Leaf surface development and the plant fossil record: stomatal patterning in Bennettitales.

Stomata play a critical ecological role as an interface between the plant and its environment. Although the guard-cell pair is highly conserved in land plants, the development and patterning of surrounding epidermal cells follow predictable pathways in different taxa that are increasingly well understood following recent advances in the developmental genetics of the plant epidermis in model taxa. Similarly, other aspects of leaf development and evolution are benefiting from a molecular-genetic approach. Applying this understanding to extinct taxa known only from fossils requires use of extensive comparative morphological data to infer 'fossil fingerprints' of developmental evolution (a 'palaeo-evo-devo' perspective). The seed-plant order Bennettitales, which flourished through the Mesozoic but became extinct in the Late Cretaceous, displayed a consistent and highly unusual combination of epidermal traits, despite their diverse leaf morphology. Based on morphological evidence (including possession of flower-like structures), bennettites are widely inferred to be closely related to angiosperms and hence inform our understanding of early angiosperm evolution. Fossil bennettites - even purely vegetative material - can be readily identified by a combination of epidermal features, including distinctive cuticular guard-cell thickenings, lobed abaxial epidermal cells ('puzzle cells'), transverse orientation of stomata perpendicular to the leaf axis, and a pair of lateral subsidiary cells adjacent to each guard-cell pair (termed paracytic stomata). Here, we review these traits and compare them with analogous features in living taxa, aiming to identify homologous - and hence phylogenetically informative - character states and to increase understanding of developmental mechanisms in land plants. We propose a range of models addressing different aspects of the bennettite epidermis. The lobed abaxial epidermal cells indicate adaxial-abaxial leaf polarity and associated differentiated mesophyll that could have optimised photosynthesis. The typical transverse orientation of the stomata probably resulted from leaf expansion similar to that of a broad-leaved monocot such as Lapageria, but radically different from that of broad-leafed eudicots such as Arabidopsis. Finally, the developmental origin of the paired lateral subsidiary cells - whether they are mesogene cells derived from the same cell lineage as the guard-mother cell, as in some eudicots, or perigene cells derived from an adjacent cell lineage, as in grasses - represents an unusually lineage-specific and well-characterised developmental trait. We identify a close similarity between the paracytic stomata of Bennettitales and the 'living fossil' Gnetum, strongly indicating that (as in Gnetum) the pair of lateral subsidiary cells of bennettites are both mesogene cells. Together, these features allow us to infer development in this diverse and relatively derived lineage that co-existed with the earliest recognisable angiosperms, and suggest that the use of these characters in phylogeny reconstruction requires revision.