A number of dispersed Cretaceous and Tertiary pollen grains with exine morphologies suggestive of Asteraceae have been assigned to the form genus Tubuliforidites. Zavada and de Villiers (2000) carefully examined the ultrastructure of two species of Tubuliforidites with Anthemoid type morphology from the Paleocene-Eocene of South Africa. The fossil grains, as well as Calyceraceae, Barnadesioideae, and Mutisieae, have a reticulate wall layer. Unlike Tubuliforidites, members of Barnadesioideae have thick columellae that extend through the exine. In Mutisieae a distinct intertectum is present. Like Barnadesioideae, some members of Mutisieae have basal columellae that extend through the exine. The sister group to Asteraceae, Calyceraceae, and a second family Valerianaceae, exhibit wall structure types most similar to that of Tubuliforidites. It is possible that the reticulate layer in Barnadesioideae and Mutisieae is not homologous to the condition found in Calyceraceae, Valerianaceae, and Tubuliforidites. Grains of Calyceraceae, Valerianaceae, and Tubuliforidites appear to have a reticulate exine layer that is derived from divisions of the upper portion of columellae. In considering both wall structure and phylogenies of Asteraceae and closely related families, it appears that the Anthemoid pattern has evolved multiple times both within Asteraceae and in related families. The presence of this pattern in fossil grains is not a good character to use to recognize Asteraceae in the fossil record. The challenge for the paleopalynologist interpreting the ultrastructure of grains believed to have affinities with Asteraceae is to determine if the reticulate layers are not derived from distal divisions of the columellae. Also, the presence of columellae extending through the wall would also suggest an affinity to Asteraceae.

Key words: Asteraceae, Barnadesioideae, Calyceraceae, Mutisieae, Pollen, Tubuliforidites