Forest Ecology and Management 303 (2013) 148–159

Contents lists available at SciVerse ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier .com/ locate/ foreco

The influence of ground disturbance and gap position on understoryplant diversity in upland forests of southern New England

0378-1127/$ - see front matter � 2013 Published by Elsevier B.V.http://dx.doi.org/10.1016/j.foreco.2013.04.018

⇑ Corresponding author. Address: Yale University School of Forestry and Envi-ronmental Studies, Marsh Hall, 360 Prospect St., New Haven, CT 06511, UnitedStates. Tel.: +1 203 650 9118.

E-mail address: Marlyse.duguid@yale.edu (M.C. Duguid).

Marlyse C. Duguid a,⇑, Brent R. Frey b, David S. Ellum c, Matthew Kelty d, Mark S. Ashton a

a Yale School of Forestry & Environmental Studies, New Haven, CT, United Statesb Department of Forestry, Mississippi State University, Mississippi State, MS, United Statesc Environmental Studies Department, Warren Wilson College, Asheville, NC, United Statesd Department of Natural Resources Conservation, University of Massachusetts, Amherst, MA, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 6 December 2012Received in revised form 8 April 2013Accepted 15 April 2013

Keywords:EvennessForest harvestingHerbaceousMicrositeRichnessSoils

The forest understory contains the majority of vascular plant diversity in eastern temperate forests, andits diversity, composition, and dynamics contribute directly to ecosystem function. Forest managers havetraditionally viewed the understory as primarily affecting forest regeneration or wildlife habitat, but thegrowing recognition of goods and services the understory provides (e.g., ecosystem function, ecologicalresiliency, non-timber forest products) has increased concerns about the impacts of forest managementon understory diversity. We monitored response of understory diversity to microsite position and degreeof ground-level disturbance within experimental gaps for 10 years. We did this at four sites with distinctsoil types and topographic positions of a glacial geology in southern New England that were categorizedas (i) mesic, (ii) mid-slope, (iii) outwash, and (iv) sandy-skeletal. We analyzed differences in patterns ofspecies richness, Shannon diversity, and evenness across sites and through time. Understory species rich-ness was generally enhanced by gap formation. Gap position was the primary factor influencing speciesrichness across all sites, but the patterns of diversity and evenness within gaps was site specific.Ground-disturbance was influential on drier sandy sites, and more pronounced earlier in the experiment.Temporal differences were also evident across sites, with richness stabilizing at all sites 10 years after gapcreation. The one exception was the sandy-skeletal site, which was still increasing in richness. Resourcemanagers interested in protecting and enhancing understory species diversity need to consider underly-ing site, specifically soil type when planning silvicultural treatments, as the response of the understorycommunity to disturbance can vary greatly with site.

� 2013 Published by Elsevier B.V.

1. Introduction

The importance of maintaining biodiversity has been widelyrecognized at both national and international levels (U.S., 2000;Brooks et al., 2006; UNEP, 2010). Plant diversity is a fundamentalcomponent of ecosystem diversity, contributing to both habitatstructure and ecosystem function (Srivastava and Vellend, 2005).In eastern deciduous forests, the majority of the vascular plant spe-cies diversity is found in the herbaceous layer (Whigham, 2004).Diversity within the herbaceous layer increases structural com-plexity, which has a beneficial effect on compositional diversityof many insects, small mammals, birds, amphibians and reptiles(Ricketts, 1999; Dauber et al., 2003). Indeed, studies have demon-strated that the richness of birds, butterflies, and certain mammalsis better correlated with understory rather than overstory richness

(Ricketts, 1999). The herbaceous layer plays an important role inecosystem function, contributing organic matter, aiding in decom-position, and conserving nutrients (Muller and Bormann, 1976;Peterson and Rolfe, 1982; Zak et al., 1990; Roberts and Gilliam,1995; Muller, 2003; Falk et al., 2008).

The structure of the forest understory has direct implications forforest succession and management. Herbaceous layer competitioninfluences germination, establishment, and thus, spatial arrange-ment of regenerating tree species (Maguire and Forman, 1983;Berkowitz et al., 1995; George and Bazzaz, 2003). Structural charac-teristics, such as density of the forest understory, can determineforest regeneration processes. For example, vigorous monodomi-nant clonal understories, such as hayscented fern (Dennstaedtiapunctilobula Michx.) can severely inhibit regeneration (Beckageet al., 2000; De La Cretaz and Kelty, 2002). In managed forestsbiodiversity can increase economic and ecological resiliency,productivity, and community stability (Burton et al., 1992). In manyregions, understory species provide opportunities for alternativerevenue streams through non-timber forest products (NTFPs)(Hammett and Chamberlain, 1998). For all of these reasons forest

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 149

management should strive to preserve and, where possible, enhanceunderstory plant diversity (Roberts and Gilliam, 1995).

Species diversity is driven by disturbance, forest cover type, andsite history (Bormann and Likens, 1979; Whitney and Foster, 1988;Singleton et al., 2001; Bellemare et al., 2002; Ellum et al., 2010).Forest harvesting can significantly alter edaphic and microclimateconditions (through increased light, soil moisture and nutrientavailability), and in turn shape the diversity and composition ofthe herbaceous layer (Bhatti et al., 2000; Gilliam, 2002; Robertsand Gilliam, 2003; Zenner et al., 2006). Much depends on intensityof canopy removal (amount of basal area removed or gap size) anddegree of ground disturbance. The amount of canopy removal mayalter understory diversity and composition (Reader and Bricker,1992; Battles et al., 2001; Jackson et al., 2006), or may have limitedimpacts (Hughes and Fahey, 1991; Ruben et al., 1999; Schumannet al., 2003; Kern et al., 2006). While it is apparent that ground dis-turbance can substantially alter understory composition (Armestoand Pickett, 1985; Peltzer et al., 2000; Roberts and Zhu, 2002; Freyet al., 2003; Aikens et al., 2007), the extent of its influence on diver-sity in respect to increased light availability is not well understood.

There has been significant work in temperate forests examiningunderstory response to clearcuts with inconsistent results (see re-views by Roberts and Gilliam (2003) and Moola and Vasseur(2008)). Many studies examining group selection treatments havefound increases in understory diversity (Jenkins and Parker, 1999;Falk et al., 2008), although the temporal component must also beconsidered. Additionally, the mechanisms driving these patternsmay shift with succession (Gilliam et al., 1995), and it is unclearwhat the permanent effects of forest harvesting are on understorydiversity (Duffy and Meier, 1992; Meier et al., 1995). Many studiesonly look at a moment in successional time, but examining longertime frames is necessary to isolate treatment effects (Falk et al.,2008). Further, understory cover and compositional change maybe more influenced by gap dynamics than diversity (Moore andVankat, 1986). Few studies have been able to carefully assess theimpacts of ground-disturbance intensity and gap position, usingboth pre- and post-disturbance data, controlled and comparedacross varying soil types over successional time.

To truly understand the biological diversity of a communityspecies abundance measures must be incorporated; evennessmay represent a different suite of ecological functions than speciesrichness (Magurran, 2004). Our objective was to examine patternsin understory plant species diversity in response to microsite posi-tion and two levels of ground-disturbance intensity. The distur-bance levels included ‘‘lethal’’ treatments (all vegetation removedand mineral soil exposed), and ‘‘release’’ (only the overstory re-moved). We conducted the study at four distinct sites in southernNew England (different soils and canopy compositions) to examinewhether patterns are similar across the region’s common foresttypes. We examined data over a multi-year time period to capturesuccessional changes in diversity. We hypothesize that both

Table 1Summary of the four study sites in southern New England.

Site Location Coordinates Elevation(m)

Soil series

Mesic Yale-Myers Forest,Eastford, CT

41�560N,72�070W

200 Charlton and Leice

Mid-slope Yale-Myers Forest,Eastford, CT

41�570W,72�070W

265 Brookfield/BrimfiePaxton/Montauk

Sandy-skeletal Cadwell Forest,Pelham, MA

42�220N,72�240W

325 Gloucester

Outwash Adam’s Brook,Amherst, MA

42�230N,72�290W

90 Merrimac

a Total pre-harvest understory plant species richness sampled values are followed by

ground layer disturbance and gap position influence understorydiversity, but expect gap position will be more influential. We pre-dict that diversity will be highest in gap positions that offer thegreatest levels of resources (light) and the least competition (fromedge trees), and that these trends will be parallel across the foursoil types studied. Across all sites we predict diversity to increaseinitially before stabilizing and then eventually decrease with re-source limitations due to canopy development and shading effects.

2. Methods

2.1. Study sites

We conducted this study at four sites in southern New Englandselected to represent the variety of soils found in the region. Threeare glacial till soils and one is of glacial–fluvial origin (Table 1). Theclimate throughout the region is cool-temperate and humid;approximately 110 cm of precipitation is evenly distributedthroughout the year.

The first two sites are located at Yale-Myers Forest (41�560N,72�70W), a 3213-hectare research and demonstration forest innortheastern, Connecticut. The forest belongs to the region classi-fied as Central Hardwood–Hemlock–Pine (Westveld, 1956). Theforest consists primarily of mixed-deciduous second-growthdeveloping on abandoned agricultural land from the mid 19th cen-tury (Meyer and Plusnin, 1945). The topography is ridge-valleywith an elevation range between 170 m and 300 m above sea level.The soils are glacial tills composed of moderate to well-drainedstony loams overlying bedrock. Average temperatures at Yale-Myers Forest in July and January are 21.2 �C and �4.1 �C, respec-tively (Ashton and Larson, 1996; McKenna, 2007).

The first site, labeled ‘‘Mesic’’, has a gentle slope (<10% slope)with an easterly aspect. Soils are well-drained coarse-loamy tills.The canopy is composed primarily of Quercus rubra L., with compo-nents of Acer rubrum L., Acer saccharum Marsh., Betula alleghaniensisBritton., Betula lenta L., Betula papyrifera Marsh., Carya ovata (Mill.) K.Koch, Fraxinus americana L., Liriodendron tulipifera L., and Tsugacanadensis L. The midstory–woody species that will never grow intocanopy trees – includes Carpinus caroliniana Walter, and Hamamelisvirginiana L. The understory is fairly diverse, with Carex spp., Aralianudicaulis L., and a variety of ferns (Thelypteris noveboracensis (L.)Nieuwl., Polystichum acrostichoides (Michx.) Schott, D. punctilobula,Athyrium filix-femina (L.) Roth) as the dominant species. The mesicsite had the highest levels of pre-treatment understory richness.

The second site, labeled ‘‘Mid-slope’’, has a slight northwest as-pect (<10% slope). Soils are well-drained coarse-loamy till withcoarse unsorted rocks of varying sizes. Overstory composition isprimarily Q. rubra, the midstory consists of A.rubrum, and B. lentawith an intermittent shrub layer of Kalmia latifolia L. The dominantherbaceous understory species are Carex spp., D. punctilobula, andTrientalis borealis Raf.

Drainage class Richnessa Canopy composition

ster Well-drained to poorlydrained

39 (49) Mixed mesic hardwoods

ld and Well drained to excessivelywell-drained

14 (20) Red oak, with mixedhardwoods and pine

Excessively well-drained,heterogenous

14 (14) Upland oak with pine

Excessively well-drained 16 (22) White pine with oak

a first-order jackknife estimator in parentheses.

Fig. 1. Trends in total site understory plant species richness over time as estimatedwith a first-order jackknife estimate. 1999 Indicates pre-treatment levels, gapcreation in 2000.

150 M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159

The third site, labeled ‘‘Sandy-skeletal’’, is located in Pelham,Massachusetts in the Cadwell Memorial Forest (42�220N,72�240W). Soils are well-drained ablation till, heterogeneous withisolated perched wetlands. Elevation is approximately 325 m, thehighest of the four sites. Overstory composition is primarily mixedupland oak (Quercus alba L., Quercus coccinea Muench., Q. rubra,Quercus velutina Lam.), with minor components of A. rubrum andPinus strobus (L.) Small. Dominant understory species include Gaul-theria procumbens L., D. punctilobula, A. nudicaulis, Maianthemumcanadense Desf., Medeola virginiana L., and Uvularia sessilifolia L.Prior to gap creation the sandy-skeletal site had the lowest under-story species richness (Table 1).

The fourth site, labeled ‘‘Outwash’’, is located in the AdamsBrook Forest in Amherst, Massachusetts (42�230N, 72�290W) on akame terrace. Soils are deep, excessively well-drained, sandy out-wash. With an elevation of approximately 90 m, the outwash siteis the lowest of the four sites. The overstory consists primarily ofP. strobus. Common understory plants include Chimaphila maculata,Lycopodium spp., M. canadense, and Carex spp. Average tempera-tures at the Massachusetts sites in July and January are 22.1 �Cand �4.6 �C, respectively (McKenna, 2007).

2.2. Experimental design

At each site, during the summer of 1999 we delineated a rectan-gular area 108 m � 30 m running east–west to serve as the exper-imental gap. The experimental gaps were harvested the followingwinter (1999–2000), removing all vegetation greater than 2 m inheight. We designed the gap with the intention to create a lightgradient including understory conditions, side shade, and directsun. We left a 15 m buffer at both the eastern and western endsof the gap with no plots to mitigate east–west edge effects. Withinthe gap there were four parallel east–west environmental zones–north edge, north center, south center, and south edge. In addition,there were two zones located 5 m from the northern and southerngap edges in the adjoining forest understory parallel to the centerand edge zones. Each of the six zones had four blocks of three con-tiguous rectangular plots 4 m � 6.5 m. Our study included twotreatment plots (lethal and release) randomly assigned within eachblock, totaling 48 plots per site. The third plot was part of anotherstudy examining planted tree seedling response and not includedin this study (McKenna, 2007). Measurement plots 2 m � 4 m werecentrally located within each treatment plot to limit edge effects.We laid out plots in 1999 and took initial measurements prior tocanopy removal and site treatments.

All sites were harvested using a Timbco feller-buncher (TimbcoHydraulics, Shawano, WI). To minimize ground disturbancemachinery traveled between the strips. As much material was car-ried off site as possible; remaining slash was arranged between thetwo center zones of the gap avoiding measurement plots. We trea-ted the vegetation in the lethal plots with glyphosate herbicide(Roundup�, Monsanto Corporation, St. Louis, MO) in June 2000,and four weeks later followed up by scarifying the soil surface.Scarification was achieved by mixing the organic pad with mineralsoil to 10 cm in depth with root rakes. The release plots received noadditional ground-level treatment. We installed a 60 high electricfence for deer exclusion around the perimeter of the study site,and inspected and repaired every spring for the first 6 years ofthe study. Qualitative observations revealed no noticeable differ-ence in browse between fenced and unfenced areas, and in thesummer of 2006 the fencing was removed.

2.3. Field sampling

We sampled each plot in the summer of 1999 before the har-vest, and then subsequently once each summer in 2002, 2004,

2006 and 2010. Additionally, we visited the sites in early springto assure no spring ephemeral species were missed during summersampling. We identified herbaceous and shrubby plants to specieslevel for the majority of specimens. We used congeners for grasses,sedges, and mosses – and in rare cases used morpho-species. Whenwe did use morpho-species we were confident that they were dif-ferent than all other species sampled. We used USDA–NRCS fornomenclature (USDA, 2013), and recorded percent cover usingthe six point scale described by Daubenmire (1959).

2.4. Data analysis

The four sites in this study are not true replicates, therefore weindividually analyzed each site as a case study. To examine diver-sity response we performed two-way analysis of variance (ANOVA)models for species richness (total count of species per plot), Shan-non diversity (H0 = �

PRi¼1pi log pi), and evenness (J0 = H0/ln S). Our

variables were ground-disturbance, gap position, and distur-bance � gap position. Individual models were run for each site, ineach post-harvest year (2002, 2004, 2006, and 2010). We carriedout Tukey HSD post hoc comparisons on gap position to determinewhich positions were different from others. We assumed normalitywhen / > .01 in Shapiro–Wilk normality tests. We ran non-para-metric Kruskal–Wallis tests on both richness and Shannondiversity for each variable on pre-harvest (1999) data to look forpre-existing trends. We estimated total site richness and comparedtrends across sites using first-order jackknife richness estimators.We also examined species abundance curves based on bootstrap-ping with 100 permutations (Magurran, 2004). To visualizechanges in richness and evenness we constructed rank abundancecurves for each site by year. All statistics were carried out using R2.11.1 (R Development Core Team, 2010) with the additional pack-ages Vegan and BiodiversityR (Kindt and Coe, 2005; Oksanen et al.,2010).

3. Results

3.1. Richness: across sites and time

In the mesic site we identified 39 species prior to gap creation,which increased after harvesting to 79 in 2002, 74 in 2004, and 86in 2006. The highest richness observed was at 89 species in 2010, a128% increase over pre-harvest levels. Throughout the study a total

Fig. 2. Species accumulation curves showing extrapolated number of total understory plant richness based on bootstrapping with 100 permutations (±2SD) for each site.Initial levels (1999) are shown in gray, 10 years after gap creation (2010) is shown in black.

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 151

of 129 species (including unidentified specimens) were observed.Percent cover of the majority of species was low, with the excep-tion of Carex pennsylvanica Lam., ferns (e.g. Dennstaedtia punctilo-bula; T. noveboracensis), and Rubus spp. The only species presentin 1999 that failed to appear again were Impatiens capensis Meerb.,and Smilax rotundifolia L., of these only S. rotundifolia was presentat more than two plots initially.

At the mid-slope site there were 14 species present before gapcreation, which increased after harvesting to 36 in 2002, peaking in2004 with 38 species (171% increase), then declining to 37 speciesin 2006, and 33 in 2010. Overall, 62 different species were identi-fied at this site throughout the study. Amphicarpaea bracteata (L.)Fernald, C. maculata (L.) Pursh, G. procumbens, and P. acrostichoideswere present only during the pre-harvest sampling, and only inone or two plots.

At the sandy-skeletal site we identified 14 species prior to gapcreation, which increased to 25 in 2002, 28 in 2004, and 26 in2006. Greatest richness was observed in 2010 with 44 species, a214% increase over pre-harvest richness. Forty-eight species wereobserved over the study and no species that were initially presentdisappeared completely after the disturbance.

Finally, at the outwash site, 16 species were present initially,increasing to 37 in 2002, and to 42 in both 2004 and 2006. Again,greatest richness was observed in 2010 with 44 species, a 175% in-crease over pre-harvest levels. In total 62 species were observedduring the course of the study. The only species present pre-har-vest to disappear from the site was Spiraea alba Du Roi.

Total site richness via jackknife estimation exhibited differencesin successional trends in richness amongst sites. At the mesic andmid-slope sites species richness peaks in 2006 and begins to levelthen decrease. At the two drier sites (sandy-skeletal and outwash)richness continued to increase (Fig. 1). Species accumulationcurves show differences in slope and magnitude between pre-and post-treatment richness across sites, with trends are similar

to the jackknife estimates. The two drier sites have flatter slopesin 1999, but become steeper than the moister till sites by 2010(Fig. 2).

3.2. Richness: gap position and disturbance

With one exception (sandy-skeletal, 2002), gap position was asignificant predictor of species richness for all sites in all post-har-vests years (Table 2). In the mesic, sandy-skeletal and outwashsites the general trend in all post-harvest years was greatest rich-ness in the southern edge and southern center gap positions, andlowest in the understory positions (Fig. 3 & Table 3). The only sig-nificant pre-disturbance trend was in the sandy-skeletal site,where richness increased toward the northern gap positions,which is opposite of the post-harvest pattern at that site. Themid-slope site showed opposite results of all other sites, withgreatest diversity values in the center and northern locations.

Intensity of ground disturbance was only significant on the san-dy-skeletal site initially after harvest (2002), but on the outwashsite lethal ground disturbance produced higher species richnessin 2004, 2006, and 2010 (Table 2). On the two drier, sandy siteslethal treatments had higher species richness initially; no differ-ence by disturbance treatment was observed for the mesic andmid-story sites. Only the outwash site showed consistently higherrichness on the lethal treatments throughout the study. The onlyinteraction effect was on the mid-slope site in 2006.

3.3. Shannon diversity

The effect on Shannon diversity was less consistent across sitesthan species richness. On the mesic site disturbance type wasimportant initially, and then the influence of gap position becamemore important – first as an interaction in 2004, then gap positiononly for 2006 and 2010. The mid-slope site was opposite with gap

Table 2Significant ANOVA results for species richness, Shannon diversity, and evenness.

Year Factora DfN DfD Sum sq. Mean sq. F Pr(>F)

MesicSpecies richness2002 GAP 5 15 384.00 76.90 8.65 0.002004 GAP 5 15 350 69.90 4.04 0.022006 GAP 5 15 589.00 117.70 8.03 0.002010 GAP 5 15 477.35 95.47 6.61 0.00

Shannon diversity2002 DIST 1 18 1.31 1.305 5.6 0.032004 GAP � DIST 5 18 2.45 0.49 4.23 0.012006 GAP 5 15 2.37 0.48 4.18 0.012010 GAP 5 15 3.25 0.65 10.34 0.00

J evenness2002 DIST 1 18 0.186 0.1864 11.7 0.002004 GAP � DIST 5 18 0.2889 0.06 6.59 0.002006 GAP � DIST 5 18 0.1326 0.02651 3.96 0.012010 GAP 5 15 0.10 0.02 4.43 0.01

Mid-slopeSpecies richness2002 GAP 5 15 118.7 23.74 4.3 0.012004 GAP 5 15 86.9 17.38 5.05 0.012006 GAP 5 15 56.9 11.37 6.54 0.00

GAP � DIST 5 18 38.9 7.77 2.82 0.052010 GAP 5 15 95.6 19.1 3.98 0.02

Shannon diversity2002 GAP 5 15 3.43 0.69 4.12 0.022004 DIST 1 18 0.568 0.57 4.68 0.042006 – – – – – – –2010 – – – – – –

J evenness2002 GAP 5 15 0.98 0.2 6.75 0.00

DIST 1 18 0.25 0.25 6.9 0.022004 GAP 5 15 0.6 0.12 3.58 0.032006 – – – – – – –2010 – – – – – – –

Sandy-skeletalSpecies richness2002 DIST 1 18 133.3 133.3 31.79 0.002004 GAP 5 15 89.6 17.92 5.04 0.012006 GAP 5 15 74.1 14.82 3.59 0.032010 GAP 5 15 94.8 18.95 6.15 0.00

Shannon diversity2002 – – – – – – –2004 – – – – – – –2006 – – – – – – –2010 GAP 5 15 0.75 0.15 3.67 0.02

J evenness2002 GAP 5 15 0.28 0.06 4.86 0.01

DIST 1 18 0.18 0.18 16.96 0.002004 GAP 5 15 0.23 0.05 8.74 0.00

DIST 1 18 0.02 0.02 6.38 0.02GAP � DIST 5 18 0.06 0.01 3.28 0.03

2006 GAP 5 15 0.17 0.03 8.08 0.00GAP � DIST 5 18 0.13 0.03 5.39 0.00

2010 GAP � DIST 5 18 0.05 0.01 5.14 0.00

OutwashSpecies richness2002 GAP 5 15 169.4 33.9 9.57 0.002004 GAP 5 15 227.4 45.5 14.7 0.00

DIST 1 18 20 20.02 6.28 0.022006 GAP 5 15 202 40.4 7.39 0.00

DIST 1 18 27 27 4.42 0.052010 GAP 5 15 190 38 3.07 0.04

DIST 1 18 50 50 8.9 0.01

Shannon diversity2002 GAP 5 15 3.41 0.68 3.66 0.02

DIST 1 18 1.4 1.4 6.26 0.022004 GAP 5 15 3.86 0.77 14.2 0.002006 GAP 5 15 2.41 0.48 3.49 0.03

152 M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159

Table 2 (continued)

Year Factora DfN DfD Sum sq. Mean sq. F Pr(>F)

2010 – – – – – –

J evenness2002 GAP 5 15 1.85 0.37 18.1 0.00

DIST 1 18 0.59 0.59 10.27 0.002004 GAP 5 15 0.47 0.09 5.63 0.002006 GAP 5 15 0.84 0.17 11.4 0.002010 – – – – – – –

a Significant factors in the model, GAP indicates microsite position; DIST indicates intensity of ground disturbance.

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 153

position significant in 2002 followed by disturbance type in 2004(Table 2). The sandy-skeletal site showed gap position significantonly in 2010 with no disturbance effects. The outwash site wassimilar to the mesic site, with disturbance type significant onlyimmediately post-harvest (2002), and gap position significant in2002, 2004, and 2006 (Table 2). Across all sites the general trendswere similar for Shannon diversity as for species richness, but thedifferences were less pronounced or absent.

3.4. Species evenness

On the mesic site evenness followed the same pattern as Shan-non diversity, with disturbance initially important, followed by itsinteraction with gap position, and then eventually only gap posi-tion (Figs. 4 and 5). Additionally, gap position was significant innearly all post-harvest years among the other sites. Decreasingby 2006 at the mid-slope site, and then across the other sites by2010. Evenness was highest in the shadiest gap positions (northernunderstory and southern understory) and lowest in the sunniestpositions (north central and northern edge) across all sites. Distur-bance was significant immediately post-harvest (2002) across allsites, only on the sandy-skeletal site was it significant beyond2004. Interaction effects were evident on the sandy-skeletal sitein 2004, 2006, and 2010 (Table 2).

Fig. 3. Mean understory species richness by gap position for all sites in 2006. Gap positnorthern center, NE – northern edge, and NU – northern understory. Error bars repressignificantly different (Tukey HSD). 2006 demonstrates the pattern seen across all post-

Rank abundance plots show different evenness responses todisturbance type across sites (Fig. 6). The mesic, mid-slope, andsandy-skeletal sites all initially had a few species with higherabundance than the majority of species present, while the outwashsite exhibited a higher level of initial evenness. The lines divergepost-harvest for all sites, but the magnitude and effect is differentacross sites.

On the mesic site the two species with the highest ranking –Carex spp. and New York fern (T. noveboracensis) had maintainedor regained their dominance by 2010. Interestingly, Carex spp.had lower abundance in the final sampling, while the majority ofspecies experienced higher abundance than pre-harvest. Themid-slope site had less difference between pre and post-harvestcurves. Similar to the mesic site, the dominant species, hay-scentedfern (D. punctilobula), maintained its dominance, although moun-tain laurel (K. latifolia) increased in dominance to ranked second-supplanting starflower (T. borealis) and Carex spp. The pre- andpost-harvest curves converge around the 20th ranked species withlow abundances seen across the majority of species. The sandy-skeletal site had fewer species initially; therefore, the post-harvestcurve is longer and they never converge. A few species shareslightly higher abundances than on the mid-slope and outwashsites, wintergreen (G. procumbens) holds a prominent place on bothcurves (first in 1999, second in 2010). Interestingly, hay-scented

ions are SU – southern understory, SE – southern edge, SC – southern center, NC –ent one standard error. Within each site, columns sharing the same letter are notharvest years.

Table 3Mean (standard error) plot level species richness for each year by gap position and disturbance type. For each year, columns sharing the same letter are not significantly different(Tukey HSD).

1999a 2002 2004 2006 2010

MesicSU 6.63 (0.68) 12.5 (0.87) 13.75 (1.44) 13.88 (1.41) 13.25 (1.83)SE 7.5 (0.76) 19.5 (1.1) 21.63 (1.98) 24.13 (1.66) 20.25 (2.02)SC 6.38 (0.91) 18.13 (1.89) 19.25 (1.08) 22.88 (1.57) 22.5 (0.98)NC 6 (0.73) 15.63 (0.75) 18.88 (0.74) 19 (1.09) 17.38 (1.87)NE 7.63 (0.53) 13 (0.6) 15.25 (1.01) 17.75 (1.52) 15.63 (1.35)NU 6.63 (1.02) 12.25 (0.7) 15.88 (0.83) 16.88 (1.34) 15.13 (0.72)Lethal 6.33 (0.48) 15.33 (0.88) 18.08 (0.82) 19.75 (1.1) 18.38 (1.18)Release 7.25 (0.4) 15.04 (0.79) 16.79 (0.94) 18.42 (1.05) 16.38 (0.91)Total 6.79 (0.32) 15.19 (0.59) 17.44 (0.62) 19.08 (0.76) 17.38 (0.75)

Mid-slopeSU 2.88 (0.67) 4.25 (0.53) 5.5 (0.71) 6.5 (0.63) 5.38 (1.9)SE 2 (0.57) 6.13 (0.79) 7.75 (0.88) 8.13 (0.55) 7 (2.47)SC 2 (0.73) 9.25 (0.96) 9.13 (0.67) 9 (0.94) 5.75 (2.03)NC 3 (0.68) 7.38 (1) 9.38 (0.68) 8.25 (0.62) 5.75 (2.03)NE 3.38 (0.26) 8.13 (0.97) 8 (0.46) 9.88 (0.77) 8.5 (3.01)NU 4.13 (0.23) 7.25 (0.31) 9.25 (0.62) 9.38 (0.56) 9 (3.18)Lethal 2.63 (0.39) 7.04 (0.57) 8.58 (0.48) 8.88 (0.52) 6.67 (0.48)Release 3.17 (0.28) 7.08 (0.53) 7.75 (0.44) 8.17 (0.34) 7.13 (0.49)Total 2.9 (0.24) 7.06 (0.39) 8.17 (0.33) 8.52 (0.31) 6.9 (0.34)

Sandy-skeletalSU 5.25 (0.7) 7.38 (0.91) 8.88 (0.79) 9.5 (0.33) 8.5 (0.19)SE 6 (0.19) 10.38 (0.84) 12.38 (0.68) 13.25 (0.62) 13.13 (0.61)SC 6.88 (0.55) 10.25 (0.9) 12.38 (0.89) 10.88 (0.9) 10.38 (0.32)NC 6.63 (0.46) 7.75 (0.84) 12 (0.78) 12.75 (0.45) 11.5 (0.76)NE 7.38 (0.46) 8.38 (1.27) 10.88 (0.72) 12 (0.46) 11.38 (1.02)NU 7.13 (0.61) 8.13 (1.08) 9.63 (0.68) 11.25 (1.01) 11.38 (1.03)Lethal 6.71 (0.35) 7.04 (0.47) 10.67 (0.54) 11.38 (0.45) 12.33 (0.55)Release 6.38 (0.29) 10.38 (0.5) 11.38 (0.45) 11.83 (0.45) 11.38 (0.41)Total 6.54 (0.23) 8.71 (0.42) 11.02 (0.35) 11.6 (0.32) 11.85 (0.35)

OutwashSU 2.88 (0.67) 3.5 (0.5) 5.25 (0.73) 5.88 (0.72) 4.88 (0.91)SE 3.13 (0.23) 6.13 (1.08) 11.25 (0.75) 11.13 (0.67) 10.38 (0.65)SC 2.5 (0.33) 8.63 (1) 9.5 (0.93) 9.25 (1.11) 8.13 (1.22)NC 1.75 (0.31) 6.25 (0.67) 8.25 (0.98) 9.5 (1.07) 8.75 (1.54)NE 3 (0.5) 3.25 (0.67) 5.75 (0.92) 6.38 (0.94) 5.5 (0.78)NU 3.25 (0.41) 4.25 (0.88) 6.13 (1.04) 5.88 (1.04) 5.75 (1.35)Lethal 2.83 (0.25) 5.63 (0.68) 8.33 (0.63) 8.75 (0.72) 8.25 (0.67)Release 2.67 (0.27) 5.04 (0.5) 7.04 (0.69) 7.25 (0.58) 6.21 (0.76)Total 2.75 (0.18) 5.33 (0.42) 7.69 (0.47) 8 (0.47) 7.23 (0.52)

a Pre-harvest observations.

154 M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159

fern, which holds second place pre-harvest, does not rank in thetop three in 2010. The largest discrepancy in curves is on the out-wash site. The initial curve indicates an even community with themost abundant species, spotted wintergreen (C. maculata) and Can-ada mayflower (M. canadense), holding only a small dominanceover the other species when compared to the other sites. Post-har-vest trends show Carex spp. and moss with higher abundance thanany pre-harvest species. Subsequent species abundances decreaseslowly until it converges with the pre-harvest line around the20th ranked species, similar to the mid-slope site (Fig. 6).

4. Discussion

Our results suggest that underlying site conditions, specificallysoil type, and potentially aspect, affect the way understory plantdiversity will respond to ground disturbance and light availability.Gap formation alters the physical understory environment, creat-ing resource gradients; the asymmetry of light in temperate forestgaps interacts with soil moisture and nutrient availability to definethe floristic pattern of these gradients (McKenna, 2007). We foundgap position (light availability) as the strongest driver of post-dis-turbance species diversity regardless of soil type, although theunderlying patterns within these gaps are site specific. We did

not find other studies to corroborate our findings, suggesting thatmore research is needed isolating different disturbance parametersin forest harvesting.

We expected that maximum diversity would be in the positionthat minimized competition and maximized resources (Connell,1978). The south-center strip has high light availability and highavailable moisture and nutrients due to no competition from thesurrounding canopy. Although not all research corroborates this,North et al. (2005) found understory richness to be negativelycorrelated with direct light. At 41–42�N latitudes the northernedge is the brightest, driest, and therefore harshest environment(Oliver and Larson, 1996; Gray et al., 2002; McKenna, 2007), thuswe expected lower diversity in that position. Interestingly, thetwo moister till sites had opposite patterns of diversity, andneither peaked in center positions. The mesic site had highestdiversity along the southern edge, and the mid-slope along thenorth edge and understory. Depending on gap size and latitudelight may penetrate into the understory adjacent to gaps, buteven in the brightest gap positions the duration of direct sun isbrief (Canham et al., 1990). The discrepancy of the mid-slope’sdiversity pattern may be explained by: (1) The pre-harvest spe-cies composition may have provided species with a competitiveadvantage in that microsite position; (2) The northern aspectmay have provided protection from spring desiccation along the

Fig. 4. Mean evenness by gap position for all sites in 2004. Gap positions are SU – southern understory, SE – southern edge, SC – southern center, NC – northern center, NE –northern edge, and NU – northern understory. Error bars represent one standard error. Within each site, columns sharing the same letter are not significantly different (TukeyHSD).

Fig. 5. Interaction plots of mean evenness over time by gap position and ground-disturbance (4A Mesic, 4B Sandy-skeletal). Gap positions are SU – southern understory, SE –southern edge, SC – southern center, NC – northern center, NE – northern edge, and NU – northern understory. Ground disturbance levels are RE – release and LE – lethal.Asterisks indicate that significant interaction in the ANOVA.

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 155

Fig. 6. Rank abundance curves for all sites. Pre-harvest (1999) shown in black diamonds, 10 years post-harvest (2010) shown in gray triangles.

156 M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159

normally harsh northern gap edge; or (3) The resource heteroge-neity in the stony till soil provides greater resource availabilitythan uniform soils (McKenna, 2007). The fact that the two sandysoils followed a similar pattern to the mesic site confirms thatcompetition from regenerating trees is an influential spatial dri-ver of understory plant diversity. Evenness on the other handpeaked in the shadiest understory positions and was lowest alongthe harsh, competitive north central and northern edge positions.Dominance in those positions by a few strong competitors islikely responsible for the low observed diversity, and is an impor-tant consideration for forest management.

Many studies have examined herbaceous layer response to gapswith mixed results. Some found little to no relationship betweenopening size and herbaceous response, and species richness unaf-fected by gap dynamics (Moore and Vankat, 1986; Collins and Pick-ett, 1988; Reader and Bricker, 1992). Others found differences inspecies composition (Matlack, 1994; Schumann et al., 2003; Faheyand Puettmann, 2007; Naaf and Wulf, 2007), supporting the theoryof resource partitioning within gaps. These differences could be ex-plained by gap size; larger gaps possess larger resource gradientsand as a result, show more differentiation by species (Fahey andPuettmann, 2007). Understory species composition may representa better reflection of these resource gradients than associated over-story species, because of their greater diversity and niche partition-ing (Matlack, 1994; Fahey and Puettmann, 2007). Our results,standardized by gap size and orientation across sites, suggest thatthe underlying site conditions are important.

Disturbance may interact with microenvironment on these sitesto create a secondary axis of diversification and provide additionalopportunities for certain species to establish and contribute to fur-ther gap partitioning (Fahey and Puettmann, 2007). While size,shape, and orientation of gaps determine resource availability (Col-lins et al., 1985), our results show that location and underlying siteconditions also have an effect on species diversity and sensitivityto different levels of disturbance. Interestingly, we found that theintensity of ground disturbance, when combined with increasedlight, plays a more important role on sites with dry and sandy soils.

Further, ground disturbance is more influential immediately after adisturbance, but lessens in its importance as succession progresses.These effects also are more pronounced on the drier sandy sites.Interactions between gap position and disturbance have moreinfluence on species evenness and diversity than richness, butthese effects are also site specific. Our results support the idea thatfollowing a disturbance habitat heterogeneity and niche differenti-ation may be as or more important than overall site productivity ininfluencing species richness (Shmida and Wilson, 1985), at least atsmall spatial scales.

Temporal factors must also be considered. Understory responseto harvesting consists of the initial effects of the disturbance itself,and long-term effects resulting from successional change (Robertsand Gilliam, 2003). Our results from the mesic site show Shannondiversity and richness initially driven by ground disturbance, butas succession progressed the influence of gap position becamemore prominent. Environmental variation within gaps decreaseswith succession, and species cover and composition adjust tempo-rally with these changes (Moore and Vankat, 1986; Collins andPickett, 1988; Goldblum, 1997; Ford et al., 2000).

Previous work on till soils found light to be the primary influ-ence on tree sapling growth and not moisture (Pacala et al.,1994). Ashton and Larson (1996) found that height growth foroak seedlings is greatest in gap centers of valley positions in com-parison to drier ridges, suggesting moisture does play a role. Ourresults show richness in the two moister sites already beginningto decrease within 10 years of gap creation, while richness in thedrier sandy sites is still increasing. This implies that competitionfrom the regenerating stand has usurped light resources within10 years on the better sites, while the sandy sites still have ampleavailable growing space. The relationship between certain guilds(e.g., ruderal species) and known patterns of establishment withingaps over time suggests that gap effects may be short-lived (Faheyand Puettmann, 2008). In fact, Beaudet et al. (2004) found thatshade levels returned to pre-disturbance levels after just 13 years.Our data shows overall higher establishment and height growth ofseedlings on the mesic and mid-slope sites, particularly within the

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 157

lethal treatments (Frey, 2012). The role of succession is alsoevident in the interaction plots (Fig. 5). The mesic site, which hasfaster sapling growth, has more divergent interactions in 2004,becoming insignificant by 2010. The sandy-skeletal site shows alag behind the mesic, becoming more divergent as time goes on.We only examined effects 10 years after gap-creation. It wouldbe useful to revisit after a longer temporal scale to further eluci-date the difference in successional time frame amongst sites.

Across the four sites we see a positive correlation between spe-cies richness and site productivity before gap creation (Grime,1973; Grace, 1999; Mittelbach et al., 2001), though the responseto disturbance is not equal across sites. By 2010, jackknife esti-mates indicate the sandy-skeletal site almost quadrupled in rich-ness while still trending upward, while the mesic site had onlydoubled at its peak. One explanation is pre-disturbance composi-tion and species life-history traits, which play an important rolein determining post-disturbance communities (Halpern, 1989; Ai-kens et al., 2007). In a compositional study on the mid-slope site,Aikens et al. (2007) found that although species composition chan-ged in both lethal and release treatments, the release treatmentsmore closely resembled pre-disturbance communities. Gap centerswere dominated by early successional species, while edge plotswere dominated by residual species (and did not differ betweenthe north and south edge). We did not analyze species composi-tional change, but the compositional pattern Aikens et al. (2007)observed in the mid-slope location did hold true for diversity aswell. Whether compositional diversity follows the same patternas species diversity across the other sites is not clear. The impor-tance of compositional diversity to ecosystem function is becomingincreasingly clear (Diaz and Cabido, 2001). Further research isneeded that incorporates the role of life history traits in determin-ing response to disturbance. More research is also needed to eval-uate whether compositional and functional diversity is maximizedunder the same conditions as species diversity.

5. Management implications

Resource managers interested in understory species diversityneed to consider underlying site conditions (specifically soil type)when planning for treatments. Site-specific silviculture may ap-pear obvious, but it is often ignored. There are inherently differentbaseline levels of understory diversity to work with based on soiltype, and the response of the understory community to distur-bance can vary greatly. Although light availability is generallymore influential than ground disturbance in determining thedevelopment of the understory after forest harvesting, it can be sitespecific. Moderate ground-level disturbance during harvesting hasthe capacity to increase understory diversity, particularly inshaded microsites on dry sandy soils. When ground-disturbanceis varied, disturbed patches (lethal) adjacent to less disturbedpatches (release), it can promote increased diversity. This hetero-geneity allows for legacy reserves of clonal understory herbs andopportunities for ruderal species – both of which have importantecological roles. Heterogeneous ground-disturbance is consistentwith many types of forest harvesting and can be easily integratedinto silvicultural prescriptions. However, microsites with morelethal disturbances maybe more prone to invasives (Hobbs,1989), especially on rich mesic sites (Lake and Leishman, 2004).

Silvicultural treatments have the ability to create heterogeneity,which can be advantageous for biodiversity. Group-reserves withineven-aged regeneration methods, such as shelterwoods and seedtrees could be favorable to biodiversity at the stand and landscapelevels. Group reserves need to be large enough to mediate climaticconditions, and should be placed strategically so that colonizationzones for herbs can intersect and combine early in stand develop-

ment. Managers should not necessarily avoid all soil disturbanceduring harvesting operations, but encourage an intimate mixtureof site preparation treatments where scarification is done adjacentto protection of the understory. One way would be to combine pro-tected areas that may have advance regeneration, and scarifyingareas where there is no regeneration present. In selection treat-ments, larger irregular gaps with increased edge provide more hab-itat opportunities for a variety of species. The orientation of the gapis also important. When possible, gaps should be oriented so spe-cies of high conservation value can be maintained on south sidesof gaps, and care should be taken to minimize harsh north edgesthat can contribute to dominance by aggressive colonizing species.

Acknowledgements

The Mellon Foundation and the Leopold Schepp Foundationfunded this research. The authors would like to thank Yale-MyersForest and The University of Massachusetts for infrastructure andlogistical support. We are grateful to John McKenna, Melissa Ai-kens, Eli Sagor, Avril de la Cretaz, Caroline Kuebler, Samantha Roth-man, Fulton Rockwell, and Jason Nerenberg for data collection andfieldwork. Thank you to Elaine Hooper, Stella Cousins, KristoferCovey, Jeff Carroll, Karin Burghardt, the Silviculture & LMS labs atYale School of Forestry and Environmental Studies for supportand feedback during the preparation of this manuscript, and totwo anonymous referees for their comments and feedback on ear-lier drafts.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foreco.2013.04.018.

References

Aikens, M.L., Ellum, D., McKenna, J.J., Kelty, M.J., Ashton, M.S., 2007. The effects ofdisturbance intensity on temporal and spatial patterns of herb colonization in asouthern New England mixed-oak forest. Forest Ecology and Management 252,144–158.

Armesto, J.J., Pickett, S.T.A., 1985. Experiments on disturbance in old-field plantcommunities: impact on species richness and abundance. Ecology 66, 230–240.

Ashton, M.S., Larson, B.C., 1996. Germination and seedling growth of Quercus(section Erythrobalanus) across openings in a mixed-deciduous forest ofsouthern New England, USA. Forest Ecology and Management 80, 81–94.

Battles, J.J., Shlisky, A.J., Barrett, R.H., Heald, R.C., Allen-Diaz, B.H., 2001. The effectsof forest management on plant species diversity in a Sierran conifer forest.Forest Ecology and Management 146, 211–222.

Beaudet, M., Messier, C., Leduc, A., 2004. Understorey light profiles in temperatedeciduous forests: recovery process following selection cutting. Journal ofEcology 92, 328–338.

Beckage, B., Clark, J.S., Clinton, B.D., Haines, B.L., 2000. A long-term study of treeseedling recruitment in southern Appalachian forests: the effects of canopygaps and shrub understories. Canadian Journal of Forest Research 30, 1617–1631.

Bellemare, J., Motzkin, G., Foster, D.R., 2002. Legacies of the agricultural past in theforested present: an assessment of historical land-use effects on rich mesicforests. Journal of Biogeography 29, 1401–1420.

Berkowitz, A.R., Canham, C.D., Kelly, V.R., 1995. Competition vs. facilitation of treeseedling growth and survival in early successional communities. Ecology 76,1156–1168.

Bhatti, J.S., Fleming, R.L., Foster, N.W., Meng, F.R., Bourque, C.P.A., Arp, P.A., 2000.Simulations of pre- and post-harvest soil temperature, soil moisture, andsnowpack for jack pine: comparison with field observations. Forest Ecology andManagement 138, 413–426.

Bormann, F.H., Likens, G.E., 1979. Pattern and Process in a Forested Ecosystem.Springer-Verlag, New York.

Brooks, T.M., Mittermeier, R.A., da Fonseca, G.A.B., Gerlach, J., Hoffmann, M.,Lamoreux, J.F., Mittermeier, C.G., Pilgrim, J.D., Rodrigues, A.S.L., 2006. Globalbiodiversity conservation priorities. Science 313, 58–61.

Burton, P.J., Coward, A.C., Cumming, S.G., Kneeshaw, D.D., 1992. The value ofmanaging for biodiversity. The Forestry Chronicle 68, 225–236.

Canham, C.D., Denslow, J.S., Platt, W.J., Runkle, J.R., Spies, T.A., White, P.S., 1990.Light regimes beneath closed canopies and tree-fall gaps in temperate andtropical forests. Canadian Journal of Forest Research 20, 620–631.

158 M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159

Collins, B.S., Pickett, S.T.A., 1988. Response of herb layer cover to experimentalcanopy gaps. American Midland Naturalist 119, 282–290.

Collins, B.S., Dunne, K.P., Pickett, S.T.A., 1985. Response of forest herbs to canopygaps. In: Pickett, S.T.A., White, P.S. (Eds.), The Ecology of Natural Disturbanceand Patch Dynamics. Academic Press, Orlando, FL, pp. 217–234.

Connell, J.H., 1978. Diversity in tropical rain forests and coral reefs. Science 199,1302–1310.

Daubenmire, R., 1959. A canopy-coverage method of vegetational analysis.Northwest Science 50, 431.

Dauber, J., Hirsch, M., Simmering, D., Waldhardt, R., Otte, A., Wolters, V., 2003.Landscape structure as an indicator of biodiversity: matrix effects on speciesrichness. Agriculture, Ecosystems & Environment 98, 321–329.

De La Cretaz, A.L., Kelty, M.J., 2002. Development of tree regeneration in fern-dominated forest understories after reduction of deer browsing. RestorationEcology 10, 416–426.

Diaz, S., Cabido, M., 2001. Vive la différence: plant functional diversity matters toecosystem processes. Trends in Ecology & Evolution 16, 646–655.

Duffy, D.C., Meier, A.J., 1992. Do Appalachian herbaceous understories ever recoverfrom clearcutting? Conservation Biology 6, 196–201.

Ellum, D.S., Ashton, M.S., Siccama, T.G., 2010. Spatial pattern in herb diversity andabundance of second growth mixed deciduous-evergreen forest of southernNew England, USA. Forest Ecology and Management 259, 1416–1426.

Fahey, R.T., Puettmann, K.J., 2007. Ground-layer disturbance and initial conditionsinfluence gap partitioning of understorey vegetation. Journal of Ecology 95,1098–1109.

Fahey, R.T., Puettmann, K.J., 2008. Patterns in spatial extent of gap influence onunderstory plant communities. Forest Ecology and Management 255, 2801–2810.

Falk, K.J., Burke, D.M., Elliott, K.A., Holmes, S.B., 2008. Effects of single-tree andgroup selection harvesting on the diversity and abundance of spring forestherbs in deciduous forests in southwestern Ontario. Forest Ecology andManagement 255, 2486–2494.

Ford, W.M., Odom, R.H., Hale, P.E., Chapman, B.R., 2000. Stand-age, standcharacteristics, and landform effects on understory herbaceous communitiesin southern Appalachian cove-hardwoods. Biological Conservation 93, 237–246.

Frey, B.R., 2012. Spatial and temporal regeneration ecology of oak-dominatedforests of southern New England in Forestry and Environmental Studies. YaleUniversity, New Haven, CT, pp. 136.

Frey, B.R., Lieffers, V.J., Munson, A.D., Blenis, P.V., 2003. The influence of partialharvesting and forest floor disturbance on nutrient availability and understoryvegetation in boreal mixedwoods. Canadian Journal of Forest Research 33,1180–1188.

George, L.O., Bazzaz, F.A., 2003. The herbaceous layer as a filter determining spatialpattern in forest tree regeneration. In: Gilliam, F.S., Roberts, M.R. (Eds.), TheHerbaceous Layer in Forests of Eastern North America. New York, Oxford, pp.265–282.

Gilliam, F.S., 2002. Effects of harvesting on herbaceous layer diversity of a centralAppalachian hardwood forest in West Virginia, USA. Forest Ecology andManagement 155, 33–43.

Gilliam, F.S., Turrill, N.L., Adams, M.B., 1995. Herbaceous-layer and overstoryspecies in clear-cut and mature central Appalachian hardwood forests.Ecological Applications 5, 947–955.

Goldblum, D., 1997. The effects of treefall gaps on understory vegetation in NewYork State. Journal of Vegetation Science 8, 125–132.

Grace, J.B., 1999. The factors controlling species density in herbaceous plantcommunities: an assessment. Perspectives in Plant Ecology, Evolution andSystematics 2, 1–28.

Gray, A.N., Spies, T.A., Easter, M.J., 2002. Microclimatic and soil moisture responsesto gap formation in coastal Douglas-fir forests. Canadian Journal of ForestResearch 32, 332–343.

Grime, J.P., 1973. Competitive exclusion in herbaceous vegetation. Nature 242, 344–347.

Halpern, C.B., 1989. Early successional patterns of forest species: interactions of lifehistory traits and disturbance. Ecology 70, 704–720.

Hammett, A.L., Chamberlain, J.L., 1998. Sustainable use of non-traditional forestproducts: alternative forest-based income opportunities. In: Kays, J.S., Goff, G.R.(Eds.), Natural Resources Income Opportunities on Private Lands. University ofMaryland Cooperative Extension Service, Hagerstown, Maryland, pp. 141–147.

Hobbs, R.J., 1989. The nature and effects of disturbance relative to invasions. In:Drake, J., Mooney, H., di Castri, F. (Eds.), Biological invasions: a globalperspective. Scientific Committee on Problems of the Environment Series, vol.37. Wiley, pp. 389–405.

Hughes, J.W., Fahey, T.J., 1991. Colonization dynamics of herbs and shrubs in adisturbed Northern Hardwood Forest. Journal of Ecology 79, 605–616.

Jackson, S.W., Harper, C.A., Buckley, D.S., Miller, B.F., 2006. Short-term effects ofsilvicultural treatments on microsite heterogeneity and plant diversity inmature tennessee oak–hickory forests. Northern Journal of Applied Forestry 23,197–203.

Jenkins, M.A., Parker, G.R., 1999. Composition and diversity of ground-layervegetation in silvicultural openings of southern Indiana forests. The AmericanMidland Naturalist 142, 1–16.

Kern, C.C., Palik, B.J., Strong, T.F., 2006. Ground-layer plant community responses toeven-age and uneven-age silvicultural treatments in Wisconsin northernhardwood forests. Forest Ecology and Management 230, 162–170.

Kindt, R., Coe, R. 2005. Tree diversity analysis: a manual and software for commonstatistical methods for ecological and biodiversity studies. World AgroforestryCentre Eastern and Central Africa Program.

Lake, J.C., Leishman, M.R., 2004. Invasion success of exotic plants in naturalecosystems: the role of disturbance, plant attributes and freedom fromherbivores. Biological Conservation 117, 215–226.

Maguire, D.A., Forman, R.T.T., 1983. Herb cover effects on tree seedling patterns in amature Hemlock-Hardwood forest. Ecology 64, 1367–1380.

Magurran, A.E., 2004. Measuring Biological Diversity. Blackwell PublishingCompany, Malden, MA.

Matlack, G.R., 1994. Vegetation dynamics of the forest edge-trends in space andsuccessional time. Journal of Ecology 82, 113–123.

McKenna, J., 2007. The effects of multiple resources on forest regeneration:Microsite variation and seedling response. University of Massachusetts,Amherst, pp. 129.

Meier, A.J., Bratton, S.P., Duffy, D.C., 1995. Possible ecological mechanisms for loss ofvernal-herb diversity in logged eastern deciduous forests. EcologicalApplications 5, 935–946.

Meyer, W.H., Plusnin, B.A., 1945. The Yale Forest in Tolland and Windham Counties,Connecticut. In: Huber, W. (Ed.), Yale Forestry Bulletin. Yale School of Forestry,New Haven, CT, p. 54.

Mittelbach, G.G., Steiner, C.F., Scheiner, S.M., Gross, K.L., Reynolds, H.L., Waide, R.B.,Willig, M.R., Dodson, S.I., Gough, L., 2001. What is the observed relationshipbetween species richness and productivity? Ecology 82, 2381–2396.

Moola, F.M., Vasseur, L., 2008. The maintenance of understory residual flora witheven-aged forest management: a review of temperate forests in northeasternNorth America. Environmental Reviews 16, 141.

Moore, M.R., Vankat, J.L., 1986. Responses of the herb layer to the gap dynamics of amature Beech-Maple forest. American Midland Naturalist 115, 336–347.

Muller, R.N., 2003. Nutrient Relations of the herbaceous layer in deciduous forestecosystems. In: Gilliam, F.S., Roberts, M.R. (Eds.), The Herbaceous Layer inForests of Eastern North America. Oxford University Press, Oxford, pp. 15–37.

Muller, R.N., Bormann, F.H., 1976. Role of Erythronium americanum Ker. in energyflow and nutrient dynamics of a northern hardwood forest ecosystem. Science193, 1126–1128.

Naaf, T., Wulf, M., 2007. Effects of gap size, light and herbivory on the herb layervegetation in European beech forest gaps. Forest Ecology and Management 244,141–149.

North, M., Oakley, B., Fiegener, R., Gray, A., Barbour, M., 2005. Influence of light andsoil moisture on Sierran mixed-conifer understory communities. Plant Ecology177, 13–24.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., O’Hara, R.B., Simpson, G.L.,Solymos, P., Stevens, M.H.H., Wagner, H. 2010. vegan: Community EcologyPackage. R package version 1.17-4 <http://CRAN.R-project.org/package=vegan>.

Oliver, C.D., Larson, B.C., 1996. Forest Stand Dynamics. John Wiley & Sons, New York.Pacala, S.W., Canham, C.D., Silander Jr., J.A., Kobe, R.K., 1994. Sapling growth as a

function of resources in a north temperate forest. Canadian Journal of ForestResearch 24, 2172–2183.

Peltzer, D.A., Bast, M.L., Wilson, S.D., Gerry, A.K., 2000. Plant diversity and treeresponses following contrasting disturbances in boreal forest. Forest Ecologyand Management 127, 191–203.

Peterson, D.L., Rolfe, G.L., 1982. Nutrient dynamics and decomposition of Litterfallin Floodplain and Upland Forests of Central Illinois. Forest Science 28, 667–681.

R Development Core Team, 2010. R: A language and environment for statisticalcomputing. R Foundation for Statistical Computing, Vienna, Austria.

Reader, R.J., Bricker, B.D., 1992. Response of five deciduous forest herbs to partialcanopy removal and patch size. American Midland Naturalist 127, 149–157.

Ricketts, T.H., 1999. Who’s where in North America? BioScience 49, 369.Roberts, M.R., Gilliam, F.S., 1995. Patterns and mechanisms of plant diversity in

forested ecosystems: implications for forest management. EcologicalApplications 5, 969–977.

Roberts, M.R., Gilliam, F.S., 2003. Response of the herbaceous layer to disturbance inEastern Forests. In: Gilliam, F.S., Roberts, M.R. (Eds.), The Herbaceous Layer inForests of Eastern North America. New York, Oxford, pp. 302–320.

Roberts, M.R., Zhu, L., 2002. Early response of the herbaceous layer to harvesting in amixed coniferous-deciduous forest in New Brunswick, Canada. Forest Ecologyand Management 155, 17–31.

Ruben, J.A., Bolger, D.T., Peart, D.R., Ayres, M.P., 1999. Understory herb assemblages25 and 60 years after clearcutting of a northern hardwood forest, USA.Biological Conservation 90, 203–215.

Schumann, M.E., White, A.S., Witham, J.W., 2003. The effects of harvest-created gapson plant species diversity, composition, and abundance in a Maine oak–pineforest. Forest Ecology and Management 176, 543–561.

Shmida, A., Wilson, M.V., 1985. Biological determinants of species diversity. Journalof Biogeography 12, 1–20.

Singleton, R., Gardescu, S., Marks, P.L., Geber, M.A., 2001. Forest herb colonization ofpostagricultural forests in central New York State, USA. Journal of Ecology 89,325–338.

Srivastava, D.S., Vellend, M., 2005. Biodiversity-ecosystem function research: is itrelevant to conservation? Annual Review of Ecology, Evolution, and Systematics36, 267–294.

U.S., 2000. National Forest Management Act with Amendments. vol. 47 FR 43037.National Forest System Land and Resource Management Planning Authority.

UNEP, 2010. Global Biodiversity Outlook 3. Secretariat of the Convention onBiological Diversity, Montreal, pp. 94.

M.C. Duguid et al. / Forest Ecology and Management 303 (2013) 148–159 159

USDA, NRCS. 2013. The PLANTS Database. (<http://plants.usda.gov>, 26 February2013). National Plant Data Team, Greensboro, NC 27401-4901 USA.

Westveld, M., 1956. Natural forest vegetation zones of New England. Journal ofForestry 54, 332–338.

Whigham, D.F., 2004. Ecology of woodland herbs in temperate deciduous forests.Annual Review of Ecology, Evolution, and Systematics 35, 583–621.

Whitney, G.G., Foster, D.R., 1988. Overstorey composition and age as determinantsof the understorey flora of woods of Central New England. Journal of Ecology 76,867–876.

Zak, D.R., Groffman, P.M., Pregitzer, K.S., Christensen, S., Tiedje, J.M., 1990. Thevernal dam: plant-microbe competition for nitrogen in northern hardwoodforests. Ecology 71, 651–656.

Zenner, E.K., Kabrick, J.M., Jensen, R.G., Peck, J.E., Grabner, J.K., 2006. Responses ofground flora to a gradient of harvest intensity in the Missouri Ozarks. ForestEcology and Management 222, 326–334.