As indicated earlier, a very peculiar Liatris is both the most distinctive and undoubtedly the rarest of the endemics. It is abundant on only a single glade and restricted to a handful of populations within an area of less than 1.6 km², the heart of the "core area" where the Ketona Glades are the most extensive and richest in rare taxa.

Liatris oligocephala J. Allison, sp. nov. TYPE: Alabama: Bibb County, ca. 14.2 km NE of Centreville. "Starblaze Glade Southwest," Ketona Dolomite outcrop ca. 0.3 km N of the Little Cahaba River, 18 Jul 1993, James R. Allison and Timothy E. Stevens 7802 (holotype, GH; isotypes: AUA, BRIT, DUKE, FLAS, FSU, GA, IBE, JSU, MICH, MO, NCU, NY, TAMU, TENN, UARK, UNA, US, USCH, VDB). Figures 8 and 9.

Species corpore achenii sine setulis a congeneris diversa.

Perennial herb, (1.2) 2.5-4.8 (5.5) dm tall from a globose or rounded "corm" up to 3.5 cm in diameter, with 1-several stems, the corm in age short-branched at the summit, producing one or a few, close offsets. Stem green to yellowish, purplish, or brownish, striate-sulcate, to 2.5 mm thick, simple (very rarely with a slender branch from low on the main stem), glabrous below the peduncles, the latter glabrous or distally sparsely pilose with sinuous hairs 0.3-1.2 mm long. Leaves at the very base of the stem almost scalelike, broader and much shorter than the rest, usually deciduous or withered by anthesis, well developed leaves rigid, linear or the uppermost linear-lanceolate, slightly lustrous, punctate, 1.2-20 (23) cm long and 1.2-3 (4.5) mm wide, those of the lower stem much longer than those above the middle, or sometimes the submedial leaves the longest and then upwards abruptly reduced, leaf margins strongly ciliate with spreading, almost straight to curved or sinuous hairs 0.9-1.5 mm long (mostly ca. 1.1 mm), these sometimes glabrescent, the midrib, especially beneath, sometimes with similar, shorter, less spreading, more or less evanescent hairs. Inflorescence a solitary, terminal head or more often a subcorymbiform, essentially complanate, arrangement, to 9.5 (11) cm long and 7 (9.4) cm broad, of 2-7 (10) heads, sometimes with one or a few of these up to a few cm lower (sometimes one or a few additional heads lower still, these usually abortive); peduncles (0.5) 1.2-4 cm long, with or without a leaflike bract, 8-15 mm long, positioned 0.5-3.5 mm below the involucre. Heads campanulate-turbinate, mostly 1.5-2.2 cm long, (11) 14-21 (25) flowered, about equal in size or the central slightly larger; involucre 10-19 (23) mm high, (0.8) 1-1.7 (2.1) cm wide, phyllaries (14) 17-22 (25), appressed or merely a little loose at the tip, persistent, (5.3) 8.0-12.5 (14.0) mm long and (1.0) 1.5-3 (3.8) mm wide, imbricated in about 3 series, ribbed in part, the unribbed portion punctate, the outer ones lanceolate, like reduced leaves, green or purple, long-acuminate, ribbed only proximally, ciliate essentially to the base, inward a little broader, less acuminate, the ribbed portion progressively greater and the cilia more distally disposed, the innermost narrowing again, linear-oblong, short-acuminate; chaff never developed; outer flowers of the head tending to be bent outward approximately at mid-corolla, giving the heads a broad, flat top; corolla purple, (9) 11-12.5 (14) mm long, outer surface of tube with glandular punctations, inner surface of tube glabrous, lobes glabrous, very narrowly triangular, 1.6-2.2 (2.6) mm long; anthers distinctly included, apices (1.5) 3-4 mm below the sinuses of the corolla, appendages ovate; style and stigmas purple and glabrous. Fruit a several-ribbed achene, body dark brown, (3.2) 5.2-7 (7.3) mm long, 1.0-1.8 mm wide, glabrous. Pappus (7) 8-10 mm long, purplish (at least distally), barbellate, the barbels ca. 0.1 mm long. Chromosome number unknown.

Flowering late June-July (August), fruiting late July-September.

English Name: Cahaba Torch.

Paratypes. Alabama: Bibb Co., (topotype) 1 Nov 1993, A. and S. 8134 (DUKE, JSU, VDB); 14.3 km NE of Centreville, "Starblaze Glade Northeast," 6 Sep 1992, A. and S. 7119 (GH, MO, NY, UNA, US); 14.6 km NE of Centreville, "Goat Glade South," A. and S. 7814, 18 Jul 1993 (FLAS, FSU, USCH); 14.8 km NE of Centreville, "Brown's Dam South Glade III," 18 Jul 1993, A. and S. 7820 (AUA, DUKE, GA, JSU, UNA); 14.9 km NE of Centreville, "Brown's Dam North Glade West," 18 Jul 1992, A. et al. 6864 (AUA, UNA); 14.9 km NE of Centreville, "Goat Glade North," 3 Jul 1992, A. 6787 (NY, US); 15.3 km NE of Centreville, "Desmond's Glade," 18 Jul 1993, A. and S. 7817 (AUA, GA, GH, JSU, MICH, UNA, VDB).

Liatris oligocephala is found only on the sunniest, most exposed microhabitats, unlike L. cylindracea, which is able to tolerate moderate shade. The latter is actually a more characteristic plant of the Ketona Glades, found on virtually all but the easternmost site.

This new Liatris appears to be sufficiently distinctive to merit the establishment of a new series, perhaps even a new section, to accommodate it, but we will leave that decision to a future monographer. It is clearly a Liatris because of its globose "corm," epaleaceous receptacle, and the fact that it hybridizes with L. cylindracea. Past generic descriptions (e.g., Geiser 1946, King and Robinson 1987) however, must be revised because L. oligocephala is the only known Liatris lacking setulae on the ribs of its achenes. It is also the only Liatris that produces mostly one to three heads or at most a simple cyme. Liatris cymosa (Ness.) K. Schum. and L. ohlingerae (Blake) B. Rob. often have the heads arranged in a simple cyme, but vigorous individuals of both these species bear compound cymes (Geiser 1946). The two sections of the genus currently recognized were distinguished by L. O. Geiser (1946) by having a plumose vs. a barbellate pappus. If Geiser's sections are natural, the plumose pappus found in both L. cymosa and L. ohlingerae (Geiser 1946) would indicate that they are not closely related to L. oligocephala, which has a barbellate pappus.

Though the great majority of species of Liatris bear their heads in a spikelike or racemelike arrangement, true spikes and racemes are indeterminate. In the Liatris inflorescence, floral maturation begins at the apex and proceeds downwards (basipetal), and its determinate nature means that the common forms are, properly speaking, spiciform or racemiform cymes (Geiser 1946). Thus, the simple cyme of L. oligocephala would appear to be a primitive characteristic. To posit that the plant actually resembles an ancestral taxon that was a theoretical progenitor of both Liatris and the closely related Carphephorus Cass. would be too speculative, but L. oligocephala bears so similar an aspect in the field to that of Carphephorus pseudoliatris Cass. that real consideration was given to another choice of epithet: Liatris pseudocarphephorus. The lack of setulae on the achenes, however, makes L. oligocephala a discordant element within the entire subtribe in which King and Robinson (1987) placed these genera, and absence of setulae may be a derived trait, unique to the species (autapomorphy).

When two or more species of Liatris grow in proximity and flower simultaneously, hybrids are often produced (Geiser 1946). Although anthesis in Liatris oligocephala begins prior to that of the syntopic and usually more abundant L. cylindracea, there is a limited period of overlap in flowering, and we detected hybrids between them on three glades. The hybrids were always found in close proximity to both parental species, and the latter are so different from each other in head arrangement that many hybrid plants were easily recognized as such. Much greater difficulty is encountered in trying to sort out F¹ plants from putative backcrosses.

In technical characters the hybrid differs from the parents by the intermediate length of the barbels of its pappus bristles, and with the ribs of the achenes bearing fewer setulae than in Liatris cylindracea (absent in L. oligocephala). Notable also is the intermediacy in shape of corolla lobes, in the degree of internal pubescence of the corolla (both the lobes and tube), and in the pubescence of the style and stigma. Intermediacy is expressed in other characters as well, including phyllary shape and the pubescence of stem, leaves, and phyllaries.

The aggregate number of individuals and populations of Liatris oligocephala is critically small. Therefore, interspecific hybridization should be recognized as a potential threat to its survival. At the very least, multi-year monitoring of the interaction between L. cylindracea and L. oligocephala should be a conservation objective. Ultimately, since there are many more Bibb County populations of L. cylindracea than of L. oligocephala, it might prove desirable to reduce or even eliminate L. cylindracea and their hybrid at the few glades where L. oligocephala occurs.

Despite some ambivalence about the practice of assigning unique epithets to nothospecies, we acknowledge that there is a tradition of doing so in certain genera, Liatris among them (see Geiser 1946). We therefore formally describe the two hybrids we found between species of Liatris, while resisting the impulse to do the same in similar cases involving species of Coreopsis, Onosmodium, and Silphium, where no such tradition exists.

The first of these hybrids is dedicated to the memory of John D. Freeman (1941-1997), noted Trillium expert, a positive influence on his numerous students at Auburn University, and one of the first in the state to call attention to the need for the preservation of threatened and endangered plants of Alabama.

Liatris Xfreemaniana J. Allison, hybr. nov. TYPE: Alabama: Bibb County, ca. 14.5 km NE of Centreville. "Starblaze Glade Northeast," Ketona Dolomite outcrop ca. 0.6 km N of the Little Cahaba River; with both parental species, 18 Jul 1993, James R. Allison and Timothy E. Stevens 7804 (holotype, AUA; isotypes: GH, MO, NY, UNA, US, VDB). Figure 9.

Planta inter L. cylindraceam Michx. et L. oligocephalam J. Allison plus minusve intermedia, ex hybridatione harum specierum orta et his crescens; ab ambobus differt pappo subplumoso et corpore achenii intermedio numero setularum.

Paratypes. Alabama: Bibb County, (topotypes) 6 Sep 1992, A. and S. 7118 (DUKE, GA, JSU, MICH, USCH); 27 Sep 1992, A. and S. 7237 (AUA, JSU, UNA); 1 Nov 1993, A. and S. 8131 (US); 14.0 km NE of Centreville, "Starblaze Glade Southwest," 18 Jul 1993, A. and S. 7801 (AUA, GH, MO, NY, UNA, US, VDB); 15 Aug 1993, A. and S. 7878 (JSU, UNA, VDB); 1 Nov 1993, A. and S. 8133 (NY); 14.6 km NE of Centreville, "Goat Glade South," 18 Jul 1993, A. and S. 7813 (AUA, MICH, UNA).

The other Liatris hybrid detected during our surveys was found at sites where L. cylindracea occurs close to, though usually not intermixed with, L. squarrosa L. In neither Geiser (1946), King and Robinson (1987), nor the PLANTS Database (USDA 2000) is the hybrid L. cylindracea X L. squarrosa reported. While L. cylindracea is a characteristic species of the Ketona Glades. L. squarrosa is rarely found on them, and then usually in marginal ecotones. According to the PLANTS Database (USDA 2000), there are five accepted varieties within Liatris squarrosa. At the locations cited below, the squarrosa plants were all referable to var. squarrosa sensu Geiser (1946).

This hybrid is dedicated to Sidney McDaniel, Professor of Botany at Mississippi State University, a keen observer and collector of the flora of the Southeast and of the Iquitos region of Peru, and mentor to many well trained students, with wishes for a speedy recovery from a recent serious injury.

Liatris Xmacdanieliana J. Allison, hybr. nov. TYPE: Alabama: Bibb County, ca. 9.0 km NNW of Centreville. Occasionally mowed area adjacent to "Highway 5 Glade West," Ketona Dolomite outcrop just W of Ala. Hwy. 5 at jct. Ala. Hwy. 219. With Liatris squarrosa. Liatris cylindracea nearby on glade, 20 Aug 1994, James R. Allison 8559 (holotype, US; isotypes: AUA, GH, IBE, MO, NY, UNA). Figure 9.

Planta ex hybridatione L. cylindraceae Michx. et L. squarrosae L. orta et his crescens; forma divergentiaque bractearum involucrorum etiam pubescentia caulis et foliorum et involucrorum inter parentes media.

Paratypes. Alabama: Bibb County, (topotypes) 2 Sep 1996, A. et al. 9461 (AUA, FLAS, FSU, JSU, UARK, UNA, USCH); 17 Aug 1998, A. 11343 (AUA, DUKE, GA, JSU, MICH, VDB); 19 Aug 1999, A. 12166 (GA, GH, JSU, UNA); 8.9 km NNW of Centreville, "Highway 5 Glade East, "12 Aug 1993, A. and S. 7862b (US, VDB);19 Aug 2000, A. 12509 (IBE); 13.6 km NE of Centreville, "Double Glade South," 17 Jul 1993, A. and S. 7798 (NY, US); 31 Jul 1994, A. and S. 8533 (AUA, JSU, UNA); 17 Jul 1999, A. et al. 12095 (GH, MO).