The first family-wide classification of de Candolle (1821a), which is based on embryo types, fruit length/width ratio, and dehiscence, had a major influence on all subsequent tribal subdivisions of the Brassicaceae. Schulz’s (1936) tribal classification, and its subsequent minor modifications by Janchen (1942) and Al-Shehbaz (1984), was followed by all students of the family until about three years ago. In all previous classification systems, the tribes were delimited on the basis of one or a few characters, especially those of the embryo and fruit type—structures that are subject to considerable convergence. As a result, almost all of Schulz’s tribes that included more than one genus were artificially delimited.

Beilstein et al. (2006) sampled over 100 genera for the chloroplast gene ndhF and constructed the first comprehensive phylogeny of the family in which they recognized three major clades in a polytomy sister to the genus Aethionema. Based on that phylogeny, Al-Shehbaz et al. (2006) proposed a new phylogenetic tribal classification of the Brassicaceae. They recognized 25 tribes, of which only nine (Alysseae, Arabideae, Brassiceae, Euclidieae, Heliophileae, Hesperideae, Lepidieae, Schizopetaleae, and Sisymbrieae) were also treated by Schulz (1936) who recognized 19 tribes and 30 subtribes. The majority of the 25 tribes of Al-Shehbaz et al. (2006) were shown to be monophyletic based on phylogenetic studies using other markers, including mitochondrial nad4 (Franzke et al., 2009), the chloroplast trnL-F (Koch et al., 2007), and the nuclear markers ITS (Bailey et al., 2006) and PHYA (Beilstein et al., 2008). All molecular studies have clearly demonstrated the artificiality of Schulz’s (1936) tribal classification. For example, the monophyly of Lepidieae, a tribe that Schulz defined solely on the presence of angustiseptate fruits (flattened perpendicular to the septum), was rejected by statistical analysis (Beilstein et al., 2006), and genera from this tribe are now assigned to 13 different tribes in the new classification of Al-Shehbaz et al. (2006).

Warwick et al. (2007, 2008) and German et al. (2009) expanded the sampling of the tribes Alysseae, Anchonieae, Euclidieae, and Camelineae sensu Al-Shehbaz et al. (2006) and demonstrated that the first three tribes were paraphyletic and the last was polyphyletic. Monophyly in all except the last tribe was restored by the re-circumscriptions of their limits and the establishment of several additional new tribes (Al-Shehbaz and Warwick, 2007; German and Al-Shehbaz, 2008a; Warwick et al., 2009; German, 2009). The 34 tribes presently recognized (Koch and Al-Shehbaz, 2009) include well over 90% of the genera and species of the family. However, it is likely several smaller tribes will be added when unstudied genera are sampled.

As indicated by Koch et al. (2007) phylogenetic constructions based on one marker are of a limited value. Substantial segments, rather single genes, of the entire genomes of representative species of the largest tribes need to be sequenced to establish more robust phylogenies. Despite the use of multigene phylogenies from the chloroplast, nucleus, and mitochondria, there is no resolution in the skeletal backbone of the family. Koch and Al-Shehbaz (2009) suggested that such lack of resolution was probably the result of either rapid early radiation events characterized by low levels of genetic variation among the different lineages, or perhaps the product of reticulate evolution lead to conflicting gene trees that did not reflect species phylogenies. The data presented by Koch et al. (2007) and Franzke et al. (2009) favor the first view.

The following alphabetical listing of the 34 tribes includes the bibliographical citations, type genus, brief descriptions, names of component genera, and geographical centers of distribution. For detailed accounts of the tribes and their synonymy, please consult Al-Shehbaz et al. (2006), Warwick and Al-Shehbaz (2007), German & Al-Shehbaz (2008a), and Warwick et al. (2009).