R: Fit Rate-Constant Birth-Death Model to Branching Times
Fit Rate-Constant Birth-Death Model to Branching Times
Finds maximum likelihood estimates of the net diversification rate r
(speciation rate S minus the extinction rate E) and the extinction
fraction a = E/S, using branching times derived from an
ultrametric phylogenetic tree.
bd(x, ai = c(0.1, 0.5, 0.9))
a numeric vector of branching times
a vector of initial a parameterizations for the optimization algorithm
Non-linear optimization can be exceedingly difficult, and the algorithms used here can become trapped
on local (rather than global) optima. The default 'ai' parameters specified above fit the constant-rate
birth-death model to branching times using three initial a values. You should check your results
against those obtained using the pureBirth model. If the log-likelihood under bd is
less than pureBirth, you should explore alternative initial parameterizations. For example,
ai = seq(0.05, 0.99, length.out = 20) would attempt the optimization with 20 equally spaced a
values on the interval (0.05, 0.99).
I have found the default option to be satisfactory for all phylogenies I have examined.
a list with the following components:
the log-likelihood at the maximum
the Akaike Information Criterion
the net diversification rate giving the maximum log-likelihood
the extinction fraction giving the maximum log-likelihood