A total of 3801 nucleotides and 128 morphological characterswere included in one alignment of 3929 characters (Table 3). Thehighest proportion of parsimony informative characters was foundin the morphological data (100%), followed by similar values forCOI (52%), CAD (50%) and 28S (50%) and slightly lower proportionsfor ArgK (45%) and EF-1a (41%). All third positions in the proteinencoding genes were variable and all but one or two sites wereparsimony informative (Table 3). The consistency index for the var-ious partitions showed the highest value for the morphologicaldata (0.25) and 28S (0.18), and a generally higher value for firstand second positions than for 3rd positions in the coding genes.COI showed the highest level of homoplasy measured by boththe consistency (CI) and the retention index (RI). When proteinencoding genes were translated to amino acids, the most variablegenes COI and CAD also had the highest CI and RI values. The short-er alignment of 28S resulted in a lower proportion of parsimonyinformative characters (37%) and increased CI and RI values (opti-mised on the topology resulting from data set II).

3.2. Phylogenetic analyses

3.2.1. Data set IThe Bayesian analysis of all nucleotides and morphological

characters (eight partitions, long alignment) resulted in a mono-phyletic Curculionidae as sister group to a monophyletic Brentidae(Fig. 1). A grade of brachycerine taxa and a monophyletic Dryoph-thorinae formed the most basal groups within Curculionidae, how-ever with low posterior probability separating the remaining of theingroup taxa. A monophyletic Entiminae was separated from fur-ther ingroup taxa with p.p. of 0.98. Cossoninae was monophyletic,

http://www.bioportal.uio.no

Fig. 1. Phylogenetic tree resulting from the Bayesian analysis of data set I, including all nucleotides and morphological data, using the longer 28S alignment and eightpartitions (nucleotide position per genome, rDNA and morphology). PDRF = 1.0, st. dev. = 0.021. Posterior probability indicated above nodes (�� – 100; � – >95). Hatch marks onnodes indicate taxa with confirmed observations on parental care.


Fig. 2. Phylogenetic tree resulting from the Bayesian analysis of data set III, including four amino acid translated gene fragments, rDNA nucleotides (short alignment) andmorphological data, using a mixed model for direct estimation of amino acid substitutions per gene, a GTR + G + I model for rDNA, and a gamma-distributed rate model for themorphological data. PDRF = 1.01, st. dev. = 0.039. Posterior probability indicated above nodes (�� – 100; � – >95). Hatch marks indicate taxa with confirmed observations onparental care.


Fig. 3. Ancestral state reconstruction of wood boring behaviour in weevils, calculated by likelihood estimation with the Mk-1 model over tree topology 1 (Fig. 1). Only thosetaxa with confirmed observations of parental care are marked in black, those with uncertain status in parental care and those with external mating and oviposition in deadwood marked in green.


with a monophyletic Araucariini forming the sister group to theremaining cossonines with maximum probability. Psepholaxgrouped with other cryptorhynchs in a monophyletic Cryptorhyn-chini forming the sister group to Cossoninae. Conoderini wasmonophyletic with maximum probability and formed an unre-solved sister relationship to a clade consisting of Scolytinae, Platy-podinae and Coptonotus – with the latter taxon as sister to the firsttwo (p.p. = 1.0). Scolytinae was paraphyletic with respect to amonophyletic Platypodinae. Within the latter, Mecopelmus, Sched-larius and Notoplatypus formed successively more inclusive sistergroups to the remaining platypodines, with Tesserocerini paraphy-letic with respect to Platypodini. Analyses of data set I using theshorter 28S alignment resulted in many of the same relationships,mainly differing in Cossoninae and Conoderini changing position,with Cossoninae as sister to a clade consisting of Scolytinae, Cop-tonotus and Platypodinae (Table 4).

Partitioning the data by gene and morphology (six partitions)revealed nearly the same topology as in the eight partitions analy-sis (not shown). The main difference was the shift of Coptonotus,which formed an unresolved sister group to the Araucariini andthe remaining Cossoninae (p.p. = 0.96).

Parsimony searches resulted in a much less resolved topologywith paraphyletic Cossoninae, paraphyletic Platypodinae with re-spect to Mecopelmus, and a non-monophyletic Scolytinae plus Platy-podinae (not shown). Conoderini was furthermore polyphyleticwith respect to Coptonotus. Using the shorter 28S alignment hadmarginal influence on the topology compared to the Bayesian anal-ysis, with a monophyletic Platypodinae nested within Scolytinae.

3.2.2. Data set IIThe Bayesian analysis of morphological and nucleotide data,

with 3rd positions of the four protein encoding genes excluded(six partitions), resulted in a strongly supported Curculionidae,

with Brachycerinae and Entiminae forming successively moreinclusive sister groups to the remaining ingroup taxa (topologynot shown). Further changes from the previous Bayesian analysesincluded Cossoninae as sister to a clade consisting of Scolytinae,Platypodinae and Coptonotus, however with minimal posteriorprobability. Platypodinae was similarly nested within Scolytinae,but the successive sister relationships of Scolytini and Microboruswith respect to Platypodinae contained low posterior probability(p.p. 0.87 and 0.82, respectively). Analysis using the shorter 28Salignment showed similar results, with Cossoninae as a paraphy-letic grade forming successive sister groups to Scolytinae andPlatypodinae. The nested position of Platypodinae within Scolyti-nae was only weakly supported by a posterior probability 0.62–0.64 for the two nested nodes.

Parsimony analysis revealed the same relationships betweenScolytinae and Platypodinae (BS = 63) with a sister relationshipto a partial Cossoninae that included Coptonotus, but excludedAraucariini. Araucariini formed an unsupported sister relationshipto Baridinae, including Conoderini. Analysis using the shorter 28Salignment differed mainly at basal nodes being Brachycerinaeand Entiminae, and by the paraphyletic Cryptorhynchini (Table 4).

3.2.3. Data set IIIBayesian analysis of amino acid translated data, the shortest

alignment of 28S nucleotides and morphological data resulted ina strongly supported Curculionidae which included several basalnodes with maximum posterior probability (Fig. 2). The first splitoccurred between a monophyletic Brachycerinae and remainingtaxa, followed by Entiminae. Cossoninae was monophyletic andformed the sister group to a subset of molytine taxa, however withmoderate posterior probability. Followed by several weakly sup-ported relationships, the wood boring taxa of Conoderini (Campy-loscelina) formed the sister group to Scolytinae and Platypodinae

0

20

40

60

80

100

120

140

160

Curculionidae

Plat+Scol+Cop

Cossoninae

PlatypodinaeAraucariini

Nod

e ag

e (M

yr)

Scolytinae 100 AScolytinae 100 BScolytinae 55 AScolytinae 55 BScolytinae excludedBrentidae only

0

20

40

60

80

100

120

140

160

180

BrentidaeScolytinae

Dryophthorinae

PolydrususPlatypodini

Cenocephalus

DryocoetesHylastini

Nod

e ag

e (M

yr)

fossil ageScolytinae 100Scolytinae 55

Fig. 4. Node age estimates. Above: estimation of node ages for nodes with known associated fossils, estimated by successively excluding one fossil date for each search.Below: estimation of nodes with unknown fossil age, based on all fossils with Scolytinae fossil set to 55 or 100 Ma, or excluding Scolytinae altogether, or including only theBrentidae fossil age. Estimates including all fossils were run twice (A and B) to indicate variation between analyses.


(p.p. = 0.95), with Coptonotus as weakly supported sister group(p.p. = 0.75) to a monophyletic Scolytinae (p.p. = 0.92). The sisterrelationship between Platypodinae and a monophyletic Scolytinaeplus Coptonotus obtained maximum posterior probability. WithinScolytinae, Scolytini formed the sister group to the remaining Sco-lytinae (p.p. = 0.92). Platypodinae showed generally the sametopology as in the other analyses. Analysis using the longer 28Salignment produced a slightly different topology, with a paraphy-letic Cossoninae forming a grade of successively closer sistergroups to the reciprocally monophyletic Platypodinae andScolytinae.

Parsimony analysis revealed a sister relationship between thereciprocally monophyletic Scolytinae and Platypodinae, with Cop-

tonotus as their sister taxon. Cossoninae was paraphyletic andformed a grade of successive sister groups to the previous men-tioned clades, with Araucariini as the more basal group. A mono-phyletic Cryptorhynchini included Psepholax. Also in this analysisthe Brachycerinae and Entiminae were found in the most basalpositions of a monophyletic Curculionidae (BS = 89). Node supportwas generally very low in the parsimony analysis of these data (Ta-ble 4). Analysis using the longer 28S alignment resulted in a nearlyidentical topology, differing only by a paraphyletic Brachycerinae.

3.2.4. Morphological data onlyThe Bayesian analysis of the morphological data provided a

topology with little resolution between subfamilies and tribes

Table 3Properties of the data included in the parsimony analysis of all nucleotides andmorphology (data set I) and amino acid translation of four genes, rDNA nucleotidesand morphology (data set III). PI, parsimony informative characters; % PI, inproportion to total number of characters; CI, consistency index (informativecharacters only); RI, retention index. Indices for data set I and III were measuredover tree topologies shown in Figs. 1 and 2, respectively.

Partition Characters PI % PI CI RI

Data set I: ‘nucleotides’Morphology 128 128 1.00 0.25 0.81COI 690 361 0.52 0.10 0.27

1st pos 230 92 0.40 0.13 0.392nd pos 230 40 0.17 0.30 0.573rd pos 230 229 0.99 0.08 0.22

EF-1a 857 353 0.41 0.13 0.341st pos 286 53 0.19 0.16 0.382nd pos 286 28 0.10 0.21 0.393rd pos 285 272 0.96 0.12 0.33

ArgK 801 362 0.45 0.13 0.401st pos 267 71 0.27 0.18 0.452nd pos 267 30 0.11 0.24 0.513rd pos 267 261 0.98 0.11 0.39

CAD 675 337 0.50 0.12 0.381st pos 225 78 0.35 0.18 0.482nd pos 225 38 0.17 0.29 0.513rd pos 225 221 0.98 0.11 0.36

28S ‘long’ 778 391 0.50 0.18 0.48

Total 3929 1932 0.49 0.13 0.39

Data set III: ‘ amino acids’Morphology 128 128 1.00 0.26 0.82COI 230 77 0.33 0.27 0.51EF-1a 285 46 0.16 0.22 0.33ArgK 267 62 0.23 0.24 0.43CAD 225 63 0.28 0.36 0.5028S ‘short’ 407 149 0.37 0.22 0.49

Total 1542 525 0.34 0.26 0.60


(Supplementary information: Fig. S1). Platypodinae was monophy-letic but unresolved with respect to Scolytinae. These two taxagrouped by maximum posterior probability, with Conoderini form-ing a grade of successively distant sister lineages. Cossoninae in-cluded Coptonotus with high probability (p.p. = 1.0), the latternested within Araucariini, but with low probability value. The clos-est relatives to the Cossoninae according to these data were theAmorphocerini (p.p. = 0.92).

Table 4Bootstrap support and posterior probability for selected groups of weevils. Abbreviations: I,P denote Bayesian and parsimony analyses, respectively. Endash (–) implies nodes not pre

Clade I

L S

B P B

Curculionidae 1.00 74 0.98Brachycerinae – – –Curculionidae, ex Brachycerinae – – 0.99Curculionidae, ex Brachycerinae & Entiminae – – 0.55Curculionidae, ex Brachycerinae & Dryophthorinae 0.80 <50 –Entiminae 1.00 94 1.00Dryophthorinae 1.00 100 1.00Molytinae – – –Cryptorrhynchini 1.00 52 1.00Conoderini 1.00 – 1.00Scolytinae – – –Scolytinae + Coptonotus – – –Scolytinae + Platypodinae 0.99 – 1.00Platypodinae 1.00 – 1.00Platypodinae + Scolytinae + Coptonotus 1.00 – 1.00Platypodinae + Scolytinae + Coptonotus + Conoderini <0.50 – –Platypodinae + Scolytinae + Coptonotus + Cossoninae – – 1.00Cossoninae 1.00 – 1.00Araucariini 1.00 <50 1.00

Parsimony analysis resulted in more than 10,000 most parsimo-nious trees of limited resolution. The consensus topology con-firmed most of the Bayesian results, with a monophyleticCurculionidae (BP = 92). A few noteworthy exceptions from theBayesian analysis included Dryophthorinae as the basal sistergroup to all other Curculionidae. Platypodinae was nested withina paraphyletic Scolytinae; however, the nested position was notsupported in the bootstrap analysis. The clade consisting of Scolyt-inae and Platypodinae received a bootstrap value of 69, in whichPlatypodinae was monophyletic (BP = 97).

3.2.5. Amino acid data onlyAt the subfamily level, Brachycerinae, Dryophthorinae, Entimi-

nae and the core Platypodinae (excluding Schedlarius and Mecopel-mus) were the only monophyletic subfamilies in the Bayesiananalysis, and Scolytinae nearly so with one Baridinae species in-cluded (Supplementary information: Fig. S2). The position of Cop-tonotus was unresolved with respect to Scolytinae. All othersubfamilies were paraphyletic. Brachycerinae and Platypodinaeformed two successively basal sister groups to the remaining Cur-culionidae taxa and the node separating the core Platypodinaefrom the remaining taxa received maximum posterior probability.We note that Mecopelmus and Schedlarius were both nested withinthe remaining Curculionidae (p.p. = 1.0). The parsimony topologywas similarly unresolved for many taxa.

3.2.6. Pruning transitional taxaExclusion of taxa had generally a significant influence on tree

topology in analyses of all three data sets (Supplementary informa-tion: Table S4). The most apparent changes in topology were re-lated to the position of Platypodinae which in data set I and IImoved from a nested position in Scolytinae to a reciprocally mono-phyletic relationship with Scolytinae in two cases, and to a sisterrelationship with Dryophthorinae in nine cases, while it remainedunchanged in 13 cases (see Supplementary information: Fig. S3). Infour cases, Platypodinae and Dryophthorinae formed a clade sub-tending the node between Brachycerinae and Entiminae. In dataset III, Platypodinae moved from a sister group position with Sco-lytinae to form a sister relationship with Dryophthorinae in 5 outof 12 taxon exclusion analyses, and moved to a basal position be-tween the nodes subtending Brachycerinae and Entiminae in 4 of

II and III indicate data sets I–III; S and L denote short and long alignment of 28S; B andsent.

II III

L S L S

P B P B P B P B P

<50 1.00 96 1.00 85 1.00 98 1.00 89– 1.00 – – – – – 0.84 62

<50 1.00 – 0.80 <50 1.00 <50 1.00 <50<50 1.00 – 1.00 <50 1.00 <50 1.00 –

– – <50 – – – – – –96 1.00 99 1.00 98 1.00 97 1.00 9696 1.00 95 1.00 97 1.00 99 1.00 99

– – <50 – – – – – –<50 0.63 <50 0.79 – – – – <50

– 0.99 <50 0.99 <50 0.99 63 – <50– – – – – 0.52 <50 0.92 <50– – – – – – – 0.75 –

<50 1.00 63 1.00 <50 1.00 71 – 59<50 1.00 <50 1.00 <50 1.00 <50 1.00 68

– 1.00 – 0.99 – 1.00 <50 1.00 <50– 0.50 – – – – – 0.91 –– – – 1.00 – 0.99 – – –– <0.50 – – – – – 0.95 –

54 1.00 <50 1.00 64 1.00 62 0.99 79


the 6 cases where Dryophthorinae were excluded. Overall for datasets I–III, in 11 of 14 cases where Platypodinae formed a sistergroup relationship with Dryophthorinae, a reversal to a Platypodi-nae–Scolytinae relationship occurred after excluding alsoDryophthorinae.

3.3. Ancestral state reconstruction of subsociality

The various tree topologies reported resulted in at least five ori-gins of subsociality in the ingroup, with multiple origins in Cosson-inae (Fig. 3). The lowest number of transitions from no care toparental care was found in the topology based on combined aminoacid translated data for four genes, rDNA and morphology (Fig. 2).In this topology the transition to subsociality at the node joiningCampyloscelina and the Platypodinae–Scolytinae clade obtaineda likelihood score of 0.97.

If we assume that most, if not all, cossonines have parental care,the number of origins is reduced to 2 or 3. The constrained mono-phyly of Cossoninae, Platypodinae, Coptonotus and the four subso-cial taxa of Campyloscelina results in 43 extra steps (data set I) or 6extra steps (data set II) in the parsimony analysis. Their monophylycould not be rejected by the KH-test (data set I, P = 0.18; data set II,P = 0.67) or the Templeton test (data set I, P = 0.17; data set II,P = 0.73, respectively).

3.4. Dating of nodes

We used the most conservative phylogenetic estimate for dat-ing nodes, resulting from the Bayesian analysis of morphologicaldata, amino acid coded protein genes, and the short unambiguousalignment of rDNA nucleotides (Fig. 2).

The fit between fossil dates and independent assessment ofnode age were generally in close agreement, depending on thechoice of Scolytinae fossil age (Fig. 4; Supplementary information:Table S3). Using the oldest 100 Ma old scolytine fossil, the age esti-mates for Brentidae and Dryophthorinae were much older thanwhat their currently oldest fossils implies. When the younger scol-ytine fossil was used, the difference between fossil age and nodeestimates were closer, but underestimated the age of most fossilcalibration nodes.

Five additional nodes were estimated using the two alternativeScolytinae fossil dates, set either to 100 or 55 Ma, and compared toestimates with Scolytinae fossils excluded, or including onlyBrentidae as a single calibration point. The oldest Scolytinae fossilled to consistently older estimates for all nodes, while using theyounger fossil resulted in consistently younger estimates com-pared to estimates with Scolytinae fossils excluded (Fig. 4). Brenti-dae alone estimated ages very similar to estimates based on allexcept Scolytinae fossils. For all estimates, using the oldest knownScolytinae fossil, the age of Curculionidae ranged from 108 to134 Ma, the node subtending Scolytinae, Platypodinae and Copton-otus from 100 to 105 Ma, Cossoninae from 87 to 102 Ma and Platy-podinae from 72 to 78 Ma.

4. Discussion

4.1. Phylogeny of wood boring weevils

This study clearly shows that Scolytinae and Platypodinae arederived lineages within Curculionidae and thus reject the hypoth-esis proposed by Wood (1973, 1986) and Morimoto and Kojima(2003) that these taxa evolved before the origin of the advancedweevils. A derived position is supported by morphological andmolecular data analysed both separately and in combination, andthus, corroborate recent phylogenetic studies based on morpholog-

ical data (Kuschel, 1995; Marvaldi, 1997), molecular data(McKenna et al., 2009) or both (Marvaldi et al., 2002). These previ-ous studies, and the one reported here, included overall a largeamount of data from different genetic loci, and morphologicalcharacters taken from all life stages and body parts. One can there-fore firmly conclude that Scolytinae and Platypodinae belong to theadvanced weevils.

Whether or not Platypodinae is nested within, or as the sistergroup to, or being more distantly related to Scolytinae, is stillambiguous given their ephemeral position across the various anal-yses. This was particularly evident from the taxon exclusion anal-yses where Platypodinae sometimes moved from a sister to ornested position in Scolytinae to a more basal position in Curculion-idae. In many cases the core Platypodinae (excluding Mecopelmusand Schedlarius) grouped with Dryophthorinae with high supportin a similar manner to a recent study based on molecular dataalone (McKenna et al., 2009). However, the exclusion of Dryoph-thorinae frequently reversed the position of Platypodinae, furtheremphasising the sensitivity of taxon composition. Such changesin topology point out the importance of taxon sampling and previ-ous studies have demonstrated the potentially large influence ontree topology by this parameter (e.g. Pick et al., 2010; Pollocket al., 2002). Although comprehensive taxon sampling is not al-ways critical for successful resolution of a phylogeny (e.g. Jordaland Hewitt, 2004; Rosenberg and Kumar, 2001), this parameterapparently becomes much more important when missing taxaare also basal members of long branched clades (Pick et al.,2010). Indeed, Platypodinae had the longest branches of all ingrouptaxa and it is therefore quite likely that this taxon is more affectedby biased taxon sampling than other weevil taxa. Given the broadand evenly distributed taxon sampling in our study, the suggestedsister relationship to Scolytinae therefore seems more likely than abasal platypodine position potentially generated by missing data.

A close relationship between Scolytinae and Platypodinaehinges largely on morphological characters (see also Kuschelet al., 2000; Marvaldi et al., 2002), in particular from adult headstructures (e.g. Morimoto and Kojima, 2003). Parallel evolution ofsuch adaptive traits is a likely outcome of tunnelling in woodand considerable doubts therefore exists about the homology ofhead characters (see e.g. Lyal, 1995). Reduction in rostrum lengthis observed in several unrelated weevil groups such as Conoderin-i-Campyloscelina, Cossoninae (several tribes) and Cryptorhynchini(especially Psepholacina-Psepholax), suggesting that Platypodinaeand Scolytinae head structures cannot alone support a sister rela-tionship. Characters from other body parts and life stages suggestdifferent relationships, and a more basal position in the neighbour-hood of Dryophthorinae is concordant with e.g. larval characters(Marvaldi, 1997), but also some molecular data (McKenna et al.,2009). Examination of the primitive platypodine taxa Notoplatypus,Tesserocerus and Schedlarius shows that their apodemes of the malegenitalia are straight and thus well developed (Jordal, 2011), whichmay indicate affinities to the orthocerous weevils. The knownorthocerous weevils included in this study, Brachycerinae (Erirhi-nini) and Entiminae, branched off early in the resulting topologiesas expected from current classification (Oberprieler et al., 2007;McKenna et al., 2009).

Although a close relationship to Scolytinae could not be firmlysupported with the current data, the combined molecular and mor-phological data did favour either a nested position or a sister rela-tionship between Platypodinae and Scolytinae when a broadestpossible sample of taxa and characters was included. We are in-clined to prefer analyses based on the largest possible sampleand to those analyses that give less weight to synonymous substi-tutions. Saturation by excessive back mutations is a prevalentfeature associated with ancient divergence such as the mid-Creta-ceous origin of the advanced weevils (Fig. 5), providing ample time

Fig. 5. Weevil egg galleries. Left, Ctonoxylon (Scolytinae: Xyloctonini) female moving egg into egg niche. Right, male and female Homoeometamelus (Baridinae: Conoderini).Arrow points to eggs laid in egg niches and packed with boring frass.


for extensive saturation at synonymous sites (Jordal, 2007). Wenote that the support for a nested Platypodinae within Scolytinaewas much reduced with the exclusion of nucleotides in third posi-tion, and furthermore supported a sister relationship when thesesame data were translated into amino acids that evolve at a muchslower rate.

A sister relationship between Scolytinae and Platypodinae alsoconcurs with several recent phylogenetic studies (Kuschel, 1995;Kuschel et al., 2000; Marvaldi et al., 2002) and thus excludes theprevalent hypothesis that Cossoninae represents a putative ances-tor that gave rise to the bark beetle behaviour characteristic forScolytinae (Lawrence and Newton, 1995; Marvaldi, 1997; Sequeiraet al., 2001). On the other hand, they may be related to a clade con-sisting of Platypodinae and Coptonotus in addition to Scolytinae, assuggested by three of the six Bayesian analyses using alternativerDNA alignments (see Table 4). We can therefore not exclude thepossibility that Cossoninae forms a very old lineage associatedwith the origin of very primitive wood boring behaviour. The coss-onine hypothesis was originally motivated by a typical bark beetlebehaviour observed in species of the cossonine tribe Araucariini(Kuschel, 1966) that have true socketed spines along the tibialmargins similar to Scolytinae (Mecke, 2005). However, socketedspines and tunnelling in bark and wood are not unique featuresfor these two groups of weevils, and is known also from other Cos-soninae, and in some little known bark and wood boring Conode-rini (Campyloscelina: Fig. 5). There are also scolytine generalacking socketed spines altogether, such as Scolytus and Scolyto-platypus, which emphasises the homoplastic nature of charactersassociated with adaptations to burrowing activities (CI, socketedspines = 0.29). Despite having some features that are not typicalfor Cossoninae, most analyses quite strongly supported a sisterrelationship between a monophyletic Araucariini (here includingthe taxa Araucarius, Coptocorynus, and Xenocnema) and the remain-der of Cossoninae. There is therefore little reason to doubt theircossonine affinities.

The historical difficulty in placing the genus Coptonotus in theweevil tree is illustrated well in this study by its ephemeral posi-tion across various analyses. This genus has been placed in Platy-podinae (-idae) or in a separate subfamily or family Coptonotidae(-inae), and more lately been transferred to Scolytinae (Thompson,1992). Morphological characters indicate affinities with Cossoni-nae (Supplementary information Figs. S1 and S2) where particu-larly the tibial characters show a transition between Scolytinae

and Cossoninae. Our analyses confirm the transitional status of thisgenus, varying here between sister to Scolytinae (Fig. 2), and Platy-podinae and Scolytinae combined (Fig. 1), but was also on one oc-casion included in Cossoninae (data set II). Given the ephemeralposition of Coptonotus, not being nested within Scolytinae or Platy-podinae in any analysis, a separate monotypic subfamily Coptonot-inae seems like a prudent solution.

Amorphocerini with the two cycad associated genera Porthetesand Amorphocerus were recently moved from Cossoninae toMolytinae (see Oberprieler et al. (2007) for a detailed overview).Our data support the exclusion of Amorphocerini from Cossoninae,but were not decisive in placing these genera in another highertaxon. Molytinae was not well sampled, however, and it is not un-likely that Amorphocerini will group with other molytines if morebroadly sampled. In the molytine tribe Cryptorhynchini, 8 of the 12combined analyses grouped Psepholax together with the two spe-cies in Cryptorhynchina, which possibly indicate that Psepholacinashould be part of Cryptorhynchini. The key distinguishing charac-ter in Psepholacina, the absence of sclerolepidia, is probably nota very reliable character as we have seen that absence of thesecharacter occur multiple times, e.g. in Scolytinae (Lyal et al., 2006).

4.2. Old fossils – derived clades

It seems paradoxical that the weevil group with the oldestknown fossils (Cognato and Grimaldi, 2009; Kirejtshuk et al.,2009) is not among the first splitting lineages in Curculionidae(see Figs. 1 and 2). Our data strongly support the derived positionirrespective of taxon sampling and concurs with some previousstudies showing that Scolytinae is among the most derived weevilsubfamilies (Kuschel, 1995; Marvaldi et al., 2002; McKenna et al.,2009). Even though several groups such as the Platypodinae andDryophthorinae are still ambiguously placed in the weevil topol-ogy, the derived position of Scolytinae is remarkably stable. Sohow can these beetles be the first appearing in the fossil recordof advanced weevils?

Taxon age based on fossils is obviously biased given the frag-mentary record of fossils, in particular in Cretaceous and oldertimes. Because fossils tend to appear in the geological strata firstwhen species reach high abundance, they will normally indicateonly the minimum age of a taxon, allowing for the inference of old-er ages in most lineages (Forest, 2009). Bark and ambrosia beetlesare generally among the most frequently trapped insects in resins


due to these insects actively seeking damaged trees (Bright andPoinar, 1994; Grimaldi and Engel, 2005). Thus it is likely that otherweevils typically associated with non-woody plant resources areless frequently fossilised in this medium and thus found at lowerabundance in the amber fossil record. It is therefore possible thatolder fossils of Curculionidae are yet to be discovered. Such fossilsdo not necessarily need to be much older than the presently oldestscolytine fossils of 100 Ma, however, as the short internodes sepa-rating the various subfamilies of advanced weevils indicate thatthe radiation of main lineages occurred in a relatively narrow timeframe.

Variation in the oldest known age of a fossil is not trivial fornode calibration, even if a large number of other fossils are in-cluded. For the current data, when using the previously oldestknown age of Scolytine from London clay (55 Ma), most nodeshad their age underestimated (Fig. 4). Time estimates were morein concordance with other studies when the recently discoveredfossil age of 100 Ma was used. These estimates harmonise withthe current consensus on weevil ages, showing that Curculionidaeis older than 100 Ma, that the ancestor of Scolytinae and Platypo-dinae originated about 100 Ma, and also places Brentidae 170 mil-lion years before present, well before the main radiation inCurculionidae (Farrell, 1998; Hunt et al., 2007; Marvaldi et al.,2002; McKenna et al., 2009; Oberprieler et al., 2007).

Some authors have proposed a much older age for Scolytinaebased on fossilised engravings from the lower Cretaceous (re-viewed by Labandeira et al., 2001), a hypothesis also claimed bya few other authors (Morimoto and Kojima, 2003; Wood, 1973).Surely this hypothesis would fit well with a primitive associationwith older gymnosperm lineages such as Araucaria. However,molecular as well as morphology-based phylogenies (Farrell,1998; Kuschel, 1995; Marvaldi, 1997; Marvaldi et al., 2002;McKenna et al., 2009) clearly show that these engravings couldnot have been made by scolytines since those did not evolve beforethe advanced weevils. Crude engravings may be made by bark bor-ing Brentidae which has some species with this habit (Sforzi andBartolozzi, 2004). Claims about a primitive association with oldconifer lineages in Scolytinae, comparable with that of ancientNemonychidae, are therefore unsubstantiated as previously shownin less broadly sampled studies on conifer associated scolytinesthat suggest a more recent re-colonisation of these ancient hostsat the origin of the Scolytinae lineage (Sequeira et al., 2000, 2001).

4.3. The origin of wood boring and social family structures

Excavation of tunnels for food ingestion or fungus cultivation,and for reproduction, is typical for all Scolytinae and Platypodinae.In these groups the parents excavate tunnels where the mother layeggs and one or both parents care for their offspring by removingtunnel frass or relocate eggs according to fungus growth (Fig. 5)(Kirkendall et al., 1997). This behaviour is less known from otherwood boring weevils where mothers generally oviposit externallyand only the larvae and young tenerals are found under the barkor in wood. However, some cossonines do excavate nuptial cham-bers and excavate niches along the tunnels where a single egg islaid in each niche, e.g. in species of Araucarius (Kuschel, 1966),Inosomus (May, 1993), Eurycorynophorus (see Mecke, 2002), Rhync-olus, Stenoscelodes, Stenancylus and related genera (Jordal andOberprieler, 2011). The same behaviour is also observed in threeundescribed species of the conoderine wood boring genus Homoe-ometamelus in the subtribe Campyloscelina (Fig. 5). All other spe-cies of wood boring Conoderini, mainly in Campyloscelina, arenest parasites of ambrosia beetles, mostly platypodines (Schedl,1972; Thompson, 1996), but also some large scolytines in the tribeXyleborini (Jordal, unpublished data). The level of social organisa-tion in nest parasitic conoderines is not documented, but appears

to be similar to the reproductive behaviour in Platypodinae (Jordal,unpublished data). Our phylogeny did not provide sufficient reso-lution to infer the direction of parental care evolution in Conode-rini, but it seems likely that nest parasitism and nestconstruction evolved from a single tunnel dwelling ancestor suchas in Homoeometamelus (see Fig. 5).

All our phylogenetic analyses indicated multiple origins of sub-social breeding systems associated with the wood boring habit.None of the analyses indicated monophyly of Platypodinae, Scolyt-inae, Cossoninae and the conoderine subclade Campyloscelina.Although monophyly of such a group could not be rejected statis-tically in a parsimony framework, these alternative tree topologiesare still much longer by an additional number of character stepsand thus less likely to reflect evolutionary scenarios. The mostlikely scenario thus involves the independent evolution of egg tun-nel galleries in Cossoninae (multiple times), Conoderini (Campylo-scelina), and in Platypodinae–Scolytinae–Coptonotinae. Hamilton(1978, 1979) noted that social systems occur particularly fre-quently in closed niches under bark of dead trees. It is thereforenot surprising that we can infer multiple origins of subsocialbehaviour within a single large insect group such as the weevils.Having in mind that reproductive behaviour has never been ob-served in nature for most tropical weevils, one should expect thatmore taxa are undiscovered with respect to similar kinds ofbehaviour.

4.4. Conclusion and further phylogenetic progress

Resolving weevil phylogeny is a daunting task and our revisedapproach in this study resolved a limited number of nodes at theinter-subfamily or tribal level. We nevertheless managed to docu-ment some progress in our understanding of wood boring weevilclassification – and hence the origin of wood boring – much dueto a comprehensively sampled morphological data matrix, andthe broad sample of intergrading taxa. Thus, we have documentedthe likely monophyly of Scolytinae with respect to Platypodinaeand Cossoninae, and we have confirmed the likely basal relation-ship of Brachycerinae, Entiminae, and perhaps Dryophthorinae.Among the more critical relationships yet to be resolved includethe hyperdiverse Baridinae and Molytinae that have not been ana-lysed sufficiently thorough to understand the possible evolutionarytransitions between these and the wood boring Cossoninae.

Phylogenetic progress in weevils seems most of all dependenton a broader sampling of informative genetic loci. Several studieshave shown that a relatively modest number of extra proteinencoding loci will significantly improve topological stability andnode support in insects (Wiegmann et al., 2009), although a largernumber of loci is more realistically needed (i.e. 10,000 bp, 10–20loci) (see e.g. Rokas et al., 2003; Zwick et al., 2010). Our resultbased on four protein encoding genes, partial LSU, and morpholog-ical data, is nevertheless a major step forward compared to studiesbased on a limited number of genes, especially those relying on asingle category such as rDNA genes (Hundsdoerfer et al., 2009;Marvaldi et al., 2002). The addition of new and little used proteinencoding genes such as CAD and ArgK have contributed consider-ably in this respect (see also Cognato et al., 2011; Dole et al., 2010).Finding similarly useful loci for phylogenetic analyses is not a triv-ial issue, however, and several popular loci used in other insectphylogenies have proven problematic for weevils (Jordal, 2007).Highly conserved markers such as Histone H3 and Polymerase IIare tormented by a multitude of paralogous copies in these beetles.Other less problematic loci such as part of the Potassium–Sodiumpump gene (NaK) and Enolase are evolving at rates not suitable be-yond genus level. A more prosperous and so far less exploredsource of new loci are found among the many ribosomal proteins


in which limited taxon sampling among beetles has shown prom-ising phylogenetics utility (Longhorn et al., 2010).

Acknowledgments

This study was funded, in part, by grants from the NorwegianResearch Council (170565/V40) and Meltzer research grant (UiB-2009) to BHJ, and the National Science Foundation, US (DEB-0328920) to AIC. For highly valuable taxon sample contributions,we thank D. Downie, T. Ekrem, H. Goto, V. Grebennikov, J. Hulcr,D. Kent, L. Kirkendall, M. Mandelshtam, R. Mecke, J. Pedersen, K.Voolma, and F. Ødegaard. Photographs (Fig. 5) were generouslyprovided by A. Breistöl.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ympev.2011.03.016.

References

Alonso Zarazaga, M.A., Lyal, C.H.C., 1999. A World Catalogue of Families and Generaof Curculionoidea (Insecta: Coleoptera) excluding (Scolytidae and Platypodi-dae). Entomopraxis, Barcelona, 315pp.

Beaver, R.A., 1989. Insect–fungus relationships in the bark and ambrosia beetles. In:Wilding, N., Collins, N.M., Hammond, P.M., Webber, J.F. (Eds.), Insect–FungusInteractions. Academic Press, London, pp. 121–143.

Bright, D.E., Poinar Jr., G.O., 1994. Scolytidae and Platypodidae (Coleoptera) fromDominican Republic amber. Annals of the Entomological Society of America 87,170–194.

Britton, E.B., 1960. Beetles from the London Clay (Eocene) of Bognor regis, Sussex.Bulletin of the British Museum (Natural History) (Geology) 4, 27–50.

Browne, F.G., 1972. Larvae of the principal old world genera of the Platypodinae(Coleoptera: Platypodidae). Transactions of the Royal Entomological Society ofLondon 124, 167–190.

Calder, A.A., 1989. The alimentary canal and nervous system of Curculionoidea(Coleoptera): gross morphology and systematic significance. Journal of NaturalHistory 23, 1205–1265.

Cognato, A.I., Grimaldi, D., 2009. 100 million years of morphological conservation inbark beetles (Coleoptera: Curculionidae: Scolytinae). Systematic Entomology34, 93–100.

Cognato, A.I., Hulcr, J., Dole, S.A., Jordal, B.H., 2011. Phylogeny of haplo-diploid,fungus growing ambrosia beetles (Curculionidae: Scolytinae: Xyleborini)inferred from molecular and morphological data. Zoologica Scripta 40, 174–186.

Dole, S.A., Jordal, B.H., Cognato, A.I., 2010. Polyphyly of Xylosandrus Reitter inferredfrom nuclear and mitochondrial genes (Coleoptera: Curculionidae: Scolytinae:Xyleborina). Molecular Phylogenetics and Evolution 54, 773–782.

Downton, M.A., Austin, A.D., 1998. Phylogenetic relationships among themicrogastroid wasps (Hymenoptera: Braconidae): combined analysis of 16Sand 28S rDNA genes and morphological data. Molecular Phylogenetics andEvolution 10, 354–366.

Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis bysampling trees. BMC Evolutionary Biology 7, 214.

Edgar, R.C., 2004. MUSCLE: multiple sequence alignment with high accuracy andhigh throughput. Nucleic Acids Research 32, 1792–1797.

Farrell, B.D., 1998. ’’Inordinate fondness’’ explained: Why are there so manybeetles? Science 281, 555–559.

Farrell, B.D., Sequeira, A., O’Meara, B., Normark, B.B., Chung, J., Jordal, B., 2001. Theevolution of agriculture in beetles (Curculionidae: Scolytinae andPlatypodinae). Evolution 55, 2011–2027.

Forest, F., 2009. Calibrating the tree of life: fossils, molecules and evolutionarytimescales. Annals of Botany 104, 789–794.

Franz, N.M., Engel, M.S., 2010. Can higher-level phylogenies of weevils explain theirevolutionary success? A critical review. Systematic Entomology 35, 597–606.

Gratshev, V.G., Zherikhin, V.V., 2003. The fossil record of weevils and relatedfamilies (Coleoptera, Curculionoidea). Acta Zoologica Cracoviensia 46, 129–138.

Grimaldi, D., Engel, M.S., 2005. Evolution of the Insects. Cambridge University Press,Cambridge.

Hamilton, W.D., 1978. Evolution and diversity under bark. In: Mound, L.A., Waloff,N. (Eds.), Diversity of Insect Faunas. Blackwell Scientific, Oxford, pp. 154–175.

Hamilton, W.D., 1979. Wingless and fighting males in fig wasps and other insects.In: Blum, M.S., Blum, N.A. (Eds.), Reproductive Competition, Mate Choice, andSexual Selection in Insects. Academic Press, New York, pp. 167–220.

Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogeny.Biometrics 17, 754–755.

Hundsdoerfer, A.K., Rheinheimer, J., Wink, M., 2009. Towards the phylogeny of theCurculionoidea (Coleoptera): reconstructions from mitochondrial and nuclearribosomal DNA sequences. Zoologischer Anzeiger – A Journal of ComparativeZoology 248, 9–31.

Hunt, T., Bergsten, J., Levkanicova, Z., Papadopoulou, A., John, O.S., et al., 2007. Acomprehensive phylogeny of beetles reveals the evolutionary origins of asuperradiation. Science 318, 1913–1916.

Jordal, B.H., 2002. Elongation factor 1a resolves the monophyly of the haplodiploidambrosia beetles Xyleborini (Coleoptera: Curculionidae). Insect MolecularBiology 11, 453–465.

Jordal, B.H., 2007. Reconstructing the phylogeny of Scolytinae and close allies:major obstacles and prospects for a solution. In: 3rd Genetics Workshop on BarkBeetles and Associated Microorganisms. USDA Forest Service Proceedings,RMRS-P, pp. 3–8.

Jordal, B.H., 2011. Platypodinae. In: Beutel, R., Leschen, R. (Eds.), Handbook of Zoology,Band IV Arthropoda: Insecta. Part 38: Coleoptera, Beetles. V. deGruyter Press.

Jordal, B.H., Hewitt, G.M., 2004. The origin and radiation of Macaronesian beetlesbreeding in Euphorbia: the relative importance of multiple data partitions andpopulation sampling. Systematic Biology 53, 711–734.

Jordal, B.H., Oberprieler, R.G., 2011. Cossoninae. In: Beutel, R., Leschen, R. (Eds.),Handbook of Zoology, Band IV Arthropoda: Insecta. Part 38: Coleoptera, Beetles.V. deGruyter Press.

Jordal, B.H., Gillespie, J.J., Cognato, A.I., 2008. Secondary structure alignment anddirect optimization of 28S rDNA sequences provide limited phylogeneticresolution in bark and ambrosia beetles (Curculionidae: Scolytinae). ZoologicaScripta 37, 1–14.

Kirejtshuk, A.G., Azar, D., Beaver, R.A., Mandelshtam, M.Y., Nel, A., 2009. The mostancient bark beetle known: a new tribe, genus and species from Lebanese amber(Coleoptera, Curculionidae, Scolytinae). Systematic Entomology 34, 101–112.

Kirkendall, L.R., Kent, D.S., Raffa, K.A., 1997. Interactions between males, femalesand offspring in bark and ambrosia beetles: the significance of living in tunnelsfor the evolution of social behaviour. In: Crespi, B.J., Choe, J.C. (Eds.), SocialBehaviour in Insects and Arachnids. Cambridge University Press, Cambridge, pp.181–215.

Kuschel, G., 1966. A cossonine genus with bark-beetle habits, with remarks onrelationships and biogeography (Coleoptera, Curculionidae). New ZealandJournal of Science 9, 3–29.

Kuschel, G., 1995. A phylogenetic classification of Curculionoidea to families andsubfamilies. Memoirs of the Entomological Society of Washington 14, 5–33.

Kuschel, G., Leschen, R.A.B., Zimmerman, E.C., 2000. Platypodidae under scrutiny.Invertebrate Taxonomy 14, 771–805.

Labandeira, C.C., LePage, B.A., Johnson, A.H., 2001. A Dendroctonus barkengraving (Coleoptera: Scolytidae) from a middle Eocene Larix (Coniferales:Pinaceae): early or delayed colonization? American Journal of Botany 88,2026–2039.

Larsson, S.G., 1978. Baltic Amber – A Palaeobiological Study. Scandinavian SciencePress Ltd., Klampenborg, Denmark.

Lawrence, J.F., Newton, A.F., 1995. Families and subfamilies of Coleoptera (withselected genera, notes, references and data on family-group names). In:Pakaluk, J., Slipinski, S.A. (Eds.), Biology, Phylogeny, and Classification ofColeoptera, Papers Celebrating the 80th Birthday of Roy A. Crowson. Muzeumi Instytut Zoologii PAN, Warzawa, pp. 779–1006.

Lekander, B., 1968. Scandinavian Bark Beetle Larvae; Descriptions and Classification.Department of Forest Zoology, Royal College of Forestry, Stockholm.

Li, C., Lu, G., Orti, G., 2008. Optimal data partitioning and a test case for ray-finned fishes (Actinopterygii) based on ten nuclear loci. Systematic Biology57, 519–539.

Longhorn, S.J., Pohl, H.W., Vogler, A.P., 2010. Ribosomal protein genes ofholometabolan insects reject the Halteria, instead revealing a close affinityof Strepsiptera with Coleoptera. Molecular Phylogenetics and Evolution 55,846–859.

Lyal, C.H.C., 1995. The ventral structures of the weevil head (Coleoptera:Curculionoidea). Memoirs of the Entomological Society of Washington 14,35–51.

Lyal, C.H.C., Douglas, D.A., Hine, S.J., 2006. Morphology and systematic significanceof sclerolepidia in the weevils (Coleoptera: Curculionoidea). Systematics andBiodiversity 4, 203–241.

Maddison, W.P., Maddison, D.R., 2010. Mesquite: a modular system for evolutionaryanalysis. Version 2, 74.

Marvaldi, A.E., 1997. Higher level phylogeny of Curculionidae (Coleoptera:Curculionoidea) based mainly on larval characters, with special reference tobroad nosed weevils. Cladistics 13, 285–312.

Marvaldi, A.E., Sequeira, A.S., O’Brien, C.W., Farrell, B.D., 2002. Molecular andmorphological phylogenetics of weevils (Coleoptera, Curculionoidea): doniche shifts accompany diversification? Systematic Biology 51, 761–785.

Mathew, G., 1987. Insect borers of comercially important stored timber in the stateof Kerala, India. Journal of Stored Products Research 23, 185–190.

May, B.M., 1993. Larvae of Curculionoidea (Insecta: Coleoptera): a systematicoverview. Fauna of New Zealand 28, 1–1221.

May, B.M., 1997. Larvae of two weevil genera (Coleoptera: Curculionidae)associated with Crossotarsus armipennis Lea (Platypodidae) from New SouthWales. New Zealand Entomologist 20, 7–13.

McKenna, D.D., Sequeira, A.S., Marvaldi, A.E., Farrell, B.D., 2009. Temporal lags andoverlap in the diversification of weevils and flowering plants. Proceedings of theNational Academy of Science USA 106, 7083–7088.

Mecke, R., 2002. Insetos do Pinheiro Brasileiro. Attempto Service GmbH, Tübingen,Germany.

Mecke, R., 2005. Tibial thorns or true hairs of corresponding structure in both, barkbeetles (Scolytidea) and bark inhabiting weevils (Curculionidae: Cossoninae:Araucariini). Entomologia Generalis 28, 17–21.



Miller, K.B., Bergsten, J., Whiting, M.F., 2009. Phylogeny and classification of thetribe Hydaticini (Coleoptera: Dytiscidae): partition choice for Bayesian analysiswith multiple nuclear and mitochondrial protein-coding genes. ZoologicaScripta 38, 591–615.

Morimoto, K., 1962. Comparative morphology and phylogeny of the superfamilyCuculionoidea of Japan. Journal of the Faculty of Agriculture, Kyushu University11, 331–374.

Morimoto, K., Kojima, H., 2003. Morphologic characters of the weevil head andphylogenetic implications (Coleoptera, Curculionoidea). Esakia 43, 133–169.

Muskus-Arrieta, R.A., Atkinson, T.H., 1992. Anatomia externa de larvas deescarabajos descortezadores (Coleoptera: Scolytidae) y su utilidad en lataxonomia. Agrociencia serie Proteccion Vegetal 3, 47–59.

Normark, B.B., Jordal, B.H., Farrell, B.D., 1999. Origin of a haplodiploid beetle lineage.Proceeding of the Royal Society, London, Series B 266, 2253–2259.

Nylander, J.A.A., 2004. MrModeltest. Evolutionary Biology Centre, UppsalaUniversity.

Oberprieler, R.G., Marvaldi, A.E., Anderson, R.S., 2007. Weevils, weevils, weevilseverywhere. Zootaxa 1668, 491–520.

Pick, K.S., Philippe, H., Schreiber, F., Erpenbeck, D., Jackson, D.J., Wrede, P., Wiens, M.,Alie, A., Morgenstern, B., Manuel, M., Worheide, G., 2010. Improvedphylogenomic taxon sampling noticeably affects nonbilaterian relationships.Molecular Biology and Evolution 27, 1983–1987.

Pollock, D.D., Zwickl, D.J., McGuire, J.A., Hillis, D.M., 2002. Increased taxonsampling is advantageous for phylogenetic inference. Systematic Biology 51,664–671.

Rokas, A., Williams, B.L., King, N., Carroll, S.B., 2003. Genome-scale approaches toresolving incongruence in molecular phylogenies. Nature 425, 798–804.

Rosenberg, M.S., Kumar, S., 2001. Incomplete taxon sampling is not a problem forphylogenetic inference. Proceedings of the National Academy of Science USA 98,10751–10756.

Schedl, K.E., 1972. Monographie der Familie Platypodidae, Coleoptera. Verlag Dr. W.Junk, den Haag.

Scudder, S.H., 1893. Tertiary Rhynchophorus Coleoptera of the United States. USGeological Survey Monograph 21.

Sequeira, A.S., Normark, B.B., Farrell, B.D., 2000. Evolutionary assembly of theconifer fauna: distinguishing ancient from recent associations in barkbeetles. Proceedings of the Royal Society of London Series B 267, 2359–2366.

Sequeira, A.S., Normark, B.B., Farrell, B.B., 2001. Evolutionary origins of Gondwananinteractions: how old are Araucaria beetle herbivores? Biological Journal of theLinnean Society 74, 459–474.

Sforzi, A., Bartolozzi, L., 2004. Brentidae of the World (Coleoptera, Curculionoidea).Museo regionale di Scienze Naturali, Torino.

Simon, C., Frati, F., Beckenbach, A., Crespi, B., Liu, H., Flook, P., 1994. Evolution,weighting and phylogenetic utility of mitochondrial gene sequences and acompilation of conserved polymerase chain reaction primers. Annals of theEntomological Society of America 87, 651–701.

Swofford, D., 2002. PAUP⁄. Phylogenetic Analysis Using Parsimony (⁄and othermethods) version 4. Sinauer Associates, Sunderland, Massachusetts.

Thompson, R.T., 1992. Observation on the morphology and classification of weevils(Coleoptera, Curculionoidea) with a key to major groups. Journal of NaturalHistory 26, 835–891.

Thompson, R.T., 1996. The species of Phaenomerus Schönherr (Coleoptera:Curculionidae: Zygopinae) of the Australian region. Invertebrate Taxonomy10, 937–993.

Whiting, M.F., Carpenter, J.C., Wheeler, Q.D., Wheeler, W.C., 1997. The Strepsipteraproblem: phylogeny of the holometabolous insect orders inferred from 18S and28S ribosomal DNA sequences and morphology. Systematic Biology 46, 1–68.

Wiegmann, B., Trautwein, M., Kim, J.-W., Cassel, B., Bertone, M., Winterton, S.,Yeates, D., 2009. Single-copy nuclear genes resolve the phylogeny of theholometabolous insects. BMC Biology 7, 34.

Wood, S.L., 1973. On the taxonomic status of Platypodidae and Scolytidae(Coleoptera). Great Basin Naturalist 33, 77–90.

Wood, S.L., 1986. A reclassification of the genera of Scolytidae (Coleoptera). GreatBasin Naturalist Memoirs 10, 126.

Wood, S.L., 1993. Revision of the genera of Platypodidae (Coleoptera). Great BasinNaturalist 53, 259–281.

Zablotny, J.E., 2003. Sociality. In: Resh, V.H., Cardé, R.T. (Eds.), Encyclopedia ofInsects. Academic Press, Amsterdam.

Zimmermann, E.W., 1994. Australian weevils (Coleoptera: Curculionoidea), Vol. I:Orthoceri: Anthribidae to Attelabidae: The Primitive Weevils.

Zwick, A., Regier, J.C., Mitter, C., Cummings, M.P., 2010. Increased gene samplingyields robust support for higher-level clades within Bombycoidea(Lepidoptera). Systematic Entomology. doi:10.1111/j.1365-3113.2010.00543.x.

Zwolinski, J.B., Geldenhuys, C.J., 1988. Insect damage to timbers from the indigenousforests – preliminary results. South African Forestry Journal 146, 7–11.

The age and phylogeny of wood boring weevils and the origin of subsociality

Documents

Transcript of The age and phylogeny of wood boring weevils and the origin of subsociality