## Similar protocols

## Protocol publication

[…] We performed a series of analysis to 1) statistically (PERMANOVA, PERMDISP) and visually (NMDS) explore the independent influence of biome and land use on patterns of taxonomic, functional and phylogenetic beta diversities (2x2 factorial design); 2) explore the individual ant and bird species contributing mostly to differentiate land uses within and between biomes (SIMPER); and 3) examine the associations among the different facets of diversity (simple and partial Mantel tests) and the existence of phylogenetic signal in trait data (D-Statistic).The three beta diversity facets were estimated using the appropriate distance measure between sites. To estimate TBD between land uses (cropfields and tree plantations) and biomes (Forest and Grassland) we first built an incidence matrix (sites x species) for each of birds and ants, in which species presence/absence was recorded for each study site; then we calculated the 1-Sorensen index as a measure of taxonomic dissimilarity. To estimate FBD and PBD for both birds and ants, we first built phenotypic dendrograms and phylogenetic ultrametric trees and then calculated the 1-Sorf and 1-PhyloSor indices using **PICANTE** []. Sorf and PhyloSor represent the proportion of branch lengths shared by two assemblages []; they are analogous to the Sorensen taxonomic similarity index, and consequently we minimized the potential lack of correlation between TBD, FBD and PBD due to differences in index construction.The phenotypic distance (euclidean) matrix (species x species), used to calculate the 1-Sorf index, was built using the selected species functional traits as variables (). All traits were defined as binary categorical variables, and multi-state traits (such as diet, habitat type, worker ant body-size, or ant functional group) were analyzed as multiple binary characters (0 = no, 1 = yes) []. Using phenotypic distances between species, we performed a hierarchical clustering procedure (UPGMA) in R [] to obtain the functional dendrogram. To calculate 1-Phylosor, phylogenetic ultrametric trees and phylogenetic distance matrices (species x species) were obtained after branch length adjustment in each constructed super-tree. Adjustments were conducted using bladj algorithm (**Phylocom**, []), which minimizes the variance between branch lengths within the constraints imposed by the dating of tree internal nodes. Node ages were obtained from available information for birds [] and ants [–] compiled on the Time Tree of Life website [].After estimating TBD, FBD and PBD we tested for significant associations between taxonomic, functional and phylogenetic distance matrices with biome and land use. To do that, we first visually explored site ordination based on taxonomic, functional and phylogenetic distances between assemblages in the Euclidean space, by performing non-metric multidimensional scaling (NMDS). First and second axes were plotted to evaluate whether site ordination was associated with biome or land use. Then, we performed permutational multivariate analysis of variance (PERMANOVA) [] that can be computed for any distance index and allows to test for interaction effects between factors (i.e. biome x land use). However, the interaction term may have significant effects on distances when simple effects are different in direction or magnitude; thus, we tested for significant simple effects and for homogeneity of multivariate dispersion using PERMDISP []. Significance was obtained for each test by 9999 Monte Carlo permutations. Finally, the Similarity Percentage analysis (SIMPER) [] was performed to identify species that contributed mostly to the taxonomic 1-Sor distances between treatments. SIMPER performs pairwise comparisons of groups of sampling units (treatments; i.e. cropfields in Grassland) and ranks all species according to the average contribution of each one to the overall average distance index. Using presence data, species that occur in most sites within treatments are those that contribute the most to the similarity within and dissimilarity between treatments. Thus, SIMPER allows identifying bird and ant species that better discriminate between treatments []. NMDS and SIMPER were implemented in R [].To analyze the association between the three beta diversity facets, we performed correlations between pairs of distances matrices (i.e., simple Mantel tests) [] to test the association between each pair of taxonomic (1-Sor), functional (1-Sorf) and phylogenetic (1-PhyloSor) beta diversities. Then, we performed partial correlations (i.e., partial Mantel tests) between pairs of distance matrices [] to remove the effects of the third distance matrix. Tests were performed using the VEGAN package [] applicable in R [], and 1000 permutations of the distance matrices to obtain the significance level.A high correlation between 1-Sorf and 1-Phylosor is a strong indicator of phylogenetic signal in trait data []. Thus, the phylogenetic conservation of species traits was explored by testing the degree of phylogenetic signal of each bird and ant species trait using D-statistic for binary traits []. Starting from trait values randomly distributed along a phylogenetic tree (D ~ 1), the D-statistic approaches zero as trait phylogenetic signal increases. When traits are more conserved than expected by the Brownian evolutionary model (i.e. trait values differ proportionally to species divergence times []), then the D-statistic is significantly less than zero. Observed and expected distributions of the D-statistic, and significance level for each test were obtained using the CAPER package [] applicable in R []. […]

## Pipeline specifications

Software tools | Picante, Phylocom |
---|---|

Application | Phylogenetics |

Diseases | Zellweger Syndrome |