Aligned sequences represent 212 bp of the ATPase gene and covers the short stretch of overlap between regions coding for ATPase 6 and ATPase 8 (Fig. 2). The first nucleotide corresponds to position 256 in the Cottus kessleri ATPase sequence (Grachev et al., 1992). Aligned sequences represent 301 bp of 12S rRNA gene. The first nucleotide corresponds to position 233 in the Mertensiella luschani 12S rRNA sequence (Titus and Larson, 1995). No insertions or deletions (‘indels’) had to be inferred to align the ATPase sequences. Alignment of the 12S sequences required two indels at one position among the Triturus species and two indels covering four positions in the out-group taxa. More indels on nine positions were required when published sequences were added to the data set, lengthening the fragment to 305 positions, with no regions of ambiguous alignment (sensu Titus and Larson, 1995).
One-hundred and sixty-one of the 212 ATPase nucleotide positions were identical across the four species studied and 38 variable positions differed by a single substitution in one taxon (Fig. 2). This left 13 (6.1 %) nucleotide positions with potentially, phylogenetically relevant information. The sequence difference between T. boscai and in-group taxa ranged between 18.9 % and 19.3 %. Within the vulgaris species group the distances were considerably lower, ranging from 3.3 % between T. montandoni and T. vulgaris to 10.4 % between T. montandoni and T. helveticus. Taking the short stretch of sequence overlap between both ATPase genes into account, the transition: transversion ratio among in-group taxa was 3.2 and the number of substitutions at the first, second, and third codon position were 5, 3, and 17, respectively. Following these relationships, the weights assigned to transitions versus transversions were 3 and 21 whereas the weights applied to the first, second and third codon positions were 8, 14, and 2. At the level of amino acid codon usage, 16 variable characters were found with nine synapomorphic character states for the T. helveticus - T. montandoni - T. vulgaris clade, four autapomorphic characters states for T. helveticus and three synapomorphies were found for T. montandoni - T. vulgaris. The DNA sequences of T. vulgaris and T. montandoni differed only by silent substitutions.
Two-hundred and thirty-five of the 301 nucleotide positions on the 12S fragment were identical across the 12 taxa studied and 25 variable positions differed by a single substitution in one taxon, leaving 41 (13.6%) phylogenetically informative sites (Fig. 2). Half the number of variable sites are found in one-third of the 12S fragment, at the 3’-end. The sequence difference between out-group and in-group taxa ranged between 6.0 % and 14.0 %. Within the genus Triturus the distances were considerably lower, ranging from 0.7 % between T. montandoni and T. vulgaris [which is less than that found within T. alpestris (3.0 %) and T. vittatus (1.7 %)] to 10.6 % between T. cristatus and T. alpestris. The average sequence difference between in-group species and an out-group taxon ranged from 7.4 % (Cynops) to 11.2 % (Neurergus). The transition:transversion ratio among in-group taxa was 2.8 and weights determined were 4 and 20.
Given Cynops, Neurergus and Paramesotriton as out-groups, the monophyly of Triturus appears to be strongly supported with a 89 % bootstrap replication score (Pb) (Fig. 3). Two major in-group clusters are formed. The first, with Pb = 85 %, consists of T. vittatus as a sister taxon to the T. marmoratus species group (Pb = 98 %). The second group contains T. alpestris and all small-bodied Triturus species, with a low bootstrap support (Pb = 59 %). Within this group, the monophyly of the T. vulgaris species group is strongly supported (Pb = 100 %), whereas the relationship of T. alpestris, T. boscai, and T. italicus among themselves is poorly resolved (Pb < 55 %). On the basis of 12S sequence data no firm conclusion can be drawn regarding the branching order within the vulgaris species group. However, the ATPase sequence data provided strong support for the sister taxon status of T. montandoni and T. vulgaris (Pb = 96, Fig. 3) and the same bootstrap value was observed when amino acid sequence data were analyzed instead of the DNA sequence. Under the weighting scheme topologically similar results were obtained, with three reservations : 1) with the ATPase gene fragment, a score of Pb = 100 was observed for the vulgaris - montandoni clade, while with the 12S rRNA gene fragment; 2) generally less phylogenetic resolution and lower bootstrap replication scores were observed on well-established branches; and, 3) an equally parsimonious solution was found in which the genus Triturus is paraphyletic. Sequences derived from the same or congeneric species by different authors (Caccone et al., 1994; Titus & Larson, 1995; Hay et al., 1995 and present study) are very similar to one another and consequently the taxa they represent are placed close together in the phylogenetic tree that has maximum parsimony (i.e., Cynops sp., Neurergus strauchii, Notophthalmus viridescens, Paramesotriton sp., Pleurodeles waltl, Salamandra salamandra, Triturus alpestris, T. cristatus superspecies and T. vulgaris).
With a wide range of out-groups included in the analysis, the monophyly of the genus Triturus was not supported. The sister clade to the subgenus Palaeotriton would not be the subgenus Triturus (i.e., T. vittatus, T. marmoratus, and the T. cristatus superspecies), but the subgenus Triturus plus Taricha and Notophthalmus and the sister clade to this group would be composed of Cynops, Euproctus, Neurergus, Pachytriton, and Paramesotriton (Fig. 4A). Templeton’s test indicated that the difference between the shortest tree and the tree in which Triturus is monophyletic was significant (P < 0. 05).