Panicum sensu stricto is a genus of grasses (Poaceae) with nearly, according to this study, 163 species distributed worldwide. This genus is included in the subtribe Panicinae together with Louisiella, the latter with 2 species. Panicum and subtribe Panicinae are characterized by including annual or perennial taxa with open and lax panicles, and spikelets with the lower glume reduced; all taxa also share a basic chromosome number of x = 9 and a Kranz leaf blade anatomy typical of the NAD-me subtype photosynthetic pathway. Nevertheless, the phylogenetic placements of many Panicum species, and the circumscription of the genus, remained untested. Therefore, phylogenetic analyses were conducted using sequence data from the ndhF plastid region, in an extensive worldwide sampling of Panicum and related genera, in order to infer evolutionary relationships and to provide a phylogenetic framework to review the classification of the genus. Diversification times, historical biogeography and evolutionary patterns of the life history (annual vs. perennial) in the subtribe and Panicum were also studied. Results obtained provide strong support for a monophyletic Panicum including 71 species and 7 sections, of which sections Arthragrostis and Yakirra are new in the genus; 7 new combinations are made here. Furthermore, 32 species traditionally assigned to Panicum were excluded from the genus, and discussed in other subtribes of Paniceae. Our study suggested that early diversification in subtribe Panicinae and Panicum occurred through the Early-Mid Miocene in the Neotropics, while the subsequent diversification of its sections mainly occurred in the Late Miocene-Pleistocene, involving multiple dispersals to all continents. Our analyses also showed that transition rates and changes between annual and perennial life history in Panicum were quite frequent, suggesting considerable lability of this trait. Changes of the life history, together with C4 photosynthesis, and the multiple dispersal events since the Mid Miocene, seem to have facilitated a widespread distribution of the genus. All these findings contribute to a better understanding of the systematics and evolution of Panicum.
Panicinae_ndhF_alignment
ndhF alignment for subfamily Panicoideae (Poaceae) including members of subtribe Panicinae
Panicinae(RAxML)_bestTree
Maximum likelihood tree for subtribe Panicinae generated with RAxML 8.2.4 using the ndhF alignment under the GTRCAT model across codon positions.
Panicinae(BEAST)CAL1
xml file for phylogenetic analyses and divergence time estimation in BEAST 1.8.4 using the nucleotide substitution models unlinked across codon position, a birth-death model with incomplete sampling as tree prior, an uncorrelated lognormal clock model for the rate variation among tree branches, and a calibration based only on external angiosperm fossils (eudicots and non-grass monocots) (calibration scheme 1). We used secondary calibrations in normal prior distributions for the following six crown nodes: subfamily Panicoideae (scheme 1: mean= 38.18 Mya, SD= 3.86), most recent common ancestor (MRCA) of supertribes Andropogonodae –Panicodae (mean= 30.31 Mya, SD= 3.27), supertribe Andropogonodae (mean= 28.5 Mya, SD= 3.33), tribe Andropogoneae (mean= 11.79 Mya, SD= 2.95), tribe Paspaleae (mean= 22.6 Mya, SD= 3.13), and supertribe Panicodae (mean= 25.46 Mya, SD= 7.76). Analyses were conducted using three independent runs of 100 million generations, sampling every 50,000, and discarding the first 25% of each run as burn-in.
Panicinae(BEAST)CAL2
xml file for phylogenetic analyses and divergence time estimation in BEAST 1.8.4 using the nucleotide substitution models unlinked across codon position, a birth-death model with incomplete sampling as tree prior, an uncorrelated lognormal clock model for the rate variation among tree branches, and a calibration based on external angiosperm fossils (eudicots and non-grass monocots) together with the controversial phytolith microfossils of Poaceae (calibration scheme 2). We used secondary calibrations in normal prior distributions for the following six crown nodes: subfamily Panicoideae (mean= 48.09.18 Mya, SD= 4.94), most recent common ancestor (MRCA) of supertribes Andropogonodae –Panicodae (mean= 36.65 Mya, SD= 3.64), supertribe Andropogonodae (mean= 34.29 Mya, SD= 3.73), tribe Andropogoneae (mean= 14.45 Mya, SD= 2.71), tribe Paspaleae (mean= 26.25 Mya, SD= 3.61), and supertribe (mean= 30.74 Mya, SD= 2.97). Analyses were conducted using three independent runs of 100 million generations, sampling every 50,000, and discarding the first 25% of each run as burn-in.
Panicinae_CAL1_MCCT_BEAST
Maximum clade credibility (MCC) tree of Panicoideae obtained from BEAST analyses with the ndhF sequences, using the uncorrelated lognormal relaxed clock model and secondary calibrations based only on external angiosperm fossils (calibration scheme 1).
Panicinae_CAL2_MCCT_BEAST
Maximum clade credibility (MCC) tree of Panicoideae obtained from BEAST analyses with ndhF sequences using the uncorrelated lognormal relaxed clock model and secondary calibrations based on external angiosperm fossils together with the phytolith microfossils of Poaceae (calibration scheme 2).
Panicinae_CAL1_MCCT_BEAST_Pruned
Maximum clade credibility (MCC) tree of Panicoideae obtained from BEAST analyses with the ndhF sequences, using the uncorrelated lognormal relaxed clock model and secondary calibrations based only on external angiosperm fossils (calibration scheme 1); and pruned to include only the subtribe Panicinae. This tree was used in the biogeographic analyses with the package BioGeoBEARS 0.2.1
Panicinae(BEAST)CAL1_1000trees
1000 trees randomly subsampled from the posterior distribution of chronograms obtained in BEAST analyses under the calibration scheme 1, and pruned to include only the subtribe Panicinae. These trees were used in the analyses of the evolution of life history in the subtribe Panicinae.
FIGS1
Maximum clade credibility (MCC) tree of Panicoideae obtained from BEAST analyses with the ndhF sequences, using the uncorrelated lognormal relaxed clock model and secondary calibrations based on external angiosperm fossils together with the phytolith microfossils of Poaceae (calibration scheme 2, see materials and methods). Red boxes indicate phylogenetic placement of Panicum species recovered outside subtribe Panicinae. Posterior probability ≥ 0.9 are shown on the branches and horizontal bars on the nodes indicate the 95% HPD of ages. Subtribe Panicinae are shown in detail in Fig. S2. Mya, million years ago; Pli, Pliocene; Plei, Pleistocene. Results from divergence time estimation based only on external angiosperm fossils calibration are shown in Fig. 1.
FIGS2
Divergence time estimations for subtribe Panicinae. Maximum clade credibility (MCC) tree of Panicoideae obtained from BEAST analyses with ndhF sequences using the uncorrelated lognormal relaxed clock model and secondary calibrations based on external angiosperm fossils together with the phytolith microfossils of Poaceae (calibration scheme 2, see materials and methods). Only subtribe Paniceae is shown in detail; for the remaining clades see Fig. S1. Posterior probabilities ≥ 0.9 are shown on the branches and horizontal bars on the nodes indicate the 95% HPD of ages. Vertical bars indicate sections within Panicum. Paniceae 1 refers to tribe Paniceae excluding subtribes Cenchrinae, Melinidinae, and Panicineae. Mya, million years ago; Pli, Pliocene; Plei, Pleistocene. Results from divergence time estimation based only on external angiosperm fossils calibration are shown in Fig. 2.
FigS3
Ancestral range estimation (ARE) on the Panicinae chronogram using the BayArea+J model in BioGeoBEARS. Pie charts at nodes show the relative probability of the possible states (areas or combination of areas) before the instantaneous speciation event, whereas those on branches represent probability of the descendant lineage immediately after speciation. Boxes to the left of taxon names indicate areas of tip species.
Table S1
Areas and dispersal probabilities used in BioGeoBEARS analyses of subtribe Panicinae.
TableS2
Number of biogeographical events estimated in the history of the subtribe Panicinae using biogeographical stochastic mapping. Event counts were averaged across 1000 BSMs and are presented here with the standard deviations in parentheses. Total event counts are given for range-expansion events (anagenetic dispersal) (a), founder events (cladogenetic dispersal) (b), and Sympatry speciation events (c).
