When I was a student, and for several decades after, a remarkable group of very knowledgeable botanists progressively put forward their ideas about how the 250,000 flowering plant species were related to each other. Foremost in this group were Armen Takhtajan (USSR), Arthur Cronquist & Ledyard Stebbins (USA), and Rolf Dahlgren (Denmark). Bill Philipson (University of Canterbury) was New Zealand’s main contributor of ideas to this ongoing programme, and Bruce Sampson (Victoria University of Wellington) contributed critical information on flower anatomy and pollen grain structure. In spite of their considerable combined intellect and knowledge, they never agreed on an overall scheme that was stable.
Two things changed all that in the 1980s and 90s. First, biologists realised that the centuries-long search for a natural classification meant seeking to discover and name only those groups that had evolved from a single common ancestor. That significant insight came from East German entomologist Willi Hennig in the 1950s, and it took a while to be applied to plants. Secondly, cheap and rapid DNA sequencing techniques gave us the abundant and reliable data we needed to begin this search. One advantage of this explicit, rather than intuitive, approach was that different ideas could be evaluated and compared objectively.
In 1993, a remarkable coalition of 52 botanists led by Mark Chase (now at Kew) published a ground-breaking paper, based on the variation of a gene called rbcL. This gene, found in chloroplasts, is the DNA code for the structure of the large subunit of a protein called rubisco, which is essential to photosynthesis and very abundant in plants. Their huge (for its time) data set took over a month to analyse on a Sun workstation. Later, similar collaborations used other genes, in particular atpB and ribosomal genes, and found very similar patterns.
In the 18 years since Chase et al. (1993), the family level classification of the flowering plants has been pretty well established and formalised as the Angiosperm Phylogeny Group (APG) system, but a few important problems remained. In part, these problems could be caused by evolutionary events that followed each other so rapidly that evidence of their history wasn’t left in the DNA of any living species, or they could arise because more recent changes to the DNA have overwritten and hidden that signal.
One solution was to sequence more genes, widely sampling across the three plant genomes (in the cell nucleus, chloroplast, and mitochondrion). This month, yet another such grand coalition (just 28 scientists this time) led by Doug & Pam Soltis at the University of Florida, Gainesville, has assembled a remarkable data set of 17 genes for 640 species to address the remaining problems. Their findings are broadly in agreement with the APG system, but they clarify some important issues.
Since the early 1990s it’s been clear the traditional classification of flowering plants into two large groups must be abandoned. Most of us learned at school that the flowering plants divide neatly into monocotyledons (with 1 seed leaf) and dicotyledons (with 2). Modern classifications still recognise the monocotyledons (monocots), including grasses, sedges, orchids, palms, and lilies, which have a single seed leaf, parallel leaf veins, and lack true wood, leaf stalks and taproots. The problem is that using the opposites of these characteristics to define a second group, the dicotyledons, simply won’t work. Plants that have 2 cotyledons, wood, taproots, leaf stalks, and net-like leaf veins include not just the remaining flowering plants, but many cone-bearing plants as well. To find natural groups, we must look for species that share uniquely evolved features. In this way, our ideas about classification are based on evolutionarily meaningful evidence, rather than conjecture. If we look for natural groups, the closest we can get to the traditional dicots is a group that’s defined by the unique feature of having three, rather than one, pore or groove in their pollen grains, a group that’s now called the eudicots or the tricolpates. Together the monocots and eudicots make up most of the flowering plants.
But about 3% of the flowering plants are neither monocots nor eudicots. These fall into a number of small groups and collectively they’re called basal angiosperms. Although they have two cotyledons, they're no closer related to eudicots than they are to monocots. One major classification problem has been the order in which the ancestors of the basal angiosperms diverged.
Most of the papers, including this new one, say the first evolutionary split of the ancestors of living flowering plants was between the ancestors of a single living species, Amborella trichopoda, and all the other flowering plants. Thus, if you wanted to divide flowering plants into two groups, one group would have to be Amborella and the other group would include everything else. Although many studies find this result, there are some that dispute it.
Amborella trichopoda grows as an understorey plant in montane forests in New Caledonia (left), and it’s become a bit of a celebrity there. Why would evolution produce two sister groups with such different evolutionary potentials? Maybe Amborella is the only surviving remnant of a group that was once much larger, but has mostly become extinct. Alternatively, there might be something about Amborella that makes it an unlikely group to radiate into lots of different habitats; perhaps it’s always been a very small lineage.
The second major split is also very uneven. On one side we have just the water-lilies and their close relatives, including a small aquatic plant called Hydatella that until very recently was thought to be a monocot; on the other side, all the remaining flowering plants.
After that split, a small group of three families, including star anise and its relatives, diverged from the ancestor of all the other flowering plants.
And then next comes a rather larger group, including among it a few familiar New Zealand plants like hutu (Ascarina lucida), horopito (Pseudowintera colorata, pictured at right), tawa (Beilschmiedia tawa) and pukatea (Laurelia novae-zelandiae); magnolias and a lot of tropical spices like nutmeg, cinnamon, and camphor belong in this group too.
Those families, from Amborella to Magnolia, are the basal angiosperms, a remarkable and divergent set of plants that have all kinds of interesting and useful features. But all they have in common is that they’re not monocots (they have 2 cotyledons, leaf stalks, taproots, and net-veined leaves), nor are they eudicots (they have single pores or grooves in their pollen). They’re not considered a natural group.
In the 17th and 18th centuries, the spices provided by these plants, particularly pepper, nutmeg and cinnamon, drove European exploration and conquest of tropical Asia. It was said that if a British ship could get to SE Asia and bring back a load of spices without being captured by the Dutch who controlled much of the trade, every member of the crew, from the captain down to the lowliest cabin boy, would be rich beyond the dreams of Croesus. Even today, spices are high value products, in terms of monetary value in proportion to weight.
Why does any of this matter? Well, if we can classify related species together in groups, we can make predictions based on those classifications. This helps in all sorts of economic botany questions, like drug discovery. It’s also useful to ecologists and evolutionary biologists; if we can classify plants according to their relationship history, we can understand how they evolved and how they have adapted to different environments. This is a continuation and an application of the questions posed by Charles Darwin: “… from so simple a beginning endless forms most beautiful and most wonderful have been, and are being, evolved.”
Soltis, D. E. & 27 other authors (2011). Angiosperm phylogeny: 17 genes, 640 taxa. American Journal of Botany 98: 704—730.