One of the most distinctive types of pollen in seed plants is the saccate type, which is found in about half of living conifer species, all of them members of the families Pinaceae and Podocarpaceae, with two (rarely three) ‘sacs’, ‘bladders’, or ‘wings’ formed by separation of the inner and outer layers of the exine, giving the characteristic Mickey Mouse silhouette of pine pollen. Despite the conspicuousness of the sacs, their function and evolutionary significance have been the subjects of much confusion. Early botanists assumed that they functioned to increase buoyancy in the air and thereby favored longer wind dispersal. However, in his classic treatise on pollen, Wodehouse (1935) noted that the sacs fold together when the grain is shed, ‘So that if the bladders are organs of flight, pollen grains are possibly the only flying organisms of which it can be said that they fold up their wings and fly away.’ He suggested instead that the sacs reduce desiccation by sealing off the germination aperture located between them. In a series of classic studies on pollination in conifers (summarized in Doyle, 1945), Joseph Doyle (no relation) proposed a very different explanation – that the sacs are devices for flotation in the liquid pollination drop secreted by the ovule for pollen capture. He showed that the presence of sacs is correlated with downwards orientation of the ovule relative to gravity at the time of pollination, and that after capture the pollen floats upwards in the micropylar canal to the nucellus. He interpreted this system as ancestral in conifers, based on the fact that Late Paleozoic conifers and the putatively related fossil cordaites had saccate pollen (Florin, 1951), and the absence of sacs in other living conifers as a result of loss, which he associated with either loss of the drop mechanism and germination outside the micropyle or a shift to upwards orientation of the ovules, as in Cupressaceae and Taxaceae. Awareness of this work seemed to fade over the next few decades, but new experimental studies that started in the 1990s confirmed and extended Doyle’s hypothesis, demonstrating the effectiveness of sacs in flotation and yielding new insights. For example, Tomlinson et al. (1991) showed that the pollen drop may expand and ‘scavenge’ pollen that lands on areas outside the micropyle, while Salter et al. (2002) showed that saccate pollen is rapidly wetted and pulled inside the pollination drop rather than remaining on its surface. But the most definitive and statistically well-supported confirmation of the flotation hypothesis is the series of ingenious experiments reported in this issue of New Phytologist by Andrew Leslie (pp. 273–279), which clear up contrary and equivocal results of earlier studies by showing that saccate pollen is more efficiently floated up the micropylar canal than nonsaccate pollen, both in vitro and in vivo.
‘Despite the conspicuousness of the sacs, their function and evolutionary significance have been the subjects of much confusion.’
As discussed by Leslie, these observations are important for understanding the significance of the origin and loss of saccate pollen as an aspect of changes in reproductive systems. Conversely, better understanding of the phylogenetic relationships among saccate and nonsaccate taxa may clarify the direction of these changes. Considering only living plants, and assuming their most likely relationships based on molecular data (Burleigh & Mathews, 2004; Fig. 1), it would be most parsimonious to assume that saccate pollen was an independent innovation in Pinaceae and Podocarpaceae, and no special adaptive explanation would be required to explain its absence in other conifers or in cycads, Ginkgo, Gnetales and angiosperms – these would simply be groups in which sacs had never evolved. However, the fact that the oldest known fossil conifers and cordaites all had sacs (Florin, 1951; Poort et al., 1996) suggests that there was positive selection for loss in the nonsaccate conifer groups, perhaps, for example, because of a shift to upwards ovule orientation.
Saccate pollen also occurred in many so-called seed ferns, fossils that retained the ancestral fernlike leaf morphology of seed plants as a whole, which originally had spore-like, nonsaccate pollen. The case that attracted the most attention was the Late Carboniferous seed fern Callistophyton (Fig. 2), which also resembled coniferophytes in its platyspermic seeds and stem anatomy, and stimulated Rothwell’s (1982) hypothesis that coniferophytes were derived from seed ferns, rather than being a separate line of evolution from Devonian progymnosperms. However, saccate pollen has long been known in younger Permian and Mesozoic ‘seed ferns,’ namely glossopterids, some peltasperms (e.g., Autunia), corystosperms and Caytonia. Some cladistic analyses based on morphological characters have even indicated that all living seed plants belong to a clade united by a single origin of saccate pollen, implying that all living nonsaccate taxa, including cycads and angiosperms, were ultimately derived from saccate ancestors (Nixon et al., 1994; Doyle, 1996; Hilton & Bateman, 2006). However, this is uncertain because of the large number of nonsaccate taxa scattered among seed plants and the poor resolution of relationships – the arrangement of the five major lines of living seed plants is one of the most contentious issues in molecular systematics, and this is compounded by the addition of fossils, in which many characters are not preserved.
The effects of taxon sampling can be illustrated by comparing Fig. 1 with Fig. 2, a tree found when living taxa were forced into the most favored molecular arrangement and fossils were allowed to attach in their most-parsimonious positions based on morphology (Doyle, 2008). Although parsimony reconstruction indicates that sacs arose independently in glossopterids and Caytonia, they are a synapomorphy of the clade including Callistophyton and all living gymnosperms. This suggests that the further addition of fossil taxa might change the results again, especially considering the abundance of dispersed saccate pollen in Late Paleozoic sediments and the large number of poorly understood fossil seed plants. For example, although the Paleozoic conifer Emporia establishes the presence of saccate pollen on the line leading to living conifers, the subsequent history of sacs is equivocal: it is equally parsimonious to assume that sacs were retained in Pinaceae and Podocarpaceae and lost in Gnetales, Araucariaceae and the Cupressaceae–Taxaceae clade, or that they were lost at the base of the living conifers and regained in Pinaceae and Podocarpaceae. However, placement of other saccate fossils within living conifers, for example on the line to Araucariaceae, could shift the balance in favor of the former hypothesis. In light of the functional studies, the identification of saccate pollen as a synapomorphy of living seed plants would imply that all living nonsaccate taxa underwent a shift away from downwards-oriented ovules with a pollination drop somewhere in their history.
As Leslie emphasizes, his results also have implications for functional interpretation of the morphology of fossil plants. Thus, saccate pollen is correlated with downwards ovule orientation in Callistophyton, where ovules occurred on the lower surface of fernlike leaves, and probably in Caytonia, which has been identified as the sister group of angiosperms in some studies (Doyle, 1996, 2008; Hilton & Bateman, 2006) but rejected in others (Nixon et al., 1994; Rothwell et al., 2009). One argument against this relationship is the contrast between the saccate pollen of Caytonia and the nonsaccate, monosulcate pollen of basal angiosperms, but even if the common ancestor of the two groups had sacs (which is not supported by Fig. 2), their loss would be expected on the line to angiosperms, because of a shift in the site of pollen capture and germination from a pollination drop to the exposed surface of a stigma. The flotation hypothesis may also provide new insights on some contentious issues of the morphological interpretation of fossil plants, such as the Permian glossopterids, where ovules were borne on one surface of a sporophyll or cupule, which was itself attached to the adaxial side of a leaf (Fig. 2). The simplest explanation is that the sporophyll was borne on an axillary branch adnate to the midrib of the leaf. However, because of limitations of preservation, it is uncertain whether the ovule-bearing side of the sporophyll faced towards the leaf or away from it. Anatomical evidence shows that the ovule-bearing surface was adaxial, but this would be consistent with either orientation of the sporophyll relative to the leaf, depending on whether it was attached to the side of the axillary branch towards or away from the main axis (Fig. 2). Taylor (1996) argued that the ovules faced away from the leaf, on the grounds that the opposite orientation would make it difficult for pollen to reach the ovules, although pollen manages to get to the micropyle in conifers when the cone scales are only slightly separated. However, glossopterids had saccate pollen, so the flotation hypothesis would favor the view that the ovules were on the surface facing the subtending leaf, because this would mean that they were oriented downwards relative to gravity.