molecular phylogenetics of a south pacific sap beetle species … · molecular phylogenetics of a...

Molecular Phylogenetics and Evolution 64 (2012) 428–440

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

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

Molecular phylogenetics of a South Pacific sap beetle species complex (Carpophilusspp., Coleoptera: Nitidulidae)

Samuel D.J. Brown a,⇑, Karen F. Armstrong a, Robert H. Cruickshank b

a Bio-Protection Research Centre, PO Box 84, Lincoln University, Lincoln 7647, Canterbury, New Zealandb Department of Agriculture and Life Sciences, PO Box 84, Lincoln University, Lincoln 7647, Canterbury, New Zealand

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

Article history:Received 19 May 2011Revised 28 March 2012Accepted 30 April 2012Available online 18 May 2012

Keywords:OceaniaPolynesiaTaxonomybiosecurity

1055-7903/$ - see front matter � 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.ympev.2012.04.018

⇑ Corresponding author.E-mail address: [email protected] (S.D.J. Br

Several species of sap beetles in the genus Carpophilus are minor pests of fresh produce and stored prod-ucts, and are frequently intercepted in biosecurity operations. In the South Pacific region, the superficiallysimilar species C. maculatus and C. oculatus are frequently encountered in these situations. Three subspe-cies of C. oculatus have been described, and the complex of these four taxa has led to inaccurate identi-fication and questions regarding the validity of these taxa. A molecular phylogenetic study using themitochondrial gene cytochrome c oxidase I (COI) and two nuclear markers comprising the rDNA internaltranscribed spacer 2 (ITS2) and the D1–D2 region of the large (28S) ribosomal RNA subunit showed that C.maculatus, and C. o. cheesmani were easily differentiated from the two other subspecies of C. oculatus. COIalso showed differentiation between C. o. gilloglyi and C. o. oculatus, but this was not shown when thirdcodon positions were removed and when RY-coding analyses were conducted. Generalised mixed Yule-Coalescent (GMYC) models were fitted to trees estimated from the COI data and were analysed using amultimodel approach to consider the evidence for three taxonomic groupings of the C. oculatus group.While the arrangement with the highest cumulative weight was not the arrangement ultimatelyaccepted, the accepted taxonomy also had an acceptable level of support. ITS2 showed structure withinC. oculatus, however C. o. oculatus was resolved as paraphyletic with respect to C. o. gilloglyi. COI showedevidence of sequence saturation and did not adequately resolve higher relationships between speciesrepresented in the dataset. 28S resolved higher relationships, but did not perform well at the specieslevel. This study supports the validity of C. maculatus as a separate species, and provides sufficient evi-dence to raise C. o. cheesmani to the level of species. This study also shows significant structure withinand between C. o. gilloglyi and C. o. oculatus, giving an indication of recent speciation events occurring.To highlight the interesting biology between these two taxa, C. o. gilloglyi is retained as a subspecies ofC. oculatus. These results give clarity regarding the taxonomic status of C. maculatus and the subspeciesof C. oculatus and provide a platform for future systematic research on Carpophilus.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

The sap beetle genus Carpophilus (Coleoptera: Nitidulidae)contains over 200 species found throughout the world. Both adultsand larvae live and develop in organic matter, including rottingfruit and stored products, which has made some species pests inorchards and warehouses. Carpophilus davidsoni Dobson, C. hemi-pterus (Linnaeus) and C. mutilatus Erichson have emerged as seri-ous pests of stone fruit in Australia with the decrease in use ofbroad-spectrum insecticides (Hossain and Williams, 2003; Jameset al., 1997; James and Vogele, 2000), while C. lugubris Murray isan important pest of corn in the United States (Dowd, 2000) andC. sayi Parsons has been implicated in the transmission of oak wilt

ll rights reserved.

own).

disease (Cease and Juzwik, 2001). The species most commonlyfound in stored products are C. hemipterus, C. dimidiatus (Fabricius),C. ligneus Murray and C. obsoletus Erichson (Dobson, 1955). A num-ber of species are also associated with flowers and are importantpollinators, particularly of the Annonaceae (Nagel et al., 1989). Inthe tropical South Pacific region, C. maculatus Murray and C. ocula-tus Murray are widespread species that are commonly found inagricultural situations, and have been intercepted in exported freshproduce (Archibald and Chalmers, 1983; Leschen and Marris,2005).

There has been much published on the ecology and control ofthe group, as it is considered to be the most economically impor-tant genus within the Nitidulidae (Connell, 1981). However, thetaxonomy of the genus remains problematic, with the last globalrevision of the group dating to the 19th Century (Murray, 1864).Since Murray’s seminal work, several new species have been

http://dx.doi.org/10.1016/j.ympev.2012.04.018

mailto:[email protected]


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

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

S.D.J. Brown et al. / Molecular Phylogenetics and Evolution 64 (2012) 428–440 429

described, most notably by Dobson and Kirejtshuk (e.g. Dobson,1952, 1993a; Kirejtschuk, 2001). Not withstanding this, the genusremains in need of a thorough taxonomic revision, as most speciesare poorly defined and morphologically variable, with subtle char-acters defining the species. This taxonomic uncertainty impacts onthe ability to understand their relative invasiveness and risk.

Within the Nitidulidae, Carpophilus is placed in the subfamilyCarpophilinae (Kirejtschuk, 2008). While no formal systematicstudies have yet been published on the relationships of higher taxawithin the Nitidulidae, it is believed that the Carpophilinae form asingle lineage with the Epuraeinae and Amphicrossinae (Kirejts-chuk, 2008), an arrangement that is consistent with preliminarymolecular systematic results from the family (A. Cline pers. comm.).The genus is easily identified by having two exposed abdominaltergites and a button-like male 8th sternite (Gillogly, 1962; Leschenand Marris, 2005). It is morphologically conservative betweenspecies within the genus and there are a number of groups contain-ing several species with very similar appearance and habits. Carpo-philus is classified into nine subgenera: Carpophilus Stephens sensustricto, Megacarpolus Reitter, Semocarpolus Kirejtshuk, Gaplocarpo-lus Kirejtshuk, Askocarpolus Kirejtshuk, Plapennipolus Kirejtshuk,Ecnomorphus Motschulsky, Caplothorax Kirejtshuk and MyothoraxMurray (Kirejtschuk, 2008). However, the exact limits of these sub-genera have not been expanded on in the context of an overall re-view of Carpophilus systematics. The majority of these subgeneraare fairly small, consisting of only a few species, most of whichare confined to South-East Asia. The exceptions are Carpophilus s.str., Ecnomorphus and Myothorax, which are large and cosmopolitan.

Carpophilus maculatus and C. oculatus are widely distributedthrough the South Pacific, with C. maculatus also being found inSoutheast Asia, South America and the West Indies. The two spe-cies are similar, with the key character that differentiates betweenthem being the extent of red colouration on the elytra. However,this patterning is very variable and there is significant overlap inthe patterns possessed by both species. Dobson (1993b) reviewedC. oculatus and described three subspecies based on differencesin pronotal punctation and male genitalia. The nominate subspe-cies, C. o. oculatus is found from Tahiti through the central Pacificto New Caledonia. There is significant overlap in geographic rangebetween this subspecies and C. o. gilloglyi Dobson which is foundfrom Easter Island to Fiji and Micronesia. The last subspecies, C.o. cheesmani Dobson is confined to Vanuatu. Morphologicalevidence for monophyly of the C. oculatus subspecies include thedistinctive ring-shaped elytral colour pattern and the pronotallength/width ratio, however there are no clear and unambiguoussynapomorphies that define C. oculatus. A sister taxon relationshipis suggested between C. oculatus and C. maculatus by the sculptur-ing of the prosternum and by the shape of the male genitalia. Thegenitalic similarity is particularly striking between C. maculatusand C. o. gilloglyi, with the shape of the parameres of these two spe-cies being nearly identical. These similarities, and the variation incolour pattern has led some to surmise that these two species forma single, hyper-variable species (Leschen and Marris, 2005).

In this study, the monophyly of C. oculatus and its relationshipwith C. maculatus was tested using molecular systematic tech-niques. The opportunity was also taken to provide a preliminaryglimpse into the systematics of Carpophilus, by sampling a rangeof species within the genus and with an emphasis on species with-in the subgenus Myothorax. A molecular phylogeny of Carpophiluswas constructed based on sequences of mitochondrial and nucleargene regions from as many species as it was possible to obtain. Thiswas used primarily to test the monophyly of C. oculatus and itssubspecies, and to test for a sister-taxon relationship betweenC. oculatus and C. maculatus. It also provides a basis for the system-atics of Carpophilus in general and will be useful for guiding futuretaxonomic work on the genus.

2. Methods

2.1. Taxon coverage and specimens

To test the monophyly of C. oculatus, the three subspecies of thisspecies were collected preferentially. The relationship of C. oculatuswith C. maculatus was also targeted, given the morphological sim-ilarity between them. Beyond this, other Carpophilus species werecollected opportunistically by the senior author and colleagues,primarily from the Pacific.

Specimens were collected by hand from rotting fruit and vege-tables and preserved in propylene glycol in the field for transport-ing back to New Zealand. In the laboratory, specimens were sortedand temporarily stored in 100% ethanol. A single leg was removedfrom each specimen to be used for molecular analysis and theremainder card-mounted as a voucher specimen and deposited inthe Lincoln University Entomology Museum (LUNZ).

2.2. Molecular methods

DNA was extracted using the prepGEM DNA extraction kit (Zy-GEM Ltd., Hamilton, New Zealand). Incubation consisted of 30 minat 75 �C and 5 min at 95 �C. This longer incubation period ensuredthat dehydrated tissues had sufficient time to rehydrate and lyse.DNA was amplified using a 10 ll PCR reaction containing 0.25 UExpand High-Fidelity Taq (Roche Applied Science, Indianapolis,IN, USA), 0.2 mM dNTPs, 2 mM MgCl2 and 0.3 mM of both forwardand reverse primers. To encourage DNA amplification in difficultspecimens, 2 ll GC Rich mixture (Roche) was added to the reactionwhen necessary. The 50 end of the cytochrome c oxidase subunit I(COI) mitochondrial gene, the D1–D2 region of the 28S ribosomalRNA gene and the internal transcribed spacer 2 (ITS2) region inthe nuclear ribosomal encoding cistrons were amplified using theprimers listed in Table 1. For COI the combination LCO1490/HCO2198 was used preferentially, with TY-J-1460/C1-N-2191 usedwhen these amplifications were unsuccessful. ITS2 amplificationsrequired a primer concentration of 0.5 mM. Reactions were runon a GeneAmp 9700 thermocycler (Applied Biosystems, Foster City,CA, USA) with an initial denaturation of 94 �C for 2 min, followedby 40 cycles of 94 �C (15 s), 45 �C (30 s) and 72 �C (75 s), and witha final extension at 72 �C for 7 min. An annealing temperature of54 �C was used for 28S reactions. Successful amplification was con-firmed by running PCR products in 1.0% agarose gels made using aNaOH/borate buffer with a pH of 8.0; these were run at 170 V and50 mA for 15 min. Gels were stained during casting with SYBR SafeDNA gel stain (Invitrogen, Carlsbad, CA, USA) and viewed with aGeneWizard Gel imaging system (SynGene, Cambridge, England).PCR products were sequenced in 10 ll reactions containing 0.3 llof PCR product, 0.5 ll BigDye� Terminator (v3.1) (Applied Biosys-tems), 2 ll BigDye� Sequencing buffer (v1.1/3.1) and 0.8 lM ofthe primers used for amplification. Products were sequenced inboth directions. PCR cleanup was done with CleanSEQ� Dye-Terminator Removal kit (Agencourt Bioscience Corporation,Beverly, MA, USA). Sequences were read using a Long ReadSequencing Protocol on an ABI Prism� 3100-Avant GeneticAnalyzer (Applied Biosystems).

2.3. Analysis

Sequences were aligned by eye using the manual alignmentsoftware BIOEDIT (Hall, 1999). COI data were analysed using fourmethods, (1) Original nucleotide (NT) coded data, (2) Third codonpositions excluded, (3) Third codon positions RY-coded, and (4)All data RY-coded (Phillips et al., 2004), while the other two lociwere unaltered. Loci were analysed independently, due to the

Table 1Markers and PCR primer combinations used in this research.

Marker Primer name Primer sequence Reference

COI LCO1490 50-GGT CAA CAA ATC ATA AAG ATA TTG G-30 Folmer et al. (1994)HCO2198 50-TAA ACT TCA GGG TGA CCA AAA AAT CA-30 Folmer et al. (1994)TY-J-1460 50-TAC AAT TTA TCG CCT AAA CTT CAG CC-30 Simon et al. (1994)C1-N-2191 50-CCC GGT AAA ATT AAA ATA TAA ACT TC-30 Simon et al. (1994)

28S LSUfw2 50-ACA AGT ACC DTR AGG GAA AGT TG-30 Sonnenberg et al. (2007)LSUrev1 50-TAC TAG AAG GTT CGA TTA GTC-30 Sonnenberg et al. (2007)

ITS2 CAS5p8sFc 50-TGAA CAT CGA CAT TTY GAA CGC ACA T-30 Ji et al. (2003)CAS28sB1d 50-TTC TTT TCC TCC SCT TAY TRA TAT GCT TAA-30 Ji et al. (2003)

430 S.D.J. Brown et al. / Molecular Phylogenetics and Evolution 64 (2012) 428–440

inconsistencies in phylogenetic inference when genes with differ-ing histories are concatenated (Kubatko and Degnan, 2007), andthe lack of multiple samples for many species precluded the useof coalescent-based approaches (Heled and Drummond, 2010).

Maximum likelihood (ML) analyses for all loci were run usingPHYML ALRT (Anisimova and Gascuel, 2006; Guindon and Gascuel,2003), and support for the topology calculated using parametricbootstrap procedures with 1000 replicates. JMODELTEST (Posada,2008) was used to determine the appropriate models of molecularevolution, based on the Akaike information criterion (AIC). Satura-tion plots of the p-distance against ML distances calculated fromtrees, were constructed in R using functions from the APE package(Paradis et al., 2004; R Development Core Team, 2012). Pairwisedistances are reported as both raw p-distances and as model-cor-rected distances to illustrate the uncertainty in the derivation ofgenetic distance estimates (Collins et al., in press).

Ultrametric COI trees for generalised mixed Yule-Coalescent(GMYC) analysis were estimated from the NT coded COI data usingBEAST 1.6.2 (Drummond and Rambaut, 2007; Drummond et al.,2006) using an HKY + I + C model of evolution, a relaxed lognormalmolecular clock and default priors. A lognormal molecular clockmodel was justified with the ucld.stdev value of 1.384 indicatingconsiderable rate heterogeneity. The MCMC chain was run for a to-tal of 50.7 million generations, and trees were sampled every 1000generations. A maximum clade credibility (MCC) tree was madeusing TREEANNOTATOR from the sampled trees after removing a stan-dard burn-in of the first 10% trees of each run. GMYC (Pons et al.,2006) was conducted on the MCC tree, and on 2,000 other ran-domly selected trees generated by BEAST. The number and compo-sition of clusters estimated using both single and multiple GMYCthreshold were calculated. The composition of clusters, AIC, AICweights, model-weighted number of entities (Powell, 2012) andother statistics on GMYC objects were calculated using SPIDER V.1.1-2 and SPIDERDEV V. 0.0-3 (Brown et al., 2012).

To investigate the probability of trees based on a priori expecta-tions (i.e. monophyly of C. oculatus and Carpophilus), SH-tests (Shi-modaira and Hasegawa, 1999) were conducted using the SH.test

function in the PHANGORN package for R (Schliep, 2011).

Table 2Sequence polymorphism data.

COI 28S ITS2

Sequences 104 43 53Species 29 22 6Length 576 734 504Gaps 0 135 169Variable sites (%) 219 (38) 226 (38) 49 (15)Parsimony informative sites 201 133 27

Base frequenciesA 28.0 20.8 24.9C 19.6 25.5 20.2G 16.9 31.5 23.2T 35.5 22.2 31.7

3. Results

3.1. Taxon sampling

Specimens of 17 species of Carpophilus were sequenced. Theserepresented four subgenera, as proposed by Kirejtschuk (2008).Semocarpolus was represented by a single species, C. marginellus;Carpophilus by three species (C. lugubris, C. obsoletus and C. hemipte-rus); Ecnomorphus by four (C. antiquus Melsheimer, C. discoideusLeconte, C. bakewelli Murray and C. corticinus Erichson); andMyothorax by eight species (C. dimidiatus, C. mutilatus, C. davidsoni,C. gaveni Dobson, C. nepos Murray, C. maculatus, C. oculatus, C. robu-stus Murray and C. schiodtei Murray). A single species of Urophorus

(U. humeralis (Fabricius)) was also included in the dataset. COI and28S were sequenced from all species, with the exception of C.bakewelli and C. obsoletus for which 28S was unable to be amplified.ITS2 was only sequenced from selected species within the Myotho-rax subgenus. Outgroups were selected from available specimensand previously published sequences of other species within theNitidulidae. These represented three subfamilies outside of theCarpophilinae and included Epuraea signata Broun and E. ocularisFairmaire (Epuraeinae), unidentified Conotelus and Brachypeplusspecies (Cillaeinae), and Aethina concolor (Macleay), Omosita discoi-dea (Fabricius), three Meligethes species and unidentified specimensof Phenolia and Stelidota (Nitidulinae).

3.2. Alignments and variation

A total of 200 sequences over the three markers were obtained.Sequences have been submitted to Genbank with accession num-bers GU217433–GU217530 for COI, GU217391–GU217432 for28S and GU217338–GU217390 for ITS2. Specimen details areavailable as Supplementary Material (Table 6). Sequence polymor-phism data for each gene are presented in Table 2.

28S and ITS2 alignments required the addition of alignmentgaps because of the presence of indels. While there are no indelsin the COI dataset, 11 specimens had missing values from shortsequencing runs. The maximum number of missing bases in anyone specimen was 141, with the median number being 9. Base fre-quencies of the three markers are typical for insects (Lin and Dan-forth, 2004).

Although COI and 28S had roughly equivalent levels of variablesites (Table 2), COI had more parsimony-informative sites than 28S,and was better at distinguishing between species and subspecies ofCarpophilus (Fig. 2 c.f. Fig. 3). Distribution of parsimony-informativesites in COI across first, second and third codon positions respec-tively was 37, 5, and 159, a result consistent with most protein-coding genes (Xia, 1998). The saturation plot (Fig. 1) for COI showeddeviation from linearity, suggesting that the marker is unsuitablefor inferring higher relationships. Both nuclear markers are

0.0 0.2 0.4 0.6 0.8 1.0

0.00

50.

00.

100.

150.

20

COI

ML distance

p−di

stan

ce

0.0 0.2 0.4 0.6 0.8 1.0

0.00

0.05

0.10

0.15

0.20

ITS2

ML distance

p−di

stan

ce

0.0 0.2 0.4 0.6 0.8 1.0

0.00

0.05

0.10

0.15

0.20

28S

ML distance

p−di

stan

ce

Fig. 1. Saturation plots of p-distances against ML distances based on the optimal model. The solid line shows the 1:1 relationship between p-distances and ML distances.


essentially linear, showing no evidence of saturation at the levelinvestigated here.

3.3. Phylogenetic analyses

3.3.1. COI3.3.1.1. Maximum likelihood analyses. In all four COI ML analyses, C.o. gilloglyi + C. o. oculatus are shown to form a monophyletic clade.This clade is supported by >50% bootstrap values in all analyses,with the exception of the dataset where third codon positions wereexcluded. None of the analyses resolved the sister-group relation-ship of C. o. cheesmani with statistical support or placed them withthe other two C. oculatus subspecies. The full NT-coded datasetresolves C. o. gilloglyi and C. o. oculatus as being reciprocally mono-phyletic, while all other datasets resolve C. o. gilloglyi as paraphy-letic with respect to C. o. oculatus.

Maximum likelihood of the nucleotide-coded COI data inferreda single tree of -ln likelihood 8179.84071 from a TPM3uf + I + Cmodel with nucleotide frequencies and rate parameters as shownin Tables 2 and 3. The tree (Fig. 2) grouped conspecific taxa witheach other, and were supported with high bootstrap values. Inparticular, the monophyly of C. o. oculatus and C. o. gilloglyi issupported by a bootstrap of 72%. However, higher relationshipsbetween species and between genera are not resolved with anydegree of certainty. Support for deeper nodes were not increasedby the removal of third codon positions or RY coding the data(see Supplementary Material).

The distances between the subspecies of C. oculatus are given inTable 4. In addition, C. o. gilloglyi was shown to have a very deep split(16.5% mean ML pairwise distance; 7.41% mean p-distance) be-tween western specimens collected in Fiji (including Rotuma) andeastern populations from the Kermadec Islands to French Polynesia,including Tonga. The distinction between these two groups is

Table 3Rate parameters of models used in analyses.

Analysis Model likelihood f (A)

COI nt-coded TPM3uf + I+C �8179.84071 0.27986COI 3ex TIM2 + I+C �1660.78625 0.21536COI 3RY TrN + C �4266.50916 0.21838COI RY TIM3ef + C �3020.17188 0.2500028S GTR + C �3603.59084 0.20800ITS2 GTR �1478.1793 0.24900

f (A–C) f (A–G) f (A–T)COI nt-coded 0.37773 2.99728 1.00000COI 3ex 2.46898 4.07567 2.46898COI 3RY 1.00000 1.50433 1.00000COI RY 0.00001 0.62655 1.0000028S 0.58770 2.09683 1.48477ITS2 0.89063 2.24518 2.55499

retained throughout the modifications to the dataset, though theexact relationships vary between the modifications (seeSupplementary Material). When third codons are removed, theFijian clade of C. o. gilloglyi is nested within C. o. oculatus, whilethe eastern clade is paraphyletic to the clade formed by the formergroups. These relationships, however, do not have bootstrap sup-port, while the monophyly of C. o. cheesmani is supported with abootstrap of 74.3%. In both RY-coded datasets, the Fijian clade issister to the Eastern clade + C. o. oculatus (see SupplementaryMaterial).

A deep split is also seen between Australian and Pacific popula-tions of C. maculatus, with a mean ML pairwise distance betweenthe clades of 14.52% (6.01% mean p-distance). There was also sig-nificant structuring within the Pacific clade, with a mean ML pair-wise distance of 4.67% (1.9% mean p-distance). Within a single,small island (Wallis), a specimen was 3.87% ML distance (2.62%p-distance) different from its conspecifics from the same island.

Within other species with a sample size that allows the calcula-tion of intra-specific distances, C. nepos had a mean ML pairwisedistance of 4.18% (1.59% mean p-distance) from three specimenscollected in New Caledonia, Tahiti and a laboratory culture fromthe USA. In ML analyses, the specimen from New Caledonia was re-solved as basal to the other two specimens, with support of 59.4%,while in the Bayesian tree, the specimen from the USA was basal,making the Pacific specimens monophyletic (posterior probability(PP) = 0.57). The ML distance between the Pacific specimens was3.62% (0.71% p-distance). Carpophilus mutilatus was representedby three specimens from Hawaii, Tahiti and Nauru, and had a meanML pairwise distance of 0.42% (0.16% mean p-distance). In both MLand Bayesian analyses, the Nauru specimen was basal to the othertwo specimens (bootstrap: 57.9%; PP = 0.76). The relationships ofthese two species were not resolved in any of the analyses withany degree of support, with the exception of a C. nepos–C. davidsoni

f (C) f (G) f (T) C

0.19606 0.16908 0.35499 0.687000.22705 0.22740 0.33019 0.149000.22453 0.23058 0.32651 0.126000.25000 0.25000 0.25000 0.117000.25500 0.31500 0.22200 0.326000.20200 0.23200 0.31700 –

f (C–G) f (C–T) f (G–T) Invariant0.37773 2.99728 1.00000 0.4281.00000 38.39180 1.00000 0.4161.00000 3.06616 1.00000 –0.00001 0.33507 1.00000 –0.46896 3.41398 1.00000 –0.71150 1.93285 1.00000 –

Aethina concolor VANAethina concolor VAN

Conotelus sp. USAStelidota sp. VAN

Brachypeplus sp. AUS

Meligethes coracinus

Meligethes aeneusMeligethes gracilis

Phenolia sp. VANOmosita discoidea

C. obsoletus BISUrophorus humeralis AUS

C. robustus BIS

C. maculatus TUVC. maculatus TUVC. maculatus TUVC. maculatus TUV

C. maculatus NAUC. maculatus BIS

C. maculatus SAMC. maculatus TUVC. maculatus SOCC. maculatus HAW

C. maculatus AUSC. maculatus AUS

C. gaveni NZL

C. schioedtei BISC. schioedtei BIS

C. davidsoni AUSC. davidsoni AUS

C. mutilatus NAU

C. mutilatus SOCC. mutilatus HAW

C. o. gilloglyi TUVC. o. gilloglyi FIJ

C. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi FIJ

C. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi FIJ

C. o. gilloglyi FIJC. o. gilloglyi FIJ

C. o. gilloglyi KER

C. o. gilloglyi KERC. o. gilloglyi TON

C. o. gilloglyi TUBC. o. gilloglyi SOCC. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi TUBC. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi SOC

C. o. gilloglyi SOCC. o. gilloglyi SOC

C. o. gilloglyi TUBC. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi KERC. o. gilloglyi KER

C. o. oculatus VANC. o. oculatus VANC. o. oculatus VANC. o. oculatus VANC. o. oculatus VANC. o. oculatus SOC

C. o. oculatus TUVC. o. oculatus FIJ

C. o. oculatus TONC. o. oculatus TON

C. o. oculatus VAN

C. o. oculatus FIJC. o. oculatus FIJ

C. o. oculatus VANC. o. oculatus VAN

C. o. oculatus NAUC. o. oculatus VAN



C. nepos NCL

C. nepos USAC. nepos SOC

C. dimidiatus SOCC. o. cheesmani VAN

C. o. cheesmani VANC. o. cheesmani VAN

C. corticinus USA

C. hemipterus AUSC. hemipterus NZL

Epuraea ocularis SOC

Epuraea signata NZLEpuraea signata NZL

C. discoideus USAC. antiquus USA

C. marginellus FIJC. marginellus TUV

C. bakewelli AUS

C. lugubris USAC. lugubris USA

100

99.5

85.669.6

96.4

98.8

66.969.1

70.675.3

90.662.6

63.7

98.2

100

99.7

10057.9

71.9

65.8

97.473.2

62.571.681.4

98

66.3

76.7

87.5

66

99.959.4

99.8

100

98.9

99.9

64.7100

0.1

Fig. 2. ML tree of nucleotide-coded COI sequences. Bootstrap values are shown above nodes with greater than 50% support. Three letter codes signify general locality ofspecimens: AUS–Australia, BIS–Bismarck Islands, FIJ–Fiji, HAW–Hawaii, KER–Kermadec Islands, NAU–Nauru, NCL–New Caledonia, NZL–New Zealand, SAM–Samoa, SOC–Society Islands (French Polynesia), TON–Tonga, TUB–Austral Islands (French Polynesia), TUV–Rotuma Island (close to Tuvalu), USA–United States of America, VAN–Vanuatu.


Table 4COI mean pairwise distances between C. oculatus subspecies. Upper triangle:p-distances; Lower triangle: ML distances; Diagonal: ML distance/p-distance.

C. o. oculatus C. o. gilloglyi C. o. cheesmani

C. o. oculatus 0.0091/0.0037 0.0669 0.0921C. o. gilloglyi 0.2063 0.0910/0.0415 0.1137C. o. cheesmani 0.3959 0.4662 0/0


sister group relationship being resolved in the ITS2 analysis with abootstrap value of 66%. Unfortunately, C. mutilatus is not repre-sented in this dataset.

3.3.1.2. Bayesian analyses. Consistent with the ML analyses, C. ocul-atus was not resolved as being monophyletic on the Bayesian MCCtree (Fig. 6). The combination of C. o. oculatus + C. o. gilloglyi formeda strongly supported clade (PP = 0.99), whereas C. o. cheesmani wasfound to have a weakly supported relationship with C. dimidiatus(PP = 0.56). Within C. o. gilloglyi, both western and eastern cladeswere well supported (both with PP = 1), however, the node linkingthe two clades did not have strong support (PP = 0.47). The sistertaxa of C. o. oculatus + C. o. gilloglyi were not resolved with any de-gree of support.

Unlike ML analyses, the members of the Myothorax subgenuswere resolved as monophyletic (PP = 0.9), however the other sub-genera were not monophyletic, and Epuraea and Urophorus arenested within Carpophilus. The latter has strong support for a rela-tionship with C. obsoletus (PP = 0.93), while the former has moder-ate support for being in a clade that includes C. marginellus, C.lugubris, C. bakewelli, C. discoideus and C. antiquus (PP = 0.63).

3.3.2. 28S rDNAMaximum likelihood inferred a single tree with a �ln likelihood

of 3603.59084 from a GTR + C model with nucleotide frequenciesand rate parameters as shown in Table 3.

The 28S ML analysis showed resolution at the subgeneric level,but did not adequately resolve lower relationships (Fig. 3). Themonophyly of the C. oculatus group was supported by a bootstrapof 85%; C. o. cheesmani was sister to the other two subspecies,the separation supported by a bootstrap of 84.3%, while C. o. gillo-glyi was paraphyletic with respect to C. o. oculatus and did notshow the geographic structure found in the mitochondrial data. Be-yond this, resolution within the genus was extremely poor. C.gaveni, C. maculatus, C. schioedtei and C. robustus were identical intheir 28S sequences and were barely different from C. davidsoni,C. mutilatus and C. nepos. Carpophilus dimidiatus was shown to havean extremely divergent 28S sequence, and a relationship with Epu-raea signata was supported with a bootstrap of 73%.

28S results are much more congruent with subgeneric classifi-cation than COI, though subgenera were not strictly monophyleticin all cases. Species within Carpophilus s. str. came out as monophy-letic, though it also included U. humeralis. The single species withinSemocarpolus, C. marginellus, was sister to, but strongly differenti-ated from the Urophorus–Carpophilus s. str. clade. Ecnomorphuswas paraphyletic with respect to a monophyletic Myothorax(excluding C. dimidiatus).

3.3.3. ITS2Analysis of 53 ITS2 sequences produced a tree with a ML of

� ln = 1478.1793 using a GTR model (Table 3). Adding a gammadistribution to the model did not result in a different topology.

ITS2 data was only gathered for select taxa within the Myotho-rax subgenus. At this level it showed good variation between spe-cies (Fig. 4). Carpophilus o. cheesmani was again shown to be a sistertaxon to the other C. oculatus subspecies, supported by a bootstrap

value of 78.6%. In contrast to the 28S data there was sufficient var-iation in the marker to show intra-specific genetic structure. With-in C. o. oculatus there was found to be a lot of structuring, thoughno geographic influence is apparent with these analyses. Thesespecimens were resolved as being paraphyletic with respect to amonophyletic and very homogeneous C. o. gilloglyi that shows nogeographic structuring.

Unexpectedly, C. schioedtei came out within C. o. oculatus. It sitson a reasonably long branch within the group and may be a case ofhomoplasy, possibly combined with insufficient taxon sampling.

3.4. SH tests

SH-tests of topology were conducted on the COI dataset todetermine the individual and combined significance of Carpophilusand C. oculatus non-monophyly. ITS2 was tested for C. oculatusmonophyly only. Results show that constraining monophyly re-sulted in a significantly worse likelihood score for both markers(Table 5). The change in likelihood score of enforcing C. oculatusmonophyly was little different from that of forcing Carpophilusmonophyly, despite being the result of a single change in the tree,namely the change in position of C. o. cheesmani. Likewise, thecombined changes to the tree were substantially different fromthe optimal, but little different from the individual changes.

3.5. GMYC analysis of COI

Running GMYC with a single threshold on the Bayesian MCCtree resulted in a significantly better fit than the null model(logL = 604.55, 2DL = 22.73; v2 test, d.f. = 4, P < 0). Using multiplethresholds did not result in a significantly better model(logL = 606.35, 2DL = 3.6; v2 test, d.f. = 12, P = 0.99), and the fol-lowing results use the single threshold as the standard. The modelwith the highest likelihood inferred 34 entities, with a confidenceinterval of 14–45 entities. The entities estimated by the highestlikelihood model were broadly congruent with the species bound-aries accepted here, though it overestimated the number of entitieswithin the C. maculatus clade (Fig. 6; leftmost, green bar). Within C.oculatus, the best GMYC model recognised the eastern and westernclades of C. o. gilloglyi as separate entities. When multiple thresh-olds were considered, GMYC greatly overestimated the numberof distinct entities, resulting in dividing C. maculatus and the east-ern clade of C. o. gilloglyi into six entities each, and C. o. oculatusinto three.

An alternative method to reporting GMYC uncertainty is themodel-weighting approach of Powell (2012). Using the resultsfrom both single and multiple-threshold GMYC runs, the model-weighted average number of entities is 32.5 with a variance of63.11, from a total of 155 models. None of the models gave the ex-act taxonomic arrangment proposed here, but models that sup-ported single C. maculatus and C. o. oculatus + C. o. gilloglyientities had a sum of Akaike weights (W+(comb)) of 0.056 (n = 4);compared with W+(subsp) = 0.087 (n = 7) where a C. o. oculatus/C.o. gilloglyi split was supported; and W+(split) = 0.26 (n = 10) whereC. o. gilloglyi was further subdivided into the eastern and westernclades and C. maculatus is split into an Australian and a Pacificclade.

Results from the 2000 randomly selected trees were very vari-able. The number of clusters estimated using a single thresholdranged from 2 to 103. The number most commonly inferred was34, with 141 trees having this number of clusters. The range ofthe number of clusters inferred using multiple thresholds wassmaller than the single threshold, from 2 to 84, but the inter-quar-tile range was greater (21 vs 19.25) (Fig. 5). The number most com-monly inferred was 48, with 70 trees having this number ofclusters. The number of entities covered by the MCC confidence

C. maculatus TUV

C. maculatus AUS

C. maculatus NCL

C. maculatus TUV

C. maculatus TUV

C. maculatus TUV

C. maculatus AUS

C. maculatus AUS

C. schioedtei BIS

C. maculatus NCL

C. maculatus NCL

C. robustus BIS

C. maculatus TUV

C. gaveni NZL

C. o. cheesmani VAN

C. o. gilloglyi KER

C. o. gilloglyi FIJ

C. o. gilloglyi FIJ

C. o. gilloglyi FIJ

C. o. gilloglyi FIJ

C. o. oculatus TON

C. o. oculatus FIJ

C. o. oculatus TUV

C. nepos NCL

C. nepos VAN

C. mutilatus SOC

C. davidsoni AUS

C. mutilatus NAU

C. corticinus USA

C. antiquus USA

C. discoideus USA

C. dimidiatus SOC

Epuraea signata NZL

C. marginellus TUV

C. marginellus FIJ

C. lugubris USA

Carpophilus sp. AY310664

Urophorus humeralis FIJ

C. hemipterus AUS

Stelidota sp. VAN

Aethina concolor VAN

Conotelus sp. USA

Phenolia sp. VAN

100

55

100

88.3

63.8

72.8

60.5

84.9

84.3

89.9

56.5

59

87.6

73.6

55.8

100

59.2

99

0.01

Fig. 3. ML tree of 28S sequences. Bootstrap values are shown above nodes with greater than 50% support. Geographic codes as for Fig. 2.


interval accounted for a large proportion of the variation in thesampled trees. The proportion of the single threshold results with-in the interval was 0.697, while 0.5615 of the variation in themultiple threshold models was within the interval. Likelihood ratiotests between single and multiple threshold models showed thatmultiple thresholds did not result in significantly better modelsin all but two cases.

4. Discussion

Data presented here demonstrate that the subspecies of C. ocul-atus are genetically well differentiated from each other, with COImodel-corrected distances ranging between 20–46%. In all analy-ses, C. o. oculatus and C. o. gilloglyi emerged as sister taxa. Beyondthat however, relationships are unclear. Carpophilus o. cheesmani

C. o. oculatus VANC. o. oculatus FIJ

C. o. oculatus VANC. o. oculatus VANC. o. oculatus NAU



C. schioedtei BISC. schioedtei BISC. schioedtei BISC. o. oculatus FIJ

C. o. oculatus SOCC. o. oculatus FIJ

C. o. oculatus VANC. o. oculatus VANC. o. oculatus FIJ

C. o. oculatus TON

C. o. oculatus TUVC. o. gilloglyi KER

C. o. gilloglyi SOC

C. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi FIJ

C. o. gilloglyi FIJC. o. gilloglyi TUBC. o. gilloglyi FIJC. o. gilloglyi FIJC. o. gilloglyi TUVC. o. gilloglyi FIJ

C. o. gilloglyi TUBC. o. gilloglyi TUBC. o. gilloglyi TUBC. o. gilloglyi KERC. o. gilloglyi KERC. o. gilloglyi TUB

C. o. oculatus VANC. o. oculatus VANC. o. oculatus VANC. o. oculatus VAN

C. o. cheesmani VAN

C. o. cheesmani VANC. o. cheesmani VAN

C. o. cheesmani VANC. gaveni NZL

C. davidsoni AUSC. nepos NCL

C. maculatus SOCC. maculatus AUS

C. maculatus NAUC. maculatus SAMC. maculatus TUV

100

78.6

63.8

73.2

99.186.8

56.699.1

63.3

79.176.2

66

0.01

Fig. 4. ML tree of ITS2 sequences. Bootstrap values are shown above nodes with greater than 50% support. Geographic codes as for Fig. 2.

Table 5SH test results for COI and ITS2 datasets.

Trees log likelihood (ln L) Difference (ln L) p-Value

COIMost likely �7788 0 0.5483C. oculatus monophyly �9118 1330 <0.0001Carpophilus monophyly �9125 1337 <0.0001Both �9120 1332 <0.0001

ITS2Most likely �1491 0 0.4870C. oculatus monophyly �1529 38 0.0046


was closely related to the other two subspecies according to thenuclear datasets, but was placed well outside the oculatus–gilloglyiclade in COI analyses. C. o. oculatus and C. o. gilloglyi are reciprocallymonophyletic in the NT-coded COI dataset, however this mono-phyly breaks down with the modification of the COI dataset, andboth nuclear genes show paraphyly of these two taxa.

Removal of third-codon positions and RY coding are usuallyundertaken to resolve deeper phylogenetic nodes as they reducenoise from faster-evolving sites and eliminate transition bias (Phil-lips et al., 2004). In this instance they were not successful, with nodeeper nodes being resolved with any greater certainty than in thefull NT-coded dataset. However, modified datasets did resolve C. o.gilloglyi as being paraphyletic with respect to C. o. oculatus, anobservation that is consistent with 28S data. This paraphyly hints

Number of clusters

Freq

uenc

y

0 20 40 60 80 100

040

8014

0

Number of clusters

Freq

uenc

y

0 20 40 60 80

020

4060

Fig. 5. Frequency of the number of clusters estimated by GMYC models for 2000 randomly selected trees from the Bayesian MCMC chain. Solid vertical lines show theconfidence interval of the number of clusters estimated by GMYC on the MCC tree. Left: single GMYC threshold; Right: Multiple GMYC threshold.


that incomplete linage sorting may be affecting the gene trees pre-sented here, or suggests an ancestor–descendant relationship be-tween C. o. gilloglyi and C. o. oculatus. Removal of third-codonpositions also led to a decrease in genetic structuring within C. o.gilloglyi and C. o. oculatus, showing that the bulk of the differencesbetween these two taxa are in the faster-evolving third-codon po-sition, while differences between these taxa and C. o. cheesmani in-clude first and second codon positions, confirming that C. o.cheesmani is an older lineage than the other two taxa.

Despite having genitalia and colouration very similar to C. o.gilloglyi, C. maculatus was not inferred as being the sister taxon toC. oculatus. COI analyses did not show any support for nodesbeyond species-level groupings, while analysis of 28S resolved asister group relationship with a clade including C. maculatus,C. robustus, C. schioedtei and C. gaveni.

While the sample sizes and geographic coverage of other Carpo-philus species are not large enough to make definitive conclusionsregarding their evolutionary history, it is interesting to note thecomparative amounts of variation in C. nepos, C. mutilatus and C.maculatus. Carpophilus nepos shows distinct differences in COI be-tween the three specimens from Tahiti, New Caledonia and froma laboratory culture in the USA, with average intra-specific ML dis-tances of 4.2%. This degree of genetic variation suggests that thisspecies may have been present in the Pacific for some time. Bycomparison, C. mutilatus specimens showed a much lower levelof variation with an average ML intra-specific distance of 0.42%, de-spite specimens being collected over a similarly large geographicarea. This much lower level of genetic variation suggests that C.mutilatus may be a relatively recent arrival in the Pacific comparedwith C. nepos, or that it may be more readily dispersed, resulting ingreater genetic connectivity between populations. Carpophilusmaculatus shows genetic structure between island groups andwithin single islands, suggesting that this species has been presentin these areas for a substantial period of time. The greatest degreeof structuring exists between a clade representing two Australianspecimens and a clade composed of Pacific Islands specimens;these two groups differing by a genetic distance of over 10%. Thisseparation is not shown by the nuclear genes sampled, and noother evidence currently exists that would justify consideringthese populations different species. The level of genetic structuringmay represent long isolation of the Australian and Pacific popula-tions of C. maculatus.

Species assigned to the subgenus Myothorax formed a mono-phyletic group in most analyses, though in COI data it was presentwith low support. 28S was reasonably congruent with subgenericclassifications, but did not resolve subgenera as being strictlymonophyletic. These preliminary results suggest that the classifi-cation of Kirejtschuk (2008) may represent natural clades, howevermore comprehensive species sampling within these subgenera isrequired before this can be confirmed.

Our molecular data consistently resolve Urophorus and Epuraeawithin Carpophilus sensu lato. While the first of these genera wasoriginally described as a subgenus of Carpophilus, the second has

always been considered to be distinct, usually being placed in a dif-ferent subfamily to Carpophilus. Urophorus was elevated to a fullgenus by Gillogly (1962), but later workers have debated the valid-ity of this elevation and have continued to consider it as a subge-nus (e.g. Ewing and Cline, 2005). Though the structure of themale terminalia in U. humeralis looks similar to that in Carpophilus,closer investigation shows that the 8th tergite makes up the entirepygidium in U. humeralis, as opposed to the small, ventrally situ-ated button-like struction found in Carpophilus. Urophorus pos-sesses a round patch with significantly different sculpture in thesame location as the 8th tergite in Carpophilus, which completesthe illusion of similarity. This morphological feature provides sig-nificant evidence against the placement of Urophorus within Carpo-philus as inferred from our molecular data. Greater taxon samplingwithin Urophorus and other Carpophilus subgenera and the use of agenetic marker with a conservative rate of evolution is needed toinvestigate this situation further.

Despite being proposed as a nuclear equivalent of the COI re-gion for the identification of species (Sonnenberg et al., 2007),the 28S D1–D2 region proved to be unreliable for distinguishingbetween species-level taxa in this research. This was particularlythe case within the Myothorax subgenus, members of which weredistinctly different in their COI sequences. However, 28S did showpromise for the detection of higher-level phylogenetic relation-ships. The placement of U. humeralis and Epuraea within Carpophi-lus may be explained by insufficient taxon sampling or long branchattraction. The level of divergence of C. dimidiatus is very high.More specimens of this species are required to confirm whetherthis divergence is an error. As a member of the Myothorax subge-nus, it is expected that the species would have been placed withthe remainder of species in that subgenus.

GMYC results were largely congruent with morphological spe-cies. In some instances the model with the highest likelihood over-estimated the number of species when intra-specific geneticstructuring was evident, i.e. within C. o. oculatus + C. o. gilloglyiand C. maculatus. However, models within the confidence intervalof the analysis were more inclusive. The GMYC confidence intervalof the MCC tree adequately reflected the variability in the sample,as reflected by the confidence interval encompassing the bulk ofthe density of the distribution of the results on the sampled trees.The use of the AIC and model weights extends the use of GMYC fromits current use as comparing two models, to enabling the weight ofevidence for different scenarios to be considered in taxonomic deci-sions. In this instance, while there is more evidence from GMYC forrecognising the East/West split in C. o. gilloglyi and the Australia/Pacific split in C. maculatus, the evidence ratios against recognisingthe subspecies (Esplit,subsp = 2.981), or having a combined C. o. ocula-tus + C. o. gilloglyi clade (Esplit,comb = 4.604) is not so high as to be be-yond the realms of possibility. GMYC was originally developed toprovide estimates of species numbers as part of large-scale commu-nity ecology projects where in-depth taxonomic expertise was notavailable and not necessary. In this context it can be a useful tool(Monaghan et al., 2009; Pinzon-Navarro et al., 2010); however its

C. mutilatus HAW

Meligethes aeneus

Meligethes coracinusMeligethes gracilis

C. maculatus HAW

Omosita discoidea

Epuraea signata NZLEpuraea signata NZL

C. davidsoni AUS

C. hemipterus AUS

C. marginellus FIJ

Urophorus humeralis AUS

C. o. gilloglyi FIJ

C. nepos NCL

C. maculatus SAM

C. o. gilloglyi FIJ

C. o. gilloglyi FIJ


C. hemipterus NZL

Stelidota sp. VANBrachypeplus sp. AUS

Epuraea ocularis SOC

C. maculatus TUV

C. maculatus TUVC. maculatus TUV

C. o. oculatus TUV

C. maculatus TUVC. maculatus TUV

C. o. gilloglyi FIJ

C. maculatus NAU

C. mutilatus NAU

C. o. oculatus NAU

C. marginellus TUV

C. maculatus AUSC. maculatus AUS

C. o. gilloglyi SOC

C. maculatus SOC

C. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi TUB

C. o. gilloglyi FIJ

C. o. gilloglyi TUV

C. o. oculatus SOC

C. maculatus BIS

C. schioedtei BIS

C. robustus BIS

C. mutilatus SOC

C. nepos SOC

C. dimidiatus SOC

C. schioedtei BIS

C. o. gilloglyi SOC

C. o. gilloglyi SOC

C. o. gilloglyi SOC


C. o. gilloglyi TON

C. obsoletus BIS

C. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi KER

C. o. gilloglyi FIJ

C. o. gilloglyi FIJ


C. corticinus USA

C. antiquus USA

C. lugubris USAC. lugubris USA

C. nepos USA

C. o. oculatus FIJC. o. oculatus FIJ

C. bakewelli AUS

C. davidsoni AUS

C. gaveni NZL

Conotelus sp. USA

C. discoideus USA

Phenolia sp. VAN

Aethina concolor VANAethina concolor VAN

C. o. oculatus VAN

C. o. oculatus VAN

C. o. oculatus VAN

C. o. oculatus VAN


C. o. gilloglyi TUB

C. o. oculatus VAN

C. o. oculatus VAN


C. o. oculatus VAN

C. o. cheesmani VAN

C. o. oculatus VAN

C. o. oculatus VAN

C. o. cheesmani VAN

C. o. cheesmani VAN

C. o. oculatus FIJ

0.1

1

0.76

0.9

0.561

0.33

0.14

10.57

0.49

10.76

0.11

0.11

1

0.71

1

1

0.730.94

10.420.99

0.950.220.99

1

0.640.35

1

0.99

1

0.460.170.120.08

00

0.08

0.010.08

0.08

0.46

0.70.960.22

0.22

0.210.99

0.160.15

0.47

10.80.1

0.640.960.780.030.11

0.020.030.11

1

10.33

0.14

0.250.230.96

0.360.53

0.030.2

00.07

0.150.83

10.34

0.560.83

1

0.93

0.25

0.56

0.470.66

11

0.820.41

1

0.03

0.63

0.260.33

0.911

0.460.88

1

1

Fig. 6. Bayesian maximum clade credibility tree of NT-coded COI sequences. Posterior probabilities are shown above all nodes. 95% highest posterior density error bars fornode heights are given for nodes with a posterior probability greater than 0.5. Geographic codes are the same as in Fig. 2. Bars to the right of the tree show the composition ofclusters estimated by single-threshold GMYC models (from left to right): highest likelihood GMYC model (green), the lower confidence interval (light blue), and the upperconfidence interval (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)



estimates are very broad, thereby limiting its utility in the contextof a systematic treatment of a taxon.

In the light of the morphological and genetic differences be-tween the subspecies of C. oculatus, it is justifiable to elevate C. o.cheesmani to a full species while retaining C. o. gilloglyi and C. o.oculatus as subpecies. In all analyses, C. o. cheesmani is monophy-letic and substantially different from other C. oculatus subspecies,and can separated by the shape of male genitalia, and by the differ-ences in pronotal punctation (Dobson, 1993b). The elevation of thissubspecies to full species follows the precedent of a Californian ti-ger beetle, Cicindela lunalonga Schaupp, which was elevated frombeing a subspecies of Ci. terricola following molecular evidence ofstrict monophyly and a high pairwise divergence in phylogeneticanalyses (Woodcock et al., 2007). The two other subspecies of C.oculatus are retained at that rank as they show paraphyly, and asan indication that this complex has some interesting biology wor-thy of further study.

There may be some justification for the species determination ofthe C. o. gilloglyi geographic clades. However, no lines of evidenceoutside of COI currently point to species-level differences betweenthe two clades. Specimens from these areas, which appear to beidentical morphologically, have been found in the same habitatand appear to have the same ecological niche. It is possible thatthese populations are in the early stages of allopatric speciation.It is surprising that these two clades are so distinct consideringthe high frequency of human movement between Fiji and Tonga.It is even more puzzling when the large geographical distances be-tween islands inhabited by the eastern clade are considered, overwhich it appears to have maintained recent gene flow. Because ofthe limited sampling in Tonga, more focused collecting throughoutTonga and the Lau archipelago would be needed to determinewhether the apparent geographic separation of the two C. o. gilloglyiclades is real, or if there is some overlap in the ranges of these cladesand to what extent that overlap occurs. More detailed specimensampling, particularly with the use of rapidly evolving geneticmarkers such as microsatellites, may provide clues to the creationand persistance of this deep divergence within C. o. gilloglyi. Shouldthe two clades be found to exist in sympatry, the proportion of indi-viduals belonging to each clade plus the genetic diversity of thesetwo populations and extent to which their distributions overlapmay offer clues as to the processes (e.g. invasion) that currentlyinfluence the distribution of each clade.

5. Conclusion

This research set out to resolve the taxonomic questions sur-rounding the subspecies of C. oculatus, and to determine the degreeof similarity between C. oculatus and C. maculatus. Consistentevidence from three gene regions reveal that C. maculatus andC. o. cheesmani are distinct species, while C. o. oculatus andC. o. gilloglyi are closely related taxa that show paraphyletic struc-ture with genetic markers. These species therefore need to be trea-ted as such in biosecurity operations. While this distinction willnecessitate increased vigilance by biosecurity organisations, thelevel of genetic structuring between islands of the South Pacificsuggests that C. oculatus and C. maculatus may not be as invasiveas previously assumed.

At a higher level, this study suggests that Carpophilus may notbe monophyletic, potentially incorporating the genera Urophorusand Epuraea. This is unexpected as it runs against several lines ofevidence from morphology, but is to be taken with caution asfew specimens and species of Urophorus and Epuraea were sampledfor this study, and these taxa appear with the representatives of apoorly sampled subgenus of Carpophilus. A formal investigation ofthis possibility will require representative samples of all threegenera. It will also need to use genetic markers that have a slower

rate of evolution, as COI showed saturation at the subgenus levelwithin Carpophilus. A somewhat more robust result is that the sub-genus Myothorax appears to be monophyletic. Unfortunately, noother Carpophilus subgenera showed any indication of monophyly,but the sampling scheme was not designed to explicitly test thishypothesis.

Finally, this study further confirms that the South Pacific regionis a dynamic region, and shows that widespread species within theregion can have intruiging biology and can be as interesting andinformative as single island endemics.

5.1. Taxonomy of C. maculatus and C. oculatus sensu lato

The length of specimens were measured from the anterior mar-gin of the pronotum to the posterior margin of the elytra at the ely-tral suture, as the abdomen in preserved specimens can besignificantly distended or contracted. Specimens were receivedon loan from the Oxford University Museum of Natural History(Oxford, Britain) (OUMNH), New Zealand Arthropod Collection(Auckland, New Zealand) (NZAC), MAF Biosecurity New Zealand(Auckland, New Zealand) (MAF), Queensland Museum (Brisbane,Australia) (QM), the University of the South Pacific (Suva, Fiji)(USP) and the Hunterian Museum (Glasgow, Scotland) (HM).

5.1.1. Carpophilus maculatus (Murray, 1864)Diagnosis: L: 1.6–2.2 mm, W: 1–1.4 mm. Colour variable from

yellow–brown to dark brown, often with lighter, T-shaped markingon the elytra. Pronotum length/width ratio 1:1.5, punctured withreniform punctures (Fig. 7a). Prosternal process expanded behindprocoxae. Female pygidium truncate.

Distribution: Bismarck Archipelago (Gillogly, 1969, LUNZ), Car-oline Islands (Gillogly, 1962), Cook Islands (NZAC, Dobson un-publ., LUNZ), Easter Island (HM), Fiji (MAF, Dobson unpubl.,NZAC, LUNZ), Gilbert Islands (Dobson unpubl., Gillogly, 1962),Guam (Gillogly, 1962), Hawaii (Ewing and Cline, 2005; Murray,1864; Nishida and Beardsley, 2002), West Papua (Dobson unpubl.),Kiribati (Tarawa) (NZAC), Marquesas Islands (Dobson unpubl.),Mariana Islands (Gillogly, 1962), Marshall Islands (Gillogly,1962), Nauru (LUNZ), New Caledonia (NZAC, LUNZ, Dobson un-publ., QM), Niue (NZAC, Dobson unpubl., HM), Palau (Gillogly,1962), Papua New Guinea (Dobson unpubl.), Samoa (NZAC, LUNZ,Dobson unpubl., HM), Society Islands (LUNZ, Dobson unpubl.), Sol-omon Islands (Dobson unpubl.), Tokelau (NZAC, HM), Tonga(NZAC, MAF, Dobson unpubl.), Tuamotu Archipelago (Dobson un-publ.), Austral Islands (Dobson unpubl., LUNZ), Tuvalu (NZAC,LUNZ, HM), Vanuatu (Dobson unpubl., OUMNH, NZAC, LUNZ).Christmas Is., Australia, Philippines, Cuba, Cocos-Keeling Is,Indonesia, Nicobar Islands, Mollucas.

Comments: Abundant throughout the Pacific and found on awide variety of fruit and vegetables. Very variable in colouration.This species be distinguished from C. oculatus and C. cheesmani bythe wider pronotum. Illustration of the male genitalia can be foundin Dobson (1955).

5.1.2. Carpophilus cheesmani Dobson 1993b stat. rev.Diagnosis: L: 1.8–2.4 mm, W: 1.0–1.4 mm. Colour dark brown

to black, with lighter, reddish ring on each elytron. Pronotumlength/width ratio 1:1.3, coarsely and densely punctured withreniform punctures (Fig. 7b), microsculpture minutely reticulatecoriarious. Prosternal process rounded. Female pygidium truncate.

Distribution: Vanuatu (Malekula, Ounua, Tanna, Erromango,Efate) (Dobson, 1993b, LUNZ).

Comments: Restricted to Vanuatu where it is sympatric withC. o. oculatus from which it can be differentiated by the sculptureof the pronotum. The two species can be found syntopically,however C. cheesmani is the rarer of the two (SDJ Brown pers.

Fig. 7. Habitus photographs of Carpophilus maculatus, C. cheesmani and C. oculatus, showing detail of pronotum. (a) C. maculatus Rarotonga, Cook Islands. (b) C. cheesmaniEfate, Vanuatu. (c) C. o. oculatus Tongatapu, Tonga. (d) C. o. gilloglyi Rapa, Austral Islands, French Polynesia.


obs.). This species has been collected from the rotting fruit of bread-fruit (Artocarpus altitis), oranges (Citrus sp.) and guava (Psidium guaj-ava), and sweet potato (Ipomoea batatas) tubers.

5.1.3. Carpophilus oculatus oculatus (Murray, 1864)Diagnosis: L: 1.8–2.6 mm, W: 1.0–1.6 mm. Colour variable, from

light brown to black, usually with lighter ring on each elytron.Pronotum length/width ratio 1:1.3, pronotal punctures sparse,round and fine on disc, becoming reniform and coarser towardmargin (Fig. 7c), no impunctate medial line; microsculpture min-utely reticulate coriarious; usually less pubescent. Female pygid-ium truncate.

Distribution: Cook Islands (Rarotonga) (Dobson, 1993b, NZAC),Fiji (Quisenberry, Taveuni, Kadavu, Viti Levu, Vanua Levu) (Dobson,1993b, LUNZ, MAF, USP, NZAC), Hawaii (Dobson, 1993b), Marque-sas Islands (Fatu Hiva) (Dobson, 1993b), Nauru (LUNZ), New Cal-edonia (Grande Terre) (QM, NZAC), Niue (Niue) (NZAC), Samoa(Upolu, Tutuila) (HM, NZAC), Society Islands (Bora Bora, Tahiti)(Dobson, 1993b, LUNZ), Tokelau (Nukunono) (NZAC), Tonga (Ton-gatapu) (Dobson, 1993b, NZAC, MAF, HM, LUNZ), Tuvalu (Rotuma)(LUNZ), Vanuatu (Espiritu Santo, Efate) (LUNZ).

Comments: Widespread throughout the South Pacific and foundon a wide variety of fruit and vegetables.

5.1.4. Carpophilus oculatus gilloglyi Dobson 1993bDiagnosis: L: 1.5–2.4 mm, W: 0.9–1.4 mm. Colour variable, from

light brown to black, usually with lighter ring on each elytron.Pronotum length/width ratio 1:1.3, pronotal punctures on vertexand disc of pronotum close and reniform (Fig. 7d), impunctatemedian line anterior of scutellum; microsculpture reticulate coria-rious, coarser than in other subspecies. Female pygidium emargin-ate medially.

Distribution: Caroline Islands (Ponape Island, Truk) (Dobson,1993b), Cook Islands (Rarotonga, Atiu) (Dobson, 1993b, NZAC,MAF), Easter Island (NZAC), Fiji (Vanua Levu, Taveuni, Viti Levu,Kadavu, Ovalau, Lakeba) (LUNZ, Dobson, 1993b, USP, NZAC),Kermadec Islands (Meyer Island, Raoul Island, Chanters Islets)(Dobson, 1993b, NZAC, LUNZ), Niue (Dobson, 1993b, NZAC),Samoa (Upolu, Savai’i) (NZAC, HM), Society Islands (Bora Bora,Moorea, Tahiti) (Dobson, 1993b, LUNZ), Tonga (Tongatapu, Vava’u,

Eua, Niuafo’ou) (Dobson, 1993b, NZAC, MAF, HM, LUNZ), AustralIslands (Rapa, Rimatara) (LUNZ), Tuvalu (Rotuma) (LUNZ).

Comments: Widespread throughout the South Pacific and foundon a wide variety of fruit and vegetables. While the ranges of C. o.oculatus and C. o. gilloglyi overlap over a significant part of theirrange, they are rarely found syntopically (SDJ Brown pers. obs.).

Acknowledgments

Nick Porch (Deakin University, Australia), Mofakhar Hossein(Department of Primary Industries, Victoria, Australia), Kiteni Kuri-ka (NARI, Papua New Guinea), Christian Mille (IAC, New Caledo-nia), Linette Berukilukilu (VQIS, Vanuatu), Sada Nand Lal (SPC,Fiji), Will Haines (University of Hawaii), Katharina Pötz (Denmark),Maja Poeschko (Ministry of Agriculture, Cook Islands), KuateaneTupola (Ministry of Agriculture, Samoa) and Rudolph Putoa(Département de la Recherche Agronomique, Tahiti) providedspecimens for DNA sequencing. Museum specimens were loanedby James Hogan (Oxford University Museum of Natural History,UK), Shepherd Myers (Bishop Museum, Hawaii), Geoff Monteith(Queensland Museum, Australia), Geoff Hancock (Hunterian Mu-seum, UK), Olwyn Green (MAF Biosecurity, New Zealand) and HildaWaqa (University of the South Pacific, Fiji). Ron Dobson (Glasgow)provided unpublished distribution records of South Pacific Carpo-philus. Alexander Kirejtschuk (Russia) assisted in identifying C.robustus and C. schioedtei. Curtis Ewing (UC Berkeley, United Statesof America) allowed use of unpublished Carpophilus sequences.Rich Leschen (Landcare Research, New Zealand), Rupert Collins(Bio-Protection Research Centre, New Zealand) and John Marris(Lincoln University, New Zealand) provided valuable commentson the research and manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ympev.2012.04.018.




References

Anisimova, M., Gascuel, O., 2006. Approximate likelihood ratio test for branches: afast, accurate and powerful alternative. Systematic Biology 55 (4), 539–552.

Archibald, R.D., Chalmers, I., 1983. Stored product Coleoptera in New Zealand. NewZealand Entomologist 7 (4), 371–397.

Brown, S.D.J., Collins, R.A., Boyer, S., Lefort, M.-C., Malumbres-Olarte, J., Vink, C.J.,Cruickshank, R.H., 2012. Spider: an R package for the analysis of species identityand evolution, with particular reference to DNA barcoding. Molecular EcologyResources 12, 562–565.

Cease, K.R., Juzwik, J., 2001. Predominant nitidulid species (Coleoptera: Nitidulidae)associated with spring oak wilt mats in Minnesota. Canadian Journal of ForestResearch 31 (4), 635–643.

Collins, R.A., Boykin, L.M., Cruickshank, R.H., Armstrong, K.F., in press. Barcoding’snext top model: an evaluation of nucleotide substitution models for specimenidentification. Methods in Ecology and Evolution. <http://dx.doi.org/10.1111/j.2041-210X.2011.00176.x>.

Connell, W.A., 1981. Bibliography of Carpophilus humeralis (Fab.) in support of arevision of the genus Carpophilus Stephens. Bulletin of the Entomological Societyof America 27 (4), 263–266.

Dobson, R.M., 1952. A new species of Carpophilus Stephens (Col. Nitidulidae) fromAustralia. The Entomologist’s Monthly Magazine 88, 256–258.

Dobson, R.M., 1955. The species of Carpophilus Stephens (Col. Nitidulidae) associatedwith stored products. Bulletin of Entomological Research 45, 389–402.

Dobson, R.M., 1993a. New species and subspecies of Carpophilus Stephens(Coleoptera: Nitidulidae) from the Australasian region. Storkia 2, 1–12.

Dobson, R.M., 1993b. A reappraisal of Carpophilus oculatus Murray 1864,(Coleoptera: Nitidulidae) with descriptions of two new subspecies. Storkia 2,26–29.

Dowd, P.F., 2000. Dusky sap beetles (Coleoptera: Nitidulidae) and other kerneldamaging insects in Bt and non-Bt sweet corn in Illinois. Journal of EconomicEntomology 93 (6), 1714–1720.

Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogeneticsand dating with confidence. PLoS ONE 4 (5), e88.

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

Ewing, C.P., Cline, A.R., 2005. Key to adventive sap beetles (Coleoptera: Nitidulidae)in Hawaii, with notes on records and habits. The Coleopterists Bulletin 59 (2),167–183.

Folmer, O., Black, M., Hoeh, W., Lutz, R., Vrijenhoek, R., 1994. DNA primers foramplification of mitochondrial cytochrome c oxidase subunit I from diversemetazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 294–299.

Gillogly, L.R., 1962. Coleoptera: Nitidulidae. Insects of Micronesia 16 (4), 133–188.Gillogly, L.R., 1969. Nitidulidae (Col.) collected by the Noona Dan expedition in the

Philippine and Bismarck Islands. Entomologiske Meddeleiser 37, 242–252.Guindon, S., Gascuel, O., 2003. A simple, fast and accurate algorithm to estimate

large phylogenies using maximum likelihood. Systematic Biology 52 (5), 696–704.

Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor andanalysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41,95–98.

Heled, J., Drummond, A.J., 2010. Bayesian inference of species trees from multilocusdata. Molecular Biology and Evolution 27 (3), 570–580, <http://dx.doi.org/10.1093/molbev/msp274>.

Hossain, M.S., Williams, D.G., 2003. Phenology of carpophilus beetle populations(Coleoptera: Nitidulidae: Carpophilus spp.) in a fruit dump in northern Victoria.Australian Journal of Experimental Agriculture 43, 1275–1279.

James, D.G., Faulder, R.J., Vogele, B., Bartelt, R.J., Moore, C.J., 1997. Phenology ofCarpophilus spp. (Coleoptera: Nitidulidae) in stone fruit orchards as determinedby pheromone trapping: implications for prediction of crop damage. AustralianJournal of Entomology 36, 165–173.

James, D.G., Vogele, B., 2000. Development and survivorship of Carpophilushemipterus (L.), Carpophilus mutilatus Erichson and Carpophilus humeralis (F.)(Coleoptera: Nitidulidae) over a range of constant temperatures. AustralianJournal of Entomology 39, 180–184.

Ji, Y.-J., Zhang, D.-X., He, L.-J., 2003. Evolutionary conservation and versatility of anew set of primers for amplifying the ribosomal internal transcribed spacerregions in insects and other invertebrates. Molecular Ecology Notes 3, 581–585.

Kirejtschuk, A.G., 2001. Notes on the systematics of the African Nitidulidae(Coleoptera). Annales Historico-Naturales Musei Nationalis Hungarici 93, 17–89.

Kirejtschuk, A.G., 2008. A current generic classification of sap beetles (Coleoptera:Nitiduldae). Zoosystematica Rossica 17 (1), 107–122.

Kubatko, L.S., Degnan, J.H., 2007. Inconsistency of phylogenetic estimates fromconcatenated data under coalescence. Systematic Biology 56 (1), 17–24, <http://dx.doi.org/10.1080/10635150601146041>.

Leschen, R.A.B., Marris, J.W.M., 2005. Carpophilus (Nitidulidae) of New Zealand andnotes on Australian species. Tech. Rep. FMA121, MAF Contract report.

Lin, C.-P., Danforth, B.N., 2004. How do insect nuclear and mitochondrial genesubstitution patterns differ? Insights from Bayesian analyses of combineddatasets. Molecular Phylogenetics and Evolution 30, 686–702.

Monaghan, M.T., Wild, R., Elliot, M., Fujisawa, T., Balke, M., Inward, D.J.G., Lees, D.C.,Ranaivosolo, R., Eggleton, P., Barraclough, T., Vogler, A.P., 2009. Acceleratedspecies inventory on Madagascar using coalescent-based models of speciesdelineation. Systematic Biology 58 (3), 298–311, <http://dx.doi.org/10.1093/sysbio/syp027>.

Murray, A., 1864. Monograph of the family Nitidulariae, pt I. The Transactions of theLinnaean Society of London 24, 211–414.

Nagel, J., Peña, J.E., Habeck, D., 1989. Insect pollination of atemoya in Florida. FloridaEntomologist 72 (1), 207–211.

Nishida, G.M., Beardsley, J.W., 2002. A review of the insects and related arthropodsof Midway Atoll. Bishop Museum Occasional Papers 68, 25–69.

Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics andevolution in R language. Bioinformatics 20, 289–290.

Phillips, M.J., Delsuc, F., Penny, D., 2004. Genome-scale phylogeny and the detectionof systematic biases. Molecular Biology and Evolution 21 (7), 1455–1458.

Pinzon-Navarro, S., Barrios, H., Murria, C., Lyal, C.H.C., Vogler, A.P., 2010. DNA-basedtaxonomy of larval stages reveals huge unknown species diversity inneotropical seed weevils (genus Conotrachelus): relevance to evolutionaryecology. Molecular Phylogenetics and Evolution 56, 281–293, <http://dx.doi.org/10.1016/j.ympev.2010.02.022>.

Pons, J., Barraclough, T., Gomez-Zurita, J., Cardoso, A., Duran, D.P., Hazell, S.,Kamoun, S., Sumlin, W.D., Vogler, A.P., 2006. Sequence-based speciesdelimitation for the DNA taxonomy of undescribed insects. SystematicBiology 55 (4), 595–609.

Posada, D., 2008. jModeltest: phylogenetic model averaging. Molecular Biology andEvolution 25, 1253–1256.

Powell, J.R., 2012. Accounting for uncertainty in species delineation during theanalysis of environmental DNA sequence data. Methods in Ecology andEvolution 3 (1), 1–11.

R Development Core Team, 2012. R: A language and environment for statisticalcomputing. R Foundation for Statistical Computing. Vienna, Austria. ISBN 3-900051-07-0. Available from: < http://www.R-project.org/>.

Schliep, K., 2011. Phangorn: phylogenetic analysis in R. Bioinformatics 27 (4), 592–593.

Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods withapplications to phylogenetic inference. Molecular Biology and Evolution 16 (8),1114–1116.

Simon, C., Frati, F., Beckenbach, A.T., 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 AppliedBiology 87, 651–701.

Sonnenberg, R., Nolte, A.W., Tautz, D., 2007. An evaluation of LSU rDNA D1-D2sequences for their use in species identification. Frontiers in Zoology 4 (6), 12.

Woodcock, M.R., Kippenhan, M.G., Knisley, C.B., Foster, J.A., 2007. Moleculargenetics of Cicindela (Cylindera) terricola and elevation of C. lunalonga tospecies level, with comments on its conservation status. Conservation Genetics8, 865–877.

Xia, X., 1998. The rate heterogeneity of nonsynonymous substitutions inmammalian mitochondrial genes. Molecular Biology and Evolution 15 (3),336–344.

http://dx.doi.org/10.1111/j.2041-210X.2011.00176.x

http://dx.doi.org/10.1111/j.2041-210X.2011.00176.x

http://dx.doi.org/10.1093/molbev/msp274

http://dx.doi.org/10.1093/molbev/msp274

http://dx.doi.org/10.1080/10635150601146041

http://dx.doi.org/10.1080/10635150601146041

http://dx.doi.org/10.1093/sysbio/syp027

http://dx.doi.org/10.1093/sysbio/syp027



http://www.R-project.org/

molecular phylogenetics of a south pacific sap beetle species … · molecular phylogenetics of a...

Documents