BackgroundDiatoms are one of the most ecologically important aquatic micro-eukaryotes. As a group unambiguously recognized as diatoms, they seem to have appeared relatively recently with a limited record of putative remains from oldest sediments. In contrast, molecular clock estimates for the earliest possible emergence of diatoms suggest a considerably older date. Depending on the analysis, Paralia and Leptocylindrus have been recovered within the basal molecular divergences of diatoms. Thus these genera may be in the position to inform on characters that the earliest diatoms possessed.FindingsHere we present auxospore development and structure of initial and post-auxospore cells in a representative of the ancient non-polar centric genus Paralia. Their initial frustules showed unusual, but not unprecedented, spore-like morphology. Similarly, initial frustules of Leptocylindrus have been long considered resting spores and a unique peculiarity of this genus. However, even though spore-like in appearance, initial cells of Paralia readily resumed mitotic divisions. In addition, Paralia post-auxospore cells underwent several rounds of mitoses in a multi-step process of building a typical, “perfect” vegetative valve. This degree of heteromorphy immediately post-auxosporulation is thus far unknown among the diatoms.ImplicationsA spore-related origin of diatoms has already been considered, most recently in the form of the “multiplate diploid cyst” hypothesis. Our discovery that the initial cells in some of the most ancient diatom lineages are structurally spore-like is consistent with that hypothesis because the earliest diatoms may be expected to look somewhat similar to their ancestors. We speculate that because the earliest diatoms may have appeared less diatom-like and more spore-like, they could have gone unrecognized as such in the Triassic/Jurassic sediments. If correct, diatoms may indeed be much older than the fossil record indicates, and possibly more in line with some molecular clock predictions.