Early evolutionary history of the seed

The seed is an essential stage in the life history of gymnospermous and angiospermous plants, facilitating both their survival and dispersal. We reappraise knowledge of the evolutionary history of the gymnospermous seed, from its origin in the late Devonian through to the well‐known end‐Permian extinctions – an interval encompassing the origins of most major lineages of seed‐bearing plants. The framework for our broader discussions is a novel cladistic analysis of anatomically preserved Palaeozoic seeds, analysing 79 seed‐species for 89 morphological characters in a matrix containing only 24% missing values. The resulting consensus tree is weakly but fully resolved and compatible with traditional division into three informal seed groups: paraphyletic lagenocarps, paraphyletic trigonocarps and monophyletic cardiocarps. Three seed‐genera – Rhychosperma, Albertlongia and Muricosperma – are revealed as potential ‘missing links’ between groups, and modest re‐circumscription of seed‐genera is required. Although the value of single‐organ phylogenies remains controversial, the present seed‐tree topology receives general support from the dated sequence of first appearances of seed‐species in the fossil record, and from the topologies of morphological cladistic studies that combined conceptually reconstructed fossil plants with primitive extant lineages, notably ginkgos and cycads. Branch lengths in the tree and phenetic distances in ordinations of the matrix indicate similar overall rates of character change through the Palaeozoic, rather than a fractal pattern reflecting progressively increasing constraint, although early changes in architectural and pollination‐related characters gradually give way to greater experimentation with the internal layering and external topography of the testa. Our process‐based evolutionary inferences are informed by extant gymnosperms, particularly Cycas and Ginkgo. The origin of the true seed is attributed primarily to (1) the complex biochemical signalling needed to allow the sperm to reach the archegonia through the megasporangium wall and (2) the localised apoptosis of the megasporangium hypothesised to have simultaneously allowed hollowing out of the nucellar apex to form a sophisticated pollen‐receiving apparatus (the pollen chamber) and secretion of a pollination drop to capture air‐borne (pre)pollen. Subsequent potential key innovations include transfer of function of both pollination‐drop channelling and pollen chamber sealing from the nucellar salpinx to the integumentary micropyle, and introduction of a haustorial pollen tube to direct spermatozoa towards the archegonia. Assuming that the seed‐plant megasporangium terminates an axis, synorganisation has played a key role in seed evolution, leaf‐like lateral organs being repeatedly pulled towards the apex and incorporated into the terminal structure. Lateral webbing of integumentary lobes eventually almost fully enclosed the nucellus, while a similar synorganisation process affecting a lower set of vegetative organs formed a cupule as yet another protective layer surrounding one or more ovules. Our tree refutes viewing these evolutionary developmental trends as linear transition series. The earliest seeds were small but soon increased to reach the maximum size achievable by gymnosperms. Dehiscence and dormancy mechanisms were likely primitive at best, while increasingly complex layering and sculpting of the testa may have aided both abiotic and biotic dispersal. The end‐Permian extinction of plants bearing lagenocarps and trigonocarps is attributed tentatively to one or more of several features of reproductive biology identified as being vulnerable to desiccation.