Dichromacalles algecirasensis, Stüben, Peter E., Schütte, André & Astrin, Jonas J., 2013

Stüben, Peter E., Schütte, André & Astrin, Jonas J., 2013, Molecular phylogeny of the weevil genus Dichromacalles Stüben (Curculionidae: Cryptorhynchinae) and description of a new species, Zootaxa 3718 (2), pp. 101-127 : 105-107

3.2. Dichromacalles tree

( Fig. 2 )

Molecular analysis is based on 32 specimens from 9 out of 12 known Dichromacalles species ( D. algecirasensis sp. n. included) and 5 outgroup species from other Cryptorhynchinae genera, all also included in the Cryptorhynchinae tree ( Fig. 1 ). See Table 2 for collecting and vouchering information and GenBank accession numbers as well. A phylogeny of some Dichromacalles species in combination with the Madeiran clade and Canary clade of Acalles s.l. as sister group can be found in STÜBEN & ASTRIN (2010b: fig. 1A.). Voucher specimens and extracted genomic DNA are deposited at the Biobank of the Zoologisches Forschungsmuseum Alexander Koenig, Bonn, Germany (ZFMK). The laboratory routine followed ASTRIN & STÜBEN (2008). PCR primers were taken from ASTRIN & STÜBEN (2008), COI is based on the FOLMER et al. (1994) region; 16S is based on the CRANDALL & FITZPATRICK (1996) region, 28S is used from ASTRIN & STÜBEN (2008). For detailed primer information see Table 3 .

DNA sequence alignments for COI, 16S and 28S gene were performed with BioEdit 7.0.9 using ClustalW algorithm with default parameters. Primer sequences were trimmed; single nucleotide polymorphisms and gaps of 16S and 28S alignments were manually shifted to minimize differences between sequences, especially to prevent gaps at the beginning or end of a sequence. Missing data (whole gene or missing nucleotides at the beginning or end of a sequence) were filled up with “n”. 16S sequence data were not available for one specimen ( Dichromacalles diocletianus , 587). 28S sequence data were not available for four specimens ( D. boroveci , 921; D. diocletianus, 1057 ; D. dromedarius , 862; D. tuberculatus , 743).

Alignment length was 658 nt for COI, 524 nt for 16S (poorly aligned sequence data not counted), and 365 nt for 28S (poorly aligned sequence data not counted).

The nucleotide substitution model was determined for every single gene alignment using jModelTest ver. 0.1.1 (POSADA 2008) implementing the Bayesian Information Criterion (BIC; SCHWARZ 1978), ambiguous sequence data of 16S and 28S was not provided to jModelTest. jModelTest calculation resulted in three submodels of GTR (LANAVE et al. 1984, COI: TrN+I+G, 16S: TIM3+I+G, 28S: TPM2uf+G), so we implemented the GTR+I+G model for COI and 16S and the GTR+G model for 28S.

Afterwards a concatenated sequence block was built from COI, 16S and 28S alignments, providing 1708 nucleotide positions. Ambiguous data of 16S and 28S were kept in the concatenated data block, but were excluded in the phylogenetic analysis. The 16S data contained four poorly aligned positions (699–705, 924–933, 1020–1026, 1208), the 28S data contained two (1244–1262 1493–1627), so 1528 nucleotide positions were used for phylogenetic analysis. Table 4 provides additional PCR product length information.

We ran MrBayes ver. 3.1.2 (RONQUIST & HUELSENBECK 2003) in two independent replicates, each with 1 cold chain and 3 chains of different temperature (standard setting). For the COI sequence block, the genetic code for metazoan mitochondrial DNA was used. All gene partitions were unlinked in shape, revmat, statefreq and pinvar. The analysis was run for 40 million generations (average standard deviation of split frequencies: 0.001395), sampling 40.000 trees. Negative log-likelihood score stabilisation was determined in a separate visualisation in Microsoft Excel 2003. Accordingly, we retained 39.990 trees after burn in (10.000 generations were discarded), from which a 50%-majority rule consensus tree with posterior probabilities was built ( Fig. 2 ). FigTree 1.3.1 (RAMBAUT et al. 2009) was used for graphical display of the tree. MEGA5 (Tamura et al. 2011) was used to calculate uncorrected p-distance values for all species involved in the dataset. We combined two datasets in one table, see Table 6: the lower left area shows the results of COI calculation only, with a total of 658 nucleotide positions in the COI dataset. The upper right area shows the results of the combined COI+16S+28S dataset, which were a total of 1528 nucleotide positions. All ambiguous positions and gaps were removed for each sequence pair to ensure the p-distance values are based on exactly the same data as the tree.

To determine whether the selection of slightly different ambiguous data blocks leads to a different tree topology we have also used the following software to determine the ambiguous positions to exclude: Aliscore (KÜCK et al. 2010) with options -r -n, Gblocks (CASTRESANA 2000; TALAVERA & CASTRESANA 2007) with standard settings and with three options activated for less stringent selection: allowing smaller final blocks, allowing gap positions within the final blocks, allowing less strict flanking positions. Gblocks standard setting was closest to our manual selection. The topology of the Dichromacalles tree has been preserved between all generated trees (data not shown), regardless of which ambiguous data has been excluded in Bayesian analysis. The reason is that COI has the highest differentiation power of all genes used, and did not have any areas to exclude. In our Dichromacalles dataset COI shows up to 23% of genetic difference between species of the same genus (Table 6).

The phylogenetic analysis shows that the subgenus Dichromacalles s.str. is paraphyletic. However, at the present point in time we do not want to describe new subgenera as long as definite molecular analyses of the missing species (see cataloque above) are not yet available. Referring to the International Rules of Zoological Nomenclature (IRZN) it seems appropriate to speak of informal groups for the present.

Figures

FIGURE 1. Phylogenetic hypothesis for the western Palaeartic genera of the subfamily Cryptorhynchinae Schoenherr, 1825, based on type species (gray shaded) and others. Bayesian 50 % majority rule consensus tree built from mitochondrial COI + 16 S and nuclear 28 S gene sequences. Numbers next to nodes indicate posterior probabilities in percent. Name of each species is followed by internal extraction number in brackets (in three cases complemented sequences derived from two individuals with different extraction numbers). See Table 1 for collecting data. Photos on the right side show some of the species listed in the tree.

FIGURE 2. Phylogenetic hypothesis for the Palaeartic species of the genus Dichromacalles Stüben, 1998. The clade of the new species D. algecirasensis is gray shaded. Bayesian 50 % majority rule consensus tree built from mitochondrial COI + 16 S and nuclear 28 S gene sequences. Numbers next to nodes indicate posterior probabilities in percent, country abbreviation and internal extraction numbers are given in brackets (CI: Canary Islands, E: Continental Spain, F: France, GR: Greece, I: Italy, Mo: Morocco, P: Portugal). See Table 2 for collecting data. Fotos on the right side show most of the species listed in the tree together with their host plants.