runMedusa {geiger} | R Documentation |
Fits a series of diversification models to phylogenetic and taxonomic data using stepwise AIC
runMedusa(phy, richness, estimateExtinction=T, modelLimit=20, cutAtStem=T, startR = 0.05, startE = 0.5, ...) summaryMedusa(phy, richness, out, cutoff=4, plotTree=T, useCorrection=F, cutAtStem=T)
phy |
Phylogenetic tree in phylo format |
richness |
A table with taxon names in column 1 and extant diversities in column 2 |
estimateExtinction |
If false, fits a series of pure-birth models; otherwise birth-death models are used |
modelLimit |
Maximum number of rate shifts to place on the tree |
cutAtStem |
When rate shifts are placed on branches in Medusa, they could be placed at either the beginning or the end of the stem branch. cutAtStem=T cuts at the beginning of the stem branch, while cutAtStem=F cuts at the end. |
... |
Additional arguments for nlm |
out |
Output from medusa, sent to summaryMedusa to compile results |
cutoff |
Cutoff value for differences in AIC scores when comparing models. More complex models with an AIC score more than this number of units lower than simpler models are retained. |
plotTree |
Plot phylogenetic tree with branches under different models shown in different colors. |
useCorrection |
If true, use AICc instead of AIC |
startR |
Starting guess for net diversification rate. Try tweaking if Medusa crashes |
startE |
Starting guess for relative extinction rate. Try tweaking if Medusa crashes |
MEDUSA fits a series of diversification models to a combination of phylogenetic and taxonomic data. The input is a phylogenetic tree with branch lengths proportional to time showing the relationship among clades, and the diversity for living species in all of those clades. All taxa missing from the tree have to be assigned to one of the tip clades in the richness matrix.
The algorithm first fits a Single diversification model to the entire dataset. Then, a series of breaks are added, so that different parts of the tree evolve with different parameter values (birth and or death rates). MEDUSA first compares all single-breakpoint models with the overall model, and selects the best one. Then all possible two-breakpoint models are compared with the best single-breakpoint model, and so on.
summaryMedusa summarizes the results of MEDUSA. Here one must choose a cutoff value for improvement in AIC score necessary to accept more complex models.
From medusa: a matrix where each row summarizes the next more complex model selected by the stepwise
procedure. Columns record: the node where the next break was found, the likelihood, the number of
parameters, the AIC score, and the AICc score (corrected for small sample size).
From summaryMedusa: a list summarizing the model selected using AIC (or AICc) scores and the cutoff
value. Each item in the list gives the likelihood for that part of the tree, two parameter estimates,
r=b-d and epsilon=d/b, and the tip taxa included in that part of the tree.
Luke J. Harmon and Daniel Rabosky
Alfaro, M. E., F. Santini, C. Brock, H. Alamillo, A. Dornburg, D. L. Rabosky, G. Carnevale, and L. J. Harmon. 2009. Nine exceptional radiations plus high turnover explain species diversity in jawed vertebrates. Proceedings of the National Academy of Sciences 106:13410-13414.
data(geospiza) attach(geospiza)