Printer Friendly

Sandstone rockhouses of the Eastern United States, with particular reference to the ecology and evolution of the endemic plant taxa.

II. Introduction

Rockhouses are semicircular recesses that extend far back under cliff overhangs ([ILLUSTRATION FOR FIGURE 1 OMITTED]; Braun, 1950), mostly of sandstone and limestone (Sauer, 1927; Willard, 1960), and they are large enough to provide shelter for humans (Sauer, 1927). In eastern United States, other terms that are synonymous, or nearly so, with sandstone rockhouse are alcove (Bates & Jackson, 1980; Vaughn, 1989), bluff shelter (Willard, 1960), cliff overhang (Bates & Jackson, 1980), grotto (Wolfe et al., 1949), recess cave (Hall, 1953), rockshelter (McFarlan, 1954; Bates & Jackson, 1980; Donahue & Adovasio, 1990), rockshelter cave (Carman, 1946), sandstone cave (Griggs, 1914; Bates & Jackson, 1980), and shelter cave (DeLong, 1967). Sandstone rockhouses are not unique to eastern United States; they occur also in Australia (Baker, 1990), Tasmania (Coates & Kirkpatrick, 1992), southwestern United States (e.g., Welsh, 1989), and perhaps elsewhere.

Like other rock outcrops of eastern United States (e.g., shale barrens, cedar glades, granite outcrops), a number of plant taxa are endemic, or nearly so, to sandstone rockhouses (Table I), but none is endemic to limestone rockhouses. However, unlike the plant assemblages of other rock outcrops of eastern United States, those of sandstone rockhouses have not been studied or documented thoroughly in the botanical/ecological literature. Thus, the primary purpose of this paper is to synthesize information on the geographic distribution, ecology, evolutionary origin, and conservation biology of the endemic plant taxa of sandstone rockhouses in the eastern United States. But first, a general overview of the landscape ecology; geology, formation, and age; human uses; habitats and physical environmental factors; flora; and fauna of sandstone rockhouses in the eastern United States is given.

III. Landscape Ecology

Sandstone rockhouses occur in several physiographic provinces in the eastern United States, but most occur in the Appalachian Plateaus, Interior Low Plateaus, and Ozark Plateaus physiographic provinces (Table II). The highest concentrations of endemic taxa are found in the Appalachian Plateaus and Interior Low Plateaus physiographic provinces, with the greatest number in the Red River and Cumberland River drainages on the Cumberland Plateau [ILLUSTRATION FOR FIGURE 2 OMITTED].

Rockhouses occur in gorges of highly dissected uplands and nearly-level plateaus where streams have cut downward through sandstone, producing vertical cliffs and steep-sided valleys as in the Cliff Section of the Cumberland Plateau [ILLUSTRATION FOR FIGURE 3 OMITTED] (cf. Fenneman, 1938; Smalley, 1986). In addition, rockhouses occur on the sides of longitudinal ridges of the dip slope on monoclinal mountains, such as the Cumberland Mountains (Braun, 1935; J. Walck, pers. obs.), and on outliers ("knobs") from the main escarpment of the Cumberland Plateau and Shawnee Hills (Quarterman & Powell, 1978).

Thousands of rockhouses at varying densities may occur in the landscape. For example, 2080 rockhouses were documented in the 1641.8 ha Big South Fork Archaeological Project on the Cumberland Plateau of Kentucky and Tennessee (Ferguson et al., 1986; Hoffman, 1987). Archaeological surveys record 0.021 to 1.33 rockhouses [ha.sup.-1] on the Cumberland Plateau of eastern Kentucky (Table III). Further, Sussenbach (1990) [TABULAR DATA FOR TABLE II OMITTED] reported a mean density of 0.79 [+ or -] 0.75 rockhouses [(100 ac).sup.-1] [(40 ha).sup.-1] in eastern Kentucky; the large standard deviation reflects high variation between survey units and stream drainages.

Rockhouses can occur anywhere from the base of the gorge adjacent to stream bottoms almost to ridgecrest (Sussenbach, 1990). In eastern Kentucky, archaeological surveys record elevations above mean sea level (a.m.s.1.) of rockhouses from 1000 to 1420 ft [305-433 m (Sussenbach, 1990)], 880 to 1160 ft [268-354 m (Wyss & Wyss, 1977)], and 770 to 1295 ft [235-395 m (Hoffman, 1987)]. This multi-layered elevation effect corresponds to the stratigraphy of geologic formations, i.e., several thick, resistant sandstone layers separated by siltstone, shale, coal beds (Sussenbach, 1990), or less-resistant sandstone layers. Rockhouses occur on a variety of slopes and aspects (Wyss & Wyss, 1977; Ferguson & Gardner, 1986), and within a cliffline they may be separated by cliffs or slopes.

Rockhouses are found primarily in the mixed mesophytic, western mesophytic, and oak-hickory forest regions of Braun (1950). In these regions, oak-hickory or oak forests are dominant on uplands and mixed mesophytic forests in gorges (Braun, 1950; Hinkle, 1989; Bryant et al., 1993). Floristically and vegetationally, gorges in the Shawnee Hills of western Kentucky are similar to those in the Cumberland and Unglaciated Allegheny Plateaus. The rockhouse endemics "Silene rotundifolia, Heuchera parviflora var. Rugellii [H. parviflora var. parviflora (Wells, 1984)], and Thalictrum clavatum [probably T. mirabile] on the sandstone cliffs accentuate the similarity" (Braun, 1950). Vegetation outside rockhouses varies depending on elevation and aspect, but rockhouses are recorded in hemlock, mixed oak, oak-hickory, mixed mesophytic, beech/yellow poplar, and mixed pine communities (Segars et al., 1951; Van Stockum, 1979; Medley, 1980; Ferguson & Gardner, 1986; Campbell et al., 1989).

IV. Geology, Formation, and Age

Sandstone rockhouses in eastern United States are formed mostly in Mississippian and Pennsylvanian quartzose sandstone [i.e., ca. 95% quartz and little or no silt- or clay-size matrix (Rice, 1984)] and quartz-pebble conglomerate (e.g., Jacobson, 1992; Chesnut, 1992) (Table IV). High concentrations of various iron minerals (W. Anderson, pers. comm.) may be present in narrow, irregular seams (McFarlan, 1954), and tafoni (cavities) may be formed by honeycomb weathering (Jennings, 1968) in the sandstones (e.g., Cumings, 1922; Wolfe et al., 1949).
Table III

Density and dimensions of sandstone rockhouses reported in
archaeological surveys in eastern Kentucky

 Survey

Measured variables 1(a) 2(b) 3(c)

Density (per [ha.sup.-1]) 0.047(d) 1.33(e) 0.021(f)

Length (m)

N 137 956 54
Range 4.5-121.9 2-110 4-50
Median 18.3 -(g) 12
Mean 23.5 15.7 12

Depth (m)

N 131 952 54
Range 2-30.5 1-46.5 2-10
Median 7.0 - 4
Mean 7.9 4.7 4

Height (m)

N 126 926 54
Range 1-45.7 0.5-40 0.85-9.9
Median 8.0 - ca. 2.2
Mean 11.2 2.5 2.7

Floor area ([m.sup.2])

N - 911 54
Range - 2-1850 12-144(h)
Mean - 68.4 -

a Rockhouses in Menifee (Webb & Funkhouser, 1936; Wyss & Wyss,
1977), Lee (Funkhouser & Webb, 1929), Wolfe, and Powell (Funkhouser
& Webb, 1930) counties in the Cliff Section of the Cumberland
Plateau (sensu Braun, 1950).

b Rockhouses in McCreary County (Hoffman, 1987) in the Cliff Section
of the Cumberland Plateau (sensu Braun, 1950).

c Rockhouses in Breathitt County (Sussenbach, 1990) in the Low Hills
Belt of the Cumberland Plateau (sensu Braun, 1950).

d Based on 115 rockhouses previously occupied by humans on 6000 ac
(2428 ha) (Wyss & Wyss, 1977).

e Based on 958 rockhouses previously occupied and unoccupied by
humans (Hoffman, 1987) on 720.7 ha (Ferguson et al., 1986).

f Based on 61 rockhouses previously occupied and unoccupied by
humans on 7222 ac (2923 ha) (Sussenbach, 1990).

g - = no data.

h Over 50% = [less than]44 [m.sup.2].


Three factors appear to be essential for the development of sandstone rockhouses: 1) thick-bedded sandstone; 2) basal or intercalated, less resistant rocks in the sandstone sequence; and 3) rapid valley incision through the sandstone sequence, causing development of a vertical cliff. During valley incision, lateral erosion of basal or intercalated, less resistant rocks in the sandstone sequence forms a small recess in the sandstone cliff (Donahue & Adovasio, 1990). The less-resistant rock may be shale (Donahue & Adovasio, 1990), dolostone (Willard, 1960), or a loosely cemented, cross-bedded [TABULAR DATA FOR TABLE IV OMITTED] sandstone layer (Hall, 1953). The initial development of a rockhouse can be rapid. Donahue and Adovasio (1990) observed that a recess 0.5 m in depth developed within 26 years along an artificial drainage ditch cut to the sandstone/shale interface.

Rockfall and attrition contribute to the enlargement of the recess. Release of rock fragments, which is greatest in the initial stages of rockhouse formation, is attributed to the gradual widening of joints and bedding planes by groundwater discharge, freeze-thaw conditions, root growth, and/or the thermal contraction and expansion of sandstone in response to daily and seasonal temperature changes. Attrition or granular disintegration of sand-sized sediment is likely due to physical and chemical (hydrolysis, oxidation, and dissolution of various minerals) weathering (Vento, 1985; Donahue & Adovasio, 1990). Weathering processes have varied in importance over geologic time [e.g., role of periglacial conditions (Morris, 1986)] and with geographic location [e.g., freeze-thaw conditions are more intense in northern than in southern latitudes (Vento, 1985)].

The landscape contains rockhouses at various stages of development, including vertical cliffs, slight overhangs, large overhangs ("rockhouses"), and deep recesses resembling caves (Carman, 1946; Bart, 1961). Thus, rockhouses vary widely in their dimensions (Table III). However, the largest rockhouses are located at the heads of gorges, where groundwater sapping is most active (Vaughn, 1989). Rockhouses are geologically stable structures (Young & Young, 1992), and rarely does the roof of one collapse (Sussenbach, 1990). Otherwise, they ultimately may become arches/bridges (McFarlan, 1954; Corgan & Parks, 1979).

Most rockhouses occur south of the Wisconsinan Glacial Boundary, in stream valleys downcut during the Pleistocene, especially during Illinoian and Wisconsinan Stages. This includes rockhouses in western Pennsylvania, southeastern Ohio, northeastern Kentucky (Hall, 1953; Ray, 1974; Donahue & Adovasio, 1990), northeastern Mississippi (Vento, 1985; Donahue & Adovasio, 1990), south-central Indiana (Miller et al., 1990), southern Illinois (Willman & Frye, 1980), and northern Arkansas (Quinn, 1958). For example, the initial development of the Meadowcroft Rockshelter in southwestern Pennsylvania occurred during the Sangamonian Stage, and after the middle Late Wisconsinan Stage (ca. 21,300 Y.B.P.) it was large enough for human occupation (Adovasio et al., 1984)

In only a few cases do rockhouses occur in gorges that were glaciated, and thus they have formed since the retreat of the Wisconsinan Glacier [e.g., southwestern New York (cf. Parks & Farrar, 1984)]. Other evidence suggests that some rockhouses formed in the Late Tertiary. In the Red River Gorge of eastern Kentucky, rockhouses occur in gorges downcut between the Eocene and Pleistocene (Jillson, 1969). Recently, Outerbridge (1987) concluded that dissection of a portion of the Cumberland Plateau southeast of the Red River Gorge occurred during the past 1.5 million years. Using his method for determining the age of erosion surfaces and his erosion rate of quartzitic sandstone [10 m [(million year).sup.-1]], we determined that the erosion surface in the Red River Gorge (relief = ca. 180 m), and possibly in the Cumberland River area, developed during the past 9 million years.

V. Prehistoric and Historic Uses by Humans

The endemic plant taxa occur in some rockhouses that were used and impacted heavily by humans during prehistoric and historic times. Since the Paleo-Indian Period (13,000-10,000 Y.B.P.), humans have utilized rockhouses as temporary bivouacs, special-activity loci, seasonal and long-term encampments, and burial sites (Funkhouser & Webb, 1929; Pace et al., 1986; Donahue & Adovasio, 1990). In addition, the Meadowcroft Rockshelter has yielded radiocarbon dates associated with materials of human manufacture as old as 19,600 Y.B.P. (Adovasio et al., 1984).

For the most part, prehistorically occupied rockhouses differ little from unoccupied ones, except in distance to chert outcrops, east/west aspect (Sussenbach, 1990), and size (Ferguson & Gardner, 1986). Today, the most obvious signs of visitation or occupation by prehistoric humans are hominy holes and petroglyphs (Coy & Fuller, 1971). The petroglyphs probably date to the Mississippian Period [1100-300 Y.B.P. (Pace et al., 1986)].

In historic time, rockhouses have been used as dwellings and as barns by lumbermen and local families, campsites by hunters and trappers, shelters for whiskey stills (e.g., Funkhouser & Webb, 1928, 1929, 1930; Howell, 1981), a site for graveside services during inclement weather, a schoolhouse (Howell, 1981), an amphitheater (Ruchhoft, 1986), and a springhouse (Shaver, 1954). They also have been impacted heavily by the removal of ashes left from fires of Native Americans for use as fertilizer (Funkhouser & Webb, 1930); mining of saltpeter, especially during the War of 1812 and the Civil War (e.g., Webb & Funkhouser, 1936); and the searching by local people for a lost silver mine and treasure (Funkhouser & Webb, 1930) and Native American artifacts (Ferguson & Gardner, 1986). Remnants of whiskey stills, saltpeter mining equipment, and houses still are present in some rockhouses (e.g., Gardner, 1986).

Besides noting prehistoric and historic artifacts in rockhouses, archaeologists also have analyzed faunal and floral remains to better understand resource utilization (Cowan, 1979; Hoffman, 1987), cultural processes (Sussenbach, 1990), and paleovegetation outside of rockhouses (Cowan et al., 1981; Adovasio et al., 1984). Moreover, remains of cultivated plants retrieved from rockhouses have increased our understanding of the development of horticulture in eastern North America, particularly the Eastern Agricultural Complex (Cowan et al., 1981; Smith, 1992).

VI. Habitats and Physical Environmental Factors

Three distinct habitats occur in rockhouses: ceiling, backwall, and floor [ILLUSTRATION FOR FIGURE 4 OMITTED]. The ceiling typically is highest at the outer edge of the rockhouse and slopes downward to form the backwall. The backwall is the darkest part of the rockhouse and is usually moist due to groundwater seepage. Crevices, ledges, and tafoni (cavities) are common on the ceiling and backwall. The floor is bounded by the backwall and the dripline. The dripline forms where water and sheet-wash sediments fall from the outermost edge of the ceiling. A plunge basin may be present at the dripline. The floor consists of sandy soil and/or small to large boulders, and leaf litter usually accumulates only at the dripline. Topography of the rockhouse floor may be level to fairly steep. The steepest part of the floor in some rockhouses occurs at an active spring or at the beginning of a stream course. A ceiling, backwall, and/or floor may occur also under large sandstone boulders (slump blocks) on the colluvial slope below the cliffline, under slight overhangs, and under arches.

The interior of rockhouses is shaded by the ceiling and by the tree canopy adjacent to the dripline. At the backwall of a rockhouse with Trichomanes boschianum and on a boulder with T. intricatum, illuminance in summer mostly was 30-60 ft-c (323-646 lux; full sun = ca. 10,000 ft-c or ca. 107,650 lux). However, in winter, due to the absence of leaves in the surrounding forest canopy, illuminance increased to ca. 500 ft-c (5382 lux).

In winter, these ferns may receive 1-3 hours of direct sunlight (Farrar, 1971). Illuminance in various rockhouse habitats of Vittaria appalachiana ranged from [less than]50 to 250 ft-c (538-2691 lux), but most readings were ca. 60 ft-c (645 lux) (Farrar, 1978). Voigt and Swayne (1955) determined that the highest illuminances in habitats of Dodecatheon frenchii under a bluff and of D. meadia under an open canopy were 8-300 ft-c (86-3230 lux) and 5000-6000 ft-c (53,825-64,590 lux), respectively; 10,000 ft-c (107,650 lux, or full sun) was recorded in an area with no canopy. One population of D. frenchii in an open woods received 5000-6500 ft-c (53,825-69,972 lux).

Although light quantity is lower inside than outside rockhouses, plants at the dripline receive more light than those at the back of the rockhouse. For example, in the rockhouse shown in Figures 1 and 4, the photosynthetic photon irradiance (PPI) recorded above plants of Solidago albopilosa near the dripline, under partly cloudy conditions at 3:30 P.M. (EDT) on 1 July 1994, was 30-37 [micro]mol [m.sup.-2] [s.sup.-1], whereas the PPI near plants of Trichomanes boschianum and above plants of Thalictrum mirabile at the backwall was 3-5 [micro]mol [m.sup.-2] [s.sup.-1]. However, the PPI of a sunfleck above S. albopilosa near the dripline in the same rockhouse was 1100 [micro]mol [m.sup.-2] [s.sup.-1] (full sun = ca. 2000 [micro]mol [m.sup.-2] [s.sup.-1]) (J. Walck et al., unpubl.).

Temperatures inside rockhouses are higher during winter and lower during summer, and the difference between monthly maximum and minimum temperatures smaller, than those outside rockhouses (Tables V, VI). The longest frost-free period in a rockhouse in Ohio was 276 days (9 March to 29 November), whereas that recorded at a nearby weather station was only 160 days (22 April to 29 September). However, the longest continuous frost period was 373 hours in the rockhouse, compared to 246 hours in a "frost pocket" in the valley below the rockhouse (Wolfe, 1951). Temperatures of groundwater flowing from the backwall in rockhouses in northeastern Ohio averaged 8-9 [degrees] C year round (Cusick, 1983); in southern Illinois, groundwater temperature in a rockhouse with Trichomanes boschianum was 57 [degrees] F (13.9 [degrees] C) on 11 August (Mohlenbrock & Voigt, 1959).

Voigt and Swayne (1955) reported higher relative humidity and lower evaporative rate in a rockhouse with Dodecatheon frenchii than in the surrounding forest with D. meadia. In addition, in southern Ohio the per-season evaporation rate in a "frost pocket" in the valley below a rockhouse was five times greater than that in the rockhouse (Wolfe, 1951).

Although the inside of rockhouses is protected from direct precipitation (Wolfe, 1951), water can be blown into them during rains (Braun, 1940) and from waterfalls at a plunge basin (Farrar, 1971). In addition, water can flow far back into the rockhouse along the underside of the ceiling, especially during a heavy rain or during rain of long duration (Adovasio et al., 1977). Farrar (1971) observed that during summer, air moving into a rockhouse from the gorge can be cooled, and moisture can condense upon plants. Otherwise, a perennial water supply is available, especially at the backwall, from groundwater seepage. Soil moisture varies spatially and temporally within and between rockhouses, but no seasonal patterns exist [ILLUSTRATION FOR FIGURE 5 OMITTED]. In an archaeological survey in eastern Kentucky, the floors in rockhouses averaged 50% dry, 39% damp (visible moisture but not total saturation), and 11% wet (standing water) (Hoffman, 1987).

Soil depth to bedrock in rockhouses in eastern Kentucky ranged from 0 to 90 cm and averaged 48 cm (Hoffman, 1987; Sussenbach, 1990). Soils are composed of ca. 72-95% sand, with minor amounts of clay and silt (Table VII; Cowan et al., 1981; Vento, 1985; Morris, 1986). Percentages of silt, clay, sand, and larger fractions vary depending on depth of strata (Cowan et al., 1981; Morris, 1986; Donahue & Adovasio, 1991), and they are dependent on the mechanism of sedimentation: rockfall, attrition, sheet wash, flooding (Donahue & Adovasio, 1990), and eolian deposition (Morris, 1986). Paleo- and modern-day sedimentation rates, paleodriplines, and paleotopographies of the floor have been determined at Meadowcroft Rockshelter (Donahue & Adovasio, 1990).

Levels of soil nutrients vary within and between rockhouses and with soil depth in rockhouses. In addition, differences between levels of soil nutrients inside and outside the dripline vary between rockhouses (Table VII; Vento, 1985). High levels of nitrogen and phosphorus may be expected in rockhouses, since saltpeter earth occurs in them. Ammonium transported in seeping groundwater from surface soils via an evaporative moisture gradient is oxidized to nitrate at the atmosphere-cave wall surface. Phosphorus may be derived from seeping groundwater or the clay component of underlying shale (Hill, 1981). High levels of calcium, potassium, phosphorus, and magnesium, and high pH, may occur in rockhouses; these levels are associated with materials (e.g., mussel shells, bone, charred and uncharted wood, calcareous cherts) left by prehistoric occupants (Vento, 1985).

Soil pH generally is between 4 and 5 (Table VII; Cowan et al., 1981; Vento, 1985; Morris, 1986), and soil organic matter is between 0.5 and 21.6% (Table VII). Morris [TABULAR DATA FOR TABLE V OMITTED] [TABULAR DATA FOR TABLE VI OMITTED] (1986) reported that organic matter content was highest in top layers of sediment profiles (2.3-10%) but decreased sharply to 0% with increase in depth. The depth at which the sharp decrease occurred ranged from 12 to 33 cm.

VII. Flora

Algae (including Cyanophyta), bryophytes, and vascular plants have been studied in sandstone rockhouses of the eastern United States. However, we are unaware of any published studies on bacteria, fungi, or lichens in rockhouses.

A. ALGAE AND BRYOPHYTES

Although algae and bryophytes are common on the backwall of sandstone rockhouses in eastern United States, none is known to be endemic to them. Algal communities on sandstone cliffs, including those under overhangs and in rockhouses, have been studied in south-central Ohio (Koch, 1976) and eastern Kentucky (Camburn, 1982). In general, algae are found as a gelatinous mass composed primarily of taxa belonging to Cyanophyta, Chlorophyta, and Bacillariophyceae (Camburn, 1982). Koch (1976) recorded 91 taxa of Chlorophyta, 64 of Cyanophyta, 6 of Chrysophyta (excluding diatoms), and 1-3 each of Pyrrhophyta, Euglenophyta, and Xanthophyta. Further, Camburn (1982) recorded 113 taxa of diatoms. Several taxa in Koch's (1976) study were relicts with northern affinity. Koch (1976) concluded that moisture was the major factor influencing community composition. Algal growth on cliffs occurs slowly, and large algal mats therefore require many years to develop (Koch, 1976).

Some of the most commonly mentioned mosses include Bryoxiphium norvegicum (Brid.) Mitt., Hookeria acutifolia Hook. & Grev., Syrrhopodon texensis Sull., and Tetraphis pellucida Hedw. (Cranfill, 1980, 1991). In addition, Scopelophila cataractae (Mitt.) Broth., which is a "metal moss" endemic to substrates containing high levels of copper, zinc, lead, and/or iron, has been recorded in rockhouses in Kentucky (Studlar & Snider, 1989) and under an overhanging sandstone ledge in Tennessee (Shaw, 1987).

[TABULAR DATA FOR TABLE VII OMITTED]

Mohlenbrock and Voigt (1959) listed four mosses and five hepatics associated with Trichomanes boschianum in southern Illinois, and Farrar (1978) noted that H. acutifolia occasionally was associated with Vittaria appalachiana. In the Red River Gorge of eastern Kentucky, Studlar and Snider (1989) recorded 41 mosses and 22 hepatics in sandstone "caves" (crevices, rockhouses, under arches). The sandstone "cave" bryofloras in eastern Kentucky (Studlar & Snider, 1989) and the Ozarks (Redfearn, 1987) contain Tertiary/Pleistocene-relict taxa with boreal, temperate, and tropical affinities.

B. VASCULAR PLANTS

Vascular plants more commonly found in habitats outside than inside rockhouses are associated with endemic taxa on the floor. In addition, epipetric ferns (e.g., Asplenium spp.) occur on the backwall and on the ceiling. The vegetation in most rockhouses consists of an herbaceous stratum with low plant cover and an occasional tree (adult, juvenile, and/or seedling), shrub, and/or woody vine. Some rockhouses have few or no plants growing inside them. However, in other rockhouses the endemic taxa (e.g., Ageratina luciae-brauniae, Solidago albopilosa) may form dense, monoculture stands on the floor (Medley, 1980; Wofford & Smith, 1980; J. Walck et al., pers. obs.).

Vascular plant taxa associated with Ageratina luciae-brauniae and Solidago albopilosa were determined from floristic surveys of 15 rockhouses containing A. luciae-brauniae (from Palmer-Ball et al., 1988, and Campbell et al., 1989) and of 28 other rockhouses containing S. albopilosa (J. Walck & S. Francis, unpubl.). Separate lists were kept for all 15 sites of A. luciae-brauniae and for 17 of the 28 sites of S. albopilosa; however, for the other 11 sites of S. albopilosa, the number of sites in which each taxon occurred was recorded. Thus, all 28 rockhouses with S. albopilosa could be used to calculate percent presence; only the 17 sites for which separate lists were kept could be used to calculate Sorensen's index (see below).

A total of 99 taxa was recorded in the 15 rockhouses with Ageratina luciae-brauniae, and the number of taxa observed in each of the rockhouses ranged from 9 to 39 (mean [+ or -] SE = 19 [+ or -] 2) (Table VIII). A total of 85 taxa was recorded in the 28 rockhouses with Solidago albopilosa, and the number of taxa observed in each of the 17 rockhouses for which separate floristic lists were made ranged from 4 to 31 (mean [+ or -] SE = 15 [+ or -] 2). The [TABULAR DATA FOR TABLE VIII OMITTED] majority of taxa recorded in both surveys are native, and they commonly occur in the surrounding forest; only Dactylis glomerata L., Digitaria ischaemum (Schreber) Muhl., Microstegium vimineum (Trin.) A. Camus, Polygonum cespitosum Blume, Rumex aceto-sella L., and Stellaria media (L.) Villars are non-native.

Percentage of the 15 (and 28) rockhouses in which a taxon occurred was calculated, and each was assigned to one of five presence classes (Cain & Castro, 1959): 1 (1-20%); 2 (21-40%); 3 (41-60%); 4 (61-80%); and 5 (81-100%) (Table VIII). Most taxa associated with Ageratina luciae-brauniae [ILLUSTRATION FOR FIGURE 6A OMITTED] or Solidago albopilosa [ILLUSTRATION FOR FIGURE 6B OMITTED] occurred sporadically among rockhouses. Many of the endemics co-occur (e.g., Reed, 1951; Wofford & Smith, 1980); however, Heuchera parviflora var. parviflora was the only endemic taxon consistently associated with both of these taxa (Table VIII). Although not recorded in Table VIII, Farrar's (1992) data indicate the occurrence of Trichomanes intricatum in a rockhouse with S. albopilosa.

A photosynthetic pathway ([C.sub.3], [C.sub.4], CAM) was assigned to each taxon based on information published in numerous papers. Photosynthetic pathways were inferred from information at the familial and/or generic level for taxa whose pathways had not been determined. All woody plants and ferns were assumed to have the [C.sub.3] pathway. Carter and Martin (1994) reported that sporophytes of Vittaria lineata (L.) Smith have CAM, but the photosynthetic pathway of gametophytes of this species has not been determined. Although sporophytes of the fern Pyrrosia longifolia (N. L. Burm.) Morton have CAM, gametophytes of this species use the [C.sub.3] pathway (Martin et al., 1995). Therefore, gametophytes of Vittaria appalachiana were assumed to have [C.sub.3] photosynthesis. Of the taxa associated with Ageratina luciae-brauniae and/or Solidago albopilosa, 96% have the [C.sub.3] photosynthetic pathway of carbon fixation, 4% have the [C.sub.4] pathway, and none has CAM. Only Andropogon sp., Digitaria ischaemum, Microstegium vimineum, Muhlen-bergia schreberi J. F. Gmelin, and M. tenuiflora (Willd.) BSP. have the [C.sub.4] pathway.

Life forms of spermatophytes (sensu Raunkiaer, 1934) and growth forms (trees, shrubs, woody vines, forbs, graminoids, ferns) were determined for each taxon from information in Gibson (1961). Life form and growth form spectra for each of the two surveys were compared to the Kentucky state flora (Gibson, 1961). The majority of taxa associated with Ageratina luciae-brauniae and/or Solidago albopilosa are phanerophytes (40%) or hemicryptophytes (36%); 2% are chamaephytes, 13% are cryptophytes, and 9% are therophytes. Forbs composed 33% of the taxa; trees were 22%, shrubs 10%, woody vines 5%, graminoids 15%, and ferns 15%. Percentages of phanerophytes are higher and those of hemicryptophytes and therophytes lower in the rockhouse flora than in the Kentucky flora (Table IX). Percentages of trees and ferns are higher, and that of forbs lower, in the rockhouse flora than in the Kentucky flora (Table IX).

Similarities in total species composition and in endemic taxa composition between all pairwise combinations of the 15 and the 17 (for which separate floristic lists were kept) rockhouses with Ageratina luciae-brauniae and Solidago albopilosa, respectively, were calculated using Sorensen's index (SIs),

[SI.sub.s] = 2c / A + B x 100,

where A is the total number of taxa in rockhouse A, B is the total number of taxa in rockhouse B, and c is the total number of taxa in both A and B. Values of SIs [greater than]50% between the floras of two rockhouses indicate that they represent the same community (Barbour et al., 1987).

[SI.sub.s] based on total species composition ranged from 8% to 53% (mean [+ or -] SE = 29 [+ or -] 1) among the 15 Ageratina luciae-brauniae sites and from 7% to 57% (mean [+ or -] SE = 30 [+ or -] 1) among the 17 Solidago albopilosa sites. Only 2% of the 105 pairwise comparisons between sites of A. luciae-brauniae and only 1% of the 136 pairwise comparisons between sites of S. albopilosa had a similarity of [greater than]50%. On the basis of total species composition, then, the majority of rockhouses surveyed had a distinct assemblage of plant taxa.

However, [SI.sub.s] values based only on the five endemic taxa at the 15 Ageratina luciae-brauniae sites and on the six endemic taxa at the 17 Solidago albopilosa sites were much higher than those for the entire vascular floras at these sites. Among the A. luciae-brauniae sites, [SI.sub.s] for the endemics ranged from 40% to 100% (mean [+ or -] SE = 78 [+ or -] 2), with 93% of the pairwise comparisons [greater than]50%. Among the S. albopilosa sites, SIs for the endemics ranged from 33% to 100% (mean [+ or -] SE = 69 [+ or -] 2), with 68% of the pairwise comparisons [greater than]50%; an additional 24% of the comparisons were equal to 50%. Thus the endemic taxa have a much greater tendency to form a distinct plant assemblage than does the flora as a whole.

Some additional associates of the endemic taxa not recorded in our compilation (Table VIII) are Anemonella thalictroides (L.) Spach (Timme & Lacefield, 1991), Aquilegia canadensis L., Lycopodium porophilum Lloyd & Underw. (Wofford & Patrick, 1980), Botrychium virginianum (L.) Sw., L. lucidulum Michx. (Medley, 1980), Cardamine pensylvanica Muhl. (Kral, 1983), Carex pedunculata Muhl., Circaea alpina L. (Campbell et al., 1989), Cystopteris bulbifera (L.) Bernh. (Campbell et al., 1994), Decumaria barbara L., Osmunda regalis L. (Gunn, 1991), Lygodium palmatum (Bernh.) Sw., Saxifraga michauxii Britton (Wofford & Smith, 1980), Stenanthium gramineum (Ker Gawler) Morong (J. Walck & J. Baskin, pers. obs.), and Sullivantia sullivantii (T. & G.) Britton (Wolfe, 1951).
Table IX

Percentages of Raunkiaer life forms of spermatophytes and of growth
forms of all plant taxa associated with Ageratina luciae-brauniae
and Solidago albopilosa in eastern Kentucky compared to the flora of
Kentucky (determined from Gibson, 1961)

Associated with
 A. luciae-brauniae S. albopilosa Kentucky

Life form

Phanerophyte 39.1 51.4 16.3
Chamaephyte 1.1 2.9 1.4
Hemicryptophyte 37.9 28.6 51.2
Cryptophyte 12.6 10.0 15.3
Therophyte 9.2 7.1 15.8

Growth form

Trees 19.0 26.7 7.7
Shrubs 10.0 14.0 6.8
Woody vines 7.0 4.7 2.5
Forbs(a) 40.0 20.9 61.9
Graminoids 10.0 16.3 16.8
Ferns 13.0 17.4 4.3

a Includes all herbaceous spermatophytes, except graminoids.


VIII. Fauna

The fauna of sandstone rockhouses in eastern United States has not been studied per se, but rockhouses are known to provide shelter for animals (Funkhouser, 1925; Funkhouser & Webb, 1928). Several noteworthy animals, none endemic, occur in rockhouses (Campbell et al., 1989, 1994): green salamander [Aneides aeneus (Cope & Packard)], eastern woodrat [Neotoma floridana (Ord)], Virginia big-eared bat [Plecotus townsendii (Cooper) subsp. virginianus (Handley)], Rafinesque's big-eared bat [Plecotus rafinesquii (Lesson)], and small-footed myotis [Myotis leibii (Audubon & Bachman)].

IX. Endemic Plant Taxa

The most distinctive aspect of the flora of sandstone rockhouses is the 11 plant taxa endemic to them (Table I); these taxa occur exclusively, or nearly so, behind the dripline of rockhouses [ILLUSTRATION FOR FIGURE 4 OMITTED] throughout their geographic range. These endemics are confined to a certain habitat (habitat endemism sensu Anderson, 1994), and three of them are known only from single river gorges (geographic endemism sensu Anderson, 1994).

In this paper, we classify the endemics following the cytologically based scheme of Favarger and Contandriopoulos (1961) with modifications by Keener (1983). The classes of endemics and their characteristics are given in Table X, and their related taxa and chromosome numbers in Table XI.

[TABULAR DATA FOR TABLE X OMITTED]

A. PALEOENDEMICS

A paleoendemic is an ancient relict with no apparent closely related extant taxon (Table X; Keener, 1983). The parental taxa of Trichomanes intricatum and Vittaria appalachiana either became extinct during the Late Tertiary or Pleistocene or still may exist in tropical America. Other species of Trichomanes and Vittaria that occur in eastern United States are not related closely to T. intricatum or to V. appalachiana (Farrar & Mickel, 1991; Farrar, 1992). Until more phylogenetic studies are done, it is best to consider T. intricatum and V. appalachiana as paleoendemics, especially in light of their unusual life cycle.

Trichomanes intricatum Farrar and Vittaria appalachiana Farrar & Mickel are perennials that reproduce exclusively by gemmae, even though gametangia may be present. These two species exist only as gametophytes (Farrar, 1978, 1992); however, Farrar (1978) observed abortive sporophytes of V. appalachiana in Ohio. Trichomanes intricatum ranges southward from southern New Hampshire and central Vermont to [TABULAR DATA FOR TABLE XI OMITTED] northern Georgia, northern Alabama, southern Illinois, and central Indiana (Farrar, 1992). Vittaria appalachiana ranges southward from southwestern New York and northeastern Ohio to southeastern Pennsylvania, northeastern Maryland, northern Georgia, northern Alabama, western Kentucky, and south-central Indiana (Parks, 1987; Farrar & Mickel, 1991). Elevations at which both species occur are in the range of ca. 150-1800 m a.m.s.l. (Farrar, 1993a, 1993b).

These two ferns occur in rockhouses, under overhanging cliffs, under rock outcrops, and on cliffs, and in these habitats they grow on ledges, in crevices, and on cliff faces. In addition, the two ferns grow on boulders and occasionally on bases of trees (Farrar & Mickel, 1991; Farrar, 1992). Habitats are moist, cool, and deeply shaded, and those in northeastern United States are more highly buffered from climatic extremes than are those in southeastern United States (Farrar et al., 1983). Trichomanes intricatum and Vittaria appalachiana have been recorded from a variety of noncalcareous rocks - sandstone/quartzite, conglomerate, grauwhacke, metaconglomerate, coal, schist, gneiss, granite, slate, and shale - and on decaying wood and soil (Parks, 1987; Farrar, 1978, 1992; Farrar & Mickel, 1991).

When the two species occur at the same site, Trichomanes intricatum hangs from the upper surface or vertical wall of crevices and Vittaria appalachiana grows on the lower surface of crevices; they usually occur in different crevices. In addition, T. intricatum is found at sites subject to more desiccation than those of V. appalachiana (Parks, 1989); however, T. intricatum may require wetter habitats in the northern than in the southern part of its range (Farrar et al., 1983). Both species can form dense, mat-like colonies of 100 [cm.sup.2] or more; however, if light is limiting, gametophytes of V. appalachiana tend to grow in rows (Farrar, 1978, 1992).

Reilly (1992) reported that, compared to plants that grow in high-light environments, Vittaria appalachiana has a low chlorophyll a/chlorophyll b ratio and a low activity of photosystem I, both of which are characteristic of shade plants (Ludlow & Wolf, 1975; Bjorkman, 1981; Nasrulhaq-Boyce & Mohamed, 1987). Gametophytes of V. appalachiana grow with their free surface perpendicular to the direction of the strongest light source, and they are etiolated in the darkest areas of the habitat. Competition for light with bryophytes or other plants appears to be low (Farrar, 1978). Yatskievych et al. (1987) suggested a partial zonation with respect to light: V. appalachiana occupies sites receiving less light than do most mosses and hepatics, but more light than do some species of Chlorophyta and Cyanophyta.

Morphological variations occur within and between populations of Vittaria appalachiana. Within a population of V. appalachiana, gametophytes growing on soil are more robust than those growing on rock. In addition, gametophytes growing on soil frequently produce archegonia, whereas those growing on rock have high gemmae production (Farrar, 1978). Differences in gemmae form and gametophyte size occur between populations. No distributional pattern exists in gemmae form, but gametophytes are smaller in the westernmost than in other parts of the range. While differences in gemmae form are maintained in culture, differences in robustness and gemmae production are not (Farrar, 1978, 1990).

Farrar (1978) found that gametophytes of Vittaria appalachiana were more cold hardy than are those of two tropical species, V. lineata and V. graminifolia Kaulfuss, that occur in southeastern United States. In addition, he showed that V. appalachiana gametophytes grew best on sandstone, and little or no growth occurred on palm bark, sea sand, or mica schist; gametophytes of V. lineata had high growth rates on all substrates.

The gametophyte and gemmae are more tolerant of desiccation or freezing than are the sporophytes of other taxa in Trichomanes and Vittaria (Farrar, 1985). If gametophytes grow where sporophytes cannot grow, the selective pressure for maintenance of alternation of generations and production of sex organs would be lessened while each gametophyte survived independently via asexual reproduction. This situation may have led to the loss of sexual reproduction. The inability of Trichomanes intricatum and Vittaria appalachiana to produce sporophytes is fixed genetically and thus is not due entirely to the environment (Farrar, 1971). These two species probably are remnants of taxa that once had a normal life cycle in the area in which the sporophyte phase now is greatly reduced or extinct (Farrar, 1985).

Trichomanes intricatum and Vittaria appalachiana occur up to ca. 400 and 90 km, respectively, north of the Wisconsinan Glacial Boundary ([ILLUSTRATION FOR FIGURE 2 OMITTED]; Cusick, 1983; Farrar et al., 1983). Since dispersal of gemmae is greatly restricted, the present geographical ranges of these two species indicate dispersal by spores when sporophytes were present. Farrar (1992) speculated that sporophytes of T. intricatum existed until after the Wisconsinan Glaciation. The distribution of this species may have been extended northward to New England during the Hypsithermal Interval by dispersal of spores. On the other hand, sporophytes of V. appalachiana may have been eliminated by a cooling climate either at the close of the Miocene or, at latest, during the Illinoian Glaciation (Farrar, 1990).

Vittaria appalachiana [n = 120 (Gastony, 1977)] originated possibly by interspecific hybridization from parental diploid species (Farrar, 1990). Such hybrid ferns produce viable diploid spores that form gametophytes. However, the gametophytes usually are incapable of producing sporophytes except by apogamy, which has been observed only rarely in cultures or in populations of V. appalachiana (Farrar, 1978, 1990).

Genetic variation in Vittaria appalachiana is about equal to that in sexually reproducing angiosperms. However, a very high degree of genetic uniformity occurs within populations, indicating very little gene flow between populations (Farrar, 1990).

A third species of fern with similar life cycle and evolutionary history as Trichomanes intricatum and Vittaria appalachiana that might be expected to occur in sandstone rockhouses is Hymenophyllum tayloriae Farrar & Raine, which is found in southwestern North Carolina and adjacent northwestern South Carolina and in northwestern Alabama. Hymenophyllum tayloriae occurs only as a gametophyte in the Carolinas, but both the sporophyte and gametophyte occur in Alabama (Raine et al., 1991; Farrar & Davison, 1994). The species occurs in crevices, on ledges, on boulders, and under or on cliffs of noncalcareous rocks other than sandstone in the Carolinas (Wagner et al., 1970; Raine et al., 1991); however, in Alabama it occurs on sandstone (D. Farrar, pers. comm.).

B. SCHIZOENDEMICS

In schizoendemics, adaptive radiation and vicariant evolution results in at least two population-complexes with identical chromosome numbers. Schizoendemics may be of varying ages: neoschizoendemics, which have restricted geographic ranges due to their youth; and holoschizoendemics, which are wide-ranging, "mature" or ancient taxa (Table X; Keener, 1983).

1. Neoschizoendemics

Ageratina luciae-brauniae, Arenaria cumberlandensis, Solidago albopilosa, and Thalictrum mirabile are classified as neoschizoendemics. All but Arenaria cumberlandensis are thought to be youthful because of their narrow geographic ranges and the presence of the putative ancestor in neighboring habitat outside of rockhouses. Evolution in these taxa probably proceeded by adaptive radiation, i.e., diversification of form in response to pressures of a different habitat (Bramwell, 1972). The origin of A. cumberlandensis remains obscure, but for now this taxon is classified as a neoschizoendemic.

Ageratina luciae-brauniae (Fern.) King & Robinson [Eupatorium luciae-brauniae Fern. (King & Robinson, 1970)] is a proto-hemicryptophyte perennial without runners (Gibson, 1961). It occurs in the Cumberland River drainage of Tennessee [four adjacent counties (Wofford & Patrick, 1980)] and Kentucky [five adjacent counties (Palmer-Ball et al., 1988; Campbell et al., 1991)], and in the Red River drainage of Kentucky [one county (Campbell et al., 1989)]. Approximately 100 sites are known in Tennessee and Kentucky (Campbell et al., 1994), and populations range from three flowering stems in four clumps (Wofford & Patrick, 1980) to "1000s of Eupatorium plants, of good vigor" (Palmer-Ball et al., 1988). The species first was described by E. L. Braun (1940) as E. deltoides, but since this name was a later homonym (Wofford, 1976), E. luciae-brauniae was designated by Fernald (1942).

Ageratina luciae-brauniae occurs mostly on the floor of rockhouses (Wofford & Patrick, 1980; Campbell et al., 1991), but it also occurs at the base of a slight overhang (Wofford & Patrick, 1980). It grows in moist soil (Wofford & Patrick, 1980; Kral, 1983; Campbell et al., 1990) and in shaded to relatively well-lighted areas (Wofford & Patrick, 1980; Campbell et al., 1990). The species usually grows in deep soil, but has been observed "on ledges and among boulders with little sandy fill" (Campbell et al., 1991) and on a "large slab of sandstone" (Wofford & Patrick, 1980). Campbell et al. (1990) noted that the species does not have a preference for any particular aspect. It is restricted to sandstone.

Ageratina luciae-brauniae [n = 17 (Wofford, 1976)] is related closely to A. altissima (L.) King & Robinson [Eupatorium rugosum Houtt. (King & Robinson, 1970)] [2n = 34 (e.g., Grant, 1953)], which occurs in woods and clearings from western Nova Scotia to southeastern Saskatchewan, south to northwestern Florida and central Texas (Clewell & Wooten, 1971). Ageratina altissima occurs in the forest outside of rockhouses (Wofford & Patrick, 1980). Ageratina luciae-brauniae differs from A. altissima by its more delicate and slender stem, very thin leaves, the leaf blade being about as long as the petiole (vs. leaf blade longer than petiole) (Gleason & Cronquist, 1991), somewhat smaller heads, hairy-ribbed (vs. smooth) achenes (Kral, 1983), glabrous (vs. pubescent) stems, and cordate-deltoid (vs. ovate) leaves (Wofford, 1976). Braun (1940) remarked that the "whole plant is so delicate that it [A. luciae-brauniae] probably could not withstand the impact of heavy rain." The bicellular foliar secretory and tubular cavities in A. altissima (Curtis & Lersten, 1986; Lersten & Curtis, 1986) are similar to those in A. luciae-brauniae (N. Lersten, pers. comm.).

Clewell and Wooten (1971) included "luciae-brauniae" in the geographically highly variable Ageratina altissima var. altissima, since plants of the former taxon were "bizarre plants showing extreme symptoms of etiolation from growing under limestone [sic] ledges in Kentucky." However, Wofford (1976) concluded from a common garden experiment that the "delicate appearance of these plants is a genetically based adaptation and not simple etiolation." Plants of A. luciae-brauniae grown from seed in the greenhouse expressed the same phenotype as those in populations in nature (Wofford, 1976). The two species have identical chloroplast DNA restriction site patterns (Schilling, 1996), which indicates either recent divergence or recent gene flow through hybridization (E. Schilling, pers. comm.).

Most of the freshly matured achenes of Ageratina luciae-brauniae are nondormant; however, only a small percentage of the achenes germinate in darkness. Achenes cold stratified at 5 [degrees] C germinate to higher percentages in light and in darkness than do freshly matured achenes. Freshly matured achenes of A. altissima germinate only at high temperatures in light, and none germinate in darkness (i.e., they are conditionally dormant). Cold-stratified achenes also germinate to high percentages at low temperatures in light, but to only 0-1% in darkness. Achenes of A. luciae-brauniae, then, probably do not form a long-lived (persistent) soil seed bank, but those of A. altissima probably can. Peak germination of A. luciae-brauniae achenes occurred in autumn in an unheated greenhouse, whereas peak germination of A. altissima achenes occurred in spring (Walck et al., in press). Walck et al. (in press) concluded that achene germination characteristics do not explain why A. luciae-brauniae is an endemic nor why A. altissima is widespread.

Arenaria cumberlandensis Wofford & Kral [Minuartia cumberlandensis (Wofford & Kral) McNeill (McNeill, 1980; Wofford, 1981)], is a passive chamaephyte (cf. Ennis, 1928) perennial described in 1979, even though the first collections were made in 1941 (Wofford & Kral, 1979). The species is restricted to the watershed of the Big South Fork of the Cumberland River; it occurs in four adjacent counties in Tennessee and a neighboring county in Kentucky. Twenty-seven occurrences are known in Tennessee and one in Kentucky. In Tennessee, 20 of these are within 2 mi (3 km) of each other, and all 27 are within 25 mi (40 km) of each other. The number of plants at a site ranges from [less than or equal to]6 clumps to thousands of clumps (Wofford & Smith, 1980; U.S. Fish and Wildlife Service, 1996a).

This species grows on the floor and ledges of rockhouses and overhanging cliffs and in "solution pockets" on cliffs; it occurs only on sandstone. The most critical habitat requirements appear to be shade, moisture, relatively constant and low temperatures, and high relative humidity (Wofford & Smith, 1980).

No contiguous populations of a closely related, perennial Arenaria occur close to A. cumberlandensis (Wofford & Kral, 1979). Further, all other members of the genus in southeastern United States occur on nonshaded rock outcrops (Weaver, 1970). Arenaria cumberlandensis [n = 10 (Wofford & Kral, 1979)] is related most closely to A. groenlandica (Retz.) Sprengel var. groenlandica [n = 10 (Weaver, 1970)], a densely tufted perennial of arenaceous rock exposures from Greenland to Labrador, south to the crests of the Appalachian Mountains in Virginia, North Carolina, and Tennessee (Maguire, 1951; Wofford & Kral, 1979). In addition, A. groenlandica (variety undetermined) is disjunct to Brazil (Hulten, 1964). Arenaria cumberlandensis differs from A. groenlandica var. groenlandica in leaf morphology and general appearance, number of pedicels, seed morphology, and habitat (Wofford & Kral, 1979; Wofford, 1981); however, flowering times overlap between the two taxa (Wofford & Kral, 1979).

Arenaria cumberlandensis is sympatric with A. glabra Michx. [A. groenlandica var. glabra (Michx.) Fern.] [n = 10, 2n = 20 (Weaver, 1970)], which occurs on nonshaded sandstone rock outcrops on the Cumberland Plateau above rockhouses (Perkins, 1981). However, A. glabra is distinguished by the absence at flowering of a basal rosette, thinner and narrower leaves, life form (winter annual), and flowering time (Weaver, 1970; Wofford & Smith, 1980).

Solidago albopilosa Braun is a semi-rosette hemicryptophyte perennial with vegetative reproduction by rhizomes (Gibson, 1961; Andreasen & Eshbaugh, 1973). It occurs only along tributaries of the Red River, in three adjacent counties in eastern Kentucky (Campbell et al., 1989; U.S. Fish and Wildlife Service, 1993a). Ninety occurrences containing an estimated 45,000 stems were reported by the U.S. Fish and Wildlife Service (1993a).

The species occurs almost exclusively on the floors of rockhouses (Braun, 1942; Medley, 1980) with stable ceilings (U.S. Fish and Wildlife Service, 1993a) and of "sheltered cliffs" (Andreasen & Eshbaugh, 1973). Additionally, it grows on ledges in the ceiling and backwall of rockhouses (Medley, 1980) and has been found at the base of a sandstone boulder downslope from a rockhouse (J. Walck, pers. obs.). Solidago albopilosa occurs in rockhouses containing little leaf litter (Medley, 1980) and grows in almost all possible moisture regimes, aspects (Campbell et al., 1989), and light regimes (J. Walck, pers. obs.). It occurs at 800-1300 ft (244-396 m) a.m.s.1. (U.S. Fish and Wildlife Service, 1993a) and only on sandstone. Braun (1942) noted that the habitat was similar to, but drier than, that of Ageratina luciae-brauniae; the two species have not been found growing together.

Solidago albopilosa is morphologically similar to S. flexicaulis L., which occurs in woods from Nova Scotia to North Dakota and south to Georgia and Arkansas (Semple et al., 1984; Gleason & Cronquist, 1991). In the Red River Gorge area, S. flexicaulis occurs on slopes in the forest below rockhouses, especially on soils derived from limestone (U.S. Fish and Wildlife Service, 1993a). Andreasen and Eshbaugh (1973) found that S. albopilosa and S. flexicaulis are similar in leaf shape and size, inflorescence pattern, number of ray flowers, and trichome morphology. However, the rockhouse habitat, decumbent stems, white-pilose stems and lower leaf surfaces, and thinner leaves are important differences between S. albopilosa and S. flexicaulis (Braun, 1942; Andreasen & Eshbaugh, 1973). Braun (1942) remarked that the leaves of S. albopilosa are "so thin that coarse print is readable through [them]."

Fernald (1950) considered Solidago albopilosa an "ecological development of [S. flexicaulis]." Based on chromosome number and on garden transplants, Beaudry (1959) concluded that S. albopilosa (2n = 36) was a distinct species and not an ecad. However, Beaudry's (1959) argument of different chromosome numbers between the two taxa was based on incomplete data (Semple et al., 1984). Diploids of S. flexicaulis (2n = 18) occur east of a line running through eastern Ontario to New Jersey and eastern Maryland; tetraploids (2n = 36) occur west of this line (Semple et al., 1984). In eastern Kentucky, including one of the three counties in which S. albopilosa occurs, S. flexicaulis is tetraploid (Semple et al., 1993). Semple et al. (1984) concluded that "Solidago albopilosa could be treated as a variety or subspecies of S. flexicaulis without philosophically necessitating a merger of all the species of the [S. flexicaulis] complex into a single species."

Solidago albopilosa is cross compatible with S. flexicaulis and S. caesia L. [2n = 18 (Semple et al., 1984)], another species sympatric with S. albopilosa (E. Esselman, pers. comm.). Likewise, plants near the dripline appear to be intermediate between S. albopilosa and S. flexicaulis (Braun, 1942; Andreasen & Eshbaugh, 1973) or between S. albopilosa and S. caesia (Medley, 1980). Braun (1942) reported plants intermediate in leaf shape, inflorescence pattern, growth form, and degree of hairiness. Andreasen and Eshbaugh (1973) reported plants that were larger than typical S. albopilosa and had dark green, flexicaulis-like leaves. However, they also reported that plants in one population "inside the rockhouse are short and weak, and light green or almost yellow in color while the plants out in the open on the ledge, are larger and darker in color." Thus, intermediate plants may not be hybrids; they may be true S. albopilosa plants that exhibit a phenotypically plastic response to increased light. Population analysis indicated significant morphological differences among S. albopilosa plants of four sites (Andreasen & Eshbaugh, 1973).

Based on a lack of colonization of available habitats and high endemism, Andreasen and Eshbaugh (1973) concluded that Solidago albopilosa was of recent origin. They considered that if it were an old species, glaciation could account for its absence in rockhouses in Ohio, but not in Tennessee. Esselman (E. Esselman, pers. comm.) found that S. albopilosa is sufficiently genetically distinct from S. flexicaulis to indicate that it is not of very recent origin. However, it does appear that S. albopilosa is colonizing new rockhouses in its range. Today, about 3500 plants of S. albopilosa occur in a rockhouse (Medley, 1980) in which Andreasen and Eshbaugh (1973) reported no plants of this species in 1969.

Solidago albopilosa is self-incompatible (E. Esselman, pers. comm.) with fragrant flowers that attract bees, moths, and syrphid flies (Braun, 1942; Medley, 1980; U.S. Fish and Wildlife Service, 1993a). Achene production is about equal to that of other, more-common Solidago species (E. Esselman, pers. comm.). Several age classes can be found within populations. Seedlings occur at densities of 100-200+ [m.sup.-2], and it takes 2 years or longer for plants to flower (S. Francis, unpubl.). Herbivores have been noted to damage plants (U.S. Fish and Wildlife Service, 1993a).

Esselman (E. Esselman, pers. comm.) found that the species is not genetically depauperate, but levels of diversity are lower than those found in more-common Solidago species. Highest levels of genetic diversity are found among populations, and isolated populations are more genetically distinct than are those occurring in close proximity to others.

Thalictrum mirabile Small is a semi-rosette hemicryptophyte without runners (Gibson, 1961) that occurs in eastern and western Kentucky and possibly in northwestern Alabama (Boivin, 1944; Keener, 1976; Cranfill & Medley, 1981; C. Keener, pers. comm.). Thalictrum mirabile reported from Tennessee and Alabama by Kral (1966, 1973) is now considered to be T. clavatum (R. Kral, pers. comm.).

Thalictrum mirabile is difficult to distinquish from T. clavatum DC., which occurs in rich mountain woods, on cliffs and seepage slopes, and along mountain streams from western Virginia and southern West Virginia to eastern Kentucky, south to western South Carolina, northern Georgia, and northern Alabama (Keener, 1976, pers. comm.). Compared to T. clavatum, T. mirabile is more delicate and lax, and has longer achene stipes, straighter dorsal achene margins, and shorter filaments (Small, 1900; Keener, 1976).

Keener (1976) concluded that "Population studies are needed to assess the comparative differences between this species [Thalictrum mirabile] and T. clavatum." In a letter to J. Walck dated 15 June 1993, R. Kral stated that T. mirabile is not distinct from T. clavatum since "characters [achene margin shape and achene stipe length] run together in the same populations in AL and TN, and also in my NC material." In southeastern Kentucky, Palmer-Ball et al. (1988) reported that the two species did not appear distinct but that the plants generally had "mirabile"-type seeds. Further, Boivin (1944) observed, "Fruits and stamens of the Kentucky specimens tend to be larger [than those of T. mirabile from Alabama], but all characters overlap to some extent."

If Thalictrum mirabile is a good species, Keener (pers. comm.) would consider T. clavatum [n = 7 (Jensen, 1944)] to be its closest relative. However, if these two species are lumped, then the closest relative of T. clavatum may be in eastern Asia. Boivin (1944) states that T. clavatum "evidently belongs to the same section [Physocarpum] as all these eastern Asiatic species but it is not especially closely related to any of them. As far as it is possible to judge, with only the original description at hand, the fruits of T. tenerum Huth might resemble those of T. clavatum more than those of any other species except T. mirabile Small."

Cranfill and Medley (1981) considered Thalictrum mirabile to be endemic to sandstone rockhouses. The species occurs on the floor and is most abundant around plunge basins (Palmer-Ball et al., 1988), groundwater seeps/springs, and at the heads of streams; however, it also occurs on wet cliffs with slight overhangs. Rarely does the species occur outside the dripline of a rockhouse (J. Walck, pers. obs.). Thalictrum clavatum has been reported as an associate of other rockhouse endemics in eastern Kentucky (Medley, 1980; Campbell et al., 1992), Tennessee (Wofford & Patrick, 1980; Wofford & Smith, 1980; Kral, 1983), and Alabama (Gunn, 1991).

Fifteen of the 18 populations of Thalictrum clavatum (probably T. mirabile) surveyed by Melampy and Hayworth (1980) in eastern Kentucky, had [less than]50 individuals. Plants were growing only in shaded (2.3% [+ or -] 0.8% full sunlight), moist sandy (58% [+ or -] 14% soil moisture), and generally low soil-nutrient conditions; in 17 populations the substrate was wet sandstone. Apparently, plants are not capable of growing where irradiances are much higher than about 6-10% of full sunlight; at high irradiances, leaves turn brown and little growth occurs (Melampy & Hayworth, 1980). Melampy and Hayworth (1980) noted very few other vascular plants growing in the habitat of T. mirabile.

Thalictrum mirabile/clavatum is autogamous but not apogamous. Outcrossing probably is due to visits by syrphid flies [Sphegina sp(p).], but flower visitation rate is only 0.09 visits per [flower-hour.sup.-1]; the plant is nectarless. Biomass allocation to seeds is similar to that in T. dioicum L. and T. polygamum Muhl. [T. pubescens Pursh (Keener, 1976)], but seeds are smaller; root/shoot ratio in T. mirabile/clavatum is 0.8 (Melampy & Hayworth, 1980). Melampy and Hayworth (1980) suggested that autogamy may have evolved in this species because of the unreliability of insect pollination. In addition, to ensure its chances of colonizing suitable but patchy habitat, the species produces a large number of seeds by fertilizing and maturing practically all ovules and by producing small seeds, thereby maximizing the number of seeds per unit of energy expended. However, the narrow habitat requirements of the species may be due to autogamy (i.e., high inbreeding): genes necessary to exploit different habitats may have been eliminated from the gene pool (Melampy & Hayworth, 1980).

2. Holoschizoendemics

Dodecatheon frenchii, Heuchera parviflora vat. parviflora, Silene rotundifolia, Thelypteris pilosa var. alabamensis, and Trichomanes boschianum are classified as holoschizoendemics. The origin of D. frenchii is problematic; it may have evolved via polyhaploidy (Olah & DeFilipps, 1968) or by vicariance. Until more studies are done, we will be conservative and consider D. frenchii a holoschizoendemic with a possible vicariad taxon in western United States. Heuchera parviflora vat. parviflora and S. rotundifolia are sympatric with their closest relatives and are considered "mature" holoschizoendemics that evolved via adaptive radiation.

Thelypteris pilosa var. alabamensis and Trichomanes boschianum may have evolved by vicariance, i.e., divergent evolution in which geographic isolation is important, and genetic drift and weak selection result in the establishment of distinctive characters in populations that occupy similar habitats as their parents (Bramwell, 1972). Both taxa are ancient and thus could be classified as palcoendemics; however, each has a possible vicariad taxon in tropical America. The biogeographic histories of T. pilosa var. alabamensis and T. boschianum are similar to those of Trichomanes intricatum and Vittaria appalachiana. Even though T. pilosa var. alabamensis has a narrow geographic range, it probably is best to consider it an ancient holoschizoendemic.

Dodecatheon frenchii (Vasey) Rydb. [D. meadia L. var. frenchii Vasey, D. meadia L. subsp. membranaceum R. Knuth (Mohlenbrock, 1978)] is a rosette hemicryptophyte perennial without runners (Hansen, 1952) that occurs from south-central Indiana (two adjacent counties) to southern Illinois (six adjacent counties), east-central Missouri (one county), and western Kentucky (seven adjacent counties) (Harker et al., 1981). Disjunct populations occur in northeastern Kentucky [two counties (Campbell et al., 1992)], northwestern Alabama [one county (Timme & Lacefield, 1991)], and northern Arkansas [two counties (Smith, 1988)]. The species does not occur in Wisconsin as reported by Fassett (1944) and others (Iltis & Shaughnessy, 1960). Further, the variation in interpretation of its range [e.g., "Ill.-Minn.-Ark.-Pa. (?)" (Rydberg, 1932)] was attributed to the inclusion of plants belonging to D. meadia L. var. genuinum Fassett (Fassett, 1944) [D. meadia subsp. meadia (Thompson, 1953)].

Dodecatheon frenchii grows on the floor and on ledges in rockhouses (Olah & DeFilipps, 1968; Mohlenbrock, 1978; Harker et al., 1981; Cranfill, 1991) and at the bases of cliffs with little or no overhang (Tucker, 1984). The species is found mostly on mesic north- and east-facing aspects (Voigt & Swayne, 1955; Tucker, 1984); it occurs only on sandstone. In southern Illinois, D. frenchii has been reported in open, well-lighted upland woods 50 yd (46 m) from the cliff (Voigt & Swayne, 1955; Olah & DeFilipps, 1969). The species occurs at ca. 420-1700 ft (128-518 m) a.m.s.l. (Swayne, 1973; Tucker, 1984).

Dodecatheon frenchii is sympatric with D. meadia, which can occur on the summit of cliffs above rockhouses (Fassett, 1944; Olah & DeFilipps, 1968). The geographic range of D. meadia is from Washington, DC, to western Wisconsin and south to Georgia and Texas. Its habitat includes meadows, open woods, mesic and dry prairies, wooded to open bluffs or cliffs (Klotz & Walck, 1993), and a rockhouse in Ohio (Wolfe, 1951). Dodecatheon frenchii is smaller, paler, and more delicate than D. meadia and has thinner leaves (Olah & DeFilipps, 1968) that are cordate or abruptly narrowed to the petiole, rather than tapering as they do in D. meadia (Fassett, 1944).

Fassett (1944) reported that when grown under increased light and longer photoperiod, the leaf shape of Dodecatheon frenchii resembled that of D. meadia; however, when grown under reduced light, the leaf shape of D. meadia did not resemble that of D. frenchii. He concluded that plasticity in leaf shape was due to reduced light and not to dripping water, with which D. frenchii often is associated. However, Voigt and Swayne (1955) showed that leaves of D. frenchii remained distinct over three growing seasons in a common garden experiment under various light regimes and in a reciprocal transplant experiment, whereas in shaded conditions leaves of D. meadia were thinner than were those in nonshaded conditions. After 30 years, D. frenchii plants in the transplant experiment retained the original leaf character (Mohlenbrock, 1987). In addition, retention of leaf characteristics on plants of D. frenchii growing in an open, well-lighted upland woods show that reduced lighting is not required for this leaf shape to be expressed (Voigt & Swayne, 1955), although these plants have larger leaves than those found in rockhouses (Olah & DeFilipps, 1969).

Based on morphological, ecological, and chromosome-number differences between Dodecatheon frenchii and D. meadia, Olah and DeFilipps (1968) concluded that D. frenchii was a distinct species. Dodecatheon frenchii is diploid (n = 22) and D. meadia tetraploid (n = 44). In addition, the two taxa must be reproductively isolated, since it is not possible to cross Dodecatheon plants with different chromosome numbers (Thompson, 1953).

Two possible scenarios may explain the origin of Dodecatheon frenchii. Olah and DeFilipps (1968) concluded that this species is a polyhaploid derived from the autopolyploid D. meadia by abrupt ecospeciation. However, Stebbins (1980) argued that diploid populations may be derived from neighboring tetraploids by polyhaploidy only if the tetraploids are "unaltered autotetraploid strains." Otherwise, diploids are much more likely to be relictual disjuncts evolved from other diploid species.

Iltis and Shaughnessy (1960) suggested that Dodecatheon frenchii is related more closely to the D. pulchellum (Raf.) Merr. complex [D. radicatum Greene (Welsh et al., 1993)] of western North America than to D. meadia. Dodecatheon radicatum subsp. radicatum is diploid (n = 22), but tetraploids (n = 44) and hexaploids (n = 66) are reported in the complex (Beamish, 1955). The similarity of D. frenchii to D. pulchellum was based on the preference of D. frenchii for moist, rocky habitats; its geographic distribution south of the Wisconsinan Glacial Boundary; a more delicate habit, fewer-flowered inflorescences, and much thinner capsules than D. meadia; and general appearance. However, fruit shape, sepal length, and flower color of D. frenchii were more similar to D. meadia than to D. pulchellum (Iltis & Shaughnessy, 1960). Fassett (1944) and Thompson (1953) treated "frenchii" as a variety of D. meadia, which has a thick capsule wall.

The distribution, and in some respects the habitat, of Dodecatheon frenchii is similar to that of D. amethystinum (Fassett) Fassett [2n = 88 (Kawano, 1965)], which consists of widely disjunct populations in eastern United States that occur mostly on rock outcrops (Klotz & Walck, 1993). Likewise, D. amethystinum is more closely related to D. radicatum than to D. meadia (Fassett, 1944; Iltis & Shaughnessy, 1960; but see Ugent et al., 1982), and D. amethystinum has been included in D. radicatum (Thompson, 1953; Gleason & Cronquist, 1991).

Dodecatheon frenchii probably originated prior to Illinoian Glaciation, and populations were destroyed where they were covered by glacial ice or melt-water lakes. Since the Illinoian Stage, the species has not extended its geographic range northward (Swayne, 1973; see also Fassett, 1944). The distributional history of D. frenchii may be similar to that of D. amethystinum. As suggested by Iltis and Shaughnessy (1960), "D. frenchii may represent a population which, like D. amethystinum, migrated east from the western mountains under more favorable (i.e. glacial?) conditions and has since become isolated and restricted to local and specialized habitats." Schwegman (1984) proposed that D. amethystinum arrived in Illinois from western North America during the Pleistocene and that present-day populations of this species are relicts of a former widespread distribution during the Sangamonian Stage or during the Wisconsinan Stage. Disjunct occurrences of D. frenchii may be adventive, of independent genetic origin (Swayne, 1973), or relict.

Swayne (1973) suggested that the limited range and restricted habitat of Dodecatheon frenchii is due to its ecological adaptations, inability to compete successfully, limited occurrence of continuous suitable habitat, and glacial destruction of a previously widespread distribution. Voigt and Swayne (1955) reported no appreciable differences in soil organic matter, moisture content, or pH between the habitats of D. frenchii and D. meadia in southern Illinois.

About 50 populations of Dodecatheon frenchii are known, with the majority in unglaciated southern Illinois. Populations are small ("about 25 to 75 or more" individuals), but several age groups are present; however, population size fluctuates. Density of individuals in most populations is sparse, and seedling density is 1-12 [m.sup.-2] (Tucker, 1984). The floor of a rockhouse in which D. frenchii grows is almost completely bare of leaf litter, and few other species are present (Tucker, 1984; Swayne, 1985; Timme & Lacefield, 1991). Thus the species could be characterized as a "pioneer" (Tucker, 1984). Tucker (1984) suggested that intraspecific root competition in D. frenchii may be greater than intraspecific shoot competition, because of shallow, dry soil in rockhouses, and that interspecific competition would be detrimental.

Other species of Dodecatheon are almost exclusively insect pollinated (e.g., Macior, 1970); however, Tucker (1984) observed few insects (small members of Diptera and small bees) visiting flowers in an Arkansas population of D. frenchii. Dodecatheon frenchii has high pollen fertility, but developmental irregularities in sporogonial tissue, resulting in sterile anthers, have been observed (Olah & DeFilipps, 1968). In all species of Dodecatheon, vegetative reproduction occurs by branching and separation of the caudex and by separation of roots from the caudex (Thompson, 1953). The extent to which clones develop in D. frenchii is unknown.

Heuchera parviflora Bartling var. parviflora [H. rugelii Shuttlw. ex Kunze; H. parviflora vat. rugelii (Shuttlw. ex Kunze) Rosend., Butt. & Lak.; H. missouriensis Rosend. (Wells, 1984)] is a rosette hemicryptophyte perennial without runners (Hansen, 1952; Gibson, 1961). This is the only eastern Heuchera species that is not evergreen (Wells, 1984). It ranges from southern West Virginia and south-central Ohio to southern Indiana, southern Illinois, and southeastern Missouri, south to northern Georgia and Alabama (Furr, 1971; Wells, 1984).

This taxon grows on the floor and on ledges and in crevices of rockhouses and on ledges and in crevices of cliffs (Mohr, 1901; Wherry, 1939; Winterringer, 1949; Gillespie, 1956; Wells, 1984; Schmalzer et al., 1985; Cranfill, 1991; Tobe et al., 1992). It prefers shade (Mohr, 1901; Wells, 1984) and north-facing aspects (Winterringer, 1949). The typical variety occurs mostly on sandstone, but it has been recorded on granite in Georgia, gneiss in North Carolina (Furr, 1971), and limestone in Kentucky (Martin et al., 1979) and Missouri (Steyermark, 1934; Rosendahl, 1951). In a survey of vegetation on sandstone ledges in southern Illinois, Heuchera parviflora was present at 44% of the sites (Winterringer, 1949). At six sites it was "sparse," and at one site "very sparse"; at five sites it had a frequency of 1-20%, and at two sites a frequency of 21-40%.

A second variety of Heuchera parviflora, var. puberula (Mack. & Bush) Wells, occurs mostly in Arkansas and Missouri, but it has been collected in western Kentucky and southern Indiana. The typical variety differs from var. puberula by longer, less dense pubescence and by teeth on floral bracts (Furr, 1971; Wells, 1984). The variety grows in similar habitats as var. parviflora, but most occurrences are on limestone, with a few on dolomite and sandstone (Rosendahl, 1951; Steyermark, 1934, 1963; Furr, 1971; J. Logan, pers. comm.).

Heuchera parviflora [2n = 14 (Soltis, 1980)] is related closely to H. villosa Michx. [2n = 14 (Soltis, 1980)] (Wells & Bohm, 1980; Wells, 1984; but see Rosendahl et al., 1936), which occurs on shaded rocks, ledges, and shallow rocky soil. Two varieties of H. villosa are recognized. Variety villosa ranges from western Virginia, southern Ohio, and southern Indiana south to northern Georgia, northern Alabama, and northeastern Mississippi, with a disjunct population in western New York. Variety arkansana (Rydb.) E. B. Smith occurs in Arkansas. Heuchera parviflora differs from H. villosa in having rounded (vs. triangular) leaf lobes; oblanceolate, reflexed (vs. linear, coiled) petals; smooth (vs. echinate) seeds; longer pedicels and internodes of floral branches; and slightly shorter free hypanthia (Wells, 1984). Leaves of H. parviflora are rounder and thinner than those of H. villosa (Fernald, 1950). Isozyme data suggest a high degree of similarity between H. parviflora and H. villosa (Soltis, 1985).

Heuchera parviflora and H. villosa are moderately interfertile when artificially crossed (Wells, 1979). The two species are sympatric, and they flower at the same time. However, Wells (1984) suggested that they do not hybridize in nature because of spatial barriers and ecological differences: H. villosa grows chiefly on granite and gneiss, and H. parviflora is restricted to heavily shaded ledges and rock undercuts on sandstone. However, hybrids have been observed in Tennessee (Furr, 1971) and in Ohio and Kentucky (A. Cusick, pers. comm.). In addition, H. villosa occurs on sandstone boulders on the colluvial slopes in the forest below rockhouses (J. Walck, pers. obs.) and on sandstone cliffs close to rockhouses (J. & C. Baskin, pers. ohs.).

Wells and Bohm (1980) suggested that Heuchera parviflora and H. villosa were derived from H. micrantha Douglas [2n = 14 (Soltis & Kuzoff, 1995)], a species of western North America, or from a H. micrantha-like ancestor. A nuclear ribosomal DNA phylogeny supports the close relationship of the three species (Soltis & Kuzoff, 1995; D. Soltis, pers. comm.). Heuchera parviflora and H. villosa may have had separate origins in the west and then spread eastward independently, or H. parviflora could have diverged from a H. villosa-like prototype after migrating to the east. Geographical and ecological isolation probably was essential for speciation. Isolation of the eastern Heuchera species may have occurred before the Pliocene uplift of the Rocky Mountains (Wells, 1984).

Heuchera parviflora shows adaptation to its habitat in flavonoid composition, morphological characters, and type of breeding system. There have been 24 flavonoids identified in H. parviflora var. parviflora, whereas 44 have been identified in H. villosa var. villosa (Wells & Bohm, 1980). Wells and Bohm (1980) suggested that lower diversity of flavonoids in H. parviflora is an adaptation to a more severe ecological setting and tighter energy budget. In addition, leaves of H. parviflora are reddish beneath (Rosendahl et al., 1936), which may prevent photoinhibition and increase maximum photosynthesis (cf. Gould et al., 1995). Seeds of H. parviflora are smooth and not echinate like all other eastern Heuchera species, and the leaves in the inflorescence are viscid from an exudate secreted by gland-tipped trichomes (Wells, 1984). Wells (1984) observed that seeds, when shed, adhere to the viscid plant body and remain on the ledge with the parent plant, instead of being dispersed into the nearby forest. Loss of spines increases the surface area of the seed that contacts the exudate. Further, unlike other eastern species of Heuchera, H. parviflora is self-compatible. This allows sexual reproduction between an individual and its own progeny sharing a ledge, which may be essential for significant seed production. All eastern species of Heuchera are protandrous. (Additional information on anatomy and morphology of H. parviflora is given in Wells, 1984.)

Silene rotundifolia Nutt. [Melandryum rotundifolium (Nutt.) Rohrb. (Rohrbach, 1868)] is a semi-rosette wintergreen hemicryptophyte perennial without runners (Beatley, 1956; Gibson, 1961) that ranges from western West Virginia and south-central Ohio to western Kentucky, south to western Virginia, northwestern Georgia, and northern Alabama (Hitchcock & Maguire, 1947; Heaslip, 1948). It grows on the floor and on ledges and in crevices of rockhouses and on ledges and in crevices of cliffs (Heaslip, 1948; Andreasen & Eshbaugh, 1973; Campbell & Meijer, 1989; Schmalzer et al., 1985; Cranfill, 1991; Cusick, 1994). Gilbert (1938) and Caplenor (1955) observed the species growing under or on boulders. The species is associated with dry soil conditions (Gilbert, 1938; Campbell & Meijer, 1989). The substrate may be limestone, but it is predominantly sandstone (Heaslip, 1948; Fernald, 1950).

Even though plants of this species inhabit areas of high light that appear dry and seem to receive little precipitation, water is not a limiting factor. The tap root of Silene rotundifolia often exceeds 6 ft (1.8 m) in length, and thus it reaches water in joints and bedding planes (Heaslip, 1948; Wolfe et al., 1949). Temperatures are less extreme in rock crevices with S. rotundifolia than temperatures in the surrounding environment: Mean minimum and maximum weekly temperatures for one year in crevices with S. rotundifolia were 8 [degrees] and 16 [degrees] C, respectively, whereas temperatures outside of crevices were 1 [degree] and 23 [degrees] C, respectively [converted from [degrees]F (Wolfe et al., 1949)].

The species is related most closely to, and probably derived from, Silene virginica L., which occurs in rich or open woodlands, thickets, and rocky slopes from central New York to southeastern Minnesota, south to Georgia and southeastern Oklahoma. Both species have "similar large scarlet flowers, pubescence, and peculiar saccate appendages below which is a conspicuous clear area [extending down the claw]." However, S. rotundifolia "has become a weak trailing, more viscid plant, with broader leaves [than S. virginica], and is apparently restricted to a drier, more exposed habitat" (Hitchcock & Maguire, 1947). Although the tetraploids S. rotundifolia (2n = 48) and S. virginica (2n = 48) can hybridize, the hybrid is sterile (Heaslip, 1950, 1951; Kruckeberg, 1963). Hybrids never have been observed in nature (Heaslip, 1948).

Silene rotundifolia also is cross fertile with the eastern North American species S. regia Sims (2n = 48) and the western North American species S. laciniata Cav. (2n = 48, 96); however, [F.sub.1] hybrids are sterile (Heaslip, 1951; Kruckeberg, 1963). Heaslip (1951) concluded that S. rotundifolia, S. virginica, S. regia, and S. laciniata were interrelated through common ancestors, and that each could be classified as an ecospecies belonging to the same cenospecies. These species or their ancestors probably originated as allopolyploids, rather than as autopolyploids (Heaslip, 1951). In addition, Hitchcock and Maguire (1947) suggested that S. virginica is related closely to the western species S. californica Durand [2n = 48, 72, 96 (Kruckeberg, 1954, 1960)], which is related closely to S. laciniata (Wilken, 1993).

Flowers of Silene rotundifolia and S. virginica are protandrous (Heaslip, 1950; Fenster et al., 1996) and are pollinated by ruby-throated hummingbirds [Archilochus colubris (L.)] (Austin, 1975). In addition, syrphid flies and small bees visit flowers of S. virginica (Fenster et al., 1996). Heaslip (1950, 1951) suggested that in both species self-pollination occurs; however, Fenster et al. (1996) reported that S. virginica does not autogamously self-fertilize, although it is self-compatible (Antonovics et al., 1996).

Plants of Silene rotundifolia require vernalization to flower (Heaslip, 1948; Wolfe et al., 1949), but they are day-neutral (Heaslip, 1948). Seeds are only slightly dormant, and Heaslip (1948) obtained 65% germination in light and 66% in darkness at room temperature; cold treatment (1 [degrees] C for 7 days) did not affect total germination percentages in either light or darkness. Seeds of S. virginica were more dormant than those of S. rotundifolia. In fact, only those with broken (scarified) seed coats germinated readily. However, for both species some afterripening may have occurred in the laboratory before the seeds were tested. More recent studies (J. Walck et al., unpubl.) show that S. rotundifolia seeds are conditionally dormant (sensu Baskin & Baskin, 1985) at maturity, and that cold-stratified seeds germinate to higher percentages in light and in darkness than do nonstratified ones. Seeds of S. rotundifolia have been tested for the possibility of cryopreservation, but those immersed in liquid nitrogen before cold stratification did not germinate, even though the control (cold stratified, no nitrogen) seeds did (Pence, 1991).

Heaslip (1948) suggested that lack of seed dormancy was one of the factors contributing to the limited distribution of Silene rotundifolia. She hypothesized that "seeds [such as those of S. virginica] which remain viable and germinate gradually over a long period of time are more apt to yield more seedlings which yield more plants than are those [such as S. rotundifolia] which germinate rapidly over a short period of time." However, this probably is not the case, since freshly matured seeds of some widespread species are nondormant (Baskin & Baskin, 1988).

Thelypteris pilosa (Mart. & Gal.) Crawford var. alabamensis Crawford [Leptogramma pilosa (Mart. & Gal.) Underw. var. alabamensis (Crawford) Wherry (Wherry, 1964); Stegnogramma pilosa (Mart. & Gal.) Iwatsuki var. alabamensis Iwatsuki (Iwatsuki, 1964)] is a perennial, evergreen fern with rhizomes that produces spores year-round; it is the only taxon of Thelypteris in southeastern United States lacking an indusium (Crawford, 1951; Kral, 1983). The taxon first was discovered in 1949, along the West Sipsey Fork of the Black Warrior River in Winston County, Alabama. The type locality was destroyed by bridge construction and by flooding associated with the Lewis Smith Dam. It was rediscovered upstream in 1972, and presently occurs along a 3.25 mi (5.25 km) segment of the West Sipsey Fork in Winston County, Alabama (Short & Freeman, 1978; Gunn, 1991). The number of plants in 15 known populations ranges from 1 to 6500; seven populations have [less than or equal to] 12 individuals, but one has ca. 1500 individuals and another 6000. The other populations contain 20-600 individuals (Gunn, 1991).

Plants are rooted in crevices or fissures in sandstone or directly to rough rock surfaces on the ceilings of rockhouses, on ledges beneath overhangs, and on flat, exposed cliff faces. Population sites vary in aspect, moisture (very wet to very dry), and amount of light (shaded to exposed). In addition, a few sites are very near water, where they are likely to be inundated during seasonal floods (Gunn, 1991). The habitat is characterized by a combination of high relative humidity, high substrate moisture, and shade (Kral, 1983).

The typical variety of Thelypteris pilosa (n = 36) ranges from northwestern Mexico to Honduras (Smith, 1981). Variety alabamensis is smaller than the typical variety, and it has a linear (vs. lanceolate) leaf blade, spreading (vs. ascending) pinnae with rounded (vs. acute to acuminate) tips, sinuses of the pinnule margins reached by one (vs. two) lateral vein, and a free lobe at the base of basal pinnae (Kral, 1983; Smith, 1993). When first described, one specimen each collected from Sonora and Chihuahua, Mexico, ca. 2000 km from the type locality in Alabama, was identified as var. alabamensis (Crawford, 1951; Short & Freeman, 1978; Smith, 1993). Although Mickel and Smith (U.S. Fish and Wildlife Service, 1992) concluded that the Alabama plants were distinct from all Mexican material, Smith (1993) included the Sonoran and Chihuahuan specimens in var. alabamensis.

The population of Thelypteris pilosa var. alabamensis in Alabama could have arisen via long-distance dispersal of spores from Mexico or Central America. More likely, however, the existing populations are remnants of T. pilosa that had a wider distribution in the early Tertiary, when tropical species occurred in southeastern United States (cf. Graham, 1993).

Trichomanes boschianum Sturm [Vandenboschia radicans (Sw.) Copeland in part (Copeland, 1938)] is a perennial, evergreen, rhizomatous fern (Wherry, 1957; Mohlenbrock & Voigt, 1959; Gleason & Cronquist, 1991) that ranges from central West Virginia and south-central Ohio south to western South Carolina, northeastern Georgia, northern Alabama, and northeastern Mississippi, and from south-central Indiana and southern Illinois to western Kentucky (Lowe, 1921; Strausbaugh & Core, 1978; Duncan & Kartesz, 1986; Farrar, 1990). Disjunct populations occur in northern Arkansas and Chihuahua, Mexico (Farrar, 1993a). Fertile diploids (2n = 72), tetraploids (2n = 144), and sterile triploids of this species have been identified: Diploids occur in Arkansas, Illinois, and western Kentucky, and polyploids occur primarily east of these states (Farrar, 1971, 1993a).

Trichomanes boschianum is epipetric and occurs mostly on the backwall of sandstone rockhouses and overhangs (e.g., Wherry, 1939; Reed, 1951; Evers, 1961; Dean, 1964), but it also has been recorded on a vertical cliff (Darling, 1955), in a crevice of a cliff (J. Schwegman, pers. comm.), and on the floor at the backwall of a rockhouse (Evers, 1961). Schwegman (1982) reported that it grows on sandstone around the vertical entrance to a limestone pit-type cave. In North Carolina, the species is recorded on granitic gneiss (Coker, 1938), and in South Carolina on schist and gneiss (Tobe et al., 1992). Coker (1938) reported that the pH of the granitic gneiss was 5.45 on the surface and 6.54 in the interior. The species grows at 560-3000 ft (171-914 m) a.m.s.l. (Mohlenbrock & Voigt, 1959).

In Indiana, Trichomanes boschianum occurs in moist (but not wet) sites and in habitats receiving slightly more light than those of Vittaria appalachiana (Yatskievych et al., 1987). Mohlenbrock and Voigt (1959) suggested that "the leaves of the filmy fern [T. boschianum] are adapted to moist shaded conditions by their one-celled thickness and absence of stomates." Mortality due to low temperatures (Cranfill, 1980; Yatskievych et al., 1987) and sporadic droughts (Farrar, 1993a) have been reported.

Unlike Trichomanes intricatum, sporophytes and gametophytes of T. boschianum occur in nature. In addition, gametophytes of T. boschianum may exist independently of the sporophyte; e.g., in Arkansas, gametophytes occur up to 50 km from the sporophyte (Farrar, 1992). Although the gametophytes cannot be distinguished morphologically, isozyme patterns of the two species are distinctive (Farrar, 1985, 1992). Sporophytes of T. boschianum are "very slow" growing, but growth is continuous throughout the year. The species seldom shows evidence of sexual reproduction, even though gametophyte colonies may be found in the vicinity of fertile sporophytes (Farrar, 1971, 1993a). In addition, Farrar (1971) noted that reproduction by spores, if it occurs, probably is limited to diploid populations, since the spores of polyploid populations appear to be abortive. Asexual reproduction occurs by gemmae (Farrar, 1993a), rhizomes (Gleason & Cronquist, 1991), and vegetative budding (Farrar, 1971). Vegetative buds are formed on leaves that have come into contact with a moist substrate. The buds differentiate into a rhizome and first-leaf, and then they are detached from the parent plant and dispersed.

Photosynthesis and respiration rates were measured on sporophytes of Trichomanes boschianum at different illuminances and temperatures (Farrar, 1971). Photosynthetic rate was light-saturated at ca. 500 ft-c (5380 lux), and the light compensation point was 30 ft-c (323 lux). Rate of photosynthesis was highest at 15 [degrees] to 25 [degrees] C; however, Farrar (1971) concluded that it may be relatively constant throughout the year. In summer, when temperatures are optimum, photosynthesis is limited to ca. 30% of the maximum rate by low illuminance. In winter, when illuminance is optimum, photosynthesis is limited to ca. 40% of the maximum rate by low temperatures. Thus, the effect of low temperatures in winter is compensated by an increase in illuminance due to the absence of a leaf canopy in the surrounding forest. In addition, photosynthesis in T. boschianum is characteristic of shade plants [i.e., low compensation point, low light-saturation point, and low maximum rate of photosynthesis (Larcher, 1995)].

Trichomanes boschianum probably is a relict that evolved from a tropical taxon that occurred in southeastern United States during the Early Tertiary (cf. Graham, 1993). Swayne (1973) concluded that T. boschianum is a relict of the Eocene. It most closely resembles T. radicans Swartz (n = 36), a species of the New and Old World tropics (Smith, 1981). Wherry (1961) regarded T. boschianum as an unnamed variety of T. radicans, since T. boschianum differed from T. radicans only in its smaller size. In addition, diploid plants of T. boschianum tend to be smaller and less robust than polyploids of this species, and thus polyploids more closely resemble T. radicans (Farrar, 1971). Recent authors have agreed that T. boschianum is a distinct taxon endemic to North America (Farrar, 1971, 1993a).

X. Conservation Biology

Three of the 11 endemic taxa are listed as federal endangered or threatened: Arenaria cumberlandensis as endangered (U.S. Fish and Wildlife Service, 1996a) and Solidago albopilosa and Thelypteris pilosa var. alabamensis as threatened (U.S. Fish and Wildlife Service, 1992, 1993a). Further, Ageratina luciae-brauniae previously was considered a "C2" candidate for listing as endangered or threatened (U.S. Fish and Wildlife Service, 1993b, 1996b). In addition, several of the endemic taxa are state-listed.

Damage from hiking, camping, and rock climbing/rappelling (e.g., trampling, smoke and heat from campfires); digging for prehistoric artifacts; flooding caused by dam construction; and collection by scientists and wildflower enthusiasts threaten the endemic taxa (Tucker, 1984; U.S. Fish and Wildlife Service, 1993a, 1996a). In addition, weedy competitors, although not abundant in rockhouses, occur with some endemics, e.g., Microstegium vimineum with Solidago albopilosa; however, their effects on the endemic taxa are unknown.

The ranges of six endemic taxa are almost exclusively in recreational areas (e.g., state park, national river and recreation area, national forest). Although some populations of these taxa occur in isolated sites not easily accessible, others occur in rockhouses heavily visited by tourists. In addition, many of the endemics occur in small populations. Archaeological surveys in the Daniel Boone National Forest in Kentucky (Wyss & Wyss, 1977) and in the Big South Fork National River and Recreation Area in Kentucky and Tennessee (Ferguson & Gardner, 1986) showed that in ca. 35% of the rockhouses, 0-10% of the floor area in the rockhouse was disturbed; in ca. 35%, 10-70% was disturbed; and in ca. 30%, [greater than] 70% was disturbed. Ease of access, rather than visibility of rockhouses, was correlated significantly with degree of disturbance (Ferguson & Gardner, 1986). Population declines due to recreation have been noted for Trichomanes boschianum (Cranfill, 1980), Ageratina luciae-brauniae (Palmer-Ball et al., 1988), and Solidago albopilosa (U.S. Fish and Wildlife Service, 1993a). On the other hand, S. albopilosa recolonized disturbed areas once they were left undisturbed (U.S. Fish and Wildlife Service, 1993a).

Timbering or other large-scale activities (e.g., mining) may change the rockhouse environment by altering water and nutrient flow, and the cutting of trees directly in front of rockhouses could alter temperature, humidity, and light conditions (U.S. Fish and Wildlife Service, 1992, 1996a). Population extinctions of Trichomanes boschianum have been attributed to changes in the rockhouse environment caused by logging (Shaver, 1954). Current policy in the Daniel Boone National Forest in Kentucky prohibits logging from the base of the cliff downslope to the nearest break in the slope or to 50 ft (15 m) from the dripline; the forest above the cliff is maintained as needed to protect the site's hydrologic characteristics (U.S. Fish and Wildlife Service, 1993a). However, populations of endemic plants occur on private property receiving no protection.

XI. Summary

Sandstone rockhouses, an unusual plant habitat in eastern United States, are semicircular recesses that extend far back under cliff overhangs and are large enough to provide shelter for humans. In eastern United States, most of them occur in gorge-type topography in the Appalachian Plateaus, Interior Low Plateaus, and Ozark Plateaus physiographic provinces. Formed mostly in Mississippian- and Pennsylvanian-age rocks, rockhouses vary in size but are largest at heads of gorges. Lateral erosion of basal or intercalated, less resistant rocks forms a recess during stream/valley incision; rockfall and attrition contribute to enlargement of the recess. Most rockhouses occur in stream valleys cut during the Pleistocene. They have been impacted heavily by humans since Paleo-Indian time.

Three distinct habitats occur in rockhouses: ceiling, backwall, and floor. The interior of rockhouses is shaded and, compared to habitats outside of rockhouses, has higher relative humidities and lower rates of evaporation and is warmer during winter and cooler during summer. Although rockhouses are protected from direct precipitation, water enters them primarily by groundwater seepage and by dripping from the ceiling. Soil consists mostly of sand, and pH generally is low. High levels of some nutrients are associated with saltpeter earth or with ecofactual and artifactual remains left by human occupants of rockhouses.

Algae and bryophytes are common on the backwall. Vascular plants from habitats in the vicinity of rockhouses occur with endemic taxa on the floor and backwall. A total of 139 vascular plant taxa have been recorded as associates of the rockhouse endemics Ageratina luciae-brauniae and/or Solidago albopilosa. Most of the taxa are native C3 phanerophytes or hemicryptophytes that occur sporadically among rockhouses. The flora has a higher percentage of phanerophytes and lower percentages of hemicryptophytes and therophytes than the Kentucky state flora. Similarities in species composition among population sites of A. luciae-brauniae and of S. albopilosa are low, suggesting that rockhouses support many different assemblages of plant species. However, the endemic taxa have a much greater tendency to form a distinct plant assemblage than does the flora as a whole.

Eleven taxa belonging to 8 families and 10 genera are endemic, or nearly so, to sandstone rockhouses in eastern United States; 36% are members of Polypodiophyta and the rest belong to Magnoliophyta. Hymenophyllaceae, Caryophyllaceae, and Asteraceae each contain two taxa; all other families contain one taxon. Trichomanes is the only genus with two species.

Six endemic taxa occur in seven or more states, and five taxa occur only in one or two states. Only one taxon occurs far north of the Wisconsinan Glacial Boundary, and three taxa occur only in single river drainages. The highest concentration of endemic taxa occurs in the Red River and Cumberland River drainages on the Cumberland Plateau.

All 11 endemics are perennials. Of the dicots, the majority are hemicryptophytes without vegetative reproduction. Two ferns exist exclusively as gametophytes reproducing by gemmae. Relatively little is known about the life history ecology of the endemics.

Vicariance and adaptive radiation have played an important role in speciation of the endemic taxa, but polyploidy has not. The cause(s) of endemism is unknown. The endemic ferns are most likely Tertiary relicts derived from tropical taxa. Apparently, they survive in a humid temperate climate because of the tropical-like microhabitat in rockhouses [i.e., moderated temperatures, constant moisture, year-round growing season (Farrar, 1971)]. In contrast, the majority of endemic flowering plants are derived from taxa that grow in forest habitats in the vicinity of rockhouses. The endemics have characteristics more typical of sciophytes [i.e., thin leaves, large leaf area, weak and etiolated stems (Daubenmire, 1974; Larcher, 1995)] than do their parental taxa, The relative age of the endemic flowering plants ranges from the Late Tertiary to the Recent.

Rockhouses provide an opportunity for the study of speciation, cause(s) of endemism/rarity, and microclimates and microhabitats in eastern United States. Because some of the most narrowly endemic taxa in eastern United States occur in rockhouses, and since many rockhouses are disturbed by recreational use, they should receive high priority for conservation and preservation.

Table I

Plant taxa endemic, or nearly so, to sandstone rockhouses in eastern United States

VITTARIACEAE

Vittaria appalachiana Farrar & Mickel

HYMENOPHYLLACEAE

Trichomanes boschianum Sturm T. intricatum Farrar

THELYPTERIDACEAE

Thelypteris pilosa (Mart. & Gal.) Crawford var. alabamensis Crawford

RANUNCULACEAE

Thalictrum mirabile Small

CARYOPHYLLACEAE

Arenaria cumberlandensis Wofford & Kral Silene rotundifolia Nutt.

PRIMULACEAE

Dodecatheon frenchii (Vasey) Rydb.

SAXIFRAGACEAE

Heuchera parviflora Bartling var. parviflora

ASTERACEAE

Ageratina luciae-brauniae (Fern.) King & Robinson Solidago albopilosa Braun

XII. Acknowledgments

We thank W. Anderson, A. Cusick, E. Esselman, D. Farrar, M. Homoya, J. Johansen, L. Klotz, N. Lersten, J. Logan, C. Keener, R. Kral, E. Schilling, J. Schwegman, D. Soltis, T. Sussenbach, E. Wells, and B. Wofford for providing information used in this paper.

XIII. Literature Cited

Adovasio, J. M., J. Donahue, R. C. Carlisle, K. Cushman, R. Stuckenrath & P. Wiegman. 1984. Meadowcroft Rockshelter and the Pleistocene/Holocene transition in southwestern Pennsylvania. Pages 347-369 in H. H. Genoways & M. R. Dawson (eds.), Contributions in Quaternary vertebrate paleontology: A volume in memorial to John E. Guilday. Carnegie Mus. Nat. Hist. Special Publ. No. 8.

-----, J. D. Gunn, J. Donahue & R. Stuckenrath. 1977. Meadowcroft Rockshelter: Retrospect 1976. Pennsylvania Archaeol. 47: 1-93.

Anderson, S. 1994. Area and endemism. Quart. Rev. Biol. 69: 451-471.

Andreasen, M. L. & W. H. Eshbaugh. 1973. Solidago albopilosa Braun, a little known goldenrod from Kentucky. Castanea 38: 117-132.

Antonovics, J., D. Stratton, P. H. Thrall & A.M. Jarosz. 1996. An anther-smut disease (Ustilago violacea) of fire-pink (Silene virginica): Its biology and relationship to the anther-smut disease of white campion (Silene alba). Amer. Midl. Naturalist 135: 130-143.

Atherton, E. & J. E. Palmer. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Illinois. U.S. Geol. Surv. Profess. Pap. 1110-L.

Austin, D. F. 1975. Bird flowers in the eastern United States. Florida Sci. 38: 1-12.

Baker, V. R. 1990. Spring sapping and valley network development, with case studies by R. C. Kochel, V. R. Baker, J. E. Laity & A. D. Howard. Pages 235-265 in C. G. Higgins & D. R. Coates (eds.), Groundwater geomorphology; The role of subsurface water in Earth-surface processes and landforms. Geol. Soc. Amer. Special Pap. 252.

Barbour, M. G., J. H. Burk & W. D. Pitts. 1987. Terrestrial plant ecology. Ed. 2. Benjamin/Cummings, Menlo Park, California.

Barr, T. C., Jr. 1961. Caves of Tennessee. Tennessee Div. Geol. Bull. 64.

Bartlett, C. S., Jr. 1963. The Tom's Brook Site - 3J01, A preliminary report. Pages 15-65 in C. R. McGimsey III (ed.), Arkansas archeology 1962. Arkansas Archeol. Soc., Fayetteville.

Baskin, C. C. & J. M. Baskin. 1988. Germination ecophysiology of herbaceous plant species in a temperate region. Amer. J. Bot. 75: 286-305.

Baskin, J. M. & C. C. Baskin. 1985. The annual dormancy cycle in buried weed seeds: A continuum. BioScience 35: 492-498.

Bates, R. L. & J. A. Jackson (eds.). 1980. Glossary of geology. Ed. 2. American Geological Institute, Falls Church, Virginia.

Beamish, K. I. 1955. Studies in the genus Dodecatheon of northwestern America. Bull. Torrey Bot. Club 82: 357-366.

Beatley, J. C. 1956. The winter-green herbaceous flowering plants of Ohio. Ohio J. Sci. 56: 349-377.

Beaudry, J. R. 1959. Solidago albopilosa Braun and S. flexicaulis L. Castanea 24: 53-54.

-----. 1963. Studies on Solidago L. VI. Additional chromosome numbers of taxa of the genus Solidago. Canad. J. Genet. Cytol. 5: 150-174.

Berg, T. M., M. K. McInerney, J. H. Way & D. B. MacLachlan. 1983. Stratigraphic correlation chart of Pennsylvania. Pennsylvania Bur. Topogr. Geol. Surv. Gen. Geol. Rep. 75.

Beveridge, T. R. 1990. Geologic wonders and curiosities of Missouri. Ed. 2. Revised by J. D. Vineyard. Missouri Div. Geol. Land Surv. Educ. Set. No. 4.

Bicker, A. R., Jr. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Alabama and Mississippi. Carboniferous outcrops of Mississippi. U.S. Geol, Surv. Profess. Pap. 1110-I.

Bjorkman, O. 1981. Responses to different quantum flux densities. Pages 57-107 in O. L. Lange, P.S. Nobel, C. B. Osmond & H. Ziegler (eds.), Physiological plant ecology I: Responses to the physical environment. Encyclopedia of plant physiology. New series, vol. 12A. Springer-Verlag, Berlin.

Blackburn, K. B. 1928. Chromosome number in Silene and the neighbouring genera. Pages 439-446 in Proc. Fifth Int. Congr. Genetics.

Boivin, B. 1944. American Thalictra and their old world allies. Rhodora 46: 337-377, 391-445, 453-487.

Bramwell, D. 1972. Endemism in the flora of the Canary Islands. Pages 141-159 in D. H. Valentine (ed.), Taxonomy phytogeography and evolution. Academic Press, London.

Braun, E. L. 1935. The vegetation of Pine Mountain, Kentucky. Amer. Midl. Naturalist 16:517-565.

-----. 1940. New plants from Kentucky. Rhodora 42: 47-51.

-----. 1942. A new species and a new variety of Solidago from Kentucky. Rhodora 44: 1-4.

-----. 1950. Deciduous forests of eastern North America. Blakiston, Philadelphia.

Bryant, W. S., W. C. McComb & J. S. Fralish. 1993. Oak-hickory forests (western mesophytic/oak-hickory forests). Pages 143-201 in W. H. Martin, S. G. Boyce & A. C. Echternacht (eds.), Biodiversity of the southeastern United States: Upland terrestrial communities. John Wiley, New York.

Bush, W. V., B. R. Haley, C. G. Stone, D. F. Holbrook & J. D. McFarland III. 1977. A guidebook to the geology of the Arkansas Paleozoic area. Arkansas Geological Commission, Little Rock.

Butler, B. M. 1971. Hoover-Beeson Rockshelter, 40Cn4, Cannon County, Tennessee. Tennessee Archaeol. Soc. Misc. Pap, No, 9.

Cain, S. A. & G. M. de Oliveira Castro. 1959. Manual of vegetation analysis. Harper and Brothers, New York.

Camburn, K. E. 1982. Subaerial diatom communities in eastern Kentucky. Trans. Amer. Microscop. Soc. 101: 375-387.

Campbell, J. J. N. & W. Meijer. 1989. The flora and vegetation of Jessamine Gorge, Jessamine County, Kentucky: A remarkable concentration of rare species in the Bluegrass Region. Trans. Kentucky Acad. Sci. 50: 27-45.

-----, J. R. Abbott, R. R. Cicerello, J. D. Kiser, J. R. MacGregor & J. G. Palis. 1994. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, London Ranger District. Kentucky State Nature Preserves Commission, Frankfort.

-----, S. A. Bonney, J. D. Kiser, L. E. Kornman, J. R. MacGregor, L. E. Meade & A. C. Risk. 1992. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, Morehead Ranger District. Kentucky State Nature Preserves Commission, Frankfort.

-----, J. E. Flotemersch, J. R. MacGregor, D. Noe, A. C. Risk, M. D. Studer & D. T. Towles. 1991. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, Berea Ranger District. Kentucky State Nature Preserves Commission, Frankfort.

-----, A. C. Risk, V. L. Andrews, B. Palmer-Ball, Jr. & J. R. MacGregor. 1990. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, Stearns Ranger District. Kentucky State Nature Preserves Commission, Frankfort.

-----, D. T. Towles, J. R. MacGregor, R. R. Cicerello, B. Palmer-Ball, Jr., M. E. Medley & S. Olson. 1989. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, Stanton Ranger District. Kentucky State Nature Preserves Commission, Frankfort.

Caplenor, D. 1955. An annotated list of the vascular plants of the gorges of the Fall Creek Falls State Park. J. Tennessee Acad. Sci. 30: 93-108.

Carman, J. E. 1946. The geologic interpretation of scenic features in Ohio. Ohio J. Sci. 46: 241-283.

Carter, J. P. & C. E. Martin. 1994. The occurrence of crassulacean acid metabolism among epiphytes in a high-rainfall region of Costa Rica. Selbyana 15: 104-106.

Chesnut, D. R., Jr. 1992. Stratigraphic and structural framework of Carboniferous rocks in the central Appalachian Basin in Kentucky. Kentucky Geol. Surv. Ser. 11, Bull. 3.

Childress, J. D. 1992. Soil survey of Bell and Harlan Counties, Kentucky. U.S.D.A. Soil Conservation Service & U.S.D.A. Forest Service, in cooperation with the Kentucky Agric. Exp. Sta. & Kentucky Nat. Resources Environ. Protect. Cab.

Clark, M. B. 1962. Trichomanes boschianum in Madison County, Arkansas. Amer. Fern J. 52: 85-86.

Clewell, A. F. & J. W. Wooten. 1971. A revision of Ageratina (Compositae: Eupatorieae) from eastern North America. Brittonia 23: 123-143.

Coates, F. & J. B. Kirkpatrick. 1992. Environmental relations and ecological responses of some higher plant species on rock cliffs in northern Tasmania. Austral. J. Ecol. 17: 441-449.

Coker, W. C. 1938. A filmy fern from North Carolina. J. Elisha Mitchell Sci. Soc. 54:349-350 + plates 40-41.

Collins, H. R. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Ohio. U.S. Geol. Surv. Profess. Pap. 1110-E.

Copeland, E. B. 1938. Genera Hymenophyllacearum. Philipp. J. Sci. 67:1-110.

Corgan, J. X. & J. T. Parks. 1979. Natural bridges of Tennessee. Tennessee Div. Geol. Bull. 80.

Cowan, C. W. 1979. Prehistoric plant utilization at the Rogers Rockshelter, Powell County, Kentucky. M.A. thesis, University of Kentucky, Lexington.

-----, H. E. Jackson, K. Moore, A. Nickelhoff & T. L. Smart. 1981. The Cloudsplitter Rockshelter, Menifee County, Kentucky: A preliminary report. Southeastern Archaeol. Conf. Bull. 24: 60-76.

Coy, F. E., Jr. & T. C. Fuller. 1971. Petroglyphs of Powell County, Kentucky. Central States Archaeol. J. 18: 112-122.

Cranfill, R. 1980. Ferns and fern allies of Kentucky. Kentucky Nature Preserves Commission Sci. Techn. Set. No. 1.

-----. 1991. Flora of Hardin County, Kentucky. Castanea 56: 228-267.

----- & M. E. Medley. 1981. Notes on the flora of Kentucky: New and interesting plants in Kentucky. Rhodora 83:125-131.

Crawford, L. C. 1951. A new fern for the United States. Amer. Fern J. 41: 15-20.

Cumings, E. R. 1922. Nomenclature and description of the geological formations of Indiana. Pages 403-570 in W. N. Logan, E. R. Cumings, C. A. Malott, S. S. Visher, W. M. Tucker & J. R. Reeves (eds.), Handbook of Indiana geology. Indiana Department of Conservation Publ. No. 21.

Curtis, J. D. & N. R. Lersten. 1986. Development of bicellular foliar secretory cavities in white snakeroot, Eupatorium rugosum (Asteraceae). Amer. J. Bot. 73: 79-86.

Cusick, A. W. 1983. Vittaria gametophytes discovered in a new physiographic province. Amer. Fern J. 73: 33-38.

-----. 1994. Noteworthy collections - West Virginia. Castanea 59: 79-80.

Darling, T., Jr. 1955. A red-letter day in South Carolina. Amer. Fern J. 45: 138-142.

Daubenmire, R. F. 1974. Plants and environment: A textbook of plant autecology. Ed. 3. John Wiley, New York.

Dean, B. E. 1964. Ferns of Alabama and fern allies. American Southern Publications, Northport, Alabama.

DeLong, R. M. 1967. Bedrock geology of the south Bloomingville quadrangle, Hocking and Vinton Counties, Ohio. Ohio Div. Geol. Surv. Rep. Invest. No. 63.

Donahue, J. & J. M. Adovasio. 1990, Evolution of sandstone rockshelters in eastern North America; a geoarchaeological perspective. Pages 231-251 in N. P. Lasca & J. Donahue (eds.), Archaeological geology of North America. Geol. Soc. Amer. Centennial Special Vol. 4.

Duncan, W. H. & J. T. Kartesz. 1981. Vascular flora of Georgia, an annotated checklist. University of Georgia Press, Athens.

Eaton, S. W. & E. F. Schrot. 1987, A flora of the vascular plants of Cattaraugus County, New York. Buffalo Soc. Nat. Sci. Bull. 31.

Englund, K. J., H. L. Smith, L. D. Harris & J. G. Stephens. 1961. Geology of the Ewing quadrangle, Kentucky and Virginia. U.S. Geol. Surv. Map GQ-172.

Ennis, B. 1928. The life forms of Connecticut plants and their significance in relation to climate. Connecticut Geol. Nat. Hist. Surv. Div. Biol. Bull. No. 43.

Evers, R. A. 1961. The filmy fern in Illinois. Illinois Nat. Hist. Surv. Notes No. 44.

Farrar, D. R. 1971. The biology of ferns with asexually reproducing gametophytes in the eastern United States. Ph.D. diss., University of Michigan, Ann Arbor.

-----. 1978. Problems in the identity and origin of the Appalachian Vittaria gametophyte, a sporophyteless fern of the eastern United States. Amer. J. Bot. 65: 1-12.

-----. 1985. Independent fern gametophytes in the wild. Proc. Roy. Soc. Edinburgh 86B: 361-369.

-----. 1990. Species and evolution in asexually reproducing independent fern gametophytes. Syst. Bot. 15: 98-111.

-----. 1992. Trichomanes intricatum: The independent Trichomanes gametophyte in the eastern United States. Amer. Fern J. 82: 68-74.

-----. 1993a. Hymenophyllaceae Link - filmy fern family. Pages 190-197 in N. R. Morin et al. (eds.), Flora of North America north of Mexico. Vol. 2. Oxford University Press, New York.

-----. 1993b. Vittariaceae Ching - shoestring fern family. Pages 187-189 in N. R. Morin et al. (eds.), Flora of North America north of Mexico. Vol. 2. Oxford University Press, New York.

----- & P. G. Davison. 1994. Hymenophyllum tayloriae: Sporophyte and gametophyte together again after 58 years. Amer. J. Bot. 81, Suppl. to No. 6:128 (abstract).

----- & J. T. Mickel. 1991. Vittaria appalachiana: A name for the "Appalachian gametophyte." Amer. Fern J. 81: 69-75.

-----, J. C. Parks & B. W. McAlpin. 1983. The fern genera Vittaria and Trichomanes in the northeastern United States. Rhodora 85: 83-92.

Fassett, N. C. 1944. Dodecatheon in eastern North America. Amer. Midl. Naturalist 31: 455-486.

Favarger, C. & J. Contandriopoulos. 1961. Essai sur l'endemisme. Ber. Schweiz. Bot. Ges. 71: 384-408.

Fenneman, N. M. 1938. Physiography of eastern United States. McGraw-Hill, New York.

Fenster, C. B., C. L. Hassler & M. R. Dudash. 1996. Fluorescent dye particles are good pollen analogs for hummingbird-pollinated Silene virginica (Caryophyllaceae). Canad. J. Bot. 74: 189-193.

Ferguson, T. A. & J. W. Gardner. 1986. Results of field investigations. Pages 74-268 & G1-G42 in T. A. Ferguson, R. A. Pace, J. W. Gardner & R. W. Hoffman (eds.), An archaeological reconnaissance and testing of indirect impact areas within selected development sites of the Big South Fork National River and Recreation Area. Final report submitted to U.S. Army Corps of Engineers, Nashville District, Nashville, Tennessee.

-----, R. A. Pace, J. W. Gardner & R. W. Hoffman (eds.). 1986. An archaeological reconnaissance and testing of indirect impact areas within selected development sites of the Big South Fork National River and Recreation Area. Final report submitted to U.S. Army Corps of Engineers, Nashville District, Nashville, Tennessee.

Fernald, M. L. 1942. The seventh century of additions to the flora of Virginia. Rhodora 44: 341-405, 416-452, 457-479 + plates 717-744.

-----. 1950. Gray's manual of botany. Ed. 8. American Book Co., New York.

Froelich, A. J. & J. F. Tazelaar. 1974. Geologic map of the Pineville quadrangle, Bell and Knox Counties, Kentucky. U.S. Geol. Surv. Map GQ-1129.

Funkhouser, W. D. 1925. Wild life in Kentucky. Kentucky Geol. Surv. Ser. 6, Vol. 16.

----- & W. S. Webb. 1928. Ancient life in Kentucky. Kentucky Geol. Surv. Ser. 6, Vol. 34.

----- & -----. 1929. The so-called "Ash Caves" in Lee County, Kentucky. Univ. Kentucky Rep. Archaeol. Anthrop. 1: 37-112.

----- & -----. 1930. Rock shelters of Wolfe and Powell Counties, Kentucky. Univ. Kentucky Rep. Archaeol. Anthrop. 1: 239-306.

Furr, R. S. 1971. A biosystematic study of the subsections Villosae and Micranthae of the genus Heuchera from the southeastern United States. Ph.D. diss., University of Tennessee, Knoxville.

Gardner, J. W. 1986. Historic site and feature descriptions. Pages C1-C40 in T. A. Ferguson, R. A. Pace, J. W. Gardner & R. W. Hoffman (eds,), An archaeological reconnaissance and testing of indirect impact areas within selected development sites of the Big South Fork National River and Recreation Area. Final report submitted to U.S. Army Corps of Engineers, Nashville District, Nashville, Tennessee.

Gastony, G. J. 1977. Chromosomes of the independently reproducing Appalachian gametophyte: A new source of taxonomic evidence. Syst. Bot. 2: 43-48.

Gibson, D. 1961. Life-forms of Kentucky flowering plants. Amer. Midl. Naturalist 66: 1-60.

Gilbert, F. A. 1938. Notes on plants from southern West Virginia III. Castanea 3: 81-84.

Gillespie, J. P. 1956. The genus Heuchera in Tennessee. J. Tennessee Acad. Sci. 31: 232-241.

Gleason, H. A. & A. Cronquist. 1991. Manual of vascular plants of northeastern United States and adjacent Canada. Ed. 2. The New York Botanical Garden, Bronx.

Gould, K. S., D. N. Kuhn, D. W. Lee & S. F. Oberbauer. 1995. Why leaves are sometimes red. Nature 378: 241-242.

Graham, A. 1993. History of the vegetation: Cretaceous (Maastrichtian)-Tertiary. Pages 57-70 in N. R. Morin et al. (eds.), Flora of North America north of Mexico. Vol. 1. Oxford University Press, New York.

Grant, W. F. 1953. A cytotaxonomic study in the genus Eupatorium. Amer. J. Bot. 40: 729-742.

Griggs, R. F. 1914. A botanical survey of the Sugar Grove region. Ohio Biol. Surv. Bull. 3.

Gunn, S. C. 1991. An update on the status of Thelypteris pilosa var. alabamensis. Report submitted to U.S. Fish and Wildlife Service, Jackson, Mississippi.

Haley, B. R., E. E. Glick, W. M. Caplan, D. F. Holbrook & C. G. Stone. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Arkansas. U.S. Geol. Surv. Profess. Pap. 1110-O.

Hall, J. F. 1953. The geology of Hocking State Park. Ohio Div. Geol. Surv. Inform. Circ. No. 8.

Hammond, H. D. 1992. Corrections and additions for and errors found in the Manual of vascular plants of northeastern United States and adjacent Canada, second edition, by Henry A. Gleason (late) and Arthur Cronquist. Bull. Torrey Bot. Club 119: 88-89.

Hansen, C. E. 1952. The life-forms of the flowering plants of Illinois. M.S. thesis, Northwestern University, Evanston, Illinois.

Harker, D. F., Jr., M. E, Medley, L. R. Phillippe, R. R. Hannah & A. Phillippi. 1981. Rare plants of eastern Kentucky and the Daniel Boone National Forest. Kentucky Nature Preserves Commission, Frankfort.

Hayes, W. C. & R. D. Knight. 1961. Cambrian System. Pages 14-20 in J. W. Koenig (ed.), The stratigraphic succession in Missouri. Missouri Div. Geol. Surv. Water Resources Ser. 2, Vol. 40.

Heaslip, M. B. 1948. Autecological studies of Silene rotundifolia Nutt. M.S. thesis, Ohio State University, Columbus.

-----. 1950. Cytoecological studies of Silene rotundifolia Nutt., S. virginica L., and hybrid. Ohio J. Sci. 50: 97-101.

-----. 1951. Some cytoecological aspects in the evolution of certain species of the plant genus Silene. Ohio J. Sci. 51: 62-70.

Hill, C. A. 1981. Origin of cave saltpeter. Natl. Speleol. Soc. Bull. 43: 110-126.

Hinkle, C. R. 1989. Forest communities of the Cumberland Plateau of Tennessee. J. Tennessee Acad. Sci. 64: 123-129.

Hitchcock, C. L. & B. Maguire. 1947. A revision of the North American species of Silene. Univ. Washington Publ. Biol. 13: 1-73.

Hoffman, R. W. 1987. Taphonomy and zooarcheology of rockshelters of the Big South Fork Area of the Cumberland Plateau, Tennessee and Kentucky. Ph.D. diss., University of Tennessee, Knoxville.

Homoya, M. A., D. B. Abrell, J. R. Aldrich & T. W. Post. 1984. The natural regions of Indiana. Proc. Indiana Acad. Sci. 94: 245-268 + colored map of the natural regions of Indiana by M. A. Homoya.

Howell, B. J. 1981. A survey of folklife along the Big South Fork of the Cumberland River. Univ. Tennessee Dept. Anthrop. Rep. Invest. No. 30.

Hulten, E. 1964. Remarkable range extension for Minuartia groenlandica (Retz.) Ostenf. Sv. Bot. Tidskr. 58: 432-435.

Iltis, H. H. & W. M. Shaughnessy. 1960. Preliminary reports on the flora of Wisconsin No. 43. Primulaceae - Primrose family. Trans. Wisconsin Acad. Sci., Arts & Letts. 49: 113-135.

Iwatsuki, K. 1964. An American species of Stegnogramma. Amer. Fern J. 54: 141-143.

Jacobson, R. J. 1992. Geology of the Goreville quadrangle, Johnson and Williamson counties, Illinois. Illinois Geol. Surv. Bull. 97.

Jennings, J. N. 1968. Tafoni. Pages 1103-1104 in R. W. Fairbridge (ed.), The encyclopedia of geomorphology. Reinhold Book, New York.

Jensen, H. W. 1944. Heterochromosome formation in the genus Ilex. Amer. Naturalist 78: 375-379.

Jillson, W. R. 1969. The geology of Powell County, Kentucky. Roberts Printing, Frankfort, Kentucky.

Kawano, S. 1965. Application of pectinase and cellulase in an orcein squash method. Bot. Mag. (Tokyo) 78: 36-42.

Keener, C. S. 1976. Studies in the Ranunculaceae of the southeastern United States. II. Thalictrum L. Rhodora 78: 457-472.

-----. 1983. Distribution and biohistory of the endemic flora of the mid-Appalachian shale barrens. Bot. Rev. (Lancaster) 49: 65-115.

King, R. M. & H. Robinson. 1970. Studies in the Eupatorieae (Compositae). XIX. New combinations in Ageratina. Phytologia 19: 208-229.

Klotz, L. H. & J. L. Walck. 1993. Rare vascular plants associated with limestone in southwestern Franklin County, Pennsylvania. Bartonia, Suppl. to No. 57: 16-41.

Koch, A. R. 1976. Floristics and ecology of algae on sandstone cliffs in east-central and southeastern Ohio. Ph.D. diss., Ohio State University, Columbus.

Kral, R. 1966. Observations on the flora of the southeastern United States with special reference to northern Louisiana. Sida 2: 395-408.

-----. 1973. Some notes on the flora of the southern states, particularly Alabama and middle Tennessee. Rhodora 75: 366-410.

-----. 1983. A report on some rare, threatened, or endangered forest-related vascular plants of the south. 2 vols. U.S.D.A. Forest Service Techn. Publ. R8-TP 2.

Kruckeberg, A. R. 1954. Chromosome numbers in Silene (Caryophyllaceae): I. Madrono 12: 238-246.

-----. 1960. Chromosome numbers in Silene (Caryophyllaceae). II. Madrono 15: 205-215.

-----. 1963. Artificial crosses involving eastern North American Silenes. Brittonia 16: 95-105.

Larcher, W. 1995. Physiological plant ecology. Ed. 3. Springer-Verlag, Berlin.

Lersten, N. R. & J. D. Curtis. 1986. Tubular cavities in white snakeroot, Eupatorium rugosum (Asteraceae). Amer. J. Bot. 73: 1016-1021.

Lowe, E. N. 1921. Plants of Mississippi: A list of flowering plants and ferns. Mississippi Geol. Surv. Bull. No. 17.

Ludlow, C. J. & F. T. Wolf. 1975. Photosynthesis and respiration rates of ferns. Amer. Fern J. 65: 43-48.

Macior, L. W. 1970. Pollination ecology of Dodecatheon arnethystinum (Primulaceae). Bull. Torrey Bot. Club 97: 150-153.

Maguire, B. 1951. Studies in the Caryophyllaceae - V. Arenaria in America north of Mexico. A conspectus. Amer. Midl. Naturalist 46: 493-511.

Malott, C. A. 1924. The Upper Chester of Indiana. Proc. Indiana Acad. Sci. 34:103-132.

Manning, R. 1993. The historic Cumberland Plateau: An explorer's guide. University of Tennessee Press, Knoxville.

Martin, C. E., M. T. Allen & C. H. Haufler. 1995. [C.sub.3] photosynthesis in the gametophyte of the epiphytic CAM fern Pyrrosia longifolia (Polypodiaceae). Amer. J. Bot. 82: 441-444.

Martin, J. A., R. D. Knight & W. C. Hayes. 1961. Ordovician System. Pages 20-32 in J. W. Koenig (ed.), The stratigraphic succession in Missouri. Missouri Div. Geol. Surv. Water Resources Ser. 2, Vol. 40.

Martin, W. H., W. S. Bryant, J. S. Lassetter & J. B. Varner. 1979. The Kentucky River Palisades flora and vegetation. Report submitted to Kentucky Chapter of The Nature Conservancy, Frankfort.

McDowell, R. C. 1981. Correlation chart for units on the geologic map of Kentucky. U.S. Geol. Surv. Misc. Field Studies Map MF-1291.

McFarlan, A. C. 1954. Geology of the Natural Bridge State Park area. Kentucky Geol. Surv. Ser. 4, Special Publ. No. 4.

McNeill, J. 1980. The delimitation of Arenaria (Caryophyllaceae) and related genera in North America, with 11 new combinations in Minuartia. Rhodora 82: 495-502.

Medley, M. E. 1980. Status report on Solidago albopilosa E. L. Braun. Report submitted to U.S. Fish and Wildlife Service, Atlanta.

Melampy, M. N. & A. M. Hayworth. 1980. Seed production and pollen vectors in several nectarless plants, Evolution 34: 1144-1154.

Milici, R. C., G. Briggs, L. M. Knox, P. D. Sitterly & A. T. Statler. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Tennessee. U.S. Geol. Surv. Profess. Pap. 1110-G.

Miller, J. R., D, F. Ritter & R. C. Kochel. 1990. Morphometric assessment of lithologic controls on drainage basin evolution in the Crawford Upland, south-central Indiana. Amer. J. Sci. 290: 569-599.

Mohlenbrock, R. H. 1978. The illustrated flora of Illinois. Flowering plants: Hollies to loasas. Southern Illinois University Press, Carbondale.

-----. 1987. Alum Cove, Arkansas. Nat. Hist. 96(4): 60-62.

----- & J. W. Voigt. 1959. A study of the filmy fern Trichomanes boschianum. Amer. Fern J. 49: 76-85.

Mohr, C. 1901. Plant life of Alabama. Contrib. U.S. Natl. Herb. 6: 1-921.

Moore, J. E. 1965. A study of the vegetation of Petit Jean Mountain in central Arkansas. Castanea 30: 1-37.

Morris, M. W. 1986. A sediment analysis of three rockshelter sites within the drainage of the Big South Fork. Pages F1-F32 in T. A. Ferguson, R. A. Pace, J. W. Gardner & R. W. Hoffman (eds.), An archaeological reconnaissance and testing of indirect impact areas within selected development sites of the Big South Fork National River and Recreation Area. Final report submitted to U.S. Army Corps of Engineers, Nashville District, Nashville, Tennessee.

Morton, J. K. 1981. Chromosome numbers in Compositae from Canada and the U.S.A. J. Linn. Soc., Bot. 82: 357-368.

Nasrulhaq-Boyce, A. & M. A. H. Mohamed. 1987. Photosynthetic and respiratory characteristics of Malayan sun and shade ferns. New Phytol. 105: 81-88.

Olah, L. V. & R. A. DeFilipps. 1968. A cytotaxonomic study of French's shooting star. Bull. Torrey Bot. Club 95: 186-198.

----- & -----. 1969. Oversize Dodecatheon frenchii in Illinois. Bull. Torrey Bot. Club 96: 378-379.

Ormerod, D. E. 1983. White Rocks: A Woodland rockshelter in Monroe County, Ohio. Kent State Res. Pap. Archaeol. No. 4.

Outerbridge, W. F. 1987. The Logan Plateau, a young physiographic region in West Virginia, Kentucky, Virginia, and Tennessee. U.S. Geol. Surv. Bull. 1620.

Pace, R. A., R. W. Hoffman & J. W. Gardner. 1986. Regional overview. Pages 5-54 in T. A. Ferguson, R. A. Pace, J. W. Gardner & R. W. Hoffman (eds.), An archaeological reconnaissance and testing of indirect impact areas within selected development sites of the Big South Fork National River and Recreation Area. Final report submitted to U.S. Army Corps of Engineers, Nashville District, Nashville, Tennessee.

Palmer-Ball, B., Jr., J. J. N. Campbell, M. E. Medley, D. T. Towles, J. R. MacGregor & R. R. Cicerello. 1988. Cooperative inventory of endangered, threatened, sensitive and rare species, Daniel Boone National Forest, Somerset Ranger District. Kentucky Nature Preserves Commission, Frankfort.

Parks, J. C. 1987. First record of Vittaria gametophytes in eastern Maryland. Bartonia No. 53: 61.

-----. 1989. Distribution of gametophytic populations of Vittaria and Trichomanes in Pennsylvania. Rhodora 91: 201-206.

----- & D. R. Farrar. 1984. A first report of the fern genus Vittaria in New York. Rhodora 86: 421-423.

Paul, C. & W. Moffitt. 1959. The Quaker State Site - An Indian rock shelter. Pennsylvania Archaeol. 29: 51-65.

Pence, V. C. 1991. Cryopreservation of seeds of Ohio native plants and related species, Seed Sci. Techn. 19: 235-251.

Perkins, B. E. 1981. Vegetation of sandstone outcrops of the Cumberland Plateau. M.S. thesis, University of Tennessee, Knoxville.

Prufer, O. H. 1981. Raven Rocks: A specialized Late Woodland rockshelter occupation in Belmont County, Ohio. Kent State Res. Pap. Archaeol. No. 1.

Quarterman, E. & R. L. Powell. 1978. Potential ecological/geological natural landmarks on the Interior Low Plateaus. U.S. Department of the Interior, Washington, DC.

Quinn, J. H. 1958. Plateau surfaces of the Ozarks. Proc. Arkansas Acad. Sci. 11: 36-43.

Raine, C. A., D. R. Farrar & E. Sheffield. 1991. A new Hymenophyllum species in the Appalachians represented by independent gametophyte colonies. Amer. Fern J. 81: 109-118.

Rau, J. L. 1970. Pennsylvanian System of northeast Ohio. Pages 69-89 in P. O. Banks & R. M. Feldman (eds.), Guide to the geology of northeastern Ohio. Northern Ohio Geological Society, Cleveland.

Raunkiaer, C. 1934. The life forms of plants and statistical plant geography. Clarendon Press, Oxford.

Ray, L. L. 1974. Geomorphology and Quaternary geology of the glaciated Ohio River Valley - a reconnaissance study. U.S. Geol. Surv. Profess. Pap. 826.

Redfearn, P. L., Jr. 1987. Habitat and distributional notes on selected mosses of the Interior Highlands of North America. Mem. New York Bot. Gard. 45: 232-236.

Reed, C. F. 1951. Notes on the ferns of Kentucky, II. Trichomanes boschianum. Amer. Fern J. 41: 120-123 + plate 8.

Reilly, M. J. 1992. Photosynthetic pigment analysis and determination of the pigment protein complexes in gametophytes of the low light intensity - adapted fern, Vittaria appalachiana. M.S. thesis, Indiana University of Pennsylvania, Indiana.

Rice, C. L. 1984. Sandstone units of the Lee Formation and related strata in eastern Kentucky. U.S. Geol. Surv. Profess. Pap. 1151-G.

----- & R. G. Ping. 1989. Geology of the Middlesboro north quadrangle, Kentucky. U.S. Geol. Surv. Map GQ-1663.

-----, E, G. Sable, G, R, Dever Jr. & T. M. Kehn. 1979, The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States - Kentucky. U.S. Geol. Surv. Profess. Pap. 1110-F.

Rohrbach, P. 1868. Monographie der Gattung Silene. W. Engelmann, Leipzig.

Rosendahl, C. O. 1951. A new Heuchera from Missouri together with some notes on the Heuchera parviflora group. Rhodora 53: 105-109.

-----, F. K. Butters & O. Lakela. 1936. A monograph on the genus Heuchera. University of Minnesota Press, Minneapolis.

Ruchhoft, R. H. 1986. Kentucky's land of the arches. Pucelle Press, Cincinnati, Ohio.

Rydberg, P. A. 1932. Flora of the prairies and plains of central North America. The New York Botanical Garden, Bronx.

Sauer, C. O. 1918. Geography. Pages 1-83 in C. O. Sauer, G. H. Cady & H. C. Cowles (eds.), Starved Rock State Park and its environs. Geogr. Soc. Chicago Bull. No. 6.

-----. 1927. Geography of the Pennyroyal. Kentucky Geol. Surv. Ser. 6, Vol. 25.

Schilling, E. E. 1996. Chloroplast DNA restriction site analysis of Eupatorieae (Asteraceae) from eastern North America. Amer. J. Bot. 83, Suppl. to No. 6: 190 (abstract).

Schmalzer, P. A., T. S. Patrick & H. R. DeSelm. 1985. Vascular flora of the Obed Wild and Scenic River, Tennessee. Castanea 50: 71-88.

Schwegman, J. E. 1973. Comprehensive plan for the Illinois nature preserves system, part 2: The natural divisions of Illinois. Illinois Nature Preserves Commission, Springfield.

-----. 1982. Additions to the vascular flora of Illinois. Castanea 47: 243-247.

-----. 1984. The jeweled shooting star (Dodecatheon amethystinum) in Illinois. Castanea 49: 74-82.

Segars, C. B., L. C. Crawford & A. M. Harvill. 1951. The occurrence and distribution of hemlock in Alabama. Ecology 32: 149-151.

Semple, J. C., G. S. Ringius, C. Leeder & G. Morton. 1984. Chromosome numbers of goldenrods, Euthamia and Solidago (Compositae: Astereae). II. Additional counts with comments on cytogeography. Brittonia 36: 280-292.

-----, J. Zhang & C. Xiang. 1993. Chromosome number determinations in fam. Compositae, tribe Astereae. V. Eastern North American taxa. Rhodora 95: 234-253.

Shaver, J. M. 1954. Ferns of Tennessee with the fern allies excluded. Bureau of Publications, George Peabody College for Teachers, Nashville, Tennessee.

Shaver, R. H., C. H. Ault, A. M. Burger, D. D. Carr, J. B. Droste, D. L. Eggert, H. H. Gray, D. Harper, N. R. Hasenmueller, W. A. Hasenmueller, A. S. Horowitz, H. C. Hutchison, B. D. Keith, S. J. Keller, J. B. Patton, C. B. Rexroad & C. E. Wier. 1986. Compendium of Paleozoic rock-unit stratigraphy in Indiana - A revision. Indiana Geol. Surv. Bull. 59.

Shaw, J. 1987. Evolution of heavy metal tolerance in bryophytes II. An ecological and experimental investigation of the "copper moss," Scopelophila cataractae (Pottiaceae). Amer. J. Bot. 74: 813-821.

Short, J. W. & J. D. Freeman. 1978. Rediscovery, distribution and phytogeographic affinities of Leptogramma pilosa in Alabama. Amer. Fern J. 68: 1-2.

Small, J. K. 1900. Notes and descriptions of North American plants. - II. Bull. Torrey Bot. Club 27: 275-281.

Smalley, G. W. 1986. Classification and evaluation of forest sites on the northern Cumberland Plateau. U.S.D.A. Forest Service, Southern Forest Exp. Sta. Gen. Techn. Rep. SO-60.

Smith, A. R. 1981. Pteridophytes. Pages 1-370 in D. E. Breedlove (ed.), Flora of Chiapas. California Academy of Sciences, San Francisco.

-----. 1993. Thelypteridaceae Ching ex Pichi-Sermolli - marsh fern family. Pages 206-222 in N. R. Morin et al. (eds.), Flora of North America north of Mexico. Vol. 2. Oxford University Press, New York.

Smith, B. D. 1992. Rivers of change: Essays on early agriculture in eastern North America. Smithsonian Institution Press, Washington, DC.

Smith, E. B. 1988. An atlas and annotated list of the vascular plants of Arkansas. Ed. 2. Published by the author, Fayetteville, Arkansas.

Solecki, M. K. 1983. Vascular plant communities and noteworthy taxa of Hawn State Park, Ste. Genevieve County, Missouri. Castanea 48: 50-55.

Soltis, D. E. 1980. Karyotypic relationships among species of Boykinia, Heuchera, Mitella, Sullivantia, Tiarella, and Tolmiea (Saxifragaceae). Syst. Bot. 5: 17-29.

-----. 1985. Allozymic differentiation among Heuchera americana, H. parviflora, H. pubescens, and H. villosa (Saxifragaceae). Syst. Bot. 10: 193-198.

----- & R. K. Kuzoff. 1995. Discordance between nuclear and chloroplast phylogenies in the Heuchera group (Saxifragaceae). Evolution 49: 727-742.

Stebbins, G. L. 1980. Polyploidy in plants: Unsolved problems and prospects. Pages 495-520 in W. H. Lewis (ed.), Polyploidy: Biological relevance. Plenum Press, New York.

----- & J. Major. 1965. Endemism and speciation in the California flora. Ecol. Monogr. 35: 1-35.

Steyermark, J. A. 1934. Some features of the flora of the Ozark Region in Missouri. Rhodora 36: 214-233.

-----. 1963, Flora of Missouri. Iowa State University Press, Ames.

Strausbaugh, P. D. & E. L. Core. 1978. Flora of West Virginia. Ed. 2. Seneca Books, Grantsville, West Virginia.

Studlar, S. M. & J. A. Snider. 1989. Bryophytes of the Red River Gorge of Kentucky: Floristics and phytogeography. Castanea 54: 133-152.

Sussenbach, T. 1990. Archaeological site distributions on the Cumberland Plateau of eastern Kentucky. Univ. Kentucky Program Cult. Resource Assessment Archaeol. Rep. 218.

Swayne, J. R. 1973. Paleoecology of southern Illinois and Pleistocene glaciation effect as indicated by modern distribution of disjunct species, French's shooting star, and cane. Ph.D. diss., Southern Illinois University, Carbondale.

-----. 1985. Dodecatheon frenchii. Endangered species survey. Report submitted to Illinois Department of Conservation, Natural Heritage Division, Springfield.

Szmuc, E. J. 1970. The Mississippian System. Pages 23-67 in P.O. Banks & R. M. Feldman (eds.), Guide to the geology of northeastern Ohio. Northern Ohio Geological Society, Cleveland.

Tesmer, I. H. 1954, Geology of Chautauqua County, New York. Hobbies, Buffalo Mus. Sci. 35: 28-35.

-----. 1975. Geology of Cattaraugus County, New York. Buffalo Soc. Nat. Sci. Bull. 27.

Thom, R. H. & J. H. Wilson. 1980. The natural divisions of Missouri. Trans. Missouri Acad. Sci. 14: 9-23.

Thomas, W. A. 1972. Mississippian stratigraphy of Alabama. Alabama Geol. Surv. Monogr. 12.

Thompson, H. J. 1953. The biosystematics of Dodecatheon. Contrib. Dudley Herb. 4: 73-154.

Timme, S. L. & J. Lacefield. 1991. Dodecatheonfrenchii (Vasey) Rydb. (Primulaceae) in Alabama. Rhodora 93: 189-191.

Tobe, J. D., J. E. Fairey III & L. L. Gaddy. 1992. Vascular flora of the Chauga River Gorge, Oconee County, South Carolina. Castanea 57: 77-109.

Tucker, G. E. 1984. Status report on Dodecatheon frenchii (Vasey) Rydb. Report submitted to U.S. Fish and Wildlife Service, Atlanta.

Ugent, D., M. Verdun & M. Mibb. 1982. The jeweled shooting star (Dodecatheon amethystinum): A postglacial migrant in the Mississippi Valley. Phytologia 51: 323-329.

U.S. Fish and Wildlife Service. 1992. Endangered and threatened wildlife and plants; threatened status for the plant Thelypteris pilosa var. alabamensis (Alabama streak-sorus fern). Fed. Regist. 57: 30164-30168.

-----. 1993a. White-haired goldenrod recovery plan. U.S. Fish and Wildlife Service, Atlanta.

-----. 1993b. Endangered and threatened wildlife and plants; review of plant taxa for listing as endangered or threatened species. Fed. Regist. 58: 51144-51190.

-----. 1996a. Cumberland sandwort recovery plan. U.S. Fish and Wildlife Service, Atlanta.

-----. 1996b. Endangered and threatened wildlife and plants; review of plant and animal taxa that are candidates for listing as endangered or threatened species. Fed. Regist. 61: 7596-7613.

Van Stockum, R. R., Jr. 1979. Hemlock - mixed mesophytic forest communities in southern Indiana, western Kentucky and Highlands, North Carolina. Ph.D. diss, University of Louisville, Louisville, Kentucky.

Vaughn, D. M. 1989. Valley modification by sapping processes in southern Indiana. Phys. Geogr. 10: 72-84.

Vento, F. J. 1985. The geology and geoarchaeology of the Bay Springs Rockshelters, Tishomingo County, Mississippi. Ph.D. diss., University of Pittsburgh, Pittsburgh.

Voigt, J. W. & R. H. Mohlenbrock. 1964. Plant communities of southern Illinois. Southern Illinois University Press, Carbondale.

----- & J. R. Swayne. 1955. French's shooting star in southern Illinois. Rhodora 57: 325-332.

Wagner, W. H., Jr., D. R. Farrar & B. W. McAlpin. 1970. Pteridology of the Highlands Biological Station area, southern Appalachians. J. Elisha Mitchell Sci. Soc. 86: 1-27.

Walek, J. L., C. C. Baskin & J. M. Baskin. In press. Comparative achene germination requirements of the rockhouse endemic Ageratina luciae-brauniae and its widespread close relative A. altissima (Asteraceae). Amer. Midl. Naturalist.

Weaver, R. E., Jr. 1970. The Arenarias of the southeastern granitic flat-rocks. Bull. Torrey Bot. Club 97: 40-52.

Webb, W. S. & W. D. Funkhouser. 1936. Rock shelters in Menifee County, Kentucky. Univ. Kentucky Rep. Archaeol. Anthrop. 3: 105-147.

Weir, G. W. 1974. Geologic map of the Slade quadrangle, east-central Kentucky. U.S. Geol. Surv. Map GQ-1183.

Wells, E. F. 1979. Interspecific hybridization in eastern North American Heuchera (Saxifragaceae). Syst. Bot. 4: 319-338.

-----. 1984. A revision of the genus Heuchera (Saxifragaceae) in eastern North America. Syst. Bot. Monogr. 3: 45-121.

----- & B. A. Bohm. 1980. Chemotaxonomic studies in the Saxifragaceae s.l. 15. The flavonoids of subsection Villosae section Heuchera in the genus Heuchera. Canad. J. Bot. 58:1459-1463.

Welsh, S. L. 1989. On the distribution of Utah's hanging gardens. Great Basin Naturalist 49: 1-30.

-----, N. D. Atwood, S. Goodrich & L. C. Higgins (eds.). 1993. A Utah flora. Ed. 2 rev. Brigham Young University, Provo, Utah.

Wherry, E. T. 1939. Recent fern finds in West Virginia. Castanea 4: 1-4.

-----. 1957. Which northeastern ferns are evergreen? Amer. Fern J. 47: 93-95.

-----. 1961. The fern guide: Northeastern and midland United States and adjacent Canada. Doubleday and Co., Garden City, New York.

-----, 1964. Some new name-combinations for southeastern ferns. Amer. Fern J. 54: 143-146.

Wilken, D. H. 1993. Silene. Pages 488-493 in J. C. Hickman (ed.), The Jepson manual: Higher plants of California. University of California Press, Berkeley.

Willard, R. W. 1960. Beaver Reservoir IV - Nature fashions the stage. Arkansas Archeol. Soc. Newslett. 1(8): 1-5.

Willman, H. B. & J. C. Frye. 1980. The glacial boundary in southern Illinois. Illinois Geol. Surv. Div. Circ. 511.

-----, E. Atherton, T. C. Buschbach, C, Collinson, J. C. Frye, M, E. Hopkins, J. A. Lineback & J. A. Simon. 1975. Handbook of Illinois stratigraphy. Illinois Geol. Surv. Bull. 95.

Winterringer, G. S. 1949. Rock-ledge vegetation in southern Illinois. Ph.D. diss., University of Illinois, Urbana.

Wofford, B. E. 1976. The taxonomic status of Ageratina luciae-brauniae (Fern.) King & H. Robins. Phytologia 33:369-371.

-----. 1981. External seed morphology of Arenaria (Caryophyllaceae) of the southeastern United States, Syst. Bot. 6: 126-135.

----- & R. Kral. 1979. A new Arenaria (Caryophyllaceae) from the Cumberlands of Tennessee. Brittonia 31: 257-260.

----- & T. S. Patrick. 1980. Status report on Eupatorium luciae-brauniae Fernald. Report submitted to U.S. Fish and Wildlife Service, Atlanta.

----- & D. K. Smith. 1980. Status report on Arenaria cumberlandensis Wofford & Kral. Report submitted to U.S. Fish and Wildlife Service, Atlanta.

Wolfe, J. N. 1951. The possible role of microclimate. Ohio J. Sci. 51: 134-138.

-----, R. T. Wareham & H. T. Scofield. 1949. Microclimates and macroclimate of Neotoma, a small valley in central Ohio. Ohio Biol. Surv. Bull. 41.

Wyss, J. D. & S. K. Wyss. 1977. An archaeological assessment of portions of the Red River Gorge, Daniel Boone National Forest, Kentucky. Report submitted to U.S.D.A. Forest Service, Daniel Boone National Forest, Winchester, Kentucky.

Yatskievych, G., M. A. Homoya & D. R. Farrar. 1987. The fern genera Vittaria and Trichomanes in Indiana. Proc. Indiana Acad. Sci. 96: 429-434.

Young, R. & A. Young. 1992. Sandstone landforms. Springer-Verlag, Berlin.
COPYRIGHT 1996 New York Botanical Garden
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Walck, Jeffrey L.; Baskin, Jerry M.; Baskin, Carol C.
Publication:The Botanical Review
Date:Oct 1, 1996
Words:21137
Previous Article:Occurrence and taxonomic significance of ruminate endosperm.
Next Article:Azolla: a review of its biology and utilization.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters