O R I G I N A L P A P E R

Klaus von Wilpert Æ Ju¨rgen Scha¨ffer

Ecological effects of soil compaction and initial recovery dynamics:
a preliminary study

Received: 26 November 2004 / Accepted: 13 May 2005 / Published online: 10 December 2005
Springer-Verlag 2005

Abstract Skid trails of regular harvest operations with
time delays up to 24 years between tracking and exam-
ination were investigated in order to characterize the
status of recovery of essential soil functions. The study
was focused on the ability of soils to serve as an effective
rooting space. The gas diffusion coefficient and the fine
root distributions of comparable sensitive silty loams
were used to describe the disturbances of soil functions
still detectable after decades. Tracking with heavy loa-
ded machines severely reduced the soil aeration and in
consequence the ability of the soil to act as a rooting
space. Reduction of gas diffusivity and rooting was
found on the whole skidding trail area and even ex-
panded to the close vicinity of the margin zone. Up to
14 years after machine impact, gas diffusion coefficients
and root densities beyond 4 cm depth under wheel
tracks showed no signs of restoration. Soil aeration and
root densities comparable to the control plot were found
in the top soil layer 18 years after tracking at the site
Ettenheim. At that location 24 years after machine im-
pact, significantly reduced root densities occurred only
in soil depths beyond 54 cm. In the long run, only
concentration of machine traffic on permanent skid trail
systems guarantees an adequate soil preservation. This
applies especially under Middle European site condi-
tions and management practices. This prescription must
be underlined in guidelines for machine use in forests.

Keywords Soil compaction Æ Soil aeration Æ Fine root
distribution Æ Soil structure recovery

Introduction

Harvesting and skidding machines are prepared to
handle heavy loads under off-road conditions. Thus the
unprotected forest soil has to serve as a weak receptor
body for static and dynamic forces caused by tracking
machines (Bredberg and Wa¨sterlund

1983

; Horn

1988

).

The resulting soil deformation occurs in two ways. On
the one hand, the soil pore volume is destructed until an
equilibrium between external impact and counter-forces
of the compacted soil is reached. On the other hand, if
traffic takes place on soils close to the water saturation
point, incompressible water fillings preserve the pore
volume, especially of the macro-pores. In this case, the
machine ‘‘swims’’ in the weak soil and deformation
primarily leads to a soil displacement and loss of macro-
pore continuity rather than a reduction of soil pore
volume. In both cases, it is very likely that the naturally
high continuous pore system generated and maintained
by the edaphon is severely damaged having negative
effects on transfer processes like, for example, gas and
water fluxes (Hildebrand

1987

). Site conditions like the

water status of the soil or soil texture vary the ecological
soil properties only in a narrow range. In any case, it has
to be feared that compaction and de-structuration of
soils is responsible for a long-lasting reduction of root-
ing capacity below skidding trails (Hildebrand

1994

).

The leading hypothesis of this project states that the

reduction of soil aeration is the key factor limiting the
ability of soils to serve as rooting space. Soil compaction
and soil displacement predominantly occur at the soil
surface, which is the interface for oxygen supply and
carbon dioxide release between soil and atmosphere. In
comparison with other plant compartments, the demand
for oxygen of growing roots is over-proportionally high.
In germination and planting tests, Hildebrand (

1983

,

1986

) found negative compaction effects on the seedbed

function and fine root production for beech and spruce
seedlings. He interpreted these findings as a consequence
of reduced oxygen supply. Murach et al. (

1993

) reported

Communicated by Hans Pretzsch

K. von Wilpert Æ J. Scha¨ffer (

&)

Department of Soil Science, Forest Research Institute
Baden-Wuerttemberg, Wonnhaldestr. 4, 79100 Freiburg, Germany
E-mail: [e-mail]
E-mail: juergen.schaeff[e-mail]
Tel.: [phone]
Fax: [phone]

Eur J Forest Res (2006) 125: 129–138
DOI 10.1007/s10342-005-0108-0

higher fine root growth rates after artificial oxygen
enrichment in the deeper mineral soil. Qi et al. (

1994

)

demonstrated that the growth rate of Douglas fir roots
decreased, under laboratory conditions, exponentially
with increasing CO

2

concentration, which corresponds

with dropping oxygen availability.

The result of reduced oxygen supply and carbon

dioxide disposal is a reduction of rooting intensity and
above all, a retreat of roots from deeper soil horizons
because the threshold for sufficient oxygen supply is
already reached in the upper soil layers. As long as
sufficient water and nutrient resources are still available
in the upper soil layers, trees may compensate the loss of
rooting space by higher uptake rates, and timber growth
along skid trails will be maintained although the stability
of the stands and single trees will be affected because the
risk of water and nutrient shortage is increased when
storage capacity is reduced.

Apart from the basic relation between soil deforma-

tion, soil aeration, and the fine root densities, this pro-
ject observed the recovery of the soil as rooting space. In
the context of forest conversion, the demands on
machines are very high in mixed, intensively structured
stands. A predominant aim of silviculture in these stands
is the natural regeneration. Because of the patchiness of
young tree groups in the uneven aged forests and the
intensive horizontal mixture of trees with different ages
the equipment must be able to handle heavy loads over
wide distances with a crane in order to protect these
centres of the next forest generation. On the contrary the
machines have to be extraordinarily heavy with the
consequence of higher static and dynamic forces trans-
mitted to the soil, which leads to deeper soil deformation
and longer recovery periods.

First qualitative results of measurements on former

tracking experiments showed that an effective recovery of
soil aeration seemed to need at least decades. At a con-
trolled tracking experiment at Emmendingen, even
12 years after wheel passage a highly significant reduction
of the gas diffusion coefficient persisted in the upper soil
layers (Schack-Kirchner

1994

). In our project it was in-

tended to identify the length of the recovery period by
investigating tracking trails with different time delays
after the tracking event. It has to be emphasized that this
study was based on machine impacts in practice and not
on tracking experiments under controlled side conditions.

Methods

For identification of the soil regeneration processes,
nearly stone-free loamy soils (Luvisols) were selected. In
all cases the A

h

and E-horizons had a soil texture

between loam and silty loam and stone contents of less
than 10%. The transition to the more compacted and
clayey B

t

was situated at about 40–50 cm depth.

In numerous investigations it has been shown that

substrates with grain size distributions reaching from the
clay to the sand fraction and especially silty loams are

very sensitive to soil compaction induced by forest
machines (Hildebrand

1983

,

1986

; Horn

1988

). Therefore

research on recovery dynamics on loams can be inter-
preted as ‘‘worst-case scenarios’’ in the context of sub-
strate-specific sensitivity.

Apart from standardized

substrate conditions it was important that the impacts of
the tracking machine, the load, and the machine type
were comparable. This aspect restricted the length of the
observation period, since specialized forest machines with
total loads of 10 Mg and more, as they are used today,
appeared in practice in the 1960s (SkogForsk

1997

).

Before that time agricultural tractors with total loads up
to 5 Mg as ‘‘all-round machines’’ had been used. In all
cases the skid trails for our investigations were part of
equidistant skid trail systems. All trackings during the
regular logging and harvesting operation were concen-
trated on these trails. So the impact on the soil was a
result of multi-passing during practical forest operations.
Because most of the soil compaction is realized within the
first passages (Loeffler

1985

), we can assume that the soil

damages on the sites were comparable even if the number
of passages differed among the sites. An important point
was that a possible later strain by forest machines after a
spatially defined and well-documented tracking event
should be excluded. Since forest operations and the initial
impact on the investigation sites were not documented, an
inquiry was done at the forest offices in Baden-Wuert-
temberg to ensure that no further passage had taken
place. The four investigation sites presented in this paper
represent time ranges of natural soil structure recovery of
less than 10 years, between 10 and 20 years, and between
20 and 30 years.

Recovery stages

A survey of the investigation sites and the conditions of
logging operations is given in Table

1

. The canopies of

the young stands were closed so that growth of herbal
vegetation was impossible. The youngest stage was
represented by the investigation site ‘‘Weil im Schoen-
buch’’, where a thinning campaign with the medium-
load harvester Timberjack 1270 and the processor Ponse
HS10 in combination with a Valmet forwarder took
place in 1996 in an about 45-year-old mixed beech/
spruce stand. The investigation took place 6 years after
the machine impact. All machines were characterized by
total machine weights of 10–20 Mg including load. It
could be assumed that within the 6-year period after the
machine impact the fine root propagation reached a
quasi-equilibrium with the new physical properties.

A medium-term recovery stage was supposed to be

realized at ‘‘Wolfegg’’ where the thinning campaign took
place in 1989. After this machine impact no further
tracking occurred. Today the forest type is an about
50-year-old spruce stand. At Wolfegg a harvester of the
type FMG 746/250 O¨SA ‘‘Super Eva’’ was used in com-
bination with the forwarder FMG 678, ‘‘Mini Bruunett’’.
Both machines had typical loads up to 15 Mg (Table

1

).

130

The harvester generated an incomplete brushwood mat,
which could not effectively protect the soil from defor-
mation (Kremer

1998

). Under comparable stand condi-

tions, Scha¨ffer et al. (

1991

) showed that an effective soil

protection by a sufficiently thick brush mat (more than
25 cm thickness after machine passage) was only realized
on 21% of the skid trail length.

The longest recovery phases after soil compaction

were given at ‘‘Ettenheim’’, where 18 and 24-year-old
skidding trails were investigated. Skidders had been used
there, which were comparable to actual machine types
(Welte O¨konom and Unimog U90). The machine impact
took place during regular harvesting in the form of strip
cutting. For this machine impact also, typical loads of
10 Mg were assumed on the basis of usual machine
configurations at that time (Table

1

). Even if only

skidding tractors and no harvesters were used in this
example the soil compaction was supposed to be com-
parable to the other examples. Additional soil compac-
tion by further machine impact at the same track area
does only occur if loads get substantially higher (Horn

1988

). This could not be expected for small and medium

wheeled harvesters in comparison to skidding machines
as they were used here. During harvesting and logging
the machine activities were concentrated on skidding
trails. But it could not be doubtlessly excluded that an
additional machine impact was caused by farmers with
agricultural tractors collecting fuel wood. At this site a
forest regeneration in the pole age stadium preserved the
area from further skidding. The location of the logging

trails could be unambiguously identified with the help of
a local forest official.

Observations and measurements

At each site two transects with orthogonal orientation
had been dug across either one or two skidding trails.
The trench walls at the skidding trails were 3.8–4.2 m
wide and 0.6–0.8 m deep. The evaluation of soil physical
measurements and counts of rooting densities were
stratified in wheel tracks (normally 80 cm wide), the
median strip (about the same width as the wheel tracks),
and the margin zones of the skidding trails (about 80–
100 cm at both margins). These strata were identified in
the field according to the position of wheel tracks, which
were visible even after decades at the soil surface as
depressions of several centimetres. At each investigation
site a corresponding control plot (varying width of 0.4 m
at Weil im Schoenbuch, 1.0 m at Ettenheim, and 2.0 m
at Wolfegg, depth according to skid trail profiles) was
identified where tracking of machines could be excluded
with high security. In the elder stands these were pref-
erentially locations with several trees in such close
vicinity that machine passage was impossible. For the
identification of the control area in the young stand of
Ettenheim, only the absence of visible tracks at the soil
surface could be used as an indicator. Thus there was a
higher degree of uncertainty in the definition of the local
reference.

Table 1 Description of site conditions and conditions of logging operations

Weil im Schoenbuch

Wolfegg

Ettenheim 1

Ettenheim 2

Location

Triassic Keuper landscape

Prealpin landscape

Premountains of the Black
Forest

Geology

Loess over Angulaten sandstone

Glacial loam (Wu¨rm ground moraine)

Loess over Buntsandstein

Soil type

Luvisol

Luvisol

Luvisol

Altitude in m asl

485

755

490

Stand description

Planted spruce

a

-dominated pole-sized

stand with beech

b

and larch

c

Planted spruce

a

pole-sized stand

Beech

b

-dominated

young

growth with oak

d

, spruce,

and pine

e

, mainly originat-

ing from seedling regenera-
tion

Stand age

35–40

50–55

18

24

Operation

Thinning

Thinning

Stripwise clear-cut

Harvesting

Timberjack 1270 (16 Mg weight

and 600 mm tires); Ponse HS10
(

15 Mg weight and 600 mm tires)

FMG 746/250 O¨SA Super Eva

(12.6 Mg weight and 600 mm tires)

Motor manual

Skidding

Valmet Forwarder (weight >15 Mg

including load, tires 500 mm)
(Pfeil

2003

)

FMG 678 Mini Bruunett (8.3 Mg

weight, maximum load of 7.5 Mg,
500 mm tires) (Kremer

1998

)

Welte O¨konom and Unimog
U90 (weight of

6–7 Mg,

loads up to 3 ton, 300–
500 mm tires) (Pfeil

2003

)

Date of logging

1996

1989

1984

1978

Time of soil structure

recovery (years)

6

12

18

24

a

Norway spruce [Picea abies (L.) Karst.]

b

European beech (Fagus sylvatica L.)

c

European larch (Larix decidua Mill.)

d

Sessile oak (Quercus petraea Liebl.)

e

Scots pine (Pinus sylvestris L.)

131

For counting of roots at the transects the profile wall

method described by Bo¨hm (

1979

) was used. After

machine-aided digging and manual fine preparation of
the profile walls the fine roots (<2 mm) were counted
using 4

·4 cm grids to characterize root densities and

distribution patterns.

At every skid trail strata of one transect and at the

control area five replicates of undisturbed soil columns
(100 cm

3

) were gained in four depth layers down to

34 cm (0–4, 10–14, 20–24, and 30–34 cm) in order to
analyse soil physical parameters.

Apart from the ‘‘classical’’ soil physical parameters

like bulk density (determined by gravimetry and drying
at 105

C, Hartge and Horn

1992

), porosity (vacuum

pycnometry according to Danielson and Sutherland

1986

), and pore class distribution of the wide pores

(determination of the pF-function = relation between
water suction and water content on the sand bed, Hartge
and Horn

1992

), the relative apparent gas diffusion

coefficient (D

s

/D

0

) was determined using the static

chamber method of Frede (

1986

). In water-filled pores,

gas diffusion is reduced to a negligible dimension.
Therefore D

s

coefficients have to be determined in rela-

tion to known water suctions. We determined D

s

at

water suctions of 60 hPa representing a status of high
water saturation in spring and at 300 hPa as a stage of
rather dry conditions.

The D

s

/D

0

ratio is a proportionality factor without

dimension, which characterizes the diffusion velocity of
air or other gases in the soil pore system in comparison
to their diffusion velocity in the free atmosphere.
Because the gas diffusion coefficient is mainly influenc-
ing the gas exchange between soil and atmosphere, the
presentation of the soil physical investigations is focused
on this parameter.

For the ability of the soil to serve as rooting space the

gas permeability at the interface between soil and
atmosphere is a key factor, because this soil layer acts as
a ‘‘bottleneck’’ for the oxygen supply and the carbon
dioxide disposal (Hildebrand

1994

). In deeper soil

horizons D

s

indicates the depth to which the machine

impact was forwarded in maximum by the impulse
transfer through the soil matrix.

The examinations of the soil physical properties and

root counting were done in 2001 and 2002.

Statistical analysis

For the multiple comparisons of the apparent gas dif-
fusion coefficients (D

s

/D

0

) of the different skid trail sit-

uations the non-parametric DUNN test was performed
(Zar

1999

). Differences between the groups were tested

at significance levels of P<0.05.

For the statistical comparison of the rooting densities

between wheel tracks and control plots of the different
regeneration stages a general linear model (GLM) was
used with the observed root densities in the sampled grid
cells as dependent variable and the depth of the grid cell

and the ‘‘treatment’’ as independent variable (McCulloch
and Searle

2001

). The treatment is a categorical variable

indicating if the cell belongs to wheel track or not.
Interactions of depth and treatment were included.
Because of the spatial characteristics of the sampling
design, correlation between neighbouring cell counts has
to be expected. Thus the standard assumption of inde-
pendent observations is probably not met in this case.
Another violation of the classical linear regression model
is the fact that the response is a count, which questions
the normality assumption. A Poisson distribution might
be more appropriate to describe the variability in the
observed data with increasing variance for higher count
data.

The spatial dependence was modelled through a

covariance function of the form

Kov

ðy

i

; y

j

Þ ¼ r

2

exp

d

ij

h

;

where d

ij

is the Euclidean distance between the two grid

points s

i

, s

j

of the sampled root densities y

i

, y

j

. The

covariance parameter h may be interpreted as the range
of spatial dependence. According to Wackernagel (

2003

)

the practical range is 3h, which is the distance where
95% of the maximum of the corresponding semivario-
gram is reached. For the regeneration stages the
covariance parameter h ranges between 2.53 (Ettenheim
1984) and 3.25 (Weil im Schoenbuch). As a consequence
we can state that the spatial dependency is limited to the
neighbouring 10–13 cm and does not exceed three grid
cells.

For the parameter estimation of the GLM model the

Macro Glimmix in SAS (Littell et al.

1996

) was used.

The testing of the fixed effect of tracking in the 4 cm
deep layers was done with the Tukey test, using the least-
square means option. The four wheel track situations on
the two transects were regarded as spatially independent
replications.

Results

Diffusive air permeability

The depth profile of mean D

s

/D

0

values of the control

plot was compared to the wheel track situation, the
median strip, and the margin zones of the corresponding
skidding trail in the same soil depth. The skidding trail
strata at the study site Weil im Schoenbuch presented in
Fig.

1

showed the most distinct differences among the

strata. This was the site with the most recent machine
impact.

The control plot at this study site provided the

highest D

s

/D

0

values of up to 0.07 in the upper soil,

which indicated a porous, highly permeable soil struc-
ture. In no soil horizon could a significant difference
between the D

s

/D

0

values between the control plot and

the margin zone of the skidding trail as well as between

132

the wheel track and the median strip of the skidding trail
be identified. The wheel track and median strip strata
provided significantly lower D

s

/D

0

values than the data

from the control plot and the margin zone strata, with
minimum values at the wheel track being one order of
magnitude lower than those at the control plot. The level
of D

s

/D

0

at a soil depth of 30–34 cm was comparable for

the different strata. The margin strip displays the highest
D

s

/D

0

values in the mineral soil between 10 and 34 cm

depth. That is obviously due to micro-scaled heteroge-
neities in primary soil texture, which is a common
finding in Triassic claystone areas. Thus the depth of 30–
34 cm can be cautiously assumed to indicate the lower
boundary of the deformed soil zone at that site.

Comparable relations existed among the skid trail

strata at the other sites. For the depth of 10–14 cm with

the most pronounced differences between wheel track
strata, the D

s

/D

0

values are presented in Table

2

. For the

younger stages Weil im Schoenbuch and Wolfegg, the
values were consistent showing the highest impact at the
wheel tracks (significant reduction of D

s

/D

0

by a factor

of about 5 as compared to control) and an intermediate
situation at the median strips (decrease by a factor of 2–
3). The levels of the diffusion coefficients at the margin
zones were comparable to the levels at the control sites.
At the sites Ettenheim 1984 and Ettenheim 1978, this
distinct sequence of soil aeration reduction was less
obvious. Besides the longer period of recovery at these
sites probably a higher number of passages during the
harvesting operation in the clear-cut situation could be
responsible for spreading out the machine impact to the
median strip and the margin zone.

In Fig.

2

depth profiles of D

s

/D

0

at the control plots

(left part) and the wheel tracks (right part) are compared
for the recovery stages represented by Weil im Schoen-
buch, Wolfegg, and Ettenheim.

At all sites the D

s

/D

0

values below the wheel tracks

were significantly lower, down to 25 cm soil depth as
compared to the values of corresponding control plots
except Ettenheim 1984 in 10–14 cm. The depth profiles
of D

s

/D

0

values below the wheel tracks at Weil im

Schoenbuch and Wolfegg were comparable. The slightly
increasing values with increasing soil depth at Wolfegg
may be interpreted as a result of reduced machine im-
pact at this site caused by lower load or protection by
brush mat. At the older recovery stages at Ettenheim,
which were generated in 1984 and 1978, the skidding
trails showed a remarkable increase of the D

s

/D

0

value

above 15 cm, which was accompanied by an increase of
variability. Both these findings may be cautiously ex-
plained by biological recovery processes beginning at the
soil surface and spreading stepwise into deeper soil
horizons.

At 30–34 cm soil depth the D

s

/D

0

values below wheel

tracks were still slightly lower compared with the control
plots, but only for Ettenheim 1984 the testing showed a
significant difference. This indicated that soil deforma-
tion was still detectable in 35 cm, but did not cause
significant ecological damage there.

At the control plots of Weil im Schoenbuch and

Wolfegg the D

s

/D

0

values were about one order of

magnitude higher than below the tracks. The findings at
Ettenheim partially did not fit into that pattern. The
control D

s

/D

0

values in 10–14 cm lay in the range of

areas with machine impact at this site. We cannot
exclude that this might be the result of a ‘‘hidden
wheeling damage’’ at the control area.

The soil as rooting space

A pronounced example for rooting patterns below a
skidding trail is presented for the study site Weil im
Schoenbuch with the most recent machine impact in
Fig.

3

.

Fig. 1 Depth profiles of the relative gas diffusion coefficient (D

s

/D

0

)

at a water suction of 60 hPa for the strata of the skidding trail in
Weil im Schoenbuch. Small characters indicate significant differ-
ences among the strata at levels of P<0.05 tested with the multiple
non-parametric Dunn test. Horizontal lines indicate the single
standard deviation

133

At the transect across the skidding trail, fine root

density and rooting depth had their lowest values below
the wheel tracks. Even in the uppermost, commonly very
intensively rooted 4 cm layer, several grid cells provided
no fine roots at both wheel tracks. The intensively roo-
ted zone with fine root densities above 20

· 100 cm

2

was very shallow or partially not existing at the wheel
track areas and was not deeper than 10–12 cm
throughout the whole skidding trail area. The lower
boundary of the extensively rooted zone followed the
wheel tracks in a depth from 12 to 20 cm, whereas at the
median strip it reached the depth of about 32 cm. At the
margin strips the intensively rooted zone was about
30–40 cm deep, the extensively rooted zone expanded
5–10 cm deeper. At the control plot the intensively
rooted zone exceeded with at least 40 cm depth the
boundary of margin zones and skidding trails. This
would indicate that the disturbance of soil aeration and,
as a consequence, reduced rooting densities are not
restricted to the soil directly below the heavily com-
pacted soil surface areas.

As a consequence of different time delays between

soil impact by forest operations and the investigation of
soil physical properties and rooting densities, these
parameters were likely to differ between the four

recovery stages. The rooting intensities below wheel
tracks for the recovery stages represented in the study
are shown in Fig.

4

.

The most recent skidding trail at Weil im Schoenbuch

showed throughout the whole depth of 60 cm lower fine
root densities than the control plot (Fig.

4

). The statis-

tical test routine, which considered the spatial depen-
dency

between

neighbouring

grid

cells,

indicated

significant differences between wheel tracks and control,
in the layers down to a depth of 46 cm. Compared to the
control plot, root density was still reduced below this
depth, but the test did not reach significance level.

At the study site Wolfegg with the 14-year-old skid-

ding trail, a comparable reduction of root density was
found with the exception of the uppermost 8 cm of the
mineral soil where the root densities reached the range of
the control. Still in the depth of 66 cm, root density
showed a significant lower value in the wheel track sit-
uation. At the 18-year-old skidding trail of Ettenheim,
significant differences in fine root densities between
wheel tracks and control occurred at soil depths below
18 cm. Towards the soil surface the rooting intensity
was in the same range as at the control plot. The
24-year-old skidding trail of Ettenheim did only display
reduced fine root densities at soil depths between 54 and

Fig. 2 Depth profiles of the
relative apparent gas diffusion
coefficient (D

s

/D

0

) at a water

suction of 60 hPa for wheel
tracks (left) and control plots
(right) at the study sites Weil im
Schoenbuch (6 years), Wolfegg
(14 years), Ettenheim
(18 years), and Ettenheim
(24 years). In Ettenheim same
control was used for both
treatments. At the right margin
significance levels of the paired,
non-parametric Wilcoxon test
for differences between wheel
tracks and control plots are
presented [***P<0.001,
**P<0.01, *P<0.05,
(*)P<0.1, n.s. not significant,
horizontal lines

represent the

single standard deviation]

Table 2 Mean values of D

s

/D

0

(60 hPa, five replications) for the skid trail strata at the depth of 10–14 cm

Site

Wheel track

Median strip

Margin zone

Control site

Weil im Schoenbuch (1996)

0.0067 (0.0057)a

0.0153 (0.0103)ab

0.0476 (0.0174)b

0.0355 (0.0323)ab

Wolfegg (1988)

0.00439 (0.0032)a

0.0078 (0.0017)ab

0.0179 (0.0081)ab

0.0240 (0.0120)b

Ettenheim (1984)

0.0286 (0.0163)a

0.00709 (0.0046)a

0.0082 (0.0063)a

0.0109 (0.0047)a

Ettenheim (1978)

0.00388 (0.0025)a

0.0099 (0.0195)ab

0.0206 (0.0127)b

0.0109 (0.0047)ab

The standard deviation is shown in brackets. Different characters indicate significant differences among the strata, tested by non-
parametric Dunn test at a level of P<0.05

134

66 cm (not significant). Above that depth, root densities
were comparable or even higher than at the control plot
(in the soil layer of 14–22 cm: significance of testing).

As a synoptic conclusion it can be stated that

recovery of rooting capacity of compacted soils starts at

the soil surface and obviously propagates vertically step
by step. With some precaution the time range for
recovery at the loamy sites included in the study can be
assessed as follows: 6–14 years after the machine impact,
hardly any effective natural recovery was visible. Below

Fig. 3 Pattern of fine root densities (fine roots <2 mm) at the
study site Weil. The 40 cm wide and 60 cm deep profile wall of the
control plot (right side) and the 4 m wide and 60 cm deep profile
wall of the transect across a skidding trail (left side). The number of

fine roots (n

·100 cm

2

) was counted in 4

·4 cm

2

grid cells. Bold

smoothed contour line

= threshold of the intensively rooted zone

(20 fine roots

· 100 cm

2

), fine smoothed contour line = threshold

of the extensively rooted zone (10 fine roots

· 100 cm

2

)

Fig. 4 Depth profiles of relative
fine root density (percentage of
the control plot) below wheel
tracks at the study sites Weil im
Schoenbuch (1996), Wolfegg
(1988), Ettenheim 2 (1984), and
Ettenheim 1 (1978). Vertical
reference line
(100%) = control plot. At the
right side

results of the non-

parametric Wilcoxon test for
the difference between the fine
root densities below the wheel
track and the control plot at the
corresponding soil depth are
shown [***P<0.001,
**P<0.01, *P<0.05,
(*)P<0.1, n.s. not significant]

135

the wheel tracks the gas diffusion coefficients were
reduced to the same extent but rooting intensity begins
to recover at the uppermost 8 cm layer. After 18 years
recovery reached a depth of 18 cm, and after 24 years
symptoms of recovery could be detected throughout the
rooting zone. At the site with the longest period of
recovery the variability of fine root densities was still
extremely high. This indicates that even 24 years after
machine impact a transient situation was achieved rather
than a complete recovery.

Discussion

Several reasons led to a limitation of the material which
could be used for the identification of recovery dynam-
ics. First of all, the time span during which comparable
machine types were used reaches back no longer than
40 years. It was in the 1960s when first specialized log-
ging machines like Welte or Timberjack were introduced
into practice in Middle Europe (SkogForsk

1997

). Sec-

ondly, the documentation of the side conditions during
the logging campaign contained more and more uncer-
tainties when the time delay between logging and mea-
suring of the ecological damage was longer. In the
following paragraphs the results of recovery processes
will be discussed considering these uncertainties as well
as relevant literature.

Relation of diffusion coefficients between skidding trail
strata

The relation between relative gas diffusion coefficients
(D

s

/D

0

) of the wheel tracks, median strips, and margin

zones was plausible and showed different intensities of
machine impact. At the wheel tracks with the most
intensive impact the reduction of D

s

/D

0

was highest,

measuring 1–2 orders of magnitude lower values com-
pared to the control sites. These findings correspond
with results from comparable skidding experiments
(Schack-Kirchner

1994

). However, we did not expect to

discover that the D

s

/D

0

values at the median strip would

be in the same range and statistically not distinguishable
from the wheel tracks since no tracking had taken place
in that area. This result can be explained by the effect of
a twig-mat, which enlarged the impact zone from the
wheel tracks to the sides. It can be suggested that the
same effect caused superficially reduced D

s

/D

0

values in

the uppermost soil layer at the margin zones. In the
deeper soil layers no differences between control area
and skid trails could be discovered, with the exception of
the margin zone at ‘‘Weil im Scho¨nbuch’’, which is to
judge as a site-dependent outlier.

Spatial distribution of fine roots as monitor of aeration

Fine root distribution can be interpreted as a very sen-
sitive indicator for the soil aeration status because roots

are the most oxygen-demanding tissue of plants (Qi et al.

1994

).

Schack-Kirchner et al. (

1993

) stated, based on an

oxygen distribution model and measured diffusion
coefficients for soil aeration below skid trails at a loamy
study site, that zones with deficiencies in oxygen supply
propagate to the margin zones with increasing depth.
The similarity of the lower boundary of the intensively
and the extensively rooted soil zone at Weil im
Schoenbuch (Fig.

3

) with the shape of the isograms of

oxygen supply in the two-dimensional oxygen diffusion
model suggests that indeed the fine root densities can be
interpreted as a ‘‘biological monitor’’ of soil aeration.
Apart from that spatial effect, which is the result of the
isotropic tendency of gas transport processes by diffu-
sion, it can be assumed that the forces of machine load
were partly propagated to the sides as well. The har-
vester generated a solid brushwood-mat, which might be
responsible for the diversion of a part of the machine
impact to the neighbourhood of the wheel tracks for
some decimetre. This could explain the low D

s

/D

0

values

and low rooting densities at the median strip (Figs.

1

,

3

).

Thus it can be stated that the function of the soil to serve
as rooting zone was disturbed more or less at the whole
area of the skidding trail. This applies even more
explicitly if we suppose that the compacted soil area
adds up when subsequent harvesting campaigns had
been performed. The finding that the intensively rooted
zone was not much deeper than 10 cm throughout the
whole skidding trail whereas it was at least 40 cm deep
at areas with no machine impact has to be judged as a
nearly total loss of rooting capacity due to insufficient
availability of essential resources like oxygen, water, and
nutrients.

How plausible is the interpretation of the results
as effects of recovery?

The results of D

s

/D

0

measurements and fine root inten-

sities at skidding trails differing in age show an interre-
lationship which may be interpreted as a first effect of
soil structure recovery beginning at about 15–25 years
after the machine impact. Uncertainties in this inter-
pretation arise from the question if the control areas
were really free from machine impacts. Especially at the
oldest skidding trails of the site Ettenheim, it cannot be
excluded that tracking also occurred besides the trails on
the area of the strip-shaped clear-cut. The noticeable low
gas diffusion coefficient at the control plot in a depth of
10–14 cm of this site can be judged as a remnant of a
former machine impact outside the regular skid trail
system. But apart from this uncertainty concerning the
older recovery stages, the pattern of rooting intensities,
gas diffusion coefficients, their depth gradients, and
relations among the different recovery stages are con-
sistent. They support the hypothesis of a recovery pro-
cess which begins at the soil surface and after decades
propagates gradually to the deeper soil layers.

136

Conclusions

A lot of research approaches (Hildebrand

1983

;

Scha¨ffer et al.

1991

; Schack-Kirchner et al.

1993

;

Schack-Kirchner

1994

; Gaertig et al.

1999

) as well as

the results of this study demonstrated that harvesting
and logging with heavy forest machines causes dra-
matically reduced aeration in forest soils from the first
impact on. In this study it could be shown that this
applies for the whole width of the skidding trail and
even a few decimetre to the sides. Based upon the long
chain of results concerning the physical soil damage
and the restriction of rooting intensity of forest trees in
compacted soils (Bredberg and Wa¨sterlund

1983

; Hil-

debrand

1986

; Hildebrand et al.

2000

; Kremer

1998

;

Gaertig et al.

1999

,

2000

) in the state of Baden-

Wuerttemberg a soil preservation guideline was estab-
lished in forest practice (MLR

2003

) which restricts

machine traffic to regular skidding trail systems. The
distances of the skid trails range from 20 to 40 m.
Between these more or less parallel tracks any wheeling
of heavy machines is forbidden. These prescriptions are
justified by the fact of the high time persistence of soil
damages caused by soil compaction. They can be
judged as the central points of a consequent soil pres-
ervation strategy in forests.

The demand on dense skid trail systems in young

stands is rather high compared to skid trail distances
needed for thinning operations in old growth. The first
results concerning recovery processes, especially the
finding of this study that an effective recovery of soil
compaction is possible within several decades (30 years
and more), would allow denser patterns of skidding
trails (e.g. 20 m distance) in young stands (up to
40 years) compared to those in older stands (e.g. 40 m
distance). If the preliminary results of this study can be
substantiated by further research, a temporary usage of
skid trails should not absolutely be excluded. During
the conversion period from mono-cultured coniferous
stands to highly structured mixed broadleaved stands
temporary usage of skid trails would give more free-
dom to realize silvicultural aims. In this case it would
be necessary that a clearly defined period of machine
impact is followed by a long phase of biological
recovery.

Helal (

1990

) stated that there is a large demand for

new research concerning recovery processes in com-
pacted soil zones to be able to judge the duration of
the ecological damage caused by soil compaction.
Because a lot of soil processes are involved in the
recovery of soil structure and soil aeration, a site-spe-
cific prediction of the duration which is necessary for
the restoration of soil functions is not possible. With
the results from studies like this the knowledge about
the potentialities and boundaries of soil recovery may
be improved.

Acknowledgements This study was funded by the Bundesministe-
rium fu¨r Bildung und Forschung (BMBF No. 0339760).

References

Bo¨hm W (1979) Methods of studying root systems. Ecological

studies 33. Springer, Berlin Heidelberg New York

Bredberg C-J, Wa¨sterlund I (1983) Wurzel- u. Bodenscha¨den durch

Fahrzeuge. Forstw Cbl 102:86–98

Danielson RE, Sutherland PL (1986) Porosity. Methods of soil

analysis. Part 1. Physical and mineralogical methods, 2nd edn.
Agronomy monograph, vol 9, pp 443–450

Frede HG (1986) Der Gasaustausch des Bodens. Go¨ttinger

Bodenkundl Ber 87, 130 pp

Gaertig T, v Wilpert K, Schack-Kirchner H (1999) Bodenbelu¨ftung

als Steuergro¨ße des Feinwurzelwachstums in Eichenbesta¨nden.
Allg Forst- u J-Ztg 170:81–87

Gaertig T, Hildebrand EE, Scha¨ffer J, v Wilpert K (2000) Wirkung

mechanischer Bodenlockerung auf Bodenbelu¨ftung und Dur-
chwurzelung. AFZ/DerWald 55:1124–1126

Hartge KH, Horn R (1992) Die physikalische Untersuchung von

Bo¨den. Spektrum akademischer Verlag Thieme, Stuttgart

Helal M (1990) Bodengefu¨ge, Wurzelentwicklung und Wurzel-

funktionen. Z Pflanzenerna¨hrung Bodenkunde 154:403–407

Hildebrand EE (1983) Der Einfluss der Bodenverdichtung auf

die Bodenfunktionen im forstlichen Standort. Forstw Cbl
102:111–125

Hildebrand EE (1986) Der Einfluss der Strukturscha¨digung von

Feinlehmen auf die Wurzelentwicklung zweier Fichtenklone.
Mitt Verein Forstl Standortskunde Forstpflanzenzu¨chtung
32:50–56

Hildebrand EE (1987) Die Struktur von Waldbo¨den—ein gefa¨hr-

detes Fließgleichgewicht. Allg Forst Zeitschr 42:424–426

Hildebrand EE (1994) Forest soils—medium for root growth. In:

Proceedings from ECE-FAO-ILO interactive workshop and
seminar FORSITRISK soil, tree, machine interactions, Felda-
fing, Germany, 4–8 July 1994

Hildebrand EE, Puls C, Gaertig T, Schack-Kirchner H (2000)

Fla¨chige Bodenverformung durch Befahren. AFZ/DerWald
55:683–686

Horn R (1988) Compressability of arable land. Catena Suppl

11:53–71

Kremer J (1998) Befahrungsbedingte Strukturvera¨nderungen von

Waldbo¨den und ihre Auswirkungen auf das Wachstum von
Fichten, Kiefern und Buchen auf ausgewa¨hlten Standorten.
GCA-Verlag, Herdecke

Littell RC, Milliken GA, Stroup W, Wolfinger RD (1996) SAS

system for mixed models. SAS Institute Inc., Cary

Loeffler H (1985) Bodenscha¨den bei der Holzernte—Bedeutung

und Erfassung. Forst Holzwirt 40:379–383

McCulloch CE, Searle SR (2001) Generalized, linear, and mixed

models. Wiley, New York

MLR (Ministerium fu¨r La¨ndlichen Raum Baden-Wu¨rttemberg)

(2003) Richtlinie der Landesforstverwaltung Baden-Wu¨rttem-
berg zur Feinerschließung von Wald-besta¨nden

Murach D, Isle L, Klaproth F, Parth A, Wiedemann H (1993)

Rhizotron-Experimente zur Wurzelverteilung der Fichte. For-
starchiv 64:191–194

Pfeil C (2003) Oral information about historical machine configu-

rations. Department of Forest Utilization, FVA Baden-
Wuerttemberg

Qi J, Marshall JD, Mattson KG (1994) High soil carbon dioxide

concentrations inhibit root respiration of Douglas fir. New
Phytol 128:435–442

Schack-Kirchner H (1994) Struktur und Gashaushalt von Wald-

bo¨den, vol 112. Berichte Forschungszentrum Waldo¨kosysteme
Univ. Go¨ttingen, Reihe A

137

Schack-Kirchner H, Hildebrand EE, v Wilpert K (1993) Bode-

nsauerstoffgehalt unter Fahrspuren, Einsatz eines Simulations-
modells. Allg Forst Zeitschr 48:118–121

Scha¨ffer J, Hildebrand EE, Mahler G (1991) Bodenverformung

beim Befahren. Allg Forst Zeitschr 46:550–554

SkogForsk (The Forestry Research Institute of Sweden) (1997)

Forestry in Sweden, pp 28–38

Wackernagel H (2003) Multivariate geostatistics. Springer, Berlin

Heidelberg New York

Zar JH (1999) Biostatistical analysis, 4th edn. Prentice-Hall, Upper

Saddle River

138