Forest Ecology and Management 217 (2005) 54–66 www.elsevier.com/locate/foreco

The utility of Trillium and Maianthemum as phyto-indicators of deer impact in northwestern Pennsylvania, USA Chad D. Kirschbaum a,*, Brian L. Anacker b a

Wayne National Forest, Sand County Foundation, Ironton Ranger District, 6518 State Route 93, Pedro, OH 45659, USA b Sand County Foundation, 223 Kentucky Street, Petaluma, CA 94952, USA Received 23 June 2004; received in revised form 1 May 2005; accepted 2 May 2005

Abstract Trillium spp. and Maianthemum canadense Desf. are preferred deer forage throughout their range and have potential as indicators of deer impact throughout the eastern United States. This paper evaluates the use of these plants as indicators at a landscape level within the Kinzua Quality Deer Cooperative (KQDC,
30,000 ha) in McKean County, Pennsylvania where a high level of deer impact has been sustained over the last 60 years. We hypothesized that deer impacts to Maianthemum and Trillium spp. populations across the KQDC landscape would resemble those reported at much smaller scales in previous studies and would indicate spatial variation in deer impact within the KQDC area. We found that most of the Trillium and Maianthemum characteristics measured in the KQDC were below the ‘‘healthy’’ standards established in the literature and comparable to characteristics associated with high deer impacts: the percentage of Trillium flowering in the KQDC was 9.1%, compared to the standard of 21–34%; the average Trillium heights in 27% of the KQDC plots were below the 12–14 cm recommended for healthy Trillium populations; Maianthemum mean leaf length and percent flowering were 39 mm and 0.3%, compared to standards of 42 mm and 0.5% for plants in browse accessible areas and 55 mm and 20% flowering for plants in refugia. Despite these high levels of impact, no significant correlations were found between the indicator variables and a direct measure of deer impact, a browse index based on browsing measurements of preferred woody plant species. Based on these results we conclude that these two species do not track spatial variation in deer impact across the KQDC. Relationships between deer impact and indicator characteristics could be clouded by factors such as environmental variables not measured in our study and legacy effects of high deer abundance. Both species will bear further investigation in a long-term monitoring project within the KQDC and could serve as a measurement of recovery in areas that have experienced high deer impact when deer populations are lowered and monitoring sites are analyzed through time. # 2005 Elsevier B.V. All rights reserved. Keywords: White-tailed deer; Indicator plants; Trillium; Maianthemum; Pennsylvania; Kinzua Quality Deer Cooperative; McKean County; Herbivory; Deer impact; Browse; Plant–animal interaction

* Corresponding author. Tel.: +1 740 534 6500. E-mail address: [email protected] (C.D. Kirschbaum). 0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2005.05.001

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

1. Introduction Evidence of damage to forest ecosystems resulting from browsing by white-tailed deer (Odocoileus virginianus Zimm.) across the northeastern states has been accumulating for several decades. This damage includes reductions in the abundance of preferred woody species, alterations in the size and reproductive status of herbaceous species and the spread of invasive species (Leopold et al., 1947; Hough, 1965; Alverson et al., 1988; DeGraaf et al., 1991; Anderson, 1994; Balgooyen and Waller, 1995; McShea et al., 1997; Ruhren and Handel, 2000; Vellend, 2002; Horsley et al., 2003). Deer are also implicated in changes to species richness and abundance in understory plant communities (Tilghman, 1989; Rooney and Dress, 1997). Intensive deer impact has been sustained over the last 60 years in northwestern Pennsylvania (Hough, 1965; Redding, 1995) and this area has been the focal point of much of the research above. Redding (1995) reported that deer populations have been above densities associated with adequate forest regeneration since at least the 1950s. Fencing can be used to protect areas from these impacts (Horsley et al., 2003). Fences create temporary islands of zero deer density, which are presumably effective for plants that can grow out of the reach of deer by the time of fence removal. However, permanent fencing is required for the protection or restoration of species spending their entire life cycle in the forest understory. Long-term fence maintenance and vegetation monitoring can be prohibitively expensive. 1.1. Indicators of deer impact As part of managing deer density to sustain or restore understory layers, managers need effective monitoring programs and effective variables to monitor. A number of studies suggest that preferentially browsed plants may be used as indicators of deer impact to a plant community. At least twenty species have been studied (Table 1), though few of the proposed indicators have been tested in multiple localities throughout their ranges. Ideally, an indicator plant would occur frequently across a site and have easily measurable attributes that are strongly correlated with measures of deer impact such as deer browse and deer density (Noss and Cooperider, 1994; Hilty and Merenlender, 2000). Thus, an indicator plant

55

species must be sufficiently resilient to deer browsing to persist in systems with a range of deer impacts. Trillium spp. (Liliaceae) and Maianthemum canadense Desf. (Liliaceae) have potential as indicators of deer impact in the hardwood forests of the Allegheny Plateau region since they occur frequently (Rhoads and Klein, 1993) and are preferred deer food species (Skinner and Telfer, 1974). At high deer densities, deer impact to Trillium is manifested in lower stem heights (Anderson, 1994), lower density of flowering plants (Anderson, 1994) and lower leaf area (Augustine and Frelich, 1998). Rooney (1997) found that Maianthemum shoot densities and leaf lengths were significantly greater on high boulders (no deer access) as opposed to low boulders where deer have access. Also Maianthemum found on tall boulders were more likely to flower than Maianthemum on low boulders. In Wisconsin, the frequency of Maianthemum was positively correlated with recent deer densities and percent cover of Maianthemum was negatively correlated with historic deer densities (Balgooyen and Waller, 1995). In Pennsylvania, neither Trillium spp. nor Maianthemum have been tested specifically as phyto-indicators of deer impact at a landscape scale where deer densities (and thus impact) are variable across the landscape. 1.2. The Kinzua Quality Deer Cooperative The Kinzua Quality Deer Cooperative (KQDC) is 30,628-ha adaptive management demonstration project in the Allegheny Plateau region of northwestern Pennsylvania (Fig. 1). The objectives of the project are to restore ecosystem health to an area long impacted by overabundant deer while improving deer habitat and the quality of hunting in the region (KQDC Management Plan, 2002). The county-wide deer density in McKean County averaged 10 deer/km2 during the period 1990–2001 (KQDC Management Plan, 20021), 20% above the PA Game Commission’s goal density of 8 deer/km2. In earlier decades, regional deer densities reached levels of 15–23 deer/km2 (Redding, 1995).

1

Sand County Foundation. P.O. Box 3186, Madison, WI 537043186. Contact Kevin McAleese, KQDC project director. http:// www.sandcounty.net.

56

Table 1 Potential phyto-indicator species for evaluating deer impact in the northeastern United States State

Studied variable

Results

Reference

Acer saccharum

MI, WI WI WI MI, WI

Browse Browse Browse Browse

Correlated with hemlock saplings 30–99 cm tall Highly correlated with recent deer densities Weakly correlated with frequency of Clintonia borealis Significant interaction with hemlock reproduction (0.3–1.4 m)

Rooney et al. (2000) Balgooyen and Waller (1995) Balgooyen and Waller (1995) Rooney et al., 2000

Acer spicatum

WI

Frequency

() Correlation with recent and historic*recent deer densities

Balgooyen and Waller (1995)

Actea pachypoda

IN

Height

Significant correlations with an herbaceous deer impact index

IN

Height

() Correlation with deer harvested per hunter day

Webster and Parker (2000), Webster et al. (2001) Webster and Parker (2000)

Arailia nudecalis

WI

Frequency Percent cover

() Correlation with ‘‘previous’’ deer densities () Correlation with historic deer densities

Balgooyen and Waller (1995) Balgooyen and Waller (1995)

Arisaema tryphyllum

IN

Height

Significant correlations with a herbaceous deer impact index

IN IN VA VA NJ

Height Height Stem density Fecundity Fecundity

Significantly taller stems in hunted areas () Correlation with deer harvested per hunter day Significantly higher stem densities inside fences Correlated with high and low deer densities Lower than expected fruit and flower reproduction

Webster and Parker (2000), Webster et al. (2001) Webster and Parker (2000) Webster and Parker (2000) Fletcher et al. (2001) Fletcher et al. (2001) Ruhren and Handel (2000)

Aster divaricatus

PA

% Browseable units browsed

Browsed disproportionally more than the rest of the herb. layer

Williams et al. (2000)

Aster prenanthoides

PA

% Browseable units browsed

Browsed disproportionally more than the rest of the herb. layer

Williams et al. (2000)

Betula allegheniensis

WI

Frequency

() Correlation with recent and historic deer densities

Balgooyen and Waller (1995)

Chelone glabra

PA PA

% Stems browsed % Browseable units browsed

(+) Correlation with % browseable units of the herb. layer browsed Browsed disproportionally more than the rest of the herb. layer

Williams et al. (2000) Williams et al. (2000)

Clintonia borealis

WI WI WI WI WI WI

Frequency Percent cover No. of leaves/plant Mean scape height Size Pedicels/umbel

() () () () () ()

Balgooyen Balgooyen Balgooyen Balgooyen Balgooyen Balgooyen

Impatiens capensis

PA

% Browseable units browsed

Browsed disproportionally more than the rest of the herb. layer

Williams et al. (2000)

Laportea canadensis

MN

Abundance (density)

Correlated with deer density

Augustine and Jordan (1998)

Maianthemum canadense

PA PA PA WI WI

Leaf length Shoot density Flowering Frequency Frequency

Significantly longer on tall boulders than on low boulders Significantly greater on tall boulders than on low boulders Shoots more likely to flower on tall boulders than low boulders () Correlation with historic deer densities (+) Correlation with recent deer densities

Rooney (1997) Rooney (1997) Rooney (1997) Balgooyen and Waller (1995) Balgooyen and Waller (1995)

Correlation Correlation Correlation Correlation Correlation Correlation

with with with with with with

historic and recent deer densities historic and recent deer densities recent deer densities historic and recent deer densities historic densities recent deer densities

and and and and and and

Waller Waller Waller Waller Waller Waller

(1995) (1995) (1995) (1995) (1995) (1995)

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

Species

WI

Percent cover

() Correlation with historic deer densities

Balgooyen and Waller (1995)

Orchis specatabilis

VA VA

Stem density Reproductive activity

Significantly higher stem densities inside fences Correlated with high and low deer densities

Fletcher et al. (2001) Fletcher et al. (2001)

Osmorrhiza claytonii

IN IN IN

Height Height Height

Significantly taller stems in hunted areas () Correlation with deer harvested per hunter day Significant correlations with an herbaceous deer impact index

Webster Webster Webster Webster

Smilicina spp.

VA VA

Stem density Reproductive activity

Significantly higher stem densities inside fences Correlated with high and low deer densities

Fletcher et al. (2001) Fletcher et al. (2001)

Sorbus decora

WI

Frequency

() Correlation with recent and historic deer densities

Balgooyen and Waller (1995)

Taxus canadensis

WI

Frequency

() Correlation with recent and historic deer densities

Balgooyen and Waller (1995)

Trillium grandiflorum

IL IL IL MN MN

Stem height Stem height Flowering density Grazing intensity Fecundity

Anderson (1994) Anderson (1994) Anderson (1994) Augustin and Frelich (1998) Augustin and Frelich (1998)

MN

Leaf area

Significantly higher in exclosure than in unfenced areas () Correlation with % of perennial herbaceous spp. browsed Significantly higher in exclosure than unfenced areas Significantly higher in a high deer density area Significantly higher grazing of flwring individuals than overall population Significantly higher in a high deer density area

VA VA

Stem density Reproductive activity

Significantly higher stem densities inside fences Correlated with high and low deer densities

Fletcher et al. (2001) Fletcher et al. (2001)

Augustin and Frelich (1998)

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

Uvalaria and Polygonatum

and Parker (2000) and Parker (2000) and Parker (2000), et al. (2001)

57

58

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

in the KQDC area will resemble those reported by Anderson (1994) and Rooney (1997); and (2) variation in the current demographic and morphological characteristics of Maianthemum and Trillium spp. will be correlated with variation in deer impact within the KQDC area. Based on previous studies of indicator species (Table 1) we predict that areas with high deer browsing will have plants with reduced stem height or leaf lengths, lower densities of indicator plants and lower flower frequencies.

2. Methods 2.1. Study sites

Fig. 1. The Kinzua Quality Deer Cooperative in Pennsylvania.

In the Trillium and Maianthemum studies mentioned above, Trillium spp. and Maianthemum were used to distinguish between deer densities in broad classes (i.e., fenced versus unfenced or high versus low deer densities). In practice, however, phytoindicators of deer impact are only effective as management tools if they correlate with deer impact (measured directly as browse or indirectly as deer density) in space and time. For the KQDC monitoring program, we seek plants that will indicate both temporal and spatial variation in deer impact across a broad landscape. In this paper we evaluate Trillium spp. and Maianthemum as potential indicators of deer impact at a landscape scale. To verify that deer have impacted these populations of plants we compared the morphological and demographic characteristics of Maianthemum and Trillium spp. in the Latham site, an area protected from deer for over 60 years, to that of the KQDC and the values from the studies above. We tested two hypotheses: (1) deer impacts to morphology and demographics of Maianthemum and Trillium spp.

The KQDC is located in northwestern Pennsylvania within the unglaciated Allegheny High Plateau section of the Appalachian Plateau Province (Harrison, 1970; McNab and Avers, 1994). The landscape is dominated by contiguous forest and is near the communities of Bradford, Marshburg, and Westline, PA (418450 –428000 N and 788370 –788550 W). The Pennsylvania Game Commission maintains an exclosure erected in the late 1940s in northwestern Pennsylvania. This exclosure is named the Latham site, after the ecologist Roger Latham, who originally installed it. The Latham site, is located in Game Lands #30 in southeastern McKean Co, Pennsylvania (418380 N, 788190 W). The fenced and unfenced plots at the Latham site are located within the same ecoregion, climatic zone and soil types, about 35 km southwest of the KQDC in contiguous mature northern hardwood forest. 2.2. Plot layout and survey Using ArcView GIS version 3.3 (ESRI 2002), we superimposed a grid across the KQDC and selected twenty-six 2.6 km2 squares at random for data collection. For vegetation sampling, we laid out six outer plots in 2001, and added a seventh central plot in 2003, in each grid square (Fig. 2). For the outer plots, the first plot was placed 400 m out from the center of the grid square along a randomly chosen azimuth; the other outer plots were placed 400m from the center of the grid square along azimuths incremented by 608 from the previous plot. Plots contained four 2 m radius

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

59

quadrats placed 25 m from the center point of the plot in each cardinal direction (Fig. 2). The field crew located the plots using UTM coordinates provided by ArcView GIS and GPS. A total of 113 and 147 plots were measured in 2001 and 2003, respectively. Pellet groups were counted in 1.22 m radius plots located at 30.5 m intervals on five parallel 1.6 km transects spaced 914 m apart within each sampled grid square. Data were collected between 1 April and 1 May 2002 and 2003 and deer densities were calculated by D. deCalesta following deCalesta (unpublished report, 20022). 2.3. KQDC phyto-indicator measurements In 2001, Maianthemum leaf length to the nearest millimeter and evidence of flowering (i.e., inflorescence, seeds, or rachis present) were recorded for each individual present in the 2 m radius quadrat. In 2001, measurements of Trillium populations were also conducted in 2 m radius quadrats. In 2003, measurements were taken at a larger scale to increase the number of Trillium stems encountered. For Trillium, stem height to the nearest centimeter, stem density and evidence of flowering of each individual of Trillium spp. were recorded for the entire plot (1451 m2, Fig. 2). The height and number of Trillium spp. individuals with one leaf was also recorded for each plot. Within each of the 2 m radius quadrats deer browsing on each woody species was assessed. The proportion of browsed growing tips (BGT) was assigned a BGT number based on the following classes: 0 = no browse, 1 = 1–33%, 2 = 34–66%, and 3 = 67–100%. Browseable growing tips were defined as apical growth of twigs between heights of 5 and 200 cm above the ground under the assumption that these twigs had been available to deer during the previous winter. Browse was only recorded on lignified tissue from the previous growing season. BGT numbers were averaged for each species in each plot and one browse index value was calculated for each plot by averaging only the browse indices of preferred woody plant species. Browse preference was 2 deCalesta, D.S., 2002. Kinzua Quality Deer Cooperative Demonstration: deer density estimates, roadside counts, and checking station operation. Unpublished report on file at Sand County Foundation, Madison, WI.

Fig. 2. Kinzua Quality Deer Cooperative plot layout for vegetation sampling. Figure is not drawn to scale.

determined by comparing the relative abundance (frequency and density) of all species across the landscape to the average browse index and browsing frequency of each species (Fig. 3). Preferred species were species in which browsing frequency and the average browse index exceeded relative frequency and relative density (Fig. 3). The preferred species were Fagus grandifolia Ehrh, Acer pensylvanicum L., Fraxinus americana L., Betula sp., and Amelanchier sp. Sampling of the KQDC began on 23 May and ended 14 August in 2001 and began on 21 May and ended 20 August in 2003. Species nomenclature followed Gleason and Cronquist (1991). 2.4. Latham site phyto-indicator measurements Thirty-one 1 m2 quadrats were placed systematically based on a random starting point on a temporary grid inside the 0.41 ha Latham site. Another grid was laid out in a 0.41 ha area in a nearby forest of the same forest type, topography and disturbance history, but open to deer for the last 60 years—the unfenced Latham site. At each quadrat we recorded Trillium spp. heights,

60

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

Fig. 3. Average browse index (S.E.M.), relative frequency, relative density and browse frequency of the nine most abundant woody plant species in the KQDC. Preferred browse species (asterisk) are those in which browsing frequency and the average browse index exceed relative frequency and relative density.

the number of ramets of Maianthemum in a 0.25 m2 nested quadrat and the occurrence of flowering or fruiting plants of both species. The plots were sampled on the 13–14 May 2002. 2.5. Statistical analysis Means of the indicator variables were calculated for each sampling unit at each site (i.e., plots for the KQDC and quadrats for the Latham sites). Trillium is found in several morphologically distinct life stages (Kawano et al., 1986), flowering, non-flowering, and ‘‘juvenile’’ (plants with a single-leaf, see Section 4). Thus, for Trillium, three different height variables were calculated from the total population of Trillium at each plot: (1) height of flowering and non-flowering plants; (2) mean height of single-leaf Trillium (‘‘juvenile’’ plants, see Section 4); and (3) height of non-flowering and non-single-leaf Trillium. One-way analysis of variance (ANOVA) was used to test for differences in mean Trillium height (nonflowering and non-single-leaf Trillium) and stem density and mean Maianthemum ramet density between the KQDC, the Latham fenced site, and the Latham unfenced site. These phyto-indicator variables and flowering frequencies for both species in each site were graphed along with the published

phyto-indicator values for Trillium in Anderson (1994) and for Maianthemum in Rooney (1997). Post-hoc comparisons between the KQDC, Latham fenced and Latham unfenced sites were made using Tukey’s HSD multiple comparison tests. These analyses were performed using SYSTAT1 version 10.2 (Systat Software Inc., 2002). Spearman-rank correlation analysis, a non-parametric test, was used to test for correlations between the phyto-indicator variables and the preferred deer browse index using SPSS version 11.5 (LEAD Technologies Inc., 2002). The critical significance levels for these tests were adjusted from a = 0.05 using Bonferroni adjustment methods (Samuels and Witmer, 1999). For Maianthemum two comparisons were made; 2001 deer browse versus mean leaf length and mean ramet density and the critical value for the correlation analyses were adjusted to a = 0.025. For Trillium, 2003 deer browse was correlated against mean height of flowering and non-flowering plants, mean height of single-leaf Trillium, mean height of non-flowering and non-single-leaf Trillium, mean Trillium density, proportion of flowering plants, and proportion of singleleaf plants, for a total of six comparisons. For these comparisons the critical value was adjusted to a = 0.008. Since several plots did not contain Trillium and/or Maianthemum the analyses were also conducted with these cells as missing values.

3. Results 3.1. Variation in deer densities Pellet-group censuses collected in the KQDC area prior to the beginning of each sampling season suggested that deer densities ranged from 3 to 23 deer/ km2 with a mean of 10.9 (1.1) deer/km2 in 2001 and 3 to 22 deer/km2 with a mean of 11.0 (0.9) deer/km2 in 2003. Deer densities did not differ significantly between years (t = 0.087, p = 0.931). 3.2. Characteristics of Trillium spp. and Maianthemum Trillium spp. heights (F 2,148 = 5.7, p < 0.01) and stem densities (F 2,208 = 146.0, p < 0.001) and Maianthemum ramet densities (F 2,189 = 75.7,

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

p < 0.001) were significantly different between the KQDC, the Latham fenced site, and the Latham unfenced site. Trillium spp. heights and stem densities and Maianthemum ramet density were significantly greater in the Latham fenced site but not significantly different between the Latham unfenced site and the KQDC (Figs. 4 and 5). Average densities of Trillium and Maianthemum are at least 100 times lower in the KQDC and non-fenced Latham site versus the fenced Latham site. The number of Trillium spp. stems measured was 5169 (mean density = 0.03  0.004 stems/m2) at the KQDC, 416 (mean density = 14.5  2.0 stems/m2) at the fenced Latham site, and 4 (mean density = 0.13  0.075 stems/m2) at the unfenced Latham. Flowering frequency in the KQDC was 9.1% versus 61% in the fenced Latham site. No Trillium spp. individual was flowering in the quadrats of the unfenced Latham site. Of the 5169 Trillium in KQDC, 66% (3410) had only one leaf. One-leafed plants had an average height of 7.8  0.05 cm. Flowering individuals had an average height of 26.2  0.5 cm.

61

Fig. 4. Comparisons of mean (S.E.M.) Trillium height and density for Illinois (Ryerson exclosure and averaged values for unfenced plots in Anderson, 1994), the KQDC and the Latham sites. Average heights and average densities with different uppercase and lower case letters, respectively, are significantly different ( p < 0.001). N/ A: data not available.

Fig. 5. Comparisons of mean (S.E.M.) Maianthemum leaf length and stem densities for high and low boulders in PA (data from Rooney, 1997), the KQDC and Latham sites. Average ramet densities with different uppercase letters are significantly different ( p < 0.001). Data were not collected on leaf length at the Latham sites. N/A: data not available.

62

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

In the KQDC, 8510 Maianthemum leaves were measured and counted. At the fenced and unfenced Latham site, 2259 and 20 Maianthemum were recorded, respectively. Mean leaf length in the KQDC was 39  0.1 mm. The proportion of flowering plants was 0.3% in the KQDC and 4.7 and 0% in the fenced and unfenced Latham sites, respectively. Mean ramet densities were 0.01  0.014, 27.9  5.08, and 0.258  0.202 ramets per 0.1 m2 in the KQDC, fenced and unfenced Latham sites, respectively. 3.3. Phyto-indicators and deer impact No significant correlations were found between the average preferred browse index and Trillium mean height of flowering and non-flowering plants (r = 0.146, p = 0.08, N = 147), mean height of single-leaf plants (r = 0.112, p = 0.18, N = 147), mean height of non-flowering and non-single-leaf (r = 0.062, p = 0.46, N = 147), mean Trillium density (r = 0.126, p=0.13, N = 147), proportion of flowering plants (r = 0.021, p = 0.80, N = 147), proportion of single-leaf plants (r = 0.168, p = 0.04, N = 147), mean Maianthemum leaf length (r = 0.095, p = 0.32, N = 113) or mean Maianthemum ramet density (r = 0.140, p = 0.14, N = 113). The lack of correlation is exemplified by scatter plots of the indicator variables versus the browse index (Figs. 6 and 7). When the points were labeled by plot, plots of the same cluster were not grouped together, indicating no effect of clustering of plots into squares. These results were not influenced by the plots without indicator plants (i.e., zeros in the analysis) as no significant correlations were found when only plots with Trillium and/or Maianthemum were analyzed.

4. Discussion 4.1. Trillium spp. and Maianthemum as phyto-indicator species The mean morphological characteristics of both Trillium spp. and Maianthemum in the KQDC are quite similar to or below the values reported by Anderson (1994) and Rooney (1997) for populations heavily impacted by white-tailed deer, with the exception of Trillium stem density (Figs. 4 and 5).

This is consistent with the fact that deer densities in the KQDC region exceeded the goals established by the Pennsylvania Game Commission for most of the 20th century (Redding, 1995). The Trillium spp. and Maianthemum in the Latham fenced site were similar to those reported for populations experiencing low deer impact. In addition to having mean phyto-indicator characteristics associated with high deer impacts, many of the individual plots on which plants were sampled also exhibited characteristics associated with high deer impacts. For example, the mean height of Trillium spp. in 27% of the KQDC plots was below the 12–14 cm recommended for healthy Trillium spp. populations by Anderson (1994). Interestingly, the KQDC plots were dominated by single-leaf Trillium individuals (66%). Anderson (1994) does not present data on single-leaf individuals for comparison. However, in a population of Trillium grandiflorum a stable stage population model predicted that
15% of the population should display the single-leaf or ‘‘juvenile’’ morphology (Rooney and Gross, 2003). It has been suggested that single-leaf Trillium are young plants (Kawano et al., 1986), however inspection of the rhizome constrictions of several plants indicated that many of the single-leaf plants are at least 10 years old (Kirschbuam, personal observation). Knight (2003) found that reproductive plants were more likely to regress to non-flowering individuals the following season when they were naturally and experimentally browsed. Given the high densities of single-leafed Trillium in the KQDC and the ability of Trillium to regress to non-flowering life stages due to herbivory, it seems likely that the high proportion of single-leaf Trillium plants observed in this study are the results of long-term exposure to high deer densities. Only 9% of the Trillium spp. found on KQDC showed evidence of flowering, also well below the 21– 34% flowering recommended by Anderson (1994). Low flowering frequency of Trillium was also found in Wisconsin, Michigan, Minnesota and Pennsylvania in areas of high deer impact (Augustine and Frelich, 1998; Knight, 2003a,b; Rooney and Gross, 2003). Based on similar comparisons, deer impact to Maianthemum was more evident than that of Trillium spp. The mean leaf length was 39 mm, close to the 42 mm reported by Rooney (1997) for Maianthemum

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

63

Fig. 6. Scatter plots of Trillium indicator variables (stem heights in cm and density is number of stems per plot) vs. a browse index calculated using preferred browse species. Correlation analyses revealed no significant relationships among the variables. NF: non-flowering, NSL: nonsingle-leaf.

64

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

Fig. 7. Scatter plots of average Maianthemum leaf length (mm) and ramet densities (number of ramets per plot) vs. a browse index calculated using preferred browse species. Correlation analyses revealed no significant relationships among the variables.

individuals exposed to deer browsing on low boulders, and below the 55 cm leaf length reported for Maianthemum protected from deer browsing on high boulders. Of the plots sampled, 100% had mean Maianthemum densities <32 ramets per 0.1 m2 (0.1 m2 = high boulder plot size), and 97% of the plots had mean leaf lengths <55 mm. Flowering percent for Maianthemum in the KQDC (0.3%) was also low compared to percentage flowering on high boulder tops (20%). Maianthemum flowering percentage in the fenced Latham site (5%) was lower than that reported for protected sites in the literature. Based on the above comparison our first hypothesis was confirmed. The morphological characteristics of the tested indicators, Trillium spp. and Maianthemum are similar to those reported in the literature for areas with high deer impact, a condition that this area has experienced for 60 years. 4.2. Indicator plants, deer impact and browse indices The correlations between the indicator characteristics and deer browse are equivocal with regard to the potential use of these species as indicators of deer impact on a landscape scale with a diverse range of deer densities. Despite the evidence provided by this study that the morphological characteristics of these species are good indicators of the high deer impact in an area that has sustained over the past several decades, we found no significant correlations between

direct measures of deer impact (browse) and the indicator species characteristics. Likely explanations for the lack of these correlations include: environmental factors not directly measured in our study which affects the growth morphology and demographics of indicator plants, potential legacy effects of high deer abundance, and difficulty in accurately quantifying deer impact. In the central Appalachians, Trillium frequency is greater in soils of high pH and net nitrification (Christ et al., 2002) and in Oregon Trillium populations can experience >95% mortality in intensely managed areas (Jules, 1998). Maianthemum appears not to be affected by soil chemistry (Silva et al., 1982; Demchik and Sharpe, 2001). Maianthemum ramet density tends to be higher; however, where there is higher light and soil moisture (Silva et al., 1982). Despite restricting our sample areas to those which had dry, acidic soils and had not been harvested in the last several decades, variation in edaphic factors and management history could still cloud relationships between deer impact and phyto-indicator characteristics. Our study area has experienced high deer impact levels for many decades. Balgooyen and Waller (1995) found that morphological characteristics were often correlated in different directions with current versus historical deer densities; the same might prove true of our study area over time. Our data suggest that historically high deer populations in our study site have led to low densities of Trillium plants. Research on pollen limitation due to low flower densities of T.

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

grandiflorum by Knight (2003) suggests density dependent effects of pollination success. Thus, in the KQDC if historically high deer densities have led to declines in Trillium densities, recovery of these populations may take many years. Indeed, even in the absence of deer browsing, growth rates of Trillium populations in Wisconsin and Michigan could continue to decline—1.61% per year (Rooney and Gross, 2003). Little is understood about mechanisms of recovery from deer overabundance for either of these species, although Anderson (1994) showed a positive trend in mean Trillium height over a three-year period inside a deer exclosure. Long-term monitoring of the KQDC and Latham sites will produce information regarding changes of both Trillium spp. and Maianthemum characteristics with deer impact over time. This information will go some distance in describing the mechanisms of recovery of these and other forest herbs from successive years of over-browsing by deer, and in determining whether these species recover quickly enough to serve as good phyto-indicators. 4.3. Implications for management This study implies that the ability of managers to use indicator species as reliable estimates of deer impact at a particular time in space may be clouded by factors such as available soil resources, site management history or legacy impacts. Future work should focus on environmental factors that influence morphological and demographic characteristics of phytoindicator populations that could be considered covariables with deer impact in this system. The lack of correlations in space however does not rule out the potential for indicators to gauge the recovery of a plant community after deer populations have been restored to levels that are compatible with healthy forest understories. Research in the KQDC will continue to explore the utility of indicator species in the context of a long-term adaptive management project in which increased hunting pressure will lead to lower deer populations. By monitoring indicator species and other ecosystems parameters we will be able to learn how phyto-indicators will change in response to lower deer impact across a landscape and how to employ them in an effective ecosystem monitoring program. Until further tests can validate the utility of phyto-

65

indicators across landscapes with variable deer impact in space and time, we recommend managers continue to monitor deer impacts using browsing surveys or methods that directly measure deer populations.

Acknowledgements The authors wish to thank Kevin McAleese of the Sand County Foundation, who was instrumental in the initiation, operation, and completion of the project. We would also like to acknowledge our field assistants Tim Bischoff, Sarah Farley, Rebecca Schultze, and Veronique VanGheem. We are grateful for the manuscript reviews and statistical advice provided by Catherine Bach, Stephen Horsley, Kari Jensen-Kirschbaum, Todd Ristau, Ajejandro Royo and J. Angel Soto-Centeno. Invaluable support and resources were made available by the Northeastern Research Station, Forestry Sciences Lab and by Open Sesame Locksmithing. Funding was provided by the Bradley Fund for the Environment of the Sand County Foundation, the Vira I. Heinz Endowment, and the National Fish and Wildlife Foundation. The KQDC is owned and managed by the USDA Forest Service Allegheny National Forest (20,283 ha), Bradford City Water Authority (4858 ha), Commonwealth Forest Investments (4069 ha), Kane Hardwood, a Collins Pine Company (1215 ha), and RAM Forest Products (202 ha).

References Alverson, W.S., Waller, D.M., Solheim, S.L., 1988. Forests too deer: edge effects in Northern Wisconsin. Conserv. Biol. 2, 348– 358. Anderson, R.C., 1994. Height of white-flowered trillium (Trillium grandiflorum) as an index of deer browsing intensity. Ecol. Appl. 4, 104–109. Augustine, D.J., Frelich, L.E., 1998. Effects of white-tailed deer on populations of an understory forb in fragmented deciduous forests. Conserv. Biol. 12, 995–1004. Augustine, D.J., Jordan, P.A., 1998. Predictors of white-tailed deer grazing intensity in fragmented deciduous forests. J. Wildl. Manage. 62, 1076–1085. Balgooyen, C.P., Waller, D.M., 1995. The use of Clintonia borealis and other indicators to gauge impacts of white-tailed deer on plant communities in northern Wisconsin, USA. Nat. Areas J. 15, 308–318. Christ, M.J., Peterjohn, W.T., Cumming, J.R., Adams, M.B., 2002. Nitrification potentials and landscape, soil and vegetation char-

66

C.D. Kirschbaum, B.L. Anacker / Forest Ecology and Management 217 (2005) 54–66

acteristics in two Central Appalachian watersheds differing in NO3 export. Forest Ecol. Manage. 159, 145–158. Demchik, M.C., Sharpe, W.E., 2001. Forest floor plant response to lime and fertilizer before and after partial cutting of a northern red oak stand on an extremely acidic soil in Pennsylvania, USA. Forest Ecol. Manage. 144, 239–244. DeGraaf, R.M., Healy, W.H., Brooks, R.T., 1991. Effects of thinning and deer browsing on breeding birds in New England oak woodlands. Forest Ecol. Manage. 41, 149–191. Fletcher, J.D., McShea, W.J., Shipley, L.A., Shumway, D.L., 2001. Use of common forest forbs to measure browsing pressure by white-tailed deer (Odocoileus virginianus Zimmerman) in Virginia USA. Nat. Areas J. 21 (2), 172–176. Gleason, H.A., Cronquist, A., 1991. Manual of Vascular Plants of Northeastern United States and Adjacent Canada. The New York Botanical Garden, Bronx, New York, USA. Harrison, R.E., 1970. Physiographic Map of the United States in National Atlas of the United States of America. U.S. Geological Survey, Washington, DC, USA. Hilty, J., Merenlender, A., 2000. Faunal indicator taxa selection for monitoring ecosystem health. Biol. Conserv. 92, 185–197. Horsley, S.B., Stout, S.L., deCalesta, D.S., 2003. White-tailed deer impact on the vegetation dynamics of a northern hardwood forest. Ecol. Appl. 13, 98–118. Hough, A.F., 1965. A twenty-year record of understory vegetation change in a virgin Pennsylvania forest. Ecology 46, 370–373. Jules, E.S., 1998. Habitat fragmentation and demographic change for a common plant: Trillium in old-growth forest. Ecology 79 (5), 1645–1656. Kawano, S., Ohara, M., Utech, F.H., 1986. Life history studies on the genus Trillium (Lilliaceae). 2. Reproductive biology and survivorship of four eastern North American species. Plant Species Biol. 1, 47–58. Knight, T.M., 2003a. Effects of herbivory and its timing across population of Trillium gradniflorum (Liliaceae). Am. J. Bot. 90 (8), 1207–1214. Knight, T.M., 2003b. Floral density, pollen limitation, and reproductive success in Trillium grandiflorum. Oecologia 137, 557– 663. Leopold, A., Sowls, L.K., Spencer, D.L., 1947. A survey of overpopulated deer ranges in the United States. J. Wildl. Manage. 11, 162–177. McNab, W.H., Avers, P.E. (compilers), 1994. Ecological subregions of the United States: section descriptions. USDA Forest Service Administrative Publication WO-WSA-5. USDA Forest Service, Washington, DC, USA. McShea, W.J., Underwood, H.B., Rappole, J.H., 1997. The Science of Overabundance, Deer Ecology and Population Management. Smithsonian Institution Press, Washington, DC, USA. Noss, R.F., Cooperider, A.Y., 1994. Saving Nature’s Legacy: Protecting and Restoring Biodiversity. Island Press, Washington, DC.

Redding, J., 1995. History of deer population trends and forest cutting on the Allegheny National Forest. 214-224 In: Gottschalk, K.W., Fosbroke, S.C.L. (Eds.), Proceedings, Tenth Central Hardwood Forest Conference. U.S.D.A. Forest Service General Technical Report NE-197. Northeastern Forest Experiment Station, Radnor, PA, USA. Rhoads, A.F., Klein Jr., W.M., 1993. The Vascular Flora of Pennsylvania: Annotated Checklist and Atlas. American Philosophical Society, Philadelphia, PA. Ruhren, S., Handel, S.N., 2000. Considering herbivory, reproduction, and gender when monitoring plants: a case study of Jack-inthe-pulpit (Arisaema triphyllum [L.] Schott) Nat. Areas J. 20, 261–266. Rooney, T.P., Gross, K., 2003. A demographic study of deer browsing impacts on Trillium grandiflorum. Plant Ecol. 168, 267–277. Rooney, T.P., 1997. Escaping herbivory: refuge effects on the morphology and shoot demography of the clonal forest herb Maianthemum canadense. J. Torrey Bot. Soc. 124, 280–285. Rooney, T.P., Dress, W.J., 1997. Species loss over sixty-six years in the ground layer vegetation of Heart’s Content, an old growth forest in Pennsylvania. PA Nat. Areas J. 17 (4), 297–305. Rooney, T.P., McCormick, R.J., Solheim, S.L., Waller, D.M., 2000. Regional variation in recruitment of hemlock seedlings and saplings in the upper Great Lakes, USA. Ecol. Appl. 10 (4), 1119–1132. Samuels, M.L., Witmer, J.A., 1999. Statistics for the Life Sciences. Pretince Hall, Upper Saddle River, NJ. Silva, J.F., Kana, T.M., Solbrig, O.T., 1982. Shoot demography in New England population of Maianthemum canadense Desf. Oecologia 52 (2), 181–186. Skinner, W.R., Telfer, E.S., 1974. Spring, summer, and fall foods of deer in New Brunswick. J. Wildl. Manage. 38, 210–214. Tilghman, N.G., 1989. Impacts of white-tailed deer on forest regeneration in northwestern Pennsylvania. J. Wildl. Manage. 53, 524–532. Vellend, M., 2002. A pest and an invader: white-tailed deer (Odocoileus virginianus Zimm.) as a seed dispersal agent for honeysuckle shrubs (Lonicera L.) Nat. Areas J. 22 (3), 230–234. Webster, C.R., Parker, G.R., 2000. Evaluation of Osmorhiza claytonii (Michx.) CB Clarke, Arisaema triphyllum (L.) Schott, and Actaea pachypoda Ell. as potential indicators of white-tailed deer overabundance. Nat. Areas J. 20 (2), 176–188. Webster, C.R., Jenkins, M.A., Parker, G.R., 2001. A field test of herbaceous plant indicators of deer browsing intensity in mesic hardwood forests of Indiana, USA. Nat. Areas J. 21 (2), 149– 158. Williams, C.E., Mosbacher, E.V., Moriarity, W.J., 2000. Use of turtlehead (Chelone glabra L.) and other herbaceous plants to assess intensity of white-tailed deer browsing on Allegheny Plateau riparian forests, USA. Biol. Conserv. 92 (2), 207–215.