When Robert Brown created the family Limnanthaceae in 1833 he declared that, "the place of this new family is not absolutely determined; but it is suggested that in two remarkable points of its structure, namely, the presence of glands subtending the alternate filaments, and the existence of a gynobase, it more nearly approaches to Hypogynous families than to Perigynous, with which it has hitherto been associated". Ultimately he demurred from assigning a systematic position to his family. Subsequent taxonomists promoted various divergent placements. Bentham and Hooker (1862) decided that given the great similarity of
Limnanthes and
Floerkea to taxa in the Geraniaceae, family status was not warranted; accordingly, they placed both genera in the tribe Limnantheae in that family. Engler and Prantl (1896), however, disagreed, noting that the position of the seed indicated the family should be placed in the Sapindales. Most recent authors of taxonomic treatises have maintained the family Limnanthaceae but placed it in the order Geraniales, including Thorne (1976) Cronquist (1988), who nevertheless states that Limnanthaceae are, "without...obvious affinities..." and Takhtajan (1980). Hutchinson (1973), proposed that Limnanthaceae should be included in Geraniales, but its similarity to Caryophyllales (which he believed is derived from Geraniales) suggested that Limnanthaceae form a link between these two groups. A number of studies over the years have noted various morphological, developmental and embryological characters at variance with the Geraniales (as well as other groups to which Limnanthaceae have been assigned). Maheshwari and Johri (1956) conducted an extensive investigation of the morphology of
Floerkea noting that, among other things, the herbaceous habit, gynobasic style, unusual type of
tetrasporic embryo sac and basal parietal placentation of the unitegmic, tenuinucellate ovules differ from the Geraniales which has among its woody to herbaceous members, (at most) lobed syncarpous gynoecia, monosporic embryo sacs, generally axile placentation and bitegmic, crassinucellate ovules. Additionally the fruit type of Limnanthaceae, a schizocarpic nutlet, is unlike anything in the Geraniales most of which produce capsules. They also found a number of key differences between Limnanthaceae and Sapindales, and concluded that Limnanthaceae should be given their own order. Hofmann and Ludewig (1985) undertook a similar detailed study of the morphology of
Limnanthes douglasii and made systematic inferences from their findings. They concluded that there is nothing in the morphology to suggest a relationship to the Coriariaceae (suggested by Chatin in 1856), Geraniales or the Sapindales (
sensu Cronquist), though they allow that the, "level in evolution is about the same...." They also rejected any meaningful similarity between Limnanthaceae and Tropaeolaceae, as suggested by Dahlgren, who noted the shared possession of certain phytochemicals (glucosinolate, myrosinase, erucic acid and eicosenoic acid). They rejected this connection on the grounds that these phytochemicals also occur in the Brassicaceae which is assumed to be unrelated and suggest that the secondary compounds must be convergent. They concluded that the systematic position of Limnanthaceae is uncertain and cannot be presently determined. Dahlgren (1975), who believed strongly in the systematic value of secondary chemicals, placed Limnanthaceae in Capparales with Brassicaceae. He later (1980) placed Limnanthaceae together with Tropaeolaceae in the order Tropaeolales. Buchner, Halbritter, Prundner and Hesse (1990) investigated the pollen morphology of Limnanthaceae and discovered that the zonosulcate morphology is unlike the pollen of any known angiosperm and therefore relationships cannot be inferred from it. They reiterated Maheshwari and Johri's suggestion that a new order, the Limnanthales, should be created to contain the family. Rodman (1991a, b) included Limnanthaceae in a twin phenetic and cladistic analysis of all 15 taxa then known to produce glucosinolates. In
UPGMA phenograms, Limnanthaceae clustered with Balsaminaceae. Similarly, cladistic analysis showed Limnanthaceae either in a clade with Balsaminaceae and sometimes Pentadiplandraceae or in a
polytomy with Balsaminaceae, Pentadiplandraceae, Caricacaeae, Centrospermae, and a clade which includes all the rest of the glucosinolate taxa (except
Drypetes). Rodman et al. (1993, 1997) assembled additional DNA sequence data sets for glucosinolate taxa and applied cladistic methods to generate hypotheses about relationship. They discovered that all the glucosinolate-producing taxa save
Drypetes, form a clade. Despite the paucity of morphological synapomorphies (and therefore, to the surprise of many taxonomists),
rbcL sequences have provided well supported evidence for the group's monophyly. 18S nuclear ribosomal DNA sequences have similarly, though less certainly, indicated a common ancestor for all the glucosinolate containing plants except
Drypetes. The Angiosperm Phylogeny Group places Limnanthaceae in Brassicales, which is in the Malvid/Rosid II lineage. ==References==