Table 4.
Clade | Character state | Posterior probability | log(harmonic mean of likelihoods) | BF test statistic 2 [logA-logB] |
---|---|---|---|---|
A | non-retention | 0.005 ± 0.016 | -31.9672 | 7.31 |
short-term retention | 0.891 ± 0.180 | -28.3106 | - | |
long-term retention | 0.103 ± 0.179 | -30.2700 | 3.92 | |
B | non-retention | 0.002 ± 0.011 | -35.4964 | 14.43 |
short-term retention | 0.526 ± 0.261 | -28.2813 | - | |
long-term retention | 0.472 ± 0.261 | -30.4022 | 4.24 |
Taxa were coded as having non-retention, short-term retention, or long-term retention of functional chloroplasts based on PAM data. The evolution of kleptoplasty was then modeled using Bayesian Inference via Markov-Chain Monte Carlo methods to estimate the posterior probability of alternative character states. Posterior probabilities (means ± one standard deviation) were calculated from an MCMC chain run for 20 million generations. Bayes factor (BF) tests were used to compare log-likelihood scores of models in which the ancestor was assigned one of the possible character states; values >2 represent positive evidence and values >5 represent strong evidence for models fixing the ancestral state as short-term retention.