Forest Ecology and Management 261 (2011) 342–352

Contents lists available at ScienceDirect

Forest Ecology and Management

journal homepage: [Link]/locate/foreco

Accumulation of dead wood in abandoned beech (Fagus sylvatica L.) forests in

northwestern Germany
P. Meyer a,∗ , M. Schmidt b
a
Nordwestdeutsche Forstliche Versuchsanstalt, Sachgebiet Waldnaturschutz/Naturwald, Grätzelstr. 2 D-37079, Göttingen, Germany
b
Nordwestdeutsche Forstliche Versuchsanstalt, Sachgebiet Informatik/Waldwachstumsmodellierung, Grätzelstr. 2, D-37079, Göttingen, Germany

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

Article history: Currently, there is much debate about what strategy is most suitable for increasing old-growth attributes
Received 12 May 2010 in forests that have been managed intensively for wood production in the past. Passive restoration, i.e.
Received in revised form 18 August 2010 cessation of forestry interventions, should be considered when the old-growth attributes desired can be
Accepted 19 August 2010
restored within a feasible period of time.
Available online 4 December 2010
Our study focuses on standing and lying coarse dead wood (≥20 cm diameter) in beech-dominated
forests in northwestern Germany. We analyzed monitoring data of 545 sample plots (sized 500–1000 m2 )
Keywords:
from 12 strict forest reserves (SFRs). The SFRs had been without forestry intervention for up to 28 years.
Nature conservation
Saproxylic species
Both, number of dead objects and volume of dead wood (m3 ha−1 ) increased significantly with ongoing
Old-growth time since abandonment from forestry interventions. The mean amount doubled from 9 to 18 m3 ha−1
Set aside forests within 10 years. The proportion of standing dead wood was about 40% of the total dead wood pool ≥20 cm
diameter.
With mixed linear modeling we showed that dead wood increased by a mean net rate of about
1 m3 ha−1 a−1 . Therefore, after three decades critical values for restoring the dead wood pool could be
reached. We hypothesized that the rate of dead wood input is mainly determined by disturbance driven
tree mortality such as oak decline, bark beetle infestations and storms.
A comparison with primeval forests or reserves abandoned more than 100 years ago showed that the
SFRs studied are at the beginning of a long process of dead wood accumulation.
Based on our results, the abandonment of forest activities in harvestable pure and mixed beech stands
is an effective strategy for restoring the dead wood pool.
© 2010 Published by Elsevier B.V.

1. Introduction Meanwhile, dead wood is recognized as a key element of forests.

It contributes significantly to biodiversity (Maser et al., 1979;
In general, natural forests are characterized by large amounts Speight, 1989), carbon sequestration (Harmon, 2009) and nutrient
of dead wood with high structural diversity (Peterken, 1996; cycling (Harmon et al., 1986; Müller et al., 2005; Müller-Using and
Christensen et al., 2005; Wirth et al., 2009). Due to intensive man- Bartsch, 2007).
agement and exploitation of temperate and boreal forests in Europe In the MCPFE (Ministerial Conference on the Protection of
in the past, dead wood was relatively scarce or entirely absent Forests in Europe) process, volume of dead wood has been adopted
over prolonged periods (Grove and Stork, 1999; Siitonen, 2001; as a quantitative indicator for the maintenance, conservation and
Christensen et al., 2005). As a result, saproxylic species became appropriate enhancement of biological diversity in forest ecosys-
extinct on a local or regional scale (Speight, 1989; Albrecht, 1991; tems. The amount and proportion of dead wood is now monitored
Odor et al., 2006). Today, the number of endangered saproxylic in various countries to assess the ecological state of forests (Schuck
species is still very high. Therefore, the protection of saproxylic et al., 2004; Woodall et al., 2009).
organisms plays a major role in conservation programs, e.g. in the In recent years profound insight has been gained into the
European Union Natura 2000 Networking Programme (European management of old-growth biological communities in forests. For
Communities, 2003). instance it is well known and understood that logging carried out
without necessary precautionary measures is a predominant threat
to the saproxylic biodiversity in forests (Martikainen et al., 2000;
Siitonen, 2001; Müller et al., 2007). Additionally, thresholds of dead
∗ Corresponding author. Tel.: +49 551 69401 180; fax: +49 551 69401 160. wood volume necessary for the survival of endangered saproxylic
E-mail address: [Link]@[Link] (P. Meyer). species have been derived recently; minimum levels amount to

0378-1127/$ – see front matter © 2010 Published by Elsevier B.V.

doi:10.1016/[Link].2010.08.037
 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352 343

20–30 m3 ha−1 (Bütler et al., 2004; Müller et al., 2005; Müller and 2. Materials and methods
Bütler, 2010).
However, the question of which management strategy is most 2.1. Selected study sites
suitable for obtaining these minimum levels, remains unanswered.
Broadly, active and passive restoration strategies can be distin- In Lower Saxony, Germany, there are 106 designated SFRs, com-
guished (Bauhus et al., 2009). Keeton (2006) showed that active prising an area of about 4500 ha (Meyer et al., 2007). Some SFRs
restoration is more successful for creating old-growth features than have been left to develop without forestry intervention since the
conventional silvicultural systems. Nevertheless girdling, felling or early 1970s. However, most were abandoned from the late 1980s
pulling down trees to be left on the forest floor may be a costly into the late 1990s.
measure. Moreover, it requires in-depth understanding of natu- In many of these reserves, live forest structure has been moni-
ral habitats to avoid the creation of artificial structures. Therefore tored for several decades (Meyer, 2005). However, dead wood was
passive restoration, i.e. the cessation of forestry interventions (thin- not monitored until 1988.
ning, harvesting, planting, etc.), should be considered when the In this study, all SFRs comprising beech as a co-dominant or
restoration process can be expected to achieve the desired forest dominant tree species were selected when forest structure and
state within a feasible time period (Bütler and Lachat, 2009). dead wood had been re-measured once. Based on these criteria,
In Germany, predominantly state forest enterprises have begun 545 sample plots in 12 SFRs covering a total area of 46.6 ha were
to launch conservation programs for old-growth biological commu- selected.
nities by leaving resident trees or setting aside forest stands from The 12 SFRs selected are spread evenly across the eastern region
forest management (Schaber-Schoor, 2008). For several decades, an of north-west Germany (Fig. 1) and comprise a wide range of forest
increasing number of strict forest reserves (SFRs), which cover areas types and site conditions (Table 1). They had been abandoned for up
ranging from less than one hectare to many hundreds of hectares, to 28 years at the time of the first survey (Table 2). Since then, some
have been abandoned from forestry interventions (Bücking et al., SFRs have been extended. The SFRs were divided into 2 groups of
2000; Meyer, 2005; Meyer et al., 2007). This trend is complemented sample plots based on time since abandonment.
by the increasing designation of larger unmanaged forests as nature Most selected SFRs encompass different forest stands. The mean
reserves and national parks. Consequently, the area of protected tree age indicates that mainly older stands of harvestable age have
forests in Germany with no active, or only minimal intervention been designated as SFRs (Table 1). However, younger stands are
has increased from 910 km2 (=0.9% of the forest area) in 2000 to also represented in most SFRs.
1570 km2 (=1.4% of forest area) in 2005 (UNECE, 2010). Yet, in
Germany the area of forests set aside from forestry interventions
is still small, as it is in almost all European countries (Hanski and 2.2. Monitoring forest structure
Walsh, 2004).
Effective restoration programs for old-growth forests require a In all selected SFRs, circular sample plots of either 500 or
thorough understanding of the ongoing ecological processes and 1000 m2 were established based on a grid spacing of either
how these affect biological communities. As yet, little is known 50 m × 50 m or 100 m × 100 m which covered the whole reserve.
about processes in the old-growth stage of natural forests in Cen- Within the sample plots, all living and dead standing trees with
tral Europe because primeval forests are virtually non-existent and, dbh ≥7 cm were assessed. Tree species was recorded, and coordi-
in the designated reserves, only a brief period of time has elapsed nates (azimuth, distance from plot centre), slope, dbh and height
since management ceased. (all dead trees and snags; living trees sampled) were measured.
Notwithstanding this, there has been an increase in the num- The typology of all trees, based on vitality (living, dying, dead) and
ber of studies concerned with effects of forest abandonment on tree type (tree intact, tree with partial crown breakage, snag) was
species diversity (Graae and Heskjaer, 1997; Brunet et al., 1996, recorded. In addition, the decay stage of dead trees and snags was
1997; Aude and Lawesson, 1998; Oheimb et al., 2004; Winter determined.
et al., 2005; Paillet et al., 2010), the mosaic of development phases Lying dead wood pieces were recorded only within the sam-
(Oheimb et al., 2005; Meyer and Schmidt, 2008) and dead wood ple plot area and when diameter at the larger end exceeded 20 cm.
(Christensen et al., 2005; Vandekerkhove et al., 2009). Among the The coordinates of each end of lying dead wood in the plot were
forest sites studied, SFRs are of special interest because they allow recorded. When pieces touched the plot boundary, the coordinates
the early self-restoration processes of previously managed forests of the point in the plot and the point on the boundary were recorded
and the development of old-growth structures to be investigated (Fig. 2). The first diameter was measured at the larger end and the
(Vandekerkhove et al., 2009). second diameter at the smaller end, or later 3 m from the larger end
This study draws on the substantial database resulting from (assessment protocol changed with time). In addition, decay stage,
the monitoring program of SFRs (Meyer, 2005), which provides a tree species and typology of the dead wood (whole tree, crown
unique source of time series data from recently abandoned forest breakage, branch) was determined. Stumps below 1.3 m height
reserves in Central Europe, to investigate dead wood dynamics. were excluded.
We analyzed coarse dead wood in pure and mixed beech forests, The first sample plot inventories were conducted in 1988
being the dominant natural forest vegetation in Central Europe (Table 2). Re-measurement began in 1997.
(Bohn and Gollub, 2007). The study addresses the following ques-
tions:
2.3. Data analysis

1. Does the amount of standing and lying dead wood increase Data analysis and presentation was carried out with SAS 9.1 (SAS
markedly in formerly managed forests in the first decades after Institute, 2009) and R (R Development Core Team, 2009).
abandonment? After transforming polar coordinates into three dimensional
2. Is the net change in dead wood amount (NRCD = net rate of Cartesian coordinates, the length of lying dead wood was calcu-
change) mainly a stochastic process, or do certain factors influ- lated using the Pythagoras formula. The height of dead trees and
ence NRCD significantly? snags was measured directly in the field. Subsequently, dead wood
3. If the latter is true, which factors influence NRCD significantly? volume was calculated according to Meyer (1999).
344 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352

Fig. 1. Location of SFRs studied in Lower Saxony in relation to forest cover.

 
The net rate of change of dead wood volume per year (NRCD)  ln 2 
T2;0.5 =   (4)
was calculated according to Eq. (1) (symbols explained below). k
V2 − V1
NRCD = (1) where %NRCD, percentage annual rate of change of dead wood vol-
t
ume; T2;0.5 , half-life of dead wood; V1 , volume of dead wood in first
Based on the assumption of an exponential increase or decrease survey in m3 ha−1 ; V2 , volume of dead wood in second survey in
arising from a constant relative rate of change (Eq. (2)), the per- m3 ha−1 ; k, constant; t, length of monitoring period in years.
centage annual rate of change (NRCD%, see Eq.(3)) and the half-life, First, NRCD was calculated for each sample plot. Subsequently,
or doubling-time (T2;0.5 , see Eq. (4)) were calculated. based on the mean NRCD for the 12 selected SFRs, the mean, stan-
dard deviation and maximum and minimum values were calculated
V2 = V1 · ek·t (2)
for NRCD and %NRCD (see Table 3). Undefined values of %NRCD and
ln v2 − ln v1 k occurred when dead wood volume was zero in both the first and
%NRCD = × 100 (3)
t second inventory. In this case, these observations were set to zero

Table 1
Main site and stand characteristics of SFRs.

Name SFR (abbreviation) Forest type Nutrient supplya Elevation Mean annual Mean annual Age of beechb
a. s. l. [m] precipitation [mm] temperature [◦ C] [min.–mean–max.]

Braken (Brak) Oak–Hornbeam/Beech–Oak Eutrophic 33–45 790 8.5 6–116–149

Großer Staufenberg (GrSta) Beech–Ash–Sycamore More mesotrophic 450–554 961 7.0 107–115–126
Hünstollen (Huen) Beech–Ash–Sycamore Eutrophic 335–425 866 7.4 42–119–128
Landwehr (Landw) Oak–Beech–Hornbeam Less mesotrophic 17 581 8.6 54–138–174
Limker Strang (Limk) Beech More mesotrophic 385–420 1015 7.3 121
Lohn (Lohn) Beech–Oak Eutrophic 55–88 663 8.4 11–103–126
Lüßberg (Luess) Beech Less mesotrophic 107–118 774 8.1 42–140–154
Meinsberg (MeinB) Beech Eutrophic 300–342 945 7.2 103
Meninger Holz (MenHo) Spruce–Pine–Beech Less mesotrophic 83–107 803 8.1 116–121–124
Sonnenkopf (SonKo) Spruce–Beech Less mesotrophic 580–730 1530 5.4 82–129–141
Weichel (Weich) Beech–Oak More mesotrophic 25–30 760 8.7 108–109–110
Winterlieth (Wint) Beech–Spruce More mesotrophic 400–500 1043 7.0 10–107–136
a
Dominant type of nutrient supply.
b
At the time of abandonment.
 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352 345

Table 2
Dates of SFR designation and inventories and time since abandonment.

Name SFR (abbreviation) N sample plots First inventory [year] Last inventory [year] Date of designation Time since abandonmentb
as SFRa [years]

Braken (Brak) 83 1996 2005 1989 12

Großer Staufenberg (GrSta) 47 1988 1997 1972 (1988) 21 (5)
Hünstollen (Huen) 50 1991 2002 1972 (1989) 24 (7)
Landwehr (Landw) 29 1990 2000 1972 (1989) 23 (6)
Limker Strang (Limk) 78 1991 2003 1972 19
Lohn (Lohn) 33 1990 2002 1974 21
Lüßberg (Luess) 96 1990 2000 1974 (1985) 18 (10)
Meinsberg (MeinB) 12 1990 2002 1974 21
Meninger Holz (MenHo) 13 1988 1999 1985 8
Sonnenkopf (SonKo) 26 1995 2005 1972 (1990) 28 (10)
Weichel (Weich) 16 1992 2002 1986 11
Winterlieth (Wint) 92 1996 2006 1994 7
a
Year in brackets indicate year of enlargement of the designated area.
b
Time from abandonment to first inventory.

was defined as age of beech in the overstorey in the year of aban-

donment.
The digital site maps were based on field mapping of site condi-
tions carried out by NFP in the State Forests of Lower Saxony over
several decades. Soil type classification was determined from soil
pits and soil coring between the pits. Nutrient supply was estimated
in the field, and graded from oligotrophic to eutrophic.
Meteorological data were derived by overlaying the maps of
sample plots with digital maps of key meteorological information
for the whole of Germany for the period 1960–1990, which were
modeled with cell size 1000 m × 1000 m (Gauer and Aldinger, 2005;
Wolff et al., 2003).
Subsequently, each sample plot was classified by geographical
region (lowland, mountainous) and nutrient supply (generally poor
vs. generally rich). Furthermore information on length of growing
season (average number of days with a mean temperature >10 ◦ C),
age of beech in the year of abandonment, percentage volume of
the 6 main species groups (beech, pedunculate oak, sessile oak,
Fig. 2. Assessment of dead wood pieces on circular sample plots. other broadleaves, Norway spruce, Scots pine, other conifers), years
since abandonment (if the last intervention was not documented:
years since designation + 5 years) and volume of living stand were
in the determination of the mean, standard deviation and maxi-
assigned to each plot. This information was combined with the
mum and minimum values of %NRCD to enable comparisons with
amount of dead wood (number of pieces and volume) per dead
NRCD.
wood type (standing, lying) and area.
In addition, dead wood input during the monitoring period was
Firstly, descriptive data analysis was carried out in each SFR for
calculated on the basis of dead wood pieces recorded exclusively
all plots. Standing, lying and total amount of dead wood (volume,
in the second inventory.
number of pieces) in the first and the second inventory were tested
The volume of living wood ≥7 cm dbh was calculated by con-
for significant deviation with the Wilcoxon–Mann–Whitney test.
structing diameter–height curves for tree species in each SFR using
A Chi-Square test was applied to determine whether the
Petterson’s function (Schmidt, 1967) with form factors from Bergel
contribution of each species group to the dead wood pool was pro-
(see Nagel, 1999), and multiplying height by form factor and basal
portional to its stake in the living stand. Average percentage volume
area at breast height.
of species groups in the living stand was used to define the expected
To determine which variables might influence NRCD, geo-data
probability of dead wood input, and compared to the percentage of
were obtained from several sources. In a first step, sample plots
the total dead wood supply during the monitoring period.
in the selected SFR were geo-referenced. The resulting maps were
Secondly, a more generalized modeling approach was applied
overlaid with digital site and forest management maps (source:
to the whole data pool. The effects of several environmental factors
NFP, Forest Planning Department, Lower Saxony) and stand age
(see above) on NRCD were quantified and tested for significance
and site characteristics were attributed to each sample plot. Age
with mixed linear models (Pinheiro et al., 2009). Mixed linear
models account for the fact that observations within a specific clus-
Table 3 ter, i.e. in a particular SFR, may be correlated, and hence violate
Mean (x), standard deviation (␴), minimum and maximum of NRCD and %NRCD the assumption of independence between observations in stan-
(computed on the basis of average values of selected SFRs). dard regression methodology. A variance function for the within
Dead wood NRCD [m3 ha−1 a−1 ] %NRCD a−1 SFR-variance was estimated simultaneously to account for het-
type eroscedasticity. The proposed model can generally be expressed
x  min. max. x  min. max. using matrix representation by Eq. (5).

Standing 0.59 1.02 −0.14 3.65 9.95 10.34 −5.22 26.94

Lying 0.48 0.42 −0.01 1.17 7.19 7.55 −0.06 24.54
yi = Xi ˇ+Zi bi +εi , i = 1, . . . , 12, bi ∼N(0, ), εi ∼N(0,  2 i),
Total 1.06 1.34 −0.11 4.77 7.65 6.59 −0.74 18.34
(5)
346 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352

Table 4
Ratios of the proportion of dead wood input of a specific tree species group to it’s
volume proportion in the living stand at the first inventory (Chi-Square test was
carried out for all species groups).

Name SFR (abbreviation) Species group p > Chisq

Beech Oak Spruce

Braken (Brak) 0.95 1.05 1.01 0.0017

Großer Staufenberg (GrSta) 0.66 0.90 3.81 <0.0001
Hünstollen (Huen) 0.71 10.73 – <0.0001
Landwehr (Landw) 0.55 1.27 2.28 <0.0001
Limker Strang (Limk) 0.94 0.21 8.74 <0.0001
Lohn (Lohn) 0.83 0.82 2.87 <0.0001
Lüßberg (Luess) 0.81 2.62 0.33 <0.0001
Meninger Holz (MenHo) 0.46 1.15 1.43 <0.0001
Sonnenkopf (SonKo) 2.79 – 0.77 <0.0001
Weichel (Weich) 0.02 5.78 8.53 <0.0001
Winterlieth (Wint) 0.88 2.79 1.50 <0.0001

Fig. 3. Mean stem number ha−1 (left column) and mean volume stock (m3 ha−1 , right the proportion of standing dead wood pieces was 0.41, and 0.44
column) by SFR. The absolute proportions of the tree species groups are indicated after the first, and second inventory respectively.
by different patterns. Similar to the number of dead wood pieces, the average dead
wood volume increased in nearly all SFRs (Fig. 5). Average dead
where yi is the response vector of NRCD for the ith SFR, Xi the fixed wood volume reached a maximum of 57 m3 ha−1 and a minimum
effects design matrix for SFR i, ˇ the fixed-effects vector for the value of 8 m3 ha−1 . Over all 545 sample plots, the average dead
population average, Zi the random effects design matrix for SFR i, bi wood amount was 9 m3 ha−1 in the first, and 18 m3 ha−1 in the sec-
the vector of random effects for SFR i, εi the vector of within-group ond inventory. The proportion of standing dead wood volume was
residuals, and and i are positive-definite matrices. 0.35, and 0.67 after the first, and second inventory respectively.
The statistical modeling was carried out in steps. Firstly, all
factors were considered. Next, non-significant variables (p < 0.05) 3.3. Net rate of dead wood change (NRCD)
were removed and the model re-run. The average residual of the
model was then used to determine the bias. The mean NRCD was calculated by averaging the mean values
for the 12 SFRs. Mean NRCD was 1.06 m3 ha−1 a−1 (Table 3). The
3. Results statistical spread of values was very broad. Although standing dead
wood was less frequent, and represented less volume than lying
3.1. Living stand dead wood, its NRCD was higher. The transition from standing dead
wood to lying dead wood over time may explain this result.
The average volume of the living stand ranged from 300 m3 ha−1 %NRCD of total dead wood volume ranged from a slight decrease
to about 700 m3 ha−1 (Fig. 3). Stem density tended to decrease of −0.7% to a rather large increase of about 18% (Table 3). On aver-
with increasing volume per ha. The six SFRs with the highest age, the volume of dead wood increased by 7.7% per year with an
standing volume exhibited a lower stem number between 200 increase in the proportion of standing dead wood and reduction in
and 300 trees ha−1 . These stands were single-layered stands, in the proportion of lying dead wood.
which oak was rarely present. In contrast, multi-layered stands
had a higher proportion of oak, a higher stem number but lower
3.4. Doubling time of total dead wood pool
standing volume. For the living stand, stem number ranged from
about 150 ha−1 in single-layered beech stands to about 500 ha−1 in
Given the mean NRCD calculated was to continue in future, total
oak-dominated mixed stands with a beech-hornbeam understorey
dead wood would double in about 9 years. T2;0.5 of lying dead wood
(Fig. 3).
amounts to 9.6 and T2;0.5 of standing dead wood to about 7 years.
Almost pure beech forests as well as mixed forests were also
These values indicate a rather fast relative increase in dead wood
represented in the database (Fig. 3). The indigenous oak species
volume.
(Quercus petraea L., Quercus robur L.) played a major role in many
stands. The proportion of conifers was quite low.
3.5. Contribution of tree species groups to the dead wood pool
3.2. Amount of dead wood
The proportion of each species in the living stand did not reflect
During the monitoring period, the mean number of dead wood their contribution to the dead wood pool (Table 4). The deviation
pieces per hectare increased in nearly all SFRs (Fig. 4). The mag- between the relative presence of species in the living stand and in
nitude of the increase differed somewhat, ranging from a slight dead wood input was significant in all cases. SFR-Meinsberg was
increase, e.g. in SFR Lohn, to a large increase, e.g., in SFR Großer excluded from this analysis as admixed species were not present
Staufenberg and SFR Landwehr. In many reserves, no significant in the dead wood pool. In many cases, the contribution of oak
difference between the first and the second inventory was found and Norway spruce to the build-up of dead wood volume was
by the Wilcoxon–Mann–Whitney test. This reflected the high varia- disproportionately high, whereas the contribution of beech was sig-
tion in NRCD between the plots of one SFR. For all plots, the average nificantly lower. This result showed the higher susceptibility of oak
density of dead wood almost doubled from 9.2 to 17.8 ha−1 within and Norway spruce to disturbance. For several decades peduncu-
10 years. This increase was highly significant. late oak and sessile oak suffered from oak decline in northwestern
In general the amount of standing dead wood (snags and whole Germany (Hartmann and Blank, 1998). Norway spruce is known to
trees) was much lower than the amount of lying pieces. On average, be highly susceptible to bark beetle infestation and windthrow.
 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352 347

Fig. 4. Mean number of dead wood pieces >20 cm diameter in the first and second inventory by SFR. Probability (p) of significant difference between means is indicated by
following symbols: n.s. = not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

3.6. Modeling NRCD

lying dead wood : NRCDij = ˇ0 + ˇ1 IADWij + ˇ2 VLSij + bi + εij ,

To derive a model of annual NRCD, selected environmental and
stand variables were tested for significance (see data analysis). Dif- (6.2)
ferent models were developed for standing, lying and total dead
wood. The initial amount of dead wood (IADW), the percentage
volume of oak (Oakperc) and the volume of the living stand (VLS) standing dead wood : NRCDij = ˇ0 + ˇ1 IADWij + ˇ2 Oakpercij
from the first inventory appeared to be significant factors (Table 5).
The models for the different components are: + bi + εij , (6.3)

where bi ∼ N (0,  1 2 ), εij ∼ N(0,  2 (ı1

+ |vij |ı2 ))
for Eqs.(6.1) and
total dead wood : NRCDij = ˇ0 + ˇ1 IADWij + ˇ2 Oakpercij
(6.2) and εij ∼ N(0,  2 ) for Eq.(6.3), and where i = 1, . . ., 12 represents
+ ˇ3 VLSij + bi + εij , (6.1) the SFRs and j the plots nested within SFRs. bi denotes the ran-
348 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352

Fig. 5. Mean volume of dead wood ≥20 cm diameter in the first and second inventory by SFR. Probability (p) of significant difference between mean values is indicated by
following symbols: n.s. = not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

dom effect of the intercept of the ith SFR, and εij , the within-group A purely random intercept model with no covariates was fitted for
residual for the jth plot nested within SFR i. The predictor variables the 3 dead wood components as well to quantify the mean NRCD
were centered by their means to improve the interpretability of the and test whether a significant increase in NRCD results:
intercept and to reduce the correlation between the intercept and
the slopes (gradient in the graph). The heterogeneous variance was for all dead wood compartments : NRCDij = ˇ0 + bi + εij , (6.4)
modeled as power function of the fitted values plus a constant. This
type of variance function is recommended by Pinheiro and Bates where bi ∼N(0, 12 ) and εij ∼ N(0,  2 )
(2000) for response variables when the data range includes zero. The IADW proved to be highly significant in all models (Table 5).
NRCD (total, lying and standing dead wood) decreased by 0.23, 0.25
 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352 349

Table 5
Parameters of NRCD mixed linear models (levels of significance are indicated as follows: n.s. = not significant, *p < 0.05, **p < 0.01, ***p < 0.001; LCL = lower confidence limit
95%, UCL = upper confidence limit 95%).

Model Effect Parameter Value Standard error p>t LCL0.95 UCL0.95

Standing Fixed Intercept 0.1355 0.2510 0.5895

Fixed Oak percentage 0.0173 0.0037 <0.001
Fixed Volume dead wood-first inventory −0.0650 0.0087 <0.001
Random 1 0.80 0.49 1.29
Random  1.53 1.44 1.63

Lying Fixed Intercept 0.3356 0.1029 0.0012

Fixed Volume living stand 0.0014 0.0005 0.0099
Fixed Volume dead wood-first inventory −0.0248 0.0054 <0.001
Random 1 0.21 0.05 0.86
Random  0.40 0.17 0.97
Variance function ı1 3.93 1.48 10.46
Variance function ı2 1.51 0.73 2.28

Total Fixed Intercept 0.8438 0.2390 <0.001

Fixed Oak percentage 0.0127 0.0053 0.0167
Fixed Volume living stand 0.0016 0.0007 0.0261
Fixed Volume dead wood-first inventory −0.0234 0.0057 <0.001
Random 1 0.71 0.31 1.64
Random  0.08 0.02 0.26
Variance function ı1 27.54 7.83 96.93
Variance function ı2 3.64 2.25 5.04

Table 6
Values of the NRCD intercept model (levels of significance are indicated as follows: n.s. = not significant, *p < 0.05, **p < 0.01, ***p < 0.001; LCL = lower confidence limit 95%,
UCL = upper confidence limit 95%).

Model Effect Parameter Value Standard error p>t LCL0.95 UCL0.95

Standing Fixed Intercept 0.59 0.29 0.044 – –

Random 1 0.96 – – 0.60 1.53
Random  1.62 – – 1.53 1.72

Lying Fixed Intercept 0.42 0.11 <0.001 – –

Random 1 0.23 – – 0.07 0.72
Random  1.81 – – 1.70 1.91

Total Fixed Intercept 1.05 0.38 0.0057 – –

Random 1 1.23 – – 0.76 2.01
Random  2.44 – – 2.30 2.59

and 0.65 m3 for each additional 10 m3 ha−1 of dead wood present However, the exact amount on small areas of 0.05–0.1 ha could not
in the first inventory. When pedunculate oak or sessile oak were be predicted since the fixed effects (Eqs. (6.1)–(6.3)) represented
present in the living stand, NRCD of total and standing dead wood only a small proportion of the variance of NRCD. For total dead
rose. Furthermore, the volume of the living stand increased NRCD wood the random variance of NRCD decreased only from 6.82 in
of total and lying dead wood significantly by 0.16 and 0.13 m3 per the pure intercept model (Eq. (6.4)) to 6.35 in the model including
additional 100 m3 ha−1 respectively. predictor variables (Eq. (6.1)). In these models, the variance of the
None of the environmental factors nor stand age or time since lying and standing dead wood decreased from 3.31 to 3.12 and from
abandonment was found to influence NRCD. For this dataset, NRCD 3.13 to 2.70, respectively.
depended solely on tree species composition, volume of the living
4. Discussion
stand and initial volume of dead wood.
The unexplained variance of NRCD was high. Thus, on a small
The question, whether the cessation of forest interventions is
scale, dead wood dynamics were mainly driven by stochastic pro-
an effective way of restoring old-growth biological communities
cesses. The variation between SFRs was much lower than variance
in forests, is currently under debate. In our study we focused on
between the plots within SFRs. The variance partition coefficient
the dynamics of dead wood in pure and mixed beech forests. We
(VPC) (Goldstein, 2003) of the variance between SFRs was 19%, 3%
showed that, after several decades of abandonment, substantial
and 21% for the total, lying and standing dead wood respectively
accumulation of dead wood is highly probable. Within this period
(Eq. (7)). In the calculation of the VPC, the variance for the total
of time, the average NRCD of about 1 m3 ha−1 a−1 will result in a
and lying dead wood within an SFR was assumed homogenous, not
stock of dead wood that exceeds recently derived threshold values
heterogeneous.
for endangered saproxylic species (Bütler et al., 2004; Müller et al.,
12 2005; Müller and Bütler, 2010).
VPC = (7) From general ecosystem theory it follows that dead wood
12 + 2
dynamics depend on the relationship between input and decay
The intercept of the pure random intercept model with no rates (Olson, 1963; Mindermann, 1968). Thus, a better understand-
covariates (Eq. (6.4)) for total (1.05 m3 ) and lying dead wood ing of dead wood dynamics requires in depth investigation of
(0.42 m3 ) was significant at the level of 0.01 (Table 6). For standing decomposition and of processes providing dead wood input.
dead wood (0.59 m3 ), the pure random intercept model was signif- Knowledge of decay processes for Central European tree species
icant at the level of 0.05. Therefore an accumulation of dead wood in general, and beech in particular, is limited. Rock et al. (2008) have
after abandonment was probable for all dead wood components. summarized decomposition rate constants referring to density
350 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352

loss. However, forest inventories primarily assess dead wood vol- discrete or periodic nature of disturbances. This means that after
ume. Only Müller-Using (2005), Kahl (2008) and Müller-Using and a disturbance event a high dead wood stock yet low NRCD would
Bartsch (2009) have derived volume-based decay constants. These result. However, if a disturbance event occurs during the moni-
studies are restricted to specific sites in Lower Saxony (Solling toring period, a high NRCD coincides with low initial dead wood
Mountains) and Thuringia (Hainich National Park), involve only a stock. This temporary effect would be smoothed out if more SFRs
small number of stands and look only at lying dead wood. Hence, were observed over longer periods of time. Thus we anticipate a
there are two major gaps in knowledge of decay of dead wood in continuous or, more likely, accelerated dead wood accumulation
Central Europe: supra-regional volume-based decay functions are from future monitoring as time since abandonment increases.
lacking, and decay rates and the life-time of standing dead wood That age did not affect NRCD in our model may be due firstly to
are still uncertain. Again SFRs provide appropriate study sites for the imbalance in stand ages in this dataset, which comprises mainly
dead wood decay because monitoring data and a wide variety of older stands. Secondly, stands in the old-growth stage are largely
dead wood pieces are available. unrepresented. Finally, the monitoring period was most likely too
In respect of processes providing input of dead wood a better short to observe age effects.
understanding of causes of mortality (Franklin et al., 1987; Lutz A comparison of our results on dead wood stocks with those
and Halpern, 2006) is required as well as the role of tree mortality from primeval forests or reserves abandoned for more than 100
or crown breakage on dead wood input (Harmon, 2001). years ago (Korpel, 1995; Tabaku, 1999; Saniga and Schütz, 2001;
The results suggest that mortality caused by disturbances such Meyer et al., 2003; Christensen et al., 2005) shows that the SFRs
as oak decline, bark beetle infestations or storms is the main influ- studied are at the beginning of a long process of dead wood accumu-
ence on NRCD (see Edman et al., 2007). This conclusion is supported lation. Christensen et al. (2005) report an average of 130 m3 ha−1
by the disproportionately high contribution to the dead wood pool dead wood in beech forest reserves. Dead wood stocks recorded
of the disturbance-prone species, sessile oak, pedunculate oak, and in primeval beech forests range from 32 to 310 m3 ha−1 (Saniga
Norway spruce, and the identification of percentage of oak in the and Schütz, 2001; Tabaku, 1999; Meyer et al., 2003) with fluc-
living stand as a significant factor in the NRCD model. tuations depending on development phase and defined threshold
The major role of disturbances in forest ecosystem dynamics diameter. Saniga and Schütz (2001) estimated an average percent-
in general (Pickett and White, 1985), and dead wood dynamics age share of 16–29% of dead wood (referring to the total stock
in particular, is beyond doubt and has already been demon- of living and dead biomass) for the whole development cycle of
strated in models (Siitonen, 2001; Harmon, 2009). Nevertheless, virgin beech forests. Korpel (1995) calculated an average wood vol-
disturbance-driven mortality needs to be differentiated from ume of about 600 m3 ha−1 in the living stand of primeval beech
density-dependent and age-related mortality, which are also forests in equilibrium. Thus, given a range of 16–29% for dead wood
important sources of dead wood. Furthermore crown and branch percentage, the volume of wood in the living stand would range
breakage also contribute to the dead wood pool. from 111 to 224 m3 ha−1 . Results modeled for abandoned beech
In contrast, Vandekerkhove et al. (2009) found higher dead forests fall within this range. On the basis of the generic model
wood accumulation rates in beech-dominated forests than in BEFORE, Rademacher and Winter (2003) estimated a mean dead
oak-dominated forests. This may be attributed to differences in dis- wood volume of 148 m3 ha−1 after 2250 years without interven-
turbance regimes, as, in this case, oak decline plays only a minor tion. All values cited exceed by far the average dead wood stock
role. measured in our SFRs.
The absence of any relationship between abiotic ecological When comparing different reserves it should be noted that even
variables (length of growing season, nutrient supply) and NRCD forests that have been abandoned from forestry interventions for a
indicates the stochastic, disturbance driven nature of the net long period of time do not represent primeval forests. From a sim-
change in dead wood within the region and period of time stud- ulation study, Rademacher et al. (2001) concluded that, in beech
ied. The ecological gradient may be too short to cause detectable forests, non-intervention must continue over several generations
variation in dead wood decay. As a matter of fact several abiotic fac- before a primeval structure can be achieved. The difference in age
tors like pH, temperature and water regime have been proved to be structure in previously managed, and primeval forests may lead to
important determinants of wood decay in many studies (Boddy and diverging dead wood stocks even after a long period of abandon-
Swift, 1984; Harmon et al., 1986; Rayner and Boddy, 1988; Garrett ment. In even-aged, formerly managed stands higher spatial and
et al., 2007) as they affect the kind and rate of microbial (fungal) temporal fluctuations in dead wood stocks can be assumed than in
wood decomposition (Boddy, 2001). Notwithstanding there is no multi-cohort primeval forests.
general agreement on which of these factors are of primary impor- The accumulation rates recorded in abandoned forests may
tance for wood decay (Mackensen et al., 2003). therefore differ to a high degree with respect to, e.g., dis-
In our model input drives NRCD because the variation of decay turbance regime, age structure and tree species composition.
between SFRs and plots is either small or depends on factors we However, all recently published values exceed the NRCD we
could not examine, e.g. dimension, decay stage or position of dead found. Vandekerkhove et al. (2009) derived a mean accumula-
wood pieces, which have been found to be important in other tion rate in beech and oak forest reserves of 2.6 m3 ha−1 a−1 .
studies (Brown et al., 1996; Naesset, 1999; Janisch et al., 2005; Oheimb et al. (2007) calculated an increase in dead wood from 2.9
Müller-Using, 2005; Beets et al., 2008; Herrmann and Bauhus, to 111.6 m3 ha−1 in SFR-Serrahn, north-eastern Germany, after a
2008; Müller-Using and Bartsch, 2009). In the SFRs studied decay 35-year period. The corresponding accumulation rate amounted
may become a more prominent process as more dead wood is accu- to 3.1 m3 ha−1 a−1 . Müller-Using and Bartsch (2007) recorded an
mulated. increase in coarse dead wood from 28.2 to 67.5 m3 ha−1 in an
In future, sustainable, or even accelerated accumulation of dead abandoned beech stand in the Solling Mountains, north-west
wood is probable as the living stand ages. This will result in increas- Germany, after 8 years, resulting in a mean accumulation rate of
ing mortality rates. However, in our model, the initial amount of 4.9 m3 ha−1 a−1 .
dead wood and NRCD are negatively correlated, which would result In respect of site conditions and tree species composition the SFR
in decreasing accumulation rates over time. Age was not found to observed in our study are comparable to the forest reserves exam-
be an explanatory factor for NRCD. ined by Müller-Using and Bartsch (2007), Oheimb et al. (2007) and
This negative correlation between initial dead wood stock and Vandekerkhove et al. (2009). However, the duration of abandon-
NRCD may be an artefact that can be attributed to the generally ment from forestry interventions is shorter. Therefore with ongoing
 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352 351

time since abandonment we assume that the accumulation rates in Bohn, U., Gollub, G., 2007. Buchenwälder als natürliche Vegetation in Europa. Natur
our SFR will also increase. und Landschaft 82 (9/10), 391–397.
Brown, S., Mo, J., McPherson, J.K., Bell, D.T., 1996. Decomposition of woody debris
in Western Australian forests. Canadian Journal of Forest Research 26, 954–
5. Application of results 966.
Brunet, J., Falkengren-Grerup, U., Tyler, G., 1996. Herb layer vegetation of south
Swedish beech and oak forests – effects of management and soil acidity during
Our results suggest that, of several management options for one decade. Forest Ecology and Management 88, 259–272.
old growth attributes (see Bauhus et al., 2009), setting aside har- Brunet, J., Falkengren-Grerup, U., Rühling, A., Tyler, G., 1997. Regional differences
in floristic change in south Swedish oak forests as related to soil chemistry and
vestable pure and mixed beech forests from forestry interventions land use. Journal of Vegetation Science 8, 329–336.
is a potentially effective way of restoring the dead wood compo- Bücking, W., Parviainen, J., Schuck, A., 2000. Netzwerk Europäischer Naturwal-
nent. In contrast to Keeton (2006), in most instances the generation dreservate. AFZ – Der Wald 11, 565–567.
Bütler, R., Angelstam, P., Ekelund, P., Schlaepfer, R., 2004. Dead wood threshold val-
of old growth structures through active management, such as
ues for the three-toed woodpecker presence in boreal and sub-Alpine forests.
girdling or pulling down trees, appears unnecessary. Exceptions Biological Conservation 119, 305–318.
may arise if highly endangered species require urgent action. Christensen, M., Hahn, K., Mountford, E., Ódor, P., Standovár, T., Rozenbergar, D.,
Diaci, J., Wijdeven, S., Meyer, P., Winter, S., Vrska, T., 2005. Dead wood in Euro-
Notwithstanding this, measures to increase dead wood during
pean beech (Fagus sylvatica) forest reserves. Forest Ecology and Management
forestry interventions are advisable, e.g. leaving resident trees or 210, 267–282.
logging residue (Siitonen, 2001; Bauhus et al., 2009). Edman, M., Jönsson, M., Jonsson, B.G., 2007. Fungi and wind strongly influence the
However, restoration of structural features does not equal temporal availability of logs in an old-growth spruce forest. Ecological Applica-
tions 17 (2), 482–490.
restoration of the old-growth biological community as a whole. European Communities, 2003. Natura 2000 and Forests: Challenges and Oppor-
Taking into consideration that most of the endangered saproxylic tunities. Interpretation Guide, Office for Official Publications of the European
species have limited dispersal abilities (Ranius, 2001; Sippola et Communities, Luxembourg.
Franklin, J.F., Shugart, H.H., Harmon, M.E., 1987. Tree death as an ecological process.
al., 2001; Hedin et al., 2008) there is the danger that colonization of BioScience 37, 550–556.
old-growth structures far away from the source population will fail. Garrett, L., Davis, M., Oliver, G., 2007. Decomposition of coarse woody debris, and
Therefore proper spatial distribution of unmanaged forests close to methods for determining decay rates. New Zealand Journal of Forestry Sciences
37, 227–240.
source populations is a further prerequisite for effective restora- Gauer, J., Aldinger, E. (Eds.), 2005. Waldökologische Naturräume Deutschlands.
tion programs (Bailey, 2007; Hedin et al., 2008; Lachat and Bütler, Forstliche Wuchsgebiete und Wuchsbezirke mit Karte 1:1.000.000. Mitteilun-
2009). gen des Vereins für Forstliche Standortskunde und Forstpflanzenzüchtung 43.
Goldstein, H., 2003. Multilevel Statistical Models, third ed. Arnold, London.
The parameterized regression models reveal a high unexplained
Graae, B.J., Heskjaer, V.S., 1997. A comparison of understorey vegetation between
variance. As a result, the prediction of NRCD and future dead wood untouched and managed deciduous forest in Denmark. Forest Ecology and Man-
amount for small areas of 500 m2 or 1000 m2 may be too uncertain agement 96, 111–123.
Grove, S.J., Stork, N.E., 1999. The conservation of saproxylic insects in tropical forests:
for practical purposes. Hence further investigations are necessary
a research agenda. Journal of Insect Conservation 3, 67–74.
to answer the following questions: does the random variation in Hanski, I., Walsh, M., 2004. How Much, How to? – Practical Tools for Forest Conser-
NRCD level out when the analysis is conducted at the stand or pro- vation, BirdLife European Forest Task Force, Finland.
tective areas levels?; could a dead wood assessment carried out Harmon, M.E., 2001. Moving towards a new paradigm for woody detritus manage-
ment. Ecological Bulletins 49, 269–278.
as part of regular forest inventories be used to calibrate mean sta- Harmon, M.E., 2009. Woody detritus mass and its contribution to carbon dynamics
tistical models?; and, to what extent can the prediction accuracy of old-growth forests: the temporal context. In: Wirth, C., Gleixner, G., Heimann,
be improved for different regional scales by using the available M. (Eds.), Ecological Studies, vol. 207: Old-Growth Forests. Springer, Berlin Hei-
delberg, pp. 159–190.
inventory data? Harmon, M.E., Franklin, J.F., Swanson, F.J., 1986. Ecology of coarse woody debris
in temperate ecosystems. In: MacFayden, A., Ford, E.D. (Eds.), Advances in
Ecological Research, vol. 15. Academic Press, London, New York, pp. 133–
Acknowledgements 302.
Hartmann, G., Blank, R., 1998. Aktuelles Eichensterben in Niedersachsen – Ursachen
We are indebted to Roland Steffens, Stefan Brößling, Friedrich und Gegenmaßnahmen. Forst und Holz 53 (24), 733–735.
Hedin, J., Ranius, T., Nilsson, S.G., Smith, H.G., 2008. Restricted dispersal in a flying
Eimer and many other fieldworkers for gathering the extensive
beetle assessed by telemetry. Biodiversity and Conservation 17, 675–684.
database we had the chance to analyze. We like to thank Helen Herrmann, S., Bauhus, J., 2008. Comparison of methods to quantify respirational
Desmond for her thorough revision of the language and two anony- carbon loss of coarse woody debris. Canadian Journal of Forest Research 38 (11),
2738–2745.
mous reviewers for their helpful comments on the manuscript. The
Janisch, J.E., Harmon, M.E., Chen, H., Fasth, B., Sexton, J., 2005. Decomposition of
study was supported by the “Deutsche Bundesstiftung Umwelt” coarse woody debris originating by clearcutting of an old-growth conifer forest.
(DBU). Ecoscience 12, 151–160.
Kahl, T., 2008. Kohlenstofftransport aus dem Totholz in den Boden., Ph.D. thesis,
University of Freiburg.
References Keeton, W.S., 2006. Managing for late-successional/old-growth characteristics
in northern hardwood-confer forests. Forest Ecology and Management 235,
Albrecht, L., 1991. Die Bedeutung des toten Holzes im Wald. Forstwissenschaftliches 129–142.
Centralblatt 110, 106–113. Korpel, S., 1995. Die Urwälder der Westkarpaten, Gustav Fischer.
Aude, E., Lawesson, J.E., 1998. Vegetation in Danish beech forests: the importance of Lachat, T., Bütler, R., 2009. Identifying conservation and restoration priorities for
soil, microclimate and management factors, evaluated by variation partitioning. saproxylic and old-growth forest species: a case study in Switzerland. Environ-
Plant Ecology 134 (1), 53–65. mental Management 44, 105–118.
Bailey, S., 2007. Increasing connectivity in fragmented landscapes: an investigation Lutz, J.A., Halpern, C.B., 2006. Tree mortality during early forest development: a long-
of evidence for biodiversity gain in woodlands. Forest Ecology and Management term study of rates, causes, and consequences. Ecological Monographs 76 (2),
238, 7–23. 257–275.
Bauhus, J., Puetmann, K., Messier, Ch., 2009. Silviculture for old-growth attributes. Mackensen, J., Bauhus, J., Webber, E., 2003. Decomposition rates of coarse woody
Forest Ecology and Management 258, 525–537. debris – a review with particular emphasis on Australian species. Australian
Beets, P.N., Hood, I.A.b, Kimberley, M.O., Oliver, G.R., Pearce, S.H., Gardner, J.F., 2008. Journal of Botany 51 (1), 27–37.
Coarse woody debris decay rates for seven indigenous tree species in the cen- Martikainen, P., Siitonen, J., Punttila, P., Kaila, L., Rauh, J., 2000. Species richness
tral north island of New Zealand. Forest Ecology and Management 256, 548– of Coleoptera in mature managed and old-growth boreal forests in southern
557. Finland. Biological Conservation 94, 199–209.
Boddy, L., 2001. Fungal community ecology and wood decomposition processes in Maser, C., Andersen, R.G., Cromack, K., Williams, J. T., Martin, R.E., 1979. Dead and
angiosperms: from standing tree to complete decay of coarse woody debris. Down Woody Material. In: Thomas, J.W. (Hrsg.), Wildlife habitats in Managed
Ecological Bulletins 49, 43–56. Forests the Blue Mountains of Oregon and Washington. U.S. Department of Agri-
Boddy, L.B., Swift, M.J., 1984. Wood decomposition in an abandoned beech and oak culture, Forest Service, Agriculture Handbook 553, 78–95.
coppiced woodland in SE England. III. Decomposition and turn over of twigs and Meyer, P., 1999. Totholzuntersuchungen in nordwestdeutschen Naturwäldern:
branches. Holartic Ecology 7, 229–238. Methodik und erste Ergebnisse. Forstwirtschaftliches Centralblatt 118, 167–180.
352 P. Meyer, M. Schmidt / Forest Ecology and Management 261 (2011) 342–352

Meyer, P., 2005. Network of Strict Forest Reserves as reference system for close to Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core team, 2009. nlme: Linear and
nature forestry in Lower Saxony, Germany. Forest Snow and Landscape Research Nonlinear Mixed Effects Models.
79 (1/2), 33–44. Pinheiro, J., Bates, D., 2000. Mixed-effects Models in S and S PLUS. Springer, New
Meyer, P., Bücking, W., Gehlhar, U., Schulte, U., Steffens, R., 2007. Das Netz der York.
Naturwaldreservate in Deutschland: Flächenumfang, Repräsentativität und Rademacher, C., Winter, S., 2003. Totholz im Buchenwald: Genetische Vorhersagen
Schutzstatus im Jahr 2007. Forstarchiv 78, 188–196. des Simulationsmodells BEFORE-CWD zur Menge, räumlichen Verteilung und
Meyer, P., Schmidt, M., 2008. Aspekte der Biodiversität von Buchenwäldern – Verfügbarkeit. Forstwiss. Zentralblatt 122, 337–357.
Konsequenzen für eine naturnahe Bewirtschaftung. Beitrag Nordwestdeutsche Rademacher, C., Neuert, C., Grundmann, V., Wissel, C., Grimm, V., 2001. Was charak-
Forstliche Versuchsanstalt 3, 159–192. terisiert Buchenurwälder? Untersuchungen der Altersstruktur des Kronendachs
Meyer, P., Tabaku, V., Lüpke, v.B., 2003. Die Struktur albanischer und der räumlichen Verteilung der Baumriesen in einem Modellwald mit
Rotbuchen-Urwälder – Abteilung für eine naturnahe Buchenwirtschaft. Hilfe des Simulationsmodells BEFORE. Forstwissenschaftliches Centralblatt 120,
Forstwissenschaftliches Centralblatt 122, 47–58. 288–302.
Mindermann, G., 1968. Addition, decomposition and accumulation of organic matter R Development Core Team, 2009. R: A language and environment for statistical
in forests. Journal of Ecology 56, 355–362. computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-
Müller, J., Strätz, C., Hothorn, T., 2005. Habitat factors for land snails in European 900051-07-0, URL [Link]
beech forests with a special focus on coarse woody debris. European Journal of Ranius, T., 2001. Constancy and asynchrony of Osmoderma eremita populations in
Forest Research 124, 233–242. tree hollows. Oecologia 126, 208–215.
Müller, J., Hothorn, T., Pretzsch, H., 2007. Long-term effects of logging intensity Rayner, A.D.M., Boddy, L., 1988. Fungal Decomposition of Wood: Its Biology and
on structures, birds, saproxylic beetles and wood-inhabiting fungi in stands of Ecology. Wiley, New York.
European beech Fagus sylvatica L. Forest Ecology and Management 242, 297–305. Rock, J., Badeck, F.-W., Harmon, M.E., 2008. Estimating decomposition rate constants
Müller, J., Bütler, R., 2010. A review of habitat thresholds for dead wood: a base- for European tree species from literature sources. European Journal of Forest
line for management recommendations. European Journal of Forest Research, Research 127, 301–313.
doi:10.1007/s10342-010r-r0400-5, published online 19 June 2010. Saniga, M., Schütz, [Link]., 2001. Dynamics of changes in dead wood share in selected
Müller-Using, S., 2005. Totholzdynamik eines Buchenbestandes im Solling. Berichte beech virgin forests in Slovakia within their development cycle. Journal of Forest
des Forschungszentrums Waldökosysteme, Reihe A, 193. Science 47 (12), 557–565.
Müller-Using, S., Bartsch, N., 2007. Totholz im Elementhaushalt eines Buchenbe- SAS Institute Inc., 2009. Version 9.1.
standes. Forstarchiv 7, 12–23. Schaber-Schoor, G., 2008. Wie viel Totholz braucht der Wald – Ergebnisse einer Lit-
Müller-Using, S., Bartsch, N., 2009. Decay dynamics of coarse and fine woody debris eraturrecherche als Grundlage für ein Alt-, Totholz- und Habitatbaumkonzept.
of a beech (Fagus sylvatica L.) forest in Central Germany. European Journal of FVA-Einblick 2, 5–8.
Forest Research 128, 287–296. Schmidt, A., 1967. Der rechnerische Ausgleich von Bestandeshöhenkurven.
Naesset, E., 1999. Decomposition rate constants of Picea abies logs in southeastern Forstwissenschaftliches Centralblatt 86, 370–382.
Norway. Canadian Journal of Forest Research 29 (3), 372–381. Schuck, A., Meyer, P., Menke, N., Lier, M., Lindner, M., 2004. Forest biodiver-
Nagel, J., 1999. Konzeptionelle Überlegungen zum schrittweisen Aufbau eines sity indicator: dead wood: a proposed approach towards operationalising the
waldwachstumskundlichen Simulationssystems für Nordwestdeutschland. MCPFE indicator. In: Marchetti, M. (Ed.), Monitoring and Indicators of For-
Schriften aus der Forstlichen Fakultät der Uni. Göttingen und der Niedersä. est Biodiversity in Europe – from Ideas to Operationality. EFI Proceedings 51,
Forstlichen Versuchsanstalt, 128. 49–77.
Odor, P., Heilmann-Clausen, J., Christensen, M., Aude, E., van Dort, K.W., Piltaver, Siitonen, J., 2001. Forest management, coarse woody debris and saproxylic organ-
A., Siller, I., Veerkamp, M.T., Walleyn, R., Standovar, T., van Hees, A.F.M., Kosec, isms: Fennoscandian boreal forests as an example. Ecological Bulletins 49,
J., Matocec, N., Kraigher, H., Grebenc, T., 2006. Diversity of dead wood inhabi- 11–41.
tating fungi and bryophytes in semi-natural beech forests in Europe. Biological Sippola, A.L., Siitonen, J., Kallio, R., 2001. Amount and quality of coarse woody debris
Conservation 131, 58–71. in natural and managed coniferou forests near the timberline in Finnish Lapland.
Oheimb, v.G., Friedel, A., Tempel, H., Westphal, C., Härdtle, W., 2004. Unter- Scandinavian Journal of Forest Research 13 (2), 204–214.
suchungen zur Struktur und zur Moos- und Flechtenflora in unbewirtschafteten Speight, M.C.D., 1989. Saproxylic invertebrates and their conservation. Council of
und bewirtschafteten Buchenwäldern des Norddeutschen Tieflandes. Beitr. Europe. Nature and Environment Series 42, 1–79.
Forstwirtschafts- und Landschaftsökologie 38 (2), 81–86. Tabaku, V., 1999. Struktur von Buchen-Urwäldern in Albanien im Vergleich mit
Oheimb, v.G., Westphal, C., Tempel, H., Härdtle, W., 2005. Structural pattern of a deutschen Buchen-Naturwaldreservaten und – Wirtschaftswäldern. Cuvilier
near-natural beech forest (Fagus sylvatica) (Serrahn, North-east Germany). For- Verlag.
est Ecology and Management 212, 253–263. UNECE, 2010. [Link] at 12th
Oheimb, v.G., Westphal, C., Härdtle, C.W., 2007. Diversity and spatio-temporal May 2010.
dynamics of dead wood in a temperate near-natural beech forest (Fagus syl- Vandekerkhove, K., De Keersmaeker, L., Menke, N., Meyer, P., Verscheide, P., 2009.
vatica). European Journal of Forest Research 126, 359–370. When nature takes over from man: dead wood accumulation in previously man-
Olson, J.S., 1963. Energy storage and the balance of producers and decomposers in aged oak and beech woodlands in North-western and Central Europe. Forest
ecological systems. Ecology 44 (2), 322–331. Ecology and Management 258, 425–435.
Paillet, Y., Bergès, L., Hjältén, J., Ódor, P., Avon, C., Bernhardt-Römermann, M., Bijlsma, Winter, S., Flade, M., Schumacher, H., Kerstan, E., Möller, G., 2005. The importance
R.J., de Bruyn, L., Fuhr, M., Grandin, U., Kanka, R., Lundin, L., Luque, S., Magura, T., of near-natural stand structures for the biocoenosis of lowland beech forests.
Matesanz, S., Mészáros, I., Sebastià, M.T., Schmidt, W., Standovár, T., Tóthmérész, Forest Snow and Landscape Research 79 (1/2), 127–144.
B., Uotila, A., Valladares, F., Vellak, K., Virtanen, R., 2010. Biodiversity differences Wirth, Ch., Messier, Ch., Bergeron, Y., Frank, D., Frankhänel, A., 2009. Old-growth
between managed and unmanaged forests: meta-analysis of species richness in forest definitions: a pragmatic view. Ecological Studies 209, 11–33.
Europe. Conservation Biology 24 (1), 101–112. Wolff, B., Erhard, M., Holzhausen, M., Kuhlow, T., 2003. Das Klima in den Wuchsge-
Peterken, G.F., 1996. Natural Woodland: Ecology and Conservation in Northern Tem- bieten Deutschlands. Mitteilungen der Bundesforschungsanstalt für Forst- und
perate Regions. Cambridge Univ. Press, New York. Holzwirtschaft Hamburg, 211 pp.
Pickett, S.T.A., White, P.S., 1985. The Ecology of Natural Disturbance and Patch Woodall, C.W., Verkerk, H., Rondeux, J., Stahl, G., 2009. Who’s counting dead wood?
Dynamics. Academic Press, Orlando. EFI News 2, 12–13.