Australasian Science: Australia's authority on science since 1938

The Ancestor of All Flowers

Credit: TKphotography64

Credit: TKphotography64

By Charles Foster

An international collaborative project has reconstructed the ancestor of all modern flowering plants. What can it tell us about the evolution of this important group?

Flowering plants (angiosperms) are among the most successful biological groups on the planet. Estimates for the number of angiosperm species vary, but there are probably at least 350,000 species representing 90% of the diversity of all living land plants.

As expected in a group of this size, angiosperms are incredibly variable. They can exist as anything ranging from small herbaceous annuals to giant woody trees that live long enough to outlast civilisations. Angiosperms can be found in nearly all environments, whether aquatic or terrestrial, and have developed remarkable adaptations to survive, including parasitism of other plants or even carnivory of small animals.

Despite the importance of angiosperms, much of their evolutionary history has remained a mystery. Many outstanding questions remain, including when angiosperms first appeared and what these early angiosperms might have looked like.

The fossil record indicates that angiosperms first appeared not long before 130 million years ago. However, ages of biological groups can also be estimated using molecular clock analyses, which combine knowledge of the fossil record, differences in DNA sequences between modern species, and models of evolution. When the age of angiosperms is estimated using molecular clocks, a far earlier origin is suggested, perhaps up to 250 million years ago. Therefore, the timing for the origin of flowering plants is still relatively uncertain.

Similarly, scientists have long sought to determine the structure of the first flowers. Could they have been showy and complex? Or were they relatively simple and unassuming? The structure of the first flowers has far-reaching implications beyond mere aesthetics, revealing not only how angiosperms might have evolved from their seed plant ancestors but also how the astonishing diversity of modern species might have diversified from this ancestral floral structure.

These important unsolved questions in angiosperm floral macroevolution are what spurred the development of the eFLOWER scientific project. Primarily coordinated by Dr Hervé Sauquet of Université Paris-Sud in France (now working in Sydney at the Royal Botanic Gardens and Domain Trust) and Dr Jürg Schönenberger of the University of Vienna, the eFLOWER project consists of a team of 36 researchers from 13 countries. Our overall goals were to reconstruct the ancestral flower and then chart its subsequent early evolution and diversification into the major groups recognised today.

Several complementary approaches can be used to address these questions, such as identifying the closest extinct relatives to angiosperms in the fossil record, or making inferences based on comparisons between the development of reproductive structures in angiosperms and their seed-producing gymnosperm relatives. For our study we chose instead to reconstruct the ancestral flower using ancestral state reconstruction analyses.

In these analyses it is possible to take the distribution of key floral traits in modern angiosperm species and trace back the evolution of these traits according to models of morphological evolution. A benefit of this approach is that it provides a reconstruction of the ancestral floral traits not just for the ancestor of all angiosperms, but also for other important groups within angiosperms. This allows inferences to be made about how these groups arose through modification to the ancestral flower of all angiosperms.

To reconstruct such an ancient ancestor with any confidence, it is necessary to have reliable estimates of the relationships among angiosperms, with this phylogeny preferably scaled to time, and floral measurements taken from a dense sample of modern species.

For the first requirement, we used the results and data set from a recent molecular dating analysis led by Dr Susana Magallón of the Universidad Nacional Autónoma de México. This analysis sampled species throughout the broad angiosperm phylogeny, and took account of temporal information from a very large sample of fossils. We also re-analysed the molecular data set with various constraints representing conflicting hypotheses about the relationships among major angiosperm groups, as well as the uncertainty surrounding the age of angiosperms. This meant that we could assess whether our reconstructed ancestral flower was robust to these competing hypotheses for angiosperm evolution.

We compiled a massive morphological data set sampling 27 morphological traits from all 792 species in the molecular dating analyses, spanning 86% of angiosperm families. This was no small task: it is the largest data set of floral traits ever assembled, and took 6 years to coordinate, double-check and analyse. The successful generation of such a large data set is a testament to the power and importance of collaboration within science: half of all data was scored in a single week by 12 advanced botanical students in July 2013.

We used three broad methods to carry out ancestral state reconstruction: maximum parsimony, in which the simplest evolutionary scenario is assumed to be the best; and maximum likelihood and Bayesian inference, both of which incorporate explicit models of morphological evolution taking time into account. While maximum parsimony has been used for this purpose previously, our study is the first to reconstruct floral macroevolution on such a large scale using evolutionary model-based approaches.

The three approaches provided remarkably congruent reconstructions of the ancestral flower for most morphological traits, although the model-based methods resolved some long-standing questions for which parsimony analyses give equivocal results. Additionally, the model-based approaches allowed us to understand which components of the ancestral flower were constructed with relative confidence, and which components remain relatively uncertain. Crucially, the uncertainty surrounding the age of angiosperms or relationships among the major clades did not impact our reconstruction.

We found that the ancestral flower:

  • was most likely bisexual, with both male and female organs on the same flower;

  • had a whorled arrangement of male organs of more than ten stamens and spirally arranged female organs of more than five carpels;

  • didn’t have petals and sepals, but instead the perianth (outer part of the flower) was radially symmetrical and undifferentiated;

  • the perianth had at least ten tepals in at least two trimerous whorls (organs in multiples of three),which implies four whorls of three tepals; and

  • had free organs (i.e. they were not fused together).

While the complex botanical terms describing the ancestral flower might seem to paint an abstract picture, each component has important ramifications for its function. For example, the arrangement of organs in spirals or whorls has impacts on structural stability.

Importantly, the ancestral flower that we reconstructed also reinforces a view that is widely accepted within evolutionary biologist: that evolution does not necessarily progress linearly towards complexity. We reconstructed an ancestral flower that possessed a complex suite of morphological traits that is not present in any living species. Most modern flowers are considerably less complex than our ancestral flower, at least when considering the number of floral parts or whorls. The complexity of modern flowers becomes apparent at a finer scale, where we can see fusion or coordination of floral parts.

Nevertheless, using our ancestral flower we can propose that the early diversification of major angiosperm lineages was driven by a successive reduction in both organ number and the number of whorls. For example, the flowers of some lineages may have arisen through the loss of entire whorls of organs, while in other lineages the flowers might have instead arisen through the fusion of these whorls. Additionally, our results demonstrate that even the earliest diverging lineages of angiosperms are evolutionarily derived in several aspects despite once being incorrectly regarded as “primitive”.

Our reconstructed pathway of floral macroevolution also makes it necessary to question the homology of organs in the major angiosperm lineages. For example, do the petals in roses and lilies derive from the same ancestral perianth whorls? If not, have they evolved convergent molecular pathways that lead to the development of petals?

An important caveat of our study is that while the methods we employ reconstruct the most recent common ancestor of all living plants, this still does not necessarily represent the first ever flower. It’s possible that more ancestral, and potentially different, flowers once existed.

In other words, if the flower we reconstructed represents the parent of all living plants, then it’s possible that the grandparents or great-grandparents remain to be discovered. Considering this, our reconstructed flower might guide the recognition of further fossil discoveries as early diverging angiosperms.

Our results represent a major milestone in the understanding of how the incredible diversity in modern angiosperms first arose. Despite this, our study cannot directly inform us of how angiosperms arose from their seed plant ancestors. We hope that our findings will guide future researchers as they continue to work towards this important question.


Charles Foster is a PhD student at The University of Sydney’s School of Life and Environmental Sciences, and co-author of the eFLOWER research published in Nature Communications (https://tinyurl.com/y8ythdrx).