An updated classification of the Brassicaceae (Cruciferae)

Abstract Based on recent achievements in phylogenetic studies of the Brassicaceae, a novel infrafamilial classification is proposed that includes major improvements at the subfamilial and supertribal levels. Herein, the family is subdivided into two subfamilies, Aethionemoideae (subfam. nov.) and Brassicoideae. The Brassicoideae, with 57 of the 58 tribes of Brassicaceae, are further partitioned into five supertribes, including the previously recognized Brassicodae and the newly established Arabodae, Camelinodae, Heliophilodae, and Hesperodae. Additional tribus-level contributions include descriptions of the newly recognized Arabidopsideae, Asperuginoideae, Hemilophieae, Schrenkielleae, and resurrection of the Chamireae and Subularieae. Further detailed comments on 17 tribes in need of clarifications are provided.


Introduction
Rapid advances in our understanding of phylogenetic relationships among taxa are driving the development of modern classification schemes that accurately reflect current knowledge. Brassicaceae (Cruciferae) is a relatively large family, currently comprising ca. 4140 species (original data), for which various classification systems have been proposed, including influential historical classifications contributed by de Candolle (1821), Hayek (1911), Schulz (1936), and Janchen (1942). The first infrafamilial classification for the Brassicaceae based on molecular phylogenetic data, proposed by Al-Shehbaz et al. (2006), included 25 tribes but no higher taxonomic units. The phylogenetic findings available at the time were based on relatively few species (e.g., ~ 100 spp.) and lacked clarity regarding the limits and relationships among the inferred major lineages (referred as to I, II, and III by Beilstein et al. 2006). Since then, numerous additional taxa have been included in phylogenetic studies and the amount, quality, and reliability of phylogenetic data has increased tremendously. This has led to the discussion of numerous informal evolutionary lineages (Huang et al. 2016;Nikolov et al. 2019) and the recognition of more than 50 tribes (e.g., Hohmann et al. 2015;Huang et al. 2020). Hence, there is an obvious need to codify the current well-supported understanding of Brassicaceae relationships (e.g., Nikolov et al. (2019), Walden et al. (2020), and especially Hendriks et al. (2022)) into an updated classification scheme that can now include robust subfamilial and supertribal groups. Distribution. Cosmopolitan, centered in temperate regions of the Northern Hemisphere.

Subfamilial division
All phylogenetic studies over the past two and a half decades identify Aethionema W.T. Aiton as sister to all other Brassicaceae, which supports the recognition of two highly unequal subfamilies, the new unigeneric Aethionemoideae with 58 species and the much bigger Brassicoideae, comprising the other 98.6% of species and the rest of the generic and tribal diversity of the family. Description. Trichomes and multicellular glands absent. Leaves entire, articulate at base. Fruits silicles, angustiseptate, bilocular, few-seeded, dehiscent, or unilocular, one-seeded, indehiscent; sometimes both types present. Most common x = 11, 12.
Tribes. Aethionemeae Al-Shehbaz, Beilstein & E.A. Kellogg. Note. For many species of Aethionema a 3-nerved petal claw has been described (e.g., Hedge 1965). Further studies are needed to verify whether this is a feature present in all members of Aethionema and whether it is unique to the genus (and then diagnostic for the subfamily).
Description. Trichomes (simple and/or variously branched) and multicellular glands absent or present. Leaves entire to variously dissected, simple or compound, not articulate at base. Fruits various in compression, dehiscence, length to width ratio, number of seeds (one to > 100), etc. Base chromosome numbers various; the lowest x = 4.
Distribution. Same as the whole family.

Supertribal division
Brassicoideae is subdivided into the following five supertribes corresponding to the main evolutionary lineages discussed in detail by Hendriks et al. (2022). Notes. Corresponds to evolutionary lineage IV of Nikolov et al. (2019) and Hendriks et al. (2022) or lineage D of Huang et al. (2016). Limits of this supertribe are not yet fully understood due to discordance in positions of tribes and their taxa in the nuclear vs. plastid phylogenies of Hendriks et al. (2022). It might be eventually restricted to Arabideae, while Alysseae and possibly Asperuginoideae would better be recognized as a separate supertribe, Alyssodae. Proper placement of Stevenieae also needs further clarification due to its grouping within Camelinodae lineage in chloroplast phylogenies (Walden et al. 2020;Hendriks et al. 2022 Franzke et al. (2011) to become known as "expanded lineage II", or lineage B of Huang et al. (2016). Cochlearieae reveals relationship with Brassicodae in nuclear-based phylogeny, though it groups with "rogue" tribes of Heliophilodae in plastid trees (details in Hendriks et al. 2022). Its supertribal assignment is therefore yet unclear.  Hendriks et al. (2022). Anastaticeae, Biscutelleae, Hillielleae, Iberideae, and Megacarpaeeae are tentatively assigned to Heliophilodae due to their partially resolved phylogenetic position (grouping with others only in nuclear-based trees; see Hendriks et al. (2022) for details and discussion). Eventually, these five tribes may be recognized as a separate supertribe, e.g., Anastaticodae, based on the most speciose tribe among them. In the latter case, Heliophilodae would become unique among supertribes being almost completely restricted to the Southern Hemisphere. Description. Trichomes usually present, simple and/or branched; multicellular glands often present. Leaves normally little divided, nearly never auriculate at base. Most common x = 7.
Distribution. Native to Eurasia (predominantly temperate and dry subtropical Asia).
Note. Corresponds to evolutionary lineage III of Beilstein et al. (2006) and subsequent studies, or lineage E of Huang et al. (2016).

New tribal adjustments
Updates at the tribal level include recognition of additional six tribes, of which four are newly described and another two are resurrected. Tribal names are followed in parenthesis by numbers of genera and species. Description. Herbs, annual or perennial. Trichomes simple, mixed with stalked 1-3(or 4)-forked. Multicellular glands absent. Cauline leaves petiolate to subsessile and cuneate to attenuate at base, not auriculate. Racemes ebracteate, often elongated in fruit. Flowers actinomorphic; sepals ascending to spreading, base of lateral pair slightly saccate or not; petals white, pink, or purple; claw obscurely differentiated from blade or distinct; filaments unappendaged, wingless; pollen 3-colpate; ovules 15-80 per ovary. Fruits siliques, linear, terete or latiseptate, unsegmented; styles obsolete or to 1 mm long; stigma entire. Seeds uniseriate; cotyledons accumbent or rarely incumbent. x = 5 and 8.

3a. Tribal assignment of Arabidopsis
Distribution. Eurasia, Africa, North America. Notes. Arabidopsideae is distinguished from the Camelineae by the lack of stellate and dendritic trichomes, though both also have simple and stalked forked trichomes, by having petiolate or subsessile cauline leaves not auriculate at base, by the lack of yellow flowers, 15-80 ovules per ovary, silique fruits, and accumbent or rarely incumbent cotyledons. By contrast, the Camelineae usually have some stellate or dendritic trichomes, always sessile and auriculate to sagittate cauline leaves, usually yellow flowers, though white to pink flowers occur just as in the Arabidopsideae, 2-40 ovules per ovary, silicle or rarely silique fruits, and incumbent or rarely accumbent cotyledons.

3b. Asperuginoides
There has been no agreement among various authors about the tribal assignment of monospecific Asperuginoides Rauschert. For example, Khosravi et al. (2009) indicated a close relationship to the Cochlearieae, whereas German et al. (2009) andWarwick et al. (2010) showed no affinity to any tribe. It was listed as an unplaced genus by Al-Shehbaz (2012). More recently, Nikolov et al. (2019) and Hendriks et al. (2022) identified a sister relationship to the Alysseae, but Španiel et al. (2015) excluded it from the tribe. Furthermore, the plastome data by Walden et al. (2020) did not support that nor indicated any relationship to the 50+ tribes. Given the current data, it appears that the best solution is to place this anomalous genus in its own tribe.

3c. Chamira
Although the tribe Chamireae was first recognized by Sonder (1846) and later accepted by Schulz (1936), it has not been widely recognized since, and Chamira Thunb. was listed as unplaced in Al-Shehbaz (2012) Walden et al. (2020), and Dogan et al. (2021) that Chamira and Heliophila are closely related genera that do not belong to the same tribe, and the former has been used as the outgroup for phylogenetic and genomic studies of the latter. A tribal description comparable to that of other tribes is provided below.
Distribution. Chamira circaeoides (L. f.) Zahlbr. is endemic to the Western Cape of South Africa.

3d. Dipoma and Hemilophia
Dipoma Franch. was first studied by Warwick et al. (2010) who did not assign it to any tribe, and together with Hemilophia Franch., they were listed as unplaced in Al-Shehbaz (2012). Nikolov et al. (2019) showed the two genera form a monophyletic clade unrelated to any tribe and suggested their placement in a new tribe. However, plastome data by Walden et al. (2020) showed Dipoma to be affiliated with the Crucihimalayeae and not with Hemilophia. The results from the nuclear genome of Hendriks et al. (2022) fully agree with those of Nikolov et al. (2019), and the new tribe Hemilophieae is proposed here to accommodate both genera, leaving incongruent chloroplast and nuclear-based phylogenies. Description. Herbs rhizomatous perennials. Trichomes simple, malpighiaceous, sometime short-stalked forked. Multicellular glands absent. Cauline leaves petiolate to subsessile and cuneate to attenuate at base, not auriculate. Racemes bracteate throughout, elongated or not in fruit. Flowers actinomorphic; sepals ascending to spreading, base of lateral pair not saccate; petals white, pink, or purple; claw obscurely differentiated from blade or distinct; filaments slender or dilated at base and sometimes strongly appendaged; pollen 3-colpate; ovules 2 or 4 per ovary, apical. Fruits dehiscent silicles, oblong to ovoid, terete or slightly angustiseptate, unsegmented, wingless or with narrow wings or crests; septum complete or absent; styles distinct, cylindrical or conical; stigma entire. Seeds aseriate; cotyledons accumbent. Base numbers various.
Distribution. Endemic to China (Sichuan and Yunnan). Note. The tribe includes narrowly distributed monospecific Dipoma and Hemilophia (6 spp.). Beilstein et al. (2006Beilstein et al. ( , 2008 Warwick et al. (2010) was the first that dealt with both Idahoa and Subularia. The former was sister clade to Petrocallis W.T. Aiton and together they were sister to Subularia. That clade was sister to many taxa of various tribes. These early studies did not resolve the relationship of both genera, and Al-Shehbaz (2012) listed both genera as unplaced.

3e. Idahoa and Subularia
The Idahoa, and further studies should resolve such a relationship. Regardless of the slight differences in the most recent plastid vs. nuclear family-wide phylogenies, it is evident that these two genera should be placed in one tribe, and the name Subularieae was validly proposed over two centuries ago. Description. Herbs scapose annuals. Trichomes absent. Multicellular glands absent. All leaves in a basal rosette, sessile or petiolate, cauline leaves absent. Racemes ebracteate throughout and elongated or not in fruit, or flowers solitary on long pedicels originating from center of rosette. Flowers actinomorphic; sepals spreading or ascending, base of lateral pair not saccate; petals white, claw obscure or undifferentiated from blade; filaments slender at base; pollen 3-colpate; ovules 4-18. Fruits dehiscent, unsegmented silicles, orbicular and strongly latiseptate or obovoid to ellipsoid and slightly angustiseptate; septum complete; styles minute or absent; stigma entire. Seeds biseriate, broadly winged and accumbent, or wingless and incumbent. x = 14 and 15.
Distribution. The tribe includes monospecific Idahoa (NW USA and Canadian British Columbia) and two aquatic or littoral species of Subularia, of which S. monticola A. Braun ex Schweinf. is restricted to tropical East Africa, and S. aquatica L. is distributed in northern North America (subsp. americana G.A. Mulligan & Calder) and temperate Eurasia (subsp. aquatica).

3f. Schrenkiella
This monospecific genus was based on Diplotaxis parvula Schrenk, a species that fluctuated between unrelated genera solely on morphological grounds. It was first shown by German et al. (2009) to occupy an isolated position among Asian Brassicaceae and was subsequently recognized by German and Al-Shehbaz (2010) as a monospecific genus that was not placed in any tribe. It was shown by Huang et al. (2016) to form a basal clade to that including Sisymbrium L. and six genera of the Brassiceae. The first robust position of Schrenkiella was shown by Walden et al. (2020) and fully supported by Hendriks et al. (2022). It is sister to a clade including the Fourraeeae and sister clade including the Brassiceae and Isatideae plus Sisymbrieae and Thelypodieae. The isolated position of monophyletic Schrenkiella strongly supports its placement in its own tribe.

Further tribal comments
The following alphabetical tribal discussions are based on the phylogenies of Hendriks et al. (2022), along with comparison of the recent family-wide phylogenies of Nikolov et al. (2019), Walden et al. (2020), and few earlier ones. Generic limits and species number closely follow BrassiBase (Kiefer et al. 2014) with some updating. As above, tribal names are followed in parenthesis by numbers of genera and species, and those that showed no conflict with previous phylogenies are not discussed here. They include Anastaticeae ( Aethionemeae (1: 58). The tribe is distributed primarily in SW Asia and the Mediterranean region, with the center of greatest diversity located in Turkey, in which 23 of the 40 species are endemic. All previous molecular studies have supported the tribal position as a sister clade to the rest of the Brassicaceae recognized above at subfamilial level.
Alyssopsideae (4: 9). A small Asian tribe distributed predominantly in Afghanistan, Azerbaijan, Iran, Tajikistan, and Turkmenistan. It is monophyletic in Hendriks et al. (2022) and a sister clade to Chrysochamela and together are sister to Pseudoarabidopsis. These two genera belong to paraphyletic Camelineae III and together are sister to the Turritideae (2 spp.). The sister relationship of Pesudoarabidopsis to the Turritideae was demonstrated earlier by Walden et al. (2020) who showed that their clade is distinct from the Camelineae including the generic type Camelina. It is clear that these taxa do not belong to the Camelineae s. str. (Hendriks et al. 2022), but further studies are needed to explore whether they are well supported within Alyssopsideae.
Anchonieae (9: 75). Except for several species of Matthiola W.T. Aiton in Europe, the tribe is distributed primarily in SW and C Asia, and Africa. Only monospecific Eremoblastus Botsch. is not covered in Hendriks et al. (2022). The generic type, Anchonium DC., has recently been reduced to synonymy of the earlier-published Sterigmostemum M. Bieb. (German and Al-Shehbaz 2017). The tribe is characterized by the presence of multicellular-multiseriate glands, though apparently these structures were independently lost in Veselskya Opiz (1 sp.), one species of Sterigmostemum, and some species of Matthiola (ca. 56 spp.). Such glands are also found in the related tribes Chorisporeae and Dontostemoneae.
Arabideae (18: 559). The tribe is the largest and most complex in the family. It includes ten monospecific genera, and Draba L. (ca. 410 spp.), Arabis L. (ca. 100 spp.), and Aubrieta Adans. (23 spp.) are the most species rich ones. The tribe has been the focal topic for the Koch lab (Heidelberg University) for about three decades and despite carving nearly a dozen segregates into several tribes, Arabis still needs further focus and taxonomic adjustments are under consideration (see Koch et al. 2022 for references).
Asteae (2: 2). The findings of Hendriks et al. (2022) strongly justify merging the Mexican monospecific tribe Scoliaxoneae with the earlier published Asteae. That clade is most closely related to the South American CES clade sensu Salariato et al. (2016). These findings are in full agreement with those of Walden et al. (2020), but not closely related to the European Kernereae, a tribe more closely related to the Cochlearieae, Conringieae, and Coluteocarpeae in Hendriks et al. (2022).
Monophyly of Brassica is established based on most recent molecular phylogenies (e.g., Hendriks et al. 2022). About a dozen species of Brassica have been transferred to Guenthera Andrz., but monophyly of the latter with additional species needs to be established. Two other genera of the tribe, Diplotaxis DC. and Erucastrum C. Presl, remain artificially delimited, and similar studies are needed to accurately define their boundaries.
Conringieae (1: 3) vs. Plagiolobeae (1: 5). The Conringieae sensu Al-Shehbaz (2012) was broadly delimited to encompass a heterogenous assembly of the genera Conringia (6 spp.) and Zuvanda (3 spp.). The findings of Hendriks et al. (2022) agree with those of Walden et al. (2020) and Nikolov et al. (2019) in that the Conringieae s.l. is not monophyletic. Based on the molecular findings and re-evaluation of morphology in light of those studies, one species, C. planisiliqua, was assigned to the genus Iljinskaea (Al-Shehbaz et al. 2021) of the Isatideae, Zuvanda and three species of Conringia are currently recognized as five species of Plagioloba of the tribe Plagiolobeae (German 2021;German 2022;Khosravi et al. 2022), and the remaining three species of Corningia are retained in the genus. The Conringieae differs from the Plagiolobeae by having 4-to 8-angled (vs. terete) fruits and entire (vs. slightly to prominently 2-lobed) stigmas with connivent (or sometimes decurrent) lobes.
Descurainieae (6: 48). Except for the monospecific Patagonian Trichotolinum O.E. Schulz, which has not yet been included in any phylogenetic studies, the position of other five genera in Hendriks et al. (2022) agrees with earlier studies.
Dontostemoneae (2: 14). Position of Dontostemoneae, Chorisporeae, and their intertribal hybrid Shehbazieae are in full agreement with the initial findings by German and Friesen (2014) and Walden et al. (2020). In contrast to these consistent findings, Liu et al. (2021) probably erroneously considered Shehbazia D.A. German as member of the paraphyletic Chorisporeae.
Eudemeae ( Hendriks et al. (2022) strongly support the recognition of this tribe.
Hillielleae (1: 11). The recently established Hillielleae was previously part of the Yinshanieae, but Chen et al. (2016) clearly showed that the two tribes are distantly related. Walden et al. (2020) confirmed the findings of Chen et al. and demonstrated that the Hillielleae is sister to a clade containing the Iberideae and Megacarpaeeae but remotely related to the Biscutelleae. However, Hendriks et al. (2022) showed that the Hillielleae is sister to the clade including the last three tribes and together are sister to the Anastaticeae.
Iberideae (2: 30). The tribe includes the primarily European Iberis L. (27 spp.) and Teesdalia (3 spp.). Only Warwick et al. (2010) included Teesdalia in their studies and showed it to form a sister clade to Iberis and thus placed both genera in the tribe Iberideae. In Hendriks et al. (2022), two species of Teesdalia and one of Iberis were sampled and the results showed them to be remotely related. Clearly a better sampling of Iberis ought to be done to check whether or not the two genera can be maintained in one tribe.

Concluding remarks
The taxonomic framework presented here reflects a growing body of phylogenetic knowledge derived from continual advances in the sampling of species, broader representation of major groups, and the extensive sampling of genomic regions needed to help robustly resolve relationships across scales (Hendriks et al. 2022). The consistent nature of those findings suggest that this classification is a considerable advance over previously available formal classifications. However, we are fully aware that further accumulation of phylogenetic data will result in additions and modifications to our understanding of relationships among a minority of Brassicaceae. Most importantly, elements of phylogenetic uncertainty, illustrated by the presence of a few "jumpy clades" and discordance between nuclear and plastid phylogenies, highlight both the need to continue to resolve Brassicaceae relationships and regions of "the family tree" that are likely to experience and require future taxonomic modifications.