Boykinia purpusii Brandegee, Bot. Gaz. 27: 447. 1899. Therofon purpusii (Brandegee) A. A. Heller, Cat. N. Amer. P1., ed. 2. 5. 1900. Sullivantia purpusii (Brandegee) C. 0. Rosend., Minnesota Stud. P1. Sci. 1(6): 407. 1927. TYPE: U.S.A. CoLoRADo: "On moist rocks in Black canyon of the Uncompahgre River, Colorado, at an elevation of 7200 ft.," Purpus 512 (HOLOTYPE: UC!, MIN- photograph!; isoTYPE: MIN-fragment from holotype!).
Moist canyon walls, wet cliffs (occasionally at high altitudes) and in close prox- imity to waterfalls. Typically (if not exclusively) occurring on metamorphic sub- strata; the water in suitable habitats often appears to be highly charged with calcium carbonate. All collections are confined to three counties (Garfield, Gunnison, and Montrose) in west-central Colorado. Variety purpusii is poorly rep- resented in major herbaria. Based on the field observations of Kaemmerer (pers. comm.), many of the populations from Garfield Co. are located in an area con- taining the largest known deposits of shale oil in the United States. The majority of collections from Montrose and Gunnison counties are located in the Black Canyon of the Gunnison. Although the number of populations present in the Black Canyon is not known, it probably is quite small. Flowering occurs from June through August.
Based on the collection data given by Purpus for the type specimen, the col- lection was made in the "Black Canyon, Uncompahgre River." The Black Canyon however, is considered to be an area along the nearby Gunnison River, and not the Uncompahgre. Subsequent to the type collection, no collections of Sullivantia have been made from the Uncompahgre River, whereas several collections (in- cluding my own) have been made from Black Canyon of the Gunnison River. The collection data given on the type specimen may be incorrect; the type was probably collected in the Black Canyon of the Gunnison River. The precise year of collection of the type is not given on the label.
Sullivantia hapemanii var. purpusii has been poorly understood taxonomically in large part because of the paucity of collections representing it. Rosendahl, for example, did not have fruiting specimens of this taxon. Variety hapemanii and var. purpusii are separable only on the basis of cryptic characters of the gynoecium and mature capsule. Living material or mature capsules are therefore needed to differentiate the two taxa. Herbarium specimens are often difficult to use because the gynoecium characters are poorly preserved in pressed flowers. This morpho- logical similarity was noted by Rydberg (1922) and Coulter and Nelson (1909), who considered collections from Colorado to be conspecific with collections from Wyoming. Despite their substantial morphological similarity, var. hapemanii and var. purpusii are clearly distinct chemically. Additional collections may morpho- logically and/or chemically bridge the gap between these two varieties, at which point it would be desirable to place var. purpusii in the synonymy of S. hapemanii.
Specimens examined: U.S.A. Colorado. Garfield Co.: Kaemmerer 157 (COLO); Soltis & Hammond- Soltis 1040, 1041 (both COLO, IND); Weber et al. 15094 (COLO). Gunnison Co.: Hall 538, 574 (both MO); Higgins 926 (COLO, RM); Langenheim 1120 (MIN); Langenheim s.n., in 1950 (MIN); Softis & Hammond-Soltis 1042, 1044 (both COLO, IND); Weber 9361 (NY); Weber & Barclay 9178 (COLO, GH); Wheeler 926 (COLO, RM). Montrose Co.: Buck 141 (ILL). County unknown: Camp 10 (GH).