Upload
kurt-m-hartman
View
214
Download
0
Embed Size (px)
Citation preview
Restoration of a Forest Understory After theRemoval of an Invasive Shrub, Amur Honeysuckle(Lonicera maackii)Kurt M. Hartman and Brian C. McCarthy1
Abstract
The recruitment of native seedlings is often reduced inareas where the invasive Amur honeysuckle (Loniceramaackii) is abundant. To address this recruitmentproblem, we evaluated the effectiveness of L. maackiieradication methods and restoration efforts using seedlingsof six native tree species planted within eradication andunmanipulated (control) plots. Two eradication methodsusing glyphosate herbicide were evaluated: cut and paintand stem injection with an EZ-Ject lance. Lonicera maackiidensity and biomass as well as microenvironmentalcharacteristics were measured to study their effects onseedling growth and survivorship. Mean biomass of Amurhoneysuckle was 361 ± 69 kg/ha, and density was21,380± 3,171 plants/ha. Both eradication treatments wereeffective in killing L. maackii (‡ 94%). The injectiontreatment was most effective on large L. maackii individuals(>1.5 cm diameter), was 43% faster to apply than cuttingand painting and less fatiguing for the operator, decreasedoperator exposure to herbicide, and minimized impact tonontarget vegetation. Deer browse tree protectors wereused on half of the seedlings, but did not affect survivorship
or growth. After 3 years, survival of native seedlings wassignificantly less where L. maackii was left intact (32± 3%)compared with the eradication plots (p< 0.002). Seedlingsurvival was significantly different between cut (51± 3%)and injected (45± 3%) plots. Species had different finalpercent survival and rates of mortality. Species survivaldiffered greatly by species (in descending order): Fraxinuspennsylvanica>Quercus muehlenbergii‡Prunus serotina‡ Juglans nigra >Cercis canadensis >Cornus florida.Survivorship and growth of native seedlings was affectedby a severe first-year drought and by site location. One siteexhibited greater spring soil moisture, pH, percent opencanopy, and had greater survivorship relative to the othersite (55± 2 vs. 30± 2%). Overall, both L. maackii eradica-tion methods were successful, but restorationists should beaware of the potential for differential survivorship of nativeseedlings depending on species identity and microenviron-mental conditions.
Key words: biodiversity, forest ecology, invasive species,non-native, Ohio, recruitment, restoration ecology, seed-lings, succession, tree tubes, understory.
Introduction
Invasions of nonindigenous species are often facilitated byanthropogenic disturbances (Hobbs & Huenneke 1992)and can be problematic in that they can cause furtherchanges in forest attributes including structure, properties,and fundamental ecosystem processes (MacMahon & Holl2001). Furthermore, if extremely successful species are theinvaders, biodiversity can be essentially reduced to a nearmonoculture resulting in a community that is low in nat-ural diversity and extremely difficult to restore (Olson &Whitson 2002).The aggressiveness of some invasive species can make
wholesale eradication quite difficult; however, their
removal at a specific site is a necessary step in the res-toration process. Eradication at local spatial scales isespecially important, because the restoration process, forthe most part, proceeds on a site-by-site basis across largeareas (Wiens et al. 1993).
After removal of invasive species, forest restorationpractices may include replanting sites with native species(Ghersa et al. 2002). This replanting step in restoration isoften necessary especially in deciduous forests of the east-ern United States where viability of seeds is relativelyshort compared with other habitat types (Barnes et al.1998) and where invasives have occupied a site longenough to result in the degradation of a normally short-lived forest seed bank (Collier et al. 2002).
Replanting may have benefits in addition to restorationof forest diversity. Replanting may inhibit further invasion,because native individuals can preempt space and acquireresources, which makes reestablishment by invasives moredifficult (Shea & Chesson 2002). Natural recovery
Department of Environmental and Plant Biology, Ohio University, Athens,OH 45701-2979, U.S.A.1Address correspondence to B. C. McCarthy, email [email protected]
� 2004 Society for Ecological Restoration International
154 Restoration Ecology Vol. 12 No. 2, pp. 154�165 JUNE 2004
processes, including succession, are often accelerated byreplanting, changing the forest from a nonindigenous,highly modified ecosystem to one with native species andself-sustaining processes. Planting natives would also cir-cumvent the problem of lack of successful recruitmentfrequently encountered in reestablishment of plant com-munities (Holmes 2001). All of these factors would resultin the restoration of a more structurally representative andfunctional forest community.
Another important benefit from conducting a res-toration experiment is that practical experience can begained in developing a protocol for reestablishing naturalforest ecosystems. A plan should successfully integrateabiotic factors (e.g., microsite variability) as well as bioticfactors (e.g., seedling requirements) influencing therestoration of the forest community (Reifsnyder & Lull1965). In addition, a consideration of the relative economiccosts of different restoration methods can help guideallocation of sponsorship in future, similar restorationprograms (Montalvo et al. 2002).
One of the most challenging invasive plants to forestrestorationists in the eastern United States is the nonindi-genous shrub, Amur honeysuckle (Lonicera maackii[Rupr.] Herder; Caprifoliaceae). This plant has spread toover 27 eastern U.S. states (Luken & Thieret 1996), theprovince of Ontario, Canada, and many counties insouthwestern Ohio (Hutchinson & Vankat 1997; Trisel1997) where this experiment was conducted. Loniceramaackii is only one member of a genus with manyknown weedy qualities (Woods 1993; Schierenbeck et al.1994).
A number of factors make L. maackii a threat to nativebiodiversity and a challenge to restoration practitionersincluding its ability to resprout after repeated cutting(Luken 1990), possible allelopathic effects on nativevegetation (Trisel 1997), and extended leaf phenology(Trisel 1997). When L. maackii invades open sites, theseareas are often converted into shrub communities (Luken &Thieret 1995), and in invaded forests, recruitment anddiversity of woody and herbaceous species is oftenreduced (Hutchinson & Vankat 1997; Gould & Gorchov2000; Collier et al. 2002; Gorchov & Trisel 2003). Despiteefforts to eradicate L. maackii, a fully integrated resto-ration protocol for the eradication of this species and sub-sequent replacement with natives has not yet beenproduced. This precipitated our investigation into theoptimal methods for restoring native vegetation after L.maackii removal.
The specific goals of this study were to: (1) quantify thebiomass of L. maackii in the study area; (2) comparemethods of L. maackii eradication in terms of effectivenessand ease of application; (3) compare survival and growthof native tree seedlings planted among L. maackii eradica-tion treatments; (4) evaluate the effects of tree protectorson native seedling survival; and (5) explain the influence ofmicroenvironmental factors on tree seedling growth andsurvival.
Methods
Study Site
This experiment was carried out at the Fernald Environ-mental Management Project (FEMP) Site (39�1802000 N3
84�4105000 W), a 425-ha facility located circa 29 km north-west of Cincinnati, Ohio. The old production facility atFEMP was used for the manufacture of high-gradeuranium and thorium to support the U.S. weapons defenseprogram until 1989 at which time remediation and restora-tion became the primary goal at the FEMP site (U.S.Department of Energy 2002). This research project wasconducted in the ‘‘North Woodlot’’ within the FEMP site,which is approximately 65 ha and is located north of theold production facility. The north woodlot area containsfour general types of habitats including ‘‘old fields, pre-viously mowed meadows, regenerating forest, and matureforest’’ (McCarthy 1999).
This area has been subject to considerable anthropo-genic related disturbance (e.g., mowing, roads, and graz-ing), which may have facilitated the influx of Loniceramaackii. A floristic study conducted on the site byMcCarthy (1999) reported that out of 332 taxa, 30.5%were nonindigenous with the most invasive species beingRosa multiflora (multiflora rose), Alliaria petiolata (garlicmustard), Festuca elatior (tall fescue), and Polygonumpersicaria (lady’s thumb smartweed); however, he reportedthat L. maackii was the most problematic non-native plantin the forested areas.
This area lies within Butler County and has a climatethat is typical of most of southern Ohio with cold wintersand hot summers. The mean temperature from Decemberto February is 0 �C, and the mean temperature from Juneto August is 22 �C (National Climate Data Center andNational Oceanic and Atmospheric Administration2002). The total annual rain precipitation is 53.3 cm, andthe total snowfall is 38.1 cm. Sixty percent of the totalprecipitation usually falls from April through September,although precipitation is present in every month. Thegrowing season is 172 days (Lerch et al. 1980).
Experimental Design
This experiment used a completely randomized blockdesign (Sokal & Rohlf 1995) with eight 5.53 13.5m repli-cate blocks. Blocks were located in two areas roughly150m apart containing Lonicera maackii stands (sitesA and B). Site A has soil that is primarily Xenia silt clayloam (XeB) and lies on a well-drained till plain. Site B lies onRagsdale silt clay loam (Ra) and has poorly drained soilswith a flat topography (Lerch et al. 1980). Site A contained atotal of five replicate blocks and was dominated byCarya laciniosa (shellbark hickory) in the overstory. Site Bcontained three replicate blocks and was dominated bya mixture of Acer negundo (boxelder), Cornus florida(flowering dogwood), Fraxinus pennsylvanica (green ash),Prunus serotina (black cherry), and Ulmus Americana
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 155
(American elm) in the overstory. Three rectangular4.53 5.5m treatment subplots were established withineach block. Treatments were randomly assigned to sub-plots and were designed to test their efficacy of killingL. maackii and ultimately their influence on the survival ofnative tree seedlings. Each block consisted of (1) a controlsubplot in which no L. maackii was removed; (2) a cutsubplot in which all L. maackii stems were cut near groundlevel, removed from plot, and stumps painted with 50%glyphosate isopropylamine salt solution (Roundup;Monsanto Company, St. Louis, MO, U.S.A.); and (3) an injec-tion subplot in which L. maackii was killed using an EZ-Ject lance (Odum Processing Engineering Consulting, Inc.;Waynesboro, MO, U.S.A.) but left standing. This lanceworks by pushing a glyphosate-filled 22-caliber capsule(Bergerud 1988) manually through the bark of the stemor swollen stem base and into the vascular system of theselected woody plant. Only stems 1.5 cm and larger wereable to be injected without operator error. On large honey-suckle individuals with two or more stems, each stemwas injected separately; otherwise, plants were injectedindividually. Before eradication treatments all honey-suckle stems in subplots were tagged and measured.Within each treatment subplot, 10 individuals each of six
species of 1-year-old indigenous tree seedlings were ran-domly planted 0.75m apart using a dibble bar. Tree specieswere chinkapin oak (Quercus muehlenbergii), black walnut(Juglans nigra), black cherry, green ash, flowering dog-wood, and redbud (Cercis canadensis). Seedlings werechosen as model midstory and overstory species to studythe effects of honeysuckle control methods on seedlingperformance. Furthermore, these species all occur withinthe county where this experiment took place (Braun 1989)and were found as canopy or subcanopy species in thestudy area (McCarthy 1999). It was observed that recruit-ment of tree species was especially poor below L. maackiistands (also found by Hutchinson & Vankat 1997; Collieret al. 2002), which justified replanting after eradication ofhoneysuckle. Half of the 1,440 tree seedlings wereenclosed in 122-cm Protex Pro/Gro Solid Tube Tree Pro-tectors (Forestry Suppliers, Inc., Jackson, MO, U.S.A.) toexclude white-tailed deer (Odocoileus virginianus) and testthe effects of deer browse on tubed versus nontubed seed-lings. Blocks were surrounded by two strands of barbedwire (50 and 100 cm height) with a 1-m buffer to excludecattle (present for the first year of the study) but not deer.Honeysuckle eradication treatments were applied on 24March 1999. Seedlings were planted from 24 March 1999to 30 March 1999, and deer tree protectors were appliedand staked on the north side from 31 March 1999 to11 April 1999. The experiment ran from 24 March 1999 to10 October 2001.
Sampling
Seedlings were measured for height (cm) and basal diam-eter (mm) at initial planting. A preliminary ANOVA
indicated that seedling size was not significantly different(p. 0.10) among blocks or treatment subplots at initialtime of planting; therefore, an analysis of covariance wasnot needed. Seedlings were also measured for height anddiameter and mortality recorded (27 May 1999, 10 October1999, 21 May 2000, 23 September 2000, 25 May 2001, and7 October 2001).
Environmental data were collected over the growingseasons of 1999 and 2000 on a treatment subplot level.Soil moisture was measured twice in 1999 (25 June and21 August) and 2000 (29 July and 29 August). Soil mois-ture was analyzed gravimetrically (McCarthy 1997). SoilpH was measured using a glass electrode method witha Corning 350 pH/ion meter (Corning Inc., Corning, NY)in a 2 : 1 water-to-soil solution (25 June 1999, 21 August1999, 29 July 2000, and 29 August 2000). Soil nitrate wasmeasured using absorbent Rexyn 300 (H�OH) beads(Fisher Scientific, Fair Lawn, NJ, U.S.A.) buried in thesoils’ A-horizon in nylon mesh bags for roughly 3 months(20 May 1999 to 21 July 1999 and 21 May 2000 to 22 July2000). Nitrate was removed from the Rexyn beads with 2MKCl solution and analyzed with a cadmium reductionmethod using NitraVer 5 Nitrate Reagent (Hach, Loveland,CO, U.S.A.). Three soil moisture, nitrate, and pH sampleswere taken at random locations in each subplot each timesampling occurred. Light availability was measured using35-mm images taken with an 8-mm hemispherical fish-eyelens at a height of 0.5m on 30 July 1999 and 27 July 2000.These images were digitized and then analyzed using GLICsoftware (Canham 1988). A full description of the protocolcan be found in Robison and McCarthy (1999). One imagewas taken per subplot (24 total photographs). Air tempera-ture and humidity were measured for eight random loca-tions inside and outside of deer tree protectors in eachsubplot on 12 July 1999 and 13 July 2000 with a Corningthermohygrometer (Corning Inc., Corning, NY).
Statistical Analysis
To assess abundance of Lonicera maackii within blocks andamong treatment plots, a regression model was constructedto estimate honeysuckle biomass (from stem density anddiameter) within blocks and treatment subplots usingrandomly selected on-site honeysuckle plants (n5 32).Honeysuckle plants were oven-dried at 105 �C for 72 hr.
An ANOVA was used to compare final native seedlingsurvival among treatments after 3 years. Percent survivalwas used as the independent variable, and the factor vari-ables were site (random factor), block within site (nested),honeysuckle eradication treatment (fixed), species (fixed),and tree protector (fixed). Normality and equal varianceassumptions were satisfied using the D’Agostino omnibustest (D’Agostino et al. 1990) and the F-Max test (Dowdy &Wearden 1991), respectively. When necessary, datatransformations (log10 or square root) were conducted tomeet these assumptions. Untransformed means and SEsare reported throughout the results.
Forest Recovery After Shrub Removal
156 Restoration Ecology JUNE 2004
The shapes of survival curves were analyzed to deter-mine whether the native seedlings died at different ratesduring the 3 years of sampling. Site, treatment, and speciessurvival distributions were compared using the log ranknonparametric test. This test was utilized because morethan 20% of the seedlings were still alive at the end ofthe experiment (i.e., censored observations; Pyke &Thompson 1986).
Loglinear modeling was used to analyze the causes ofmortality for native seedlings. This uses an n-dimensionalcontingency table and a stepdown model selection processto determine the most important factors related to whetherseedlings died by drought, fungus, handling and transplant-ing, or deer browsing. Computed chi-square values werecompared with critical values to determine statistical sig-nificance.
MANOVAs were used to examine the microenviron-mental patterns, as they related to years (1999 and 2000),eradication treatments (control, cut, and inject), and sites(A and B), which were all used as predictor variables. Allfactors were treated as fixed, except site. Spring soil mois-ture and pH, nitrate, percent open canopy, ambient airtemperature, and relative humidity were analyzed asresponse variables. Autumn soil moisture and pH weredropped from the analysis due to multicollinearity, butthis did not affect the MANOVA results. The assumptionsof multivariate normality and equal variance (Scheiner1993) were satisfied before analysis via data transfor-mations (log10). Wilks’ Lambda values were calculated forthe overall MANOVA table, and individual ANOVAswere conducted to determine which factors were importantin producing the overall MANOVA results.
A repeated measures analysis of variance (RMANOVA)was used to analyze seedling height growth. The datapassed the assumptions of residuals having normalprobability distribution and equal within-subject covari-ances (Von Ende 1993). There was a violation of thesphericity pattern of the covariance matrices (Mauchley’scriterion5 144.16, p5 0.035); therefore, a Huynh�FeldtEpsilon correction (E5 0.95) was used to create a morestringent critical F value (Crowder & Hand 1990). Seedlingdiameter was also analyzed with an RMANOVA. Like theheight data, diameters suffered from lack of sphericity ofthe covariance matrices (Mauchley’s criterion5 620.07,p, 0.001). A Huynh�Feldt correction was used to create amore conservative F test (E5 0.88). Site was a randomfactor, and individual seedling was nested within all otherfactors such as eradication treatment, species, and tubing,which were fixed. To aid in interpretation of interactions,Cicchetti contrasts were computed (Cicchetti 1972).Cicchetti contrasts enable the analysis of every unconfoundedcomparison by holding all factors constant but one.
All statistical computations of ANOVAs, loglinear mod-eling, RMANOVAs, and Bonferroni post hoc tests wereconductedusingNCSSVersion5.0 (Hintze 2000).MANOVAcalculations were conducted using SAS Version 8.0(SAS Institute 2001). Bonferroni multiple comparison tests
were conductedwhen significantF testswere found forANO-VAs. Unless otherwise stated, statistical tests were significantwhen p, 0.05.
Results
Honeysuckle Parameters
The regression model showed that honeysuckle basal areaat 5 cm aboveground explained the greatest amount ofvariance; therefore, this was the parameter that was usedto estimate honeysuckle biomass [honeysuckle biomassper individual (kg)5basal area (in cm2)3 0.9071 0.147,r25 0.91]. This equation estimates biomass very similarlyto that reported by Luken (1988). Mean biomass of Amurhoneysuckle was 361 ± 69 kg/ha. Mean stem density was65,959± 7,637 stems/ha. Mean density of Lonicera maackiiwas 21,380± 3,171 plants/ha. The mean number of stemsper plant was 3.65 ± 0.26. A preliminary ANOVA wasconducted on honeysuckle biomass to evaluate pre-treatment differences among blocks and subplots. Nosignificant (p. 0.10) differences were found.
Eradication Treatment Effectiveness on Honeysuckle
Honeysuckle mortality was assessed at the end of the 1999and 2000 growing seasons. At the end of 1999, above-ground mortality was 99% for both eradication treatments.It was difficult to inject stems smaller than 1.5 cm; there-fore, delayed resprouting did occur due to operator errorduring the initial injection treatment. For 2000, plants had98% mortality, where no operator error occurred in theinjection plots (95% if operator error was included), and94% in the cut and paint treatment plots. ANOVA resultsshowed that there were no significant differences (p. 0.10)in honeysuckle mortality among treatment plots orbetween years.
Ease of Application of Eradication Treatments, Planting, and
Tubing
The cut and paint method required more time to apply(7 hr to eradicate 66m2 or 1,060 person-hours/ha) than theinjection treatment (3 hr to eradicate 66m2 or 454 person-hours/ha). In the injection treatment one person operatedthe EZ-Ject lance and another person cleared leaf litteraway from honeysuckle stump bases when necessary.Planting 1,440 seedlings required 4 days of work by twopeople (or 64 person-hours), and assembly and applicationof tree protectors on half of the seedlings (720 plants)required 8 days of work by three people (or 192 person-hours).
Comparison of Costs of Restoration
The startup costs for the cut and paint method totaled $253USD, which included clippers ($30), loppers ($65), andglyphosate herbicide (2.5 gallons at $158). The injection
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 157
startup costs were $599, which included the EZ-Ject lance($467) and glyphosate-filled capsules (1,200 capsules at$132). Protex Pro/Gro Solid Tube Tree Protectors cost$1.74 each. Seedlings cost roughly $0.33 each for 1,440seedlings totaling $480. The time required for implementa-tion of treatments was reported, but labor costs of eradica-tion treatments, tree seedling planting, and tree protectorassembly and installation were not calculated due to thefact that our reports of eradication treatment times arelikely overestimated, as the application times for personsdoing frequent restoration treatments would likely bemuch less. Total cost including purchase of startup equip-ment, native seedlings, and tree protectors (but excludinglabor) was $831 for the cut treatment and $1177 for theinjection treatment.
Comparison of Survival for Native Seedlings
The survival of planted seedlings was 51± 5% for 1999,44 ± 5% for 2000, and 40± 5% for 2001. An analysis ofvariance of final survival of seedlings (after 3 years)found that two factors, block and tube, were nonsignificant(p. 0.05) and were not included in subsequent analyses.Site, treatment, species, and site3 species were all signifi-cant (Table 1). Overall survival at site B (56 ± 2%) wassignificantly greater (p, 0.001) than at site A (30 ± 2%).The site3 species interaction was significant (p, 0.001),indicating that species survived differently in the twosites (Table 1; Fig. 1). Survival between control andhoneysuckle treatments was also significantly differ-ent (p5 0.002; control survival5 32 ± 3%, cut survival551± 3%, injection survival5 45± 3%). Survival betweencut and paint and injection eradication treatments wassignificantly different (Fig. 2). No other two- or three-wayinteractions were found to be significant (Table 1).Because of significantly different final survival percent-
ages between sites, among years, and among treatments,survivorship was modeled to determine whether seedlingsdied at different rates within these groups. Weibull distri-
bution curves (Dodson 1994) produced the best estimatesof species survivorship. Seedlings within control treat-ments died at different rates than the cut (w25 40.67,df5 1, p, 0.001) and injection treatments (w25 23.68,df5 1, p, 0.001); however, the cut was not different thanthe injection treatment (w25 2.38, df5 1, p5 0.123)(Fig. 3). Because there was a site3 species interaction forsurvival at the end of 2 years, species survival rates wereanalyzed separately within sites. Several species (Fraxinuspennsylvanica, Prunus serotina, and Juglans nigra) hadfairly constant mortality rates during the experiment atboth sites. Other species (Quercus muehlenbergii at siteA and Cercis canadensis and C. florida at site B) had highrates of early death followed by constant mortality. Cerciscanadensis andC. florida at site A had fairly constant but highmortality for only the first 2 years of the experiment withstabilized survival during the third year (Fig. 4).
Causes of Mortality for Native Seedlings
The causes of mortality in different treatments wereassessed to determine whether seedlings died by differentmeans among sites, honeysuckle treatments, and species.Overall the most common cause of mortality was drought(39.4% of all seedlings) followed by handling and trans-planting (6.3%) and browsing (3.1%). Powdery mildew(fungus) was detected on the leaves in early spring andaccounted for a small proportion of mortality (1.2%).
Logistic regression results indicated that the factorsassociated with mortality due to drought were site, treat-ment, and species (w25 34.5, df5 10, p, 0.001). Droughtwas a greater factor of mortality at site A (29.7%) than atsite B (9.7%), and drought was a greater factor of mortal-ity in control plots (16.5%) than in injection (11.8%) or cutplots (11.1%). Fraxinus pennsylvanica had less mortalitydue to drought (2.4%) than the other five species(6.1�8.2%). Factors associated with mortality due tobrowsing were site and species (w25 7.6, df5 2,p5 0.022). Overall, mortality due to browsing was 1.9%
Table1. ANOVA of final survival for native tree seedlings.
Source Term df SS MS F ratio p Power (a5 0.05)
Site 1 4.438 4.438 84.07 ,0.001Treatment 2 1.741 0.704 331.04 0.003 1.000Control versus treatments* 1 25.27 0.002Cut versus inject* 1 8.23 0.014
Site3 treatment 2 0.005 0.003 0.05 0.951Species 5 8.339 0.668 6.65 0.029 0.771Site3 species 5 1.253 0.251 4.75 ,0.001Treatment3 species 10 0.403 0.040 0.43 0.902 0.129Site3 treatment3 species 10 0.941 0.094 1.78 0.064S 52 13.304 0.053Total (adjusted) 28 31.917Total 288
*Orthogonal contrasts.
Forest Recovery After Shrub Removal
158 Restoration Ecology JUNE 2004
with site A having a greater mortality due to browsing(2.8%) than site B (0.3%). Species’ mortalities due tobrowsing were as follows: Cercis florida (1.5%), C.canadensis (1.1%), and Quercus muhlenbergii and Prunusserotina (0.2%). Fraxinus pennsylvanica and Juglans nigra
had no incidences of browsing. Mortality due to funguswas low but best predicted by the factors of site andspecies (w25 5.4, df5 2, p5 0.067). Site A had a slightlygreater incidence of fungus (0.8%) than site B (0.3%). Theincidences of fungus mortality per species were as follows:C. florida (0.5%), C. canadensis and Q. muhlenbergii(0.14%), and F. pennsylvanica and J. nigra (0.07%). Mortal-ity due to handling was only significant for species(w25 441.31, df5 5, p, 0.001). The handling mortalityincidences were mostly associated with C. florida (4.0%)followed by C. canadensis (2.1%) and J. nigra (0.1%)(Fig. 4).
Treatment
Control Cut Inject
Fin
al s
eedl
ing
surv
ival
(%
) + S
E
0
10
20
30
40
50
60
a
b
c
Figure 2. Final seedling survival (after 3 years) in eradication
treatments. Lowercase letters indicate significant differences
(p, 0.05).
20
40
60
80
100
aa
b
CutInjectControl
Initial Spring1999
Fall1999
Fall2000
Spring2000
Spring2001
Fall2001
Wei
bull
dist
ribut
ion
estim
ates
of s
urvi
val S
(t)
Figure 3. Weibull distribution function estimates for native seedling
survival rates in three eradication treatments. Survival rates in the two
eradication treatments were greater than that in the control treatment
plots where Amur honeysuckle was left intact. Lowercase letters
indicate significant differences (p, 0.05) among treatments.
a Control
CECA COFL FRPE JUNI QUMU PRSE
Mea
n su
rviv
al (
%) +
SE
0
20
40
60
80
100
b Cut
CECA COFL FRPE JUNI QUMU PRSE
Mea
n su
rviv
al (
%) +
SE
0
20
40
60
80
100
c Inject
Species
CECA COFL FRPE JUNI QUMU PRSE
Mea
n su
rviv
al (
%) +
SE
0
20
40
60
80
100
Site ASite B
Site ASite B
Site ASite B
Figure 1. Percent final survival of planted seedlings comparing sites
across treatments (panels a�c). All species had greater survival at site
B relative to site A, except Fraxinus pennsylvanica, which had greater
survival in the two eradication treatments of site A (panels b and c).
CECA, Cercis canadensis; COFL, Cornus florida; FRPE,
F. pennsylvanica; JUNI, Juglans nigra; PRSE, Prunus serotina;
QUMU, Quercus muhlenbergii.
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 159
Seedling Growth
Seedling height growth was analyzed for each year, 1999,2000, and 2001. Growth was not significantly differentbetween sites (p5 0.40); therefore, site was subsequently
removed from the analysis. Seedlings grew equally well intubed and nontubed conditions (p5 0.79) and among eradi-cation treatments (p5 0.99). There were significant year,species, and species3 year effects (all p, 0.001) (Fig. 5).
Analysis of seedling diameters used the same factors asthose in the height RMANOVA. Site was nonsignificant(p5 0.83) and was dropped from the analysis. Diametergrowth was not significantly different between tubed andnontubed seedlings or among treatments. There wasa significant year (p, 0.001) and species3 year interaction(p5 0.01). There were no significant differences amongspecies’ diameters for 1999 and 2000; however, therewere significant species differences for 2001 (Fig. 6).
Analysis of Environmental Parameters
Only two factors, site and year, were significant (p, 0.001)in predicting microenvironmental response. There were noenvironmental differences among eradication treatments.Also, no two- or three-way interactions were significant.Follow-up ANOVAs indicated that spring moisture, pH,nitrate, and temperature were significantly different(p, 0.001) between years. Site A had significantlylower values for environmental measurements (springmoisture, spring pH, and percent open canopy) than siteB (ANOVA results). For all of these environmentalparameters 1999 had lower values than 2000. Spring mois-ture and pH and percent open canopy had significantANOVAs for site; site A had lower values than site B forthese environmental measurements (p, 0.05).
The Palmer drought severity index (PDSI) values weresubstantially different for the 3 years of the study. For 1999the PDSI values ranged from a mild drought (May andJune) to moderate and severe drought (August toOctober) to extreme drought (November and December).The 2000 PDSI values ranged from normal to slightly wetfor the entire year, and the 2001 PDSI values ranged froma mild drought (March and April) to very wet (October)(National Climate Data Center and National Oceanic andAtmospheric Administration 2002).
100
80
60
40
20
0Initial Spring
1999Fall
1999Fall
2000Spring2000
Spring2001
Fall2001
Wei
bull
dist
ribut
ion
estim
ates
of s
urvi
val S
(t)
100
80
60
40
20
0Initial Spring
1999Fall
1999Fall
2000Spring2000
Spring2001
Fall2001
Wei
bull
dist
ribut
ion
estim
ates
of s
urvi
val S
(t)
FRPEQUMUPRSEJUNICECACOFL
Site A
Site B
a
bbb
cd
abbbb
c
Figure 4. Weibull distribution function estimates for native species
survival rates in two sites. Overall survival rates at site B were greater
than those at site A (p, 0.001). Lowercase letters indicate significant
(p, 0.05) differences among species within a site. CECA, Cercis
canadensis; COFL, Cornus florida; FRPE, Fraxinus pennsylvanica;
JUNI, Juglans nigra; PRSE, Prunus serotina; QUMU, Quercus
1999 20012000
Species
ab
abab ab
a
a
c
bb b b
c
CECA COFL FRPE JUNI PRSE QUMU
Hei
ght G
row
th (
cm) +
SE
2
4
6
8
10
12
14
16
18
Species
CECA COFL FRPE JUNI PRSE QUMU
2
4
6
8
10
12
14
16
18
Species
CECA COFL FRPE JUNI PRSE QUMU
1999
Figure 5. Seedling height for 1999, 2000, and 2001. Lowercase letters denote significant (p, 0.05) differences among species within a single year. No
significant differences (p, 0.05) existed among species for 2000. CECA, Cercis canadensis; COFL, Cornus florida; FRPE, Fraxinus pennsylvanica;
JUNI, Juglans nigra; PRSE, Prunus serotina; QUMU, Quercus muhlenbergii.
Forest Recovery After Shrub Removal
160 Restoration Ecology JUNE 2004
Discussion
The successful restoration of a forest after the eradicationof invasive plants includes the restoration of the overalldiversity of the site (Sinclair et al. 1995) as well as restoringthe composition to a close approximation of the originalhabitat (Harrington 1999). Also important is the resto-ration of the community structure (Holmes 2001) andeco-system-level processes (Vitousek 1990). Not all ofthese attributes are completely restorable within a single,short-term project, but restorations should include thesegoals, which if accomplished, will set the stage for appro-priate natural successional trajectories.
The main goal of our experiment was to accelerate suc-cession (see MacDonald 1993). In our case, after eradicatingLonicera maackii, we desired to increase the rate of succes-sion by overcoming the problem of limited dispersal ofpropagules and relatively unsuccessful recruitment, whichis a frequently encountered problem in restoration efforts(Robinson & Handel 2000). Planting native species accom-plished the restoration of the forest tree composition, whichis similar to a nearby, relatively undisturbed, mature refer-ence forest (McCarthy 1999). Restoring the compositionwill likely reestablish the canopy and midstory structure aswell as recruitment processes in the future, and this will helpfacilitate the return of higher-level processes as Cairns(1986) states that most of the functional attributes ofa restored system are correlated with the replacement ofits vegetative structure and composition.
We also wanted to explain some of the results of theexperiment to aid in future restoration efforts (e.g., thesurvival of the native seedlings). Seedling establishment ismost importantly influenced by the existing environmentalconditions during the early stages of life (Walters & Reich2000). The differential survival of native seedlings inour experiment was largely due to individual species’responses to sites’ microenvironmental conditions, whichvaried according to location, L. maackii eradication treat-ment, year, and other perturbations such as fungal infec-tion, browsing, and handling mortality.
Actually species-specific responses to site conditions arequite common (Veenendaal et al. 1996; Sipe & Bazzaz2001) as is year-to-year variance in survival (Van DerMeer et al. 1999). These differences in seedling survivallikely will result in a forest composition different than theone originally intended during initial planting; therefore,follow-up procedures such as replanting may be necessaryin restorations where survival of native seedlings is rela-tively unpredictable and a particular forest composition isdesired. Our results clearly indicate the need for an aware-ness of specific site conditions along with the planting ofa diversity of tree species suited to that site.
Another factor influencing the survival of seedlingswas the presence or absence (through eradication) ofL. maackii. It is not surprising that reducing L. maackiiabundance was associated with increased survival of nativeseedlings. Other restoration projects have also foundgreater survival after clearing of invasives using herbicidetreatments (Sweeney et al. 2002). Survival of native seed-lings was clearly greater when L. maackii was killed byeither eradication methods rather than being left intact.
The mechanism for greater native seedling survival ineradication plots is not evident, because the microenviron-mental conditions that we measured were not statisticallydifferent between the eradication and control treatments.Significantly greater temperature, light, soil moisture, andpH levels have been found where L. maackii was killedversus left intact (C. Keiffer 2002, Miami University,Oxford, Ohio, personal communication). Greater lightintensity, although not statistically significant, may accountfor the increased survival that we found in eradicationplots. Work by Gorchov and Trisel (2003) may offer anexplanation for lower seedling survival in L. maackii con-trol versus removal plots. They found that native seedlingshad increased mortality when grown with L. maackii dueto mostly shoot, but also root, competition. They suspectedaboveground competition would mostly be struggle forlight. Trisel (1997) also found that L. maackii plants maybe allelopathic, as watering with L. maackii leaf extract
1999
SpeciesCECA COFL FRPE JUNI PRSE QUMU
Dia
met
er g
row
th (
mm
) + S
E
0.5
1.0
1.5
2.02001
SpeciesCECA COFL FRPE JUNI PRSE QUMU
0.5
1.0
1.5
2.02000
SpeciesCECA COFL FRPE JUNI PRSE QUMU
a
c
c
aa
b
Figure 6. Seedling diameter for 1999, 2000, and 2001. Lowercase letters denote significant differences among species within a single year. No
significant differences (, 0.05) existed among species for 1999 and 2000. CECA, Cercis canadensis; COFL, Cornus florida; FRPE, Fraxinus
pennsylvanica; JUNI, Juglans nigra; PRSE, Prunus serotina; QUMU, Quercus muhlenbergii.
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 161
had effects similar to those of a 1024 solution of juglone,a known allelopathic chemical. Generally, seedlings inL. maackii control subplots may have been chemicallyinhibited or had fewer resources than in areas whereL. maackii was eradicated. Clearly more work needs tobe conducted to elucidate the exact nature of L. maackii’sinterference mechanism.In addition to the two eradication treatments that we
used, others have used several other methods to controlL. maackii with varying levels of success. A 1% foliar gly-phosate spray has been used to control seedlings aswell as tocontrol adult L. maackii along heavily invaded edge habi-tats. The advantage of foliar spraying is that it is relativelyeasy toapply (Conover&Geiger 1993); however, it can resultinup to100%mortality tonative vegetation in the underlyingherb layer (Trisel 1997).Others have argued similarly againstthe use of herbicides in restoration areas because of negativeeffects on nontarget plants (Brockway et al. 1998).Cutting alone has been found to be a less than adequate
control method for L. maackii, because resprouting occursfrom a meristematic burl located at the base of stems(Luken 1988). Although resprouting after clipping occursless frequently in forested versus open habitats (30 vs.70%, respectively), clipping alone has been found to actu-ally increase stem numbers (Luken 1990). We do notrecommend clipping alone unless honeysuckle is growingin a closed canopy and a clipping regimen will be con-tinued to control resprouts.Cutting and applying herbicide (i.e., cutting and paint-
ing) were found to be effective in our experiment and isone of the most widely used eradication procedures forwoody invasive plants (Reinartz 1997; Olson & Whitson2002). A 20% glyphosate solution has been found to beeffective in controlling L. maackii in forest interiors, anda 50% solution is more appropriate in open habitats wherethe plant seems more resistant (T. Borgman, HamiltonCounty Park District, 2002, personal communication).Our experiment successfully controlled Amur honeysuckleusing a 50% solution for the cut and paint treatment ina young closed-canopy forest.There are several disadvantages to cutting. Cutting
alone and the cutting and herbicide method leave behindstump bases, and piles from cutting take longer than indi-vidual stems to decay. Furthermore, herbaceous vegeta-tion does not develop beneath piles. Piles can be chippedor removed, but they may have the advantage of creatingwildlife habitat for animals (J. Klein, Hamilton CountyPark District, 2002, personal communication). The biggestdisadvantage of eradication involving L. maackii cutting isthat it is very labor intensive.A number of other L. maackii eradication methods have
been reported, but each has drawbacks. Hand pulling canbe effective in areas with moist ground, but the plants willlikely resprout, if root portions remain (Gayek 2000). Gen-erally hand pulling is not difficult, if individual plants areless than 3 years old or growing on moist soils, but thespread of L. maackii from remaining roots may not make
this method effective (Gayek 2000). Other methods forcontrolling L. maackii include using a ‘‘weed wrench’’ toremove whole crowns (100% mortality) or a polaski axe(98% mortality); however, these methods are very laborintensive (Trisel 1997).
We found that the injection system may be the bestoverall method for the eradication of L. maackii. Thereare several advantages to its use. Injecting produced verylittle operator fatigue and limited exposure of the operatorto herbicide. Also less overall herbicide is used relative toother methods, and the herbicide that is used is restrictedentirely to target plants. We also found that injecting was43% faster than cutting and applying herbicide. This maybe of considerable importance in larger restoration efforts.Herbicide injections have been used by others to killwoody plants and prevent regrowth (Johansson 1985).Franz and Keiffer (2000) successfully used the injectionsystem to eradicate L. maackii in a stand in southwestOhio. They found that fall injections are more effectivethan spring and that it is more effective to inject all stems,rather than a single stem, of a L. maackii plant. Because ofthe difficulty in injecting small L. maackii stems, werecommend that injection is used only on stems 1.5 cmand larger and that the cut and paint method should beused to control smaller individuals.
Other considerations in restorations of this type includethe use of tree protectors, particularly where large mam-mal browsing is a problem. Sweeney et al. (2002) foundtree seedlings survived best when a combination of herbi-cide and tubing treatments was applied; however, we didnot find that tree protectors increased survival or growthof native seedlings. We found that despite there beingenvironmental differences in tubed and nontubed condi-tions, there were no significant survival or growth differ-ences for native seedlings. Mortality due to browsing wasonly 3.1% at this site; however, it is widely known thatdeer are keystone herbivores and can have profoundimpacts on forest composition (Rooney 2001). Barbedwire fencing around blocks may have inhibited deer brows-ing, making the browsing incidences artificially low. Per-haps if deer populations are a problem locally, then treeprotectors would be justified.
It is best to target control of L. maackii populationswhen they are small. Deering and Vankat (1999) recom-mend that early control is best, because it is done beforeplants reach reproductive maturity, which is roughly 4 or 5years. This approach does have merit, as the early controlof invasives is often less costly and has a greater chance ofsuccess than control measures taken later (Chippendale1991). Furthermore, reinvasion is likely unless all plantsare eradicated from an area. Restoration activitiesmay make site conditions more favorable for invasivereestablishment. In artificial gaps where Amur honeysucklewas removed, reinvasion by Amur honeysuckle and garlicmustard was demonstrated to be more likely than in intactstands (Luken et al. 1997). When total eradication is accom-plished, L. maackii must repopulate from another site, and
Forest Recovery After Shrub Removal
162 Restoration Ecology JUNE 2004
reinvasion is likely to occur in small, manageable amounts(Deering & Vankat 1999).
Although restoration costs are rarely reported in aca-demic articles, the economics of a restoration project arevery important (Holl & Howarth 2000). While seeminglyinitially expensive, the cost of eradicating and restoringnative communities is often recovered in a short time(Zavaleta 2000). Various restoration studies have shownthat the cost of the removal of L. maackii can be quitevariable. Gayek (2000) reported the cost of an L. maackiirestoration to be $8,200 for a single 2-ha project, whichincluded labor for a crew of 30 as well as safety equipment,7.5 gallons of herbicide, and the rental of a brush chipper.Gayek reported $2,000 for the cost of another removalproject on 4 ha of land, which was less expensive, becausevolunteers were used and no brush chipper was used.Trisel (1997) indicated lower costs of L. maackii erad-ication. He gave the startup costs for three methods ofL. maackii control including crown removal using a polaskiaxe and hand saw ($42), foliar spray using herbicide andbackpack sprayer ($165), and the stem cut and paintmethod using clippers, loppers, and herbicide ($172).However, he did not include the cost of labor in his esti-mates or account for per hectare area costs. Additionally,he states that these methods were either labor intensive ordamaging to native plants.
For our project, cutting and painting was less expensive($253) in terms of startup costs than the injection method($599); however, the most expensive part of restorationmost likely would be labor for the implementation oferadication treatments and planting of native seedlings.The amount of time required for our experiment to injectan area was roughly 43% less than the cut and painttreatment. Clearly labor is a major cost to restorationefforts of this type and should be considered.
We found that limited recruitment below L. maackiistands necessitated eradicating L. maackii and replacingit with native tree seedlings. This was successfully accom-plished, as was a thorough comparison of L. maackii era-dication methods and native seedling performance. Theend result was a successful restoration with an increase innative woody plant diversity, structure, and most likelyassociated larger-scale processes in the future.
The best situation would be to predict successfulinvaders before introduction and prevent their establishment;however, with few restrictions on introductions, invadersfrequently do enter communities and become problematic(Reichard & Hamilton 1997). It then becomes the task ofland managers and ecologists to gather information onpatterns of invasion, develop the most effective means ofcontrolling invasives, and at the same time, protect nativediversity (Byers et al. 2002). We believe that by combiningthe findings of our research with prior studies, this know-ledge base can be applied to bring about the successfulrestoration of invaded communities and improvement ofecosystem health.
Acknowledgments
We thankEricWoods, EricKroger,Harold Swiger, CarolynKeiffer, KimBrown, JamesDyer, IrwinUngar,MorganVis,Darrin Rubino, Don Miles, Jill Brown, John Klein, TomBorgman, Aswini Pai, Zachary Rinkes, Pete and KarissaCitro, and Justin and Liberty Walton for their ideas andcontributions to this project. Also we thank the U.S. andDepartment of Energy and FEMP for funding this project.
LITERATURE CITED
Barnes, B. V., D. R. Zak, S. R. Denton, and S. H. Spurr. 1998. Forest
ecology. 4th edition. Wiley Press, New York.
Bergerud, W. A. 1988. Dose�response models for stand thinning with the
‘Ezject’ herbicide injection system. Research note no. 102. British
Columbia Ministry of Forests and Lands, Victoria, British Columbia.
Braun, E. L. 1989. The woody plants of Ohio: trees, shrubs and woody
climbers, native, naturalized, and escaped. Ohio State University
Press, Columbus.
Brockway, D. G., K. W. Outcalt, and R. N. Wilkins. 1998. Restoring
longleaf pine wiregrass ecosystems: plant cover, diversity and bio-
mass following low-rate hexazinone application on Florida sandhills.
Forest Ecology and Management 103:159�175.Byers, J. E., S. Reichard, J. M. Randall, I. M. Parker, C. S. Smith, W. M.
Lonsdale, I. A. E. Atkinson, T. R. Seastedt, M. Williamson,
E. Chornesky, and D. Hayes. 2002. Directing research to reduce the
impacts of nonindigenous species. Conservation Biology 16:630�640.Cairns Jr, J. 1986. Restoration, reclamation, and regeneration of degraded
or destroyed ecosystems. Pages 465�484 in M. E. Soule, editor.
Conservation Biology. Sinauer Publishers, Ann Arbor, Michigan.
Canham, C. D. 1988. An index for understory light levels in and around
canopy gaps. Ecology 69:1634�1638.Chippendale, J. F. 1991. Potential returns to research on rubber vine
(Cryptostegia grandiflora). Master’s thesis. University of Queens-
land, Brisbane, Australia.
Cicchetti, D. V. 1972. Extension of multiple-range tests to interaction
tables in the analysis of variance: a rapid approximate solution.
Psychological Bulletin 77:405�408.Collier, M. H., J. L. Vankat, and M. R. Hughes. 2002. Diminished plant
richness and abundance below Lonicera maackii, an invasive shrub.
American Midland Naturalist 147:60�71.Conover, D. G., and D. R. Geiger. 1993. Glyphosate controls Amur
honeysuckle in native woodland restoration (Ohio). Restoration
and Management Notes 11:168�169.Crowder, M. J., and D. J. Hand. 1990. Analysis of repeated measures.
Chapman and Hall, London, United Kingdom.
D’Agostino, R. B., A. Balanger, and R. B. D’Agostino Jr. 1990. A
suggestion for using powerful and informative tests of normality.
American Statistician 44:316�321.Deering, R. H., and J. L. Vankat. 1999. Forest colonization and develop-
mental growth of the invasive shrub Lonicera maackii. American
Midland Naturalist 141:43�50.Dodson, B. 1994. Weibull analysis. ASQC Quality Press, Milwaukee,
Wisconsin.
Dowdy, S., and S. Wearden. 1991. Statistics for research. 2nd edition. John
Wiley and Sons, New York.
Franz, C. M., and C. H. Keiffer. 2000. Effectiveness of the EZJECT
capsule injection system against the invasive shrub, Amur honey-
suckle. The Ohio Woodland Journal 7:19�20.Gayek, A. 2000. Effect of habitat variation on rate and success of colon-
ization of two invasive shrubs, Lonicera maackii and Ligustrum
vulgare, in a forested glen. Master’s thesis. Department of Horticulture
and Crop Science, Ohio State University, Columbus.
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 163
Ghersa, C. M., E. de la Fuenta, S. Suarez, and R. J. C. Leon. 2002. Woody
species invasion in the Rolling Pampa grasslands, Argentina. Agri-
culture, Ecosystems, and Environment 88:271�278.Gorchov, D. L., and D. E. Trisel. 2003. Competitive effects of the invasive
shrub, Lonicera maackii (Rupr.) Herder (Caprifoliaceae) on the
growth and survival of native tree seedlings. Plant Ecology
166:13�24.Gould, A. M. A., and D. L. Gorchov. 2000. Effects of the exotic
invasive shrub Lonicera maackii on the survival and fecundity of
three species of native annuals. American Midland Naturalist
144:36�50.Harrington, C. A. 1999. Forests planted for ecosystem restoration or
conservation. New Forests 17:175�190.Hintze, J. L. 2000. NCSS user’s guide. Number Cruncher Statistical
Systems. NCSS user’s guide. Kaysville, Utah.
Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, and
invasion: implications for conservation. Conservation Biology
6:324�337.Holl, K. D., and R. B. Howarth. 2000. Paying for restoration. Restoration
Ecology 8:260�267.Holmes, P. M. 2001. Shrubland restoration following woody alien invasion
and mining: effects of topsoil depth, seed source, and fertilizer
addition. Restoration Ecology 9:71�83.Hutchinson, T. F., and J. L. Vankat. 1997. Invasibility and effects of Amur
honeysuckle in southwestern Ohio forests. Conservation Biology
11:1117�1124.Johansson, T. 1985. Herbicide injections into stumps of aspen and birch to
prevent regrowth. Weed Research 25:39�45.Lerch, N. K., W. F. Hale, and D. D. Lemaster. 1980. Soil survey of Butler
County, Ohio. United States Department of Agriculture, Soil Con-
servation Service, Washington, D.C.
Luken, J. O. 1988. Population structure and biomass allocation of the
naturalized shrub Lonicera maackii (Rupr.) Maxim. in forest and
open habitats. American Midland Naturalist 119:258�267.Luken, J. O. 1990. Forest and pasture communities respond differently to
cutting of exotic Amur honeysuckle (Kentucky). Restoration and
Management Notes 8:122�123.Luken, J. O., and J. W. Thieret. 1995. Amur honeysuckle (Lonicera
maackii; Caprifoliaceae): its ascent, decline, and fall. Sida
16:479�503.Luken, J. O., and J. W. Thieret. 1996. Amur honeysuckle: its fall from
grace. Bioscience 46:18�24.Luken, J. O., L. M. Kuddes, and T. C. Tholemeier. 1997. Response of
understory species to gap formation and soil disturbance in Lonicera
maackii thickets. Restoration Ecology 5:229�235.MacDonald, A. W. 1993. The role of seedbank and sown seeds in the
restoration of an English flood-meadow. Journal of Vegetation
Science 4:395�400.MacMahon, J. A., and K. D. Holl. 2001. Ecological restoration: a key to
conservation biology’s future. Pages 245�269 in M. E. Soule, and
G. H. Orians, editors. Conservation biology: research priorities for
the next decade. Island Press, Washington, D.C.
McCarthy, B. C. 1997. Lab protocols for the testing of eastern deciduous
forest soils (available from: http://www.plantbio.ohiou.edu/epb/soils/
soils.htm).
McCarthy, B. C. 1999. Identification and control of invasive species.
Annual Report to Fernald Environmental Management Project,
Cincinnati, Ohio.
Montalvo, A. M., P. A. McMillan, and E. B. Allen. 2002. The relative
importance of seeding method, soil ripping, and soil variables on
seeding success. Restoration Ecology 10:52�67.National Climate Data Center and National Oceanic and Atmospheric
Administration. 2002. CLIMVIS information: selection criteria for
displaying period of record (available from: http://www5.ncdc.noaa.
gov/cgi-bin/climatenormals/climatenormals.pl).
Olson, R. A., and T. D. Whitson. 2002. Restoring structure in late-
successional sagebrush communities by thinning with tebuthiuron.
Restoration Ecology 10:146�155.Pyke, D. A., and J. N. Thompson. 1986. Statistical analysis of survival and
removal rate experiments. Ecology 67:240�245.Reichard, S. H., and C. W. Hamilton. 1997. Predicting invasions of woody
plants introduced into North America. Conservation Biology
11:193�203.Reifsnyder, W. E., and H. W. Lull. 1965. Radiant energy in relation to
forests. United States Department of Agriculture, Forest Service
Technical Bulletin, No. 1344.
Reinartz, J. A. 1997. Controlling glossy buckthorn (Rhamnus frangula L.)
with winter herbicide treatments of cut stumps. Natural Areas Jour-
nal 17:38�41.Robinson, G. R., and S. N. Handel. 2000. Directing spatial patterns of
recruitment during an experimental urban woodland reclamation.
Ecological Applications 10:174�188.Robison, S., and B. C. McCarthy. 1999. Potential factors affecting the
estimation of light availability using hemispherical photography in
oak forest understories. Journal of the Torrey Botanical Society
126:344�349.Rooney, T. P. 2001. Deer impacts on forest ecosystems: a North American
perspective. Forestry 74:201�208.SAS Institute. 2001. SAS/STAT User’s Guide, Version 5.0. SAS Institute,
Cary, North Carolina.
Scheiner, S. M. 1993. MANOVA: multiple response variables and
multispecies interactions. Pages 94�112 in S. M. Scheiner, and
J. Gurevitch, editors. Design and analysis of ecological experiments.
Chapman and Hall, New York.
Schierenbeck, K. A., R. N. Mack, and R. R. Sharitz. 1994. Effects of
herbivory on growth and biomass allocation in native and introduced
species of Lonicera. Ecology 75:1661�1672.Shea, K., and P. Chesson. 2002. Community ecology theory as a frame-
work for biological invasions. Trends in Ecology and Evolution
17:170�176.Sinclair, A. R. E., D. S. Hik, O. J. Schmitz, G. G. E. Scudder,
D. H. Turpin, and N. C. Larter. 1995. Biodiversity and the need
for habitat renewal. Ecological Applications 5:579�587.Sipe, T. W., and F. A. Bazzaz. 2001. Shoot damage effects on regeneration
of maples (Acer) across an understory-gap microenvironmental gra-
dient. Journal of Ecology 89:761�773.Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and practice of
statistics in biological research. 3rd edition.W.H. Freeman, NewYork.
Sweeney B. W., S. J. Czapka, and T. Yerkes. 2002. Riparian forest
restoration: increasing success by reducing plant competition and
herbivory. Restoration Ecology 10:392�400.Trisel, D. E. 1997. The invasive shrub, Lonicera maackii (Rupr.) Herder
(Caprifoliaceae) factors leading to its success and its effect on native
species. PhD Dissertation. Department of Botany, Miami University,
Oxford, Ohio.
U.S. Department of Energy. 2002. Site Environmental Report. Fernald
Environmental Management Project. U.S. Department of Energy,
Fernald Area Office, Cincinnati, Ohio.
Van Der Meer, P. J., P. Dignan, and A. G. Saveneh. 1999. Effect of gap
size on seedling establishment, growth and survival at three years in
mountain ash (Eucalyptus regnans F. Muell.) forest in Victoria,
Australia. Forest Ecology and Management 117:33�42.Veenendaal, E. M., M. D. Swaine, V. K. Agyeman, D. Blay, I. K. Abebrese,
and C. E. Mullins. 1996. Differences in plant and soil water relations
in and around a forest gap in West Africa during the dry season may
influence seedling establishment and survival. Journal of Ecology
84:83�90.Vitousek, P. M. 1990. Biological invasions and ecosystem processes:
toward an integration of population biology and ecosystem pro-
cesses. Oikos 57:7�13.
Forest Recovery After Shrub Removal
164 Restoration Ecology JUNE 2004
Von Ende, C. N. 1993. Repeated-measures analysis: growth and other
time-dependent measures. Pages 113�137 in S. M. Scheiner, and
J. Gurevitch, editors. Design and analysis of ecological experiments.
Chapman and Hall, New York.
Walters, M. B., and P. B. Reich. 2000. Seed size, nitrogen supply, and
growth rate affect tree seedling survival in deep shade. Ecology
81:1887�1901.
Wiens, J. A., N. C. Stenseth, B. Van Horne, and R. A. Ims. 1993. Ecological
mechanisms and landscape ecology. Oikos 66:369�380.Woods, K. D. 1993. Effects of invasion by Lonicera tatarica L. on herbs
and trees in four New England forests. American Midland Naturalist
130:62�74.Zavaleta, E. 2000. The economic value of controlling an invasive shrub.
Ambio 29:462�467.
Forest Recovery After Shrub Removal
JUNE 2004 Restoration Ecology 165