Page 7 - Colomba-et-al-2011-Erctella-mazzullii-complex-Zootaxa-3134

TABLE 3. Primer list. Sequence Amplicon size (bp)
5’-GACTGTGCAAAGGTAGCATA-3’ 211
Primer name 5’-GCTGTTATCCCTAGAGTAAC-3’
MS_16S_F 5’-AAATATTTAGGGGAACTTAC-3’ 189
MS_16S_R 5’-TATTTATTACTTTTAAGTCC-3’
MS_12S_F 5’-CATTGAACATCGACATCTTGA-3’ 535
MS_12S_R 5’-CTCCGCTTAGTAATATGCTTAA-3’
ITS-2_F
ITS-2_R

Sequencing of the purified PCR products was carried out using automated DNA sequencers at Eurofins MWG
Operon (Germany). GenBank accession numbers (GQ402350–GQ402460) for all sequences generated in this
study are listed in Table 4.

Phylogenetic analyses. Sequences were visualized with BioEdit Sequence Alignment Editor 7 (Hall 1999),
aligned with the ClustalW option included in this software and double checked by eye. Standard measures of nucle-
otide polymorphism [haplotype/genotype diversity (Hd), mean pairwise differences (k), nucleotide diversity within
groups (π = Pi and πjc Pi corrected according to Jukes and Cantor) and nucleotide divergence (Dxy) between groups]
using the full set of all sequences were computed with DnaSP 4 (Rozas et al. 2003). Phylogenetic analyses were
conducted in MEGA 4 (Tamura et al. 2007), BEAST 1.4.8 (Drummond & Rambaut 2007), and PhyML 3.0 (Guin-
don & Gascuel 2003; available at http://www.phylogeny.fr/version2_cgi/one_task.cgi?task_type=phyml, see
Dereeper et al. 2008). To address the phylogenetic relationships among taxa, many different analytical methods
were used: Maximum Likelihood (ML), Neighbour Joining (NJ), Maximum Parsimony (MP) and Bayesian Infer-
ence (BI). For Maximum Likelihood analyses, best-fit models of molecular evolution were chosen using a hierar-
chical likelihood ratio test by Findmodel (available at http://www.hiv.lanl.gov/cgi-bin/findmodel/findmodel.cgi), a
web implementation of Modeltest by Posada & Crandall (1998). Models selected were: GTR+Γ (16S rDNA and
concatenation); HKY+ Γ (12S rDNA and ITS-2). Neighbour Joining trees were constructed using TN93 (16S
rDNA) and Tamura 3-parameter (12S rDNA and ITS-2) model distances; gaps were treated as missing data. MP
trees were obtained using the Close-Neighbour-Interchange (CNI) algorithm with search level 3 in which the initial
trees were achieved with the random addition of sequences (10 replicates). All positions containing gaps and miss-
ing data were eliminated from the dataset (complete deletion option). Characters were assigned equal weights. MP
trees were collapsed to obtain a 50% majority rule consensus tree. Support for the internodes was assessed by boot-
strap percentages (BP) (1,000 resampling steps for NJ and MP; 100 replicates for ML). For Bayesian analyses four
Markov Chain Monte Carlo (MCMC) chains were run for 1,000,000 generations, sampling every 100 generations;
from the 10,000 trees found the first 1,000 were discarded as “burn-in”. Finally, a 50% majority rule consensus tree
was constructed. We performed NJ, MP, ML and BI analyses on each gene segment; in addition genes were con-
catenated and analysed by partitioned analyses which allowed each gene to have its own parameters. To test incon-
gruence among genes a partition homogeneity test (Farris et al. 1994) was conducted in PAUP*4.0b10 (Swofford
2002). The test (100 replicates of random addition heuristic search option with tree-bisection reconnection branch
swapping) indicated significant heterogeneity among genes (p = 0.01). However, since a growing number of stud-
ies (Yoder et al. 2001) indicate that incongruence tests are not reliable indicators of dataset combinability and no
strong supported nodes were in conflict, genes were concatenated into a multigene dataset of about 950 bp. Analy-
ses were conducted in BEAST v1.6.1 by *BEAST (Heled & Drummond 2010) following the same steps described
in the molecular dating section (see below). All phylogenetic trees were rooted using homologous nucleotide
sequences of Cornu aspersum and Cantareus apertus (three specimens each).

Molecular dating. Molecular dating was conducted by a Bayesian MCMC approach in the program BEAST
(v 1.4.8), where the topology and divergence times can be estimated simultaneously from the data and therefore a
starting tree topology is not required, making it particularly appropriate for groups with uncertain phylogenies.
BEAST input files were generated with BEAUTi (v 1.4.8). The 16S rDNA dataset was employed and a GTR+Γ
was used to describe the substitution model, a Yule model was used to describe speciation. The likelihood ratio sta-
tistics didn’t allow to reject the molecular clock hypothesis (p = 0.9987) and, therefore, a strict molecular clock
model was used. BEAST was run for 1,000,000 generations with samples taken every 100 generations. Five inde-

FROM HELIX MAZZULLII COMPLEX TO ERCTELLA Zootaxa 3134 © 2011 Magnolia Press · 7