Computational protocol: Assessing the relationships between phylogenetic and functional singularities in sharks (Chondrichthyes)

Similar protocols

Protocol publication

[…] Four commonly available genes were collected to assess phylogenetic relationships between sharks. Three mitochondrial DNA sequences (Cytochrome‐b, hereafter cyt‐b; 12S and 16S) as well as one nuclear gene, that is, the Recombination‐Activating Gene 1 (hereafter RAG1) were obtained from GenBank (Benson et al., ). We are aware that GenBank sequences are individual specimen based, and prone to misidentification of species (see Naylor et al., ), but they were primarily chosen to combine both mitochondrial and nuclear genes in a multiloci approach, in order to avoid single‐locus analysis bias (McCormack, Hird, Zellmer, Carstens, & Brumflied, ; Nichols, ). Then, for each gene, sequences were aligned using MAFFT (Katoh, Misawa, Kuma, & Miyata, ) and phylogenies were calculated with ClustalW (Larkin et al., ). Then, a super Tree was computed based on the four trees previously assessed. Super Tree is a multiloci approach that allows to output a single tree from a set of different trees with overlapping taxa (Liu, Yu, & Pearl, ; see also Appendix for discussion on multiloci relevance for phylogenetic reconstruction). This technique transforms the topology of each tree into matrices, and combined and analyzed them with an optimization criterion, here Maximum Parsimony (Bininda‐Emonds, ; Bininda‐Emonds, Gittleman, & Steel, ). This final phylogenetic tree allowed to estimate phylogenetic relationships between 282 species representing the eight orders of sharks and 31 families (Figure  and Appendix ). [...] Different quantitative approaches were used to compare phylogeny and functions of sharks. These approaches were performed between phylogenetic data and functional data, this last being based on the original 13 functional traits database, and on the four subdatabases described before (NA‐excluded data, habitat traits, trophic traits, and behavioral traits). The first step was a comparison between the two distance matrices (pairwise cophenetic distances from the phylogenetic tree and pairwise Gower distances from functional traits) by a Mantel test (Mantel, ).The second step was the direct comparison of phylogenetic and functional trees' topologies. Two metrics of difference between trees were computed: the topological difference (Penny and Hendy, ), based on the number of branches that differ between trees, and which ranged from 0 to 2n−6, n being the number of species. As we dealt with different number of species because of the NA‐excluded database (282 as opposed to 86), the relative topological difference (RTD) was calculated as the proportion of topological difference such as: RTD=topological difference2n−6ranging from 0 (no difference) to 1 (completely different). The second metric was the branch length score (hereafter BLS, Kuhner & Felsenstein, ), which takes branch length into account (Steel and Penny, ). These two metrics were calculated on normalized trees, that is, with a total tree length equal to 1.The third step was to calculate the phylogenetic signal on functional traits taken as a whole in a measure of “functional identity.” The “functional identity” was estimated by a Brownian simulation. This simulation allows to give a quantitative state for each tip of a tree (here species in the functional tree). As a consequence, the functional identity may be defined here as the estimation of the species place in the functional tree (see Revell, , for further explanation). The phylogenetic signal of functional identity was calculated by the computation of both Moran's I and Abouheif's Cmean. These two statistics estimate the deviation from the Brownian model of evolution for traits and were recently advised to measure phylogenetic signal (Münkemüller et al., ). Moran I and Abouheif's Cmean take values comprised between −1 (no phylogenetic signal) and 1 (complete phylogenetic signal).The final step aimed to compare “species singularities.” The Evolutionary Distinctiveness index (ED, Isaac et al., ) was first calculated on the phylogenetic tree as a level of phylogenetic singularity (PS) for each species, and then calculated on the functional tree as species functional singularity (FS). This index is defined, for each branch, by its length divided by the number of species descendant from this branch. The singularity of a species is the sum of these values for all branches it descended from. To investigate the relationships between phylogenetic and functional singularities among species, a correlation of Pearson between species PS and species FS was computed. All analyses were conducted with packages “ape” (Paradis et al., ), “phytools” (Revell, ), “picante”(Kembel et al., ), and “vegan”(Oksanen et al., ) of the software R (R Core Team, ).In order to support our present work, a supplementary phylogenetic tree based on a single sequence (cyt‐b) but computed with bootstrap procedures was also confronted to the four previously described analytical steps, and results were consistent with those presented in this study (see Appendices and ). […]

Pipeline specifications

Software tools MAFFT, Clustal W, APE, Phytools, Picante
Application Phylogenetics