Page 9 - Delicado_Machordom_2015

D. Delicado et al. Evolutionary patterns of Pseudamnicola

Table 1 Mantel test parameters of correlations between the genetic phylogenetic patterns (see Fig. 2). In P. (Pseudamnicola),
distance matrix of each gene fragment and distance matrices of the splitting events appear more recent and with less-supported
abiotic variables of water conductivity, altitude and geographic dis- branches, whereas in P. (Corrosella), the branches are longer
tances between two localities. The value n represents the number and more robust, which is a possible sign of a more older
of localities included in each correlation, r is the correlation coeffi-
cient and P is the statistical significance and gradual speciation process within this group. One rea-
sonable explanation for the different topologies may be
because P. (Corrosella) species present more restricted dis-
Genetic distances Abiotic variable n r P
tribution ranges and inhabit springs and headwaters of
COI Conductivity 52 0.06 0.001
streams, which often act as isolated habitats (Wilke et al.
Altitude 88 0.32 0.001
Geographic distance 88 0.44 0.001 2010). Thus, these isolated locations may constrain gene
16S Conductivity 52 0.25 0.001 flow between populations (Br€ andle et al. 2005) and increase
Altitude 88 0.38 0.001 the degree of endemicity. In contrast, P. (Pseudamnicola)
Geographic distance 88 0.44 0.001 species and P. (P.) gasulli are euryhaline species and occur
28S Conductivity 52 0.45 0.001 in coastal streams, lakes and low river stages where the
Altitude 88 0.20 0.001
Geographic distance 88 0.39 0.001 ecological conditions are less restrictive and the waters
remain connected. Moreover, such locations are more
exposed to the presence of birds and fishes than springs
Exploring causes of diversification (Haase 2008), which may favour jump dispersal via vectors.
Mantel tests performed separately for each subgenus (Deli- In any case, these two habitat prototypes are likely associ-
cado et al. 2013, 2014) revealed no correlation between the ated with two different dispersal abilities, directly influenc-
genetic distance matrix and physical variables, such as con- ing their phylogenetic topologies.
ductivity and altitude, but a pattern of isolation by distance Despite barcoding-gap method confirmed the assignment
was found. However, when both subgenera were included of the six new species obtained by our multilocus phyloge-
in the analysis, Mantel tests showed significant correlation netic analysis, the total number of species obtained by these
with the three examined variables, namely conductivity, two approaches differs. This testifies the need of combin-
altitude and geographic distance (Table 1). Despite this, ing, at least in hydrobiids, the information yielded by COI
conductivity and altitude only had minor influences, com- sequences with multiloci analyses, morphological descrip-
pared to geographic distance, on the divergence of the tions, biogeography or ecological data (as recommended in
subgenera for the COI and 16S genes, while conductivity Puillandre et al. (2012) or Collins & Cruickshank (2013)).
had more influence for the 28S gene, followed by geo- Nevertheless, here we benefit from the information of the
graphic distance. COI fragment to objectively compare genetic divergences
between Pseudamnicola lineages and between this group and
Discussion other microgastropods. Thereby, the average pairwise
Effects of habitat transition on the evolutionary history of divergence in the COI partition between species (described
Pseudamnicola s. l through integrative taxonomy) is 1.5 times greater in
Phylogenetic patterns and distribution ranges. The applica- P. (Corrosella) than in P. (Pseudamnicola). Sequence differ-
tion of molecular tools in the systematic analysis of Pseu- ences (measured as uncorrected pairwise distances) between
damnicola has revealed the existence of three main lineages spring snails species of P. (Corrosella) ranged between 5.3%
within the genus, corresponding to the two subgenera pre- and 12% (with an average of 9%), which is similar to
viously described plus the species P. (P.) gasulli. From this ranges described for other springsnail genera, such as
molecular study, we conclude that the observed morpho- Bythinella Moquin-Tandon, 1856 (1.5–13.4% in Bichain
logical differences existing between the two subgenera (dis- et al. 2007), and Floridobia Thompson & Hershler, 2002,
cussed in Delicado et al. 2012) have a phylogenetic signal Marstonia Baker, 1926 and Pyrgulopsis Call & Pilsbry, 1883
and moreover that the anatomical differences recorded for (0.5–6.1%, 1.0–8.5% and 2.8–11.2%, respectively, in
P. (P.) gasulli in Boeters (1988) and Delicado et al. (2014) Hershler et al. 2003). Alternatively, genetic divergences for
reflect a different origin of this species with respect to P. (Pseudamnicola) species are an average of 6.7% (ranging
other P. (Pseudamnicola) species. The combined mitochon- between 0.5% and 10%), which falls between the estimated
drial and nuclear phylogeny reasonably supports each of 9% for P. (Corrosella) and 4.5% for the brackish genus
the clades; however, the relationships among them still Hydrobia (Wilke et al. 2000). To a limited extent, this gra-
remain unclear. dient of genetic divergences may be due to the type of
Although well supported as monophyletic groups, environment (freshwater vs. brackish) occupied by these
P. (Pseudamnicola) and P. (Corrosella) display different three groups.

ª 2015 Royal Swedish Academy of Sciences, 44, 4, July 2015, pp 403–417 411