Flora of North America species comparison

Arabis alpina, the generic type, is variable. It is cultivated extensively for its attractive flowers, and it sometimes escapes from cultivation. There is disagreement as to whether one or more species should be recognized in this complex. European, some African, and most North American plants are recognized as A. alpina; most of the larger-flowered, southwestern Asian plants, which are most commonly cultivated, are recognized as A. caucasica or as A. alpina subsp. caucasica (Willdenow) Briquet. In my opinion, the morphological differences between the two (see key, couplet 3) support recognition of two species, as did R. C. Rollins (1993) and G. A. Mulligan (1996).

(Discussion copyrighted by Flora of North America; reprinted with permission.)

Genera ca. 338, species ca. 3780 (97 genera, 744 species in the flora).

Of the 634 species of Brassicaceae (mustards or crucifers) native in the flora area, 616 (418 endemic) grow in the United States, 140 (12 endemic) in Canada, and 31 (1 endemic) in Greenland.

The latest comprehensive account of the Brassicaceae for North America (R. C. Rollins 1993) included Mexico and Central America and excluded Greenland. In that account, 667 native species were recognized for the continent; I place 37 of those in the synonymy of other species. Of the remaining 630 species, 111 are restricted to Mexico and Central America, and 519 are native to the flora area. This last number falls 114 species short of the 634 native species that I recognize in the flora area. Since Rollins’s account, 50 species were added to the flora in the past 15 years. Of these, 35 species were described as new, ten were added as native but previously overlooked or misidentified, and five have since become naturalized. Additionally, 72 species recognized in this treatment were treated by Rollins as either synonyms or infraspecific taxa of other species. The generic placement of 158 species in this account differs drastically from that in Rollins, though most of the changes involve the transfer of most of his species of Arabis to Boechera (59 spp.) and of Lesquerella to Physaria (54 spp.). The generic circumscriptions adopted herein are fully compatible with the rapidly accumulating wealth of molecular data, and all genera recognized here are monophyletic. Some examples demonstrate the differences between the two treatments. Arabis, in the sense of Rollins, included 80 species and 64 varieties; in this account, those 144 taxa are assigned to six genera in five tribes: Arabidopsis (2 spp.; tribe Camelineae), Arabis (16 spp.; tribe Arabideae), Boechera (109 spp.; tribe Boechereae), Pennellia (2 spp.; tribe Halimolobeae), Streptanthus (1 sp.; tribe Thelypodieae), and Turritis (1 sp.; tribe Camelineae). A similar division involves Thlaspi, a genus recognized by Rollins to include nine species, of which two are retained here in Thlaspi (tribe Thlaspideae), one is placed in Microthlaspi, and three in Noccaea (both in tribe Noccaeeae), two are reduced to synonymy of the latter genus, and one species of Noccaea is endemic to Mexico. Lepidium in this treatment includes Rollins’s Cardaria, Coronopus, and Stroganowia; Hesperidanthus includes his Caulostramina, Glaucocarpum, and Schoenocrambe (excluding its type).

The Brassicaceae include important crop plants that are grown as vegetables (e.g., Brassica, Eruca, Lepidium, Nasturtium, Raphanus) and condiments (Armoracia, Brassica, Eutrema, Sinapis). Vegetable oils of some species of Brassica, including B. napus (canola), probably rank first in terms of the world’s tonnage production. The Eurasian weed Arabidopsis thaliana (thale or mouse-ear cress) has become the model organism in experimental and molecular biology. The family also includes ornamentals in the genera Aethionema, Alyssum, Arabis, Aubrieta, Aurinia, Erysimum, Hesperis, Iberis, Lobularia, Lunaria, Malcolmia, and Matthiola. Finally, the flora includes 106 species of weeds from southwest Asia and Europe (R. C. Rollins and I. A. Al-Shehbaz 1986), of which 11 species of Lepidium have become noxious weeds in western North America.

The Brassicaceae have been regarded as a natural group for over 250 years, beginning with their treatment by Linnaeus in 1753 as the “Klass” Tetradynamia. More recently and based on a limited sampling of genera, W. S. Judd et al. (1994) recommended that the Brassicaceae and Capparaceae (including Cleomaceae) be united into one family, Brassicaceae. Molecular studies (J. C. Hall et al. 2002) suggested that three closely related families be recognized, with Brassicaceae sister to Cleomaceae, and both sister to Capparaceae. All three families have consistently been placed in one order (e.g., Capparales or Brassicales) by A. Cronquist (1988), A. L. Takhtajan (1997), and J. E. Rodman et al. (1996, 1998), as well as by the Angiosperm Phylogeny Group (APG) (http://www.mobot.org/MOBOT/research/APweb/). Brassicales includes families uniquely containing glucosinolates (mustard-oil glucosides), myrosin cells, racemose inflorescences, superior ovaries, often-clawed petals, and a suite of other characteristics (see the APG website).

Tribal classification of Brassicaceae has been subject to controversy. O. E. Schulz’s (1936) classification has been used for over 70 years, though many botanists (e.g., E. Janchen 1942; I. A. Al-Shehbaz 1984; M. Koch et al. 1999; O. Appel and Al-Shehbaz 2003; Koch et al. 2003; M. A. Beilstein et al. 2006; Al-Shehbaz et al. 2006) amply demonstrated the artificiality of that system. Schulz divided the family into 19 tribes and 30 subtribes based on characters (e.g., fruit length-to-width ratio, compression, dehiscence; cotyledonary position; sepal orientation) that exhibit tremendous convergence throughout the family. Of these, only the tribe Brassiceae was previously shown to be monophyletic.

Several molecular studies (e.g., R. A. Price et al. 1994; J. C. Hall et al. 2002; M. Koch 2003; T. Mitchell-Olds et al. 2005; C. D. Bailey et al. 2006; M. A. Beilstein et al. 2006, 2008) have demonstrated that the Brassicaceae are split into two major clades: the Mediterranean-Southwest Asian Aethionema and its sister clade that includes the rest of the family. Although Beilstein et al. showed that the family, excluding Aethionema, is divided into three major clades, such subdivision was based on only ca. 30% of the total number of genera. These three major clades still hold when nearly all genera of the family are investigated (S. I. Warwick et al., unpubl.).

Tribal assignments in the flora area are based on critical evaluation of morphology in connection with all published molecular data. To date, about 230 of the 338 genera of the family are placed in 35 tribes, including all large genera, which account for over 70% of the total species. Most of the remaining 108 genera would likely be assigned to the 35 tribes, be placed in new, smaller tribes, or be reduced to synonymy of larger genera. The delimitation of tribes for the flora area follows I. A. Al-Shehbaz et al. (2006), Al-Shehbaz and S. I. Warwick (2007), and D. A. German and Al-Shehbaz (2008) and differs from that of O. E. Schulz (1936) and the subsequent adjustments proposed by E. Janchen (1942) and Al-Shehbaz (1984, 1985, 1985b, 1986, 1987, 1988, 1988b, 1988c). Some of the tribes (e.g., Brassiceae and Lepidieae) are easily distinguished by relatively few characters; others (e.g., Arabideae, Camelineae, and Thelypodieae) are more difficult to separate unless a larger suite of characters is used. Because of the incomplete molecular knowledge on all genera of the family, the tribes, their genera, and species are listed herein alphabetically. Both R. C. Rollins (1993) and O. Appel and Al-Shehbaz (2003) arranged the genera alphabetically throughout, and the only difference in this account is the placement together of closely related genera within well-established monophyletic tribes.

Morphological data alone are sometimes unreliable in establishing phylogenetic relationships within Brassicaceae. Convergence is common throughout the family, and almost all morphological characters, especially of the fruits and embryos, which are quite heavily utilized in the delimitation of the genera and tribes, evolved independently. For example, rare character states, such as the spirolobal cotyledons, are known in at least three genera of three tribes (Bunias, Buniadeae; Erucaria, Brassiceae; Heliophila, Heliophileae), and lianas evolved independently in the South American Cremolobus (Cremolobeae), the South African Heliophila, and the Australian Lepidium (Lepidieae). Similarly, the reduction of chromosome number in the family to n = 4 occurred independently in two species of the Australian Stenopetalum and in at least 11 species of the North American Physaria. Other character states (e.g., zygomorphy, apetaly, reduction of stamen number to four, connation of median filaments, etc.) also evolved independently. Reexamination of morphology in light of molecular data is essential in order to understand the role of homoplasy and the evolution of various character states.

The literature on chromosome numbers of Brassicaceae is rather extensive, and rarely is an individual work cited herein in that regard. Instead, the recently compiled cytological data for the entire family (S. I. Warwick and I. A. Al-Shehbaz 2006) are consulted for all species.

Because the size of ovules is relatively small, it is very difficult to determine the number of ovules per ovary. The number of ovules per ovary is based on the sum of mature seeds and aborted ovules in the fruit. The length of style and type of stigma are also taken from the fruits, and the length of fruiting pedicels is measured from several proximal pedicles of the infructescence. Elevation ranges are normally given for a taxon; unfortunately, the range is not known for some taxa.

Generic delimitation in Brassicaceae is often difficult because most genera are distinguished primarily by fruit characters. The following artificial keys emphasize either flowering or fruiting characters, and the most reliable identification of a given plant to a genus can be achieved when specimens have both flowers and fruits, and when both keys are successfully used to identify it to the same genus. The keys are based on species rather than generic descriptions so that all of the morphological manifestations in a given genus are covered and, therefore, a genus may appear multiple times within one of the first four key groups. For example, genera with highly diversified vegetative and floral morphology (e.g., Cardamine, Caulanthus, Lepidium, and Streptanthus) appear in the keys to groups multiple times. Because of such coverage, keys to flowering material incorporate characteristics of a species, or groups of species, rather than of genera. Leads marked (¤) in keys for groups 1–4 indicate that mature fruits and seeds are needed for the identification of genera in their subordinate couplet(s).

(Discussion copyrighted by Flora of North America; reprinted with permission.)