2008 SOUTHEASTERN NATURALIST 7(3):515–526 Microsite Characteristics of Scutellaria montana (Lamiaceae) in East Tennessee John M. Mulhouse1, Matthew J. Gray1,*, and Charles W. Grubb1 Abstract - We surveyed 3 populations of Scutellaria montana (Large-fl owered Skullcap), a federally threatened mint, in southeastern Tennessee, and measured microsite characteristics between Large-fl owered Skullcap present and absent plots in close proximity. Large-fl owered Skullcap plots were typically associated with relatively open areas in forests. Further, some woody plants were positively associated with Large-fl owered Skullcap (e.g., Calycanthus fl oridus [Common Sweetshrub], Carya glabra [Pignut Hickory]), while others were negatively associated (e.g., Vaccinium stamineum [Gooseberry], Pinus virginiana [Virginia Pine]). Linear regression revealed that Large-fl owered Skullcap density increased with percent horizontal cover of grass (i.e., Poaceae) and decreased with percent vertical cover of vegetation. Our results suggest that suitable Large-fl owered Skullcap sites may be characterized by secondary forests with an open understory containing grass. Introduction Scutellaria montana Chapman (Large-fl owered Skullcap) is a federally threatened and globally imperiled (G2) perennial herb of the Lamiaceae family. Extremely restricted in range, Large-fl owered Skullcap has been recorded only in the forests of the Ridge and Valley and Cumberland Plateau Provinces of Georgia and Tennessee (USEPA 2002). Prior to our study, the species had been documented in 8 counties in northern Georgia and 3 in southeastern Tennessee, with populations suspected in northeastern Alabama (USFWS 1996, USEPA 2002). It has been suggested that Large-fl owered Skullcap is associated with mid- to late-successional forests, often positioned along ravines or stream bottoms (Lipps 1966). The forest composition of Large-fl owered Skullcap sites has been noted as containing an oak-hickory canopy with some occurrences of native pine species and an understory of deciduous shrubs and evergreen Vaccinium spp. (E. Bridges, Tennessee Natural Heritage Program [TNHP], Nashville, TN, unpubl. data). Sites also may contain a moderately dense herbaceous layer of mesic and xeric species in the understory (E. Bridges, unpubl. data). The substrate has been characterized as loose and well-drained soils, usually shallow, and slightly acidic (USEPA 2002). Large-fl owered Skullcap may also occur in the presence of exposed bedrock (E. Bridges, unpubl. data). Across its range, Large-fl owered Skullcap is found usually at low density, rarely exceeding more than a few plants per square meter (Cruzan 2001). 1Department of Forestry, Wildlife, and Fisheries, University of Tennessee, 274 Ellington Plant Sciences Building, Knoxville, TN 37996-4563. *Corresponding author - mgray11@utk.edu. 516 Southeastern Naturalist Vol.7, No. 3 Although the mechanisms for low density are unknown, low reproductive capacity has been cited as a contributing factor (Stirling 1983). Inasmuch as Large-fl owered Skullcap is federally threatened, there is a need to characterize microsite characteristics associated with its populations (USFWS 1996). This information is fundamental to understanding its life history and guiding management and conservation initiatives. Therefore, the objectives of our study were to: (1) quantify the microsite characteristics associated with Large-fl owered Skullcap populations, and (2) develop a predictive model for natural resource managers to assess habitat suitability. Methods Our study site was located along the escarpment of the Cumberland Plateau in southeastern Tennessee (35°21'20.65"N, 85°13'27.62"W), a region characterized by steep slopes, mountain streams, and rocky outcrops with caves interspersed. We sampled 3 previously unknown populations of Large-flowered Skullcap: 2 sites in Hamilton County and one in Bledsoe County, TN (Fig. 1). Forest fragment size at the Hamilton County sites was 29.3 and 69.5 ha (hereafter denoted as HC1 and HC2, respectively). The Bledsoe County site (BC1) was 242.4 ha. Each site was separated by >4.0 Figure 1. Locations of 3 Scutellaria montana (Large-fl owered Skullcap) populations on the Cumberland Plateau, TN, that were used in this study. 2008 J.M. Mulhouse, M.J. Gray, and C.W. Grubb 517 km and contained >300 Large-flowered Skullcap plants, which exceeds the established criteria for distinct and viable populations (i.e., >0.8 km separation and ≥50 individuals; Shea and Hogan 1998). For identification of Large-flowered Skullcap, we used Weakley (2006). Also, species confirmation was made in the field by David Lincicome (Rare Species Protection Program Administrator for the Tennessee Department of Environment and Conservation, Nashville, TN). We located populations and estimated density using belt transect sampling (Bullock 1996). Across all sites, we systematically placed transects every 91.44 m on north–south azimuths and traversed them from May–August 2006. When one or more Large-fl owered Skullcap plants were found, 400-m2 (20- x 20-m) plots were centered on the transect end-to-end and searched for the entire transect length. Density was estimated as the number of individuals per 400-m2 plot. Microsite characteristics were measured in September 2006 after transect surveys were completed. To quantify microsite characteristics for Large-flowered Skullcap, we measured variables known to potentially influence plant composition and growth. These variables included: basal area; percent canopy cover; number of canopy gaps, downed logs, and standing snags; species-specific woody plant density in the overstory, midstory, and understory; woody plant species richness; percent horizontal cover of herbaceous and woody plant life forms; percent vertical cover of vegetation; and soil properties (Higgins et al. 2005). Microsite characteristics were measured in two paired plots per transect. The first plot (hereafter called the present plot) was the 400-m2 plot associated with the greatest Large-flowered Skullcap density along a transect. The second plot (hereafter called the absent plot) was a randomly selected plot along the same transect containing no Large-flowered Skullcap plants. We measured basal area of all woody plants using a 10-basal-area-factor prism at plot center. Percent canopy cover was measured at plot center and in the 4 cardinal directions 10 m from plot center using a densiometer. Number of canopy gaps, downed logs (diameter > 11.43 cm), and standing snags (diameter-at-breast-height [DBH] > 11.43 cm) also were enumerated in each plot. Vertical structure of vegetation was quantified using a profile board. Our profile board was divided into 4 height strata (1 = 0.00–0.50 m, 2 = 0.51–1.00 m, 3 = 1.01–1.50 m, and 4 = 1.51–2.00 m), and each 0.5-m strata contained thirty 5- × 5-cm alternately colored boxes. Percent vertical structure was estimated by counting the number of boxes per strata that were covered over 50% by vegetation and dividing the number of covered boxes by 30 per strata. Vertical structure was measured 10 m from plot center in the 4 cardinal directions. Percent horizontal cover of bareground, detritus, forbs (i.e., non-Poaceae herbaceous vegetation), grasses (i.e., Poaceae), mosses, rocks, vines, and woody species was ocularly estimated in a 1-m2 plot around plot center. 518 Southeastern Naturalist Vol.7, No. 3 We measured overstory, midstory, and understory woody plant density in 3 nested square plots (400-m2, 100-m2, and 40-m2, respectively) around plot center. An overstory tree was considered to be a woody plant >1.4 m in height and >11.43 cm DBH. Overstory plants were identified to species and counted. A woody plant was counted in the midstory if it was >1.4 m height and ≤11.43 cm DBH. Density of midstory woody plants was enumerated for 4 DBH classes (1 = plants <2.54 cm, 2 = 2.55–5.08 cm, 3 = 5.09–7.62 cm, and 4 = 7.63–11.43 cm). Understory woody plants were ≤1.4 m in height and enumerated for 2 height strata: <10.0 cm and 10.0 cm–1.4 m. All woody plant species were identified using Radford et al. (1968) and Wofford and Chester (2002), and woody species richness per plot was enumerated. Soil chemical composition, percent organic matter, and temperature were measured 7 m from plot center along each of the 4 cardinal diagonals (i.e., NE, SE, SW, and NW). Soil temperature was measured using an Aquaterr® T-300 meter. Soil cores were collected and analyzed for phosphorus, potassium, pH, and percent organic matter by the University of Tennessee Agricultural Extension Service Soil Testing Laboratory (http://bioengr.ag.utk.edu/ SoilTestLab/) using standard techniques. We provide descriptive statistics (mean, standard deviation, and standard error) for all statistically significant variables in Large-fl owered Skullcap present and absent plots. Because it was not reasonable to assume that paired plots on a transect were independent, we tested for differences (α = 0.05) in microsite characteristics between present and absent plots using a two-sided paired t-test. In cases when the distribution of data was non-normal (i.e., Shapiro-Wilk W test, P < 0.05), we used a non-parametric 2-sample Wilcoxon signed-rank test to quantify differences. This test quantifies median differences; however, we present the aforementioned parametric descriptive statistics to facilitate interpretation. We also constructed a linear regression model using stepwise selection (enter and stay α = 0.05) to identify variables that were important in explaining Large-fl owered Skullcap density, and to develop a tool to identify suitable sites for Large-fl owered Skullcap management. The response variable was density of Large-fl owered Skullcap and explanatory variables were all microsite characteristics. Number of gaps and soil temperature were not used due to missing data at some plots. All analyses were performed using Microsoft® Excel, SPSS (SPSS Inc. 2006), JMP® (SAS Institute 2005), and the SAS® system (SAS Institute 2003). Results Large-fl owered Skullcap was found on 19 transects (7 at HC1, 2 at HC2, and 10 at BC1). Absolute density ranged from 9–81 plants (mean = 20.9 ± 16.3 [1 SD]) per 400-m2 at Large-fl owered Skullcap present plots. Mean density of Large-fl owered Skullcap across present and absent plots at HC1, HC2, and BC1 was 75.4, 73.9, and 51.0 plants/ha, respectively. 2008 J.M. Mulhouse, M.J. Gray, and C.W. Grubb 519 Mean basal area in Large-fl owered Skullcap absent plots (23.2 ± 7.7 m2/ha) was greater (Z = -2.06, P = 0.04) than in present plots (19.9 ± 6.3 m2/ha). Percent vertical vegetation cover in absent plots also was greater (t18, 0.05 = -2.06, P = 0.05) than in present plots for the uppermost strata (1.51–2.00 m) on the profile board (Table 1). No differences were detected for the other 3 profile board strata (t18, 0.05 ≥ -1.66, P ≥ 0.12), but there was a trend for less vertical cover of vegetation in Large-fl owered Skullcap present plots (Table 1). Large-fl owered Skullcap plots had greater percent horizontal cover of grasses (Z = -2.33, P = 0.02) and vines (Z = -2.04, P = 0.04) in the understory than absent plots (Table 2). No differences were detected for other horizontal cover categories (Z ≥ -1.76, P ≥ 0.08), although on average, over 63% of cover was detritus across all plots (Table 2). No differences also were detected (Z ≥ -1.00, P ≥ 0.32) between present Table 1. Percent vertical cover1 of vegetation in Scutellaria montana (Large-fl owered Skullcap) present and absent plots, Hamilton and Bledsoe counties, TN, September 2006. Absent Present Strata2 Mean3,4 SD SE Mean SD SE 1 73.6 16.3 3.7 70.5 16.3 3.7 2 57.0 16.3 3.7 47.2 23.8 5.5 3 50.0 19.1 4.4 39.9 23.1 5.3 4 53.5A 26.3 6.0 35.5B 21.8 5.0 Overall 58.5 15.5 3.6 48.3 18.1 4.1 1Vertical cover measured using a profile board at 10 m from plot center; each 0.5-m strata contained thirty 5 × 5 cm boxes that were considered covered if > 50% of a box was obscured by vegetation; measurements were taken in each cardinal direction and averaged by strata prior to analyses. 2Height strata on the profile board were 1 = 0.00–0.50 m, 2 = 0.51–1.00 m, 3 = 1.01–1.50 m, 4 = 1.51–2.00 m, and overall = all strata combined. 3n = 19 present and absent plots. 4Means within rows followed by unlike letters are different (P ≤ 0.05); no letters imply statistical differences were not detected. Table 2. Percent horizontal cover1 of vegetation life forms in present and absent Scutellaria montana (Large-fl owered Skullcap) plots, Hamilton and Bledsoe counties, TN, September 2006. Absent Present Life form Mean2,3 SD SE Mean SD SE Bareground 2.4 6.3 1.4 4.2 12.0 2.8 Detritus 72.2 16.3 3.7 63.9 18.0 4.1 Forb 5.8 7.3 1.7 8.1 7.3 1.7 Grass 0.1A 0.5 0.1 2.1B 4.0 0.9 Moss 0.8 3.4 0.8 0.0 0.0 0.0 Rock 5.5 10.9 2.5 3.4 7.6 1.8 Vine 3.4A 4.9 1.1 6.2B 5.1 1.2 Woody 9.8 10.4 2.4 12.1 7.2 1.6 1Percent horizontal cover was estimated in a 1-m2 plot. 2n = 19 present and absent plots. 3Means within rows followed by unlike letters are different (P ≤ 0.05); no letters imply statistical differences were not detected. 520 Southeastern Naturalist Vol.7, No. 3 and absent plots for mean number of gaps, logs, and snags (ca. <2 per 400 m2), and mean percent canopy cover (ca. >90%). Species richness of understory woody plants in Large-fl owered Skullcap present plots was greater (t18, 0.05 ≤ -2.09, P ≤ 0.05) than in absent plots for the tallest understory height stratum and both strata combined (Table 3). Understory woody plant density also was greater (t18, 0.05 ≤ -2.10, P ≤ 0.05) in present plots for the tallest understory height stratum and both strata combined (Table 4). No differences were detected (Z ≥ -1.73, P ≥ 0.08) in species richness and plant density between Large-fl owered Skullcap present and absent plots for the lowest understory height stratum, the midstory, and the overstory (Tables 3 and 4). Table 3. Species richness of understory, midstory, and overstory vegetation in present and absent Scutellaria montana (Large-fl owered Skullcap) plots, Hamilton and Bledsoe counties, TN, September 2006. Absent Present Strata1,2 Mean3,4 SD SE Mean SD SE Understory 1 3.8 2.4 0.5 3.6 2.4 0.6 Understory 2 8.3A 3.8 0.9 10.3B 2.6 0.6 Understory 1 and 2 8.7A 4.0 0.9 11.1B 2.6 0.6 Midstory 6.5 2.6 0.6 6.3 2.9 0.7 Overstory 6.1 2.0 0.4 6.0 1.7 0.4 Overall 15.1 3.5 0.8 16.4 3.9 0.9 1Understory vegetation was woody plants measured in two height strata: 1 = <10 cm and 2 = 10 cm–1.4 m, midstory vegetation was woody plants >1.4 m in height and ≤11.43 cm diameter-atbreast- height (DBH), and overstory vegetation was woody plants >1.4 m in height and >11.43 cm DBH. Overall richness was total number of unique species across all strata. 2Plot size was 40 m2, 100 m2, and 400 m2 for understory, midstory, and overstory, respectively. 3n = 19 present and absent plots. 4Means within rows followed by unlike letters are different (P ≤ 0.05); no letters imply statistical differences were not detected. Table 4. Total understory, midstory, and overstory vegetation density in present and absent Scutellaria montana (Large-fl owered Skullcap) plots, Hamilton and Bledsoe counties, TN, September 2006. Absent Present Strata1,2 Mean3,4 SD SE Mean SD SE Understory 1 13.8 14.8 3.4 13.2 12.3 2.8 Understory 2 48.1A 29.3 6.7 62.9B 21.5 4.9 Understory 1 and 2 61.9A 33.3 7.6 76.1B 23.6 5.4 Midstory 26.5 12.7 2.9 25.6 15.4 3.5 Overstory 14.4 7.1 1.6 12.7 5.4 1.2 1Understory vegetation was woody plants measured in two height strata: 1 = <10 cm and 2 = 10 cm–1.4 m, midstory vegetation was woody plants >1.4 m in height and <11.43 cm diameter-atbreast- height (DBH), and overstory vegetation was woody plants >1.4 m in height and >11.43 cm DBH. 2Plot size was 40 m2, 100 m2, and 400 m2 for understory, midstory, and overstory, respectively. 3n = 19 present and absent plots. 4Means within rows followed by unlike letters are different (P ≤ 0.05); no letters imply statistical differences were not detected. 2008 J.M. Mulhouse, M.J. Gray, and C.W. Grubb 521 Density of several woody plant species was different between Largefl owered Skullcap present and absent plots. In the understory, density of Calycanthus fl oridus L. (Common Sweetshrub) and Vitis vulpina L. (Frost Grape) was greater (Z ≤ -2.33, P ≤ 0.02) in present plots than in absent plots (Table 5). In contrast, Vaccinium stamineum L. (Gooseberry) density was greater (Z = -2.07, P = 0.04) in absent plots than in present plots (Table 5). Red Maple (Acer rubrum L.) seedlings were relatively dense in the understory (ca. 10 seedlings/40-m2 plot), although no differences were detected (Z = -0.28, P = 0.78) between present and absent plots. In the midstory, density of Tsuga canadensis (L.) Carr. (Eastern Hemlock) was greater (Z = -2.35, P = 0.02) in Large-fl owered Skullcap absent plots than in present plots (Table 5). Similar to the understory, Red Maple had a relatively high density in the midstory (ca. 6 saplings/100-m2 plot), but differences were not detected (Z = -0.03, P = 0.98) between present and absent plots. In the overstory, density of Carya glabra (Miller) Sweet (Pignut Hickory) was greater (Z = -2.56, P = 0.007) in Large-fl owered Skullcap plots than in absent plots (Table 5). The density of Quercus montana Willd. (Chestnut Oak) and Pinus virginiana Miller (Virginia Pine) was greater (Z ≤ -2.06, P ≤ 0.04) in Large-fl owered Skullcap absent plots than in present plots (Table 5). Red Maple, Quercus alba L. (White Oak), and Oxydendrum arboreum (L.) DC (Sourwood) had relatively high densities in the overstory (ca. 2 trees/400-m2 plot), but differences were not detected (Z ≥ -1.85, P ≥ 0.07) between Largefl owered Skullcap present and absent plots. Fifty additional woody plant species were recorded during sampling, most at relatively low density in the Table 5. Density of overstory, midstory, and understory woody vegetation that differed statistically (P < 0.05) between present and absent Scutellaria montana (Large-fl owered Skullcap) plots, Hamilton and Bledsoe counties, TN, September 2006. Absent Present Story1,2/species Mean3,4 SD SE Mean SD SE Under Common Sweetshrub 3.2A 5.1 1.2 12.7B 13.3 3.1 Gooseberry 10.7A 19.0 4.4 2.2B 3.9 0.9 Frost Grape 0.2A 0.5 0.1 1.2B 1.5 0.4 Mid Eastern Hemlock 1.2A 2.3 0.5 0.1B 0.3 0.1 Over Pignut Hickory 0.7A 1.2 0.3 1.9B 1.4 0.3 Virginia Pine 1.5A 3.1 0.7 0.1B 0.3 0.1 Chestnut Oak 3.4A 4.4 1.0 1.0B 1.3 0.3 1Understory vegetation was woody plants ≤1.4 m in height, midstory vegetation was woody plants >1.4 m in height and ≤11.43 cm diameter-at-breast-height (DBH), and overstory vegetation was woody plants >1.4 m in height and >11.43 cm DBH. 2Plot size was 40 m2, 100 m2, and 400 m2 for understory, midstory, and overstory, respectively. 3n = 19 present and absent plots. 4Means within rows followed by unlike letters are different (P ≤ 0.05); density of non-signifi- cant species were not presented. 522 Southeastern Naturalist Vol.7, No. 3 understory (<3 plants/40-m2 plot), midstory (<2.5 plants/100-m2 plot), and overstory (<1.5 plants/400-m2 plot). No differences in density were detected (Z ≥ -1.85, P ≥ 0.06) between present and absent sites for these woody plant species. The only difference detected in soil characteristics between Large-fl owered Skullcap present and absent plots was soil pH; it was 6% greater (t18, 0.05 = 3.20, P = 0.005) on average in present plots (5.0 ± 0.2) compared to absent plots (4.7 ± 0.3). No differences were detected (Z ≥ -1.00, P ≥ 0.30) for the other soil variables, although mean phosphorus level was 5.6 kg/ha in Largefl owered Skullcap absent plots compared to 7.3 kg/ha in present plots. Mean percent organic matter was approximately 5% and mean soil temperature was approximately 19 ºC in both plot types. Stepwise linear regression produced a final model with two forest structure variables that explained significant variation (F2,35 = 51.67, P < 0.001) in Large-fl owered Skullcap density (Table 6). There was a strong positive relationship (P < 0.001; partial R2 = 0.70) between percent horizontal cover of grass and Large-fl owered Skullcap density. Large-fl owered Skullcap density was negatively related (P = 0.02; partial R2 = 0.05) with percent vertical vegetation cover. The overall coefficient of determination adjusted for the two parameters in the model was 0.73, and there was no evidence of multicollinearity between variables (i.e., all variance infl ation factors < 10; Table 6). The final model was: Large-fl owered Skullcap density = 15.76 + 4.58 (% grass cover) - 0.19 (% vertical vegetation cover) (Table 6). Discussion The discovery of Large-fl owered Skullcap in Bledsoe County, TN represents the northernmost record of this species within its known range. A voucher specimen from this colony is filed with the University of Tennessee Herbarium (JMM 02005). In general, Large-fl owered Skullcap sites could be characterized by an intact forest with an open understory that had a grass and vine component. Common sweetshrub and frost grape were common species in the understory at Large-fl owered Skullcap sites. The most common overstory trees were red maple, white oak, sourwood, and pignut hickory. Red maple also was common in the midstory and understory. Percent horizontal cover of grass was significantly greater in Large- Table 6. Linear regression statistics and the final prediction model (i.e., parameter estimates, bi) that explained significant variation in large-fl owered skullcap density in east Tennessee. Variable1,2 bi SE t-value p-value Partial r2 VIF3 Intercept 15.76 4.26 3.70 0.007 * 0 Percent grass cover 4.58 0.45 10.10 <0.001 0.70 1.02 Percent VC -0.19 0.08 -2.50 0.018 0.05 1.02 1Variables selected using stepwise selection and parameters estimated using least-squares estimation. 2VC = Vertical vegetation cover measured using a profile board. 3VIF = Variance infl ation factor; VIF > 10 is suggestive of multicollinearity. 2008 J.M. Mulhouse, M.J. Gray, and C.W. Grubb 523 flowered Skullcap present sites and explained 70% of the variation in Large-flowered Skullcap density. Percent vertical cover of vegetation (as measured with a profile board) explained an additional 5% of the variation in Large-flowered Skullcap density. Our linear model indicated that if percent grass cover equaled zero, Large-flowered Skullcap would not occur if total percent vertical cover of vegetation was >83%. Both of these results illustrate the importance of an open understory for Large-flowered Skullcap with a grass component. Initial studies suggested Largeflowered Skullcap was a late-successional forest species, but more recently it has been considered a mid-successional species that co-occurs with trees averaging <60 years old, persisting in areas where light penetrates to the forest floor (Shea and Hogan 1998, Sutter 1993, USEPA 2002). Although we did not measure forest floor light intensity, and canopy cover was not significantly different between Large-flowered Skullcap present and absent sites, the higher grass (and vine) component at Large-flowered Skullcap present sites may suggest light availability is greater at these sites. More research is needed to make conclusive inferences on the relationship between Large-flowered Skullcap locations, forest floor light intensity, and state of succession. Percent vegetation cover in the 1.51–2.00 m vertical strata of the profile board and overstory basal area were significantly less in Large-fl owered Skullcap present sites, providing additional evidence that this species may be associated with relatively open forests. Fail and Sommers (1993) reported that Large-fl owered Skullcap sites in northwest Georgia were characterized by low basal area and vegetative biomass. Canopy disturbance also has been considered beneficial for Large-fl owered Skullcap (Sutter 1993, USEPA 2002). Nix (1993) stated that light afforded by broken canopy might be the most important parameter in determining growth and survival of Large-fl owered Skullcap. Although average canopy cover at Large-fl owered Skullcap present plots was 92% in September when forest microsite variables were measured, it would have been less in early spring when Large-fl owered Skullcap was germinating. The role of natural and man-created canopy gaps in the establishment of Large-fl owered Skullcap needs further investigation. The highest density of Large-flowered Skullcap was 81 plants in a 400- m2 plot, or 2025 individuals per ha. Interestingly, at this plot, the canopy was disturbed (percent canopy = 39.4%), further suggesting the possible importance of light penetration to the forest floor. However, it has been suggested that canopy disturbance that occurs in conjunction with soil disturbance may result in Large-flowered Skullcap being out-competed by other herbaceous species (Sutter 1993, USEPA 2002). Faulkner (1993) documented Large-flowered Skullcap in an area where logging and lowintensity ground fires had occurred, but suggested that those plants probably established prior to disturbance and colonization of additional plants was unlikely. Hamel and Somers (1992) also reported a dense population of 524 Southeastern Naturalist Vol.7, No. 3 Large-flowered Skullcap along a logging road on the Cumberland Plateau Escarpment. We made similar observations of Large-flowered Skullcap in cutover areas, near ATV trails, and along old roads (J.M. Mulhouse, pers. observ.). It is unknown whether Large-flowered Skullcap occurring in these disturbed sites will persist. Additional research is needed to quantify the extent of canopy and soil disturbance that induces reduced survival and reproduction of Large-flowered Skullcap. Some positive and negative associations between woody species and Large-flowered Skullcap were documented. In the overstory, Pignut Hickory was significantly more dense in Large-flowered Skullcap plots while chestnut oak was significantly less dense. Similar results have been reported by others and may be a consequence of different microsite conditions created by these tree species (Fail and Sommers 1993). White Oak was the second densest species in Large-flowered Skullcap plots, which Fail and Sommers (1993) also reported. However, we also found that there were no significant differences in White Oak overstory density between present and absent plots, thus the association of White Oak with Large-flowered Skullcap may be less important at our sites. Fail and Sommers (1993) reported that Sourwood was absent at Large-flowered Skullcap sites, but it was a relatively common overstory tree at our sites. Red Maple was common in the overstory, midstory, and understory, which is consistent with Fail and Sommers (1993) findings. Lastly, we found that Virginia Pine occurred at significantly higher densities at Largeflowered Skullcap absent sites. This result differs from Fail and Sommers (1993), who suggested that conditions underneath Pinus echinata Miller (Shortleaf Pine) trees could be beneficial for Large-flowered Skullcap. On the other hand, it has been found (E. Bridges, unpubl. data) that pine is a minor canopy component where Large-flowered Skullcap occurs. The negative or lack of association between Large-flowered Skullcap and pines needs further investigation. Few midstory plant associations with Large-flowered Skullcap were apparent. Eastern Hemlock density in the midstory was significantly less at Large-flowered Skullcap plots. This result is not surprising because Eastern Hemlock is known to be allelopathic (Ward and McCormick 1982). In the understory, we found that Common Sweetshrub and Frost Grape had significantly greater density at Large-flowered Skullcap present plots. Higher density of Common Sweetshrub at Large-flowered Skullcap sites is similar to others (E. Bridges, unpubl. data) that suggested a deciduous shrub layer commonly occurs with Large-flowered Skullcap. The co-occurrence of Large-flowered Skullcap and Vaccinium spp. also has been found (E. Bridges, unpubl. data), which differs from our results. We found that Gooseberry plant density was greater at Large-flowered Skullcap absent sites. These results suggest there may be differences in Large-flowered Skullcap plant associations depending on site conditions. Lastly, we found that density and richness of understory seedlings was significantly greater at Large-flowered Skullcap plots, which may have 2008 J.M. Mulhouse, M.J. Gray, and C.W. Grubb 525 been a result of increased light availability to the forest floor at these sites (Goldberg and Miller 1990, Tilman 1993). Mean pH at Large-fl owered Skullcap absent and present plots was slightly acidic (4.7 and 5.0, respectively), and within the range (4.5–6.3) suggested by J.L. Collins (Vanderbilt University, unpubl. data) for Largefl owered Skullcap occurrence. Soils were uniformly low in phosphorus and potassium content and high in organic matter. Other soil features were comparable between plots; thus, it does not appear that soil is responsible for the distribution of Large-fl owered Skullcap at our sites. Conclusion Our linear model can be used to identify the suitability of sites for Largefl owered Skullcap by measuring two simple variables in the field: (1) percent horizontal cover in a 1-m2 plot and (2) percent vertical cover measured using our design for a profile board. Field measurements of these variables can be entered into the model to predict Large-fl owered Skullcap density (see model parameter estimates in Table 6). Negative or zero predictions suggest a site that is unsuitable, whereas positive values suggest potentially suitable sites. Ideally, the accuracy of our model should be validated prior to field use. Also, we caution against use of our model outside the region and ecosystem where it was developed (i.e., the forested Cove-Gulf Region of east Tennessee). In addition, given the clear association of Large-fl owered Skullcap with light availability, judicious thinning of trees and tall underbrush may not negatively impact the species and may even increase habitat quality. This hypothesis needs to be tested. Acknowledgments We thank Bowater Incorporated for providing funding for this research and access to research sites. We especially thank the following Bowater employees: Barry Graden for assistance in securing funds, Don King for extensive help with site access, and Kevin Gallagher and Mike Williford for providing field maps and transect coordinates for our research. We also thank David Lincicome at Tennessee Natural Heritage for assisting with taxonomy, Dr. Nicholas Herrmann at the University of Tennessee for providing GIS data for our sites, and Justin Geise and Andrea Moodhart for field assistance. Finally, we thank 2 anonymous referees for comments on our manuscript. Literature Cited Bullock, J. 1996. Plants. Pp. 111–138, In W.J. Sutherland (Ed.). Ecological Census Techniques: A Handbook. Cambridge University Press, Cambridge, UK. 336 pp. Cruzan, M.B. 2001. Population size and fragmentation thresholds for the maintenance of genetic diversity in the herbaceous endemic Scutellaria montana (Lamiaceae). Evolution 55:1569–1580. 526 Southeastern Naturalist Vol.7, No. 3 Fail, J., Jr., and R. Sommers. 1993. Species associations and implications of canopy change for an endangered mint in a virgin oak-hickory-pine forest. The Journal of the Elisha Mitchell Scientific Society 109:51–54. Faulkner, J. 1993. A survey of Scutellaria montana in the Hick’s Gap Area of Marion County, Tennessee. Tennessee Department of Environment and Conservation, Nasvhille, TN. 22 pp. Goldberg, D.E., and T.E. Miller. 1990. Effects of different resource additions on species diversity in an annual plant community. Ecology 71:213–225. Hamel, P., and P. Somers. 1992. Final report: Biological survey of the North Chickamauga River Gorge and synopsis of significant elements. Prepared for the Nature Conservancy, Georgia Field Office, Atlanta, GA. 18 pp. Higgins, K.F., K.J. Jenkins, G.K. Clambey, D.W. Uresk, D.E. Naugle, J.E. Norland, and W.T. Barker. 2005. Vegetation sampling and measurements. Pp. 524–553, In C.E. Braun (Ed.). Techniques for Wildlife Investigation and Management. The Wildlife Society, Bethesda, MD. 974 pp. Lipps, E.L. 1966. Plant communities of a portion of Floyd County, Georgia— Especially the Marshall Forest. Ph.D Dissertation. University of Tennessee, Knoxville, TN. 206 pp. Nix, T.L. 1993. Intensive monitoring of Scutellaria montana Chapman in the Marshall Forest. Unpublished report. The Nature Conservancy, Georgia Field Office, Atlanta, GA. Radford, A.E., H.E. Ahles, and C.R. Bell. 1968. Manual of the Vascular Flora of the Carolinas. University of North Carolina Press, Chapel Hill, NC. 1183 pp. SAS Institute. 2003. SAS/STAT®. Version 9.1. SAS Institute, Inc. Cary, NC. SAS Institute. 2005. JMP Version 6.0.0. SAS Institute, Inc. Cary, NC. Shea, A., and T. Hogan. 1998. Status survey report on Scutellaria montana Chapman (Large-fl owered Skullcap). Prepared for the US Fish and Wildlife Service, Asheville, NC. 98 pp. SPSS Inc. 2006. SPSS for Windows version 15.0 SPSS Inc. Chicago, IL. Stirling, B.K. 1983. Reduced reproductive capacity found in the rare mint Scutellaria montana Chapman. Abstract in Georgia Journal of Science 41:9. Sutter, R. 1993. Memo to Chickamauga–Chattanooga National Military Park: Pat Reed, Sam Weddle, and Charlie Spear, and the National Park Service. The Nature Conservancy, Southeast Regional Office, Chapel Hill, NC. Tilman, D. 1993. Species richness of experimental productivity gradients: How important is colonization limitation? Ecology 74:2179–2191. United States Environmental Protection Agency (USEPA). 2002. Endangered and threatened wildlife and plants; Reclassification of Scutellaria montana (Largefl owered Skullcap) from endangered to threatened. Federal Register Environmental Documents 67:1662–1668. United States Fish and Wildlife Service (USFWS). 1996. Large-fl owered Skullcap recovery plan. Asheville, NC. 31 pp. Ward, H.A., and L.H. McCormick. 1982. Eastern Hemlock allelopathy. Forest Science 28:681–686. Weakley, A.S. 2006. Flora of the Carolinas, Virginia, Georgia, and surrounding areas. Working Draft of 6 January 2006. University of North Carolina, Chapel Hill, NC. 1082 pp. Wofford, B.E., and E.W. Chester. 2002. Guide to the Trees, Shrubs, and Woody Vines of Tennessee. The University of Tennessee Press, Knoxville, TN. 286 pp.