We constructed a genus-level phylogeny comprising 1,100 angiosperm genera found in lowland tropical South America, following protocols developed by Dexter & Chave¹. We used two chloroplast DNA gene regions: rbcL and matK. These genes were chosen based on their universality, data availability, typical sequence quality, degree of genus-level discrimination, sequencing costs and because they are recommended for standard DNA barcoding in plants^2-4. We generated 198 novel rbcL and 264 novel matK sequences from leaf fragments collected during extensive fieldwork across South America (all sequences generated in this study have an associated voucher; see Appendix 3 for a list of accession numbers and voucher information). Further sequences were also obtained through Genbank (http://www.ncbi.nlm.nih.gov/), and restricted to accessions that had an associated voucher. We had both rbcL and matK sequences for 808 genera (73%), only rbcL for 128 genera (12%) and only matK for 163 genera (15%). Sequences that were unavailable for a single region for a given genus were left as missing data. Exploratory sequence alignments and phylogenetic reconstructions enabled us to exclude sequences that were likely to represent taxonomic misidentifications. The details of DNA extraction, PCR, and DNA sequencing protocols can be found in Gonzales et al.⁵ A list of sampled genera, their respective family and GenBank accession numbers are available in Appendix 3.

We conducted multiple sequence alignments, separately for each region, using MAFFT v.6.822⁶, followed by manual adjustments in Mesquite (http://mesquiteproject.org). After manual alignments, we reduced remaining alignment issues by removing all sites which were missing data for >99% of genera. All rbcL and matK sequences were then combined to generate a starting maximum likelihood tree using RAxML v.7.2.7 in the CIPRES Science Gateway (https://www.phylo.org). A topological constraint specifying the major relationships among angiosperm orders was imposed based on the Angiosperm Phylogeny Group⁷. The early-branching angiosperm Nymphaea alba L. (Nymphaeaceae) was specified as an outgroup. This initial phylogeny was made ultrametric by using nonparametric rate smoothing method⁸ implemented in the ape package⁹ in the R statistical environment¹⁰, and then used as a starting tree in a Bayesian Markov Chain Monte Carlo (MCMC) approach to simultaneously estimate tree topology and divergence times of taxa¹¹. These analyses were performed using the BEAST software v.1.8.2 on the CIPRES server (https://www.phylo.org). An uncorrelated lognormal relaxed molecular clock was implemented, and the tree prior was a Birth-Death Incomplete Sampling model of speciation¹². We used 86 previously compiled fossil-based age constraints to calibrate node ages^13,14. Internal nodes were constrained using a log-normal distribution with a mean value equal to the fossil age, a standard deviation of 2 and a hard constraint for a minimum age equal to 80% of the estimated fossil age. No constraints were placed on the root age of the tree. We optimized operator settings before conducting the final runs by using a preliminary tree in test runs of 10⁶ generations.

We carried out three independent MCMC runs for 70.2 x 10⁵, 80.3 x 10⁵ and 58.6 x 10⁵ generations, under the same estimation conditions. We excluded burn-ins of 10³ and 2 x 10³ generations for the first two and third runs, respectively. We used LogCombiner to combine the three independent runs before sampling 282 trees evenly spaced across the posterior distribution, which were used to assemble a consensus tree. The consensus tree was assembled following the all compatible consensus rule; i.e., we choose the topology of a given node based on the relationship found in a plurality of trees from across the posterior distribution. This ensures a fully bifurcating topology that represents the most probable relationships of taxa. Lastly, we used TreeAnnotator (http://beast.bio.ed.ac.uk/treeannotator) to assign branch-lengths and divergence times (node heights) as the mean values from across the posterior distribution.

We also generated a species-level phylogeny, using the genus-level phylogeny as a basis. This consisted of imputing all 8,174 species in the floristics dataset by simulating a random birth-death phylogeny for each genus, using a speciation rate of 1 and an extinction rate of 0.9. We conducted these simulations in the TreeSim¹⁵ package in R¹⁰. This procedure produced a fully bifurcating phylogeny for each genus with the number of tips (i.e. species) corresponding to the number of species in the genus in our dataset. Importantly, we retained the stem age of the genus as estimated using our temporally calibrated phylogeny, while the crown age of genera was a product of the simulation (the mean expectation under a coalescent process is an age half that of the crown age, but there is variability around this expectation). The result of our approach is that the phylogenetic diversity represented by an individual species is proportional to the stem age of the genus divided by the species richness of the genus.

References

1. Dexter, K. G. & Chave, J. Evolutionary patterns of range size, abundance and species richness in Amazonian trees. PeerJ, 2043v1 (2016).

2. CBOL Plant Working Group. A DNA barcode for land plants. Proc. Natl. Acad. Sci. USA 106, 12794-12797 (2009).

3. Kress, W. J. & Erickson D. L. DNA barcodes: methods and protocols. Methods Mol. Biol. 858, 3-8 (2012).

4. Kress, W. J., Lopez, I. C. & Erickson, D. L. Generating plant DNA barcodes for trees in long-term forest dynamics plots. Methods Mol. Biol. 858, 441-458 (2012).

5. Gonzalez, M. A. et al. Identification of Amazonian trees with DNA barcodes. PLoS One 4, e7483 (2009).

6. Katoh, K., Misawa, K., Kuma, K. & Miyata, T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30, 3059-3066 (2002).

7. The Angiosperm Phylogeny Group. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG IV. Bot. J. Linn. Soc. 181, 1-20 (2016).

8. Sanderson M. J. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Mol. Biol. Evol. 19, 101-109 (2002).

9. Paradis, E., Claude, J. & Strimmer, K. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289-290 (2004).

10. R Core Team. R: a language and environment for statistical computing. Version 3.1.0. R Foundation for Statistical Computing, Vienna (2016). Available at: http://www.Rproject.org/

11. Drummond, A. J. & Rambaut, A. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, p. 214 (2007).

12. Stadler, T. On incomplete sampling under birth-death models and connections to the sampling-based coalescent. J. Theor. Biol. 261, 58-66 (2009).

13. Baker, T. R. et al. Fast demographic traits promote high diversification rates of Amazonian trees. Ecol. Lett. 17, 527-536 (2014).

14. Magallon, S., Gomez-Acevedo, S., Sanchez-Reyes, L. L. & Hernandez-Hernandez T. A metacalibrated time-tree documents the early rise of flowering plant phylogenetic diversity. New Phytol. 207, 437-453 (2015).

15. Stadler, T. TreeSim: Simulating Phylogenetic Trees. R package version 2.3 (2017). Available at: https://CRAN.R-project.org/package=TreeSim