Phylogenetic uncertainty and fossil calibration of Asteraceae chronograms

Barreda et al. (1) claim a Cretaceous fossil pollen type is an extinct Asteraceae. Concluding this pollen type is “nested within Dasyphyllum (crown representative),” they calibrate a Dasyphyllum + Barnadesia crown node (Dasyphyllum crown absent) and estimate an 85.9-Ma Asteraceae crown age that potentially compresses asterid evolution by tens of millions of years. However, the bootstrap majority consensus topology reported could not be reproduced from the data; instead, the fossil resolved in a trichotomy with Calyceraceae and Asteraceae. Thus, unambiguous assignment of these pollen grains to Asteraceae is premature.

Paleocene, not Cretaceous, mean ages of Asteraceae result from calibration placement consistent with the fossil’s phylogenetic position in the reproduced bootstrap tree. Calibration at the Asteraceae + Calyceraceae crown node (second calibration scenario; figure S5A and table S2 of ref. 1) is not consistent with the bootstrap, because it excludes the possibility that the fossil is a stem member of the Asteraceae + Calyceraceae clade. The third scenario placement is consistent with the fossil + Asteraceae + Calyceraceae trichotomy supported by 69% bootstrap proportion. The study’s six calibration scenarios illustrate the ambiguous phylogenetic position (identity) of Tubulifloridites lilliei type A pollen that hinders unequivocal placement in chronogram construction.

Furthermore, we should be cautious of sensational conclusions underpinned by an inferred phylogenetic relationship supported by few characters. Crucial character scoring and encoding are particularly difficult from taxa of the dispersed pollen fossil record (2). Extant Dasyphyllum, Chuquiraga, and Doniophyton have an exine bilayer (3), but when the same character observed under light microscopy is scored (character 21), Dasyphyllum spp. share a single layer state with the fossil as opposed to other Barnadesioideae. Encoding “columellate layer visibility under light microscopy” (character 19) results in unintentional character weighting. Exine thickness (character 22) is much smaller in Dasyphyllum inerme and Dasyphyllum velutinum than in other Dasyphyllum spp. (3) that would be scored as other Barnadesioideae and different from the fossil had they been sampled. Characters 19, 21, and 22 clearly contribute to place the fossil with Dasyphyllum. Character 17 assumes that concavities distributed asymmetrically (sometimes absent) along the intercolpal region in the fossil are homologous with symmetrically distributed intercolpal concavities in extant taxa and not the result of compression forces during fossilization (figure 4 in ref. 1). This character is scored as “present” in the fossil (table S1 in ref. 1) but described as “present or absent” [supporting information (p. 2) in ref. 1]. The authors did not explore the robustness of phylogenetic results to alternative scoring, encoding, or Dasyphyllum sampling.

Readers should not construe finding Paleocene-Eocene hothouse climate coincident with diversification of South American lineages as primarily due to the Dasyphyllum + Barnadesia calibration. Ages for these divergences result chiefly from the placement here of the Mutisiapollis telleriae + Raiguenrayun constraint as rationalized by Panero et al. (4), not at the Asteraceae crown node shown by Barreda et al. (5). These two calibrations placed in Asteraceae result in contrasting early evolutionary patterns in Barnadesioideae vs. the sister clade (stem lengths in figure 5 in ref. 1). The authors do not comment on this interesting consequence of their calibration placement, an effect that shrinks when the T. lilliei type A calibration is placed outside Asteraceae (figure S5 in ref. 1).