Dicotyledonous wood anatomy and the APG system of angiosperm classification

The recently proposed classification by the Angiosperm Phylogeny Group (APG) of angiosperms based on monophyletic groups as recognized mainly by molccular analysis is used here to discuss wood anatomical diversity patterns at the ordinal level. The APG orders are compared with the most recent ‘classical’ orders as listed in the second edition of 7 h e Plant Book for ‘improved’ or ‘deteriorated’ wood anatomical coherence. Although homoplasy in wood anatomical characters, largely due to ecoloqical adaptations, limits the value of wood anatomy at higher levels of classification, many families and orders tend to have characteristic combinations of microscopic wood features. Out of the 29 APG dicot orders, seven (Aquifoliales, Cucurbitales, Gentianales, Geraniales, Myrtales, Sapindales, Saxifragales) show an increase in wood anatomical homogeneity relative to their ‘classical’ predecessors; four APG dicot orders (Apiales, Ericales, Fabales, and Rosales) show a drcrrase, although within the orders several suprafamilial subclades are homogeneous. For the remaining 18 orders. wood anatomical diversity remains about the same as in previous classfications or the APG ordinal composition is almost identical to the ‘classical’ composition. The results support the value of both molecular markers and wood anatomical characters in phylogenetic classification. Because the ‘classical’ ordinal classifications have been partly inspired by wood anatomical information, one might have expected a ?greater wood anatomical coherence in them than in the largely molecularly delimited APG orders if wood anatomy did not provide significant phylogenetic sisgnals at higher taxonomic levels. The reverse appears to be the case. Among the wood anatomical characters included in the comparison, vestured intervessel pits, large and simple ray parenchyma pits, and sometimes also wide and tall rays appear to characterize orders. Some orders tend to be characterized by a combination of ‘primitive’ features in the Baileyan sense: scalariform perforations, fibres with distinctly bordered pits, apotracheal parenchyma, and heterocellular rays. This raises the question as to whethcr in thcsc cladcs this entire combination of characteristics should not he viewrd as synapomorphic rather than symplesiomorphic.