When considering where angiosperms came from, we are trying to bridge the gap between other known seed plants, living and extinct, and the earliest angio- sperms. What were the predecessors of the unique angiosperm stamens, carpels, and ovules? How and why did these organs evolve? Who were the ancestors of the angiosperms?
Modern phylogenetic studies confirm that angiosperms are not closely related to any modern gymnosperms, though until fairly recently gnetophytes were thought to be their closest cousins (Taylor and Kirchner 1996). Recall that gneto- phytes are now considered most closely related to the conifers (Chaw et al. 2000). The split between modern gymnosperms and the line that led to the angiosperms evidently occurred in some ancient group of seed ferns. During the Mesozoic Era (age of dinosaurs), seed ferns diversified and developed many specialized pollen- and ovule-bearing structures. The precursors of the angiosperms must be sought among them.
Several groups of Mesozoic plants derived from ancient seed ferns are fairly well known and were themselves quite diverse: Bennettitales, Caytoniales, Corystospermales, and Glossopteridales. Of course, if the actual precursors of angiosperms lived in dry upland habitats, it is not likely that any were fos- silized. The fossils we do have were perhaps descendants of early angiosperms that had migrated into moister environments (“museum” habitats). They can, however, provide clues as to the reproductive features that were developing at that time.
The arrangement of male and female organs into a bisexual flower seems quite similar to structures in the Bennettitales, which flourished from the early Triassic to the late Cretaceous. Members of this order were generally stocky plants with compound leaves, and strongly resembled cycads (Fig. 6.8). The pollen-bearing organs were large compound structures, each with numerous pollen sacs, and these surrounded a central spike of ovules. Some had large bracts below the reproductive organs that could be interpreted as tepals, but in others, the back- sides of the large pollen organs fit together to form a hard casing around the whole “flower.” This elaborate structure most likely served to channel pollen-eating bee- tles in an orderly path so as to repeatedly carry pollen from one cone to the ovules in another, just as in cycads and early flowering plants.
The main differences between a Bennettitalean flower and an angiosperm flower are that the ovules were lined up naked along the central axis (there were no carpels), they were straight, and they had only one integument, as in other gymnosperms. The pollen structures could have conceivably simplified into typi- cal angiosperm stamens, but there was no obvious starting point for the peculiar structure of the angiosperm ovules or the formation of carpels. For these reasons,
the “flowers” of the Bennettitales are considered a parallel development and unre- lated to those in the angiosperms.
The term “Mesozoic seed ferns” is often applied to a loosely defined group of seed plants descended from seed ferns that shared the earth with the dinosaurs. Like more ancient seed ferns, leaves of these Mesozoic seed plants were complex, and both ovules and pollen sacs were borne on specialized structures that some- times resembled leaves and sometimes resembled shoots. These extinct orders of plants are considered to be branches of the angiosperm stem group, which in turn probably split off from other seed plants as much as 350 million years ago (Renner 2009). The Bennettitales are sometimes included under the seed fern umbrella, sometimes not. Although the ovules in this specialized group are on simple stalks, the leaves and pollen structures resemble those of earlier seed ferns.
Some other Mesozoic seed ferns did not have flowerlike structures, but in one way or another their stamens, ovules, and ovule-bearing structures are more like those of angiosperms. We do not, however, have enough evidence to clearly point to any particular group as directly ancestral to them. The extinct order Caytoniales has been of greatest interest, but some arguments can be made for ancestry among the Corystospermales and Glossopteridales (Retallack & Dilcher 1981).
Figure 6.8 The Bennettitales included plants that superficially resembled the cycads but had rather different and more complex reproductive cones that contained both ovules
In the Caytoniales, pollen was borne on complex branching structures (Fig. 6.9), in sacs at the end of slender twigs, seeming to contradict the notion that early angiosperm stamens were flat and leaflike. In some members of this order, the pollen sacs were bundled in groups of four, suggesting that the angiosperm anther might represent an extreme simplification of such a structure. Stamens would have thus been more like twigs than leaves from the beginning. The pollen sacs in Caytoniales, however, were arranged uniformly around a central axis, not positioned two on each side of a somewhat flattened platform as in angiosperms (see Fig. 6.5B). So we still don’t know whether the first angiosperm stamens were twiglike or leaflike.
Ultimately, all reproductive structures in angiosperms and other seed plants trace back to the large fronds of seed ferns. The shoot-like system of branching pollen structures in the Caytoniales probably represents a modification of such an ancient frond. As an analogy, many ferns (such as the cinnamon fern, Osmunda cin- namomeum) have specialized spore-bearing fronds that are more shoots than leaves. Finally, the fossil Caytoniales we have are most likely not the direct ancestors of the angiosperms but closely related cousins that may have been specialized for different habitats and different lifestyles. We do not have to try to tie angiosperm structures directly back to any particular fossil form. The various available fossils just represent clues from which we might hypothesize a more generalized ancestor.
Figure 6.9 In the Caytoniales, both pollen (A) and cupules (C) were borne on complex, branching shoots that were ultimately derived from corresponding leafy structures in ancient seed ferns. In some Caytoniales, pollen sacs are bundled in groups of four (B), reminiscent of the anthers in angiosperms. Drawings from Brown 1935 (A, C) and redrawn after Harris 1937 (B).
The Caytoniales may give us a more direct clue about the origin of the pecu- liar angiosperm ovules, which with their double integument, simplified gameto- phytes, double fertilization, and bent configuration, seem to have no precedent. Fossils do not reveal much about the cellular structure or fertilization, but we can see a possible scenario for the double integument and bent configuration.
In the Caytoniales, as well as in the Corystospermales and Glossopteridales, the ovules were enclosed in cupules, which, like carpels, formed protective chambers around the ovules. Upon their discovery, it was proposed that cupules could have evolved directly into carpels. However, the ovules within all of these cupules were like those of other gymnosperms: straight and with just one integu- ment. If cupules had evolved directly into carpels, angiosperm ovules would have remained the same. In addition, cupules were shaped differently: typically more like curved, downward facing vases, with the ovules attached in a single row or cluster along the back side—quite different from the folded structure with two rows of ovules in the angiosperm carpel.
However, the message of the Caytonialean cupule might be completely dif- ferent. It was curved with a small opening facing downward toward the base (Fig. 6.10A). The ovules were in a cluster attached to the back side of the cupule, and hung downward with their tips directed toward the opening.
This orientation of the cupules led Stebbins (1974) and others to suggest that in some member of the Caytoniales, the cupule became, not a carpel, but the second integument of the ovule. In this particular species, the number of ovules within each cupule was reduced to just one, probably in adaptation to a harsh environment. The small opening of the single ovule then faced downward
Figure 6.10 According to the Caytonialean hypothesis championed by Stebbins, a cupule with several ovules (A) was reduced to a structure with a single ovule (B), and the cupule wall eventually became a second integument for the ovule (C) through continued reduction. The bent configuration of the angiosperm ovule was thus inherited from the original, downward-facing cupule. Redrawn after Brown 1935.
and more-or-less lined up with the opening of the cupule. Over time, the cupule “shrank down” tightly around the single ovule, effectively becoming another integument (Fig. 6.10B, C). Since the cupule was already bent downward, the new ovule/cupule fusion was also bent downward, just as the typical angiosperm ovule is today. The Caytonialean model simultaneously explains the two anoma- lies of the angiosperm ovule: its bent orientation and its two integuments.
Recent reviews of all available information (Doyle 2006 and Frohlich & Chase 2007), including new cladistic analyses combining DNA data and morphologi- cal characters, support the Caytoniales as the sister group (nearest cousin) of the angiosperms. Though this is still controversial, it considerably strengthens our hypothesis as to how the peculiar structure came about, as well as strengthen- ing the connection with the pollen-bearing structures of the Caytoniales. What about the carpel?