The function scans a phylogenetic tree looking for morphological convergence between entire clades or species evolving under specific states.
search.conv(RR=NULL,tree=NULL,y,nodes=NULL,state=NULL,aceV=NULL,
min.dim=NULL,max.dim=NULL,min.dist=NULL,declust=FALSE,nsim=1000,rsim=1000,
clus=0.5,filename=NULL)If convergence between clades is tested, the function returns a list including:
$node pairs comparison: pairwise comparison between significantly convergent pairs (all pairs if no instance of significance was found) performed on the distance from group centroids (the mean phenotype per clade).
$average distance from group centroids: smaller average distances mean less variable phenotypes within the pair.
If convergence between (or within a single state) states is tested, the function returns a list including:
state.res a dataframe including for each pair of states (or single state):
ang.state: the mean theta angle between species belonging to different states (or within a single state).
ang.state.time: the mean of theta angle between species belonging to different states (or within a single state) divided by time distance.
p.ang.state: the p-value computed for ang.state.
p.ang.state.time: the p-value computed for ang.state.time.
plotData a dataframe including
data to plot the results via plotConv
.
an object produced by RRphylo. This is not indicated
if convergence among states is tested.
a phylogenetic tree. The tree needs not to be ultrametric or fully dichotomous. This is not indicated if convergence among clades is tested.
a multivariate phenotype. The object y should be either a
matrix or dataframe with species names as rownames.
node pair to be tested. If unspecified, the function
automatically searches for convergence among clades. Notice the node number
must refer to the dichotomic version of the original tree, as produced by
RRphylo.
the named vector of tip states. The function tests for convergence within a single state or among different states (this latter case is especially meant to test for iterative evolution as for example the appearance of repeated morphotypes into different clades). In both cases, the state for non-focal species (i.e. not belonging to any convergent group) must be indicated as "nostate".
phenotypic values at internal nodes. The object aceV
should be either a matrix or dataframe with nodes (referred to the
dichotomic version of the original tree, as produced by RRphylo) as
rownames. If aceV are not indicated, ancestral phenotypes are
estimated via RRphylo.
the minimum size of the clades to be compared. When
nodes is indicated, it is the minimum size of the smallest clades in
nodes, otherwise it is set at one tenth of the tree size.
the maximum size of the clades to be compared. When
nodes is indicated, it is min.dim*2 if the largest clade in
nodes is smaller than this value, otherwise it corresponds to the
size of the largest clade. Without nodes it is set at one third of
the tree size.
the minimum distance between the clades to be compared. When
nodes is indicated, it is the distance between the pair. Under the
automatic mode, the user can choose whether time distance or node distance
(i.e. the number of nodes intervening between the pair) should be used. If
time distance has to be considered, min.dist should be a character
argument containing the word "time" and then the actual time distance to be
used. The same is true for node distance, but the word "node" must precede
the node distance to be used. For example, if the user want to test only
clades more distant than 10 time units, the argument should be "time10". If
clades separated by more than 8 nodes has to be tested, the argument
min.dist should be "node8". If left unspecified, it automatically
searches for convergence between clades separated by a number of nodes
bigger than one tenth of the tree size.
if species under a given state (or a pair of states) to be
tested for convergence are phylogenetically closer than expected by chance,
trait similarity might depend on proximity rather than true convergence. In
this case, by setting declust = TRUE, tips under the focal state (or
states) are removed randomly until clustering disappears. A minimum of 3
species per state is enforced to remain anyway.
number of simulations to perform sampling within the theta random distribution. It is set at 1000 by default.
number of simulations to be performed to produce the random distribution of theta values. It is set at 1000 by default.
the proportion of clusters to be used in parallel computing. To
run the single-threaded version of search.conv set clus = 0.
is deprecated. search.conv does not return plots
anymore, check the function plotConv instead.
Silvia Castiglione, Carmela Serio, Pasquale Raia, Alessandro Mondanaro, Marina Melchionna, Mirko Di Febbraro, Antonio Profico, Francesco Carotenuto, Paolo Piras, Davide Tamagnini
Castiglione, S., Serio, C., Tamagnini, D., Melchionna, M., Mondanaro, A., Di Febbraro, M., Profico, A., Piras, P.,Barattolo, F., & Raia, P. (2019). A new, fast method to search for morphological convergence with shape data. PLoS ONE, 14, e0226949. https://doi.org/10.1371/journal.pone.0226949
if (FALSE) {
data("DataFelids")
DataFelids$PCscoresfel->PCscoresfel
DataFelids$treefel->treefel
DataFelids$statefel->statefel
cc<- 2/parallel::detectCores()
RRphylo(treefel,PCscoresfel,clus=cc)->RRfel
## Case 1. searching convergence between clades
# by setting min.dist as node distance
search.conv(RR=RRfel, y=PCscoresfel, min.dim=5, min.dist="node9",clus=cc)
# by setting min.dist as time distance
search.conv(RR=RRfel, y=PCscoresfel, min.dim=5, min.dist="time38",clus=cc)
## Case 2. searching convergence within a single state
search.conv(tree=treefel, y=PCscoresfel, state=statefel,declust=TRUE,clus=cc)
}
Run the code above in your browser using DataLab