Page 5 - Korn_alii_2006

M. Korn et al. • Sister species within Triops cancriformis

species pairs (calculated with MS Excel). MEGA version 2.1 clock enforced resulted in P = 0.254 > 0.05, i.e. it was not sig-
was used to illustrate parsimony-informative characters and nificantly different at the 5% level; Shimodaira & Hasegawa
singletons (Kumar et al. 2001). The program ForCon 1.0 1999), we assumed clock-like evolution of the sequences of
(Raes & Van de Peer 1998) was used to interconvert input T. cancriformis. Approximate times of diversification for selected
files between formats. To assess saturation effects in this data clades were calculated by converting pairwise genetic dis-
set, pairwise comparisons of transitional and transversional tances into units of time, following the divergence range of
changes were plotted against pairwise distances in DAMBE inferred crustacean molecular clocks for mtDNA (16S)
version 4.2.13 (Xia & Xie 2001; whereby the result that the sequence divergence published in Schubart et al. (2000):
data were not saturated was stable with all distance correction 0.65–0.88% per million years.
methods implemented).
Morphological re-analyses
Phylogenetic analyses For comparison with the sequence data, we investigated
To investigate relationships among the subspecies, several three key morphological characters that have been used to
data sets were used for calculations of phylogeny recon- discriminate among subspecies of T. cancriformis (Longhurst
struction. First, a 16S dataset comprising 107 T. cancriformis 1955; Table 3): (1) size of furcal spines located postero-laterally
sequences and two outgroup sequences from Lepidurus a. apus on the telson; (2) dorsal carina spines; and (3) number of
and T. longicaudatus (GenBank accession numbers in Table 2 apodous abdominal segments.
and Fig. 4) was analysed using maximum parsimony (MP;
settings gapmode = new; add = cl) as implemented in PAUP* Furcal spines. Furcal spines are part of the telson armature,
4.0b10 (Swofford 1998) and maximum likelihood (ML) using which comprises four sets of spines (see Longhurst 1955 for
PHYML (Guindon & Gascuel 2003; via the online Web inter- details). They are positioned around the bases of the furcal
face http://atgc.lirmm.fr/phyml/). As a measure of branch rami and are few and large in T. cancriformis (Longhurst
support, bootstrap values were calculated with MP and 1955). The two most prominent furcal spines are typically
neighbour-joining (ML-corrected distances) in PAUP* (settings situated dorso-laterally, to each side of the telson, followed
nreps = 1000, maxtree = 1000) and with PHYML (nreps = 500; ventrally by several smaller spines (in some specimens, a
presented in percent). The best evolutionary model for the smaller spine may also be positioned dorsally from one of
data was established by hierarchical likelihood testing, the prominent spines). For practical reasons, only the most
performed with the program ModelTest (Posada & Crandall prominent spines (one from each side of the telson) were taken
1998). A second 16S dataset consisting of the subset of ingroup into account and are referred to below as the furcal spines.
samples for which 12S sequence data were additionally We used the ratio of furcal spine length to the distance
available (30 T. cancriformis) was analysed with 10 outgroup between furcal spine tip and the anterior-lateral edge of the
sequences in the same manner. Similarly, the 12S sequences telson (henceforth called telson length ratio) to characterize
of this selection of samples, as well as the combined 16S and the size of the furcal spines. Because the spines show no
12S sequences, were analysed (also as described above) as clearly identifiable starting point at their base in most speci-
third and fourth datasets. mens, we used subsidiary lines to define the anterior starting
point. One subsidiary line was drawn from the foremost ante-
Timing of diversification events rior edge-point of one furcal ramus to the corresponding
To estimate whether rates of mtDNA molecular evolution edge-point of the other furcal ramus in dorsal view (see
are equivalent among the sequences of T. cancriformis (a con- Fig. 1). Further subsidiary lines were directed along the distal
dition necessary for dating cladogenetic events), we compared sides of each furcal ramus. The distance from the spine tip to
the 16S maximum-likelihood trees (excluding all outgroup the point where the subsidiary lines meet was defined as the
taxa and some very close taxa of T. cancriformis to reduce furcal spine length. Telson length ratio was measured for
computation time; neighbour-joining starting tree; in PAUP*) both sides of the telson to form a mean value, except for speci-
obtained without (option multrees in effect) and with (setting mens with a damaged spine on one side.
maxtree = 1) the assumption of a molecular clock. The latter Measurements were made on digital photographs of the
analysis was undertaken by enforcing the molecular clock telson (taken in dorsal view) using Scion Image for Windows
option in PAUP* (using a UPGMA starting tree rooted on a (Release Alpha 4.0.3.2 available at www.scioncorp.com).
sample of T. c. mauritanicus from southern Spain as outgroup). Subsidiary lines were drawn in Adobe Photoshop Elements 2.0.
We used the S–H test in PAUP* to compare alternative trees
(unrooted to enable comparison). As no significant difference Dorsal carina spines. In all specimens, carina spines were counted
was detected (the best tree was the first of the two obtained using a stereomicroscope at ×50 magnification. Very small
without the molecular clock enforced, and the one with the spines were included in the counts. Only bulges with increased

© 2006 The Authors. Journal compilation © 2006 The Norwegian Academy of Science and Letters • Zoologica Scripta, 35, 4, July 2006, pp301–322 305