The availability of the complete cucumber genome sequences facilitated the search for NBS-encoding genes. At present, two cucumber inbred lines, 9930 (the ‘Chinese long’, commonly used in modern cucumber breeding) and Gy14 (the gynoecious inbred line), have been sequenced. The former was sequenced using a combination of traditional Sanger and next-generation Illumina GA sequencing technologies 10 and a database has been established for the sequence formation (http://cucumber.genomics.org.cn/page/cucumber/index.jsp). The latter was sequenced de novo with an appropriate mixture of random shotgun and paired-end shotgun reads using a 454-XLR technology; these sequences have been uploaded to the JGI Genome database (http://genome.jgi-psf.org/cucumber/cucumber.home.html). In this study, the two databases were used to search for NBS-encoding genes.

NBS-encoding genes were identified for the first time in three steps using the JGI Cucumis sativus Genome database. The first step involved a BLASTN search using A. thaliana and rice NBS-encoding sequences 31 32 33 as the query. The second step aimed at a complete search of the candidate NBS-encoding genes in the cucumber gynoecious inbred line Gy14, and was performed using the amino acid sequence of the nucleotide-binding adaptor shared by APAF-1, R-proteins, and the CED-4 (NB-ARC) domain (Pfam: PF00931) as a query to find possible encoded homologues 22 . In the third step, based on the above results, the search of candidate NBS-encoding genes in the cucumber genome was repeated using BLASTN searches. The overall analysis reveals that the NBS-encoding gene family is composed of 57 members in the ‘Gy14’ cucumber gynoecious inbred line (Table  1). The predicted nucleotide and protein sequences of all 57 NBS-encoding genes from ‘Gy14’ are provided in Additional file 1.

*These sequences are pseudogenes. ^#These genes contained no complete conserved motifs of P-loop to GLPL.

To classify these NBS-encoding proteins, each was identified based on N terminal CC motifs and TIR domains, as well as LRRs. A total of seven categories (TNL, CNL, TN, CN, N, NL and RPW8-NL) were identified (Table  2). Twenty-three proteins were shown to possess only NBS-LRRs. Two proteins, Cucsa.102240 and Cucsa.123410, were predicted to have a domain with a sequence similar to the RPW8 Arabidopsis powdery mildew resistance gene family 34 . Thirteen genes were identified with TIR domains at the N-terminal, and two of them, Cucsa.338660 and Cucsa.091460, were shown to lack LRRs. Eighteen genes were predicted to encode the CC motifs, and one of these genes lacked LRRs. The gene, Cucsa.237530, possesses only NBS domains (Table  2; Additional file 2).

Previously, the number of NBS-encoding genes from the cucumber inbred line ‘Chinese Long’ 9930 had already been reported 10 . In this study, the methods mentioned above were also implemented to identify NBS-encoding genes from the ‘Chinese Long’ cucumber 9930 database (http://cucumber.genomics.org.cn/page/cucumber/index.jsp). These NBS-encoding genes are shown in Additional file 1. A comparison of Gy14 and 9930 cucumber NBS-encoding genes showed that most of the genes were highly similar to one another (Additional file 3). Therefore, in order to perform a comprehensive analysis of the NBS-encoding genes within the cucumber genome for comparison purposes, only NBS-encoding genes from the cucumber ‘Gy14’ genome were selected for further analyses.

Generally, 5’ region preceding the NBS and 3' region following the NBS have high variability and are not included for construction. The NBS region, however, is highly conserved and is often used to generate multiple sequence alignments and phylogenetic tree constructions 30 . In order to elucidate the relationships among the cucumber NBS-encoding genes, sequences of the NBS domain of these genes were used to construct a neighbor-joining (NJ) phylogenetic tree. These genes were divided into two families, the CC-NBS and TIR-NBS, which are supported by the high bootstrap values (Figure  1A). This was consistent with the results reported by Pan et al., which showed that both the TIR-NBS and CC-NBS families of genes occur in dicotyledonous plant species 24 . Moreover, each of these two families was separated into two subfamilies (CC-I/CC-II and TIR-I/TIR-II) with high bootstrap values (Figure  1A). In addition, it was observed that the members in CC-NBS were more numerous than those in the TIR-NBS family, which are composed of 34 and 23 members, respectively. This distribution is incongruent with that found in Arabidopsis. In the Arabidopsis genome, the member of the CC-NBS family was less than that of the TIR-NBS family 31 . Recently, the melon genome sequence was available 35 , this distribution was also found in the NBS-LRR family (data not shown).

The exon/intron positions and phases of the 57 NBS-encoding genes in cucumber were further analyzed. The exon/intron structures were obtained using the online Gene Structure Display Server (GSDS: http://gsds.cbi.pku.edu.cn) with both coding sequences and genomic sequences 36 . Figure  1B provides a detailed illustration of the relative lengths of the introns and conservation of the corresponding exon sequences within each NBS-encoding gene in the cucumbers. The number of introns in all of these genes ranged from zero to seven. No intron was found in the Cucsa.338110, Cucsa.338190 or Cucsa.237530, whereas seven were found in Cucsa.132370. More than half of the genes had one to three introns (Additional file 4). Moreover, in the CC-I subfamily, all of the genes contained the lowest amount of introns (zero to three), except for Cucsa.132370, which contained seven introns. The genes in the TIR-II had the greatest number of introns (three to six), except for Cucsa.338660, which had only one intron. In the remaining two subfamilies, CC-II and TIR-I, the NBS-encoding genes contained zero to six introns (Figure  1B). These findings, together with the phylogenetic tree, indicate that some intron loss and intron gain events may have occurred during the structural evolution between the two families of cucumber NBS-encoding genes.

In order to understand the fine structure of the NBS-encoding genes in the cucumbers, CC- and TIR-NBS families were analyzed separately using the MEME and Clustal X programs. Moreover, these proteins usually have an N-terminal region, NBS domain, and LRR and C-terminal regions. Therefore, each of the two families of proteins were separated into three parts and subjected to protein domain and motif analyses.

The GY14 cucumber genome encoded 57 members. Of them, 55 NBS-encoding genes could be located on all seven chromosomes (Figure  2). The remaining two genes (Cucsa.326910 and Cucsa.155730) could not be located on any chromosomes and were assigned to Scaffold03139 and Scaffold03138, respectively. As in Arabidopsis and rice, the distribution of cucumber NBS-encoding genes among chromosomes is nonrandom 31 32 33 . There were 10 and 20 genes which were localized on chromosomes 5 and 2, respectively, whereas only five genes were located on chromosomes 1 and 6. The remaining 20 NBS-encoding genes were located on chromosomes 3, 4 and 7.

In addition, it was determined that the NBS-encoding gene clusters by a combination of three approaches described in previous studies 31 37 38 . A gene cluster is defined as a region in which two neighboring homologous genes are less than 200 kb and fewer than eight non-NBS-encoding genes between NBS-encoding genes. It was also found that most of the NBS-encoding genes were clustered on the chromosomes. A total of 33 genes were localized within nine clusters. The largest cluster contained 10 NBS-encoding genes, while the smallest included only two genes, both of which were located on chromosome 2. Shared phylogenetic clades confirmed the NBS-encoding gene chromosomal duplication, with four and seven NBS-encoding gene segment and tandem duplication events in cucumber (Figure  2), and these genes concern chromosomes 2, 3, 6 and 7.

To gain an insight into the phylogenetic relationship of NBS-encoding genes in Cucurbitaceae crops, database mining and PCR amplification were employed to identify the NBS-encoding genes in melon, bottle gourd, luffa, watermelon, and squash (Table  1, Additional file 7). A total of 165 NBS sequences were obtained. Among these sequences, 43 belonged to bottle gourd, 20 to luffa, 24 to melon, 64 to watermelon, and 14 to squash. For bottle gourd and luffa, the NBS sequences were obtained by PCR amplification; for squash, the NBS sequences were derived from database mining; for melon and watermelon, 10 and 31 sequences were respectively drawn from PCR amplification, and the remaining 14 and 33 sequences were obtained by database mining. The NBS sequences from watermelon originated from two species. One was Citrullus lanatus, consisting of 58 NBS sequences. Among them, one NBS sequence, GU124560, is a pseudogene (gene with stop codon), and two NBS sequences, GU124553 and GU124554, lacked the complete conserved motifs from P-loop to GLPL. The remaining six NBS sequences were found in Citrullus colocynthis, in which GU124552 and GU124555 were pseudogenes. Among the 24 NBS sequences from melon, two are pseudogenes (AF354508 and AF354512), which were found in the C. melo species. The AY583855 sequence has been identified as a Fusarium R gene 39 . In addition, the NBS sequences were obtained from bottle gourd (L. siceraria), luffa (L. cylindrica), and squash (C. moschata). In the current study, the pseudogenes and sequences without conserved domains from P-loop to GLPL were excluded from further analysis. Among the 165 NBS sequences from the Cucurbitaceae crops, 103 NBS sequences (JN230598 to JN230701) were identified via degenerate PCR amplification. The remaining 62 were derived from the GeneBank database (Table  1).

Phylogenetic analysis via the NJ method was conducted in order to determine the relationships among NBS-encoding genes and RGHs in the Cucurbitaceae crops. The consensus phylogenetic tree (Figures  3, Additional file 8) indicated that there were two distinct families, namely TIR-NBS and CC-NBS, which were consistent with the pattern previously described 23 40 . TIR-NBS and CC-NBS were further subdivided into nine and four distinct subfamilies (TIR1 to TIR9 and CC1 to CC4), respectively.

TIR1, TIR3 and TIR6 were the largest among the subfamilies, which were composed of 43, 35 and 40 sequences, respectively. TIR1 contained 29 sequences from bottle gourd, 5 from watermelon, 7 from luffa, and 2 from cucumber. The TIR3 subfamily included 13 sequences from cucumber, 14 from melon, and 8 from watermelon. The TIR6 subfamily contained 27 sequences from watermelon, 10 from squash, 2 from cucumber, and 1 from luffa. The TIR2, TIR5, TIR8 and TIR9 subfamilies are small, composed of 2, 2, 4, and 1 sequence, respectively. The remaining two subfamilies, TIR4 and TIR7, consisted of 10 and 15 sequences, respectively. Within the CC-NBS family, the largest subfamilies, CC1 and CC4, consist of 20 and 28 sequences, respectively, while CC2 and CC3 were relatively small subfamilies, composed of 3 and 12 sequences, respectively (Figures  3 and S4).

Interestingly, all NBS sequences obtained via PCR amplification in luffa belonged to the TIR (TIR1, TIR6 and TIR7) families, whereas the NBS sequences from the other five species were distributed widely among TIR- and CC-NBS families (Figures  3, Additional file 8).

In the following analysis, A. thaliana, a member of Brassicaceae and the model system for genomic comparisons, was selected for phylogeny. The phylogenies were constructed using the P-loop to GLPL motifs. Previous phylogenetic analyses showed that all NBS sequences in A. thaliana are separated into two distinct families, CC and TIR 23 . A great divergence was observed between the two families. Therefore, separate analyses of these families were performed, and the results are shown in this paper. Figure  4 and Additional file 9 show the results of the phylogenetic analysis of the CC-NBS family between the Cucurbitaceae crops and A. thaliana. All members from the CC1, CC2, CC3 and CC4 subfamilies were detected in four clades (N2, N1, N4, and N3, respectively), using reference sequences from Arabidopsis as described by Cannon et al. 41 . Among the four clades, N2 was the largest, being composed of 12 members from cucumber, 3 from melon, 3 from bottle gourd, and 10 from watermelon. The other three clades were relatively small. N1 clade included all CC3 subfamilies from the Cucurbitaceae crops and two members, At3g14460 and At3g14470, from A. thaliana. However, relatively lower bootstrap values were observed in N1 and N2 clades. Low scores are usually an indication that the observed patterns must be analyzed with caution, and are more often observed for large datasets. Such poor scoring does not necessarily imply unreliable branching, but instead indicates that not all members may be assigned to a particular group with high confidence. The NBS-LRR superfamily accounts for the largest number of known disease resistance genes, and is one of the largest gene families among plant genomes 42 . High diversification of these genes was observed in the plants 30 38 40 . Therefore, it is believed that the diversification of NBS-encoding genes was responsible for the poor bootstrap score. This phenomenon has occurred frequently in constructing phyologenetic trees of NBS-encoding resistance genes from other plant species 7 43 44 .

Clade N3 included 8 sequences from cucumber, 8 from bottle gourd, 3 from squash, and 1 from watermelon. Clade N4 only contained three sequences from cucumber. Within the TIR family (Figures  5, Additional file 10), the nine subfamilies from the Cucurbitaceae crops remained different when compared with the TIR family of the sequences from A. thaliana. Sequences from TIR4 and TIR9 were detected in the At-TIR-NBS-B and At-TIR-NBS-A subfamilies, respectively. Each of the remaining seven subfamilies from the Cucurbitaceae crops formed a group.

The general features of the phylogenetic tree are presented in Figures  4 and 5. In most cases, each clade in the Cucurbitaceae crops was composed of sequences from more than one species. For example, clade N3 was composed of CC1 sequences from four species (cucumber, bottle gourd, squash, and watermelon), whereas clade N2 consisted of CC4 sequences from cucumber, melon, bottle gourd, and watermelon. A similar phenomenon was also observed in the TIR1 to TIR8 subfamilies. However, in TIR9 and CC2, squash and cucumber were species-specific subfamilies, that is, they were composed of sequences from only one species.

Previous studies regarding NBS-encoding genes from Vitis vinifera, Populus trichocarpa, Arabidopsis thaliana, and Oryza sativa revealed that there are 535, 416, 174, and 519 genes in these species, respectively 46 . However, the GY14 cucumber genome only encodes 57 NBS-encoding genes, which is less than that in the four sequenced plant genomes (Additional file 11). Recently, Huang et al. identified a total of 59 NBS-encoding genes which were located on seven different chromosomes in the ‘Chinese Long’ cucumber genome 10 . These two cucumber lines have similar numbers of NBS-encoding genes (Additional file 3). Moreover, in these two cucumber genomes, the majority of conserved NBS-LRR R genes is single-copy and/or located as a singleton. The complex clustered NBS-LRR R genes contribute greatly to the rich genetic variation (data not shown). Recently, it was reported that there are 81 NBS-encoding genes in the melon genome 35 . It was found that there are significant differences in R gene counts among them. This result indicated that the degree of interspecific variation is greater than that of intraspecific variation in Cucumis.

Recently, Porter et al. reported that only 54 NBS-encoding genes are present in Carica papaya 47 , which was similar to the number of these genes in the cucumbers (Additional file 11). Although the difference between the genome size and total number of predicted protein encoded genes among the six sequenced plant species was observed clearly (Additional file 11), the number of NBS-encoding genes does not increase or decrease proportionally. The obtained data indicate that, similar to Carica papaya, NBS-encoding genes in cucumber are a relatively small gene family. Based on the information regarding the cucumber genome 10 , there are at least two explanations to this phenomenon. The first is the absence of the recent whole-genome duplication (WGD) in the small cucumber genome 10 . WGD is common in angiosperm plants and produces a tremendous source of raw materials for gene genesis, therefore the absence of WGD may have led to the small number of NBS-encoding genes in cucumber. The second explanation is the inclusion of a small number of tandem gene duplications and a few segmental duplications in the cucumber genome 10 . These duplications also contribute in part to the small number of NBS-encoding genes in cucumbers.

Cucumber genome encodes 57 NBS-encoding genes and maintains both the TIR and CC families (Additional file 11 and Figure  1A), which suggests that the cucumber has relatively few, albeit diverse, NBS-encoding genes. The sequences of TIR- and CC-NBS family of NBS-encoding genes were aligned separately. The results revealed the presence of six previously identified motifs (P-loop, Kinase-2, RNBS-B, RNBS-C, GLPL, and MHDV) in most genes (Additional file 12). Motifs RNBS-A-TIR and RNBS-D-TIR occur exclusively in the TIR-NBS R proteins (Additional file 12A), whereas RNBS-A-nonTIR and RNBS-D-nonTIR are specific to the CC-NBS R proteins (Additional file 12B). However, some motifs present in the sequence alignment were not detected using the MEME software, suggesting that these motifs were poorly conserved in some of these proteins, such as the P-loop motif in Cucsa.017490, Cucsa.088220, Cucsa.094560, Cucsa.239860 and Cucsa.318890. This phenomenon was observed only in the CC-NBS family of NBS-encoding genes in cucumbers, suggesting that these genes are more conserved in the TIR-NBS family than those in the CC-NBS family. In previous studies, some researchers found that the CC family is highly diverse and originated prior to the split between gymnosperms and angiosperms. In contrast, the TIR family is more homogeneous and was found only in dicotyledon, suggesting that it arose after the divergence of monocotyledon and dicotyledon 41 48 49 . Therefore, the results of this study were consistent with those of previous reports.

The last residue of the kinase-2 motif, D (Aspartate) or W (Tryptophan), in the NBS-encoding genes in plants has also been used to predict (95% accuracy) whether they belong to the TIR- or CC-NBS family of the NBS-encoding genes 24 . In the current paper, the last residue in most of the kinase-2 motifs of TIR-NBS family of genes is “D”, except in Cucsa.237410, Cucsa.237540, Cucsa.237520, and Cucsa.091460, in which it was substituted for “Asparagine”, “Glutamine acid”, “Threonine” and “Asparagine”, respectively. For the CC-NBS family, the last residues in the kinase-2 motifs is “W” except in Cucsa.337190, Cucsa.338110, and Cucsa.338190, in which it was replaced with “Serine”, “Serine” and “Glycine”, respectively (Additional file 12). This class not only supports the results of the above phylogenetic analysis, but also the view that both the TIR- and CC-NBS families of genes occur in dicot species 50 .

Arabidopsis is the model system for genomic comparisons among dicots, due to the fact that a complete draft of its genome is available 51 . In this study, both the TIR- and CC-NBS families were identified in all genes from the Cucurbitaceae crops (Figure  3). Separate phylogenies for the 2 families were subsequently constructed (Figures  4 and 5). The phylogenetic pattern of CC-NBS is shown in Figure  4 and Additional file 9. Subfamilies CC1, CC2, CC3 and CC4 of Cucurbitaceae fall within the subfamilies N3, N4, N1 and N2, respectively, as identified by Cannon et al. 41 No species-specific expansion in the CC-NBS family after the divergence of the Cucurbitaceae species and Arabidopsis was observed. In addition, the N1 subfamily contained only two members from Arabidopsis, At3g14460 and At3g14470, whereas subfamily CC3 of Cucurbitaceae was grouped into this subfamily. Thus, the analysis has identified a region of chromosome 3 of Arabidopsis which is potentially orthologous to the CC3 subfamily of Cucurbitaceae.

The phylogenetic comparison of TIR-NBS sequences from Cucurbitaceae and Arabidopsis revealed a degree of change as opposed to the phylogenetic pattern of the CC-NBS family (Figure  5). Similar to the results shown in Figure  3, most subfamilies were also shown to be species-specific in the phylogenetic analysis of TIR-NBS sequences from Cucurbitaceae and Arabidopsis, except for TIR9 and TIR4, which were combined into the At-TIR-NBS-A and At-TIR-NBS-B subfamilies, respectively 52 . This observation is similar to those described in Solanaceae and Asteraceae, and may be typical of other plant families as well 24 51 , which suggests recent gene radiation from a common ancestral source of NBS-encoding genes or RGHs.

Cucumber (Cucumis sativus L.) assembly and annotation V1.0 were downloaded from http://www.phytozome.net/cucumber. A TBLASTN search was used to obtain all NBS-encoding genes in the cucumber (C. sativus L.) genome. First, a TBLASTN was performed using the protein coding sequences of the NBS domain of NBS-encoding sequences from A. thaliana and rice 31 32 33 as the query against the JGI Cucumis sativus genome database (http://genome.jgi-psf.org/cucumber/cucumber.home.html). Second, the amino acid sequence of the NB-ARC domain (Pfam: PF00931) was adopted as a query in TBLASTN searches for possible homologues encoded in the cucumber genome. The conserved NBS domain of these predicted NBS-encoding proteins was determined by Pfam version 22.0 ( http://pfam.janelia.org). Third, based on the results above, the searches of candidate NBS-encoding genes in the cucumber genome were repeated using BLASTN searches. The e-value used was 1e^-5. Finally, all BLAST hits in the cucumber genome, together with flank regions of 5,000–10,000 bp in the upstream and downstream of BLAST hits, were annotated using the FGENESH (http://www.softberry.com/) and GENSCAN (http://genes.mit.edu/genescan.html/) programs.

To classify these NBS-encoding genes, all candidate genes were evaluated to further verify whether they encoded TIR, CC, NBS, or LRR motifs using the Pfam database (http://pfam.janelia.org/), SMART protein motif analyses (http://smart.embl-heidelberg.de/), and COILS, with a threshold of 0.9, to specifically detect CC domains 53 .

To investigate the diversity and structure of NBS-encoding genes in cucumbers, their predicted amino acid sequences were subjected to domain and motif analyses. According to the methods of previous researchers 31 54 , NBS-encoding genes from cucumbers were divided into three components, namely the N-terminal, NBS domain, and LRR-C-terminal regions. They were then analyzed individually using the Multiple Expectation Maximization for Motif Elicitation (MEME)/Motif Alignment and Search Tool (MAST) system (http://meme.sdsc.edu/meme/website/intro.html). Furthermore, MEME motif analyses were performed on members of TIR-NBS and CC-NBS families. Conservation of each motif among the NBS-encoding genes was performed with WebLogo version 2.8.2 (http://weblogo.berkeley.edu/) using the default settings.

Gene duplication events of NBS-encoding genes were defined based on the http://criterion used by previous researchers 55 . NBS-encoding genes in cucumber were aligned using BioEdit (http://www.mbio.ncsu.edu/bioedit/bioedit.html) and calculated by MEGA 5.0 56 for homology gene calculation.

To understand the phylogenetic relationship among the NBS-encoding genes in Cucurbitaceae crops, NBS-encoding RGHs from melon, bottle gourd, luffa, watermelon and squash were also identified via degenerate PCR amplification and database mining. First, PCR was performed using genomic DNA for young leaves from melon, bottle gourd, luffa, and watermelon using 3 pairs of degenerate primers. The young leaves in the second true-leaf stage were harvested, frozen immediately in liquid nitrogen, and stored at −80°C. Genomic DNA was isolated using a plant DNA extraction kit (Tiangen, China). The primers were designed by the previous researchers based on the conserved regions of P-loop and GLPL of amino acid identity among the known NBS-LRR R genes from the other plant species (Additional file 13). The PCR amplifications were performed in 20 μL reaction mixtures with 1 U of LATaq DNA proof reading polymerase (TaKaRa, Kyoto, Japan), 1 × PCR buffer, 1.5 mM MgCl₂, 0.5 μM each of forward/reverse primers, 0.4 mM dNTP, and 50 ng of template DNA. PCR was performed in a PTC-100 thermal cycler (MJ Research, Inc., Watertown, MA). The cycling conditions consisted of an initial denaturation performed for 3 min at 94°C, followed by 35 cycles at 94°C for 30 s, 55°C for 45 s, and 72°C for 1 min. These were followed by a 10 min extension step at 72°C and 10°C to terminate the reaction.

The DNA fragments from the PCR were separated using 1.0% agarose gels. Fragments with the expected size (~500 bp) were excised and reclaimed from the gel and purified with a PCR purification kit (Qiagen, Germany). Subsequently, these fragments were combined with vector DNA to generate recombinant DNA molecules, and then transformed into competent Escherichia coli JM10⁹ cells. Plasmid DNA was purified with a PCR purification kit (Qiagen, Germany). The DNA fragments were sequenced using an ABI 3730 sequencer (Applied Biosystems, Foster City, CA, USA). Then, each of the acquired DNA sequences was trimmed of vector sequence contamination using VecScreen at the National Center of Biotechnology Information (NCBI). Identity and similarity searches of nucleotide and amino acid sequences were performed using BLAST at the NCBI GenBank database (http://www.ncbi.nlm.nih.gov/BLAST/).

Second, other RGHs in melon, watermelon, and squash were obtained from the GenBank database searches. All sequences from these species were downloaded and searched with the NBS domain of NBS-encoding sequences from A. thaliana and rice 31 32 33 as the query. The RGHs in melon were sourced from a published paper 57 . In addition, Arabidopsis NBS-encoding proteins, which were obtained from http://niblrrs.ucdavis.edu/At_RGenes/, were selected for phylogenetic relationship analysis.

Amino acid sequences of all NBS-encoding genes in the cucumber genome and RGHs from the other five Cucurbitaceae crops were aligned using Clustal X version 1.8 58 , followed by manual adjustment. The conserved domains of P-loop to GLPL of these proteins and RGHs were applied to construct a phylogenetic tree using the NJ method 59 and an NJ algorithm implemented in the Molecular Evolutionary Genetics Analysis software version 5.0 (MEGA 5.0) 56 . Bootstrapping (1000 replicates) was used to evaluate the degree of support for a particular grouping pattern in the phylogenetic tree. Branch lengths were assigned by pairwise calculations of the genetic distances, and missing data were treated by pairwise deletions of the gaps.