Page 7 - Delicado_Machordom_2015

D. Delicado et al. Evolutionary patterns of Pseudamnicola

Fig. 2 Phylogenetic relationships of Pseudamnicola species based on Bayesian inference of the combined data set (top) and the mitochondrial
fragments (COI and 16S) and the nuclear 28S (bottom). In the combined inference, bootstrap supports of the branches resulted >90% and
BPPs > 0.9, except for those branches highlighted through: black circles, in which MP and ML bootstraps range between 50% and 90%
and BPPs between 0.5 and 0.9; and black squares, in which bootstraps also range between 50% and 90%, but posterior probabilities are
>0.9. In the individual trees of the bottom, asterisks represent bootstrap values and BBPs of branches > 90% and >0.9, respectively. Arrows
point to branches with no rate constancy (result from relative rate test).

P. (Pseudamnicola) species, with a ML bootstrap value and and the relaxed molecular clock approach applied. After
BPPs near 80% and 0.8, respectively. All the species were performing the analysis, the substitution rates inferred for
grouped as a polytomy within their corresponding clades each partition (substitution/Myr) were 0.3% for 16S and
(Fig. 2). 0.17% for 28S. Given that the posterior values of the
parameters ucld.stdev and coefficients of variation in all the
Combined data set. All the multilocus inferences recovered markers were greater than zero, evolution has not been
a total of 26 species of Pseudamnicola s. l., 20 of them previ- clocklike for these genes. Furthermore, 28S had higher
ously identified through integrative methods and six discov- deviation from a strict clock model as its coefficient of vari-
ered as new taxa. ABGD analysis confirmed the assignment ation was greater than 1.
of the sequences in these new six species; however, some of The topology of the chronogram with corresponding
the previously known species were split in two groups, confidence intervals analysed in *BEAST is shown in
increasing the number of species suggested by ABGD Fig. 3. Using an estimated rate of 0.81 0.24% substitu-
analysis to 36, eight additional species for the subgenus tion/Myr for the COI fragment, and rate estimations for
Corrosella and two for Pseudamnicola. Moreover, there were 16S and 28S, the most basal split leading to the three
obvious disparities between the nuclear and mitochondrial major clades within Pseudamnicola was calculated to have
gene trees related to the phylogenetic position of P. (P.) occurred ca. 22 Ma (HPD: 28–17 Ma), during the Upper
gasulli. Because of these disparities, the ILD test showed no Miocene. Contrary to the rest of the phylogenetic infer-
congruence between the mitochondrial and nuclear data. ences, the reconstruction by coalescence showed high pos-
Nevertheless, this does not mean than the resulting phy- terior probabilities at the level of these three lineages,
logeny was incongruent. As shown in Fig. 2, the phylogeny associating the species P. (P.) gasulli to the P. (Pseudamnico-
of the genus was well resolved at basal nodes; however, the la) lineage and dating their subsequent split ca. 17 Ma (23–
evolutionary relationships among the three lineages consti- 13 Ma). However, this result requires further investigation
tuting Pseudamnicola were ambiguous. as currently only one species composes the clade of P. (P.)
The biogeographic pattern reflected by the MP, ML and gasulli. Thus, this fact may be why the relative rate test
BI topologies was more explicit in the subgenus P. (Corro- showed significant differences on evolutionary rates when
sella) than in P. (Pseudamnicola). In contrast to P. (Corrosel- comparing the P. (P.) gasulli lineage with each of the other
la), P. (Pseudamnicola) species did not show an apparent two lineages, but showed no significant differences when
biogeographic pattern, for instance: (i) the species P. (P.) comparing P. (Corrosella) and P. (Pseudamnicola). Nonethe-
beckmanni (from Majorca) that was genetically closer to the less, the topology of each of the lineages displayed dissimi-
species occurring in Tunisia, Sicily and mainland Italy than larities in tempo and mode of species diversification.
to P. (P.) meloussensis (from Minorca), which is found in the The age of the most recent common ancestor of P. (Cor-
same archipelago of islands, (ii) the same species can be rosella) species was estimated to be older (ca. 13 Ma) than
found in different regions, such as P. (P.) sp. 5, in mainland the age of the P. (Pseudamnicola) species ancestor, excluding
Italy and Sardinia, and (iii) some species of P. (Pseudamnico- P. (P.) gasulli (ca. 6 Ma). Consequently, the first cladoge-
la) seem to live sympatrically in the same locality, such as netic events that occurred during P. (Corrosella) evolution
P. (P.) sp. 2 and P. (P.) sp. 4, which both inhabit the same were likely older than those for P. (Pseudamnicola). More-
locality (Borkane ditch) of Tunisia. over, these two lineages may have experienced a radiation
event near the base of each lineage, as the relative rate test
Divergence time estimation showed rapid diversification at basal levels of both clades.
The non-constancy of the substitution rate has been dem- In any case, the radiation event that occurred during the
onstrated as the relative rate test showed significant differ- origin of P. (Pseudamnicola) involved more species and a
ences in evolutionary rates between tree branches along the relatively shorter period of time (ca. 8–3 Ma), characteriz-
phylogeny (branches that showed significant differences in ing the major cladogenetic event for this group. The most
evolutionary rates are highlighted in Fig. 2). Thus, the recent splits between sibling species of P. (Pseudamnicola)
model of constant rate of diversification may be rejected were likely to have occurred from ca. 5 Ma to 0.08 Ma. All

ª 2015 Royal Swedish Academy of Sciences, 44, 4, July 2015, pp 403–417 409