A protein sequence study of the dicotyledons and its relevance to the evolution of the legumes and nitrogen fixation

Abstract

The data were the N-terminal 40 amino acids of the small subunit of ribulose-1,5-bisphosphate carboxylase (Rubisco-SSU) from 310 dicotyledons covering about half the families and most orders. The strategy used to build an overall phylogenetic tree is described; a maximum parsimony program, reliable for up to 17 taxa, has been used. The first step was to use four modern phylogenies to divide 101 of the families into 25 Groups of uneven size leaving 20 families with controversial affinities. Ancestral sequences were derived for each group and these were built into an overall tree in five stages that included testing the reality of the original Groups and re-defining a third of them to include most of the families of controversial affinities. The high degree of taxonomically 'correct' clustering of taxa into families suggested reliability at that level, but reliability probably diminished at higher levels that are separated by shorter distances. The results should be regarded as working hypotheses for future investigations, not as firm conclusions. Evidence is presented that, of the three leguminous families, Mimosaceae and Papilionaceae are closest and that Caesalpiniaceae is very close to Rosaceae. Families in which nitrogen fixation is known appear to associate in three clusters comprising one, three and eight families.