The influence of changes in forest manag
Forest Ecology and Management
journalhomepage:www.elsevier.com/locate/foreco
The influence of changes in forest management over the past 200 years on present soil organic carbon stocks
Jana Wäldchen ⇑ , Ernst-Detlef Schulze, Ingo Schöning, Marion Schrumpf, Carlos Sierra
Max Planck Institute for Biogeochemistry, Hans Knöll Str. 10, 07745 Jena, Germany
article info
abstract
Article history: Forest ecosystems in Europe have been affected by human activities for many centuries. Here we inves- Received 25 April 2012
tigate, if current forest soil organic carbon stocks are influenced not only by present ecological conditions Received in revised form 4 October 2012
Accepted 7 October 2012 and land use, but also by land management in the past. Based on the forest management history of the
Available online 28 November 2012 Hainich-Dün region a total of 130 inventory plots were selected in age-class forest and selectively cut for- ests under present management practice. The age-class forest originated from (1) former coppice-with-
Keywords: standards, (2) former selectively cut forests and (3) afforestation. The selectively cut forest contains
Soil organic carbon ‘‘early regulated’’ forest where selective cutting has been practised for centuries, and forest, which was
Carbon stocks managed as coppice-with-standards through the 18th and the 19th centuries. We hypothesise that past Light fraction
management influences present soil organic carbon stocks. Density fractionation of soils in three physical Heavy fraction
fractions (HF: heavy fraction, o-LF: occluded light fraction, f-LF: free light fraction) was carried out to Forest management
increase the probability of detecting long-lasting effects of management history. No detectable differ- Historical forest management
ences in soil organic carbon (SOC) stocks, as measured in kg m
ground area, of the mineral soil and the heavy fractions, were found between present and historical forest management types (average total organic carbon (OC) stocks of mineral soil: 9.7 ± 2.3 kg m ; average OC stocks of the organic layer:
0.5 ± 0.3 kg m ; average total inorganic carbon (IC) stocks of mineral soil: 5.0 ± 3.7 kg m ). The varia- tion of samples was overlapping. There was no consistent trend with management history. The upper mineral soil (0–30 cm) contained about 74% of total SOC, with f-LF contributing 24% in 0–10 cm and 20% in 10–30 cm, and o-LF 9% in 0–10 cm and 6% in 10–30 cm. The HF contained 85% (0–10 cm) and 86% (10–30 cm) of SOC stocks in the bulk soil. There was a significant decrease of total SOC stocks in
the 0–10 and 10–30 cm depth increment with increasing abundance of beech. Mean 14 C concentrations in the HF were 102.0 pMC in 0–10 cm, and 93.4 pMC in 10–30 cm, corresponding to a mean 14 C age of around 100 years and 550 years, respectively. Modelling C-dynamics based on the present measurements reveals that disturbances depleting 50% of soil C-stocks would equilibrate after 80 years. Thus, there is no memory effect of 19th century forest management. We conclude that past and present management has no detectable effect on present SOC.
Ó 2012 Elsevier B.V. All rights reserved.
1. Introduction Most studies about the historical influence of human activities on forest soils have been carried out with the aim of investigating Forest use and management is generally thought to create ma-
the impact of afforestation and deforestation. These studies show jor ecosystem disturbance comparable to natural events such as
that management practices introduced in the 19th and 20th centu- storms and forest fires ( Buergi and Gimmi, 2007 ). Thus, there is a
ries reduced soil organic carbon (SOC) stocks ( Compton et al., growing interest in quantifying land-use intensity ( Luyssaert
1998; Gragson and Bolstad, 2006; Guo and Gifford, 2002; Verheyen et al., 2011 ) and the effects of human activity on forest ecosystem
et al., 1999 ). Not only agricultural use, but also different types of structure and functioning. Most forest ecosystems in Europe have
forest management, by removing different amounts of biomass been affected by human activities for centuries. At present it is
and timber could have affected soil organic carbon stocks ( Jandl not clear how long major changes in land-use intensity persist,
et al., 2007; Mund and Schulze, 2006 ).
nor if there are ‘‘memory’’ signals of past land use in present forest The separation of soil organic matter (SOM) into physical soils.
fractions with different turnover times is expected to increase the probability of detecting historic management effects. The
⇑ Corresponding author. Tel.: +49 3641 576213; fax: +49 3641 577102. light fraction (LF) of organic matter consists of weakly decom- E-mail address: jwald@bgc-jena.mpg.de (J. Wäldchen).
posed plant and animal tissues and is generally characterised 0378-1127/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.foreco.2012.10.014
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
by a rapid turnover of <10 years. The amount of organic carbon
2.2. Forestry history and forest management types (OC) held in the light fraction is therefore controlled by the pres- ent forest management. In contrast, the mineral associated or-
Our study is based on documented forest management changes ganic matter (or heavy fraction, HF) is stabilised through
between 1800 and 1900 ( Fig. 1 a). At the beginning of the 19th cen- interaction with the mineral surfaces and characterised by turn-
tury, most of the forest sites in the Hainich-Dün region were under over times ranging from decades to centuries ( Baisden et al.,
the coppice-with-standards system, a silvicultural system in which 2002; Crow et al., 2007; Ellert and Gregorich, 1995; Janzen
timber trees with an open canopy are grown above a coppiced et al., 1992; Six et al., 2002; Yamashita et al., 2006 ). Therefore
woodland. Small areas of selectively cut forests were also present, if long-term effects of historical forest management exist, they
with selective harvesting of single trees and irregular forest use. In should be visible in the HF fraction.
the 19th century, all coppice-with-standards forests were con- Forest management has changed from practices such as coppice
verted to age-class forest or to selectively cut forest ( Wäldchen forestry or coppice-with-standards, towards modern forestry,
et al., 2011 ). The forest under age-class management is character- which aims at sustainable production of wood. To our knowledge,
ised by a sequence of relatively homogenous, even-aged stands. the effects of past forest management on present SOC stocks have
Coppice-with-standards management is no longer practiced in not yet been studied.
the Hainich-Dün region. However, remnants of coppice forest There are historical reports that unregulated use of coppice
and coppice-with-standards forest still exist in other parts of Thu- and coppice-with-standards forest led to degradation of these
ringia, being managed as historic reserves, but these sites grow on stands ( Hasel and Schwarz, 2006 ). Based on these reports of for-
shallow Leptosols. In this study these sites were only used for leaf estry in the 18th and 19th century, we hypothesise that stands
area index (LAI) measurements.
which experienced intensive wood extraction 100–200 years ago, still exhibit lower carbon stocks, especially in the mineral- associated heavy fraction of SOM. Following this initial hypothe-
2.3. Plot selection
sis we argue that a former coppice-with-standards stand should now have lower SOC stocks than less intensively used, former
The experimental design is based on an initial estimate of the selectively cut forests, where only single trees are cut and re-
number of plots which would be needed to detect a certain differ- moved. In addition, stands where management changed from
ence of carbon stocks between sites at a certain variation between coppice-with-standards to a less intensive regime in the distant
samples (Minimum detectable difference approach ( Zar, 1984 )), past should have higher SOC contents than stands where man-
based on known present soil carbon accumulation rates in this re- agement changed more recently. We assume that present SOC
gion ( Schulze et al., 2010 ). We estimated that 10–30 repetitions stocks are influenced not only by present, but also by historical
differences in the Ah, 0–10 forest management.
would be sufficient to detect 1 kg m
and 10–30 cm depth increments between treatments, after We investigate SOC stocks in bulk soil and in the different den-
100 years carbon accumulation. This difference was expected to sity fractions in forest stands in Thuringia, Germany, where
exist between management types 100–200 years ago. changes in forest management can be dated from the operational
The investigation plots were distributed across four forest dis- management plans of the forest administrations since 1800
tricts, which had independent administrations and history. Thus, ( Wäldchen et al., 2011 ). Management has changed from coppice-
we are confident in having independent samples on six different with-standards forest into high forest; in comparison to forest
forest sites, where four sites are managed today as age-class forest stands, which were used for more than 250 years as high forests
(ACF) and two sites are selectively cut forests (SF). The age-class of continuous cover (selective cuttings).
forest comprised (1) 11 plots that were used as coppice-with- The overall objective of this study is to quantify the influence of
standards (CwS) until the beginning of the 19th century, (2) 35 former and present forest management on the SOC stocks of the
plots that were used as coppice-with-standards until the beginning forests of the Hainich-Dün region, thereby enhancing the under-
of the 20th century, (3) 46 plots that were used as selectively cut standing of the ecosystem processes that link forest management
forests until the beginning of the 19th century and (4) 8 plots that with changes in SOC stocks.
were afforested (AF) in the second half of the 19th century. The two presently selectively cut forest sites contain ‘‘early regulated’’ for- est where selective cutting has been applied for centuries (16 plots), and forest that was used as coppice-with-standards forest
2. Materials and methods in the 18th and 19th century (14 plots). Thus, in total 130 plots were sampled as shown in Fig. 1 a in an overview of the forest sites
2.1. Study sites and their management histories. All study plots that belonged to the same forest type were located within an area of about 10–
This study was conducted in the Hainich-Dün region of Thurin-
20 km 2 . The distance between the forest sites in the south
(140 years SF/CwS) and the forest sites at the Dün (110 years 494 m above sea level, the mean annual precipitation from 600
gia, Germany (51°12 0 N 10°18 0 E). Elevations range from 100 to
ACF/CwS) was about 20 km ( Fig. 1 b).
to 800 mm and the mean annual temperature from 6 to 7.5 °C Following our initial hypothesis the 140 year-old coniferous afforestation of degraded grassland was expected to have lower parent material is Triassic limestone, which is covered by a Pleisto-
soil carbon stocks than 110 year-old broadleaved forest following cene loess layer of variable thickness (ca. 10–50 cm) at most sites.
coppice-with-standards, and this should be lower than 140 years Main soil groups of the study area are Cambisols, Luvisols and
selectively cut forest or 210 year age-class forest following cop- Stagnosols ( Grüneberg et al., 2010 ). The humus form (type of the
pice-with-standards. Highest carbon stocks were expected in organic layer) varied between L-mull and F-mull. The climate
210 years age-class forest following a selectively cut forest regime and soil conditions of the region provide optimum growing condi-
and the more than 250 years-old selectively cut forest. The black tions for beech (Fagus sylvatica) dominated forests. Admixed tree
bar in Fig. 1 a represents the increasing OC-stocks as expected from species are Fraxinus excelsior, Acer pseudiplatanus and Acer platano-
our initial hypothesis. For clarity, we divided the six investigation ides. Grasslands have been afforested with spruce (Picea abies) in
sites into four groups: I) afforestation, II) late changes from CwS the past ( Wäldchen et al., 2011 ).
into ACF or SF, III) early changes from CwS into ACF and IV) forest
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
(a)
(b)
Fig. 1. (a) Overview of the investigation sites with their present and historical forest management types, the time of management changes and the number of observations and (b) Overview of the location of the study sites within the Hainich-Dün region.
sites which have been managed for more than 250 years as high
Age ¼ 17:5 þ 2:6 DBH
forest.
¼ 0:58; p < 0:05
2.4. Forest inventory, woody biomass, mean stand age, land use and To quantify the management intensity of these forests, the
disturbance index (LUDI) Land-Use and Disturbance Index, LUDIp was used ( Luyssaert
et al., 2011 ). This index quantifies the long-term management decision at which height and density this forest shall produce
certain products. LUDIp is expressed in relation to the natural (DBH)>7 cm, DBH, tree height and tree species were recorded. Thus,
At all 130 plots a forest inventory was carried out on circles with an area of 500 m 2 . For all trees with a diameter at breast height
self-thinning line, and in relation to the stand density (N) of each inventory point yields information about stand density and
undisturbed pristine forests. For every plot LUDIp was calculated diameter distribution. The conversion into wood volume follows
as (Eq. (2) ):
allometric relationships, which include the shape of the forest trunk ð2Þ ( Lockow, 2003; Gerold, 1977; Bergel, 1973 ). For the conversion of
woody biomass into carbon pools, we assumed an average carbon where d AB is the distance on the self-thinning line between any two concentration of 50% of dry weight ( Wirth et al., 2003 ).
points A and B with stand density N A and N B , and d max gives the Mean stand age for all plots and management types was calcu-
length of the self-thinning curve between N min and N max (N min = lated according to Schulze et al. (2012) (Eq. (1) ), which is based on
or 10 trees ha , representing the lower boundary, and measured tree ages.
0.001 m
N max = 0.36 m
or 3600 trees ha ).
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
2.5. LAI measurements sample was strongly stirred and centrifuged (90 min), where HF settled as pellets. Sodium polytungstate was removed from all
LAI was measured with the LAI-2000 (LICOR, Lincoln, Nebraska, fractions by washing with Millipore water. Samples were then USA) on the following sites: 140 years ACF/Pasture (Group I),
freeze-dried. All density fractions were analysed for total and 210 years ACF/CwS (Group III)>250 years SF and 210 years ACF/SF
organic carbon, and total nitrogen, as described for the bulk soil. (Group IV). Additionally, we measured the LAI on remnants of cop-
pice forest (Tännreisig: 50°46 0 N 11°3 0 E, Hayn: 50°48 0 N 10°54 0 E)
2.6.4. Radiocarbon analysis
After density fractionation radiocarbon ( 14 C) concentrations of These three sites are not located in the Hainich-Dün region. 15 un-
and coppice-with-standards forest (Gottesholz: 50°46 0 N 11°3 0 E).
all HF fractions were determined with accelerator mass spectros- der-canopy measurements per site were taken to achieve an aver-
copy (AMS) in Jena, Germany ( Steinhof et al., 2004 ). Ground sam- age LAI for each location. A second LAI-2000 unit, cross-calibrated
ples were combusted to CO 2 and a small part of the sample mass with the former, was used to automatically take ‘‘above-canopy’’’
was analysed for its stable carbon isotope ratio (d 13 C). The remain- readings from a nearby clearing or agricultural field.
ing major part of the CO 2 was reduced to graphite by heating a mixture of H 2 and CO 2 with Fe powder at 650 °C. The resulting
2.6. Soil analysis graphite-coated iron was pressed into targets and measured in the AMS facility for 14 C. The radiocarbon activity is expressed as
D 14 C, the difference in parts per thousand (‰) between the At each sampling point the organic layers were sampled with a
2.6.1. Soil sampling
14 C/ 12 C ratio in the sample compared to that of the standard oxalic acid ( Trumbore, 2009 ). All values were corrected for fractionation
pled down to the bedrock or 110 cm depth with a motor-driven using the d 13 C values. The 14 C content was expressed in% modern soil column cylinder (Cobra, Eijkelkamp, 8.3 cm in diameter). Due
C (pMC). The average error was 0.25 pMC. Radiocarbon ages were to the significant effect of soil-types on SOC stocks ( Batjes, 2002;
estimated using the OxCal program ( Bronk, 2001 ) and the IntCal09 Grüneberg et al., 2010 ) only Luvisols were considered for this study
calibration curve ( Heaton et al., 2009 ).
(WRB, 2006). The soil cores were divided into depth increments of 0–10, 10–30, 30–50, 50–70, 70–90, and 90–110 cm. Depending on
2.7. Statistics
the thickness of the Ah horizon, either the 0–10 cm or 10–30 cm layer was further divided into two parts to allow separate samples
Statistical analyses were performed in the R environment (R of the Ah-horizon.
Development Core Team, 2011). All measured variables were first characterised by classical descriptive statistics (means and stan-
2.6.2. Soil preparation and SOC analysis dard deviation). Prior to analyses, we examined for normality by Mineral soil samples were air-dried and sieved to <2 mm to re-
diagnostic Quantile–Quantile plots and the one-sample Kolmogo- move roots and stones. Dry root mass was measured. An aliquot of
rov–Smirnov procedure. Equality of variances was examined by the soil sample was ground and analysed for total carbon and
the Levenes test. Not normally distributed data (SOC stocks in nitrogen by dry combustion with a C:N analyser ‘‘Vario Max’’ (Ele-
10–30 cm) were log-transformed to achieve normal distribution. mentar Analysensysteme GmbH, Hanau, Germany). Inorganic car-
One-way ANOVA was used to compare study sites with different bon was determined by measuring the total amount of carbon
forest management history. We used an ANCOVA analysis to ex- after removal of SOC by ignition of samples for 16 h at 450 °C.
plore the influence of different factors (forest stand characteristics, SOC concentrations were calculated from the difference between
soil properties and management history) on SOC stocks in the bulk total and inorganic carbon concentrations. SOC stocks (kg m )
soil and in density fractions. Given the limited number of investi- were calculated as the product of SOC concentration, the weight
gation plots (n = 130 for bulk soil and 25 for the different fractions), of the air-dried fine earth and its volume. Several interlaboratory
to avoid over-fitting we restricted the number of predictors in the tests by the Routine Measurements and Analyses (Roma) Depart-
full model to 10 (see Table 7 ) after removing highly correlated ment of the Max Planck Institute for Biogeochemistry have mea-
independent variables (multi-colinearity). sured a variation coefficient of <1% for OC concentration
measurements.
2.8. Modelling soil organic carbon recovery after management changes
2.6.3. Density fractionation To model SOC recovery after management changes in the 0– For all six study sites, five plots were randomly selected for
10 cm depth increment, we used a single compartment model of measurement of the density fractionation of soil in the Ah-Horizon,
the form:
0–10 and 10–30 cm depth increment (in total, samples from 30 plots were fractionated). SOM was separated into a free light
dC
ð3Þ fraction (f-LF), an occluded light fraction (o-LF), and a mineral-
dt
associated fraction (or HF) following the procedure of Don et al. where C is the amount of carbon in the bulk soil, I is the input of or- (2007) in which 7–10 g air-dried soil (<2 mm) were placed in a
ganic carbon and k is the decay constant. We assumed that: (1) an- centrifuge with 100 ml sodium polytungstate solution with a den-
nual carbon inputs to SOC pools have been constant since sity of 1.6 g cm . The suspension was treated with an ultrasonic
management change, and (2) SOC pools are at a steady state (Inpu- beak (60 J ml ) to crack unstable macro-aggregates followed by
t = Output). Accordingly, the k constant is calculated as follows:
30 min in the centrifuge at 3000 rpm. The organic particles (f-LF), which were floating on the polytungstate solution, were trans-
ð4Þ ferred to a glass fibre filter with a pipette. The remaining sample
k¼
was treated again with an ultrasonic beak (450 J ml ) to separate where R is the heterotrophic respiration flux from the 0 to 10 cm occluded organic matter from the mineral fraction. The sample
depth increment per year. Respiration fluxes in soils of the Hainich suspension was centrifuged again (30 min at 3000 rpm). Floating
year , which were estimated particles of the o-LF were removed from the solution and trans-
region in 0–10 cm were 136 g C m
from incubation experiments of soils in the Hainich National Park ferred to a glass fibre filter. To separate the heavy, mineral associ-
( Kutsch et al., 2010 ). Dissolved organic carbon (DOC) transport ated fraction (HF) from the polytungstate solution, the remaining
within the soil profile ( Kindler et al., 2011 ) was neglected, since it
247 was shown that DOC transport did not affect SOC ( Kahl, 2008 ).
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
coppice-with-standards forest (4.0 ± 0.4 without understorey and Based on mean SOC stocks in 0–10 cm in the old selectively cut for-
5.4 ± 0.2 with understorey).
ests (Group IV) of 3.7 kg m
The correlation matrix of stand-level predictors ( Table 2 ) ran the model with two different assumptions: carbon stocks of
we estimated a k value of 0.037. We
showed a strong correlation between mean stand DBH and the the historical coppice-with-standards and pastures were depleted
three variables age, stand density, and LUDIp. Furthermore, the ma- in carbon by 25% (Scenario 1) and by 50% (Scenario 2) in compari-
trix showed the expected correlations between wood volume and son to SOC stocks in old selectively cut forests. The model reveals,
basal area and between stand age and LUDIp. We used this matrix how many years after the management change a new equilibrium
as the basis for selecting the main variables for describing the pres- will be reached. Models with greater depth and more layers would
ent forest stand structure. Only LUDIp, wood volumes and ‘‘%beech’’ have included additional processes, but could not be used here be-
(quantified as fraction of total basal area) representing the present cause they require many variables, which were not investigated in
forest structure were used for further statistical analyses. this study (e.g. DOC transport). However, this one compartment model seems to simulate the time response adequately.
3.2. Carbon stocks in the organic layer and in the bulk soil
3. Results Mean OC stocks in the organic layer of the study plots varied be- tween 0.3 ± 0.09 kg m
(old selectively cut forest) and 0.7 ±
(afforestation site). OC stocks of the organic layer con- tributes with 5.1% ± 3.0% to the sum of OC stocks of organic layer The selected sites were different in their present forest struc-
3.1. Present forest structure
0.4 kg m
and OC stocks of the mineral soil.
Average total SOC stocks (only mineral soil) in the different the 110 years ACF/CwS forest, where it was more than one-and-
ture ( Table 1 ). The wood volume at 400 m 3 ha was highest in
management types varied between 8.8 ± 1.5 kg m in the old a-half times higher than in the 210 years ACF/CwS site
selectively cut forest (>250 years SF) and 11.1 ± 2.3 kg m at the (252.7 m 3 ha ). In contrast to the wood volume, the highest stand
210 years age-class forest sites ( Table 3 ) and there was no signifi- density was found in the afforestation sites, where it was about
cant relationship between management and SOC stocks. Also, SOC five times higher than in the selectively cut forest (>250 years
stocks in the individual depth increments were not affected by SF). Accordingly, the mean tree DBHs of afforestation sites were
present management or management history ( Table 3 and Fig. 2 ). lower than in selectively cut forests. Due to the dependency of tree
Across all sites 76% of SOC stocks where in the upper 30 cm and diameter on tree age the mean stand age showed similar patterns.
24% between 30 and 100 cm. 72% of the SOC in the 0–10 cm and The selectively cut forests were the oldest stands, and contained
27% of the total SOC were stored in the Ah-Horizon. There was the highest proportion of beech. The mean LUDIp ranged between
no difference in the depth profile between management types
4.7 at the more than 250 years old selectively cut forest to 27.7 at ( Fig. 2 ). The highest variations in total SOC stocks were found in the afforestation sites. LAI was remarkably constant across all sites
the afforestation sites, the lowest in the old selectively cut forests. ranging between 5.1 and 5.2. LAI in remnants of historic manage-
The mineral soil thickness ranged from 31 to 100 cm due to the ment types was 5.4 ± 0.2 for coppice forest and 4.7 ± 0.3 for
different amounts of loess that were deposited at the different plots.
Table 1 An overview (with mean and standard deviation) of the stand characteristics of the study sites. Data are given for all trees with a DBH P 7 cm.
140 years ACF/
110 years ACF/
140 years SF/CwS
210 years ACF/
210 years ACF/SF >250 years SF
Number of plots 8 35 14 11 46 16 Present management
Age-class forest Age-class forest Selectively cut forest Historical management
Age-class forest Age-class forest
Selectively cut forest
Selectively cut forest Time of management change
Pasture
Coppice-with-standards
Before 1750 Time since management change
Wood volume (2009) (m 3 ha )
280 Stand density (trees ha ) (2009)
Wood volume (1900) (m 3 ha )
27.6 (15.1) 39.5 (13.6) Basal area (cm ha )
Mean diameter at breast height (cm) (2009) 2 22.7 (8.2)
27.3 (11.0) 27.1 (11.6) Stand age (years) (2009)
Table 2 Correlation matrix of stand-level predictors for choosing variables for further ANCOVA analysis representing the forest structure.
Age (years) % Beech LUDIp Wood volume (m 3 ha )
Wood volume (m 3 ha )
Stand density (trees ha )
Mean DBH (cm)
Basal area (cm 2 ha )
1.0 0.5 0.9 0.5 0.2 Stand density (trees ha )
1.0 0.1 1.0 Mean DBH (cm) 2
1.0 0.1 1.0 0.4 Basal area (cm ha )
1.0 0.1 0.0 Age (years)
1.0 0.4 % Beech
1.0 LUDIp
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
Table 3 Soil characteristics (mean and standard deviation) of the study sites with different management history. For the bulk density, organic carbon concentration and organic carbon stocks the Ah-horizon is presented separately. The 0–10 cm includes the Ah-horizon. Small letter (a) following a number indicates non-significant differences within rows (ANOVA, p > 0.05).
210 years ACF/SF >250 years SF Soil profile thickness (cm)
140 years ACF/Pasture
110 years ACF/CwS
140 years SF/CwS
210 years ACF/CwS
52.3 (18.1) 53.8 (14.5) Thickness of Ah-Horizon (cm)
6.4 (2.6) 10.6 (3.7) Thickness of organic layer (cm)
5.4 (1.9) 2.9 (0.6) OC-Stocks: Organic layer (kg m )
0.4 (0.4) 0.3 (0.09) OC-concentration (Bulk soil) (g kg )
50.6 (16.0) a 42.0 (17.2) a 0–10 cm
38.9 (10.9) a 40. 5 (18.1) a 10–30 cm
16.2 (6.7) a 17.6 (10.5) a OC-Stocks (Bulk soil) (kg m )
9.6 (2.2) a 8.8 (1.5) a Ah-Horizon
2.5 (1.0) a 3.2(1.2) a 0–10 cm
3.5(0.7) a 3.7(1.0) a 10–30 cm
3.4(1.0) a 3.5(0.9) a >30 cm
2.3(1.8) a 1.8(1.3) a
ranged between 20.2 ± 6.9 and 25.5 ± 5.6 g C kg in 0–10 cm and
between 9.3 ± 2.7 and 13.7 ± 3.5 g C kg
in 10–30 cm ( Table 4 ). Despite large differences among the forest types (up to 50%), the
0−10 differences were not significant (ANOVA; p > 0.05). Also, we did
not find any differences in SOC concentration between broadleaved sites and afforestation sites, which are mainly stocked with spruce. There were also no trends for HF SOC concentrations among pres- ent management types (ANOVA; p > 0.05).
Significant linear relationships between SOC and N concentra- tions were found in all density fractions independent of historic management. The correlations between C and N concentrations be- came stronger with increasing decomposition of SOC in the o-LF
and HF fraction (see Fig. 3 ). Young, less decomposed organic carbon with a high C:N ratio (29.4 ± 4.4 in 0–10 cm, 38.6 ± 6.1 in 10–
30 cm) was only found in the f-LF fraction. The C:N ratio of the Soil depth [cm]
o-LF in the 0–10 and 10–30 depth was intermediate between the 50−70
corresponding f-LF and HF. The HF represented the fraction with the lowest C:N ratio (10.4 ± 0.8 in 0–10 cm and 8.1 ± 1.3 in 10–
140 years ACF/Pasture
30 cm) of all density fractions.
110 years ACF/CwS 140 years SF/CwS
210 years ACF/CwS
3.4. SOC stocks in density fractions
210 years ACF/SF >250 years SF
The contribution of f-LF to bulk SOC stocks was about 24% and 20% in the 0–10 cm and 10–30 cm layers respectively. The coeffi- 90−110
cient of variation (CV) of the f-LF SOC stocks per site ranged be- tween 15% and 73% (mean 45%) in the 0–10 cm and 21% and 65% (mean: 41%) in 10–30 cm layer. This indicates a high variability
0 1 2 3 4 5 6 of SOC stocks within each of the investigated sites. Differences be-
SOC stocks [kg m −2 ]
tween the highest and the lowest mean f-LF SOC stocks per site were 0.2 kg m
(0–10 cm) and 0.4 kg m (10–30 cm), which also Fig. 2. Mean and standard deviation of soil organic carbon stocks (SOC) in different
showed a high variability between management types. The contri- soil depths and under different management histories.
bution of o-LF to bulk SOC stocks was 9% in the 0–10 cm and 6% in the 10–30 cm layer almost one-third of the contribution to f-LF.
There was a significant positive relationship between total SOC The CV of the o-LF SOC stocks ranged between 24% and 83% (mean: stocks and total soil depth, but the explained variability was low
58%) in the 0–10 cm and between 12% and 80% (mean: 49%) in the (R = 0.08). Inorganic carbon (IC) stocks increased with soil depth.
10–30 cm layer. Differences between the average o-LF SOC stocks In the 0–10 cm layer, average IC stocks were 0.05 ± 0.53 kg m
(0–10 cm) and 0.3 kg m (0–10 cm). As (ca. 1% of the total carbon stock). In 10–30 cm layer average IC stocks
per site were 0.2 kg m
shown for the f-LF we found a high variation between and within of 0.2 ± 0.5 kg m were measured (ca. 4% of the total carbon stock).
the investigation sites.
The light fraction (f-LF and o-LF) contributed around 30% to
3.3. SOC and nitrogen concentration in density fractions bulk SOC stocks (33% in 0–10 cm and 26% in 10–30 cm). The HF stored the largest proportion of total SOC stocks. The CV of the
The light fractions showed a much higher SOC and N concentra- HF SOC stocks ranged between 14% and 64% (mean: 33%) in the tion in different depth intervals than the heavy fraction (see Table 4
0–10 cm and between 12% and 70% (mean: 34%) in the 10–30 cm and Fig. 3 ). The average HF SOC concentration of the study sites
layer. Therefore the mean variability of the sites was smallest in
249 Table 4
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
Mean and standard deviation (in brackets) of soil organic carbon (OC) and nitrogen (N) concentration measured in the three density fractions (f-LF free light fraction, o-LF occluded light fraction, HF heavy fraction) and in different depths of the soil on sites with different historical management types. The Ah-Horizon is presented separately. The 0–
10 cm includes the Ah-horizon. Small letter (a) following a number indicates non-significant differences within rows (ANOVA, p > 0.05). GROUP
IV
I II III
210 years ACF/SF >250 years SF OC-concentration (f-LF) (g kg )
140 years ACF/Pasture
110 years ACF/CwS
140 years SF/CwS
210 years ACF/CwS
Ah-Horizon 345.8 (16.2) a
333.2 (41.9) a 336.8 (16.7) a 0–10 cm
307.0 (60.3) a 331.1 (34.6) a 10–30 cm
245.5 (64.2) a 298.1 (52.5) a OC-concentration (o-LF) (g kg ) Ah-Horizon
318.1 (94.7) a 290.1 (137.6) a 0–10 cm
257.4 (82.9) a 319.5 (72.9) a 10–30 cm
189.4 (90.6) a 237.2 (97.8) a OC-concentration (HF) (g kg ) Ah-Horizon
30.1 (7.4) a 22.2 (3.3) a 0–10 cm
25.5 (5.6) a 20.5 (4.4) a 10–30 cm
13.7 (3.5) a 9.6 (1.4) a N-concentration (f-LF) (g kg ) Ah-Horizon
13.0 (2.1) a 12.6 (1.5) a 0–10 cm
11.1 (2.6) a 11.7 (1.7) a 10–30 cm
6.8 (1.8) a 7.1 (0.5) a N-concentration (o-LF) (g kg ) Ah-Horizon
13.9 (4.3) a 13.3 (5.8) a 0–10 cm
10.3 (3.5) a 12.0 (3.2) a 10–30 cm
4.9 (2.3) a 7.2 (3.6) a N-concentration (HF) (g kg ) Ah-Horizon
2.8 (0.7) a 2.1 (0.3) a 0–10 cm
2.5 (0.5) a 1.9 (0.4) a 10–30 cm
1.6 (0.4) a 1.1 (0.2) a
the HF fraction. Average SOC stocks of the HF in upper 10 cm per Bulk SOC stocks were significantly explained by total depth of site varied between 1.6 kg m
the soil profile. Deep soil profiles contain more SOC stocks than in the age-class forest (210 years ACF/SF). Average SOC stocks of
(140 years SF/CwS) and 2.2 kg m
shallow profiles, but shallow profiles contain more SOC stocks in the HF in the 10–30 cm layer per site varied between 1.9 kg m
the upper depth increment. Root biomass in the Ah-Horizon had in 140 years SF/CwS and 4.1 kg m
significant positive effects on SOC stocks in bulk soil and in all den- ble 5 and Fig. 4 ). Contrary to our expectations we found less SOC
on the afforestation site ( Ta-
sity fractions, especially in the light fraction ( Table 7 ). Because the in the old selectively cut forest and a relative high amount of C
free light fraction consisted of large, un-decomposed or partly in the afforestation sites. This observation is the opposite of our
decomposed root and plant fragments, strongest correlations were initial hypothesis (see Fig. 1 ). Also, there was no significant relation
observed between root mass and f-LF-SOC stocks. Root mass had between SOC stocks and historic forest management.
no influences on SOC stocks in deeper soil horizons. We found a significant decrease of total SOC stocks in the 0–
3.5. Radiocarbon signatures in the HF fraction
10 cm and in the 10–30 cm layers with increasing abundance of beech, quantified as fraction of total basal area (see Fig. 5 ). The Radiocarbon ( 14 C) concentrations can be used to trace the mean
wood volume increased significantly with abundance of beech.
These results did not change, if the spruce plots were excluded tions in the HF fraction decreased with increasing soil depth at
age of SOC and therefore the turnover time. Mean 14 C concentra-
from the analysis.
all investigation sites (see Table 6 ). Mean 14 C concentrations of
102.0 pMC in the 0–10 cm and 93.4 pMC in the 10–30 cm layers
correspond to a mean 14 C age of around 100 years and 550 years,
4. Discussion
respectively. The mean 14 C concentration for the Ah horizon was
106.3 pMC, which corresponds to a radiocarbon age of younger
4.1. Expected SOC trends in different density fractions than 60 years. The radiocarbon age of the upper 10 cm was there- fore less than the time since the last management change.
The HF represents the fraction with the lowest C:N ratio of all density fractions, which indicates a higher degree of microbial
3.6. Analysis of differences in SOC stocks between individual sample decomposition ( Gregorich et al., 2006; Guggenberger et al., plots
1995 ; von Lützow et al., 2006 ). The observation that the C:N ra- tio of the HF is 20 times lower than that of the LF confirms other
In the following ANCOVA analysis the SOC stocks in the whole studies ( Rovira and Vallejo, 2003 ; Swanston et al., 2005 , Wagai soil profile, in the Ah-Horizon, in the 0–10 cm and in the 10–
et al., 2009 ). Light fractions of SOC are known to respond faster
30 cm layers were treated as dependent variables. Following the to land-use and management changes than the HF fraction correlation matrix of stand-level predictors ( Table 2 ) LUDIp, wood
( Hassink et al., 1997 ). The LF fraction is very variable in space volume, and abundance of beech in percent were independent
and time (seasonal, annual) and this limits its use as an indicator variables representing forest structure. The historical forest man-
for historic land-use changes. f-LF comprises young detritus and agement and time since the management changed was used to
fine roots and is considered to be mostly determined by C characterise forest history (see Table 1 ). OC stocks of the organic
input levels and quality of the C input. Therefore we assumed layer, root mass, and profile depth were used as independent soil
that the LF is particularly sensitive to present forest manage- parameters.
ment but not to historical forest management. This is also the
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
Fig. 3. Relationship between soil organic carbon (SOC) concentration and nitrogen (N) concentration in different density fractions (f-LF free light fraction, o-LF occluded light fraction, HF heavy fraction) in 0–10 cm and 10–30 cm soil depth increment, r 2 = Pearson correlation coefficient, l C:N ratio = average of the C:N ratio of the different density fractions in both depth increments, red crosses are the mean SOC and N concentration with the standard deviation.
case with the organic carbon in the organic layer. OC in the or-
4.2. Mechanisms explaining similarity of SOC stocks despite differences ganic layer is subject to high seasonal variability. Therefore the
in forest history
smallest OC stocks in the old selectively cut forest are mainly caused by the sampling time. While the 140 years SF/CwS forest
This study demonstrates that 100–200 years after changes in site was sampled in autumn, the old selectively cut forest site
forest management from coppice-with-standards forest to differ- was sampled in early spring. A large amount of the litter was al-
ent types of high forest, differences in soil carbon stocks and its ready decomposed. Former investigation by Mund (2004) in the
fractions are no longer detectable. There are several factors, which same area showed relatively similar litter production between
lead to this observation.
present selectively cut forest, unmanaged forest and age-class forest. Mean litter fall varied between 0.21 and 0.28 kg C m
4.2.1. Hypothesis 1: high variability of soil properties hampers year
( Mund, 2004 ).
detection of changes
Therefore we only expected trends due to historical forest man- The total SOC stocks were similar to those found in previous agement in the SOC of the bulk soil and the HF fraction.
studies in the Hainich-Dün Region ( Mund, 2004 ) and in studies
251 Table 5
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
Soil organic carbon stocks in the different density fractions (f-LF free light fraction, o-LF occluded light fraction, HF heavy fraction) and in different depths of the soil on sites with different historical management types. The 0–10 cm includes the Ah-horizon. Small letter (a) following a number indicates non-significant differences within rows (ANOVA, p > 0.05).
210 years ACF/SF >250 years SF C-Stocks (f-LF) (kg m )
140 years ACF/Pasture
110 years ACF/CwS
140 years SF/CwS
210 years ACF/CwS
0.6 (0.4) a 0.7 (0.3) a 0–10 cm
Ah-Horizon 0.5 (0.2) a
0.7 (0.4) a 0.8 (0.2) a 10–30 cm
0.7 (0.4) a 1.0 (0.2) a C-Stocks (o-LF) (kg m )
Ah-Horizon 0.2(0.2) a
0.2 (0.1) a 0.3 (0.2) a 0–10 cm
0.3 (0.2) a 0.4 (0.2) a 10–30 cm
0.3 (0.2) a 0.4 (0.2) a C-Stocks (HF) (kg m )
1.7 (0.6) a 1.3 (0.5) a 0–10 cm
Ah-Horizon 1.0 (0.7) a
2.2 (0.3) a 1.8 (0.3) a 10–30 cm
2.0 (1.0) a 4.0(1.6) a
2.0 (0.6) a
1.6 (1.0) a
1.9 (0.6) a 3.1(1.1) a
3.0(0.8) a
1.9(1.3) a
3.3 (0.8) a 2.2 (0.3) a
Fig. 4. Soil organic carbon stocks in the different density fractions (f-LF free light fraction, o-LF occluded light fraction, HF heavy fraction) in 0–10 cm and 10–30 cm soil depth increment.
Table 6 14 C measurements table with the 14 C contents (pMC) of the FH fraction (mean and standard deviation). The 0–10 cm includes the Ah-horizon.
210 years ACF/SF >250 years SF AMS 14 C (% modern C)
140 years ACF/Pasture
110 years ACF/CwS
140 years SF/CwS
210 years ACF/CwS
at the ‘‘Göttinger Wald’’, a 120-year-old beech stand, on limestone between management histories. There was no consistent trend covered with loess ( Meiwes and Beese, 1988 ). All sites of this study
a, black bar). were located within the relatively small region ‘‘Hainich-Dün’’
according to our initial hypothesis ( Fig. 1
The variation between management types was high but over- (160 km 2 ) with similar climatic and edaphic conditions. We only
lapping, reflecting the spatial heterogeneity of many soil properties sampled Luvisols to keep the soil type constant. However the bulk
such as clay content, thickness of loess layer, soil pH, or soil thick- SOC and the density fractions did not show significant differences
ness. For example, deep soil profiles contain more SOC stocks than
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
Table 7 Analysis of covariance with carbons stocks of bulks soil and different density fractions (f-LF free light fraction, o-LF occluded light fraction, HF heavy fraction) in the Ah-Horizon, 0–10 cm and 10–30 cm depth increment as response variable. In this table degrees of freedom (df), and F-values are presented.
df Total
Bulk soil
f-LF
o-LF
HF Bulk soil
f-LF
o-LF
HF Bulk soil
f-LF o-LF HF
F -value F -value F -value LUDIp
F -value
F -value
F -value
F -value
F -value
F -value
F -value
F -value
F -value
F -value
1 1.8 0.6 0.9 0.2 0.5 2.7 0.5 0.0 3.6 0.7 3.7 0.9 7.2 * Wood volume
2.7 0.3 3.3 0.1 2.3 0.1 OC stocks litter
Root mass
0.0 0.1 2.5 Historical forest
Profile depth
2 2.1 0.5 0.3 1.0 1.1 2.0 1.9 2.0 0.5 3.1 1.2 0.4 1.6 management Time since
1 1.2 0.0 0.0 0.0 0.4 0.8 0.1 0.2 0.3 2.9 0.0 0.4 2.1 management change
** Significant result indicated by p < 0.05. Significant result indicated by p < 0.01.
Significant result indicated by p < 0.001.
that information about the soil thickness is necessary when esti- mating spatial SOC stocks within the same region.
Although we sampled 130 plots the large small-scale variability of soil properties reduced the strength of the statistical analysis.
4.2.2. Hypothesis 2: SOC has recovered within a century This hypothesis is supported by the studies of Goodale and Aber (2001) who showed in a comparison between old growth and his- torically logged forest that after 80–110 years C stocks had fully recovered from any losses that may have occurred after logging. We simulated our results using a simple one-component model ( Fig. 6 ). Even if we assume historic differences in SOC stocks of 50% in the past, the soils would reach a kind of soil carbon ‘‘equi- librium’’ after 80 years on the fertile soils of the study area and despite management-related changes in stand carbon stocks. With differences of 25% the equilibrium would be reached even earlier. Thus, if there had been differences in SOC stocks 100 years ago, the change in management into high forests would have equili- brated all differences of the past in less than 100 years. In this
SOC stocks [kg m
Respiration=136.4 [g m −2 a −1 ], k=0.03
Fig. 5. Relationship between the abundance of beech expressed as percentage of the total tree basal area and bulk SOC stocks in the 0–10 and 10–30 depth
50% of inital carbon
increment and biomass C.
25% of inital carbon shallow profiles, but shallow profiles contain more SOC stocks in 012345 the upper depth increment. This was, in part, due to differences
150 200 in root distributions, which affect the vertical placement of C. Shal-
lower soil profiles have significantly more roots in the upper Time since management change [years] 0–30 cm than deep soil profiles (linear regression between root
Fig. 6. Theoretically carbon accumulation curves after running the one-component
mass 0-30cm and profile thickness: R 2 = 0.04, p = 0.02). This confirms
soil decomposition model.
253 sense, our study came too late to detect effects of land manage-
J. Wäldchen et al. / Forest Ecology and Management 289 (2013) 243–254
fraction of basal area. Thoms et al. (2010) and Guckland et al.
(2009) found significantly increased pH values and significantly HF fraction in the 0–10 cm layer, in which we expected signals of
ment in the past. The 14 C data shows similar patterns. Even the
decreased C:N ratios with increasing contribution of Acer and Frax- historic forest management, has an average age of about the same
inus in the existing forest stands of the same region. Finzi et al. as the time since the last management change.
(1998) and Neirynck et al. (2000) , found much lower pH and base saturation beneath canopies of Fagus species than under Tilia, Frax-
4.2.3. Hypothesis 3: SOC did not differ significantly between the inus and Acer species. Augusto et al. (2002) summarised effects of management types 200 years ago
tree species on soil fertility in European temperate forests and con- An alternative hypothesis is that SOC stocks did not differ sig-
cluded that the acidifying ability of F. sylvatica species was higher nificantly between the management types 200 years ago. From
than that of all other deciduous tree species. Acidification pro- the inventory reports of the 19th century it emerges that most cop-
cesses change the activity of soil biota and affect decomposition pice-with-standards forest in the Hainich-Dün was in fact under-
and nutrient turnover processes. Soil acidification is accompanied used and fully covered with trees ( Wäldchen et al., 2011 ). Govern-
by losses of base cation, nitrogen and carbon in the mineral soil. mental reports on forest degradation may also have been biased
Acidification of the soil by litter of Fagus would increase the litter because they were used to rule against the peasants ( Küster,
carbon pool, and change the fine root distribution from the Ah into 1998 ). Litter fall represents the largest flux of C from aboveground
the litter layer due to changes in biological activity ( Schulze, 2000 ; biomass to the soil where decomposing litter is a major driving
Brumme and Khanna, 2009 ) when compared to Acer or Fraxinus variable of soil organic carbon. Due to similar LAI measurements
dominated stands. This redistribution of fine roots could result in on sites with different forest management in this study ( Table 1 )
C depletion of the Ah and upper B horizon due to respiration of the annual input of leaf mass was probably similar for all forest
old C and initiate podsolisation. In this study we compare different management types. It is generally assumed that root turnover is re-
management types on the same soil type at high base cation satu- lated to leaf turnover unless additional disturbances by manage-
ration. Thus, we are confident that we reached a quasi steady-state ment interfere. For example, the coppice with standard
as explained by Brumme and Khanna (2009) in a fairly short time management systems had a higher level of stand disturbance than
even after severe disturbance (afforestation of degraded grass- the selectively cut forest. Every 12–25 years (depending on the
lands). The forest cover of this region has changed during historic rotation period) the coppice was completely removed and only
times from mixed deciduous forests that were dominated by Acer, some standards were left for future timber wood growth. Shortly
Fraxinus and Quercus, into Fagus dominated forests ( Schulze et al., after the disturbance a higher decomposition rate is expected
2012 ), and this change in forest species composition, which is in due to a higher light intensity, which causes a higher temperature
fact a typical modern land-use change, appears to have major neg- on the soil surface. However these higher decomposition rates in
ative effects on SOC.
comparison to the selectively cut system may also have been com- pensated by a higher belowground litter input (roots) in the cop- pice-with-standards forest following each rotation. Thus, we also
5. Conclusion
may assume that the understorey in a coppice-with-standards sys- tem and the grass vegetation, which develops after deforestation added more root litter than the vegetation in a less disturbed selec-
of forests in the 18th and 19th century had no detectable effect tively cut forest.
on present forest (SOC) stocks.
Considering the fact that the soils of the study sites are gener- ally very fertile and productive, it can be assumed that they have
on SOC stocks.
a higher resilience to disturbance than forest soils, e.g., on acid bedrock. Although, the historic forest inventory described different forest management systems, these differences in forest manage- ment system may not result in differences in the soil carbon bal-
Acknowledgements
ance and associated differences in soil organic carbon stocks. It can be expected that the relation between C-gains from litter and
The Max Planck Society is gratefully acknowledged for funding C-losses from harvest were similar in all management types.
this project and supplying the employment opportunities for the Our study also suggests that carbon extraction by logging of