Page 8 - Delicado_Machordom_2015

Evolutionary patterns of Pseudamnicola D. Delicado et al.

Fig. 3 Ultrametric tree obtained with *BEAST based on the combined analysis of COI (using a rate previously calibrated for hydrobiids)
and the ribosomal fragments 16S and 28S (using rates estimated in this analysis). At nodes where *BEAST posterior probabilities (BPP) are
not given, the node is supported by BPP ≥0.90; black dots indicate BPP <0.90. Bars at nodes represent confidence intervals of divergence
times. Letters from A to F at nodes correspond to the most probable ancestral area states (pie charts with the distribution of the probability
of the ancestral area states are depicted in Fig. S1) and at tips correspond to the current distribution of the species (extracted from Abell
et al. 2008). Ma, million years ago.

of these splits appeared not well supported. Rate constancy whereas in DIVALIKE (AIC = 81.42) and BayAreaLIKE
was rejected by the relative rate test in most of the compar- (AIC = 81.39), model supports were slightly lower. Esti-
isons between branches within the P. (Pseudamnicola) line- mated parameters in DEC +J resulted d = 0.0014, e = 0
age, which may affect the split frequencies (probably and j = 0.0477. The most likely ancestral area inference
accelerated) and thus justify the low BPP values at the according to this model is given in Fig. 3. This analysis
affected nodes. suggests the Iberian Peninsula as the geographic area in
which the cladogenetic event that originated the three
Ancestral area estimation main Pseudamnicola lineages occurred. Speciation by vicari-
According to a log-likelihood ratio test, all the models ance seems to be the most common evolutionary process
tested in BioGeoBEARS which assume founder event spe- within P. (Corrosella) lineage. On the contrary and
ciation (+J) displayed significantly higher likelihood values acknowledging the lack of a complete taxon sampling for
than those without (all P < 0.05). According to model P. (Pseudamnicola) species, dispersal along Mediterranean
selection criteria AIC, no significant differences were islands and peninsulas and founder event may have been
found among the three +J models. However, the DEC +J the dominant processes in the evolutionary history of this
model obtained the highest model fit (AIC = 81.01), group.

410 ª 2015 Royal Swedish Academy of Sciences, 44, 4, July 2015, pp 403–417