Computational protocol: Highly divergent mussel lineages in isolated Indonesian marine lakes

Similar protocols

Protocol publication

[…] The bivalve origin of the obtained sequences was verified through BLAST searches ( From the blast results for the 28S and 18S sequences, the most similar sequences were selected and included in the subsequent phylogenetic analysis. Sequences were aligned and handled in CodonCode aligner v4.1.1 (CodonCode Corporation, USA) and DAMBE 5.2.15 (; ). The best-fit DNA substitution model was selected by the Akaike Information Criterion deployed in jMODELTEST v. 2.1.6 (). These models (COI: Hasegawa-Kishino-Yano (); 18S: Kimura 2-parameter (); 28S: Tamura-Nei (); all with a discrete gamma distribution) were used for subsequent maximum likelihood phylogeny inferences implemented in MEGA v6.06 () using a heuristic search with 1,000 bootstrap replicates. The resulting phylogenetic trees were visualized in FigTree v1.4.0 (). Separate maximum likelihood analyses were carried out for each marker. Net nucleotide distances between lineages were calculated in MEGA v6.06 using the models obtained with jMODELTEST. Net nucleotide divergence corrects for discrepancies between gene divergence and population divergence due to ancestral polymorphism in populations (), since it subtracts the average within-group divergence from the observed between-group estimate. Estimates of genetic variation in samples pooled per location were obtained as haplotype diversities h (), nucleotide diversities π (; ), and mean pairwise differences using Arlequin version v.3.5 (). In order to create a haplotype network, separate maximum likelihood tree analyses were carried out for each lineage (following the description above), and the resulting phylogenetic trees were used as input for Haploviewer ().Two methods were used to test for signatures of recent population expansion. First, Tajima’s D tests of selective neutrality () were carried out in Arlequin to compare the observed numbers of pairwise nucleotide differences between haplotypes in a sample with expectations under an infinite-sites model of sequence evolution, and under assumptions of selective neutrality and stable population size. Significance was tested by generating 10,000 random permutations. Second, mismatch distributions were calculated in Arlequin () and DnaSP () to test for signatures of demographic expansion and to test the null hypothesis of population growth. The observed distribution of pairwise differences between sequences was compared with a theoretical distribution, as expected under a sudden expansion model () computed in DnaSP (). To test the fit of the sudden-expansion-model, the sum of squared deviations (SSD) between the observed data and theoretical model was calculated in Arlequin. Harpending’s raggedness index (rg) was used to determine the smoothness of the observed mismatch distribution, which can be used to distinguish between expanded and stationary populations (). Expanding populations generate smooth and unimodal distributions, while more stationary populations the mismatch distribution becomes more ragged and erratic. The value of the raggedness index will be low and non-significant in expanding populations, while it is usually high and significant in stationary populations (; ). [...] Brachidontes spp. mussels were photographed in a standardized orientation for geometric morphometric analyses. In total 172 digital images were stored as Nikon RAW format (.nef) and converted to 3,008 × 2,000 pixel JPEG images using Photoshop 5.0 (Adobe). JPEG images were sampled into TPS files using the program tpsUTIL (). Shell outlines were used to capture variation in shell shape of Brachidontes spp. We used a sliding semi-landmark analysis, in which semi-landmarks are allowed to slide along the outline of a shell in order to find the position that optimally matches the positions of corresponding semi-landmarks in a consensus specimen (; ). Shell outlines were drawn as curves and digitized as 68 semi-landmarks at equal distance using tpsDig (), using the beak of the mussel (umbo, see ) as a standardized starting point for drawing an outline. A “sliders file” indicating sliding semi-landmarks was made using tpsUtil (). To standardize for size and orientation we used tpsRelw () with Generalized Procrustes Superimposition (). Residuals from the superimposition were analysed with the thin-plate spline interpolating function, producing principal warps, followed by relative warp (RW) analysis. The program TpsRelw was used to obtain centroid size () and RW scores for each individual. RW axes are analogous to the eigenvectors of principal component analysis, which combine the major patterns of shell shape variation in the data. Repeatability of RW axes was tested using regression analysis and a non-parametric analysis of similarity in PAST 2.11 () of RW scores extracted from 17 specimens of Brachidontes spp., which were independently photographed. RW axes were considered repeatable when they showed a non-significant and close to zero R-value in the analysis of similarity and a strong (r > 0.7) and significant correlation. Only repeatable relative warp axes were included in further analyses of shell shape variation. Correlations of RW scores with centroid size were tested, and if significantly correlated, we used residuals of the regression in analyses of shell shape. Significant differentiation between populations was tested using a non-parametric analysis of similarity (ANOSIM, 10,000 randomizations) () based on Euclidian distance as implemented in PAST v2.11 (). […]

Pipeline specifications