The uncovering of a large soil surface preserved under sediment for 390 million years has exposed plant remains which show that the world's earliest forests were much more complex than previously thought. See Letter p.78
Think for a minute about 'early' life on land. Complexity is probably not the first thought that springs to mind. Botanists also tended to consider the earliest forests to be simple entities composed of a single type of tree. But on page 78 of this issue, Stein et al.1 challenge this view with the description of uncovered plant remains from the ancient forest of Gilboa in New York state. They identify three distinct plant types (or components) in this Middle Devonian ecosystem, which dates to around 390 million years ago, and thereby re-evaluate the ecology of this early environment.
The Gilboa forest is iconic. The discovery2 of this “oldest fossil forest” was made in the 1920s at the Riverside Quarry site in Gilboa and other nearby sites, where spectacular sandstone casts formed by the fossilization of stumps of Eospermatopteris — the earliest known trees — were found. Stein et al.3 had previously discovered specimens representing the trunk and crown of these trees, and assigned the Eospermatopteris genus to the Cladoxylopsida, an extinct class of plants related to ferns. The trees had a slender trunk six metres or more in height that bore a crown of leafless and short-lived branches. These were continuously renewed as the plant grew. Eospermatopteris trees were assumed to be the only trees of the Gilboa forest.
But when access to the forest's fossils increased after the removal of backfill from the Riverside Quarry site in 2010, researchers discovered a 1,200-square-metre palaeosol surface — one that had been preserved under sediment. Because most of the plant fossils in this surface were found still located where they had grown, Stein et al.1 were able to define plant types and reconstruct structural features of the forest.
The authors identified three large-plant components with apparent differences in growth habit, abundance and phylogenetic relationships (Fig. 1). These components were the previously identified cladoxylopsid tree Eospermatopteris; a large rhizomatous plant (one with underground stems growing horizontally4) belonging to the extinct aneurophytalean progymnosperms; and a tree with bark similar to that of the lycopsid trees that inhabited coal swamps about 310 million years ago, but for which the authors found too few and fragmentary remains for proper reconstruction.
The discovery that the Gilboa forest did not consist solely of cladoxylopsid trees is remarkable. Palaeoecological studies of other Devonian-period sites describe early vegetated terrestrial landscapes partitioned into a 'two-dimensional' suite of patches growing side by side, each composed of closely related plants with similar morphologies and life traits, and adapted to the same environmental conditions5,6. This structure of Devonian landscapes has almost become a dogma in palaeobotany, but Stein and colleagues' report1 provides the first direct evidence that some early forests contained widely divergent groups of plants. Furthermore, the differences in the growth patterns of the plants adds a third spatial dimension to the forest ecosystem structure: below the Gilboa trees was an understorey of progymnosperm plants with underground, horizontally growing stems and aerial axes.
The combination of a large exposed surface area and in situ plant remains at the Riverside Quarry provided sufficient fossil evidence to identify the vertical structures of the Gilboa forest. Such conditions are so rarely met that Stein and colleagues' findings1 have the power to cast doubt on the supposed simple organization of other early terrestrial ecosystems.
The authors' report also provides evidence to support an alternative view of the growth habit of aneurophytalean progymnosperm plants. These plants were assumed to be bushy and to have a shallow and limited root system7. However, subsequent calculations showed8 that the shoots of Tetraxylopteris plants — the progymnosperm genus probably represented at Riverside Quarry — were not biomechanically capable of the self-support seen in the shoots of bushes and trees. Stein and colleagues' new observations support this result. In addition, they show that the aneurophytalean subterranean system — consisting mainly of rhizomes up to 15 cm in diameter — comprised a large amount of wood and had significantly more mass than previously estimated.
This finding is noteworthy for at least two reasons. First, the Eospermatopteris trees — the most conspicuous aboveground members of the Gilboa forest — had a limited amount of wood (if any)9; therefore the description of the woody rhizomes adds credibility to the hypothesis that, in early land plants, wood did not evolve as an adaptation for mechanical support10. Second, the progymnosperm architecture challenges widely accepted theories that there was a straightforward correlation between a gradual increase in plant stature above ground and a gradual increase in size of the corresponding underground parts during the Devonian period7.
In addition to their findings on plant types, Stein et al.1 studied sediment deposits around the forest, which existed near the shoreline of an inland sea. Previous studies had likened the Gilboa forest environment to a tranquil swamp, but the observations of Stein et al. suggest that the forest was periodically affected by brutal episodes of sea-level rise, which killed some of its plant life. Such disturbances represent major constraints for ecosystems, and can result in evolutionary selection for specific life strategies in plant species. Notably, plants that have adapted to disturbed habitats are generally small in size11. But if Stein and colleagues' observations of a disturbed environment are correct, the occurrence of large trees in the Gilboa forest would go against this trend, and represent an especially intriguing case of plant evolution.