1. Landscape changes in agrarian landscapes in the 1990s:
the interaction between farmers and the farmed landscape.
A case study from Jutland, Denmark
Lone Søderkvist Kristensena,*, Claudine Thenailb
, Søren Pilgaard Kristensenc
a
Danish Centre for Forest, Landscape and Planning, The Royal Veterinary and Agricultural University, Rolighedsvej 23, DK-1958 Frederiksberg C, Denmark
b
INRA-SAD armorique, 65 Rue de Saint-Brieuc, F-35042 Rennes Cedex, France
c
Institute of Geography, University of Copenhagen, Østervoldgade 10, DK-1350 Copenhagen K, Denmark
Received 29 August 2002; revised 3 March 2004; accepted 3 March 2004
Abstract
Recent landscape changes in a farmed landscape are analysed and related to farm and farmer characteristics. It is assumed that farm and
farmer characteristics serve as mediators of large scale or macro driving forces of change—in the present case, a changing farming context
including demands for a more environmentally friendly farming practise and a reduced output. The results are based on multivariate analyses
of data collected from structured interviews of 160 farmers in a case study area, in central Jutland, measuring 5000 ha. The analysis shows
that farmers are highly involved in landscape changes. The investigated landscape changes include creation and removal of landscape
elements as well as certain management changes. The most common activity was creation of elements: hedgerows, small woodlands and
conversion of rotational arable land to permanent grassland, whereas removal of elements, mainly hedgerows and semi-natural grasslands,
were seen less frequently. Management changes like abandonment of permanent grassland were widespread. The results indicate a general
extensification of the land use and the authors interpret the results partly as an indication of a change from productivism to a more
multifunctional agricultural regime. The observed landscape changes at the farm level show a low, but structured relationship with the
current farm and farmer characteristics, meaning that landscape changes were undertaken by various farmers and on various farms. On a
general level, however, the age of the farmer and the duration of farm ownership seem to have a major influence on the landscape changes.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Agricultural landscapes; Landscape changes; Driving forces; Factors influencing behaviour; Productivism; Post-productivism; Multiple
correspondence analysis
1. Introduction
It is widely agreed that agriculture in Western Europe
since the mid 1980s has undergone a complex restructuring.
The main driving forces of this restructuring are linked to
the reform of Common Agricultural Policy, the WTO
negotiations, and the increasing environmental regulation of
agriculture in the EU (Bowler and Ilbery, 1999). These
initiatives are well known and imply that agriculture must:
(1) produce within a context of an increasingly competitive
international market, (2) provide environmental goods and
(3) reduce farm output (Ilbery et al., 1997).
The new farming context has been conceptualised by a
number of researchers including Bowler, 1992, Ilbery and
Bowler, 1998, Marsden et al., 1993, Marsden, 1995,
Marsden, 1998 and Shucksmith, 1993 who suggest that
agriculture is in a transition from ‘productivism’ to ‘post-
productivism’. According to Ilbery and Bowler (1998) this
transition implies that agriculture, on a general level, will
orient away from intensification, specialisation and concen-
tration, which are characteristics of the productivist farming
period, towards extensification, diversification and disper-
sal, indicators of the post-productivist farming period. The
authors question a complete reorientation of the three
polarities; rather, they envision a co-existence of both
productivist and post-productivist farming systems.
Recently Wilson (2001) has suggested that “the notion of
‘multifunctional agricultural regime’ better (than the notion
of post-productivism) encapsulates the diversity, nonlinear-
ity and spatial heterogeneity that can currently be observed
0301-4797/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2004.03.003
Journal of Environmental Management 71 (2004) 231–244
www.elsevier.com/locate/jenvman
* Corresponding author. Tel.: þ45-3528-2213; fax: þ45-3528-2205.
E-mail address: lokr@kvl.dk (L.S. Kristensen).

2. in modern agriculture and rural society”. With this notion,
the author stresses that the new farming context is not linked
to one specific goal, which all actors try to reach, but
includes a variety of goals, actions, and thoughts linked to
both productivism and post-productivism.
The consequences of productivism in terms of landscape
impacts are well known and include the intensification of
land use and removal and abandonment of landscape
elements (Adams, 1996; Agger and Brandt, 1988; Baldock
et al., 1996; Barr et al., 1993; Baudry, 1991; Ihse, 1995).
Consequences of post-productivism seem, however, to be
more ambiguous. On one hand it has been seen as offering
new opportunities for nature development and nature
enhancement (Ilbery, 1990; Gasson and Potter, 1988; Potter
et al., 1991), but on the other hand it has been pointed out that
extensification and diversification do not necessarily lead to
an improvement of the values of landscape (Wilson, 2001).
The more unknown consequences of the post-producti-
vism and the aforementioned ambiguity in development of
the contemporary agricultural and resulting diversity in the
pattern of landscape changes call for new research into
landscape development, not only on a general level, but also
in a local context allowing for an examination of the
multifaceted actions of farmers and the impact of these
actions on the landscape. Recent landscape research in
Western Europe has mostly focused on agri-environmental
policies, their possibilities in providing environmental
benefits and scheme up-take (Wilson, 1996, 1997; Rønnin-
gen, 1993, 1999; Buller et al., 2000). However, fewer studies
have focused on landscape development in general in modern
farmed landscapes in the 1990s (Battershill and Gilg, 1996;
Brandt et al., 2001; Kristensen et al., 2001; Primdahl, 1999).
On this background, the purpose of this paper is two-fold:
(1) to investigate recent landscape changes in a farmed
landscape in order to examine the landscape consequence of
the multifunctional agricultural regime recently developed
and (2) to explore the relationship between characteristics of
farms and farmers and patterns of landscape change in order
to obtain an understanding of how the diversity of these
farm level characteristics influence landscape changes.
Landscape changes include in the present study creation
and removal of landscape elements such as hedgerows and
ponds and conversion of arable land to permanent grassland.
However, changes in management of landscape elements,
e.g. abandonment of permanent grassland, are also included.
These types of changes are often referred to as land-
use/land-cover changes (Lambin et al., 2000), but will in the
following text just be named ‘landscape changes’.
Literature on the influence of farm level characteristics on
the farmers’ decision making often emphasises length of
residency, age of the farmer, dependency of income from
agriculture (in the following called farmer occupation), farm
size(FS),andamount ofunusedareasonthe farmasimportant
factors(Adamsetal.,1994;Brandtetal.,1999;McDowelland
Sparks, 1989; Potter, 1994; Potter and Lobley, 1992, 1996;
Wilson, 1992, 1997). In addition Battershill and Gilg (1996)
report that farmers’ behaviour seem to be influenced by farm
production types, e.g. mixed dairy farming, mixed stock or
simple dairy farming. Similar farm level characteristics have
been investigated in the present study.
The patterns of landscape change and the relationship
between these patterns and the farm level characteristics
have been analysed by the use of multivariate analysis
techniques. Experiences with use of these techniques are
discussed in the end of the paper.
2. Presentation of the study area
The study area, Sønder Omme parish (SOP) located in
central Jutland within the West Jutland natural-geographical
region (Fig. 1), is an old moraine landscape dissected by
glacio-fluvial plains and valleys and dominated by sandy
soils (podsols) (Jacobsen, 1975). The total size of the parish
is approximately 8500 ha, of which about 5000 ha is
agricultural land; the remainder is predominately forest
(spruce). An extensive network of hedgerows exists
(shelterbelts), originally planted as one row of mainly
white spruce (Picea glauca). Due to fungus-related
problems, the white spruce hedgerows are increasingly
being replaced by deciduous hedgerows, mostly in three
rows or broader. The hedgerows have been planted mainly
in order to protect against wind erosion. The hedgerow
density as well as the area of forest has increased since the
1900. In contrast, the number and area of semi-natural
landscape elements, such as meadows and heathlands, have
been reduced due to ploughing and abandonment (Working
group SCOPE, 1997).
Fig. 1. The location of Sønder Omme parish in central Jutland, Denmark.
L.S Kristensen et al. / Journal of Environmental Management 71 (2004) 231–244232

3. The farms of SOP were in 1995 partly husbandry farms
(mainly dairy) and partly crop farms (50%). Compared to
national averages, the animal density was low (65 cattle/ha
compared to 77 cattle/ha and 72 pigs/ha compared to 399
pigs/ha). Sixty percent of the total agricultural area was
arable land; the remainder was permanent grassland, set-
aside land, woodlands, Christmas tree plantations and
uncultivated areas (Fig. 2). The latter includes heathlands,
bogs, ponds, hedgerows, etc. as well as gardens, build-up
areas and roads, with semi-natural areas making up the
largest proportion. The proportion of uncultivated areas and
woods (including Christmas tree plantations) was high (20%)
compared to national average (10%) (Danmarks Statistik,
1996). This may be ascribed to the poor soil condition of the
area, which makes it less suitable for arable crop-production
than for example the more fertile areas in eastern Denmark.
The average property size was 31 ha (defined as the total
amount of land owned by the farmer), and the farm average
size was 36 ha (the total amount of land owned by the farmer
plus the amount of land rented or minus the land leased out),
indicating that a significant amount of land was under
leasehold; these were mainly short-term contracts.
Thirty percent of the farmers were full-time farmers,
50% were part-time/hobby farmers or other persons with no
or little relation to agriculture and 20% were pensioners.
The farmer occupation was defined on the basis of the
farmers’ personal income from farming (not the household
income): full-time farmers having more than 80% of their
income from farming, part-time farmers 50–80% and
hobby farmers having less than 50% of their income from
farming. All farmers receiving a pension (early retirement
or real retirement) have been classified as pensioners
regardless of the income derived from farming.
3. Methods
3.1. Data sources
All farmers in the parish owning at least 2 ha of
agricultural land were contacted for a structured personal
interview. Responses were obtained from 95% of all eligible
farmers. A total of 160 farmers owning 168 farms were
interviewed. The survey included questions about the socio-
economic SE situation of the individual farmer, details
about the farm production system and land use, along with
questions concerning landscape changes during 1990–1995
(a five year period). All landscape changes were mapped in
order to increase the reliability of the information.
If data were missing for the farm/farmer characteristics
or the landscape change variables, the observation was
excluded. This filtering of observations reduced the dataset
to a total of 138 farmers owning 4343 ha. Farmers excluded
from the analysis were mainly older farmers leasing out
nearly all their land, because production and land use
information were not recorded for farmers cultivating less
than 2 ha of land.
The landscape changes investigated in the survey include
changes of both patch and linear elements, e.g. conversion of
arable land to permanent grassland and the reverse, and
removal and planting of hedgerows. The full range of
landscape changes included in the survey is shown in Table 1.
In the analysis of the relationship between landscape
changes and farmer/farm characteristics, the latter have
been separated into socio-economic features and farm
production/land-use features (Table 1) in order to test the
influence of these features individually. In addition, a farm
size variable (with six classes) has been included, to
describe the farm structure.
3.2. Multivariate analyses
Multiple Correspondence Analyses (MCA) and Hierarch-
ical Cluster Analyses (HCA) were used to produce the
typologies of farm and farmer characteristics and landscape
changes as well as to analyse their relationship. The typology
of landscape changes was produced by a HCA following a
MCA on the landscape change data. From a similar procedure
two distinct typologies, one on the production characteristics
and one on the SE characteristics, were produced. Finally, the
relationship between the farm and farmer characteristics and
the landscape changes was investigated by the use of the
produced factorial planes and typologies.
MCA allow the use of both quantitative and qualitative
variables and were in the present study used in order to
investigate gradients of relationships between farm/farmer
characteristics and the landscape changes (Legendre and
Legendre, 1998). The type of HCA is an agglomerative
clustering, i.e. a procedure that successively groups the
closest objects into clusters, which then are grouped into
larger clusters of higher rank (Legendre and Legendre,
1998). The descriptors of the objects used in the clustering
were their factorial co-ordinates on the plane of the MCA.
The metric used in the MCA is based on the x2
: This was
also the metric used in the following HCA. At each step the
procedure of HCA tends to minimise the inner-cluster
variance and jointly maximises the between-cluster vari-
ance (Lebart et al., 1995). The algorithm is based on
Fig. 2. The land use in Sønder Omme in 1995, in percentages of the total
agricultural area.
L.S Kristensen et al. / Journal of Environmental Management 71 (2004) 231–244 233

4. the principle of reciprocal pairs (McQuitty, 1966, cited by
Lebart et al., 1995), which aggregated the closest objects at
the same time at each step. All analyses were performed
with Addad software (Lebeaux, 1985).
The investigation of the relationships between the
different sets of variables was made by adding the typology
of one set of variables (e.g. landscape changes), as
supplementary variables on the factorial plane of a second
set of variables (e.g. the gradient of SE characteristics).
Supplementary variables do not participate to the building
of the factorial plane. The location of these supplementary
variables along the factorial axes expresses their linkage
with the pattern of active variables displayed by these
factorial axes (Lebart et al., 1995).
4. Results
4.1. Trend in landscape changes and combination
of landscape changes at farm level
In total, 95 farmers corresponding to 69% of all
interviewed farmers were involved in one or more landscape
changes. The most common activities were hedgerow
removal/planting, conversion of arable land to permanent
grassland, planting of woods and small woodlands, and
abandonment of permanent grassland (Fig. 3).
In terms of size, the largest single type of patch change
was planting of Christmas tree plantations (106 ha),
followed by establishment of permanent grassland (80 ha),
planting of woods and small woodlands (60 ha) and
abandonment of permanent grassland (57 ha). Also arable
land was abandoned to a certain degree (26 ha). Except for
the abandonment of permanent grassland, the above-
mentioned changes (272 ha) imply a conversion of arable
land to permanent land uses (Christmas tree plantations,
woods, small woodlands, permanent grasslands and areas
for ‘free‘ succession). In contrast, only 27 ha of permanent
grasslands were converted into arable land. The remaining
patch element changes concern changes within the perma-
nent land uses. These changes included re-grassing of
abandoned grassland (permanent grassland which had been
out of agricultural use for a longer period; 42 ha) and
abandonment of Christmas tree plantations after they have
been harvested. The total area of 8.4% (365 ha) has changed
to a different land-use/land-cover in the investigated period.
In addition, 20.8 km of hedgerows were planted, 14.3 km
of hedgerows were removed, and 13 ponds were established
or cleaned. No ponds were removed. More than half of the
hedgerows were removed in order to replant the hedgerows
in the area.
The results of the MCA performed on the landscape
changes (the typology building) show that the first axis
corresponds to an eigenvalue of 0.20, which accounts for
Table 1
Variables of landscape changes and farm level characteristics: socio-economic variables, production and land-use variables and farm size, used in the analyses
Variables Description of variables: classes and categories
Farmer characteristics: socio-economic variables (SE)
Age of the farmer, year #40, ]40–50], ]50–60], .60
Duration of farm ownership, year #5, ]5–10], ]10–20], .20
Owner occupation Full-time, part-time, hobby, pensioners and others
Number of people living on the farm 1, 2, .2
Place of growing up City or countryside
Farm characteristics: production and land-use variables (PLU)
Arable land, percentage Eight classes
Land use, percentage Fourteen land use types with eight classes
Animal production, number Nine types of animal and 3–5 classes depending
on the type of animal
Farm structure: farm size (FS)
The farm size is based on the total amount of land owned
by the farmer plus the amount of land rented or minus
the land leased out, ha
]0–10], ]10–20], ]20–30], ]30–50], ]50–100], .100
Landscape change variables (PC)
Conversion of arable land to permanent grassland, ha 0, ]0–5], .5
Cultivation of permanent grassland (ploughing up), ha 0, ]0–5], .5
Re-grassing of abandoned permanent grassland, ha 0, ]0–5], .5
Planting and removal of forest and small woodlands, ha 0, ]0–5], .5
Planting and removal of hedgerows, m 0, ]0–500], .500
Type of hedgerows, number of rows 1, [2–3], [4–6], .6
Creation and removal of Christmas tree plantations, ha 0, ]0–5], .5
Establishment, cleaning and removal of ponds Yes or no, established or cleaned
Abandonment of land (arable, permanent grassland
and Christmas tree plantings), ha
0, ]0–5], .5
L.S Kristensen et al. / Journal of Environmental Management 71 (2004) 231–244234

5. 10.40% of the total inertia, and the second axis corresponds
to an eigenvalue of 0.16, which accounts for 8.30% of the
total inertia (cumulated value: 18.7%). These values are low
compared to values obtained in the case of farm and farmer
characteristics (see Section 4.2); however, they still indicate
that a certain structure of landscape changes at the farm
level can be identified.
Six landscape change types appeared from the combined
analyses of the MCA and the HCA (Table 2). The landscape
change types PC3 and PC4 are the most common; they
contain 35 and 26 farmers, respectively.
The PC1 and PC3 are characterised by planting and
removal of hedgerows, which appear to be associated
processes. PC1 corresponds to farms with a high rate of
hedgerow planting (600–900 m) and removal (100–
600 m), whereas PC3 corresponds to farms with a medium
rate of hedgerow planting (100–300 m) and removal (50–
200 m). Both types include several other changes: conver-
sion of arable land to permanent grassland, planting of
woods and small woodlands and creation of ponds;
however, PC1 alone includes re-grazing of abandoned
grassland and PC3 alone includes the ploughing up of
permanent grassland. Among farms belonging to type PC1
and PC3, respectively, 67 and 43% have undertaken more
than one type of landscape change (when planting and
removal of hedgerows are seen as one activity).
PC4 represents farms with diverse patch changes, notably
conversion of arable to permanent grassland (about half) and
planting of woods and small woodlands. The majority of
farmers in PC4 have undertaken only one landscape change.
All farmers in the groups PC5 and PC6 have abandoned
land. PC5 represents farms with 1–8 ha of abandoned land,
which includes both permanent grassland and arable land.
Forty-three percent have undertaken more than one
change. The type PC6 represents farms with land abandon-
ment (2–5 ha), mainly of grassland, and hedgerow planting
(150–300 m) and removal (0–200 m). All farmers in this
group have undertaken more than one change.
Finally, PC2 contains a single farmer, which has
converted arable land to permanent grassland and made
Christmas tree plantations. This farmer has specialised in
nursery plants and Christmas trees and has undertaken the
majority of Christmas tree planting in the area.
4.2. Farm typologies
A typology of production/land use variables (PLU) and
socio-economic variables (SE) were made from their
co-ordinates in the MCA ‘PLU’ and ‘SE’, respectively.
For the set of variables describing the production and land
use system (PLU),1
the first axis segregates farms with less
than 40% of arable land and no animals on the positive side,
from mixed crop/livestock farms with more than 40% of
arable land on the negative side. For the set of variables
describing the SE characteristics of farms,2
the first axis
expresses a gradient of increasing age of the farmers,
Fig. 3. The number of farmers in Sønder Omme involved in landscape changes. The size of the different landscape changes is given in the left column. Changes
in ponds are given in numbers.
1
MCA ‘PLU’: the first axis corresponds to an eigenvalue of 0.69, which
accounts for 14.60% of the total inertia, and the second axis corresponds to
an eigenvalue of 0.53, which accounts for 11.25% of the total inertia
(cumulated value: 25.85%).
2
MCA ‘SE’: the first axis corresponds to an eigenvalue of 0.50,
which accounts for 20.2% of the total inertia, and the second axis
corresponds to an eigenvalue of 0.33, which accounts for 13.5% of the
total inertia (cumulated value: 33.7%).
L.S Kristensen et al. / Journal of Environmental Management 71 (2004) 231–244 235

6. Table 2
The six ‘landscape change types’ based on the combination of landscape changes
Landscape changes PC1 (12 farmers) PC2 (1 farmers) PC3 (35 farmers) PC4 (26 farmers) PC5 (14 farmers) PC6 (7 farmers)
Arable land to permanent
grassland
0 ha (10), .5 ha (2) .5 ha (1) 0 ha (30), 1–5 ha (5) 0 ha (14), 1–5 ha (12) 0 ha (13), 1–5 ha (1) 0 ha (7)
Permanent grassland to
arable land
0 ha (12) 0 ha (1) 0 ha (32), 1–5 ha (2), 0 ha (21), 1–5 ha (4), NI (1) 0 ha (14) 0 ha (7)
Started re-grassing 0 ha (9), 1–5 ha (1), .5 ha (2) 0–5 ha (1) 0 ha (35) 0 ha (24), 1–5 ha (1),
.5 ha (1)
0 ha (14) 0 ha (6), 1–5 ha (1)
New woods and small
woodlands
0 ha (9), 1–5 ha (2), NI (1) 0 ha (1) 0 ha (29), 1–5 ha (5) 0 ha (18), 1–5 ha (5),
.5 ha (2), NI (1)
0 ha (11), 1–5 ha (2),
NI (1)
0 ha (5), 1–5 ha (2)
New Christmas trees 0 ha (11), NI (1) .5 ha (1) 0 ha (30), 1–5 ha (4) 0 ha (24), .5 ha (2) 0 ha (14) 0 ha (7)
New hedgerows .500 m (10), 1–500 m (2) NI (1) 1–500 m (27), 0 m (6) 1–500 m (1), 0 m (25) 1–500 m (1), 0 m (13) 1–500 m (6), .
500 m (1)
Type of hedgerow planted 3 Rows (9), diverse rows (3) NI (1) 1–2 rows (2), 3 rows (19),
.3 rows (4), diverse (2)
.3 rows (1), none (25) .3 rows (1), none (13), 1–2 rows (2),
3 row (4), none (1)
Removed woods 0 ha (12) 0 ha (1) 0 ha (34), 1–5 ha (1) 0 ha (26) 0 ha (14) 0 ha (7)
Removed hedgerows 1–500 m (5), .500 m (4), 0 m (3) 0 m (1) 1–500 m (26), 0 m (9) 0 m (26) 1–500 m (2), 0 m (12), 1–500 m (3),
.500 m (1), 0 m (3)
Removed Christmas trees 0 ha (11), 1–5 ha (1) 0 ha (1) 0 ha (35) 0 ha (26) 0 ha (13), 1–5 ha (1) 0 ha (7)
Established and cleaned
ponds
Established (2), none (10) Established
(1)
Cleaned (2), established (1),
none (32)
Established (4),
cleaned (3),
none (19)
None (14) None (7)
Removed ponds None None None None None None
Abandonment of use None (12) None (1) No (35) Ab. of, arable land, (1),
none (25)
Ab. of perm.
grassland (9), Ab.
of arable land (5)
Ab. of perm.
grassland (6), Ab.
of arable land (1)
Amount of abandoned
land
0 ha (12) 0 ha (1) 0 ha (35/35) 0 ha (25) 1–5 ha (9),
.5 ha (3), NI (2)
1–5 ha (6), .5 ha (1)
Descriptiona
Hedgerows planted: 600–900 m Arable ! permanent
grassland
Hedgerows planted: 100–300 m Diverse patch changes Abandonment:
1–8 ha
Abandonment: 2–5 ha
Hedgerows removed: 100–600 m New Christmas trees Hedgerows removed: 50–200 m Hedgerows planted:
150–300 m
Hedgerows removed:
0–200 m
The 43 farms without any changes (PC0) are not mentioned in the table. Figures in bracket indicate the number of farmers involved in each landscape change. NI ¼ no information. The most characteristic of
the individual landscape types are indicated by a grey colour.
a
The description is supplemented by more detailed information on the amount of changes from the individual observations of the cluster types.
L.SKristensenetal./JournalofEnvironmentalManagement71(2004)231–244236