We evaluated the potential of the LEAFY gene, a nuclear gene, as a phylogenetic marker in the subfamily Caesalpinioideae of the Leguminoseae. Caesalpinioideae are large tropical trees which are extremely diverse in their floral morphology. The LEAFY gene has a crucial role in the transition to flowering, and in Pisum sativum, it also is responsible for the pinnate nature of the leaves. In all angiosperms examined to date the coding sequence is separated by two introns. We cloned and sequenced the 3 portion of the second exon, the second intron, and most of the third exon of the LEAFY gene for a representative sampling of species in the Caesalpinioideae and for a few other legumes. The phylogenetic analyses tell us that the gene is in a single copy. The intron is too variable to be useful for resolving relationships across the Caesalpinioideae, but the exons can be aligned and provide phylogenetic resolution. We have found a new intron inserted in the second exon, which serves as a marker for a group of genera in tribe Detarieae. The trees recovered from the phylogenetic analyses of the exons are highly resolved compared to those obtained from chloroplast sequences (trnL intron) and the groups are strongly supported. However, the results from the phylogenetic analysis of LEAFY are not congruent with those from the trnL intron sequences, and they differ from traditional hypotheses of relationships in the subfamily. What are the biological mechanisms that lead to this incongruence? Is LEAFY a poor phylogenetic marker? Possible causes for this incongruence are explored : saturation of non-synonymous substitutions in LEAFY, presence of undetected multiple copies in certain taxa, differences in evolutionary rates among lineages.

Key words: Caesalpinioideae, incongruence, LEAFY gene, Molecular phylogeny