Content area
Full Text
adjustdirection” option, which checked for correct sequence polarity. The concatenated 26,993 base pair sequence matrix and our trees are available as a Nexus file at the online UC Berkeley DASH repository [61]. We performed maximum likelihood phylogenetic analyses using RAxML 8.2.9 [62] on the CIPRES computing infrastructure [63]. We applied the general time-reversible (GTR)?+?CAT model [64] of nucleotide substitution with 25 rate categories independently to each of the nine gene partitions, performing tree searches with a randomized stepwise addition order maximum parsimony starting tree and 1000 rapid bootstrap replicates [65]. The analyses utilized a constraint tree that enforced the following high-level relationships [66]: Isoëtales and Selaginellales were made a monophyletic clade sister to the Lycopodiales, the Lycopodiophyta were set to be monophyletic and sister to the Pteridophyta plus spermatophytes, and the Pteridophyta and spermatophytes were each constrained to be monophyletic and sister to one another. We replicated the RAxML analysis using 10 different random number seed values and compared the resulting 10 maximum likelihood trees by their log-likelihood values and Robinson-Foulds distances [67].
Molecular dating
We produced time-calibrated chronograms using penalized likelihood [68] as implemented in r8s [69] using the truncated Newton method and a smoothing parameter of 1000. Fossil calibrations were taken from Magallón et al. [70, 71], and two different calibration schemes were utilized: the first used only a root age for all vascular plants, and the second used 55 calibrations scattered across the whole phylogeny (a list of node calibrations is provided in Additional file 14: Table S4). Using the latter calibration, the stem age of our OTUs (length of the terminal branch) ranged from 0.2 to 382 my (million years) (median?=?27 my).
Modeled ranges using macroclimatic variables
The species distribution modeling algorithm Maxent [72] was used to model the range of each Californian species, using the cleaned species-level spatial dataset from Baldwin et al. [11], deposited in the UC Berkeley DASH repository [73]. Models were fit at the species level since gene flow barriers among species mean that climatic niches are relatively cohesive within species but not within OTUs; each OTU’s range was then taken to be the union of the ranges of its included species. Models were fit using four predictor variables representing major energy- and water-related variables known to...