Align or not to align? Resolving species complexes within theCaloplaca saxicolagroup as a case study

Abstract

The Caloplaca saxicola group is the main group of saxicolous, lobed-effigurate species within genus Caloplaca (Teloschistaceae, lichen-forming Ascomycota). A recent monographic revision by the first author detected a wide range of morphological variation. To confront the phenotypically based circumscription of these taxa and to resolve their relationships morphological and ITS rDNA data were obtained for 56 individuals representing eight Caloplaca species belonging to the C. saxicola group. We tested the monophyly of these eight morphospecies by performing maximum parsimony, maximum likelihood and two different types of Bayesian analyses (with and without a priori alignments). Restricting phylogenetic analyses to unambiguously aligned portions of ITS was sufficient to resolve, with high bootstrap support, five of the eight previously recognized species within the C. saxicola group. However, phylogenetic resolution of all or most of the eight species currently included as two distinct subgroups within the C. saxicola group was possible only by combining morphological characters and signal from ambiguously aligned regions with the unambiguously aligned ITS sites or when the entire ITS1 and 2 regions were not aligned a priori and included as an integral component of a Bayesian analysis (BAli-Phy). The C. arnoldii subgroup includes C. arnoldii, comprising four subspecies, and the C. saxicola subgroup encompasses seven species. Contrary to the C. saxicola subgroup, monophyly of taxa included within the C. arnoldii subgroup and their relationships could not be resolved with combined ITS and morphological data. Unequivocal morphological synapomorphies for all species except C. arnoldii and C. pusilla are recognized and presented.

First report of the pantropical species Diploschistes rampoddensis from Europe

“…The Bayesian analyses (B) were done using BAli-Phy 2.1.1 (Suchard & Redelings 2006), a software that simultaneously accounts for alignment uncertainty and phylogeny, allowing the inclusion of the whole alignment. In this case, the nrITS dataset was divided into three partitions (ITS1, 5.8S, and ITS2) as suggested in Gaya et al (2011). The analyses were run under the following settings: ITS1 and ITS2 sharing the same model of evolution (HKY+G), indel model (RS07; Redelings & Suchard 2007) and branch length estimation, while the EQU model and separate branch length estimation were applied to the partition 5.8S, which was fixed because the alignment was unambiguous.…”

Photobiont selectivity and specificity in Caloplaca species in a fog-induced community in the Atacama Desert, northern Chile

“…For checking the eventual bias that could be operating in the generation of the alignments, we made a simultaneous Bayesian estimation of alignment and phylogeny using the package BAli-Phy (Redelings & Suchard 2005) for the unaligned nrDNA ITS region matrix of both algae and fungi following Gaya et al (2011). Four independent runs of 100 000 generations each were made for both symbionts matrices, sampling the chains every ten generations.…”

Scaling statistical multiple sequence alignment to large datasets

“…Even smaller datasets can be computationally intensive (for example, a BAli-Phy analysis of a dataset with 68 sequences took about 21 CPU days [13]), and the largest dataset that BAli-Phy has analyzed may be the 117-sequence dataset studied in [14]. However, BAli-Phy may not be able to run on substantially larger datasets than this; indeed, our initial testing found that with 500 sequences, BAli-Phy failed at an early step on every run (and practical constraints tend to limit its use below even that).…”