pu

Jrurg,al Ret Eco

a r t i c l e i n f o

Article history:Received 14 May 2014Received in revised form 30 November 2014Accepted 3 December 2014

a b s t r a c t

insect outbreaks or wet snow events in central European forestshas risen measurably (Schelhaas et al., 2003; Nilsson et al.,2004). Thus, approximately 16,000 ha or 0.15% of the total forestarea in Germany are annually disturbed considering the timebetween 1950 and 2012 (EFI, 2013). At the European level such

storation of largerest recovery andty and eco). Also, th

vation of fast growing pioneer tree species such as Populor Betula ssp. (Bazzaz, 1979; Pommerening and Murphy,may facilitate the adaptation of forests to environmental exand simultaneously favor the rapid sequestration of atmosphericcarbon to mitigate climatic change (Messier et al., 2013; Branget al., 2014).

However, in forests managed on a selection basis or close tonature, silvicultural systems with pioneers are typically avoidedin central Europe (Bauhus et al., 2013) in favor of those withlong-lived, commercially important tree species such as Fagus

Corresponding author. Tel.: +49 761 203 3678.E-mail addresses: hendrik.stark@waldbau.uni-freiburg.de (H. Stark), arne.noth-

durft@boku.ac.at (A. Nothdurft), joachim.block@wald-rlp.de (J. Block), juergen.bauhus@waldbau.uni-freiburg.de (J. Bauhus).

Forest Ecology and Management 339 (2015) 5770

Contents lists availab

Forest Ecology an

.e lcial component within the set of renewable energy options(Wiesenthal et al., 2006; Stupak et al., 2007).

However, during the last few decades the intensity and fre-quency of disturbances caused by storms, droughts, res, pest

Pioneer trees could play a major role in the redisturbed areas as they can possibly accelerate fotherefore supply multiple benets for productivifunctioning (Alban, 1982; Man and Lieffers, 1999http://dx.doi.org/10.1016/j.foreco.2014.12.0030378-1127/ 2014 Elsevier B.V. All rights reserved.systeme culti-us ssp.2004)tremes1. Introduction

Climate change mitigation strategies in Europe aim to off-setfossil fuels and replace energy-intensive materials with wood.Therefore, the supply of biomass from forests is considered a cru-

disturbances are equivalent to 8% of the total annual fellings(Schelhaas et al., 2003). Natural forest disturbances are thus a rel-evant aspect of central European forest management (Schelhaaset al., 2003; Nilsson et al., 2004; Majunke et al., 2008; Mantau,2012; Albrecht et al., 2013; Kraus and Krumm, 2013).Keywords:Nurse cropsPopulus ssp.Betula ssp.Biomass productionNutrient cyclingForest restorationThe rapid re-establishment of forests following large disturbances is being seen as one option to increasethe contribution of forests to climate change mitigation. The temporary inclusion of pioneer trees asnurse crops on disturbed sites can facilitate the establishment of target tree species and may additionallybenet productivity and soil fertility. In this study we compared productivity and nutrient cyclingbetween stands of oak target species (Quercus robur and Quercus petraea) that were established withand without widely spaced Betula ssp. or Populus ssp. nurse crops. Simulation results for a full rotationof oaks (180 years) indicated that both types of forests, with and without nurse crops, have a comparabletotal productivity. However, stands with nurse crops supplied 5996 Mg ha1 harvestable biomass after20 years, whereas the rst harvest of biomass from stands without nurse crops would occur at least30 years later. Nutrient element costs associated with the removal of Betula ssp. wood were low com-pared to Populus ssp. Also, nurse crop stands had up to 2.5 times larger pools of exchangeable base cationsin top mineral soils (030 cm) compared to mono-specic oak stands. The high soil cation pools may haveresulted from reduced leaching under nurse crops or the increased recycling of cations, also from deepersoil depth, via litter fall and ne-root turnover. Our results show that forest reestablishment with pioneertree species may be a suitable tool for the rapid recovery of forest productivity and mitigation potentialfollowing disturbances while simultaneously helping to maintain or increase soil fertility.

2014 Elsevier B.V. All rights reserved.Forest restoration with Betula ssp. and Poproductivity and soil fertility

Hendrik Stark a,, Arne Nothdurft b, Joachim Block c,aChair of Silviculture, Faculty of Environment and Natural Resources, University of Freibb Institute of Forest Growth, Department of Forest and Soil Sciences, University of NaturcDepartment of Forest Monitoring and Environmental Care, Research Institute for ForesTrippstadt, Germany

journal homepage: wwwlus ssp. nurse crops increases

gen Bauhus a

Tennenbacher Str. 4, 79085 Freiburg, Germanysources and Life Sciences (BOKU), Peter-Jordan-Strae 82, 1190 Vienna, Austrialogy and Forestry Rhineland-Palatinate (FAWF), Hauptstrae 16, 67705

le at ScienceDirect

d Management

sevier .com/locate / foreco

land-Palatinate. Here, nurse crop trials were established in 1991following large scale wind throw of Picea abies (L.) Karst. The trials

nd Mssp., Quercus ssp. or Picea ssp. Often, forest restoration with thesespecies is realized by planting mono-specic stands, which can suf-fer amongst others from late frost events in large forest openings(Lundmark and Hllgren, 1987; rlander, 1993; Groot andCarlson, 1996; rlander and Karlsson, 2000; Agestam et al.,2003). Therefore, the capacity of forests to provide biomass maybe substantially reduced, if, following disturbances, regenerationis delayed and thus productivity reduced (Osterburg et al., 2013).

On the other hand, large open areas in forests offer opportuni-ties for new, alternative and more resistant silvicultural systems.Hence, pioneer trees can be used as nurse crops, which are tempo-rary mixtures of fast growing and light-demanding tree species inthe overstorey and shade-tolerant target tree species in the under-storey (Pommerening and Murphy, 2004). During the initial yearsof stand development, the pioneer trees facilitate the establish-ment of other tree species beneath their canopy. One possiblenurse effect is the rapid recovery of forest micro-climate and thusthe amelioration of environmental extremes, such as frost (Carlsonand Groot, 1997; Schmidt-Schtz and Huss, 1998). Outside of cen-tral Europe, nurse crops are therefore commonly used for large orsmall scale reforestations (Pommerening and Murphy, 2004) fol-lowing clearfelling (Keenan et al., 1995; Lieffers et al., 1996) or nat-ural disturbances (Perala and Alm, 1990; Drouineau et al., 2000;Vallauri, 2005; Nelson et al., 2012) and for the afforestation ofabandoned agricultural lands (Mander and Jogman, 2000;Gardiner et al., 2004; Uri et al., 2007).

Once trees of the target species have been successfully estab-lished and start to suffer from inter-specic competition fromnurse trees, the latter are typically removed (Cotta, 1828;Shepperd and Jones, 1985). Often, however, the nurse trees are justkilled and not commercially harvested, because it is feared thattheir felling and extraction might damage the established regener-ation of target species.

To ensure that this biomass of nurse trees is not wasted butused, a new approach to establish widely-spaced nurse crops forthe purpose of woody biomass production has been developed(Unseld et al., 2012). This approach is based on the assumption thatthe temporary inclusion of fast growing pioneer tree species mayfacilitate an early harvest of biomass and thus an earlier returnof investment of stand establishment costs to the land owner.Additionally, the overall stand productivity may increase throughcompetitive reduction between tree species (Vandermeer, 1989;Man and Lieffers, 1997) and thus a complementarity in resourceuse, which typically occurs when mixing light demanding oversto-rey with shade-tolerant understorey trees (Man and Lieffers, 1999;Richards et al., 2010; Forrester, 2014). To achieve this, nurse cropsare managed on rather short rotations and a wide tree spacing isemployed that facilitates harvesting of nurse trees with minimaldamage to the regeneration (Unseld et al., 2012).

Nurse crops may also improve nutrient cycling following forestdisturbances, when the situation is often characterized by highmineralization rates of organic matter and reduced uptake of waterand nutrients. This can lead to high rates of leaching of nutrientelements such as calcium (Ca), potassium (K), magnesium (Mg)and nitrogen (N) and consequently a reduction in soil fertility(Hornbeck et al., 1986; Hendrickson et al., 1989; Bauhus andBartsch, 1995; Yanai, 1998). These nutrient element losses maybe aggravated, if the sites are subsequently used for intensive pro-duction and harvesting of biomass in nurse crops and nutrientscontained therein (Sverdrup et al., 2006; Worrell and Hampson,1997).

Owing to the rapid growth and canopy closure of pioneer treespecies, they have a higher nutrient and water uptake, thus reduc-

58 H. Stark et al. / Forest Ecology aing leaching, soil temperatures and mineralization rates, whencompared to slower growing tree species (Prescott, 2002). Interac-tions between nurse and target tree species may further accelerateoriginally aimed to study the effects of nurse crops of pioneer treespecies on the survival, growth and quality of site adapted targettree species such as oak (here: Quercus robur and Quercus petraea)when planted at water logged sites under open eld conditions.Thus, the trials included oak stands planted as conventional mono-cultures and oak stands that were planted under a sheltering nursecrop of pioneer tree species (Schmidt-Schtz and Huss, 1998).Treatments were randomly assigned. Between 1991 and 1998weed was controlled manually, dead tree saplings were replacedand naturally regenerated trees were removed. Between 20% and70% of the nurse tree seedlings died between 1991 and 1994 andwere replaced in 1994 (Schmidt-Schtz and Huss, 1998).

For the purpose of this study, we used two of the original trials,Kirchberg (N 49.98, E 7.32; 450 m a.s.l.) and Sobernheim(N 49.86, E 7.69; 420 m a.s.l.), for soil and biomass sampling in2011. We selected 16 plots equally spread across both trials (studysites). Hence, we selected eight plots with oak monoculture andeight plots with oak growing under a nurse crop shelter (Fig. 1).

Among the plots selected in Kirchberg, two were originallyestablished as nurse crops with Populus tremula L. x Populus tremu-loides Michx. Astria and two were established as nurse crops withBetula pendula Roth x Betula pubescens Ehrh. Quercus robur L. wasthe original target species in Kirchberg. It was used to establishthe mono-specic oak plots as well as the understorey of nursecrop plots. In Sobernheim two plots were established as nursecrops with Populus tremula L. x Populus tremuloides Michx. Astriaand two as nurse crop with Betula pendula. The target tree speciesthese processes and result in greater resource availability, uptakeand growth (Rothe and Binkley, 2001; Forrester, 2014).

While there have been many studies on the productivity ornutrient cycling in mixed-species forests (Richards et al., 2010),we are not aware of any studies that have focussed on temporarymixtures of target species such as Quercus ssp., Fagus ssp. or otherwith Populus ssp. or Betula ssp.

Therefore, the rst objective of this study was to assess whetherforest restoration with widely spaced nurse crops can facilitateearly biomass harvests, increase the overall stand productivity,and whether it thereby also intensies harvest-related exports ofnutrient elements such as Ca, K, Mg, N, and P when compared toconventional restoration of forest stands with the target speciesQuercus robur and Quercus petraea. The second objective was tostudy possible effects such nurse crops may have on ecosystemnutrient cycling. Thus, we hypothesize that nurse crops of broad-leaved pioneer tree species can buffer disturbance-induced and/or harvest-related reductions in soil fertility. Additionally, thestudy also investigated options to reduce harvest-related nutrientcosts in nurse crop systems. For this purpose, the nutrient costsof biomass harvest from Populus ssp. or Betula ssp. were comparedwith those of several other important tree species and at differentharvesting intensities. Thus, this study is the rst to examine pro-ductivity as well as nutrient cycling and conservation in widelyspaced, short-lived Populus ssp. or Betula ssp. nurse crops, provid-ing the foundation for the short-term sustainable production ofwoody biomass from disturbed forest areas in central Europe.

2. Materials and methods

2.1. Study area and trial forests

This study was carried out in the German federal state of Rhine-

anagement 339 (2015) 5770here was Quercus petraea (Mattuschka) Liebl., which was likewiseplanted in monocultures as well as in the understorey of nursecrops. In our statistical analyses we examined whether the

nd MH. Stark et al. / Forest Ecology adifference with respect to tree species selection had any effect onthe nutrient cycling or the nutrient costs of biomass harvest. Ifsuch effect was present, xed intercept parameters were estimatedand further applied in the statistical modeling.

In the following we will refer to all Quercus ssp. plots that wereestablished with a Populus ssp. nurse crop as aspen-oak. Quercusssp. plots that were established under a Betula ssp. nurse crop werereferred to as birch-oak. The aspen or birch layer of these plotswill be simply called aspen and birch, although in most caseshybrids were used. Oak monocultures will be referred to asmono-specic oak in text and as oak-mono in gures and tables.

The understorey oak was planted in 1996, while aspen in aspen-oak plots, birch in birch-oak plots and mono-specic oak plotswere planted in 1991. In 2011 the aspen and birch nurse crop treeswere nally harvested, thereby releasing the understory oak treesfrom intensive competition.

Plant spacing for aspen or birch was generally 4 4 m. Under-storey oaks in Kirchberg were planted at 4 1 m and understoreyoaks in Sobernheim were planted at 4 1.5 m and in accordancewith regional silvicultural practices, respectively. Oak rows plantedunder nurse crops were aligned parallel to the nurse crop rows andoffset by 2 m. Spacing in mono-specic oak stands was 1.5 0.7 min three out of four mono-specic oak plots in Kirchberg and in twoout of four mono-specic oak plots in Sobernheim. One mono-

Fig. 1. Chart of the study sites Kirchberg and Sobernheim as originally establishedin 1991 (modied from Schmidt-Schtz and Huss, 1998). The set of plots withmono-specic oak (oak-mono), aspen-oak and birch-oak stands assessed in thepresent study were marked with bold lines and color codes.specic oak plot in Kirchberg and two mono-specic oak plots inSobernheim were planted at 2 1 m.

Stand densities remained constant in aspen-oak and birch-oakstands because individuals that had died were replaced until1998. In mono-specic oak plots the stand densities successivelydecreased due to self thinning. Thus, the mean stand density inmono-specic oak stands in Kirchberg was 7743 stems per hectare(N ha1) in 1991 and 4814 N ha1 in 2011; in Sobernheim standdensities were 7130 N ha1 in 1991 and 5127 N ha1 in 2011.Mean diameters at breast height of aspen and birch were 19.4and 17.9 cm in Kirchberg and 16.9 and 17.4 cm in Sobernheim;mean heights were 15.5 and 12.9 m in Kirchberg and 13.8 and10.9 m in Sobernheim, respectively. Understorey oaks were about2.42.5 cm in diameter at breast height and 3.63.9 m high.Mono-specic oaks in Kirchberg were on average 7.5 cm in diame-ter at breast height and 9 m high; in Sobernheim they had a meandiameter at breast height of 5.8 cm and were 7.7 m high.

The plots were mostly square in shape with an edge length of25 m in aspen-oak, 50 m in birch-oak and 25 or 50 m in mono-spe-cic oak stands. Plots in Kirchberg were fenced in 1994 while onlya central 5 5 m square was fenced within each plot inSobernheim.

The soils at both study sites originate from historical loessdeposits mixed with clay and sand stone gravels and have a waterimpermeable layer at 3050 cm below the surface. Thus, the Kirch-berg site was dominated by Pseudogley and the site in Sobernheimwas dominated by Cambisol-Pseudogley. Mean air temperatures(19882011) during the vegetation season (daily average > 5 C,AprilOctober) were 12.4 and 13.0 C with a mean precipitation(2002 until 2011) of 552.9 and 423.5 mm in Kirchberg and Sobern-heim (FAWF, 2013), respectively.

2.2. Element pools in soil and biomass

To determine relevant base cation pools in forest oor and min-eral soils, soil cores were collected from 12 randomly selected sam-pling positions per plot. We used larger cylinders for the forestoor ( 170 mm) and smaller ones for the mineral soil (50 mm). Mineral soil cores were divided into 010 and 1030 cmdepth layers. To obtain information from a greater soil depth, addi-tional samples were collected from 3070 cm ( 24 mm) at fourout of 12 sampling positions. Samples were oven dried at 40 Cand their individual dry weight was recorded.

Subsequently, to reduce laboratory costs, pairs of sampleswere bulked to yield six samples (two samples for the layer3070 cm) per layer and plot. These bulked samples were sieved(2 mm), ground and 25 ml of 1 M NH4Cl were used to extractexchangeable base cations from 2.5 g soil sub-samples. We usedthe mass of ne soil per unit of sample volume multiplied bythe respective nutrient element concentration to estimate nutri-ent element contents per sample volume and subsequently perarea. Likewise, forest oor samples were mixed, sub-sampleswere ground to powder and subsequently digested in 65%HNO3 at 190 C (pressure vessels, Loftelds Analytical Solutions,Neu-Eichberg, Germany). The diluted solutions from extractionsand digestions were analyzed for element concentrations usingICP-OES (Spectro, Kleve, Germany) (Knig, 2005). Organic carbon(C) and nitrogen (N) concentrations were measured using drycombustion (Truspec, Leco, St. Joseph, USA). Soil pH was mea-sured in CaCl2.

To determine base cation pools within the woody biomass, ateach plot and for each tree species ve trees were destructivelysampled across the entire range of all tree diameters at breast

anagement 339 (2015) 5770 59height, respectively. Diameter at breast height (cm) and tree height(m) were recorded immediately after felling, and stem discs wereextracted at 2 m intervals along aspen and birch stems. The length

tions was recorded. Stem disc fresh weight was recordedseparately for wood and bark. Oak trees with a diameter at breast

nd Mheight thinner than 4 cm were assigned to the branch compart-ment (Stark et al., 2013).

For element analysis, wood shavings of stem discs were col-lected from two perpendicularly cross-sectional transects whilestem bark fragments were chipped. Both types of sub-sampleswere oven dried at 40 C until constant weight, mixed at the treelevel, and ground. Likewise, sample branches were chipped, mixedat the tree level and ground. Finally, compartment-specic elementanalysis was undertaken as described above for forest oorsamples.

The respective compartment-specic aboveground woody bio-mass of all trees was estimated using allometric biomass equations(Stark et al., 2013). The species-specic biomass of roots between 2and 5 mm was measured using the soil cores (0 and 30 cm depth)collected for this study, while the biomass of roots larger 5 mm andstumps was estimated according to equations provided in Wang(2006). The biomass of birch regeneration with diameters at breastheight ranging between approximately 1 and 5 cm was estimatedusing equations from Uri et al. (2007). The nutrient element poolscontained within sample tree stem wood, stem bark and brancheswere nally calculated using the measured compartment-specicbiomass and element concentrations.

We did not measure nutrient element concentrations in under-storey birch regeneration. In order to estimate the nutrient ele-ment pools within the aboveground woody biomass of birchregeneration we used the concentrations measured in sample treebranches, assuming that the stem-wood to stem-bark ratio andthus the nutrient element concentrations both in sample branchesof nurse crop trees and birch regeneration were comparable. Also,we did not measure nutrient element concentrations in roots.Therefore, we used information provided in the literature to inferthe average ratio between nutrient element concentrations in treebranches and roots. Based on such comparisons, we assumed treeroots with diameters between 2 and 5 mm to have comparablenutrient element concentrations as measured in our sample treebranches (Jacobsen et al., 2003), and we assumed stumps and roots>5 mm to have 50% of these branch concentration (Rademacher,2005; Block et al., 2007).

For planted aspen and birch in aspen-oak and birch-oak plotswe calculated the biomass and nutrient element pools per ha oneach sample plot separately as

Pnj1tj

nab 10;000, with tj being thesingle tree biomass or nutrient element pool of each planted aspenand birch tree on a specic plot, and a and b being the width of theregular plant spacing in x and y direction. The biomass and nutri-ent element pools per ha of mono-specic oak, natural regenera-tion and understory oaks were estimated by means of 2 m wide,diagonal transect samples.

2.3. Nutrient element costs of wood

We interpreted the woody biomass (kg) divided by the amountof a nutrient element contained therein (kg) (Pastor et al., 1984;Wang et al., 1991) of aspen and birch sample trees as the nutrientelement costs of woody biomass harvest. Additional data for sev-of oak stems with a diameter larger than 4 cm was divided into sixequally long sections and stem discs were extracted at these rela-tive intervals. From each sample tree with a diameter at breastheight thicker than 4 cm three branches were sampled from thelower, middle and upper crown section, respectively. Subse-quently, the fresh weight of all remaining branches and stem sec-

60 H. Stark et al. / Forest Ecology aeral other tree species of major economic importance were pro-vided by Pretzsch et al. (2013) or compiled from Jacobsen et al.(2003) for comparison.2.4. Statistical analysis

Linear mixed-effect models were employed to analyze howaspen-oak and birch-oak plots may affect soil base cation and car-bon pools when compared to mono-specic oak plots. This model-ing approach accounted for the nested study design and thepartially unbalanced data set. Because the data were skewed tothe right, a log-transformation was applied prior to model tting.We used dummy variables and indicator functions to incorporatecategorical groups of xed effects such as treatment, soil depthand study site. Generally, non independence of predictor variableswas tested and 0.7 was used as threshold for the correlation coef-cient (Dormann et al., 2013).

2.4.1. Modeling of soil nutrient element poolsThe full models were of the form

Wijklm b10 b11 I treatf g b12 I depthf g b13 I site Sobernheimf g b14 I treatf g I depthf g I site Sobernheimf g b15 I treatf g I site Sobernheimf g b16 I treatf g I depthf g b17 I site Sobernheimf g I depthf g bijkl ijklm; 1

with bijkl N 0;j2

and ijklm N 0; k2

. Wijklm was the logarithmicelement pool of soil sample m of treatment i, soil depth j and studysite k from plot l. b10 denoted the intercept of the full model repre-senting pools in the forest oor under aspen-oak stands in Kirch-berg and b14 to b

17 denoted interactions among the main xed

effects. bijkl was a plot level random effect and ijklm denoted theresidual error. The 0-hypothesis H0 : b1r 0 meaning that a param-eter b1r equals 0 was tested using Wald-tests. Model selection wasdone using the likelihood based information criterions AIC and BIC.

2.4.2. Modeling the nutrient element costs of biomass removalFor assessing the nutrient element costs of wood as a function

of tree species and biomass compartment and using log-trans-formed data we tted similar models

Pijklm b20 b21 I treatf gb22 I siteKirchbergf gb23 I compartmentf gb24 I treatf g I compartmentf gb25 I treatf g I siteKirchbergf gbijklijklm: 2

Pijklm represented the logarithmic biomass removal per mass unit ofthe respective nutrient element of sample tree m under treatment iof biomass compartment j and at study site k from plot l. b20 denotedthe intercept of the full model representing the amount of stemwood (kg) biomass that can be removed per unit nutrient element(kg) from oak trees (mono-specic stands) in Kirchberg.

2.4.3. Modeling of cation exchange capacityTo assess, whether differences in soil exchangeable cation pools

between treatments may be caused by changes in the quantity orquality of organic matter, we tted a third model. Because the datawere skewed to the right a log-transformation of response and pre-dictor variables achieved the best model t. The model was

Hijkl b30 b31 logCcon b32 logpH b33 I treatmentf g b34 logCcon I treatmentf g bijkl ijklm: 3

Hijk was the logarithmic cation exchange capacity of soil sample k oftreatment i and plot j. The intercept b30 represented the logarithmic

anagement 339 (2015) 5770cation exchange capacity at a given logarithmic soil organic carbonconcentration and logarithmic pH (CaCl2) in mineral soils to 10 cmdepth under mono-specic oak stands.

nd M2.5. Modeling of early biomass and related nutrient element pools

There is very little information on the sequestration of nutrientelements in biomass of young forests during their establishmentphase. However, during this period the rapid uptake of nutrientsinto tree biomass could substantially reduce nutrient leaching. Tocompare nutrient element sequestration from the time of standestablishment until age 20 years, diameter, height and survivalwere recorded for individual trees of selected sample rows in vemono-specic oak plots and for all aspen and birch individualson four aspen-oak and birch-oak plots (Schmidt-Schtz and Huss,1998; Eggert, 2006). This resulted in plot mean diameters, heightsand stand densities available for the years 19911997, 2006, 2011.

We then estimated diameters, heights and densities for all yearsin which these variable had not been recorded. For this, we pooledthe plot means from all years in which data had been recorded anddrew 199 sub-samples with replacement from this data pool, eachcomprising 15 values per species and study site, respectively. Foreach sub-sample we tted Pettersons (1955) growth function forthe plot mean tree diameter at stem base versus stand age, andused these models to predict the successive development of plotmean diameters of each treatment between the time of establish-ment and age 20 years.

For each predicted mean diameter, the respective biomass wasestimated using specic allometric biomass equations (Stark et al.,2013) and was then multiplied by the number of trees per hectare.For nurse crop systems a constant stand density was assumed. Forpure oak stands we regressed measured mean stand density valuesagainst mean diameters using a polynomial regression model andsubsequently estimated the consecutive annual stand densitydevelopment. We used nutrient element concentrations measuredduring the eld work in 2011 to calculate nutrient element pools.For each of the 199 sub-samples the entire procedure was gradu-ally calculated to ensure continuous error propagation.

2.6. Modeling and predicting long term development of biomass andnutrient element pools

We also predicted the annual biomass and nutrient elementpools for oaks established with and without aspen and birch nursecrops between stand age 20 years and the end of their rotation per-iod. For this we used a single tree, distance-independent growthmodeling approach. Our model specication aimed to reproducethe local management concepts for oak quality wood production.In brief, such concepts rely on initially high stand densities andthe canopy is generally kept close for three to four decades follow-ing canopy closure to ensure self-pruning for the production ofhigh quality timber. Once the targeted branch free bole length isreached, selection thinning is used to release potential future croptrees from competition to stimulate their crown expansion anddiameter growth (Dong et al., 2007; Spiecker, 2007).

We simulated complete data representing the oak trial plotswithin the model according to their original stand density andage. Oak stands established under an aspen or birch nurse cropwere simulated according to their physiological age, which wasderived by matching their actual height at the time of release withthe age-specic yield table tree height. Finally, a diameter at breastheight was assigned to each oak from prior distributions.

Algorithms for tree growth, thinning and harvesting wereadopted from the WEHAM tree growth simulator (Bsch, 2013).In WEHAM, an annual diameter increment is assigned to each treeaccording to its diameter at breast height at a given reference age,and following Slobodas (1971) growth function. The increment is

H. Stark et al. / Forest Ecology athen used to obtain a diameter prognosis for the subsequent sim-ulation cycle. Tree height was inferred from each predicted diam-eter using Pettersons (1955) growth function. All modelparameters were derived from model ts to forest inventory datacollected in the federal state of Rhineland-Palatinate (Bsch, 2013).

To conform with current management practices, once the mod-eled oaks exceeded a height of 12 m, a beech (Fagus syilvatica)understorey was introduced into the model using age specicinformation on tree diameters, their distribution as well as standdensity provided in Grote (2003) and Mnder (2005). Beech heightgrowth was modeled according to data collected from understoreybeech in the German national forest inventory (BMELV, 2002). Theabove described approach provided reliable predictions for growthof beech trees older than approximately 30 years. If not statedotherwise, model results comprise the biomass and nutrient ele-ment pools of both oak and beech.

Following the removal of aspen and birch trees, we assumed ave-year release effect for oaks established under aspen and birchnurse crops. This effect was depicted as an annual increase of 8%over the originally predicted diameter increment at breast heightand lasted for a period of 5 years. This resulted in an annual diam-eter increment of 36 mm, which was in accordance with datareported by Nutto (1999) for widely spaced (3 0.7 m) youngoak trees. The respective ability of oak to temporarily increase itsdiameter increment following release was demonstrated byUtschig and Pretzsch (2001).

Based on diameter estimates, the total aboveground woody bio-mass was predicted using allometric biomass equations developedby Stark et al. (2013) for oaks with diameter at breast heightbetween 011 cm, by Zell (2008) for oaks thicker than 11 cm andby Grote (2003) for beech. Total aboveground woody biomass ofoak was then subdivided into stem wood and bark as well as thewood and the bark of branches thicker and thinner than 70 mmusing relative proportions of the total aboveground woody biomassaccording to data published by Andr et al. (2010). Compartment-specic nutrient element pools for oak were calculated using con-centrations for wood, bark and branches determined in this study;concentrations for beech were adopted from Rademacher (2005).

Thinnings from below were conducted until the age of100 years. Older oak stands were selectively thinned, and the thin-ning intensity was inferred from the difference between the mod-eled basal area and yield table values (FVA-BW, 2000). Modelinternal thinning intervals varied between nine and 10 years irre-spective of practical silvicultural treatments, but the total removalswere assumed to be in accordance with best forest managementpractice in Germany. The selection of trees to be removed wasbased on a diameter-dependent criterion (Bsch, 2013) combinedwith a random selection process. Beech individuals were removedlatest once they exceeded 70% of mean oak height to prevent inter-action with oak crowns. Harvesting was simulated as a clearfellingoperation once the mean oak diameter of 70 cm was reached butlatest at the age of 180 years.

Until a stand height of 17 m thinned trees, and thus biomassand nutrient elements contained therein, were assumed to remainin the stand. Once the critical height of 17 m was reached, a branchfree bole length of 10 m a requirement for the production of highgrade oak wood was assumed for selected target trees due tocompetition induced self-pruning (Spiecker, 2007). The biomassand nutrient element removals associated with thinnings appliedbeyond the mean height of 17 m were therefore considered to beremoved from the stand. Possible random variation of the modelpredictions was examined using a Monte-Carlo Simulation with199 iterations.

2.7. General statistics

anagement 339 (2015) 5770 61Multiple pair-wise comparisons of means between treatments(biomass and soil element concentration and pools; base cationsaturation, pH and medium root abundance in soil; nutrient

2010) including MASS (Venables and Ripley, 2002), nlme (Pinheiro

larger in aspen compared to mono-specic oak. No signicant dif-

were signicantly larger than in aspen. In Sobernheim P pools inbirch were larger than in aspen or mono-specic oak betweenthe ages 10 and 15 years.

Oaks growing in the understorey of aspen or birch could not beconsidered in this analysis. At the age of 20 years, understorey oakscontributed approximately 3 Mg ha1 to the overall stand biomass.Likewise, root, leave and grass biomass was not considered.

3.1.2. Future development of biomass and related nutrient elementpools

We also estimated the future development of biomass andrelated nutrient element pools in oak plots that were establishedwith and without nurse crops (Figs. 3 right panel, S1 and S2).Please mind that aspen and birch were harvested from aspen-oakand birch-oak plots at the age of 20 years.

According to our model simulations and depending on the mea-sured, local stand parameters in 2011, oak stands that were estab-lished as mono-specic oak would reach a mean stand height of17 m and thus the starting time of commercial thinning at theage of 54 years in Kirchberg and 70 years in Sobernheim. Oak

0 0

0

20

40

oak

mon

o

aspe

nun

ders

t. oa

k

birc

hun

ders

t. oa

k

P

0

20

40

oak

mon

o

aspe

nun

ders

t. oa

k

birc

hun

ders

t. oa

k

Fig. 2. Empirical aboveground woody biomass and nutrient element poolscontained therein displayed for aspen nurse crops and their understorey oak(underst. oak), birch nurse crops and their understorey oak and mono-specic oak(oak-mono) in Kirchberg and Sobernheim, respectively. Only the biomass of aspenand birch nurse crops was actually harvestable at the age of 20 years. Understoreyoaks would continue growth following the nurse crop removal. Error bars indicatestandard deviations among the plots of each study site. Multiple pair-wisecomparison of means across both study sites indicated no signicant differencesin the biomass pools among aspen, birch and mono-specic oak. K pools weresignicantly larger in aspen compared to birch and P pools were signicantly largerin aspen compared to oak-mono. No signicant differences among the treatmentswere found for Ca, Mg and N. Mind: left and right axes display the same scales andunits.

nd Mferences among the treatments were found for Ca, Mg and N.

3.1.1. Past development of biomass and related nutrient element poolsWe reconstructed the past development of biomass and related

nutrient element pools for all treatments. According to this analy-sis and based on the interpretation of 95% condence intervals, thebiomass accumulated in aspen, birch, and mono-specic oakstands in Kirchberg was comparable throughout the initial20 years of stand life (Fig. 3 left panel). In Sobernheim, the 95%condence intervals indicated signicantly larger biomass poolsin birch than in aspen and mono-specic oak stands between theages of 10 and 15 years; no signicant differences were observedfor the time before and after that period.

The corresponding Ca, K and Mg pools (Figs. S1 and S2), elementconcentrations measured in 2011 Tables S7 and S8in Kirchberget al., 2011) and RODBC (Ripley and Lapsley, 2011) library.

3. Results

3.1. Productivity and related nutrient element pools in biomass20 years post establishment

After two decades, the mean aboveground woody biomassyielded by aspen, birch and mono-specic oak plots in Kirchbergwas 96 (30), 79 (8) and 95 (30) Mg ha1 (numbers in parenthesesindicate standard deviations), respectively. The biomass producedby these different stand types in Sobernheim was 67 (0.3), 59 (5)and 49 (18) Mg ha1, respectively (Fig. 2, Tables S1 and S2). In Kir-chberg oaks growing in the understorey of aspen-oak plots com-prised 4.8 (0.1) Mg of biomass per ha and those growing inbirch-oak plots had 4.5 (0.1) Mg ha1; in Sobernheim understoreyoaks contributed 2.7 (0.2) Mg ha1 to the overall biomass inaspen-oak plots and 2.7 (0.02) Mg ha1 in birch-oak plots.

Mean Ca, K, Mg, N and P pools in understorey oaks in Kirchbergwere 12 (1.5), 9 (1), 2 (0.3), 17 (2) and 1.5 (0.2) kg/ha and 7 (0.1),4.5 (0.1), 1 (0), 10 (0.2) and 1 (0) kg/ha in Sobernheim. Detailedinformation on biomass and nutrient element pools in roots,stumps and natural regeneration are presented as Supplementarymaterial in Tables S1S6.

Multiple pair-wise comparisons of the above mean biomass andnutrient element pools did not consider possible study site effectsbecause of the low number of aspen-oak and birch-oak replicatesper study site. The tests indicated no signicant differences inthe biomass pools among aspen-oak, birch-oak and mono-specicoak, irrespective of whether the oak understorey was considered inaspen-oak and birch-oak plots or not. K pools were signicantlylarger in aspen compared to birch and P pools were signicantlyelement costs of wood) were performed following the hypothesisthat their means were not signicantly different from 0 using thea 0:05 level. In order to account for unequal samples sizes thetests were performed according to the generalized linear hypothe-sis using the glht function of the multcomp library in R (Hothornet al., 2008; Bretz et al., 2010) and were based on generalized linearmodels. Comparing the nutrient element costs of wood among sev-eral tree species in central Europe we assumed a gamma data dis-tribution and specied models with a log-link function betweenpredictor and response. In all other cases, a Gaussian data distribu-tion with an identity-link function was assumed and, if necessary,data were log-transformed prior to model tting. We used the pro-gramming language and statistics environment R version 2.12.0 (R,

62 H. Stark et al. / Forest Ecology awere signicantly lower in birch than in aspen or mono-specicoak between the ages 15 and 20 years, respectively. No differenceswere observed for P and N. Also, Ca pools in mono-specic oak0

50

100

Kirchberg

biom

ass

0

50

100

Sobernheim

0

150

300

Ca

0

150

300

0

100

200

K

0

100

200

0

30

60

Mg

0

30

60

150

300

N 150

300

biom

ass

[Mg/

ha] a

nd n

utrie

nt [k

g/ha

] poo

ls

anagement 339 (2015) 5770established in aspen-oak or birch-oak stands would reach thatheight at the age of 73 years in Kirchberg and 78 years in Sobern-heim (Tables S9S12).

The biomass reduction due to mortality or release of future croptrees or cutting of wolf trees prior to commercial thinning wouldbe 188 and 143 Mg ha1 in oak plots established as mono-specicoak in Kirchberg and Sobernheim and 113 and 61 Mg ha1 in oakplots that were established as aspen-oak or birch-oak in Kirchbergand Sobernheim, respectively (Figs. 3, S1 and S2 right panel, TablesS9S12). Nutrient element returns will follow a similar patternaccording to their respective concentrations.

The mean annual increment of stand height and diameter atbreast height in mono-specic oak stands would be 0.20 m year1

and 0.44 cm year1 in Kirchberg and 0.16 m year1 and0.37 cm year1 in Sobernheim; the mean annual increment ofheight and diameter at breast height in oaks stands establishedin aspen-oak or birch-oak plots would be 0.17 m year1 and0.38 cm year1 in Kirchberg and 0.16 m year1 and 0.34 cm year1

in Sobernheim.The biomass removal by thinning of competing oaks and beech

understorey trees in mono-specic oak stands would yield on aver-age 4.8 Mg ha1 year1 in Kirchberg and 4.5 Mg ha1 year1 inSobernheim, and 4.6 Mg ha1 year1 and 4.4 Mg ha1 year1 would

be yielded by oak stands planted in aspen-oak or birch-oak plots inKirchberg and Sobernheim, respectively (Figs. 3 and 4).

At the time of nal harvest the mean diameter at breast height,the mean height and the stand density of oak stands establishedwithout nurse crops would be 71 cm, 30 m and 72 trees per hain Kirchberg and 65 cm, 30 m and 92 trees per ha in Sobernheim.The mean diameter at breast height, the mean height and the stand

020406080

100

50100150200250300350

020406080

100ci 0.0250.975aspenoak meanbirchoak meanoakmono mean

50100150200250300350

0 5 10 15 20 50 100 150

ci 0.0250.975oaknursecrop meanoakmono mean

Kirc

hber

gS

ober

nhei

mbi

omas

s [M

g/ha

]

stand age [years]

Fig. 3. Left panels: Reconstructed aboveground woody biomass development of aspen, birch and mono-specic oak (oak-mono) stands between the ages 0 and 20 years atboth study sites, respectively. Mind that the biomass of understory oak could not be included here. They contributed about 3.2 Mg biomass per ha at the age of 20 years. Rightpanels: Biomass development of oak stands established with (black) and without (red) a nurse crop. These pools include a beech understorey, which is commonly installed tosuppress the development of epicormic branches in oak stems. Mind that the age of oaks established without a nurse crop corresponds to the stand age given at the x-axis,

e reis

200 0 50 100 2000

200

400

600

800aspenoak

birchoak

oakmono

e [years]

Table 1Total biomass and nutrient element removals from mono-specic oak stands andaspen-oak and birch-oak stands in Kirchberg and Sobernheim. In all cases the biomassproduced by a beech understorey is included, too.

Site Treatment Biomass Ca K Mg N P

Mg/ha kg/ha kg/ha kg/ha kg/ha kg/haK Oak-mono 898 1733 1291 255 1748 127

Aspen-oak 928 1896 1362 271 1871 152Birch-oak 911 1746 1286 257 1841 144

S Oak-mono 823 1643 1259 203 1773 122Aspen-oak 828 1728 1313 216 1855 137Birch-oak 820 1651 1234 202 1814 131

H. Stark et al. / Forest Ecology and Management 339 (2015) 5770 63while oaks established with a nurse crop were ve years younger. For details pleasintervals of 199 model iterations. (For interpretation of the references to color in th

0

200

400

600

800

1000

Kirchberg

0 50 100

biom

ass

cum

ulat

ive

rem

oval

[Mg/

ha]

agFig. 4. Cumulative biomass removals derived from three exemplied model runs for aspeIncluded is a beech understorey, which would be introduced at the oak height of 1Supplementary Tables S9S12.fer to the Supplementary Tables S9S12. Presented are means and 95% condencegure legend, the reader is referred to the web version of this article.)

1000

Sobernheimn-oak, birch-oak and mono-specic oak stands across the complete rotation period.2 m to prevent epicormic branches in oak stems. For details please refer to the

mass) would be harvested from oak stands established without

2

4

6

8

10

12

14

Ca

[mg/

g]

oak

mon

o

aspe

noa

k

birc

hoa

k

a

b

ab

Fig. 5. Ca concentrations measured in the forest oors of aspen-oak, birch-oak andmono-specic oak (oak-mono) stands. Different letters indicate signicant differ-ences between treatments and were produced from multiple pair-wise compari-sons of means.

3.2

3.4

3.6

3.8

4.0

oak

mon

o

aspe

noa

k

birc

hoa

k

pH C

aCl 2 a

ab

Fig. 6. Soil acidity (pH) measured in mineral soils in 010 cm depth under aspen-oak, birch-oak and mono-specic oak (oak-mono) plots. Different letters indicatesignicant differences between treatments and were produced from multiple pair-wise comparisons of means.

10

20

30

40

50

oak

mon

o

aspe

noa

k

birc

hoa

k

base

cat

ion

satu

ratio

n [%

]

a

b

c

Fig. 7. Base cation saturation of the exchange complex as measured in the mineralsoil in 010 cm depth aspen-oak, birch-oak and mono-specic oak (oak-mono)plots. Different letters indicate signicant differences between treatments and wereproduced from multiple pair-wise comparisons of means.

64 H. Stark et al. / Forest Ecology and Mnurse crop in Kirchberg and Sobernheim, respectively. In contrast,832 and 761 Mg ha1 biomass would be harvested from oak standsestablished under aspen or birch nurse crop in Kirchberg andSobernheim, respectively. Adding the biomass produced by aspenand birch to these values would yield in total 928 Mg ha1 fromaspen-oak and 911 Mg ha1 from birch-oak stands in Kirchberg;828 Mg ha1 would be harvested from aspen-oak and 820 Mg ha1

from birch-oak in Sobernheim (Table 1 and Fig. 4).

3.2. Nutrient element pools in soils 20 years post establishment

Ca concentrations in the forest oor beneath aspen-oak plotswere signicantly higher than in mono-specic oak plots (Fig. 5,Tables S13 and S14), whilst no signicant differences among treat-ments were found for K, Mg, N and P. The pH (Fig. 6) in the surfacemineral soil (010 cm) was signicantly higher in aspen-oak anddensity of oak stands established with nurse crops would be 66 cm,30 m and 88 trees per ha in Kirchberg and 58 cm, 20 m and 112trees per ha in Sobernheim. These values represent the oak layeronly; values for the respective beech layer are listed separatelyin Tables S9S12.

Over the complete rotation period of 180 years, a total amountof 898 and 823 Mg ha1 biomass (including oak and beech bio-

70

80

90

100

110

120

130

140

CE

C [

mol

c/g

soi

l] at

pH

3.4

40 60 80 100 120C [mg/g soil]

oakmonoaspenoakbirchoak

Fig. 8. Regression curves for cation exchange capacity (CEC) dependent on theorganic carbon concentration in the mineral soil in 010 cm depth (Eq. (3)). The pH(CaCl2) is xed at its overall mean of 3.4.

anagement 339 (2015) 5770the saturation of base cations (Fig. 7, Tables S13 and S14) was sig-nicantly higher in both aspen-oak and birch-oak than in mono-specic oak plots.

Additionally, model (3) analyzing the cation exchange capacityas a function of soil organic carbon (C) concentration as well asthe nurse crop treatment and their interactions revealed the sig-nicance of all these effects within the upper 10 cm of mineral soil(Table S15). These results suggest a different quality of organicmatter with a higher increase in cation exchange capacity per unitorganic carbon in soils under aspen-oak and birch-oak than undermono-specic oak (Fig. 8).

Finally, the pools of exchangeable Ca in the top 10 cm mineralsoil were 175 (101), 130 (112) and 62 (48) kg ha1 in aspen-oak,birch-oak and mono-specic oak plots in Kirchberg and for thesame treatments in Sobernheim 177 (67), 131 (47) and 96(28) kg ha1, all signicantly larger in aspen-oak and birch-oakthan in mono-specic oak stands (Fig. 9, Tables S3 and S4).

Exchangeable K pools in aspen-oak, birch-oak and mono-spe-cic oak plots were 76 (21), 68 (19) and 52 (12) kg ha1 in Kirch-berg and 79 (16), 55 (8) and 51 (9) kg ha1 in Sobernheim,respectively. In Kirchberg both the K pools in aspen-oak andbirch-oak were signicantly larger compared to mono-specic

nd M0

50

100

0

200

400

C*1

000

Kirchbergmeasured

[kg/ha]predicted

[%]

0

50

100

0

200

400

Sobernheimmeasured

[kg/ha]predicted

[%]

0

600

1200

0

200

400

Ca

0

600

1200

0

200

400

1000

400

1000

400

H. Stark et al. / Forest Ecology aoak, while in Sobernheim only those in aspen-oak were signi-cantly larger than in mono-specic oak stands.

In Kirchberg, the exchangeable pools of Mg were 40 (25), 25(25) and 14 (6) kg ha1 in aspen-oak, birch-oak and mono-specicoak plots, respectively; the corresponding pools in Sobernheimwere 26 (6), 20 (5) and 17 (8) kg ha1 in mono-specic oak,aspen-oak and birch-oak, respectively. Also the Mg pools inKirchberg under aspen-oak and birch-oak stands were signicantlylarger than in mono-specic oak, while in Sobernheim only inaspen-oak stands signicantly larger pools than in mono-specicoak stands were observed.

The pools of N were 1680 (417), 1613 (266) and 1595(353) kg ha1 in aspen-oak, birch-oak and mono-specic oak plotsin Kirchberg and 1122 (187), 1171 (158) and 1293 (165) kg ha1 inaspen-oak, birch-oak and mono-specic oak in Kirchberg. No sig-nicant difference were observed in Kirchberg, but in Sobernheimthe N pools in 010 cm mineral soil under mono-specic oak plotswere signicantly larger than under aspen-oak plots. P pools inmineral soils were not analyzed.

0

500

0

200

K

0

500

0

200

0

350

700

0

200

400

Mg

0

350

700

0

200

400

0

4000

8000

0

200

400

oak

mon

oas

pen

oak

birc

hoa

k

oak

mon

oas

pen

oak

birc

hoa

k

N

0

4000

8000

0

200

400

oak

mon

oas

pen

oak

birc

hoa

k

oak

mon

oas

pen

oak

birc

hoa

k

forest floor min. soil 010 cmmin. soil 1030 cm min. soil 3070 cm

Fig. 9. Left bar triplet of each panel: Measured nutrient element pools per soildepth layer with standard deviations (error bars) given for the overall element poolof all depth layers, respectively. Right bar triplet of each panel: To increase thevisibility of treatment effects, the relative difference between the treatment aspen-oak and the mono-specic oak reference stand as well as the relative differencebetween the treatment birch-oak and the mono-specic oak reference stand wascalculated from layer-specic nutrient element pool predictions using signicantlyparametrized models (1). Thus, for each depth layer (for each shade of gray) theelement pool under mono-specic oak plots (oak-mono) was set to 100% and theelement pools under aspen-oak or birch-oak systems could be smaller or largerthan that, respectively. Test results of pair-wise comparisons of means of empiricaldata can be found in Tables S3S6.anagement 339 (2015) 5770 65Inmodel (1) the xed study site effect generally explained a sig-nicant share of variation (Table S16) for all elements except Ca, andsoil depth and often the interaction of soil depth treatment andsoil depth study site were signicant xed effects, too. Accord-ing to this model the Ca pools were up to 140%, K up to 56% andMg pools up to 165% higher in mineral soils between 0 and 10 cmdepth under aspen-oak or birch-oak plots compared to soils undermono-specic oak plots (Tables S3 and S4). A similar pattern wasobserved in the mineral soil between 10 and 30 cm depth.

3.3. Nutrient element costs of wood

According to model (2) the xed effects tree species andbiomass compartment both signicantly affected the nutrient ele-ment costs of the harvestable woody biomass (Table S17). On aver-age 25%, 41%, 39% and 2% less total aboveground woody biomass

Fig. 10. Comparison of the amount of biomass (bulk wood including bark withdiameter > 7 cm) that can be harvested per mass unit nutrient element from severaltree species of central Europe. The larger the biomass per unit nutrient the lowerthe nutrient cost of biomass harvest. Error bars indicate the standard deviationamong sample trees. Different letters indicate signicant differences between treespecies and were produced from multiple pair-wise comparisons of means. Datawere compiled from this study, Jacobsen et al. (2003) and Pretzsch et al. (2013)(sample sizes: Betula pendula x Betula pubescens 12, Fagus sylvatica 195, Picea abies123, Pinus sylvestris 94, Populus tremula x Populus tremuloides 14, Pseudotsugamenziesii 72, Quercus ssp. 101).

In this study, aspen and birch at 4 4 m plant spacing in aspen-oak and birch-oak plots produced between 59 and 96 Mg of above-

aspen and birch canopy is removed.However, contrary to our initial assumptions, the total produc-

nd Mground woody biomass per ha. This biomass was harvestable after20 years (or earlier) and corresponded to a mean annual incrementbetween 3 to 5 Mg ha1 year1 (Fig. 2, Tables S1 and S2). Despitethe water logged soil conditions and the cool mountainous climate,the productivity of the studied nurse crops were comparable tothat of other widely spaced nurse crops in Germany (Nelsonet al., 2012; Unseld et al., 2012). They were also comparable to pro-ductivities reported in the literature for some aspen or birch shel-ter woods in Estonia or Sweden (Maard, 1996; Bergqvist, 1999;Johansson, 2003; Karacic et al., 2003), but signicantly lower,approximately only 50%, than the productivity reported for planta-tions in Denmark and Germany (Nielsen et al., 2014; Liesebachet al., 1999). However, the difference may be explained by themuch lower stocking of the studied nurse crops and also the lowerfertility of forest soils when compared to former agricultural soilsthat are typically used for woody biomass plantations.

Despite of their comparatively slow growth, mono-specic oakstands had produced as much biomass at the age of 20 years asaspen or birch nurse crops. This may be mostly attributable to theirmuch higher stand densities of approximately 5000 N/ha. Addi-tionally, it may also be related to the contrasting, species-specicgrowth dynamics of aspen, birch and oak. During the rst decadepost establishment, high biomass growth in aspen and birch standswould have exceeded those in mono-specic oak stands, but alsoper unit Ca, K, Mg and P could be harvested from aspen when com-pared to birch; 9% more aboveground woody biomass could beharvested per unit N from aspen than from birch. Alternatively,69%, 47%, 54% and 13% less stem wood and stem bark per unitCa, K, Mg and N might be removed from aspen than from birchand 5% more stem wood and stem bark could be harvested per unitN from aspen compared to birch. An additional analysis of varianceusing our empirical data supported these model predictions, indi-cating signicant differences in K and Mg costs between aspenand birch (Fig. 10, Tables S18 and S19).

Considering the harvest of woody biomass compartments withdiameters larger than 7 cm of major tree species in central Europe,woody biomass of Pseudotsuga menziesii had the lowest Ca costs,whilst aspen and birch took a moderate rank among most otherspecies; Ca costs of aspen and birch stems were signicantly lowerthan those of Quercus ssp. K costs, on the other hand, were moder-ately high in birch and Picea abies. In contrast, K costs of aspen andoak stems were comparatively high and Mg costs were lowest forstems of Pseudotsuga menziesii. In general, N and P costs of aspenand birch woody biomass were comparatively high.

Finally, harvesting without branches increased the amount ofwoody biomass per unit Ca, K, Mg, N and P in aspen by 12%, 11%,17%, 30% and 44% and in birch by 23%, 39%, 32%, 35% and 63% irre-spective of the study site (Eq. (2)).

4. Discussion

4.1. Effects of nurse crops on productivity

Many trials have studied the effect of tree species mixtures onstand productivity compared to pure plantations (Man andLieffers, 1999; Binkley, 2003; Forrester, 2014). Only few studieshave focused on tree species mixtures with aspen or birch (Manand Lieffers, 1999; Bergqvist, 1999; Johansson, 2003; Comeauet al., 2009). We are not aware of studies that have consideredaspen and birch as nurse crops with rotations of only two decades.

66 H. Stark et al. / Forest Ecology apeaked early to subsequently decline below the level of mono-specic oak stands leading to comparable biomass pools at theage of 20 years (Gutsell and Johnson, 2002).tion of harvestable biomass in aspen-oak or birch-oak stands inthis study was not higher than in conventional, mono-specicoak stands (Table 1 and Fig. 4). Thus, the studied nurse crop sys-tems can be regarded as sub-optimal with respect to oak as theunderstorey target tree species and also with respect to its spatialarrangement (Unseld et al., 2012). If oaks were to be used as thetarget tree species in nurse crop systems, then their establishmentin clusters in gaps between widely spaced nurse trees wouldreduce the competition for light and thereby also increase theirvitality and productivity (von Lpke, 1998; Saha et al., 2013;Forrester, 2014). Generally, nurse crop systems are likely more suc-cessful if a pronounced complementarity in light use through can-opy stratication and phenological differences through use of moreshade tolerant, evergreen understorey species such as Abies alba orPicea abies were used (Lieffers et al., 1996).

Transgressive over-yielding was observed in several studiesthat measured productivity in forests with aspen or birch growingover shade tolerant target species in North America and Scandina-via (Maard, 1996; Bergqvist, 1999; Man and Lieffers, 1999;Johansson, 2003; Kabzems et al., 2007). This strengthens our initialassumption that given that a more shade-tolerant understoreyspecies than Quercus robur or Quercus petraea was used tempo-rary nurse crops could yield additional, harvestable biomass bothin the short and long term when compared to reforestation withtarget species only.

Nevertheless, in this study, aspen or birch nurse crops increasedthe early production and supply of harvestable biomass by5996 Mg ha1, which could be used for example for energy pur-poses after 20 years or even earlier. Forest restoration with nursecrops of fast growing pioneers may therefore be an important toolto maintain or increase the mitigation potential of forests at timeswhen they are disturbed at increasing frequency and intensity(Schelhaas et al., 2003; Wiesenthal et al., 2006; Osterburg et al.,2013). Additionally, nurse crops may be an alternative source offuel wood or other wood products from abandoned agriculturallands or from areas where mono-specic coniferous forests shallbe converted into more tree species diverse, uneven-aged systems(Pommerening and Murphy, 2004; Uri et al., 2007; Unseld et al.,2012).

4.2. Effects of nurse crops on nutrient cycling

Poor forest soils may be especially negatively affected by distur-bance or intensive biomass removal in terms of soil fertility andecosystem functioning (Ballard, 2000). Silvicultural methods thatcan benet soil fertility are therefore advantageous (Sverdrupet al., 2006). In our study, we measured signicantly larger poolsof exchangeable base cations in surface mineral soil to 30 cm inaspen and birch nurse crop plots than in mono-specic oak plotsestablished without nurse crop (Fig. 9, Tables S3S6). In the con-In contrast, oaks growing in the understorey of aspen-oak orbirch-oak plots produced only 35% of the biomass in mono-spe-cic oak stands. Nevertheless, these oaks planted under nursecrops were ve years younger than aspen, birch and mono-specicoak stands and also had much lower stand densities compared tothe mono-specic oak stands. Also, between 2006 and 2011, theunderstorey oaks suffered from the strong competition by theaspen and birch overstorey (Ammer and Dingel, 1996). Yet, basedon other studies (Utschig and Pretzsch, 2001) one can expect thatthe understorey oaks will respond with increased growth once the

anagement 339 (2015) 5770text of forest restoration, these results clearly point to a benecialeffect of aspen and birch nurse crops on soil fertility, which maylast well beyond the removal of the nurse crop shelter.

nd MThere may be three processes responsible for the higher fertilityin the top soil of nurse crop stands (Alban, 1982; Binkley andGiardina, 1998): 1. reduced leaching of nutrients, 2. increaseduptake and recycling of nutrients, also from deeper soil layers,and 3. changes in the quality of organic matter.

During the initial phase of stand development our estimatesindicated that statistically signicant differences for abovegroundwoody biomass occurred between aspen and birch in nurse cropson the one side and mono-specic oak plots on the other side onlyfor the period between 10 and 20 years (Fig. 3), which would sug-gest that the nutrient sequestration effect of pioneer trees occurredtoo late. However, the biomass of grass, leaves and roots in aspen-oak and birch-oak plots might have exceeded those in mono-spe-cic oak plots much earlier because of their wide tree spacingand reduced competition for light (Comeau et al., 2006).

Also, widely spaced pioneer trees likely invested more C intoroot growth than oaks (Litton et al., 2007; Poorter et al., 2012;Stark et al., 2013) thus exploiting larger and deeper layers of soilusing a highly ramied system of ne and small roots (Fig. S3)(Strong and Roi, 1983; Gale and Grigal, 1987; Bauhus andMessier, 1999; Rothe and Binkley, 2001). Once base cations weretaken up from soils, aspen trees conserved large quantities of thesein woody tissues and thus prevented them from leaching (Fig. 2,Tables S3 and S4) (Hendrickson et al., 1987; Man and Lieffers,1999).

The recycling of Ca and Mg in leaf litter gained in importancewith stand development (Fig. 5, Tables S13 and S14) (Van Cleveet al., 1983; Pastor et al., 1984; Par and Cleve, 1993; Alrikssonand Eriksson, 1998). Nurse crops and especially aspen therebymeasurably increased base cation saturation (Fig. 7, Tables S13and S14) and pH (Fig. 6) in the surface mineral soil when comparedto mono-specic oak plots. In addition, aspen and birch nurse cropsmay have increased the abundance of earth worms (Saetre et al.,1999; Reich et al., 2005) and thus the incorporation of organic mat-ter into mineral soils, providing additional exchange sites(Alriksson and Eriksson, 1998) charged with a higher proportionof base cations. However, abundance of earthworms and other soilfauna was not measured in our study.

Moreover, the organic matter derived from aspen and birch leafand root litter appeared to have a higher exchange capacity thanorganic matter derived from oaks independent of pH (Fig. 8).

All these processes inuenced the cycling and re-distribution ofnutrients leading to higher nutrient element stocks in surface min-eral soils under aspen and birch, when compared to pure oak plots(Tables S3 and S4) (Bocock and Gilbert, 1957; Harrison, 1971; Manand Lieffers, 1999).

However, depending on the future development of the differentstand types, the re-distribution of nutrients within the ecosystemmay change upon nurse crop removal. Also, the removal of nursecrop biomass from aspen-oak and birch-oak plots (Fig. 2), whencompared to mono-specic oak plots, reduces the ecosystem nutri-ent pools (Fig. 9 and Tables S3S6). Therefore, any reductions in thenutrient costs of biomass removals, such as using nutrient-efcienttree species, will also increase the benets of nurse crop systemson soil fertility (Hendrickson et al., 1987; Ranger and Turpault,1999).

4.3. Nutrient element costs of wood

Nutrient costs of biomass removal can vary greatly betweentree species. The selection of species with comparably low nutrientcosts may be crucial to ensure nutrient sustainability (Wang et al.,1991; Sverdrup et al., 2006; Par et al., 2013). In our study the

H. Stark et al. / Forest Ecology amean Ca, K, and Mg costs of birch stem wood and stem bark wereonly 69%, 47% and 54% Mg of that of aspen (Fig. 10, Tables S18 andS19), while there was a much smaller difference between thesespecies for N and P. When compared to the nutrient element costsfor wood of several other major tree species of economic impor-tance in central Europe, nutrient element costs of birch wood weremostly lower, whereas nutrient element costs for aspen wood wereoften higher (Pastor et al., 1984; Par and Cleve, 1993).

However, in addition to the nutrient costs of wood, the nutrientaccumulation and retention potential of species in the ecosystemshould be considered. The high concentrations of base cations inaspen biomass tissues such as branches, stem wood and stem bark(Tables S7 and S8) contributed substantially to the observed highaccumulation of nutrient elements in top mineral soils (Figs. 5and 9, Tables S3S6) and thus its nutrient accumulation and reten-tion potential (Pastor et al., 1984). In nutrient rich systems, whereleaching following disturbance may take on larger dimensions asthe nutrient loss with removal of biomass (Johnson and Todd,1987), species such as aspen with woody tissues rich in nutrientelements may provide the greatest benets. In nutrient poor sys-tems, where the nutrient loss with removal of biomass may behigher than a potential loss via leaching (Hornbeck et al., 1986),species such as birch with low nutrient element costs of biomassharvest, may be the more suitable option.

Finally, reducing the harvesting intensity from total above-ground woody biomass to stem only harvest (stem wood and bark)would additionally reduce the nutrient element costs of biomassharvest (Tables S18 and S19) (Freedman et al., 1986; Meiwes,2010; Weis and Gttlein, 2012; Merino et al., 2005). Because ofthe species-specic biomass compartmentalization and nutrientelement concentrations (Jacobsen et al., 2003; Rademacher,2005; Andr et al., 2010) the on-site retention of branch biomassachieved a higher reduction of nutrient element costs in birch thanin aspen trees.

Thus, depending on the site-specic management targets, theselection of tree species and the harvest intensity strongly inu-ence the amount of nutrients that is removed during harvestingoperations and its effects on ecosystem nutrient cycling.

4.4. General conclusions

In contrast to many other regions, conventional forest manage-ment in central Europe has paid little attention to pioneer treessuch as aspen and birch. With frequent forest disturbances andincreased demand for woody biomass (Gardinier et al., 2013), pio-neer trees such as aspen and birch may gain more prominence inforest management in central Europe and beyond. In addition toprotecting seedlings of target tree species against environmentalextremes (Keenan et al., 1995; Schmidt-Schtz and Huss, 1998),nurse crops of aspen or birch can benet forest biomass productionand soil fertility. Overall benets may increase when aspen andbirch are established as mixed-species nurse crops with shade-tolerant target tree species. Given a proper management (Unseldet al., 2012), such nurse crop systems may therefore be a suitabletool to increase productivity and short-term biomass supply inlarge forest openings, abandoned agricultural lands or degradedforest sites.

Acknowledgments

Funding was provided by the Energy Research Foundation (SEf),the Ministry of Science, Research and the Arts of the German fed-eral state of Baden-Wrttemberg as well as the Research Institutefor Forest Ecology and Forestry of the German federal state ofRhineland-Palatinate (FAWF). We would like to thank DavidForrester, Partap Khanna, Merle Koelbing, Bruntje Ldtke, Renate

anagement 339 (2015) 5770 67Nitschke, Matthias Schmidt, Volker & Kerstin Stark, Tim Steinkraus,Luise Stephani, Julius Schuck, Gebhard Schler, Rdiger Unseld andKlaus v. Wilpert for their kind support and advice. The constructive

nd Mand detailed comments of the anonymous reviewers helped muchto improve the quality of the manuscript.

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.2014.12.003.

References

Agestam, E., Ek, P.-M., Nilsson, U., Welander, N., 2003. The effects of shelterwooddensity and site preparation on natural regeneration of Fagus sylvatica insouthern Sweden. For. Ecol. Manage. 176 (1), 6173.

Alban, D.H., 1982. Effects of nutrient accumulation by aspen, spruce, and pine onsoil properties. Soil Sci. Soc. Am. J. 46 (4), 853861.

Albrecht, A., Kohnle, U., Hanewinkel, M., Bauhus, J., 2013. Storm damage of douglas-r unexpectedly high compared to norway spruce. Ann. For. Sci. 70 (2), 195207.

Alriksson, A., Eriksson, H.M., 1998. Variations in mineral nutrient and C distributionin the soil and vegetation compartments of ve temperate tree species in NESweden. For. Ecol. Manage. 108 (3), 261273.

Ammer, C., Dingel, C., 1996. Untersuchungen ber den EinuSS starkerWeichlaubholzkonkurrenz auf das Wachstum und die Qualitt jungerStieleichen / Studies on the inuence of intense competition by softwoodbroadleaves on the growth and quality of young oaks (in German).Forstwissenschaftliches Centralblatt joint with Tharandter forstlichesJahrbuch 116 (16), 346358.

Andr, F., Jonard, M., Ponette, Q., 2010. Biomass and nutrient content of sessile oak(Quercus petraea (Matt.) Liebl.) and beech (Fagus sylvatica L.) stem and branchesin a mixed stand in southern Belgium. Sci. Total Environ. 408 (11), 22852294.

Ballard, T., 2000. Impacts of forest management on northern forest soils. For. Ecol.Manage. 133 (1), 3742.

Bauhus, J., Bartsch, N., 1995. Mechanisms for carbon and nutrient release andretention in beech forest gaps. I: microclimate, water balance and seepagewater chemistry. Plant Soil 168, 579584.

Bauhus, J., Messier, C., 1999. Soil exploitation strategies of ne roots in different treespecies of the southern boreal forest of eastern Canada. Can. J. For. Res. 29 (2),260273.

Bauhus, J., Puettmann, K.J., Khne, C., 2013. Is close-to-nature forest management inEurope compatible with managing forests as complex adaptive forestecosystems? In: Messier, C., Puettmann, K.J., Coates, K.D. (Eds.), ManagingForests as Complex Adaptive Systems. Routledge, Oxon, England and New York,US, pp. 187213.

Bazzaz, F.A., 1979. The physiological ecology of plant succession. Annu. Rev. Ecol.Syst. 10, 351371.

Bergqvist, G., 1999. Wood volume yield and stand structure in norway spruceunderstorey depending on birch shelterwood density. For. Ecol. Manage. 122(3), 221229.

Binkley, D., 2003. Seven decades of stand development in mixed and pure stands ofconifers and nitrogen-xing red alder. Can. J. For. Res. 33 (11), 22742279.

Binkley, D., Giardina, C., 1998. Why do tree species affect soils? The warp and woofof tree-soil interactions. Biogeochemistry 42 (1), 89106.

Block, J., Schuck, J., Seifert, T., 2007. Einuss der waldbaulichen Behandlung und derHolznutzung auf den Nhrstoffhaushalt von Traubeneichenkosystemen /Inuence of silvicultural treatment and wood utilization on the nutrientcycling in sessile oak ecosystems. In: Dong, P.H., (Eds.), Eiche im Pflzerwald /Oak in Rhineland-Palatinates Forest, Publications of the Research Institute forForest Ecology and Forestry Rhineland-Palatinate. Research Institute for ForestEcology and Forestry Rhineland-Palatinate 2007, 63, 117150 (in German).

BMELV, 2002. Zweite Bundeswaldinventur/Second German National ForestInventory (in German). .

Bocock, K., Gilbert, O., 1957. The disappearance of leaf litter under differentwoodland conditions. Plant Soil 9 (2), 179185.

Bsch, B., 2013. WEHAM Modelle und Algorithmen/WEHAM Models andAlgorithms (in German). Technical Report. Department of Biometrics andInformatics, Forest Research Institute Baden-Wrttemberg and Federal Ministryof Food, Agriculture and Consumer Protection (BMELV). .

Brang, P., Spathelf, P., Larsen, J.B., Bauhus, J., Boncna, A., Chauvin, C., Drssler, L.,Garca-Gemes, C., Heiri, C., Kerr, G., et al., 2014. Suitability of close-to-naturesilviculture for adapting temperate european forests to climate change.Forestry, cpu018.

Bretz, F., Hothorn, T., Westfall, P., 2010. Multiple Comparisons Using R. CRC Press.Carlson, D.W., Groot, A., 1997. Microclimate of clear-cut, forest interior, and small

openings in trembling aspen forest. Agric. For. Meteorol. 87 (4), 313329.Comeau, P., Heineman, J., Newsome, T., 2006. Evaluation of relationships between

understory light and aspen basal area in the British Columbia central interior.

68 H. Stark et al. / Forest Ecology aFor. Ecol. Manage. 226 (13), 8087.Comeau, P.G., Filipescu, C.N., Kabzems, R., DeLong, C., 2009. Corrigendum to:

Growth of white spruce underplanted beneath spaced and unspaced aspenstands in northeastern B.C.10 year results. For. Ecol. Manage. 257 (7), 16291636.

Cotta, H., 1828. Anweisung zumWaldbau / Instructions for silviculture (in German).Arnoldsche Buchhandlung, Dresden & Leipzig, Germany.

Dong, P., Eder, W., Muth, M. 2007. Traubeneichen-Durchforstungsversuche imPflzerwald / Thinning experiments in sessile oak stands the Rhineland-Palatine. In: Dong, P.H., (Eds.), Eiche im Pflzerwald / Oak in Rhineland-Palatinates Forest, Publications of the Research Institute for Forest Ecology andForestry Rhineland-Palatinate. Research Institute for Forest Ecology andForestry Rhineland-Palatinate, 63, 96116 (in German).

Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carr, G., Marquz, J.R.G.,Gruber, B., Lafourcade, B., Leito, P.J., Mnkemller, T., McClean, C., Osborne,P.E., Reineking, B., Schrder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013.Collinearity: a review of methods to deal with it and a simulation studyevaluating their performance. Ecography 36 (1), 027046.

Drouineau, S., Laroussinie, O., Birot, Y., Terrasson, D., Formery, T., Roman-Amat, B.,2000. Joint Evaluation of Storms, Forest Vulnerability and Their Restoration.Technical Report. European Forest Institute Joensuu. .

EFI, 2013. Database on Forest Disturbances in Europe. .

Eggert, M., 2006. Begrndung von Eiche mit und ohne Vorwald aufSturmwurfchen in Rheinland-Pfalz/Establishment of Oak with and WithoutNurse Crops at Forest Sites Disturbed by Storms in Rhineland-Palatinate (inGerman). Masters Thesis, University of Freiburg.

FAWF, 2013. Daten zu Dauerversuchschen/Permanent Plot Information. DataUsed in This Study are Available Upon Request (in German). Technical Report.Research Institute for Forest Ecology and Forestry Rhineland-Palatinate,Trippstandt, Germany. November. .

Forrester, D.I., 2014. The spatial and temporal dynamics of species interactions inmixed-species forests: From pattern to process. For. Ecol. Manage. 312, 282292.

Freedman, B., Duinker, P., Morash, R., 1986. Biomass and nutrients in Nova Scotianforests, and implications of intensive harvesting for future site productivity. For.Ecol. Manage. 15 (2), 103127.

FVA-BW, 2000. Yield Table for Oak in Baden-Wrttemberg. Technical Report. ForestResearch Station Baden-Wrttemberg.

Gale, M.R., Grigal, D.F., 1987. Vertical root distributions of northern tree species inrelation to successional status. Can. J. For. Res. 17 (8), 829834.

Gardiner, E.S., Stanturf, J.A., Schweitzer, C.J., 2004. An afforestation system forrestoring bottomland hardwood forests: biomass accumulation of Nuttall oakseedlings interplanted beneath eastern cottonwood. Restor. Ecol. 12 (4), 525532.

Gardinier, B., Schuck, A., Schelhaas, M.-J., Orazio, C., Blennow, K., Noicoll, B., (Eds.),2013. Living with Storm Damage. European Forest Institute, Joensuu, Finland.

Groot, A., Carlson, D.W., 1996. Inuence of shelter on night temperatures, frostdamage, and bud break of white spruce seedlings. Can. J. For. Res. 26 (9), 15311538.

Grote, R., 2003. Estimation of crown radii and crown projection area from stem sizeand tree position. Ann. For. Sci. 60 (5), 393402.

Gutsell, S., Johnson, E., 2002. Accurately ageing trees and examining their height-growth rates: implications for interpreting forest dynamics. J. Ecol. 90 (1), 153166.

Harrison, A., 1971. The inhibitory effect of oak leaf litter tannins on the growth offungi, in relation to litter decomposition. Soil Biol. Biochem. 3 (3), 167172.

Hendrickson, O., Burgess, D., Chatarpaul, L., 1987. Biomass and nutrients in GreatLakes-St. Lawrence forest species: implications for whole-tree and conventionalharvest. Can. J. For. Res. 17 (3), 210218.

Hendrickson, O., Chatarpaul, L., Burgess, D., 1989. Nutrient cycling following whole-tree and conventional harvest in northern mixed forest. Can. J. For. Res. 19 (6),725735.

Hornbeck, J.W., Martin, C.W., Pierce, R.S., Bormann, F.H., Likens, G.E., Eaton, J.S.,1986. Clearcutting northern hardwoods: effects on hydrologic and nutrient ionbudgets. For. Sci. 32 (3), 667686.

Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in generalparametric models. Biometrical J. 3 (50), 346363.

Jacobsen, C., Rademacher, P., Meesenburg, H., Meiwes, K.J., 2003. Gehaltechemischer Elemente in Baumkompartimenten Literaturstudie undDatensammlungs / Element concentrations in tree biomass compartments:literature review and data collection (in German). Ber. ForschungszentrumsWaldkosysteme 69 (B).

Johansson, T., 2003. Mixed stands in nordic countriesa challenge for the future.Biomass Bioenergy 24 (45), 365372.

Johnson, D., Todd, D., 1987. Nutrient export by leaching and whole-tree harvestingin a loblolly pine and mixed oak forest. Plant Soil 102 (1), 99109, ISSN 0032-079X.

Kabzems, R., Nemec, A.L., Farnden, C., 2007. Growing trembling aspen and whitespruce intimate mixtures: early results (1317 years) and future projections. J.Ecosyst. Manage. 8 (1).

Karacic, A., Verwijst, T., Weih, M., 2003. Above-ground woody biomass productionof short-rotation Populus plantations on agricultural land in Sweden. Scand. J.

anagement 339 (2015) 5770For. Res. 18 (5), 427437.Keenan, R., Lamb, D., Sexton, G., 1995. Experience with mixed species rainforest

plantations in North Queensland. Commonw. Forestry Rev. 74 (4), 315321.

nd MKnig, N. (Ed.), 2005. Handbuch Forstliche Analytik/Handbook of Forestry Analytics(in German). Federal Ministry of Food, Agriculture and Consumer Protection(BMELV). .

Kraus, D., Krumm, F., (Eds.), 2013. Integrative Approaches as an Opportunity for theConservation of Forest Biodiversity. European Forest Institute, Joenssuu.

Lieffers, V.J., Stadt, K.J., Navratil, S., 1996. Age structure and growth of understorywhite spruce under aspen. Can. J. For. Res..

Liesebach, M., von Wuehlisch, G., Muhs, H., 1999. Aspen for short-rotation coppiceplantations on agricultural sites in Germany: effects of spacing and rotationtime on growth and biomass production of aspen progenies. For. Ecol. Manage.121 (121), 2539.

Litton, C.M., Raich, J.W., Ryan, M.G., 2007. Carbon allocation in forest ecosystems.Glob. Change Biol. 13 (10), 20892109.

Lundmark, T., Hllgren, J.-E., 1987. Effects of frost on shaded and exposed spruceand pine seedlings planted in the eld. Can. J. For. Res. 17 (10), 11971201.

Maard, H., 1996. The inuence of a birch shelter (Betula spp.) on the growth ofyoung stands of Picea abies. Scand. J. For. Res. 11 (14), 343350.

Majunke, C., Matz, S., Mller, M., 2008. Sturnschden in Deutschland von 1920 bis2007/Storm Disturbance in Germany Between 1920 and 2007 (in German). AFZ-Der Wald.

Man, R., Lieffers, V.J., 1997. Seasonal photosynthetic responses to light andtemperature in white spruce (Picea glauca) seedlings planted under an aspen(Populus tremuloides) canopy and in the open. Tree Physiol. 17 (7), 437444.

Man, R., Lieffers, V.J., 1999. Are mixtures of aspen and white spruce more productivethan single species stands? Forestry Chron. 75 (3), 505513.

Mander, U., Jogman, R. (Eds.), 2000. Consequences of Land Use Changes. WessexInstitute of Technology, Boston, MA, USA.

Mantau, U., 2012. Wood Raw Material Balance Germany, Development andScenarios of Timber Supply and use From 1987 to 2015. Technical Report.Department of Wood Science, University Hamburg.

Meiwes, K.J., 2010. Versuche zur Vollbaumnutzung Ein Ansto fr eingemeinsames Konzept der Forstlichen Versuchsanstalten/Trials on whole treeharvest towards a common approach among the forest research stations inGermany (in German). In: Nagel, J. (Ed.), Section Yield Science, Annual Meeting,pp. 169172.

Merino, A., Balboa, M., Rodrguez Soalleiro, R., Gonzlez, J., 2005. Nutrient exportsunder different harvesting regimes in fast-growing forest plantations insouthern Europe. For. Ecol. Manage. 207 (3), 325339.

Messier, C., Puettmann, K.J., Coates, K.D., (Eds.) 2013. Managing Forests as ComplexAdaptive Systems: Building Resilience to the Challenge of Global Change.Routledge. .

Mnder, K., 2005. Konkurrenzuntersuchungen und Wachstumsmodellierung inWaldumbaubestnden des Mittleren Erzgebirges/Analysis of Competition andGrowth Modelling of Converted Forest Stands in the Middle Ore Mountains (inGerman). PhD Thesis. TU Dresden, Tharandt.

Nelson, A., Saunders, M., Wagner, R., Weiskittel, A., 2012. Early stand production ofhybrid poplar and white spruce in mixed and monospecic plantations ineastern Maine. New Forest. 43 (4), 519534.

Nielsen, U.B., Madsen, P., Hansen, J.K., Nord-Larsen, T., Nielsen, A.T., 2014.Production potential of 36 poplar clones grown at medium length rotation indenmark. Biomass Bioenergy 64, 99109.

Nilsson, C., Stjernquist, I., Brring, L., Schlyter, P., Jnsson, A.M., Samuelsson, H.,2004. Recorded storm damage in Swedish forests 19012000. For. Ecol. Manage.199 (1), 165173.

Nutto, L., 1999. Neue Perspektiven fr die Begrndung und Pege von jungenEichenbestnden / New Perspectives for the Establishment and Treatment ofYoung Oak Stands, Freiburger Forstliche Forschung vol. 5, p. 190 (in German)University of Freiburg and Forest Research Station Baden-Wrttemberg,Germany (in German).

rlander, G., 1993. Shading reduces both visible and invisible frost damage tonorway spruce seedlings in the eld. Forestry 66 (1), 2736.

rlander, G., Karlsson, C., 2000. Inuence of shelterwood density on survival andheight increment of Picea abies advance growth. Scand. J. For. Res. 15 (1), 2029.

Osterburg, B., Rter, S., Freibauer, A., de Witte, T., Elsasser, P., Ktsch, S., Leischner,B., Paulsen, H., Rock, J., Rder, N., Sanders, J., Schweinle, J., Steuk, J., Stichnothe,H., Stmer, W., Welling, J., Wolff, A., 2013. Handlungsoptionen fr denKlimaschutz in der deutschen Agrar- und Forstwirtschaft/Strategies forClimate Change Mitigation on Agriculture and Forestry Lands in Germany (inGerman). Thnen Report 11. Johann Heinrich von Thnen-Institut,Braunschweig, German, 2013. .

Par, D., Cleve, K.V., 1993. Aboveground biomass production and nutrientaccumulation on postharvested white spruce sites in interior Alaska. Can. J.For. Res. 23 (6), 12331239.

Par, D., Bernier, P., Laeur, B., Titus, B.D., Thiffault, E., Maynard, D.G., Guo, X., 2013.Estimating stand-scale biomass, nutrient contents, and associated uncertaintiesfor tree species of Canadian forests. Can. J. For. Res. 43 (7), 599608.

Pastor, J., Aber, J.D., Melillo, J.M., 1984. Biomass prediction using generalizedallometric regressions for some northeast tree species. For. Ecol. Manage. 7 (4),265274.

Perala, D.A., Alm, A.A., 1990. Regeneration silviculture of birch: a review. For. Ecol.

H. Stark et al. / Forest Ecology aManage. 32 (1), 3977.Petterson, 1955. The mass production of coniferous forests. Rep. Swedish For. Res.

Station 45.Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Development Core Team. 2011. nlme:Linear and Nonlinear Mixed Effects Models. R. .

Pommerening, A., Murphy, S.T., 2004. A review of the history, denitions andmethods of continuous cover forestry with special attention to afforestation andrestocking. Forestry 77 (1), 2744.

Poorter, H., Niklas, K.J., Reich, P.B., Oleksyn, J., Poot, P., Mommer, L., 2012. Biomassallocation to leaves, stems and roots: meta-analyses of interspecic variationand environmental control. New Phytol. 193 (1), 3050.

Prescott, C.E., 2002. The inuence of the forest canopy on nutrient cycling. TreePhysiol. 22 (1516), 11931200.

Pretzsch, H., Block, J., Bttcher, M., Dieler, J., Gauer, J., Gttlein, A., Moshammer, R.,Schuck, J., Weis, W., Wunn, U., 2013. Entscheidungssttzungssystem zumNhrstoffentzug im Rahmen der Holzernte - Nhrstoffbilanzen wichtigerWaldstandorte in Bayern und Rheinland-Pfalz. / Decision support system fornutrient element removals during biomass harvest - nutrient element balancesof important forest sites in Bavaria and Rhineland-Palatinate. Final projectreport 25966-33/0, Deutsche Bundesstiftung Umwelt (DBU), pp. 204 (inGerman). .

R. R: A Language and Environment for Statistical Computing. R Foundation forStatistical Computing, Vienna, Austria, 2010. .

Rademacher, P., 2005. Contents of nutrient elements in tree components ofeconomical important species in relation to their residual utilisation. Eur. J.Wood Prod. 63 (4), 285296.

Ranger, J., Turpault, M.-P., 1999. Input-output nutrient budgets as a diagnostic toolfor sustainable forest management. For. Ecol. Manage. 122 (12),139154.

Reich, P.B., Oleksyn, J., Modrzynski, J., Mrozinski, P., Hobbie, S.E., Eissenstat, D.M.,Chorover, J., Chadwick, O.A., Hale, C.M., Tjoelker, M.G., 2005. Linking littercalcium, earthworms and soil properties: a common garden test with 14 treespecies. Ecol. Lett. 8 (8), 811818.

Richards, A.E., Forrester, D.I., Bauhus, J., Scherer-Lorenzen, M., 2010. The inuence ofmixed tree plantations on the nutrition of individual species: a review. TreePhysiol. 30 (9), 11921208.

Ripley, B., Lapsley, M., 2011. RODBC: ODBC Database Access. R Package Version1.3-3.

Rothe, A., Binkley, D., 2001. Nutritional interactions in mixed species forests: asynthesis. Can. J. For. Res. 31 (11), 18551870.

Saetre, P., Brandtberg, P.O., Lundkvist, H., Bengtsson, J., 1999. Soil organisms andcarbon, nitrogen and phosphorus mineralisation in norway spruce and mixednorway sprucebirch stands. Biol. Fert. Soils 28 (4), 382388.

Saha, S., Kuehne, C., Bauhus, J., 2013. Intra- and interspecic competition differentlyinuence growth and stem quality of young oaks (Quercus robur L. and Quercuspetraea (Mattuschka) Liebl.). Ann. For. Sci., 113.

Schelhaas, M.-J., Nabuurs, G.-J., Schuck, A., 2003. Natural disturbances in theEuropean forests in the 19th and 20th centuries. Glob. Change Biol. 9 (11),16201633, ISSN 1365-2486.

Schmidt-Schtz, A., Huss, J., 1998. Wiederbewaldung von Fichten-Sturmwurfchen auf vernssenden Standorten mit Hilfe vonPioniergehlzen/Restoration of Norway Spruce-Storm Catastrophic Areas onHydromorphic Soils with Pionieer Crops (in German). Technical Report.Silviculture Institute, University Freiburg, Freiburg.

Shepperd, W.D., Jones, J.R., 1985. Nurse crop. In: DeByle, N., Winokur, R. (Eds.),Aspen: Ecology and Management in the Western United States. GeneralTechnical Report RM-119. USDA Forest Service, Rocky Mountain Forest andRange Experiment Station, Fort Collins, Colorado, pp. 181188.

Sloboda, B., 1971. Zur Darstellung von Wachstumsprozessen mit Hilfe vonDifferentialgleichungen erster Ordnung/Representation of Growth Processeswith the Help of First Order Differential Equations (in German). Publications ofthe Department of Biometrics and Informatics, Forest Research Institute Baden-Wrttemberg. Department of Biometrics and Informatics, Forest ResearchInstitute Baden-Wrttemberg.

Spiecker, H., 2007. Zur