Phylogeny of the genus Hegeter (Tenebrionidae, Coleoptera) and its colonization of the Canary...

©1996 The Genetical Society of Great Britain.392

*Correspondence: Department de Biologia Ambiental, Universi-tat de les Illes Balears, 07071 Palma de Mallorca, Spain.

Heredity 76 (1996) 392–403 Received 22 August 1995

Phylogeny of the genus Hegeter(Tenebrionidae, Coleoptera) and its

colonization of the Canary Islands deducedfrom Cytochrome Oxidase I mitochondrial

DNA sequences

CARLOS JUAN*, PEDRO OROMI† & GODFREY M. HEWITTPopulation Biology Sector, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, U.K. and

†Departamento de Zoologıa, Facultad de Biologıa, Universidad de La Laguna, Tenerife, Spain

The genus Hegeter comprises 23 species of darkling beetles (Tenebrionidae) endemic to theMacaronesian archipelagos, with 21 of them exclusive to the Canary Islands. We havesequenced 438 bp of the mitochondrial Cytochrome Oxidase I gene in 17 species (24 taxa) ofCanarian Hegeter. Estimates of nucleotide composition, transition/transversion ratios andnucleotide change frequencies are very similar to those found in another tenebrionid Canariangenus Pimelia, indicating that similar molecular mechanisms are driving the sequence evolu-tion. The sequence variation found allows phylogenetic analyses of the genus and the deduc-tion of colonization patterns. These involve sequential island invasion with more rapidestablishment and radiation than found in the related beetles of the genus Pimelia.

Keywords: beetle, Canary Islands, colonization, Cytochrome Oxidase I, DNA sequence,phylogeny.

Introduction

The Canary archipelago appears to be an excellentmodel for studying colonization and speciationprocesses. Endemic Canarian lizards and beetleshave recently been used to construct molecularphylogenies and deduce sequential interislandcolonizations; these can then be compared with thegeological age of the islands (Thorpe et al., 1993,1994; Juan et al., 1995). Several independent datingsstrongly support a decrease in the age of the islandsfrom east to west, probably related to the volcanicactivity produced by a propagating fracture zoneoriginating from the Atlas mountains in NorthAfrica (Anguita & Hernan, 1975). The maximumdatings range from 20 My for Fuerteventura tofewer than 1 My for Hierro (Fig. 1) (Anguita &Hernan, 1975). Variation of mitochondrial Cyto-chrome Oxidase I (CO I) DNA sequences (Juan etal., 1995) of darkling beetles of the genus Pimelia

produces a phylogeny which suggests the sequencein which they colonized the Canary Islands. Thiscomprises an ancestral species, related to the extantP. lutaria endemic to Fuerteventura and Lanzarote,which probably colonized Tenerife. Representativesof its radiation on these islands in turn colonizedGran Canaria and Gomera. From the latter, rela-tively recent colonizations populated La Palma, andfinally from here Hierro was colonized. This patternfits well with the sequential formation and age of theislands, i.e. as they emerged they were colonizedfrom the closest older island, with the exception ofGran Canaria which is older than Tenerife andshould be colonized first according to this model.Clearly, factors like extinctions and secondarycolonizations may modify the original simple overallpattern.

The genus Hegeter comprises darkling beetlesendemic to the Macaronesian islands, with 22species in the Canaries, one species in the SalvagemIslands and another species, H. tristis, distributedfrom the Azores to Cabo Verde including theCanaries and the Saharan coast (Espanol, 1957a,b;

mtDNA PHYLOGENY OF HEGETER CANARIAN BEETLES 393

Oromı, 1982a). The genus has therefore experienceda process of considerable diversification and specia-tion in the Canary archipelago. All but three specieson the Canaries are endemic to a single island; H.tristis is present in all main islands except GranCanaria, where H. grancanariensis occupies the nicheof the former. Hegeter amaroides is found in thewestern islands, Tenerife, Gomera and Hierro, butnot La Palma in which another vicariant species, H.glaber, is present. Hegeter politus is present in Fuer-teventura, Lanzarote and the islets nearby. Thebigger islands of Fuerteventura, Gran Canaria andTenerife show the highest number of exclusive ende-misms, with three, six and seven, respectively,whereas the younger, smaller and more remote LaPalma and Hierro have one endemic species on theformer and none on the latter (Table 1).

The beetles of this genus are flightless sapropha-gous insects adapted to different habitats in theislands. Most are xerophilic and present at low ormedium altitude, but some are present in the highaltitude volcanic zones, e.g. H. tenuipunctatus and H.lateralis of Tenerife (Table 1). Espanol (1957a,b)revised the taxonomy of the group dividing thegenus into three subgenera: Hegeter s. str. Latr.,Pseudotalphila Reitt. and Homalapipleurus Espan.according to the presence or absence of stridulatoryapparatus and epistome morphology. The subgenusHegeter s. str. includes all the species which occur inthe western and central islands (Gran Canaria,Tenerife, Gomera, La Palma and Hierro), whereasthe other two subgenera are exclusively of theeastern islands (Fuerteventura and Lanzarote). Thelarger subgenus Hegeter s. str. has been divided intofive species groups (see Table 1) according to thedifferent morphological characters considered in thetaxonomy of these beetles. Some of the species

groups, in particular the brevicollis complex ofTenerife, have been reported to show gradual transi-tions in the morphological characteristics andremarkable variability of the different taxa.

In this study, we have collected for a molecularphylogenetic analysis 17 Hegeter Canarian species(Table 1), three of them including representatives ofthe taxa present in several islands; H. amaroides andH. politus from all their island distribution and H.tristis from Hierro and the islet of Alegranza nearLanzarote. Also, sequences were obtained for H.politus (Fuerteventura) and H. brevicollis (Tenerife)from three and two individuals, respectively. In total,DNA sequences corresponding to at least 438 bp ofthe COI gene have been determined for 27 differentHegeter haplotypes. Also, two species from differentgenera have been sequenced and tentatively used asoutgroups in the phylogenetic analysis. One is themorphologically related genus Tentyria fromMallorca (Balearic Islands) and the other is aCanarian Pimelia from a previous study. Bothspecies belong to the same subfamily (Pimeliinae,sensu Doyen, 1972), and to the family Tenebrioni-dae, as does the genus Hegeter. The sequencesobtained have been used to construct gene phylo-genies and the results compared to those obtainedfor the Canarian Pimelia in order to look for generalpatterns of DNA variation and colonization of theCanaries by tenebrionids.

Materials and methods

Sampling

The Hegeter species were collected from the CanaryIslands as listed in Table 1 (a full list of the collec-tion localities can be provided on request). Singleindividuals were used for amplification and sequenc-ing, except for H. politus, H. tristis, H. amaroides andH. brevicollis, in which more than one individual wasseparately analysed. The sequences used forcomparative purposes are those of Tentyria schaumiand Pimelia laevigata. The insects were preserved inabsolute ethanol or sent alive by mail to thelaboratory.

DNA preparations, PCR amplifications andsequencing

DNA was extracted from the heads and thoraces ofindividual beetles by standard phenol extraction andethanol precipitation. Two different portions of themitochondrial COI were used for PCR amplifica-tions. The oligonucleotides used to amplify fragment

© The Genetical Society of Great Britain, Heredity, 76, 392–403.

Fig. 1 Map of the Canary archipelago. Numbers refer tothe proposed maximum geological ages estimated for themain islands (My) (Anguita & Hernan, 1975).

394 C. JUAN ET AL.

1 (255 bp long including primers) were 5p-CCT-A C A G G A A T T A A A G T T T T T A G A T G A T T A G C - 3pand 5p- A T A G G G G G A A T C A G T G A A C T A G T C C -3p, their 5p end positions corresponding to sites 2410and 2665 of the Drosophila yakuba mitochondrialgenome. Fragment 2 (311 bp long) corresponds tothe 3p end of the COI gene. The oligonucleotidesused in this case were 5p-GTAAACCTAA-CATTTTTTCCTCAACA-3p and 5p-TCCAATG-CACTAATCTGCCATATTA-3p (primers UEA9and UEA10 of Lunt et al., 1996), with positions forthe 5p ends in the D. yakuba mitochondrial genomeof 2727 and 3038, respectively. Each PCR cycle

comprised denaturation at 94°C for 30 s, annealingat 50°C for 1 min, and primer extension at 72°C for1 min. Purified single-stranded DNA was preparedby using one of the two primers biotinilated at its 5pend in the PCR reaction. The two DNA strandswere separated with magnetic beads coated withstreptoavidin (DYNAL) according to the manufac-turer’s recommendations. The sequences weregenerated using the dideoxy Sanger method withfluoresceinated sequencing primers and a PharmaciaALF automatic sequencer. Most of the individualswere sequenced more than once from independentPCR amplifications. Sequence data were analysed by


Table 1 List of Hegeter species, their taxonomic groupings (Espanol 1957a,b), island distribution, habitat and number ofindividuals used in this study with their island provenance

Species & sp. groupings Island Habitat No. individuals

Subgenus Hegeter s. str.tristis group

H. tristis † F, L, T, H, LP a, b 2 (A, H)H. grancanariensis C a, b 1H. webbianus

amaroides group C b, c 1H. amaroides T, G, H b, c, d, e 3 (T, G, H)H. glaber

brevicollis group LP b, c, d 1H. brevicollis T b 2H. brevicornis‡ G bH. gomerensis G b, c, e 1H. proximus‡ T bH. excisus‡ T b, e, dH. transversus T d, e 1H. tenuipunctatus T f 1H. lateralis T f 1H. intercedens

impressus group T a, b 1H. costipennis C b, c 1H. impressus C b 1H. subrondatus

abbreviatus group C a 1H. abbreviatus‡ C eH. latebricola‡ S —

Subgenus PseudotalpophilaH. plicifrons F b 1H. fernandezi F c 1

Subgenus HomalapipleurusH. politus F, L, A b 5 (F, L, A)H. gonzalezi‡ F a

†H. tristis is found in other East African (Macaronesian) archipelagos.‡Not included in the present study.Abbreviations for the islands are as follows: A, Alegranza; C, Gran Canaria; F, Fuerteventura; G, Gomera; H, Hierro; T,Tenerife; LP, La Palma; S, Salvagem Islands. Abbreviations for habitats are: a, coastal dunes; b, xerophilous low ormedium altitude; c, mountainous, deforested areas; d, mountainous, pine forest; e, humid forest; f, mountainous, highaltitude.


multialignment with the Lasergene program MegA-lign (DNAstar, 1992).

Phylogenetic analyses

Maximum Parsimony analysis of the sequence datawas performed using the PAUP computer softwarepackage (Swofford, 1993). Because of the number oftaxa involved in the analyses we performed heuristicbootstrap searches with random addition of taxawith 10 repetitions for each of 100 replications.Substitution and character weighting were tried inPAUP, creating step matrices that give particular‘cost’ values to transversions with respect to transi-tions and giving less weight to third codon positionswith respect to first and second ones. The MACCLADE

program (Maddisson & Maddison, 1992) was usedto test the number of extra steps required for alter-native topologies to the most parsimonious tree(MPT). The ‘Stasis and Changes’ option in theChart menu was used to estimate the number oftransitional and transversional changes deducedfrom trees.

Total substitution rates were calculated using theKimura 2-parameter model (Kimura, 1980). Pheno-grams based on sequence distances were constructedusing the Neighbour-Joining (NJ) (Saitou & Nei,1987) procedure with the Phylogeny InferencePackage (PHYLIP) (Felsenstein, 1993). Bootstrapvalues were based on 500 replications. Nonsynon-ymous substitution rates were estimated by themethod of Li et al. (1985), with the modificationsproposed by Li (1993). Relative rate tests (Sarich &Wilson, 1967) were performed according to themethod of Wu & Li (1985).

ResultsSequence variation

A total of 438 nucleotide positions of the COI genehave been examined in 22 sequences from 17 Canar-ian Hegeter species as well as two outgroup species(Fig. 2). Of these, 166 were polymorphic, giving atotal of 130 phylogenetically informative characters.Among the Canarian sequences 139 positions varied,21 of them at first codon positions whereas all theremaining changes were at third codon positions.The transition/transversion ratio estimated from theMPT obtained was 2.8, with most changes (35 percent) being from T to C. The sequences have anaverage frequency of A+T nucleotides of 65 percent across all taxa, increasing to about 80 per centat third codon positions. In a previous study includ-ing 365 bp of COI from Canarian Pimelia, a genus

belonging to the same beetle subfamily, similarfrequencies of A+T, transition/transversion ratios,types and frequencies of substitutions were found(Juan et al., 1995). When comparing the two datasets, which are coincident for approximately 200 bp,it appears that 64 per cent of the total polymorphicsites vary in both Hegeter and Pimelia at exactly thesame positions (75 per cent of these are at thirdcodon positions and 40 per cent involve TkC transi-tions only in both genera).

Comparison of the inferred amino acid sequencesof COI obtained in Hegeter showed that fragment 1is invariant among the Canarian sequences, whichshared seven amino acid differences with respect tothe outgroups. Fragment 2 is more variable at theamino acid level; there are 19 variable amino acidpositions, four of them showing amino acid differ-ences among the Canarian Hegeter. All but three ofthe amino acid substitutions are conservative. Inamino acid position number 47 (nucleotide positions139–141) all species endemic to Gran Canaria(except H. webbianus) plus H. politus from Fuerte-ventura have serine, while the remaining Canarianspecies have alanine. At amino acid position 67(nucleotide positions 199–201) all Tenerifeanendemic species except H. intercedens have threo-nine, whereas the other groups have serine orisoleucine (H. glaber, La Palma). Therefore, twoisland groups of species (Gran Canaria and Tener-ife) are defined by particular amino acidreplacements.

Intraspecific variation

Four species were sequenced for more than oneindividual. Hegeter politus sequences were obtainedfrom three individuals from the same population ofFuerteventura plus one each from Lanzarote andthe islet of Alegranza. Hegeter amaroides sequencescame from three individuals collected in Tenerife,Gomera and Hierro, respectively. Hegeter tristis wassequenced in Hierro and Alegranza individuals andH. brevicollis (Tenerife) in two individuals from thesame population. The intraspecific variation was0.8–1.5 per cent for species that occur in a singleisland (not shown) and 7–9.5 per cent for individ-uals of the same species but from different islands.In all cases in which more than one individual wassequenced for a given species, the haplotypes fromthose individuals formed a monophyletic group.

Phylogenetic analysis

An heuristic search using PAUP and assuming an


396 C. JUAN ET AL.


Fig. 2 Sequence alignment of COI DNA sequences in species of Hegeter plus two outgroup species used for comparison(Tentyria and Pimelia). Identity to the sequence of H. subrondatus is indicated by a dot. For intraspecific variation, only thesequences of species from different islands are shown. Abbreviations for the islands are as in Table 1.

equal weight for all characters produced two MPTsequal in length (580 steps, consistency index = 0.44,retention index = 0.49). The topology of the boot-strap tree supported monophyly for the Fuerteven-tura/Lanzarote species group which occupies anancestral position in the tree, with all the otherclades, namely Gran Canarian, Tenerifean andwestern islands species groups, being derived fromone another in a monotonic way (Fig. 3a). However,bootstrap values were low (less than 50 per cent) formost of the basal nodes and some of the alternativetopologies were only a few steps apart. The tree alsosuggests a polyphyletic origin of the Gran Canariangroup of species and relates H. webbianus to theTenerifean lineage. Fitch & Ye (1991) have argued

that weighted parsimony is more likely to infer thecorrect or a better tree than equally weighting allcharacters and state changes. Weighting reduces thelevel of noise in a data set by giving emphasis to therare events, which in coding sequences are generallytransversions and substitutions at first and secondcodon positions. Thus, for substitution weighting thetransition/transversion ratio estimated from theMPT was used, giving a weight of 3:1 of transver-sions vs. transitions. Similar native transversion/tran-sition ratios for COI have been estimated in othercoleopterans (Funk et al., 1995). Also, the more vari-able third codon positions were given less weightthan the conserved first and second positions.

The most consistent trees were obtained when



Fig. 2 continued.

398C

. JUA

N ET A

L.

© T

he Genetical Society of G

reat Britain, H

eredity, 76, 392–403.

Fig. 3 (a) Shortest tree showing the phylogenetic relationships within Canarian species of Hegeter based on comparisons of 438 bp of COI DNA sequencesassuming an equal weight for all characters. (b) One of the three MPTs obtained assuming a 3:1 weight of transitions over transversions and 5:1 of first andsecond codon positions vs. third positions. The number of informative nucleotide substitutions is given both for transitional (listed first) and transversional(listed second) changes below the branches. (c) Neighbour-Joining phenogram obtained using the matrix of nucleotide divergence pairwise comparisonscorrected for multiple hits by the formula of Kimura (1980). In all trees the letters in parentheses indicate the island provenance of each species (sameabbreviations as in Table 1). Tentyria and Pimelia are used as outgroup species. Numbers above the nodes refer to the percentage bootstrap values (valueslower than 50 per cent are not shown).


combining both weighting methodologies (weights of3:1 transversion vs. transition; 5:1 first and secondcodon positions vs. third codon positions). The boot-strap shortest tree obtained in this search ispresented in Fig. 3b. This tree topology is similar tothe equally weighted tree. The main groups aresupported by relatively few nucleotide substitutions(2–5) and moderate bootstrap values (Fig. 3b). Ithas been suggested that increasing the weight givento transversions results in better resolved trees butcompletely ignoring transitions can give less resolu-tion (Funk et al., 1995). In our case, the topology ofthe five MPTs obtained when only considering trans-versions was remarkably similar to the previous onesbut differed by forcing H. webbianus into the GranCanarian species group, making it monophyletic.The program MACCLADE was used to look for alter-native topologies, changing the relative positions ofthe main clades. Swapping the Fuerteventura/Lanzarote and Gran Canarian lineages implied twoextra steps in the length of the tree. All otherchanges in the relationships between groupsincreased significantly the number of nucleotidesubstitutions necessary to support alternative topolo-gies. Finally, considering nonsynonymous substitu-tions exclusively gave a poor phylogenetic resolution

because of the low number of informative charactersin the data set.

Overall, we favour the topology shown in Fig. 3b,which relates the Fuerteventura group to the GranCanarian one; this then joins to the Tenerife+wes-tern islands species group. In the latter group, thewithin lineages relationships are difficult to establishunambiguously because they are less stable whentrying different weighting strategies. Nevertheless, asubgroup consisting of closely related Tenerifeanspecies is supported by six substitutions and a boot-strap value of 100 per cent, and a group formed byH. amaroides (Tenerife, Gomera and Hierro), H.tristis (Hierro and Alegranza) and H. glaber (LaPalma) is supported by five nucleotide substitutions.

A phenetic analysis with the NJ method using thedistance matrix calculated by the 2-parameterformula of Kimura (the matrix is available from theauthors on request) was performed. The maximumand minimum evolutionary distances betweenspecies are 20.13 per cent between H. politus(Lanzarote) and H. costipennis (Gran Canaria) and3.26 per cent between H. lateralis and H. tenuipunc-tatus (Tenerife), respectively. The topology of theNJ tree obtained (Fig. 3c) is similar to the oneobtained by the parsimony analysis, resulting in the


Fig. 3 continued.

400 C. JUAN ET AL.

same relative positions of the main groups butdifferent in the position of H. webbianus, hereplaced in the Gran Canarian group. The internodelengths separating the island groups are very short,indicating a rapid speciation and radiation of thespecies in the Canaries, and therefore making itdifficult to achieve enough resolution to establishthe between island group relationships consistently.Relative rate tests were performed for 10 compari-sons between taxa which occur in the same ordifferent islands. The results showed no evidence ofrate heterogeneity either at 4-fold or 2-fold degen-erate positions (not shown), and therefore a rela-tively steady molecular clock can be assumed. Figure4 shows a plot of the maximum evolutionary distan-ces between taxa on different islands based on all (Ain Fig. 4) or only nonsynonymous (B in Fig. 4) siteswith respect to the maximum estimate of age for theyounger island in each comparison.

Discussion

Sequence variation

The estimates for sequence variation, averagenucleotide composition, transition/transversionratios and frequency of each nucleotide change areremarkably similar in the COI DNA sequences ofthe Canarian Hegeter (this paper) and Pimelia (Juanet al., 1995). Furthermore, the transition biasdescribed in the mitochondrial DNA of other insects(DeSalle et al., 1987; Tamura, 1992; Brown et al.,1994; Funk et al., 1995) is clearly present in theseCanarian tenebrionids. Also, a comparison of thesame nucleotide positions sequenced in the twobeetle data sets shows that a majority of substitu-tions are occurring at the same sites and are of thesame type. This strongly suggests that the molecularmechanisms driving the evolution of these mitochon-drial sequences are very similar in the two generaand encourages the comparison of the inferencesmade from the sequence variation of the two datasets.


Fig. 4 Relationship between time of divergence estimatedby the maximum geological age of individual Canaryislands (values in Fig. 1) and maximum sequence diver-gence comparisons between taxa in different islands. Thesequence divergence used between Tenerife and La Palmais the estimate obtained comparing the morphologicallyrelated Hegeter amaroides (Tenerife) and H. glaber (LaPalma). The sequence divergence between the two formsof H. amaroides in Gomera and La Palma was used forthe comparison between these islands. (a) Curve obtainedconsidering the total sequence distances; (b) curveconsidering the nonsynonymous distances.

The COI protein, which is involved in the mito-chondrial respiratory chain, has been divided into 25structural regions falling into five different catego-ries (Saraste, 1990; Lunt et al., 1996). The sequencevariability found among the Hegeter species ismaximum in the last 120 nucleotides of fragment 2,which is coincident with the 3p end of the gene. Highvariability of the 3p end of both subunits of CO(COI and COII) has been previously reported atdeep phylogenetic levels (different orders of insects)(Liu & Bechenbach, 1992; Lunt et al., 1996). Thepresent results show that variability is also present,at least for COI, at the level of closely relatedspecies.

Phylogenetic inferences

Parsimony analysis supports monophyly for the threeisland variants of H. politus (bootstrap value 100 percent). The evolutionary distances between thesequences of individuals of H. politus from differentislands are high (5.43–11.58 per cent), indicatingthat this species has been diverging anagenically fora long period of time in the eastern islands. Itsuggests an origin for the species in Fuerteventura,which is the oldest island in the archipelago (20My).

The group of species occurring in the easternCanary islands, H. politus, H. plicifrons and H.fernandezi (the Fuerteventura group), is a monophy-letic clade for the plicifrons–politus taxa, but the


position of fernandezi is not well resolved. Hegeterfernandezi contains the most diverged Hegeter COIsequence in the Canaries, which perhaps is relatedto the ancestral colonizer of Fuerteventura. Inter-estingly, this species occurs in a very restricted areaon the top of the mountains of the Jandıa peninsula(Fuerteventura). Hegeter politus is remarkable inhaving a stridulatory plectrum on the posteriorfemurs which is absent from all the other Hegetertaxa (Espanol, 1957a,b). Hegeter politus and H. plici-frons are probably the result of radiation of anextinct species on the eastern islands, which alsocolonized Gran Canaria. This explanation is alsosupported by the fact that the same inferred aminoacid replacement at amino acid position 47 (frag-ment 2), which defines the Gran Canarian group, ispresent in the sequence of H. politus(Fuerteventura).

Those species occurring in Gran Canaria, namelythe Gran Canarian group, H. subrondatus, H. gran-canariensis, H. impressus and H. costipennis, areprobably a monophyletic group (60 per cent boot-strap value in the parsimony analysis), although theNJ tree relates H. subrondatus to H. webbianus in 65per cent of the bootstrap replications. Hegeterwebbianus itself seems to be the result of a different,older colonization event, and maybe a member of itsradiation colonized Tenerife. This molecular phylo-geny disagrees with the opinion of taxonomists(Espanol, 1957a,b) who considered H. webbianusclosely related to H. grancanariensis and H. tristis,possibly even a single species. The phylogeneticposition of H. amaroides from Tenerife is not clearas different results have been obtained by parsimonyand distance methods, and the parsimony treesthemselves differ in the position of this taxon.Hegeter amaroides from Gomera and Hierro are,however, a monophyletic grouping (bootstrap valueof 100 per cent).

The relationships of the five exclusive endemicspecies of Tenerife with the Gomera species H.gomerensis and H. glaber from La Palma are moreobscure. Within Tenerife, H. lateralis, H. tenuipunc-tatus, H. brevicollis and H. transversus form a closelyrelated monophyletic group, the result of a relativelyrecent radiation to which H. gomerensis seems to berelated. This supports the distinctness of the latterspecies, considered a mere variety by Wollaston(1864) and Espanol (1957a). All exclusive Tener-ifean species but one show the same specific aminoacid replacement at position 67 (fragment 2) whichdefines the group in the same manner as describedabove for the Gran Canarian clade. Finally, H.glaber from La Palma is related to the hetero-

geneous group of taxa which occur in several islands(H. tristis and H. amaroides).

The phylogenetic relationships among the mainspecies groups contain many uncertainties, as shownby the weak bootstrap values obtained, which inmany cases coincide with very short internodelengths in the distance tree. This is partly caused bythe high proportion of transitional substitutions,which give information about recent divergences butintroduce significant levels of noise for inferringancient relationships (Funk et al., 1995). It seemsunlikely that the number of characters used in theanalysis is insufficient to establish the relationshipsamong groups, because a similar study with a shorterDNA sequence (365 bp) of the same COI gene wasable to resolve the interisland group relationships inthe genus Pimelia (Juan et al., 1995). Short inter-node lengths have been explained as the result of arapid origin and radiation of lineages (Kraus &Miyamoto, 1991; Barrio et al., 1994; Emerson &Wallis, 1995). Pamilo & Nei (1988) have argued thata tree constructed from DNA sequences of a singlegene can differ considerably from the species tree ifthe time of divergence between species is short. Thismakes it difficult to reach the resolution required toclarify the phylogenetic relationships between groupsin fast cladogenetic lineages, even when large DNAsequences are used in the analysis.

If we assume such an evolutionary scenario ofrapid radiation for the Hegeter taxa, it would implythat individual Canary islands were colonized soonafter they emerged by migrants from a close existingisland, followed by within-island radiations andspeciation events. In the case of the more habitat-diverse Gran Canaria and Tenerife islands, thiswould have probably occurred relatively quicklyaccording to this explanation. The COI phylogeniesobtained for the Canarian Pimelia differ with respectto the Hegeter ones in having a better resolution ofthe different clades, as shown by the higher statis-tical support of the main island species groups (Juanet al., 1995). Hegeter seems to be a better colonizerthan Pimelia of empty habitats, including lava flowsof historical age with very little vegetation, and isalso found on the Lanzarote islets (Alegranza andMontana Clara) from which Pimelia is absent. Thiscould be because of different life history character-istics and/or reproductive strategies in the twogenera, and may account for the observed differ-ences in their mitochondrial phylogenies.

Despite the uncertainties detailed above and thelimitations of using few individuals per taxon, allparsimony and distance solutions suggest a coloniza-tion in the direction FuerteventurahGran Canar-


402 C. JUAN ET AL.

iahTenerifehGomera. These sequential coloni-zations are the most plausible explanation when weconsider both the timing of the islands’ formationand their geographical distribution in the archi-pelago (see Fig. 1). Assuming that a molecular clockexists and exhibits the generally accepted rate ofsubstitution of 2 per cent/My for mtDNA (DeSalleet al., 1987), we can estimate the colonization timesfrom the maximum evolutionary distances betweenspecies on different islands. Lanzarote and GranCanaria would have been colonized by Fuerteven-turan ancestors at a maximum of 8.5–9 My, Tenerifeat about 8 My and Gomera 6.3 My. These estimatesare in all cases lower than the accepted physicaldatings for these islands. They also imply only asmall time difference between the colonizations ofGran Canaria and Tenerife (see above). In thegenus Pimelia (Juan et al., 1995) a similar coloniza-tion pattern was deduced, except that Tenerifeseems to have been colonized before Gran Canariaby ancestors in Fuerteventura (at an estimated 8 MyBP), and was the source of colonizers for GranCanaria and Gomera at maximum estimated timesof 7.5 and 5.5 My, respectively.

The evolutionary distances between the closelyrelated Tenerifean species indicate that their radia-tion is quite recent (1–3 My). The mountainousspecies H. lateralis and H. tenuipunctatus probablyappeared from coastal ancestors after the last seriesof volcanic eruptions which formed the Las Canadasvolcanic edifice, beginning around 2 My ago (Anco-chea et al., 1990).

On the other hand, the colonizations of theyounger islands of Hierro and La Palma, and therelationship of H. gomerensis to the extant Tener-ifean species are not so clear. All the estimatesobtained for time of divergence seem to be higherthan the age of the islands, which are the youngest(1–2 My) in the archipelago. This could result fromacceleration of the sequence divergence caused byfounder effects, but the relative rate test provides noevidence of this. Another possibility is that COIsequences have a faster rate of substitution than 2per cent/My. This may be so, but to explain somelevels of sequence divergence it would have to be inthe order of 5 per cent/My. An example of this is H.glaber (La Palma) and H. amaroides (Tenerife) with9 per cent divergence, with La Palma being only 2My old. Espanol (1957a,b) considered H. glaber tobe morphologically very close to H. amaroides. Itseems unlikely that the COI region will have accel-erated in certain cases (see relative rate test), and amore likely explanation is that the glaber and amar-oides lineages diverged on Tenerife and that of

glaber colonized La Palma during the last 2 My. Ifthis is so, then maybe a glaber-like lineage sequencestills exists on Tenerife: further work can explorethis possibility.

The distribution of H. tristis has been suggested tobe the result of human transportation (Espanol,1957a,b; Oromı, 1982a,b). The species is clearlyanthropophilous, and has a widespread distributionin all Macaronesian islands and in some parts of theSaharan coast. We have so far examined the H.tristis DNA sequences from two islands only(Alegranza, c. 15 My and Hierro, 1 My and they aremonophyletic with a bootstrap value of 94 per cent).No definitive proof or refutation of supposed humantransportation can be given yet, and the evolutionarydistances and time of divergence estimates obtainedfor this species have to be regarded with somecaution. But its phylogenetic position from our COIdata suggests that it has arisen relatively recently onthe western islands of the Canaries.

The overall picture of colonization of the Canariesby this tenebrionid beetle Hegeter, revealed from itsmtDNA phylogeny, is of sequential island invasionwith more rapid establishment and radiation than itsmore ponderous cousin Pimelia. It will be interestingand informative to compare this picture with datawe will obtain from nuclear sequence analysis.

Acknowledgements

We thank Barry Richardson, Brent Emerson andKamal Ibrahim for discussions and help withcomputer data analysis. Ken Wolfe provided theLi93 program, modified by adding an insect mito-chondrial genetic code option, Joan Pons collectedspecimens of Tentyria and Diane Alden helped withfigures. C. J. was supported by an European Unionfellowship. The work has been financed by an EUBiotech grant (G.M.H.) and Spanish DGCYT PB93–0419 (P.O.).

References

ANCOCHEA, E., FUSTER, J. M., IBARROLA, E. ET AL., 1990.Volcanic evolution of the island of Tenerife (CanaryIslands) in the light of new K-Ar data. J. Volcanol.Geotherm. Res., 44, 231–249.

ANGUITA, F. AND HERNAN, F. 1975. A propagating fracturemodel versus a hot spot origin for the Canary Islands.Earth Planet. Sci. Lett., 27, 11–19.

BARRIO, E., LATORRE, A. AND MOYA., A. 1994. Phylogeny ofthe Drosophila obscura species group deduced frommitochondrial DNA sequences. J. Mol. Evol., 39,478–488.

BROWN, J. M., PELLMYR, O., THOMPSON, J. N. AND HARRISON,



R. G. 1994. Phylogeny of Greya (Lepidoptera: Prodox-idae), based on nucleotide sequence variation in mito-chondrial Cytochrome Oxidase I and II: congruencewith morphological data. Mol. Biol. Evol., 11, 128–141.

DESALLE, R., FREEDMAN, T., PRAGER, E. M. AND WILSON, A.C. 1987. Tempo and mode of sequence evolution inmitochondrial DNA of Hawaiian Drosophila. J. Mol.Evol., 26, 157–164.

DOYEN, J. T. 1972. Familial and subfamilial classification ofthe Tenebrionoidea (Coleoptera) and a revised genericclassification of the Coniontini (Tentyriidae). Quaest.Entomol., 8, 357–376.

EMERSON, B. C. AND WALLIS, G. P. 1995. Phylogenetic rela-tionships of the Prodontria (Coleoptera; Scarabaeidae;subfamily Melolonthine), derived from sequence varia-tion in the mitochondrial cytochrome oxidase II gene.Mol. Phylogen. Evol. (in press).

ESPANOL, F. 1957a. Los Hegeter de las Canarias Orientales(Col. Tenebrionidae). Trabajos del Museo de Zoologıa(N. S. Zoologica), 2, 1–16.

ESPANOL, F. 1957b. Contribucion al conocimiento de losTentyriini de las Canarias Orientales: sobre el preten-dido Gen. Pseudotaltophila Reitt. ‘EOS’ Revista Espa-nola de Entomologıa, 33, 157–176.

FELSENSTEIN, J. 1993. PHYLIP (Phylogeny InferencePackage). Version 3.5c. University of Washington,Seattle.

FITCH, W. M. AND YE, J. 1991. Weighted parsimony: Does itwork? In: Miyamoto, M. M. and Cracraft, J. (eds)Phylogenetic Analysis of DNA Sequences, pp. 147–154.Oxford University Press, Oxford.

FUNK, D. J., FUTUYMA, D. J., ORTI, G. AND MEYER, A. 1995.Mitochondrial DNA sequences and multiple data sets:a phylogenetic study of phytophagous beetles (Chryso-melidae: Ophraella). Mol. Biol. Evol., 12, 627–640.

JUAN, C., OROMI, P. AND HEWITT, G. M. 1995. MitochondrialDNA phylogeny and sequential colonization of CanaryIslands by darkling beetles of the genus Pimelia (Teneb-rionidae). Proc. R. Soc. B., 261, 173–180.

KIMURA, M. 1980. A simple method for estimating evolu-tionary rate of base substitution through comparativestudies of nucleotide sequences. J. Mol. Evol., 16,111–120.

KRAUS, F. AND MIYAMOTO, M. M. 1991. Rapid cladogenesisamong Pecoran ruminants: evidence from mitochon-drial DNA sequences. Syst. Zool., 40, 117–130.

LI, W.-H. 1993. Unbiased estimation of the rates of synony-mous and nonsynonymous substitution. J. Mol. Evol.,36, 96–99.

LI, W.-H., WU, C.-I AND LUO, C.-C. 1985. A new method forestimating synonymous and nonsynonymous rates of

nucleotide substitution considering the relative likeli-hood of nucleotide and codon changes. Mol. Biol. Evol.,2, 150–174.

LIU, H. AND BECKENBACH, A. T. 1992. Evolution of themitochondrial cytochrome oxidase II gene among 10orders of insects. Mol. Phylogen. Evol., 1, 41–52.

LUNT, D. H., ZHANG, D.-X., SZYMURA, J. M. AND HEWITT, G.M. 1996. The insect COI gene: evolutionary patternsand conserved primers for phylogenetic studies. InsectMol. Biol. (in press).

MADDISON W. P. AND MADDISON, D. R. 1992. MACCLADE:Analysis of Phylogeny and Character Evolution. Version3.1. Sinauer Associates, Sunderland, MA.

OROMI, P. 1982a. Los Tenebrionidos de las Islas Canarias.Instituto de Estudios Canarios, 50 Aniversario, 1,265–292.

OROMI, P. 1982b. Distribucion de los Tenebrionidae (Cole-optera) en las Islas Atlanticas. Boletim Sociedade Portu-guesa Entomologia, 7, 215–231.

PAMILO, P. AND NEI, M. 1988. Relationships between genetrees and species trees. Mol. Biol. Evol., 5, 568–583.

SAITOU, N. AND NEI, M. 1987. The Neighbor-Joiningmethod: a new method for reconstructing phylogenetictrees. Mol. Biol. Evol., 4, 406–425.

SARASTE, M. 1990. Structural features of cytochromeoxidase. Q. Rev. Biophys., 23, 331–366.

SARICH, V. M. AND WILSON, A. C. 1967. Rates of albuminevolution in primates. Proc. Natl. Acad. Sci. U.S.A., 58,142–148.

SWOFFORD, D. 1993. PAUP: Phylogenetic Analysis UsingParsimony. Version 3.1.1. Smithsonian Institution,Washington DC.

TAMURA, K. 1992. The rate and pattern of nucleotidesubstitution in Drosophila mitochondrial DNA. Mol.Biol. Evol., 9, 814–825.

THORPE, R. S., McGREGOR, D. P. AND CUMMING, A. M. 1993.Molecular phylogeny of the Canary Island lacertids(Gallotia): mitochondrial DNA restriction fragmentdivergence in relation to sequence divergence andgeological time. J. Evol. Biol., 6, 725–735.

THORPE, R. S., McGREGOR, D. P., CUMMING, A. M. AND

JORDAN, W. C. 1994. DNA evolution and colonizationsequence of island lizards in relation to geologicalhistory: mtDNA RFLP, cytochrome B, cytochromeoxidase, 12S rRNA and nuclear RAPD analysis. Evolu-tion, 48, 230–240.

WOLLASTON, T. V. 1864. Catalogue of Canarian Coleoptera.John van Voorst, London.

WU, C.-I AND LI, W.-H. 1985. Evidence for higher rates ofnucleotide substitution in rodents than in man. Proc.Natl. Acad. Sci. U.S.A., 82, 1741–1745.


Phylogeny of the genus Hegeter (Tenebrionidae, Coleoptera) and its colonization of the Canary...

Documents

Transcript of Phylogeny of the genus Hegeter (Tenebrionidae, Coleoptera) and its colonization of the Canary...