16th May 2016: APG - Classification by Consensus

Published on 16th May 2016

Kew scientist Mike Fay discusses the issues with classifying flowering plant species, and the efforts made by the Angiosperm Phylogeny Group (APG) to reclassify species using technological advances in DNA sequencing.

Classifications of plants were historically compiled by one or a few people, who often disagreed with each other on which characters should be emphasised. Technological improvements in DNA sequencing in the 1990s allowed botanists to break away from this tradition and led to the APG system, the fourth iteration of which has just been published.

Linnaeus’ classification system

When Linnaeus classified the flowering plants (angiosperms) in his book Species Plantarum (1753), he used the numbers of male and female parts in flowers as the main characters. He realised that this inevitably led to an artificial system, with unrelated plants being put together in many cases, and many of his contemporaries were shocked by the focus on sexual organs.

Linnaeus’ sexual system by Ehret
Linnaeus’ sexual system, as represented by G. D. Ehret

Over the next two and a half centuries, many botanists endeavoured to come up with more natural systems, using more characters, with the intention of recognising groups that more closely reflected relationships. However, even readily recognisable groups such as orchids or legumes were treated as one family in some classifications and as several families in others.

Following important works on plant classification by members of the de Candolle family, de Jussieu and others, Kew botanists, George Bentham and Joseph Hooker, developed a system of classification in the 19th century that was in use in Kew’s Herbarium and elsewhere until a few years ago. In the 20th century, Cronquist (in the USA) and Takhtajan (in Russia) developed widely used systems.

DNA comes into the picture

Nelumbo nucifera
The sacred lotus (Nelumbo nucifera), previously thought to be related to the true waterlilies, but now known to be more closely related to plane trees and banksias © Maarten Christenhusz

In the early 1990s, the first large analyses of flowering plants based on DNA sequences were published. These had become possible due to major developments in DNA sequencing technology and computing power in the late 20th century. Flowering plants were the first major group on which large groups of scientists collaborated in comprehensive analyses of this type, collecting sequences for the same genes, so that the data could be combined.

In 1993, a landmark paper with an analysis of 500 flowering plants was published by Mark Chase and 41 co-authors, the year after Mark moved from the University of North Carolina to Kew. This paper was based on sequences of one of the major genes involved in photosynthesis, and the analysis involved the botanists working with the computer programmers because the program had to be rewritten to allow them to analyse such a large data set.

The resulting family tree of relationships was an interesting mix of the expected and the unexpected. The monocots (grasses, lilies, orchids etc.) appeared as a group, for example, but the dicots (magnolias, laurels, roses, daisies etc.) did not, and some families that had never previously been thought to be related appeared close to each other. One example of this was so unexpected that fresh samples of the plants were collected and the gene was re-sequenced because people just could not believe the result! The original analysis placed the sacred lotus (Nelumbo) close to plane trees (Platanus) and banksias and their relatives (Proteaceae), and the lack of obvious shared characters made the scientists nervous about the accuracy of the data. However, the new sequences gave the same result and so did analyses of other DNA regions.

On the basis of a single gene, no-one was prepared to develop a new system of classification, but over the next few years, equivalent data sets for other DNA regions were collected. All told a very similar story, and some of the scientists involved felt confident enough to work together on a new classification based largely on DNA sequences. A major difference between the resulting paper and all previous classifications was that it was a broadly collaborative effort, with three compilers and 26 contributors from five countries. Rather than name it after the compilers, the decision was taken to use the name of the larger group, and the new system was named the Angiosperm Phylogeny Group classification, APG for short. It was published in 1998 and received a lot of media attention, including a half-page spread and an editorial in The Independent newspaper, in which it was suggested that Linnaeus would be turning in his grave!

Development of the APG system

Petenaea cordata
Petenaea cordata from northern Guatemala and southern Mexico, previously thought to be a member of Tiliaceae or Elaeocarpaceae, but now treated as the sole member of Petenaeaceae © Maria Vorontsova

Three further versions of the APG classification were published in 2003, 2009 and 2016, each with multiple compilers and contributors. The most recent version, APG IV published this month, has ten compilers and 15 contributors from six countries and is the result of a workshop hosted by Kew in September 2015. An online survey also provided useful background about the views of botanists and other users of the classification around the world (Christenhusz et al., 2015).

Each version of APG has included a list of families and/or genera of uncertain position – in 1998, 25 families fell in this category, but by 2016 the list had shrunk to seven genera, demonstrating the increase in knowledge as more plants have been sampled and more DNA regions have been investigated. The number of higher-level groups, formal and informal, has also increased, as a reflection of increasing confidence in the relationships of families to each other. Changes in APG IV on the basis of published studies include placement of Petenaea in its own family (Petenaeaceae), recognition of Kewaceae for the genus Kewa (previously included in Hypertelis in Molluginaceae) and reorganisation of some families that were known to be problematic. For example, several genera had been moved from Icacinaceae into the previously monogeneric Metteniusaceae.

Many botanic gardens and herbaria (including Kew) have adopted the APG system (e.g. Wearn et al., 2013) - APG IV will not be the final version, but the decreasing number of changes means that reorganising collections to reflect changes is becoming less arduous. The resulting system allows greater predictability than previous classifications because the groups that are recognised reflect evolutionary relationships of the flowering plants.

Dr Michael F Fay, Editor, Botanical Journal of the Linnean SocietySenior Research Leader, Conservation Genetics, Royal Botanic Gardens, Kew.

Kewa acida
The St Helenan salad plant, Kewa acida, previously Hypertelis acida (Molluginaceae), but now treated as one of eight species of Kewa, the only genus in Kewaceae © Maarten Christenhusz


Angiosperm Phylogeny Group (1998). An ordinal classification for the families of flowering plants. Annals of Missouri Botanical Garden 85: 531-553

APG II. (2003). An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants. Botanical Journal of the Linnean Society 141: 399-436 

APG III. (2009). An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG III. Botanical Journal of the Linnean Society 161: 105-121 

APG IV. (2016). An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG IV. Botanical Journal of the Linnean Society 181: 120 

Byng, J. W. (2014). The Flowering Plants Handbook: a Practical Guide to Families and Genera of the World. Plant Gateway, Hertford.

Byng, J. W., Bernardini, B., Joseph, J. A., Chase, M. W. & Utteridge, T. M. A. (2014). Phylogenetic relationships of Icacinaceae focusing on the vining genera. Botanical Journal of the Linnean Society 176: 277–294

Chase, M. W., Soltis, D. E., Olmstead, R. G., Morgan, D., Les, D. H., Mishler, B. D., Duvall, M. R., Price, R. A., Hills, H. G., Qiu, Y. L., Kron, K. A., Rettig, J. H., Conti, E., Palmer, J. D., Manhart, J. R., Sytsma, K. J., Michael, H. J., Kress, W. J., Karol, K. G., Clark, W. D., Hedrén, M., Gaut, B. S., Jansen, R. K., Kim, Y.J., Wimpee, C. F., Smith, J. F., Furnier, G. R., Strauss, S. H., Xiang, Q. Y., Plunkett, G. M., Soltis, P. S., Swensen, S. M., Williams, S. E., Gadek, P. A., Quinn, C. J., Eguiarte, L. E., Golenberg, E., Learn, G. H. Jr, Graham, S. W., Barrett, S. C. H., Dayanandan, S. & Albert, V. A. (1993). Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcLAnnals of the Missouri Botanical Garden 80: 528–580

Christenhusz, M. J. M., Brockington, S. F., Christin, P.-A. & Sage, R.F. (2014). On the disintegration of Molluginaceae: a new genus and family (Kewa, Kewaceae) segregated from Hypertelis, and placement of Macarthuria in Macarthuriaceae. Phytotaxa 181: 238–242

Christenhusz, M. J. M., Fay, M. F., Clarkson, J. J., Gasson, P., Morales Can, J., Jiménez Barrios, J. B., Chase, M. W. (2010). Petenaeaceae, a new angiosperm family in Huerteales with a distant relationship to Gerrardina(Gerrardinaceae). Botanical Journal of the Linnean Society 164: 16–25

Christenhusz, M. J. M., Vorontsova, M. S., Fay, M. F. & Chase, M. W. (2015). Results from an online survey of family delimitation in angiosperms and ferns: recommendations to the Angiosperm Phylogeny Group for thorny problems in plant classification. Botanical Journal of the Linnean Society 178: 501-528 

Linnaeus, C. (1753). Species Plantarum. L. Salvius, Stockholm.

Wearn, J. A., Chase, M. W., Mabberley, D. J. & Couch, C. (2013). Utilizing a phylogenetic plant classification for systematic arrangements in botanic gardens and herbaria. Botanical Journal of the Linnean Society 172: 127–141