The Effect of Deer and Method of Removal of Invasive Lonicera maackii on the Restoration of Understory Plants

Kendra Cipollini1,2, Elizabeth Ames,1,3 and Don Cipollini4

RUNNING TITLE: Effect of Deer and Invasive Honeysuckle Removal Method

1WilmingtonCollege, Wilmington, OH45177, U.S.A.

2 Address correspondence to K. Cipollini, email

3Present Address: 1967 Beatty Rd, Wilmington, OH45177, U.S.A.

4 Wright State University, Department of Biological Sciences, Dayton, OH45435, U.S.A.

Phone: 937.382.6661 ext. 367

FAX: 937.383.8530

Abstract

Invasive Lonicera maackiiis often removed from natural areas, yet deer may thwart restoration efforts. We investigated the success of understory plants after removal of L. maackii using two removal methods in fenced and unfenced plots. The cut/paint removal method consisted of cutting the stems and painting the stumps with the herbicide triclopyr. The basal application removal method consisted of applying triclopyr to the base of each stem, leaving the dead stems in place. Light level in the cut/paint treatment was higher than in the basal application treatment, which was higher than in the control (no removal) treatment. In 2005, we transplanted Asarum canadense and Impatiens capensis into the cut/paint and basal application treatments. Fruit production and height of transplanted I. capensis and subsequent recruitment were greater in the cut/paint treatment. Impatiens capensis were taller in the fenced treatment. There was a trend for more leaves of A. canadense in the basal application treatment. In 2006 and 2007, we measured plant communities in the control, cut/paint and basal application treatments. Native species richness was generally higher in the cut/paint and basal application treatments than in the control treatment. Standing stems of L. maackii offered some protection from deer damage. When fenced, there were more I. capensis recruits, more I. capensis fruits and greater species richness in the cut/paint treatment than in the basal application treatment. When unfenced, plants responded similarly to both removal methods. There were more invasiveAlliaria petiolata and more L. maackii seedlings in the cut/paint treatment than in the control treatment, with intermediate amounts found in the basal application treatment.

Key words: Alliaria petiolata, Asarum canadense, herbivory, honeysuckle, Impatiens capensis, Odocoileus virginiana.

Introduction

Invasive plant species can have major impact on plant populations, communities, and ecosystems(Pimentel et al. 2005). The mechanism of how invasive plants affect native systems varies with invader and invaded habitat, yet is most likely a combination of more than one factor (Inderjit 2005). Once an invasive plant has become established in an ecosystem, land managers frequently seek to reduce populations of the invasive species through a variety of methods, from fire to mowing to herbicide (e.g., Carlson & Gorchov 2004; Hartman & McCarthy 2004; Krueger-Mangold et al. 2006). It is important that alternative methods of control are assessed experimentally in order to determine the best course of action for successful and efficient restoration.

Lonicera maackii (Rupr.) Maxim. (Amur honeysuckle), is an Asian shrub that is invasive in Midwestern and northeasternU. S.forests (Luken & Thieret 1995; USDA/NRCS 2007). Lonicera maackii has extended leaf phenology (Trisel 1997) and bird and mammal dispersed fruit (Vellend 2002; Bartuszevige & Gorchov 2006), both of which contribute to its invasive success. Lonicera maackii reduces native plant species richness (Collier et al. 2002), and the performance of a wide range of both understory and overstory plants (Gould & Gorchov 2000; Gorchov & Trisel 2003; Miller & Gorchov 2004; Hartman & McCarthy 2007). Effects of L. maackii on other plants can be mediated directly through such factors as shading and allelopathy (e.g., Gorchov & Trisel 2003; Cipollini et al. 2008a) but it may also have indirect effects (Meiners 2007). The impact of L. maackii is not limited to plant communities; its presence also affects birds and herptiles (Schmidt & Whelan 1999; McEvoy & Durtsche 2004).

Natural resource agencies are focusing efforts on controlling L. maackii in order to restore forest communities. Due to these efforts, much information has been gained on how to effectively kill L. maackii in natural environments (Nyboer 1992; Conover & Geiger 1993; Batcher Stiles 2000; Hartman & McCarthy 2004). Techniques to kill L. maackii include the cut-and-paint, or cut/paint, method, where L. maackii is cut at the base and removed; herbicide is subsequently applied to the cut stump (Batcher & Stiles 2000; McDonnell et al. 2005). Two other herbicide-based methods, the foliar spray method and the injection method, result in the killing of L. maackii with stems left standing in place. The foliar spray method consists of L. maackii foliage being sprayed with a herbicide, usually in the fall (Conover & Geiger 1993). The injection method uses an injection device, EZJect (ArborSystems, Omaha, NE), in which herbicide is injected directly into the L. maackii stem (Hartman &McCarthy 2004). In restoration practice, some managers have found that the foliar spray application is superior to the injection system. For example, land managers at Hamilton County Park District in Cincinnati, Ohio, and other metropolitan parks in Ohio primarily use the foliar spray method currently due to its effectiveness and ease of application (T. Borgman 2007, Hamilton County Park District, Cincinnati, OH, personal communication).

Standing dead stems of L. maackii could affect restoration success in a number of ways. Standing stems may continue to shade the understory, which may delay the development of healthy plant communities for as long as stems remain standing. There is also some evidence that L. maackii produces allelopathic compounds in its tissues (Trisel 1997; Dorning Cipollini 2006; Cipollini et al. 2008a; Cipollini et al. 2008b). Thus, standing dead stems could continue to contribute allelochemicals to the understory until they decompose completely. In contrast, the presence of standing dead stems may have some positive effects. A major challenge to restoration of Midwestern and northeastern forests is browsing or other forms of damage by white-tailed deer, Odocoileus virginiana Zimmermann, which is considered a keystone herbivore due to its direct and indirect effects on both plants and animals (Rooney 2001; Rooney & Waller 2003; Cote et al. 2004). Sweeney & Czapka (2004) suggest that protecting seedlings from herbivory should be given a greater priority than protection from plant competition during restoration. There is some evidence that standing dead stems of L. maackii may provide protection of transplanted plants from deer browsing (Gorchov & Trisel 2003), but this effect has not been quantified adequately. Hartman & McCarthy (2004) followed survival of transplanted tree seedlings after different methods of L. maackii removal, but deer impact was insufficient to examine for protective effects. The response of the native plant community to the presence of standing dead stems has not been examined.

Another challenge to forest restoration is the response of the invasive plant community to restoration efforts. Although poorly replicated, McConnell et al. (2005) found an increase in species richness after removal of L. maackii using the cut/paint method, but L. maackii seedlings and other invasive species were also increased in the cut/paint treatment. Runkle et al. (2007) also found an increase in species richness eightyears after removal of L. maackii; however, this increase was primarily the result in the increase of species with high dispersal ability, such as vines and other weedy species. Luken et al. (1997) found the vine Vitus vulpina responded positively to L. maackii removal. No comparisons of the effects of removal method on invasive plant responses have been published, but the presence of standing dead stems may slow the response of invasive plants. On the other hand, deer could facilitate invasive species responses to restoration efforts if they facilitate dispersal into restored areas or selectively damage native plants over invasive species.

The objectives of this study were to determine the effect of the presence or absence of standing dead L. maackii, the presence or absence of deer, and their interaction on understory plants. We investigated this by following the success in basal application and cut/paint treatments of transplanted Impatiens capensis Meerb. (jewelweed), and transplanted Asarum canadense L. (wild ginger). We also followed the natural recruitment of species in basal application, cut/paint and control treatments. We predicted that in the absence of deer, the cut/paint removal method would best benefit plant performance, due to the increase in light and the removal of potentially allelopathic stem and leaf material. We predicted that in the presence of deer, the basal application method, which, like the frequently-used foliar spray method, leaves the dead stems standing, would best benefit plant performance due to the protection conferred by the stems against deer damage.

Methods

We established experimental plots during the first week of April in 2005 in Hamilton County Park District’s Sharon Woods, located in Sharonville, Ohio (39°16'40” N, 84°23'56”W). We chose two sites impacted by L. maackii approximately 0.75 km apart from each other. Each site is located in mixed oak-maple-ash forest community with a relatively large amount of L. maackii, yet still contained some degree of native understory vegetation. Deer density at Sharon Woods was approximately eight deer/km2, which is the management goal at this site (John Klein 2007, Hamilton County Park District, Cincinnati, OH, personal communication). Within each site, we selected three sub-sites, with paired 8 x 6 m experimental plots at each sub-site, for a total of 12 pairs of plots over six sub-sites. Within each sub-site, we used a split-split plot experimental design, with the split plot factor of deer exclusion and the split-split plot factor of L. maackii removal method. We fenced one plot of each pair at each sub-site, placing metal fence posts into the ground and attaching six-foot-tall plastic deer fencing with five-cm mesh to the posts with plastic ties. Within each paired plot (fenced and unfenced), we created eight 2x3 m subplots. For three of the adjacent subplots, we removed L. maackii using the cut/paint method, i.e. cutting off the L. maackii with a handsaw approximately 10 cm from the soil surface, removing the L. maackii, and painting the stump with the herbicide triclopyr (Pathfinder II, 13.6% triclopyr, Dow Agrosciences, Indianapolis, IN) to prevent resprouting. For three of the adjacent subplots, we removed L. maackii using the basal application method, i.e. applying triclopyr in a band around the entire circumference of the base of each stem, which essentially girdles each stem. In practice, the basal application method has proven to be too labor intensive to use, particularly on smaller individuals (T. Borgman 2007, Hamilton County Park District, Cincinnati, OH, personal communication). We used the basal application method in our experiment rather than the more frequently-used foliar spray method due to the fact that our treatment application had to occur in the spring and the foliar spray method is more effective in the fall. The goal in our application method was to mimic the end result of the commonly-used foliar spray method, i. e., the presence of dead stems. On subsequent visits, we controlled the minimal amount of living or sprouting L. maackii with the same method as the rest of the subplot. We did not remove any L. maackii in the remaining two adjacent subplots and these served as our controls. For each of the three removal method subplots (i.e., basal application and cut/paint), one subplot was designated as the “seedling subplot,” one was designated as the “seed subplot” and one was designated as the “natural recruitment subplot” for further treatment and measurements.

On May 5, 2005, we locally transplanted four two-leaved rhizomes of the perennial A. canadense and six seedlings of the annual I. capensisinto our “seedling subplots.” Asarumcanadenseis a shade-tolerant rhizomatous perennial that typically expands into a large clump through asexual reproduction. Asarum canadense has been shown to be impacted by L. maackii (Dorning 2004), and appears to be largely resistant to deer browsing (D. Cipollini, personal observation). Plants were collected by first excavating rhizomes, which were often several internodes in length with a pair of leaves at each node. Single transplants, each containing one leaf pair, were then created by clipping the rhizome several cm on each side of a node. Impatiens capensisis an annual that is found in a range of light conditions, from forest clearings to closed woodlands (von Wettberg & Schmidt 2005) and reproduces exclusively by seed. Impatiens capensis is sensitive to deer browsing (Williams et al. 2000; Asnani et al. 2006). Seedlings were transplanted from local population when they were approximately 8-12 cm tall. Transplants were not placed in the control treatment, since we expected survival to be low from previous experience (e.g., Cipollini et al. 2008c). We added seeds to the “seed subplots” though we will not discuss the results here. For the “natural recruitment subplots,” we did not add any plants. In 2005, we measured survival and fruit number of transplanted I. capensis on 2 June and 30 June, and both height and fruit number on 21 July. We measured survival and leaf number of each A. canadense clump on 25 April 2006 and 4 May 2007. On 4 May 2007, we measured the width of the largest leaf in each A. canadense clump.

In a 1-m2 area of each of the seedling subplots (where I. capensisand A. canadensehad been transplanted in 2005), we counted the number of seedlings of I. capensis on 26 April 2006 and on 24 April 2007. We were confident that most, if not all, of the seedlings of I. capensis were from fruits of plants that we had transplanted the previous year; there were almost no I. capensis in the area except for the ones we transplanted. On 26 April 2006 and 24 April 2007, in the control and natural recruitment subplots, we counted the number of plants of each species in a 1-m2 area of the subplot. For the natural recruitment data, the richness, or number of species, was determined, removing Alliaria petiolata (garlic mustard) and other invasive species. In all the subplots, we measured light levels at 1 m height on July 6 2006, which was a bright and mostly cloudless day (Li-Cor Quantum Sensor, Lincoln, NB). All light measurements were made within 65 min of each other.

For all analyses, we used a nested split-split plot Analysis of Variance (ANOVA) or Multivariate Analysis of Variance (MANOVA) with site as a fixed factor, sub-site as the main plot factor (nested within site), fencing as the split plot factor and removal method as the split-split plot factor (SAS 1999). For MANOVAs, when significance was found using Wilk’s λ, separate univariate analyses of variance were performed on each separate date, followed by Tukey’s test to determine significance between treatment levels. We analyzed I. capensis fruit production in 2005 and number of seedlings in 2006-7, and number of A. canadense leaves using separate MANOVAs, with each date as a separate variable in the model (von Ende1993). We analyzed I. capensis height,I. capensissurvival, A. canadense leaf width and A. canadense survival separately using ANOVA. We analyzed native species richness and number of A. petiolata plants using separate MANOVAs, using each date as a separate variable in the model. We analyzed number of L. maackii seedlings in spring 2007 using ANOVA. Data were transformed as necessary to meet model assumptions. For light levels, data could not be transformed to meet model assumptions of normality, and a Kruskal-Wallis Rank Sum test was then used. The α-level used for all tests was 0.05.

Results

Light levels were significantly affected by treatment (F 2, 76 = 47.10, p < 0.001); light levels in the cut/paint treatment was significantly higher than the light levels in the basal application treatment which were in turn significantly higher than the light levels in the control treatment (Figure 1).

In the MANOVA for transplanted I. capensis fruit production across three dates, there was a significant effect of removal method (F3,42 = 4.19, p = 0.0111) and the interaction of fencing and removal method (F3, 42 = 3.34, p = 0.0280). On 30 June, there were significantly more fruits in plants found in the cut/paint treatment compared to plants in the basal application treatment (F1,44 = 11.77, p = 0.0013). On 30 June, there was a significant interaction between removal method and fencing (F1,44 = 9.58, p = 0.0034), with a greater fruit reduction from fenced to unfenced for plants in the cut/paint treatment compared to plants in the basal application treatment (Figure 2). In the ANOVA for transplanted I. capensis height on 21 July 2007, there was a significant effect of fencing (F1,3 = 10.25, p = 0.0493), with greater plant height in the fenced treatment, and a significant effect of removal method (F1,55 = 21.53, p < 0.0001), with greater plant height in the cut/paint treatment (Figure 3). There were no significant treatment effects on survival in the ANOVA.

In the MANOVA for I. capensis seedling number, there was a significant effect of removal method (F2,9 = 4.86, p = 0.0370) and the interaction of fencing and removal method (F2,9 = 9.09, p = 0.0069). In 2006 and 2007, there were more I. capensis seedlings in the cut/paint treatment (F1,10 = 4.97, p = 0.0499 and F1,10 = 10.58, p = 0.0087, respectively). In 2007, there was a significant interaction between fencing and removal (F1,10 = 7.91, p = 0.0184), with a greater reduction in seedlings in the cut/paint treatment between the fenced and unfenced treatments compared to the basal application treatment, which had a similar seedling number in fenced and unfenced treatments (Figure 4).

In the MANOVA for A. canadense leaf number, there were nosignificant treatment effects, yet the effect of removal method was approaching significance (F2,44 = 2.83, p = 0.070). We performed ANOVAs for each separate date to determine the nature of this non-significant trend. We found that, on 25 April 2006, there were more numbers of leaves in the basal application treatment (3.28 ± 0.27, mean ± SE) than in the cut/paint treatment (2.53 ± 0.15, mean ± SE; F1,45 = 5.68, p = 0.0214). In the ANOVAs for A. canadense leaf width and survival, there were no significant effects.

In the MANOVA, species richness was significantly affected by removal method (F4,38 = 3.70, p = 0.0122) and the interaction between fencing and removal method (F4,38 = 2.87, p = 0.0358). The effect of removal method on species richness was approaching significance in 2006 (F2,20 = 3.11, p = 0.0667) and was significant in 2007 (F2,20 = 7.62, p = 0.0035), with greater richness in the cut/paint and basal application treatment compared to the control treatment (Figure 5). The interaction of fencing and removal was significant in 2007 (F2,20 = 5.61, p = 0.0116). When fenced, species richness and number of plants declined from the cut/paint, basal application and control treatment. In contrast, when unfenced, there was similar to lower species richness and number of species in cut/paint treatment than in basal application treatment(Figure 5).