Phylogenetic Analysis and Trait Evolution in Australian Lineages of Drywood Termites (Isoptera, Kalotermitidae)

Molecular Phylogenetics and Evolution Vol. 17, No. 3, December, pp. 419 – 429, 2000 doi:10.1006/mpev.2000.0852, available online at http://www.idealibrary.com on

Phylogenetic Analysis and Trait Evolution in Australian Lineages of Drywood Termites (Isoptera, Kalotermitidae) Graham J. Thompson,* ,1,2 Leigh R. Miller,† Michael Lenz,† and Ross H. Crozier* ,1 *Department of Genetics and Evolution, La Trobe University, Bundoora, Victoria 3083, Australia; and †Division of Entomology, CSIRO, GPO Box 1700, Canberra, ACT 2601, Australia Received December 30, 1999; revised August 11, 2000; published online November 30, 2000

INTRODUCTION A phylogenetic analysis of Australian drywood termites (Isoptera, Kalotermitidae) based on partial sequence from the cytochrome oxidase II (COII) and cytochrome b genes is presented. In addition to providing new information on the evolutionary relationships among 25 species from seven genera, we evaluate the relative likelihoods of alternative topological hypotheses, including those derived from morphology-based classifications. We also test the applicability of a molecular clock for estimating the age of the Kalotermitidae and infer the evolution of species-specific variation for habitat type and soldier caste phragmosis by mapping this information onto the independently derived phylogeny. Maximum-likelihood analysis of both nucleotide and protein sequences from a multigene data set jointly support a single topology, which is shown to be the best estimate of the true phylogeny among the alternatives tested. Our results support the monophyly of all genera but question the discrimination between Procryptotermes and Cryptotermes. A basal dichotomy among generic groups suggests two principle lines of divergence within the family. Intergeneric relationships show mixed congruence to previous proposals, resulting in one morphology-based classification being rejected. A molecular clock hypothesis is not supported due to significant among-lineage rate heterogeneity in the COII gene. Patterns revealed through trait mapping suggest that the most recently diverged taxa tend to occupy the driest habitats and that these same taxa reflect a defensive transition away from large mandibulate soldiers toward small phragmotic soldiers. The association between habitat and defensibility supports the hypothesis that these two characters have been tightly linked throughout the social diversification of termites. © 2000 Academic Press Key Words: maximum-likelihood; mitochondrial DNA; genetic variation; insect molecular phylogeny; Mastotermes; Australian biogeographic region. 1 Present address: School of Tropical Biology, James Cook University, Townsville, Queensland 4811, Australia. 2 To whom correspondence should be addressed. Fax: 61 7 4725 1570. E-mail: [email protected].

The Kalotermitidae are the second largest of seven families in the exclusively eusocial order Isoptera. Owing to the low atmospheric moisture requirements of various taxa within this family, the Kalotermitidae as a whole are colloquially refered to as the “drywood termites” and perhaps are most noted for their economic significance as structural pests. Collectively, the over 20 extant genera show a cosmopolitan distribution (Pearce and Waite, 1994) with their greatest abundance expected to occur in coastal forests of tropical and subtropical regions. Despite their ubiquity and taxonomic diversity, however, the family has no accepted classification above the generic level. This is in contrast to other families within the order and indicates that the phylogenetic relationships among the genera themselves remain largely unknown or untested. This uncertainty may be due in part to the appearance of many transitional forms among living kalotermitids, making the clear differentiation of certain genera difficult (Ahmad, 1950). The Kalotermitidae are regarded as having a relatively simple form of social organization (Abe, 1987; Ahmad, 1950; Noirot, 1970; Roisin and Pasteels, 1990). Their colonies are essentially small, monogamous families which live entirely within drywood excavations, with no elaborate nest architecture or necessary contact with soil (Krishna, 1961). Their system of caste differentiation is flexible (Watson and Sewell, 1985), but they have no true workers and have few soldiers per colony (Noirot, 1985). Various genera do have the potential to develop neotenic replacement reproductives through the differentiation of nymphs or pseudergates (Myles and Nutting, 1988), and specific taxa among them have special karyotypic features (Luykx and Syren, 1979). The genetic composition of single colonies is, however, not known to deviate from a basic family model and they generally do not approach the same degree of social integration evident among other termite groups. Moreover, according to traditional morphologybased phylogenetic hypotheses, the Kalotermitidae form

419

1055-7903/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

420

THOMPSON ET AL.

a monophyletic group with the relictual Mastotermes and, together with it, form the sister group to all other termites (Krishna, 1970). Beyond their postulated differentiation from a widespread Mastotermes-like ancestor (Ahmad, 1950) at some time during the mid-Mesozoic (Emerson and Krishna, 1975), however, the historical origins of kalotermitid diversity remain obscure. The number of kalotermitid species indigenous to the Australian region is moderate in relation to other biogeographic zones and represents 8 of the worldwide total of 21 genera (Calaby and Gay, 1959). Yet, their biology and distribution within this region has been comparatively well documented (Abensperg-Traun and Steven, 1997; Calaby and Gay, 1959; Gay and Calaby, 1970; Hill, 1942; Luykx, 1990; Sewell, 1978; Watson and Abbey, 1993; Watson and Gay, 1991; Watson and Sewell, 1985), providing a stable platform from which to base an examination of their phylogenetic history. At the time of Hill’s (1942) monograph there were 39 species of Kalotermitidae (sensu Grasse´, 1949) described from the Australian biogeographic region in the single genus “Calotermes,” which consisted of fewer than six subgenera (Hill, 1942, pp. 7– 8). Following complete revisions of Cryptotermes (Gay and Watson, 1982) and Glyptotermes (Eldridge, 1996), as well as partial revisions of Kalotermes (Gay, 1977; Sewell and Gay, 1978), Ceratokalotermes (Krishna, 1961), Bifiditermes (Watson et al., 1984), Incisitermes (Gay, 1975a), Procryptotermes (Gay, 1975b), and Cryptotermes (Yule and Watson, 1976), there are currently 36 species in eight genera described for this region. Here we endeavour to reconstruct the phylogenetic relationships among Australian lineages of the Kalotermitidae based on mitochondrial cytochrome b (Cytb) and cytochrome oxidase II (COII) gene sequences. As Australian insect diversity is characteristically derived from both historically remote and more recent in situ diversification processes (Cranston and Nauman, 1991), patterns of cladogenesis revealed through these taxa are expected to reflect more global patterns of phylogenetic resolution. Thus, in addition to providing new information on the interrelationships and general patterns of diversification among Australian lineages, we seek to test existing morphology-based hypotheses of taxonomic monophyly and intergeneric relationships for the family as a whole. We also assess the applicability of a molecular clock for dating kalotermitid divergences and discuss the evolution of colony habitat type and soldier caste phragmosis—two attributes which are both unique and conspicuous aspects of their biology. MATERIALS AND METHODS

from alcohol-preserved private and museum collections (Table 1). In most cases, only single specimens of each species were included but, to assess the degree of intraspecific variability, two or more conspecifics were included wherever possible. Two non-Australian taxa were included in the analysis: Kalotermes brouni (New Zealand) and Neotermes sp. (Fiji). In addition, single specimens from Mastotermes darwiniensis (Mastotermitidae) and Porotermes adamsoni (Termopsidae) were employed as outgroups. These taxa are regarded as close to the kalotermitids on the basis of phenotypic (Krishna, 1970; Miller, 1997; Noirot, 1995; Thorne and Carpenter, 1992) and genotypic (Kambhampati et al., 1996) analyses. In total, 32 different kalotermitid specimens were included in the study, representing 25 species from seven genera. These taxa represent all but one of the Australian genera, Incisitermes, which is monotypic in this region. DNA Extraction, Amplification, and Purification DNA was obtained from fresh tissue using a silicabased extraction resin (Chelex 100; Bio-Rad) following manufacturer specifications. In contrast, a phenol/chloroform-based extraction was employed for preserved tissue (Saghai-Maroof et al., 1984). To avoid contamination from the intestinal symbionts present in all termites, only heads were used as source tissue. DNA was amplified using the polymerase chain reaction (PCR) (Saiki et al., 1988) and an array of oligonucleotide primers previously designed for use with insects (Liu and Beckenbach, 1992; Simon et al., 1994) but modified to optimize their effectiveness here (Table 2). Primers flanking target regions were modified in reference to Locusta migratoria, an orthopteran whose entire mitochondrial genome is known (Flook et al., 1995), whereas internal primers were based on partial termite sequences as they were obtained. All PCRs were of 50-␮l volumes containing 200 ␮M dNTPs, 1.5–3 mM MgCl 2, 1 ␮mol of each primer, and 0.4 units of Promega Taq DNA polymerase in a standard reaction buffer; 5 ␮l of the Chelex supernatant, or 2 ␮l of the phenol/chloroform resuspensions, was used as DNA template. Amplification profiles were 3 min initial denaturation at 94°C for 1 cycle, followed by 30 s at 93°C, 30 s at between 45 and 55°C, and 30 s at 72°C for 33 cycles. Finally, a single extension phase at 72°C for 5 min was added to the end of the cycle program. Prior to sequencing, the amplification product of each reaction was purified using Wizard PCR Preps DNA Purification System (Promega) and resuspended in 50 ␮l of TAE buffer (40 mM Tris–acetate, pH 8.5, 2 mM EDTA). Sequencing

Sample Collections The taxa included in this study were obtained either directly from the field, from live laboratory cultures, or

Purified DNA was sequenced either directly in 33plabeled termination reactions (Sanger et al., 1977) using the fmol Cycle Sequencing Kit (Promega) or via an

421

PHYLOGENY OF AUSTRALIAN DRYWOOD TERMITES

TABLE 1 List of Taxa Included in This Study Along with Information on Collection Locale, Source of Tissue, and GenBank Sequence Accession Nos. Collection information Species

(source, population)

Mastotermes darwiniensis b Porotermes adamsoni c Bifiditermes improbus c,1 Bifiditermes improbus c,2 Ceratokalotermes spoliator a Cryptotermes austrinus b Cryptotermes brevis b Cryptotermes cynocephalus b,1 Cryptotermes cynocephalus b,2 Cryptotermes domesticus b,1 Cryptotermes domesticus b,2 Cryptotermes domesticus b,3 Cryptotermes dudleyi b Cryptotermes primus b,1 Cryptotermes primus b,2 Cryptotermes queenslandis b Cryptotermes secundus a Cryptotermes simulatus c Cryptotermes tropicalis c Glyptotermes brevicornis c Glyptotermes iridipennis c Glyptotermes eucalypti c Glyptotermes tuberculatus c Kalotermes aemulus a Kalotermes hilli a Kalotermes pallidinotum c Kalotermes rufinotum c Kalotermes brouni c Neotermes insularis c Neotermes sp. c Procryptotermes australiensis a

Accession No.

Locality

Reference

CytB

COII

Darwin, NT Dandenongs, Vic Braidwood, NSW Kinglake, Vic Mullion Ck, NSW Alice Springs, NT Brisbane, Qld Mt Webb, Qld Mossman, Qld Mossman, Qld Darwin, NT Cooktown, Qld Thursday Is, Qld Bullock Pt, Qld Cape Cleveland, Qld Yannaman, Qld Adelaide R, NT Beecroft, NSW Danbulla, Qld Gadgarra For., Qld Black Rock, Vic Braidwood, NSW Blakehurst, NSW Western Australia Western Australia Beecroft, NSW Author’s Seat, Vic Muriwai Bch, NZ Kinglake, Vic Culo-I-Suva, Fiji Stuart Hwy, NT

ANIC21791 ANIC21782 ANIC21783 ANIC21784 ANIC18714 ANIC14835 ANIC1681 ANIC17171 ANIC16383 ANIC15581 ANIC18763 ANIC17191 ANIC16586 ANIC17186 ANIC16853 ANIC15935 ANIC19060 ANIC21785 ANIC15600 ANIC21349 ANIC21786 ANIC21787 FCNI5816 ANIC15973 ANIC15870 ANIC21788 ANIC21789 ANIC17489 ANIC21790 ANIC21792 ANIC18709

— — — AF189110 — AF189111 AF189112 AF189113 AF189114 AF189115 — — AF189116 AF189117 AF189118 AF189119 AF189120 AF189121 AF189122 — — AF189123 — — — — — — AF189124 — —

AF189108 AF189109 AF189080 AF189079 AF189089 AF189081 AF189082 AF189083 AF189084 AF189085 AF189086 AF189087 AF189088 AF189090 AF189091 AF189092 AF189093 AF189094 AF189095 AF189096 AF189097 AF189098 AF189099 AF189100 AF189101 AF189102 AF189103 AF189104 AF189105 AF189106 AF189107

Note. ANIC, Australia National Insect Collection; FCNI, Forestry Commission of New South Wales Insect Collection; NT, Northern Territory; Vic, Victoria; NSW, New South Wales; Qld, Queensland; NZ, New Zealand. a Alcohol-preserved museum specimen. b Live culture. c Field collection.

automated DNA sequencer (ABI 377) using dideoxy terminator chemistry following recommended procedures. Management of manual and automated sequence data involved MacVector 4.1.4 (IBI) and Sequencher 3.0 (Gene Codes) sequence editing software. All sequences were obtained in both directions. Preliminary analysis of DNA sequence data indicated that the COII gene itself does not provide accurate resolution across the entire range of divergences encountered here (Thompson and Crozier, 1998). For this reason, we adopted a combined data strategy obtaining additional nucleotide sequence from Cytb for all Cryptotermes spp., plus representatives from Bifiditermes, Neotermes, and Glyptotermes. All sequences reported in this study were registered in GenBank databases (Table 1), aligned using CLUSTAL W (Thompson et al., 1994), and verified by eye against existing insect alignments.

TABLE 2 Oligonucleotide Primers Used to Amplify and Sequence Termite Cytochrome Oxidase II and Cytochrome b Genes Gene Primer name COII C2-J-3096 C2-J-3360 C2-N-3365 TK-N-3807 Cytb CB-J-10493 CB-J-10787 CB-J-11219 CB-N-10801 CB-N-11251 CB-N-11561

Primer sequence

5⬘-AGAGCATCACCAATCATAGAACA-3⬘ 5⬘-GGACACCAATGATACTGAAG-3⬘ 5⬘-TTTATTGAAGTCTGAATATTC-3⬘ 5⬘-GTTTAAGAGACCATTACTTA-3⬘ 5⬘-CCAACAAACATTTCAATATGATGAAA-3⬘ 5⬘-ATAGCAACAGCATTTATAGG-3⬘ 5⬘-CACATTCAACCAGAATGATATTT-3⬘ 5⬘-TCCTCAAAATGATATTTGACCTCA-3⬘ 5⬘-ATAACTCCTCCTAATTTATTAGG-3⬘ 5⬘-ACTTCTTTTCTTATGTTTTCAAAAC-3⬘

422

THOMPSON ET AL.

FIG. 1. Hypothetical relationships among kalotermitid genera according to (A) Krishna (1961) and (B) Ahmad (1950).

Phylogenetic Analysis Maximum-likelihood (ML) analyses of nucleotide data were performed using the NUCML program of the MOLPHY, version 2.3 package (Adachi and Hasegawa, 1996b) and employed the HKY85 (Hasegawa et al., 1985) model of nucleotide substitution. Likelihood-ratio tests (Posada and Crandall, 1998) among 40 alternative hierarchically nested models indicated that HKY85 was indeed well suited to the substitutional processes characterizing this data. Similarly, ML analyses of protein data were performed using the PROTML program of the MOLPHY package and em-

ployed the mtREV24 (Adachi and Hasegawa, 1996a) model of amino acid substitution. To evaluate the extent to which these data support the intergeneric relationships proposed by Ahmad (1950) and Krishna (1961) (Fig. 1), we calculated the likelihoods of trees derived under searches constrained to uphold the relationships therein. Differences in likelihoods between these constrained trees and between those derived from other searches without any such constraints were calculated using the formulae of Kishino and Hasegawa (1989), implemented in NUCML and PROTML for nucleotide and protein sequences, respectively.

423

PHYLOGENY OF AUSTRALIAN DRYWOOD TERMITES

TABLE 3 Kishino–Hasegawa Tests for Alternative Topological Hypotheses Derived from Mitochondrial DNA Sequence Data under Constrained and Unconstrained Analyses Topology Sequence Protein ⌬lnL SE ⌬lnL/SE Nucleotide ⌬L SE ⌬lnL/SE

1

2

3

4

5

6

7

Sites

⫺5529.8 ml

⫺12.7 21.8 0.58

⫺39.4 21.9 1.79

⫺29.9 21.4 1.40

⫺13.4 16.7 0.80

⫺20.7 19.9 1.04

⫺30.7 21.9 1.40

561

⫺15531.7 ml

⫺180.4 36.7 4.91*

⫺46.8 30.1 1.55

⫺117.1 36.1 30.94*

⫺16.9 22.5 0.75

⫺33.3 24.2 1.38

⫺92.7 34.8 2.66*

1683

Note. The natural logarithm of the likelihood value (lnL) is given for the most likely (ml) topology, whereas ⌬lnL (⫾SE) shows the difference in the lnL from that of the ml tree for inferior topologies. Topologies 1– 4 were derived without constraints under different tree-building criteria and models, a whereas Topologies 5–7 were derived under maximum-likelihood (ML) criteria (HKY85) with topological constraints of morphology-based hypotheses b imposed. Values with an asterisk indicate that the corresponding hypothesis can be rejected (5% significance) by the standard criterion ⌬lnL/SE ⬎ 1.96 (Kishino and Hasegawa, 1989). a 1, ML (HKY85); 2, ML (mtREV24); 3, ML (Quartet Puzzling; Strimmer and von Haesler, 1996); 4, ML (Quarted Puzzling; protein). b 5, all taxonomic genera monophyletic; 6, Ahmad (1950); 7, Krishna (1961).

A test for the molecular clock-like behavior of the COII nucleotide data set was also performed under likelihood criteria, by constraining the mean rate of substitution to be constant among lineages and comparing this tree to the ML estimate obtained when this constraint was relaxed. If the topologies are equivalent and the likelihoods are not significantly different (likelihood-ratio test, ␣ ⫽ 0.05, df ⫽ s ⫺ 2, where s ⫽ number of taxa), the molecular clock hypothesis cannot be rejected (Huelsenbeck and Rannala, 1997). Finally, through optimization of ancestral character states over the most likely estimate of the true phylogeny, we discuss the evolutionary correlates that occur between habitat type and soldier head capsule morphology (details below) in relation to kalotermitid social diversification. RESULTS Levels and Patterns of Sequence Variation A total of 624 bp of COII was obtained for 32 specimens representing 27 termite species. In addition, a total of 1059 bp of Cytb was obtained for 17 specimens including 13 species. Because all overlapping regions were contiguous, and no frameshift or nonsense codons were apparent, we assume that these sequences are of mitochondrial origin only. Nucleotide variation for both genes show an adenine–thymine bias in their nucleotide composition (A ⫹ T ⫽ 0.64215 for COII; A ⫹ T ⫽ 0.6025 for Cytb) that is consistent with other insect mitochondrial genomes (Crozier and Crozier, 1993; Flook et al., 1995; Mitchell et al., 1993). Among-site rate variation, also common in protein-encoding mitochondrial genomes (Kumar, 1996), was evident for both

genes as the vast majority of variation occurred at third codon positions (55% for COII, 58% for Cytb), followed by positions 1 and 2, respectively. Thus, the pattern of substitutional process observed in the COII gene is similar to that found for other termite (Miura et al., 1998) and orthopteran (Maekawa et al., 1999) species. To test for the effects of intraspecific variation on phylogenetic estimation, we initially included multiple replicates from Bifiditermes improbus (⫻2), Cryptotermes cynocephalus (⫻2), C. domesticus (⫻3), and C. primus (⫻2). In each case, conspecific sequences were different, but no change occurred in the branching pattern under basic searches when either one or several individuals were included from these replicates. As a condition prior to the concatenation of the COII and Cytb data sets, Kishino–Hasegawa tests (Kishino and Hasegawa, 1989) were used to test for phylogenetic congruence between them. In relation to their two most likely topologies, no significant incongruence was found (P ⬍ 0.8232 with COII as “test” data set; P ⫽ 0.5998 with Cytb as “test” data set), thereby justifying their combination into a single data set (1683 bp). Phylogenetic Analysis The results of the constrained, as well as the unconstrained, ML analyses are summarized in Table 3. A single topology (Topology 1; Fig. 2) was invariably the most likely (ml) among the alternatives, irrespective of whether models of nucleotide or amino acid substitution were chosen to evaluate them. Among these alternatives (not shown), all were less likely and some significantly so. Namely, the HKY85 nucleotide model found that Topology 2 (ML search using the mtREV24

424

THOMPSON ET AL.

as the ml, and again Topologies 2, 4, and 7 as significantly worse by this same criterion. Figure 2 shows the ml topology obtained from the mitochondrial protein-encoding nucleotide sequences of the combined data set. The ml tree indicates that taxonomic genera tend to represent well-defined monophyletic clades that fall into one of two phyletic lines: Ceratokalotermes ⫹ Kalotermes being basally divergent from all of the remaining genera. This indicates that the Ceratokalotermes ⫹ Kalotermes clade diverged early from other living kalotermitid lineages and is a direct descendant from pre-kalotermitid ancestors. Within the Glyptotermes ⫹ Neotermes ⫹ Bifiditermes ⫹ Procryptotermes ⫹ Cryptotermes clade, Glyptotermes is the most basal, whereas the Procryptotermes ⫹ Cryptotermes assemblage represents the most apical lineage. Neotermes and Bifiditermes occupy positions intermediate between the two polar generic groups. One apparent exception to the rule of monophyletic genera is therefore Cryptotermes. The placement of Procryptotermes australiensis well within Cryptotermes questions the clear differentiation between these two genera. The placement of this single species, however, does not reduce the overall likelihood of the tree sufficiently to merit a faslification of their taxonomic validity here (Table 3). Molecular Clock FIG. 2. Most likely tree (Topology 1 in Table 3, HKY85 model, ␣/␤ ⫽ 5.43, lnL ⫽ ⫺ 15531.71) derived from a combined data set consisting of partial cytochrome oxidase II and cytochrome b nucleotide sequences (1683 bp) for 25 species of Australian kalotermitids plus two outgroup taxa. Branch lengths are proportional to the expected number of substitutions under the HKY85 model. The local bootstrap probabilities (in %), calculated by the RELL method (Hasegawa and Kishino, 1994), are given above each internal branch.

model), Topology 4 [Quartet Puzzling (Strimmer and von Haeseler, 1996) of protein sequence], and Topology 7 (Krishna (1961)-constrained ML search using the HKY85 model) were all significantly worse than the ml tree. Of these, Topologies 2 and 4 differed from the ml tree in that they failed to make Cryptotermes monophyletic, whereas Topology 7 differed in that it grouped Glyptotermes and Neotermes in a clade together with the basally divergent Ceratokalotermes and Kalotermes (see Fig. 1A). The remaining topologies, including Topology 5 (genera constrained to be monophyletic) and Topology 6 (Ahmad (1950)-constrained ML search using the HKY85 model; see Fig. 1B), while less likely, are not significantly different from the ml tree and therefore cannot be rejected by the criteria of Kishino and Hasegawa (1989). These conclusions were reinforced when Cytb data was removed (1059 bp, 13 taxa) and the relative likelihoods among Topologies 1–7 recalculated. The HKY85 model again selects Topology 1

Due to among-lineage rate heterogeneity in the COII gene, the molecular clock hypothesis is not supported. Specifically, the likelihood of the clock-like COII tree (⫺lnL ⫽ 2538.7005) and that of the unconstrained tree (⫺lnL ⫽ 2504.98413) were significantly different from each other (␹ 2 ⫽ 67.43, DF ⫽ 24, P ⬍ 0.05). This result indicates that various lineages evolved at different rates and, therefore, the use of molecular clock calibrations to estimate haplotype divergence times appears statistically invalid for this data set. DISCUSSION Relationships within the Kalotermitidae Among the taxa examined here, the most basal clade consists of the monotypic genus Ceratokalotermes and the five species of Kalotermes. Ceratokalotermes, which is unknown outside the Australian region (Emerson, 1955), was originally included in Kalotermes (Hill, 1942) but later erected to full generic rank (Krishna, 1961). By some classifications, Kalotermes is considered to be the most “primitive” of the kalotermitids, as evidenced by its wing venation and the elongate shape of the soldier head (Ahmad, 1950). Our results support the idea that it is indeed among the most anciently diverged. Within Kalotermes, a close association is shown between K. aemulus and K. Hilli, which have been noted for their similarity in morphology and dis-

PHYLOGENY OF AUSTRALIAN DRYWOOD TERMITES

tribution (Sewell and Gay, 1978). Kalotermes brouni, which is endemic to New Zealand and geographically isolated from the Australian mainland, does not appear to be especially distinct from the local Australian taxa. Glyptotermes forms a distinct clade, apparently derived from an ancestor common to the Ceratokalotermes ⫹ Kalotermes clade. This position of Glyptotermes differs from previous proposals (Fig. 1), which suggest a more derived position with respect to Neotermes. The endemic N. insularis and Fijian N. sp. group together, but emerge from the tree later than Glyptotermes. Bifiditermes was erected for the inclusion of taxa formerly in Kalotermes (Krishna, 1961), and the sole Australian species, B. improbus, was transferred to this genus from Kalotermes accordingly (Watson et al., 1984)—a move supported by the clear distinction that we see between these genera. The sole Australian representative of the genus Procryptotermes (Gay, 1975b), P. australiensis is, as expected, closely aligned with Cryptotermes. Of the 11 species of Procryptotermes, some show a mild degree of soldier head phragmotism and for this reason are expected to occur immediately basal to the Cryptotermes species (Fig. 1). However, P. australiensis appears here embedded within the Cryptotermes clade and, though our hypothesis of generic monophyly was not rejected (Table 3), this result questions the autonomy of the two genera. Further studies that include a greater representation of Procryptotermes species may help to determine whether these two genera should be synonymized. Within Cryptotermes, C. gearyi and the introduced C. brevis (Heather, 1971) are the most basal. The latter species is a particularly destructive pest and has gained a cosmopolitan distribution. Cryptotermes queenslandis was originally placed in Procryptotermes (Snyder, 1949), but later transferred to Cryptotermes (Krishna, 1961), a move supported here. Other relationships within this clade are predicted on the basis of morphological similarity. Thus, in their review of Australian Cryptotermes (Gay and Watson, 1982), C. queenslandis and C. simulatus, as well as C. primus and C. tropicalis, are noted for their similarity, and each of these pairings appear here. Testing Alternative Hypotheses Since selection will operate on protein-encoding genes to preserve the codon or amino acid rather than a nucleotide, protein sequences are necessarily more conserved than the nucleotide sequences that underlie them. The reduced variation associated with this conservation and threefold reduction in the number of sites from which comparisons are based may account for the observed reduction in power for discriminating among the alternative hypotheses from protein sequence. None of the alternative hypotheses tested could be rejected by the mtREV24 model, whereas

425

three alternatives (Topologies 2, 4, and 7) were rejected by the HKY85 model (Table 3). Interestingly, Topology 2, which was derived from the mtREV24 model, was not found to be the ml tree when evaluated against the same data set from which it was derived (Table 3). Instead, the protein model yielded a higher likelihood for that topology which was derived from nucleotide data (Topology 1), suggesting that PROTML resulted only in a local maximum. This convergence upon a single topology by nucleotide and protein substitutional models reinforces support for the phylogeny presented. Of the remaining unconstrained searches to be rejected, Topology 4 is also derived from protein data under Quartet Puzzling (QP) (Strimmer and von Haeseler, 1996). QP algorithms, which search for the ML tree, do not always find it (Cao et al., 1998) unless additional rearrangements of the QP topology are examined. We suggest then that the relatively short length of sequence and large number of taxa examined in this study act to confound these protein-based inferences. Of the two constrained searches, Topology 7, where fixed subtrees were erected to uphold the intergeneric relationships proposed by Krishna (1961) (Fig. 1), is rejected. While certain sister group relationships predicted therein are corroborated (e.g., {{Procryptotermes, Cryptotermes}, Bifiditermes}), others were not (Fig. 2). Chief among them were those concerning the interrelationships among Glyptotermes, Neotermes, Ceratokalotermes, and Kalotermes. Our results suggest that Ceratokalotermes ⫹ Kalotermes form a distinct clade separate from the clade that contains Glyptotermes and Neotermes. Two distinct and basally divergent phyletic lines are evident here but they do not correspond directly to those described in Krishna (1961). Specifically, while the “Incisitermes–Cryptotermes complex” (represented here by Bifiditermes, Procryptotermes, and Cryptotermes) is assembled into a monophyletic clade, the “Proelectrotermes–Calcaritermes complex” (represented by Ceratokalotermes, Kalotermes, Glyptotermes, and Neotermes) is not. Instead, the later forms a paraphyletic assemblege of taxa. We therefore suggest that if subfamilial taxonomy is implemented for this group as in other termite families of comparable diversity, that it correspond to the monophyletic lineages represented here by (Ceratokalotermes ⫹ Kalotermes) and (Glyptotermes ⫹ Neotermes ⫹ Bifiditermes ⫹ Procryptotermes ⫹ Cryptotermes). An additional subfamily may include fossil genera Proelectrotermes, Prokalotermes, and Electrotermes, as suggested by Emerson (1942), but this hypothesis cannot be tested here. Further studies which include a greater number of genera will test the generality of this conclusion and will lead to the refinement of relationships within this framework. Existing classifications on these relationships, such as

426

THOMPSON ET AL.

those tested here, are invaluable tools without which no tests would even be possible. Biogeographic Implications Distributional patterns of termite diversity can be predicted on the basis of geological history and contemporary ecological factors (Eggleton et al., 1994). Australia is a distinct biotic region but little is known of the biotic diversification processes that accompanied its fragmentation from Gondwana (Cranston and Nauman, 1991). Whereas some Australian termite genera restricted to the cool southeast corner are thought to have remained from this ancient connection (e.g., Stolotermes, Porotermes: Termopsidae), kalotermitid genera probably dispersed here secondarily from tropical Papuan and/or Indomalayan regions after having originated elsewhere as early as the Mid-Mesozoic (Emerson and Krishna, 1975; Kukalova-Peck, 1991). The current distribution of kalotermitid species in Australia, primarily north tropical and coastal (Watson and Abbey, 1993), as well as their taxonomic affinity to these northerns regions (Gay and Calaby, 1970), is consistent with the hypothesis of a relatively recent Indomalayan source and suggests that temperature and moisture are the major ecological factors that currently act to maintain this distribution. Trait Mapping From nesting and feeding habit information, evolutionary trends in the termite life types can readily be identified (Abe, 1987). By Abe’s criterion, kalotermitids are generally of the one-piece drywood type, indicating that colonies live entirely within a single piece of wood that provides both an expansible food-rich resource and a structural defense. These type designations are, however, not absolute, as nest site attributes, as well as defense strategies, are known to vary within these general themes. For example, among these taxa, individual species’ habitat preferences range from extremely dry, sound wood (e.g., standing dead twigs or branches on living trees, defined here as true drywoods) to relatively moist, soft wood (e.g., dead tree trunks, logs, rotting stumps) (Lenz, 1994). Similarly, soldiers range from those with elongated heads and snout, strongly toothed mandibles physically adapted for combat to those with foreshortened heads and reduced mandibles (phragmotic) physically adapted for mechanically blocking predators from entering nest galleries (Prestwich, 1984) (Fig. 3). Based on the relative characteristics of the habitats in which individual species are typically found (with reference to findings in Creffield, 1991; Eldridge, 1991, 1996; Gay, 1975a,b, 1977; Gay and Watson, 1982; Hadlington, 1987; Hill, 1942; Krishna, 1961; Krishna and Weesner, 1970; Lenz, 1994; Sewell and Gay, 1978; Watson et al., 1984; Yule and Watson, 1976), we identify those taxa which could be considered true drywoods and, from soldier

head profiles, those which are phragmotic. From this information we infer the phylogenetic association between changes in habitat type and changes in defensability through a reconstruction of ancestral character states over the phylogenetic tree. The simplified coding scheme employed here is not necessarily exclusive, as certain species could fit more than one category. However, it will serve to categorize individual species in a broad sense and will provide a first indication of the presence and extent of correlated changes between habitat type and other life history features. Figure 3 shows the relationship between habitat type and soldier type. The general pattern to emerge from this qualitative comparison is that the true drywood species tend to be smaller and show the greatest reliance on phragmotism compared to related species which are larger and rely on mandibulate soldiers for colony defense. In general, food–shelter habitats and an ability to defend them are thought to be basic conditions that underpin termite social structure (Crespi, 1994) and as such are expected to have covaried throughout their evolutionary diversification. Moreover, the association between habitat and defensive characteristics observed here supports the hypothesis that differences in ecology are tightly linked to defensive strategy in termites (Abe, 1991). Within the Procryptotermes ⫹ Cryptotermes clade, C. gearyi, C. brevis, C. domesticus, and C. cynocephalus are basal and have particularly sculptured and foreshortened heads, which indicates that this form of caste specialization occurred rapidly but was largely reversed in the branch to other Cryptotermes. CONCLUSIONS The present study provided strong statistical support by both bootstrap (Fig. 2) and Kishino–Hasegawa tests (Table 3) for the interrelationships among Australian lineages of the Kalotermitidae. The phylogenetic distribution of habitat and defensive characters among these lineages also suggests that ecological factors may account for much of the variation present in termite social systems. This work complements other studies which too have begun to evaluate the phylogenetic relationships among major taxa within termite families (Miura et al., 1998), as well as among the families themselves (Kambhampati and Eggleton, 2000). Understanding these relationships are of interest because they will help to ordinate the appearance of additional traits that have come to characterize their varied life histories and, when placed in the context of higher-order phylogenies (DeSalle et al., 1992; Maekawa et al., 1999; Thorne and Carpenter, 1992), will serve as a basis for testing any number of hypotheses concerning termite social evolution.

PHYLOGENY OF AUSTRALIAN DRYWOOD TERMITES

427

FIG. 3. Cladogram showing the phylogenetic distribution of true drywood vs drywood species and phragmotic vs mandibulate soldiers (defined in Discussion) among Australian lineages of Kalotermitidae. There is a correlation between the first appearance of the true drywood form (indicated at ancestor 1) and the first appearance of the phragmotic soldier (indicated at ancestor 2) among these taxa. This pattern indicates that the most recently diverged taxa tend to occupy the driest habitats and that these same taxa reflect a defensive shift away from large mandibulate soldiers toward small phragmotic soldiers. The association between habitat and defensability supports the hypothesis that these two characters have been linked throughout the social diversification of termites (Abe, 1991). The above tree is rooted using outgroups as indicated in Fig. 2.

428

THOMPSON ET AL.

ACKNOWLEDGMENTS We thank M. Abensperg-Traun, J. Creffield, R. H. Eldridge, and the Australian National Insect Collection (ANIC) for specimens and/or advice on sampling; M. Chapuisat, M. Senetra, L. Jermiin, and R. DeSalle for their constructive comments; and S. P. Kim for providing all termite illustrations. This work is supported by a NSERC (Canada) Postgraduate Scholarship, a La Trobe University Postgraduate Research Scholarship, and a LTU Overseas Postgraduate Research Scholarship to G.J.T., as well as Australian Research Council grants to R.H.C.

REFERENCES Abe, T. (1987). Evolution of life types in termites. In “Evolution and Coadaptation in Biotic Communities” (S. Kawano, J. H. Connell, and T. Hidaka, Eds.), pp. 125–148. Univ. of Tokyo Press, Tokyo. Abe, T. (1991). Ecological factors associated with the evolution of worker and soldier castes in termites. Ann. Entomol. 9: 101–107. Abensperg-Traun, M., and Steven, D. (1997). Latitudinal gradients in the species richness of Australian termites (Isoptera). Aust. J. Ecol. 22: 471– 476. Adachi, J., and Hasegawa, M. (1996a). Model of amino acid substitution in proteins encoded by mitochondrial DNA. J. Mol. Evol. 42: 459 – 468. Adachi, J., and Hasegawa, M. (1996b). “MOLPHY: Programs for Molecular Phylogenetics Based on Maximum Likelihood,” Version 2.3. Institute of Statistical Mathematics, Tokyo. Ahmad, M. (1950). The phylogeny of termite genera based on imago worker mandibles. Bull. Am. Mus. Nat. Hist. 95: 39 – 86. Calaby, J. H., and Gay, F. J. (1959). Aspects of the distribution and ecology of Australian termites. In “Biogeography and Ecology in Australia” (A. Keast, R. L. Crocker, and C. S. Christian, Eds.), pp. 211–223. Junk, The Hague. Cao, J., Adachi, J., and Hasegawa, M. (1998). Comment on the Quartet Puzzling method for finding maximum-likelihood tree topologies. Mol. Biol. Evol. 15: 87– 89. Cranston, P. S., and Nauman, I. D. (1991). Biogeography. In “The Insects of Australia” (CSIRO, Eds.), pp. 180 –197. Melbourne Univ. Press, Melbourne. Creffield, J. W. (1991). “Wood Destroying Insects: Wood Borers and Termites,” CSIRO, Melbourne. Crespi, B. C. (1994). Three conditions for the evolution of eusociality: Are they sufficient? Ins. Soc. 41: 395– 400. Crozier, R. H., and Crozier, Y. C. (1993). The mitochondrial genome of the honeybee Apis mellifera: Complete sequence and genome organization. Genetics 133: 97–117. DeSalle, R., Gatesy, J., Wheeler, W., and Grimaldi, D. (1992). DNA sequences from a fossil termite in Oligo-Miocene amber and their phylogenetic implications. Science 257: 1933–1936. Eggleton, P., Williams, P. H., and Gaston, K. J. (1994). Explaining global termite diversity—Productivity or history? Biodiv. Cons. 3: 318 –330. Eldridge, R. H. (1991). “Studies on the Australian Glyptotermes (Isoptera: Kalotermitidae),” MSc thesis, University of Sydney. Eldridge, R. H. (1996). Revision of Australian Glyptotermes Froggatt (Isoptera, Kalotermitidae). Aust. J. Entomol. 35: 165–176. Emerson, A. E. (1942). The relations of a relict South African termite (Isoptera: Hodotermitidae: Stolotermes). Am. Mus. Nov. 1187: 1–12. Emerson, A. E. (1955). Geographical origins and dispersions of termite genera. Fieldiana Zool. 37: 465–521.

Emerson, A. E., and Krishna, K. (1975). The termite family Serritermitidae (Isoptera). Am. Mus. Nov. 2570: 1–31. Flook, P. K., Rowell, C. H. F., and Gellissen, G. (1995). The sequence, organization, and evolution of the Locusta migratoria mitochondrial genome. J. Mol. Evol. 41: 928 –941. Gay, F. J. (1975a). An Australian species of Incisitermes Krishna (Isoptera: Kalotermitidae). J. Aust. Entomol. Soc. 14: 396 –398. Gay, F. J. (1975b). An Australian species of Procryptotermes Holmgren (Isoptera: Kalotermitidae). J. Aust. Entomol. Soc. 15: 45– 48. Gay, F. J. (1977). A new species of Kalotermes Hagen (Isoptera: Kalotermitidae) from tropical Queensland. J. Aust. Entomol. Soc. 16: 221–224. Gay, F. J., and Calaby, J. H. (1970). Termites of the Australian region. In “Biology of Termites” (K. Krishna and F. M. Weesner, Eds.), pp. 393– 448. Academic Press, New York. Gay, F. J., and Watson, J. A. L. (1982). The genus Cryptotermes in Australia (Isoptera: Kalotermitidae). Aust. J. Zool. Suppl. 88: 1– 64. Grasse´, P. P. (1949). Ordres des Isopte`res ou termites. In “Traite de Zoologie, Anatomie, Systematique, Biologie” (P. P. Grasse´, Ed.), pp. 408 –544. Masson, Paris. Hadlington, P. W. (1987). “Australian Termites and Other Common Timber Pests,” NSW Univ. Press, Kensington. Hasegawa, M., and Kishino, H. (1994). Accuracies of the simple methods for estimating the bootstrap probability of a MaximumLikelihood tree. Mol. Biol. Evol. 11: 142–145. Hasegawa, M., Kishino, H., and Yano, K. (1985). Dating of the human–ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22: 160 –174. Heather, N. W. (1971). The exotic drywood termite Cryptotermes brevis (Walker) (Isoptera: Kalotermitidae) and endemic Australian drywood termites in Queensland. J. Aust. Entomol. Soc. 10: 134 – 141. Hill, G. F. (1942). “Termites (Isoptera) from the Australian Region,” CSIRO Australia, Melbourne. Huelsenbeck, J. P., and Rannala, B. (1997). Phylogenetic methods come of age: Testing hypotheses in an evolutionary context. Science 276: 227–232. Kambhampati, S., and Eggleton, P. (2000). Phylogenetics and taxonomy of termites (Isoptera). In “Termites: Evolution, Sociality, Symbioses, Ecology” (D. E. Bignell, T. Abe, and M. Higashi, Eds.), pp. 1–32. Kluwer Academic, Berlin. Kambhampati, S., Kjer, K. M., and Thorne, B. L. (1996). Phylogenetic relationships among termite families based on DNA sequence of mitochondrial 16S ribosomal gene. Ins. Mol. Biol. 5: 229 –238. Kishino, H., and Hasegawa, M. (1989). Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29: 170 –179. Krishna, K. (1961). A generic revision and phylogenetic study of the family Kalotermitidae (Isoptera). Bull. Am. Mus. Nat. Hist. 122: 303– 408. Krishna, K. (1970). Taxonomy, phylogeny, and distribution of termites. In “Biology of Termites” (K. Krishna and F. M. Weesner, Eds.), pp. 127–152. Academic Press, New York. Krishna, K., and Weesner, F. M., Eds. (1970). “Biology of Termites,” Vol. 2, Academic Press, New York. Kukalova-Peck, J. (1991). Fossil history and the evolution of hexapod structures. In “The Insects of Australia” (CSIRO, Eds.), pp. 141– 179. Melbourne Univ. Press, Melbourne. Kumar, S. (1996). Patterns of nucleotide substitution in mitochondrial protein coding genes of vertebrates. Genetics 143: 537–548.

PHYLOGENY OF AUSTRALIAN DRYWOOD TERMITES Lenz, M. (1994). Food resources, colony growth and caste development in wood-feeding termites. In “Nourishment and Evolution in Insect Societies” (J. H. Hunt and C. A. Nalepa, Eds.), pp. 159 –209. Westview Press, Oxford and IBH Publ., Boulder/New Delhi. Liu, H., and Beckenbach, A. T. (1992). Evolution of the mitochondrial cytochrome oxidase II gene among 10 orders of insects. Mol. Phylogenet. Evol. 1: 41–52. Luykx, P. (1990). A cytogenetic survey of 25 species of lower termites from Australia. Genome 33: 80 – 88. Luykx, P., and Syren, R. M. (1979). The cytogenetics of Incisitermes schwarzi and other Florida termites. Sociobiology 4: 191–209. Maekawa, K., Kitidae, O., and Matsumoto, T. (1999). Molecular phylogeny of orthopteroid insects based on the mitochondrial cytochrome oxidase II gene. Zool. Sci. 16: 175–184. Miller, L. R. (1997). “Systematics of the Australian Nasutitermitinae with Reference to Evolution within the Termitidae (Isoptera),” PhD thesis, Australian National University. Mitchell, S. E., Cockburn, A. F., and Seawright, J. A. (1993). The mitochondrial genome of Anopheles quadrimaculatus species A: Complete nucleotide sequence and gene organization. Genome 36: 1058 –1073. Miura, T., Maekawa, K., Kitidae, O., Abe, T., and Matsumoto, T. (1998). Phylogenetic relationships among subfamilies in higher termites (Isoptera: Termitidae) based on mitochondrial COII gene sequences. Ann. Entomol. Soc. Am. 91: 515–523. Myles, T. G., and Nutting, W. L. (1988). Termite eusocial evolution: A re-examination of Bartz’s hypothesis and assumptions. Q. Rev. Biol. 63: 1–23. Noirot, C. (1970). The nests of termites. In “Biology of Termites” (K. Krishna and F. M. Weesner, Eds.), pp. 73–125. Academic Press, New York. Noirot, C. (1985). Pathways of caste development in lower termites. In “Caste Differentiation in Social Insects” (J. A. L. Watson, B. M. Okot-Kotber, and C. Noirot, Eds.), pp. 41–57. Pergamon, Oxford. Noirot, C. (1995). The gut of termites (Isoptera): Comparative anatomy, systematics, phylogeny. 1. Lower termites. Ann. Soc. Entomol. France 31: 197–226. Pearce, M. J., and Waite, B. S. (1994). A list of termite genera (Isoptera) with comments on taxonomic changes and regional distribution. Sociobiology 23: 247–263. Posada, K., and Crandall, K. A. (1998). Modeltest: Testing the model of DNA substitution. Bioinformatics 14: 817– 818. Prestwich, G. D. (1984). Defense mechanisms of termites. Annu. Rev. Entomol. 29: 201–232. Roisin, Y., and Pasteels, J. M. (1990). Evolutionary trends in neoteny and secondary reproduction in termites. In “Social Insects and the Environment” (G. K. Veeresh, B. Malik, and C. A. Viraktamath, Eds.), pp. 33–34. Westview Press, Oxford and IBH Publ., New Delhi. Saghai-Maroof, M. A., Soliman, K. M., Jorgensen, R. A., and Allard,

429

R. W. (1984). Ribosomal DNA spacer-length polymorphisms in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proc. Natl. Acad. Sci. USA 81: 8014 – 8018. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B., and Erlich, H. A. (1988). Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239: 487– 491. Sanger, F., Nicklen, F., and Coulsen, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74: 5463–5467. Sewell, J. J. (1978). “Developmental Pathways and Colony Organization in the Genus Kalotermes Hagen (Isoptera: Kalotermitidae),” PhD thesis, Australian National University. Sewell, J. J., and Gay, F. J. (1978). The genus Kalotermes Hagen in Western Australia (Isoptera: Kalotermitidae). J. Aust. Entomol. Soc. 17: 41–51. Simon, C., Frati, F., Beckenbach, A., Crespi, B., Liu, H., and Flook, P. (1994). Evolution, weighting and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved PCR primers. Ann. Entomol. Soc. Am. 87: 651–701. Snyder, T. E. (1949). “Catalogue of the Termites (Isoptera) of the World,” Smithsonian Miscellaneous Collections, Washington, DC. Strimmer, K., and von Haeseler, A. (1996). Quartet puzzling: A quartet likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13: 964 –969. Thompson, G. J., and Crozier, R. H. (1998). Molecular phylogeny of Australian kalotermitids and the evolution of drywood termites. In “Social Insects at the Turn of the Millenium” Proceedings of the XIII International Congress of IUSSI (M. P. Schwarz and K. Hogendoorn, Eds.), p. 471. XIII Congress of IUSSI, Adelaide. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994). CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids. Res. 22: 4673– 4680. Thorne, B. L., and Carpenter, J. M. (1992). Phylogeny of the Dictyoptera. Syst. Entomol. 17: 253–268. Watson, J. A. L., and Abbey, H. M. (1993). “Atlas of Australian Termites,” CSIRO Australia, Canberra. Watson, J. A. L., and Gay, F. J. (1991). Isoptera. In “The Insects of Australia,” pp. 330 –347. Melbourne Univ. Press, Melbourne. Watson, J. A. L., and Sewell, J. J. (1985). Caste development in Mastotermes and Kalotermes: Which is primitive? In “Caste Differentiation in Social Insects,” pp. 27– 40. Pergamon, Oxford. Watson, J. A. L., Gay, F. J., and Barrett, R. A. (1984). The identity of Kalotermes improbus Hagen (Isoptera: Kalotermitidae). J. Aust. Entomol. Soc. 23: 193–197. Yule, R. A., and Watson, J. A. L. (1976). Two further domestic species of Cryptotermes from the Australian mainland (Isoptera: Kalotermitidae). J. Aust. Entomol. Soc. 15: 349 –352.