[For Tyge Wittrock Böcher, 1909-1983, Danish cytogeneticist who worked on subarctic flowering plants]

Ihsan A. Al-Shehbaz, Michael D. Windham

Perennials or, rarely, biennials; (sexual or apomictic, caudex usually present, rarely absent); not scapose; usually glabrous or pubescent, rarely hirsute or hispid, trichomes simple or branched, 2-14-rayed, sometimes dendritic, not stellate. Stems erect, ascending, or decumbent, unbranched or branched distally. Leaves basal and cauline; petiolate or sessile; basal rosulate or not, petiolate, blade margins usually entire or dentate, rarely lyrate-pinnatifid; cauline usually sessile, rarely shortly petiolate, blade (base sometimes auriculate or sagittate), margins entire or dentate. Racemes (sometimes paniculate), often elongated in fruit. Fruiting pedicels erect, ascending, divaricate, or reflexed (secund or not), slender. Flowers: sepals ovate or oblong, (lateral pair slightly saccate or not basally, margins membranous); petals usually white, pink, lavender, or purple, rarely yellowish, red, or magenta, spatulate or oblanceolate, (claw shorter than sepals or undifferentiated from blade, apex obtuse); stamens tetradynamous; filaments not dilated basally; anthers ovate or oblong, (apex obtuse), [pollen ellipsoid (sexual plants) or spheroid (apomictic)]; nectar glands confluent, subtending bases of stamens, lateral glands semi-annular or annular. Fruits usually sessile, rarely shortly stipitate, usually linear, rarely oblong or lanceolate, straight or falcate, smooth or torulose; valves (papery), each with obscure or prominent midvein, usually glabrous, rarely pubescent; replum (visible), rounded; septum complete, (membranous, veinless); ovules 8-250 per ovary; (style sometimes obsolete); stigma capitate. Seeds usually uniseriate or sub-biseriate, rarely biseriate, flattened, winged, not winged, or margined, oblong or orbicular; seed coat (usually smooth or minutely reticulate, rarely papillate), not mucilaginous when wetted; cotyledons accumbent.

Species 111+ (109 in the flora): North America, n Mexico, e Asia (Russian Far East).

Boechera falcata (Turczaninow) Al-Shehbaz is known from eastern Asia (Russian Far East).

Boechera often is treated as a synonym of Arabis (e.g., R. C. Rollins 1993; S. L. Welsh et al. 2003) but it has become clear that morphological similarities between these groups are due to evolutionary convergence, not shared ancestry. Molecular analyses by M. Koch et al. (2001) and T. Mitchell-Olds et al. (2005) revealed that Arabis and Boechera belong to distantly related clades of Brassicaceae that diverged some 19-25 million years ago. A new tribal classification of the family (I. A. Al-Shehbaz et al. 2006) places them in different tribes (Arabideae and Boechereae, respectively), reflecting their substantial molecular divergence.

The taxonomic complexity of Arabis, in the broad sense, is legendary (R. C. Rollins 1941, 1993; G. A. Mulligan 1996). When the genus is split, most of the problematic taxa come to reside in Boechera. A rare confluence of hybridization, apomixis, and polyploidy makes this one of the most difficult genera in the North American flora. The sexual diploid species are relatively distinct from one another, but they hybridize wherever they come into contact. Through apomixis and polyploidy, the hybrids become stable, self-propagating lineages. Most of the hybrid derivatives in Boechera are triploids, but apomictic diploids are known as well. Thus, for any pair of sexual diploid species (e.g., AA and BB), this process can yield different intermediates, including AB apomicts and both possible apomictic triploids (AAB and ABB). The situation becomes even more challenging when a third sexual diploid enters the picture. To date, we have identified three taxa (B. divaricarpa, B. pinetorum, B. tularensis) that appear to be trigenomic triploids. Under these circumstances, even the most distinctive sexual diploid progenitors can become lost in a seemingly continuous range of morphological variability.

In a genus characterized by the presence of polyploids and apomicts, it is essential to know which taxa represent the products of primary, divergent evolution (i.e., sexual diploids) and which are the result of secondary, reticulate evolution. Fortunately, a strong correlation between pollen morphology and ploidy level/reproductive mode facilitates the separation of sexual diploids from polyploids and apomicts in Boechera. Because of differences in meiosis, sexual diploids produce small (13-16 µm diam.), ellipsoid pollen grains with symmetrical colpi. In apomictic individuals, the pollen grains are significantly larger (20-30 µm diam.), and spheroid with asymmetrical colpi. The differences in pollen size and shape are so pronounced that the ploidy level and reproductive mode of most plants with flowers can be determined using a medium power (40×) dissecting microscope (see Fig. 1 in M. D. Windham and I. A. Al-Shehbaz 2006).

To facilitate the study of ploidy level and reproductive mode in Boechera and to allow direct comparison of taxa named by previous authors, we assembled at the Missouri Botanical Garden the holotypes and isotypes of all taxa originally described in Arabis and currently placed in Boechera (over 160 published basionyms). In addition to the types, another 12,000 specimens were examined to document morphological variability and geographic distribution. During this process, we identified additional morphological features (e.g., trichome branching patterns, number of ovules per ovary, pollen and seed morphology) overlooked or underused by previous authors. The result is a substantially revised taxonomy for the genus, the nomenclatural foundation for which was established in a series of papers (M. D. Windham and I. A. Al-Shehbaz 2006, 2007, 2007b).

In many cases, the species circumscriptions adopted here deviate significantly from those of previous authors. Where R. C. Rollins (1993) accepted 63 species with varieties, we recognize 109 species and two with two subspecies. Our treatment includes a total of 71 sexual species. These represent the morphological extremes of the complex and are often easily distinguished (when separated from the apomicts using pollen characters). Although it is likely that some diploid species remain to be discovered, we feel that this portion of the treatment is relatively complete.

Our coverage of the apomictic hybrids is much less comprehensive. There are literally hundreds of hybrids in Boechera with unique genomic combinations, each of which could be recognized at species level. Our treatment includes just 38 apomictic species, primarily taxa recognized at some level by other authors. Because some hybrid combinations are not formally recognized, it is inevitable that some names will be misapplied to superficially similar hybrids of different parentage. For example, plants of B. goodrichii (= B. retrofracta × B. gracilipes), B. consanguinea (= B. retrofracta × B. fendleri), and B. pinetorum (= B. retrofracta × B. rectissima × B. sparsiflora) are sufficiently similar that they might be considered a single taxon if their respective parentages and disjunct geographic ranges were not taken into account. Indeed, all three have been called Arabis holboellii var. pinetorum (e.g., R. C. Rollins 1993; S. L. Welsh et al. 2003), although the epithet consanguinea has priority.

The best way to avoid such misidentification is to pay close attention to the geographic distribution of apomictic taxa and their sexual progenitors. Apomictic hybrids in Boechera appear to be of relatively recent origin and generally have not migrated beyond regions where their parents are sympatric. Thus, users of this treatment should be wary of major range extensions for apomictic taxa; in most cases, these will turn out to be unique hybrid combinations not represented in the keys or descriptions. Because the use of hybrid binomials is potentially misleading, the best approach to identifying a hybrid is to provide a formula name based on the hypothesized parentage (e.g., B. fendleri × B. stricta or B. fendleri hybrid). This requires an accurate understanding of the sexual diploids occurring in the region of interest, which we hope the following keys and descriptions will provide.

Given the inherent taxonomic complexity of Boechera, it has been necessary to incorporate micromorphological characters such as pollen morphology and trichome branching patterns in the identification keys. Whenever possible, we have restricted such characters to later couplets, but microscopic observations are required to distinguish some species. Effective use of the keys also depends on having complete specimens bearing both flowers and fruits. In all cases, measurements of stem length are taken from fruiting plants, those of basal leaves from the largest in the basal rosette, for the fruiting pedicels from the longest in the infructescence, and for the stem trichomes from the largest near the base. Descriptions of the pedicels, flowers, and fruits are taken from the main inflorescence rather its lateral branches, the number of seed rows per locule is determined near the middle of the fruit, and the number of ovules is observed in mature fruits by counting the number of seeds plus the abortive ovules.

SELECTED REFERENCES Al-Shehbaz, I. A. 2003b. Transfer of most North American species of Arabis to Boechera (Brassicaceae). Novon 13: 381-391. Windham, M. D. and I. A. Al-Shehbaz. 2006. New and noteworthy species of Boechera (Brassicaceae) I: Sexual diploids. Harvard Pap. Bot. 11: 61-88. Windham, M. D. and I. A. Al-Shehbaz. 2007. New and noteworthy species of Boechera (Brassicaceae) II: Apomictic hybrids. Harvard Pap. Bot. 11: 257-274. Windham, M. D. and I. A. Al-Shehbaz. 2007b. New and noteworthy species of Boechera (Brassicaceae) III: Additional sexual diploids and apomictic hybrids. Harvard Pap. Bot. 12: 235-257.