Key points are not available for this paper at this time.
Summary In aggregate, past discussions of heterospory and its role in the alternation of generations are riddled with ambiguities that reflect overlap of terms and concepts. Heterospory sensu lato can be analyzed more effectively if it is fragmented into a series of more readily defined evolutionary innovations: heterospory sensu stricto (bimodality of spore size), dioicy, heterosporangy, endospory, monomegaspory, endomegasporangy, integumentation, lagenostomy, in situ pollination, in situ fertilization, pollen tube formation, and siphonogamy (Tables 1, 2, Figs 1, 13). Current evidence suggests that the last five characters are confined to the seed‐plants. The fossil record documents repeated evolution of heterosporous lineages from anisomorphic homosporous ancestors. However, interpretation is hindered by disarticulation of fossil sporophytes, the difficulty of relating conspecific but physically independent sporophyte and gametophyte generations in free‐sporing pteridophytes, the inability to directly observe ontogeny, and the rarity of preservation of transient and/or microscopic reproductive phenomena such as syngamy and siphonogamy. Unfortunately, the rarely preserved phenomena are often of far greater biological significance than corresponding readily preserved phenomena (e.g. heterospory versus dioicy, heterosporangy versus endospory). In most fossils gametophyte gender can only be inferred by extrapolation from the morphology of the sporophyte and especially of the spores. This is readily achieved for species possessing high‐level heterospory, when the two spore genders have diverged greatly in size, morphology, ultrastructure and developmental behaviour. However, the earliest stages in the evolution of heterospory, which are most likely to be elucidated in the early fossil record of land‐plants, also show least sporogenetic divergence. It is particularly difficult to distinguish large microspores and small megaspores from the large isospores of some contemporaneous homosporous species (Figs 3–6 a , g ). Heterospory is best identified in fossils by quantitative analysis of intrasporangial spore populations. The spatial scale of the differential expression of megaspores and microspores varies from co‐occurrence in a single sporangium (anisospory) to different sporophytes (dioecy) (Figs 6–8). Studies of the relative positions of the two spore morphs on the sporophyte, and of developmentally anomalous terata (Fig. 9), demonstrate that gender is expressed epigenetically in both the sporophyte and gametophyte. Hormonal control operates via nutrient clines, with nutrient‐rich microenvironments favouring femaleness; megaspores and microspores compete for sporophytic resources. External environments can also influence gender, particularly in free‐living exosporic gametophytes. The evolution of heterospory was highly iterative. The number of origins is best assessed via cladograms, but no current phylogeny includes sufficient relevant tracheophyte species. Also, several extant heterosporous species differ greatly from their closest relatives due to high degrees of ecological specialization and/or saltational evolution; extensive molecular data will be needed to ascertain their correct phylogenetic position. Current evidence suggests a minimum of 11 origins of heterospory, in the Zosterophyllopsida (1: Upper Devonian), Lycopsida (1: Upper Devonian), Sphenopsida (?2: Lower Carboniferous), Pteropsida (?4: Upper Cretaceous/Palaeogene) and Progymnospermopsida (?3: Upper Devonian/Carboniferous). The arguably monophyletic Gymnospermopsida probably inherited heterospory from their progymnospermopsid ancestor (Table 3, Figs 11–13). No origin of heterospory coincides with the origin of (and thus delimits) any taxonomic class of tracheophytes. The actual number of origins of heterospory is probably appreciably higher, exceeding that of any other key evolutionary innovation in land‐plants and offering an unusually good opportunity to infer evolutionary process from pattern. Heterospory reflects the convergent attainment of similar modes of reproduction in phylogenetically disparate lineages. Nature repeated this experiment many times, whereas true phylogenetic synapomorphies evolve only once and require a unique causal explanation. Cladograms also offer the best means of determining the sequence of acquisition of heterosporic phenomena within lineages, here exemplified using the lycopsids (Fig. 10). Comparison of such sequences among lineages can potentially allow generalizations about underlying evolutionary mechanisms. Current evidence (albeit inadequate) indicates broadly similar sequences of character acquisitions in all lineages, though they differ in detail. Some logical evolutionarily stages were temporarily bypassed. Other lineages appear to have acquired two or more characters during a single saltational evolutionary event. Heterosporic phenomena can also be lost during evolution. Although no complete reversals to homospory have been documented, this could reflect unbreakable developmental canalization of heterospory rather than selective advantage relative to homospory. Several extant species refute widely held assumptions that certain phenomena, notably heterospory and dioicy, are reliably positively correlated. Moreover, some fossils are likely to possess combinations of heterosporic characters that are not found in their extant descendants. Fossil data have played a crucial role in understanding both the number of origins of heterospory and the ensuing patterns of character acquisition. Although non‐adaptive evolutionary events are likely in at least some lineages, the highly iterative nature of heterospory and similar patterns of character acquisition in different lineages together suggest that its evolution was largely adaptively driven. However, many previous adaptive models of heterosporic evolution confused pattern and process, and paid insufficient attention to the role of the environment as a passive filter of novel morphotypes. Linear gradualistic models were imposed on the data, often intercalating hypothetical intermediates where desired. The evolution of heterospory is best understood in terms of inherent antagonism between the sporophytic and gametophytic phases of the life history for control of sex ratio and reproductive timing. Control is achieved directly by the gametophyte, via gametogenesis, and indirectly by the sporophyte, via sporogenesis and the ability to determine to varying degrees the environment in which the gametophyte undergoes sexual reproduction. Increasing levels of heterospory (particularly the acquisition of endospory) compress the heteromorphic life history, as the increasingly dominant sporophyte progressively co‐opts the sex determination role of the gametophyte. The resulting life history is more holistic, effectively streamlining evolution by offering only a single target for selection. However, by wresting control of sex ratios from the gametophyte, the ability of the sporophyte to respond rapidly to environmental changes decreases. This competitive weakness is greatest for heterosporous species possessing exosporic but obligately unisexual gametophytes (epitomized by the pteropsid Platyzoma *). It can be alleviated in endosporic species by occupying favourable environments (e.g. the aquatic Salviniales and Marsileales), switching to an apomictic mode of reproduction (thereby incurring inbreeding depression; e.g. many selaginellaleans), or acquiring more complex pollination biologies (thereby by‐passing the environment as a selective filter: the seed‐plants). Lineages differ greatly in the maximum number of heterosporic characters that were acquired by their most derived constituent species. Several Devono‐Carboniferous lineages reached the level of reducing numbers of functional megaspores to one per sporangium (Figs 7 e , f , 8, 13), but only the putatively monophyletic gymnospermopsids broke through this apparent barrier to acquire the increasingly complex pollination biology that characterizes modern seed‐plants. Many theories have been proposed to explain the remarkable success (both in terms of species diversity and ecological dominance) of seed‐plants. The majority focus on characters that are absent from the earliest seed‐plants (the Devono‐Carboniferous lyginopterid pteridospermaleans), which were no more reproductively sophisticated than other penecontemporaneous lineages possessing advanced heterospory (particularly the most derived lycopsids, equisetaleans and progymnospermopsids). Reliable pollination was a key reproductive breakthrough, though the sophisticated economic‐vegetative characters inherited by the earliest seed‐plants from their putative progymnospermopsid ancestors were probably equally important in ensuring their success in water‐limited habitats. With the exception of some ecologically specialized pteropsids, known heterosporous lineages originated during a relatively short period in the Upper Devonian and Carboniferous (Fig. 11). They exploited a wind
Bateman et al. (Mon,) studied this question.