Introduction
In the Mediterranean area, drywood termites in the genus Kalotermes (Isoptera, Kalotermitidae) are represented by three species: Kalotermes flavicollis Fabr., with a wide range of distribution in the circum-Mediterranean lands, Kalotermes italicus Ghesini & Marini, with a small range in central Italy, and Kalotermes monticola (Sjöstedt), found on a locality in north-eastern Algeria. The Moroccan Kalotermes maroccoensis (Sjöstedt) was recently synonymized with K. flavicollis (Krishna et al., Reference Krishna, Grimaldi, Krishna and Engel2013). Kalotermes sinaicus Kemner, distributed in north-eastern Egypt and Israel, was recently assigned to the new genus Longicaputermes, because of morphological and genetic peculiarities that differentiate it from the genus Kalotermes and liken it to the genera Comatermes, Ceratokalotermes, and Paraneotermes (Ghesini et al., Reference Ghesini, Simon and Marini2014).
While K. flavicollis populations from central Mediterranean regions (from France to Crete) have been analyzed with some detail (Luchetti et al., Reference Luchetti, Bergamaschi, Marini and Mantovani2004, 2013; Velonà et al., Reference Velonà, Luchetti, Ghesini, Marini and Mantovani2011), data on eastern Mediterranean Kalotermes populations are scarce. In particular, no data exist on Cypriot and Lebanese K. flavicollis, and for Israel only records of K. flavicollis presence in some localities are available (Bodenheimer, Reference Bodenheimer1930; Kugler, Reference Kugler, Yom-Tov and Tchernov1988).
In this study, we conducted both genetic (mitochondrial DNA: COII, 16S and control region) and morphological analyses of Kalotermes sp. from Cyprus, Lebanon, and Israel, in order to characterize these populations and to clarify their phylogenetic relationships. Because these populations form a clade sister to the clade composed of K. flavicollis and K. italicus, and because diagnostic characters exist that differentiate this taxon from west-European Kalotermes species, we describe it as Kalotermes phoeniciae sp. nov.
Materials and methods
Samples of Kalotermes sp., morphologically compatible with K. flavicollis, were collected during collecting trips conducted in Cyprus, Lebanon, and Israel in the years 2009–2012. Sampling localities are shown in fig. 1. When two or more colonies were analyzed from the same locality they are identified by a capital letter (A, B, or C) added to the name of the locality. Samples from Cyprus and Lebanon contained all the castes (alates, soldiers, and pseudergates), while samples from Israel contained only pseudergates and soldiers. Samples for genetic analyses were preserved in 100% ethanol, while samples for morphological observations were preserved in 80% ethanol.
Fig. 1. Sampling localities of Kalotermes sp. in Cyprus, Lebanon, and Israel. One colony was sampled in each locality, except when differently stated (3 col.).
Total DNA was extracted from termite heads following the CTAB protocol (Doyle & Doyle, Reference Doyle and Doyle1987). Two individuals for each sample were analyzed. A 662–668-bp fragment of the cytochrome oxidase subunit II gene (COII), a 515–516-bp fragment of the large mitochondrial ribosomal subunit gene (16S), and a 297–300-bp fragment including the 3′ portion of the control region, tRNA-Ile, tRNA-Gln, and the 5′ portion of tRNA-Met (CR), were amplified. The primers used are TL2-J-3034 (5′-AAT ATG GCA GAT TAG TGC A-3′) and TK-N-3785 (5′-GTT TAA GAG ACC AGT ACT TG-3′) for COII, LR-J-12887 (5′-CCG GTC TGA ACT CAG ATC ACG T-3′ and LR-N-13398 (5′-CGC CTG TTT AAC AAA AAC AT-3′) for 16S, and AT-KR (5′-GTG GCT ATA CCC ACT ATA AA-3′) and TM-N-193 (5′-TGG GGT ATG AAC CAG TAG C-3′) for CR. Polymerase chain reaction was performed in a 50 μl mixture using GoTaq® Flexi DNA Polymerase kit (Promega, USA), following the enclosed protocol. Reaction conditions were set as follows: initial denaturation at 95°C for 5 min; 30 cycles composed by denaturation at 94°C for 30 s, annealing at 48°C for 30 s, extension at 72°C for 30 s; final extension at 72°C for 7 min. Purification and sequencing were performed by Macrogen Inc. (Amsterdam, The Netherlands). The sequences obtained in this study are deposited in GenBank under accession numbers KC914294–KC914309 and KC914316–KC914331.
The preliminary analysis and the alignment of DNA sequences were performed with MEGA version 5 (Tamura et al., Reference Tamura, Peterson, Peterson, Stecher, Nei and Kumar2011).
Haplotype networks were obtained with Network 4.6.1.1, available at www.fluxus-engineering.com (Bandelt et al., Reference Bandelt, Forster and Röhl1999), and with TCS 2.21 (Clement et al., Reference Clement, Posada and Crandall2000), based on a COII+16S+CR alignment of the Kalotermes sp. sequences obtained in this work, adding sequences of K. flavicollis (Viterbo) and K. italicus (Grosseto) as roots for the median-joining analysis.
For the construction of phylogenetic trees, COII and 16S sequences published in previous studies (Velonà et al., Reference Velonà, Luchetti, Ghesini, Marini and Mantovani2011; Ghesini & Marini, Reference Ghesini and Marini2013; Ghesini et al., Reference Ghesini, Simon and Marini2014) from K. flavicollis and K. italicus were drawn from GenBank and added to the alignment. Sequences from the Mastotermitidae Mastotermes darwiniensis (GenBank A.N. JX144929) and from the Kalotermitidae Kalotermes brouni, Neotermes insularis, Cryptotermes brevis, and Longicaputermes sinaicus (A.N. AF189104, KF840415, JX144933, FJ806880, FJ806145, KC914288, KC914292) were used as outgroups. CR sequences were not used for phylogenetic reconstruction because sequences from different genera cannot be reliably aligned.
In order to compare European Kalotermes species with other Kalotermes species for which only COII sequences are available, an alignment was built including the COII sequences used for phylogenetic reconstruction, as well as sequences of species from Australia (Thompson et al., Reference Thompson, Miller, Lenz and Crozier2000) and Madagascar (Monaghan et al., Reference Monaghan, Wild, Elliot, Fujisawa, Balke, Inward, Lees, Ranaivosolo, Eggleton, Barraclough and Vogler2009). This alignment was also used for the construction of a Bayesian tree to be used as input for the Species Delimitation plugin (Masters et al., Reference Masters, Fan and Ross2010) of the software Geneious 7.1 (Biomatters), that computes some measures of taxon distinctiveness, including inter-species tree distances, Rodrigo's test, and Rosenberg's reciprocal monophyly test. Rodrigo's test estimates the probability that a clade has the observed degree of distinctiveness due to random coalescent processes; Rosenberg's test tests the hypothesis that monophyly is a chance outcome of random branching (Rosenberg, Reference Rosenberg2007; Rodrigo et al., Reference Rodrigo, Bertels, Heled, Noder, Shearman and Tsai2008; Boykin et al., Reference Boykin, Armstrong, Kubatko and De Barro2012).
Models of nucleotide substitution were tested with JModelTest 2.1.2 (Darriba et al., Reference Darriba, Taboada, Doallo and Posada2012), with the corrected Akaike Information Criterion. A partition homogeneity test, to determine whether COII and 16S datasets could be included in the same dataset, was performed with PAUP* 4.0 (Swofford, Reference Swofford2003), and the p-value was determined based on 200 replicates. Maximum likelihood trees were obtained with PhyML 3.1 (Guindon & Gascuel, Reference Guindon and Gascuel2003), and bootstrap values were calculated after 300 replicates. Maximum parsimony analyses were performed with PAUP* 4.0. Bootstrap values were calculated with 1000 replicates. Bayesian trees were obtained with MrBayes 3.1.2 (Huelsenbeck et al., Reference Huelsenbeck, Ronquist, Nielsen and Bollback2001; Ronquist & Huelsenbeck, Reference Ronquist and Huelsenbeck2003). Convergence was reached after one million generations (average standard deviation of split frequencies <0.01), and 25% of the trees were discarded as burn-in.
Morphological observations and analyses were conducted as in Ghesini & Marini (Reference Ghesini and Marini2013). A total of 39 alates (17 from Lebanon and 22 from Cyprus) and 27 soldiers (4 from Lebanon, 9 from Israel, and 14 from Cyprus) were observed and measured. Mann–Whitney U test was conducted using the software Past (Hammer et al., Reference Hammer, Harper and Ryan2001).
Results
In Kalotermes sp. from the Levant, 16, 7, and 9 haplotypes were found, respectively, for COII, 16S, and CR sequences (table 1). The two individuals analyzed for each colony have the same haplotype, except in three colonies (Benouaiti C, Hadera C, and Rishpon B) for the gene COII. For all the three regions sequenced, the haplotypes found in Cyprus are not found in Lebanon and Israel, and vice versa. For the gene COII, Lebanese haplotypes differ from Israeli haplotypes, while for 16S and CR a haplotype (S5 and R6 respectively) is found both in Lebanon and Israel.
Table 1. COII, 16S, an CR haplotypes of Kalotermes sp. isolated in this study and corresponding GenBank accession numbers. Numbers of localities are as in fig. 1. Different colonies collected in the same locality are indicated by a capital letter (A, B, and C). Different haplotype combinations found in the same colony are indicated by a number (1, 2).
Based on nucleotide p-distances between combined COII+16S+CR haplotypes (table 2), samples from Lebanon show a high affinity with those from Israel, with distance levels (0.14–2.04%) that fall within the range of distances found among Israeli samples (maximum distance 2.04%). Samples from Cyprus are more distant (2.64–3.32% with Lebanon, 2.64–3.53% with Israel).
Table 2. Minimum and maximum percent nucleotide p-distances between combined haplotypes (COII+16S+CR) of different colonies from Cyprus, Lebanon, and Israel and within-group distances. Standard deviations are shown in parentheses. Distances with K. flavicollis (Viterbo) and K. italicus (Grosseto) are shown for comparison.
Based on COII sequences, p-distances between Kalotermes sp. from the Levant and the other European species (from 9.37 ± 1.35% to 11.03 ± 1.38% with K. flavicollis, from 10.88 ± 1.24% to 12.39 ± 1.47% with K. italicus) are comparable to or higher than several interspecific distances in the genus Kalotermes (e.g., from 6.04 ± 1.10% to 8.31 ± 1.30% between K. flavicollis and K. italicus, 7.57 ± 1.27% between Kalotermes aemulus and Kalotermes hilli, 12.01 ± 1.17% between K. aemulus and K. brouni). Based on the Bayesian COII tree, the distance between Kalotermes sp. from the Levant and the group K. flavicollis+K. italicus is 24.9%. Rodrigo's and Rosenberg's tests support the distinction of the group K. flavicollis+K. italicus from Kalotermes sp. from the Levant (P < 0.05 and P < 105 respectively).
The median-joining and the parsimony haplotype networks, based on the COII+16S+CR alignment, identify two main groups, the first includes haplotypes from Cyprus and the second includes haplotypes from the mainland (Lebanon and Israel) (fig. 2). In the parsimony haplotype network, with a 95% cutoff, the two groups are disjointed. The correspondence between the collocation of haplotypes in the network and the geographic location of sampling localities is fairly good. The Cypriot haplotype nearest to the mainland group is the one found in Nikokleia. The mainland haplotype nearest to the Cypriot group is the one found in Hadera. The sampled haplotypes nearest to K. flavicollis and K. italicus are those from the mainland group, in particular those from Hadera C.
Fig. 2. Median-joining haplotype network based on the alignment COII+16S+CR. Numbers near lines represent the number of changes between haplotypes. Small black dots represent median vectors (unsampled or hypothetical haplotypes at the intersection of two or more lines). The attribution of median vectors to a geographical area is arbitrary.
JModeltest 2.1.2 indicates TPM3uf+I+G and TPM2uf+G as the best models of nucleotide substitution, for COII and 16S datasets respectively. The partition homogeneity test allows to include the two genes in the same dataset (P = 0.29). For the combined COII+16S dataset, JModelTest 2.1.2 chooses TPM2uf+I+G as the best model of nucleotide substitution.
The topologies of maximum likelihood, maximum parsimony, and Bayesian inference trees based on the COII+16S dataset are coincident (fig. 3). In the phylogeny of central and eastern Mediterranean Kalotermes species, the samples of Kalotermes sp. from the Levant form a well supported clade, sister to the clade including K. flavicollis and K. italicus. Two main subclades are identified, one including all the samples from Cyprus, and the other including all the samples from Lebanon and Israel. The internal branching of these subclades is only partially resolved. The topology of Kalotermes sp. clade shows a good correspondence with the structure of the haplotype network (fig. 2).
Fig. 3. Maximum parsimony, maximum likelihood, and Bayesian inference tree (topologies are coincident) based on COII+16S alignment. Numbers at nodes represent bootstrap and posterior probability values (MP/ML/BI). Support values of internal nodes are shown on the left of the figure.
The results of the morphological analysis are shown in the description of the new species (see below).
Discussion
In the phylogeny of Mediterranean Kalotermes species, samples from Cyprus, Lebanon, and Israel form a well supported clade, sister to the clade including K. flavicollis and K. italicus. Both the haplotype network and the phylogenetic analysis identify two groups, one including samples from Cyprus and the other including samples from Lebanon and Israel. The high affinity between Israeli and Lebanese samples was expected due to the geographical proximity and the absence of relevant barriers to termite dispersion between the two countries.
In the haplotype network, the haplotypes nearest to the path connecting the Cypriot group to the Israeli-Lebanese group are those found in Nikokleia and Hadera, suggesting that the colonization route between Cyprus and the mainland might have connected southern Cyprus to northern Israel. In order to rule out other possible origins of Cypriot Kalotermes sp., it would be useful to analyze samples of Kalotermes taxa from nearby lands, such as Egypt, Syria, and Turkey, for which no genetic data exist to date.
Samples from Cyprus, although genetically distinguishable from those from Lebanon and Israel, are strongly related to them. This is in contrast to what is found in the case of Reticulitermes sp. from Cyprus, that is genetically distant from the Israeli Reticulitermes clypeatus, and instead shows affinities with populations from northeastern Greece (Ghesini & Marini, Reference Ghesini and Marini2012). Thus, the colonization of Cyprus by the two termite genera seems to have occurred through different routes.
Cyprus is an oceanic island, i.e., it was never connected with the mainland. It began emerging from the sea about 20 million years ago, and, even at times of minimum sea level, was probably separated from the surrounding lands by at least a 30–40 Km water gap (Simmons, Reference Simmons1999). Unlike Reticulitermes termites, Kalotermes termites can tolerate high salinity levels, such as those of sea water (Springhetti, Reference Springhetti1959). Their ‘one piece’ wood nesting behavior further facilitates over-water dispersal, scarcely exploitable by Reticulitermes termites. The colonization of islands by wood rafting is a concrete possibility for termites (Gathorne-Hardy et al., Reference Gathorne-Hardy, Jones and Mawdsley2000).
The monophyly of the Cypriot clade supports the hypothesis that Kalotermes sp. established on Cyprus following a single colonization event. This is not so surprising as it might seem, because, from a biogeographical point of view, Cyprus is one of the most isolated Mediterranean islands, with a very low rate of faunistic immigration (Corti et al., Reference Corti, Masseti, Delfino and Pérez-Mellado1999). Despite the frequent and long-standing movement of people and goods between the mainland and Cyprus, we found no evidence of human-mediated dispersal of Kalotermes sp. between the two regions.
Until the 1970s, two termite species were believed to be widely distributed along the Mediterranean coasts of Europe and Asia: the subterranean termite Reticulitermes lucifugus (Rossi) and the drywood termite Kalotermes flavicollis (Fabr.). During the last decades, morphological, chemical, and genetic analyses of Reticulitermes samples collected over most of R. lucifugus range have shown that what was once believed to be a single species is in fact a complex of about 10 taxa, five of them already described as species (e.g., Clément, Reference Clément1978; Clément et al., Reference Clément, Bagnères, Uva, Wilfert, Quintana, Reinhard and Dronnet2001; Velonà et al., Reference Velonà, Ghesini, Luchetti, Marini and Mantovani2010; Ghesini & Marini, Reference Ghesini and Marini2012), so that the range of R. lucifugus is actually limited to south-eastern France and Italy. Similarly, in the case of K. flavicollis, a more complex pattern than was previously known is now emerging from genetic and morphological analyses of long-known populations and to the acquisition of samples from new localities. The first genetic analyses, conducted on samples from Italy and the Balkan peninsula, showed a remarkable genetic homogeneity across K. flavicollis populations (Luchetti et al., Reference Luchetti, Bergamaschi, Marini and Mantovani2004). The extension of analyses to populations from southern France, Corse, and Sardinia revealed a fair degree of genetic variation (Velonà et al., Reference Velonà, Luchetti, Ghesini, Marini and Mantovani2011). Then, the analysis of an Italian dark-necked Kalotermes population lead to the description of K. italicus (Ghesini & Marini, Reference Ghesini and Marini2013). In this study, we show that eastern Mediterranean Kalotermes populations from Cyprus, Lebanon, and Israel are different from K. flavicollis, and can be considered as a new species, that we describe below.
The distribution of Mediterranean Kalotermes spp. samples whose species attribution has been verified to date through genetic analyses (Luchetti et al., Reference Luchetti, Bergamaschi, Marini and Mantovani2004, 2013; Velonà et al., Reference Velonà, Luchetti, Ghesini, Marini and Mantovani2011; Ghesini & Marini, Reference Ghesini and Marini2013; this study) is summarized in fig. 4. None of the samples collected in Cyprus, Lebanon and Israel – some of them collected in localities where K. flavicollis had been previously reported – belongs to the species K. flavicollis. Because Kalotermes sp. from the Levant and K. flavicollis are morphologically similar and have the same ecology, it is likely that the previous reports of K. flavicollis in these regions are in fact attributable to Kalotermes sp.
Fig. 4. Localities whose Kalotermes colonies have been analyzed from the genetic point of view (mitochondrial DNA and/or microsatellite analysis). Black dots: Kalotermes flavicollis; black squares: K. italicus; open squares: hybrids or probable hybrids (individuals with K. flavicollis phenotype and K. italicus mitochondrial haplotypes) between K. flavicollis and K. italicus; black triangles: K. phoeniciae sp. nov. (see the text for references).
Kalotermes phoeniciae sp. nov. (fig. 5)
Holotype (female alate): sample BEN A2. Paratypes (17 female alates, 21 male alates, 27 soldiers in total): samples AGI; BEN A, B, C; EPI; HAD A, B, C; KAP; KAT; KFA; MAK; NIK; RIS A, B, C; SDO. All these series are in the M. Marini collection, University of Bologna, Italy.
Fig. 5. Main diagnostic characters for K. phoeniciae sp. nov.: arolium of the alate (A), mandibles (B) and rudimental eye (C) of the soldier. Arrows indicate the extremities of the eye, along its major axis.
Alate
Body dark brown. Clypeus, antennae, tibiae, and tarsi, light brown-yellowish. Pronotum yellow. Wing membrane smoky brown.
Eye with 169–252 ommatidia (mean 204.33 ± 19.75). Ocellus separated from the eye by a distance of approximately a half ocellus diameter. Antennal segments 15–19 (Cyprus: 15–18; Lebanon: 16–19).
Arolia big (fig 5A), of about the same size in all legs (75–80 μm in length, nearly half the length of the claw). Wing venation variable among individuals and often different in the contralateral wings of the same individual. Radial sector sclerotized, with 9–11 branches in the forewing and 6–8 branches in the hind-wing. Media unsclerotized, usually well developed and reaching, with its ramifications, the wing edge; sometimes short, ending before the wing edge or joining the cubitus. Cubitus unsclerotized, with a variable number of branches and sub-branches, often occupying half or more of the area of the wing.
Measurements: see table 3.
Table 3. Samples used for this study: collection date, collector, geographic coordinates, elevation (m a.s.l.), and castes (P: pseudergates, S: soldiers, A: alates).
Soldier
Head light brown, darkening toward mandibles. Mandibles dark brown. Thorax and abdomen whitish.
Mandibles broad, with big basal tubercles on their outer sides (figs 5B and 6). Maximum diameter of the rudimental eye: 160–210 μm, about half the maximum external diameter of the antennal socket (fig 5C). Antennal segments 13–16 (Cyprus: 13–16; mainland: 13–15). Second and third antennal segment of about the same length; third shaped as a truncated cone; fourth shortest.
Fig. 6. Outline of the left mandible of K. flavicollis and K. italicus (left) and K. phoeniciae (right). Arrows indicate basal tubercles.
Measurements: see table 4.
Table 4. Measurements (mm) of alates of K. phoeniciae sp. nov. An asterisk (*) indicates characters with significantly different values in Cyprus and in Lebanon (Mann–Whitney test, P < 0.01).
Measurements of alates and soldiers are reported for completeness (tables 4 and 5), but should be considered with caution, because in Kalotermes spp. measurements can vary amply among colonies and even among individuals of the same colony, and the ranges of variation of measurements of different taxa often largely overlap.
Distribution
Cyprus, Lebanon, Israel
Ecology
Diverse environments, from arid or semi-arid to shaded creek banks, from 0 to 1100 m a.s.l. All colonies found in natural environment, except in the case of Rishpon (avocado orchard).
Host plants: colonies found in tree stumps, logs, or under the bark of living plants of Ceratonia siliqua L., Ficus carica L., Juniperus sp., Persea americana Mill., Pinus pinea L., Platanus orientalis L., Salix sp., Tamarix sp.
Comparison with the other Mediterranean Kalotermitidae
The pronotum of the alate of K. flavicollis is often of a duller shade of yellow (with dark margins, or even entirely dark, in K. flavicollis var. fuscicollis). Arolia are slightly smaller (about 60 μm in length, less than 4/10 of the length of the claw). The alate of K. italicus has a dark brown, almost black pronotum, unpigmented wings, and much smaller arolia (about 40–45 μm in length, less than 3/10 of the length of the claw) (Ghesini & Marini, Reference Ghesini and Marini2013). No description of the alate of K. monticola exists.
Soldiers of both K. flavicollis and K. italicus mandibles with smaller basal tubercles (fig. 6). The soldier of K. flavicollis has bigger rudimental eyes (maximum diameter longer than 230 μm, more than half the antennal socket). Most soldiers of K. italicus have smaller rudimental eyes (maximum diameter shorter than 170 μm, less than half the antennal socket). The soldier of K. monticola lacks basal tubercles on the outer sides of mandibles and has two frontal depressions (Sjöstedt, Reference Sjöstedt1925).
For the identification of the Mediterranean species in the genus Kalotermes, we provide the following keys.
Identification key for alates
-
1. Length of arolium:
-
nearly half the length of the claw: K. phoeniciae
less than half the length of the claw: 2
-
-
2. Color of the wings:
-
hyaline, whitish when folded: K. italicus
dark: K. flavicollis
-
Identification key for soldiers
-
1. Outer side of mandibles:
-
without basal tubercles: K. monticola
with basal tubercles: 2
-
-
2. Size of basal tubercles:
-
3. Maximum diameter of the rudimental eye:
-
less than 170 μm: K. italicus
more than 230 μm: K. flavicollis
-
Table 5. Measurements (mm) of soldiers of K. phoeniciae sp. nov. Measurements on soldiers from Cyprus and from the mainland do not differ significantly, except in the case of head length (Mann–Whitney test, P = 0.02).
Acknowledgements
This study was supported by Canziani Donation and Sireb S.a.s. funds. We wish to thank Dany Simon of Tel Aviv University for assistance and advice during the collecting trip in Israel, and Elie El Jiz for helping to collect samples in Lebanon.
