Management influence on environmental impacts in an apple

Agriculture, Ecosystems and Environment 114 (2006) 311–322
www.elsevier.com/locate/agee
Management influence on environmental impacts in an apple
production system on Swiss fruit farms: Combining life cycle
assessment with statistical risk assessment
Patrik Mouron a,*, Thomas Nemecek b, Roland W. Scholz a, Olaf Weber a
a
Swiss Federal Institute of Technology, Department of Environmental Sciences, Institute for Human-Environment Systems,
ETH Zentrum, CHN J76.1, CH-8092 Zurich, Switzerland
b
Agroscope FAL Reckenholz, Swiss Federal Research Station of Agroecology and Agriculture, CH-8046 Zurich, Switzerland
Received 21 September 2004; received in revised form 25 October 2005; accepted 22 November 2005
Available online 18 January 2006
Abstract
In this study we assessed to what extent management of apple-growing within a well defined farming system affects environmental
impacts. A four-year data set of 12 fruit farms from eastern Switzerland was analyzed using the life cycle assessment (LCA) method to
evaluate the variability of different environmental impacts. For the total of 445 annual data sets of apple orchards eight impact categories were
assessed. A principal component analysis (PCA) was performed to group the eight impact categories according to their correlation. A three
component solution turned out to be adequate. It indicated that the three impact categories energy use, aquatic ecotoxicity and aquatic
eutrophication were influenced independently of each other to a high degree. These three key impacts can be managed by keeping the inputs of
machinery, pesticides and fertilizers low. Production constraints were highly homogeneous within the sample. Because of this, we were able to
define the management influence on environmental impacts as the ratio of the maximum and minimum observed. On a per hectare basis, the
effect of management for energy use was factor 2, for aquatic ecotoxicity factor 4 and for aquatic eutrophication factor 1.1. In contrast, when
measured per receipts, the management influence was greater than per hectare, indicated by a range of factor 6 for each of the three key impact
categories. Further insight into the effect of management was attained by statistical risk assessment. A positive and significant correlation
between mean value (M) and the coefficient of variance (CV) indicated that the expected risk could be reduced by a low level of variability.
Such a M–CV correlation was found for the two key impact categories energy use and aquatic eutrophication if calculated per receipts. No M–
CV correlation was found for aquatic ecotoxicity. It was on the other hand observed that farms with low aquatic ecotoxicity also practiced low
energy use and low eutrophication on a per receipt basis. We conclude that the promotion of environmentally sound apple-growing is not only
a question of choosing one or the other farming system (e.g. organic versus integrated farming) but that an understanding of the system
specific management influence is crucial.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Farm management; Environmental management; Life cycle assessment (LCA); Environmental impact assessment; Statistical risk assessment;
Principal component analysis (PCA); Integrated apple-growing
1. Introduction
Farmers need information about the causes of environmental impacts in order to promote environmentally sound
agricultural production. Life cycle assessment (LCA) has
* Corresponding author. Tel.: +41 44 780 48 46; fax: +41 44 632 10 29.
E-mail address: patrik.mouron@env.ethz.ch (P. Mouron).
0167-8809/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.agee.2005.11.020
become an established tool to generate such information
(Audsley et al., 1997; Andersson, 2000; Haas et al., 2000;
FAL, 2002; Brentrup, 2003). Some agricultural LCA-studies
have compared different input intensity levels mainly for
fertilizers and pesticides by comparing farming systems
such as integrated versus organic farming. Stadig (1997),
Reganold et al. (2001) and Bertschinger et al. (2004) have
reported such comparisons of farming systems for apple-
312
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
growing. Farming systems for arable crops have been
assessed by Gaillard and Hausheer (1997), Nemecek et al.
(2001, 2002) and Bailey et al. (2003) and for grassland by
Haas et al. (2001). All those studies have addressed the
differences between the farming systems regarding environmental impacts. However, Rossier and Gaillard (2004)
found wide differences within both integrated and organic
farming systems applying LCA to 50 farms in Switzerland. This variability within the same type of farms raises
questions about the potential for improvement of
environmental impact management in individual farming
enterprises. To study the causes of the variability between
farms the performance of individual fields or orchards
should be addressed because in practice, as argued by
Keating and McCown (2001), the single field is the basic
managerial unit of a farm. In addition, Milà i Canals
(2003) pointed out the importance of site-dependency for
impact assessment, especially for apple orchard systems.
However, to our knowledge, no empirical study has
investigated the statistical distribution of environmental
impacts among individual fields of a homogeneous group
of farms over a reasonable span of years in order to
investigate the management influence on environmental
impacts. We report such a study here for the case of fruit
farming.
The objective of this paper is to investigate in which
domains the environmental performance of integrated
apple-growing in Switzerland can be improved at the farm
level. This is done by (i) using life cycle assessment
(LCA) to analyze eight impact categories, (ii) by applying
a principal component analysis (PCA) to reduce the
complexity of these impact categories, and (iii) by
analyzing the management influence based on a statistical
risk assessment. As a result it is possible to suggest how
much a successful farm manager could influence the level
and distribution of impacts among the orchards and how
far a control of the environmental impact variability is
feasible. Apart from this, knowledge about the correlation
between impact categories should allow a farm manager to
focus on particularly relevant impacts when faced with the
impact category complexity assessable by the LCA
methodology.
The farm data used in our study contains detailed
information about the variability of production inputs and
environmental impact categories of a sample with 12 Swiss
fruit farms including 445 annual data sets of apple orchards
over 4 years. All farms were participants in the environmental program of integrated farming.
The paper is structured as follows. The next section
introduces the LCA methodology and the framework used to
assess the management influence and describes the data. The
empirical results then follow. A comparison of environmental impacts of apple-growing with those of arable crops
and implications for improving the environmental management of integrated apple-growing are discussed in the two
final sections.
2. Methods and data
2.1. Methodology of life cycle assessment (LCA)
Life cycle assessment is a method to quantify the
environmental impacts of a product or process over its entire
lifespan, i.e. ‘‘from the cradle to the grave’’ or from the
extraction of the resources to the disposal of wastes. In this
study the Swiss agricultural life cycle assessment method
(SALCA) Version 1.31 was used. The life cycle inventories
of emissions and used resources were taken from the
SALCA database Version 031a according to Nemecek et al.
(2003). Direct field emissions of ammonia, nitrous oxide,
methane, phosphorus, nitrate and heavy metals were
calculated by models with situation dependent parameters
(Nemecek, 2003). The following eight environmental
categories were assessed, which characterize relevant
environmental impacts of apple-growing:
! Energy use, representing the non-renewable energy
resources including primary energy of direct forms used
in operations, i.e. diesel and electricity, as well as indirect
forms, e.g. pesticide production, according to Frischknecht et al. (1996). Fossil energy resources (coal, oil gas)
and uranium are included.
! Global warming potential (GWP) for 100 years was
computed according to Houghton et al. (1995). The main
emissions are carbon dioxide and nitrous oxide, stemming
from the use of energy resources and nitrogen fertilizers.
! Ozone formation potential was calculated according to
Heijungs et al. (1992) and Dinkel et al. (1996). The major
emissions are nitrogen oxides and volatile organic
compounds (VOC) from engine exhaust gases.
! Aquatic ecotoxicity potential and terrestrial ecotoxicity
potential were based on methods described by Jolliet and
Crettaz (1997) and Margni et al. (2002). Toxic impacts to
ecosystems are mainly caused by pesticides and heavy
metals.
! Aquatic eutrophication potential and terrestrial eutrophication potential were computed according to Heijungs
et al. (1992) and LCA Nordic (1995). Eutrophication is
the enrichment of the nutrients phosphorus and nitrogen
in sensitive aquatic and terrestrial ecosystems.
! Acidification potential was calculated according to
Heijungs et al. (1992). Acidification refers not to the
agricultural soil, but to sensitive ecosystems through
aerial emissions.
Human toxicity was not assessed due to methodical
difficulties. Ozone depletion is not considered to be relevant
for the studied system and was therefore not calculated.
Impacts on biodiversity and soil quality were not evaluated,
since the methodology is not yet developed.
An agricultural LCA was carried out with the system
boundary at the farm gate. Consequently the system
included all activities in the orchards as well as transport
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
of the apples to the farm and transport of materials from the
farm to the orchards. Not considered were the activities at
the wholesaler and retailer such as sorting, storing and
packaging of the fruits. The allocation of the inputs to the
orchards was clearly defined by the farmer’s diary. Post
harvest activities in the orchards that were linked to the next
harvest were included in the LCA for the next season. For
each orchard and each year a separate LCA was performed.
Impacts from the manufacturing of the following inputs
were taken into account in the LCA:
! Tractors and equipment, including their transport and
maintenance.
! Buildings, required for shelter of tractors, equipment and
materials.
! Energy carriers, which were diesel for machinery and
electricity for lighting the buildings.
! Pesticides.
! Mineral fertilizers including their transport; no liquid or
solid manure was applied.
! Tree nursing, including inputs for planting and 3 years of
establishing of the orchards.
! Constructions for hail protection.
! Water for irrigation.
! Application of compost.
Two functional units (FU) were used to relate the
environmental impacts to the two main goals of
environmentally sound apple-growing, according to multifunctional agriculture (van der Werf and Petit, 2002;
Nemecek et al., 2004). First, ‘1 ha’ was taken as FU,
referring to the goal of minimizing the impacts while
cultivating a certain area. This area-related FU was used
to compare the impact intensity of the farms. Second,
‘total receipts’ was used as FU, referring to the goal of
minimizing the environmental impacts while achieving a
certain production value. This receipts-related FU allows
us to measure efficiency of the input use. It represents a
measure of the eco-efficiency, according to the ecoefficiency concept which aims to minimize environmental
impacts in relation to the amount or value of produced
goods or services (Schaltegger and Burrit, 2000; Scholz
and Wiek, 2005). ‘Total receipts’ includes sales of fresh
apples of all quality-classes including processed fruits
(culls), hail insurance proceeds and direct payments for
ecological compensation through participating in the
Swiss integrated farm management program (800 $ ha"1).
The receipts were recorded in Swiss Francs (CHF). To
convert into US dollars ($) an exchange rate of CHF
1.50 = $ 1.00 was used, which is the average exchange
rate in the period of investigation. The average portions of
the different apple qualities were 64% for first class
(0.53 $ kg"1), 20% for second class (0.33 $ kg"1) and
16% for processed fruits (0.12 $ kg"1), which results in an
average price of 0.43 $ kg"1 (weighted average over all
three quality classes).
313
2.2. Assessing the management influence on inputs and
environmental impacts
In this study, the differences between the farms can be
considered to be mainly due to differences in the farm
management because of the homogeneous boundary conditions of the farms (see below). We refer to farm management
in terms of input selection, which in turn have environmental
implications. Based on that, we defined the differences
between the farms as the management influence on inputs,
respectively the management influence on environmental
impacts. As a measure for the management influence we used
the ratio of the minimum and maximum values observed. For
example, if the maximum value was twice as high as the
minimum value, the management influence was factor 2.
The management influence is understood as a relative effect
size, not implying that the lowest environmental impacts
across the different environmental categories could be
realized in reality on one single farm.
In particular, the management influence is caused by
differences in conducting the allocation of inputs (e.g.
pesticides, fertilizers) to the different apple orchards of a
farm. Regarding this input allocation, we quantified the
management influence not only by mean values (M) but also
by the coefficient of variance (CV) and the skewness of
distributions. Investigating the correlations of M with CV
and skewness, allocation strategies to reduce inputs and
accordingly environmental impacts can be formulated.
Regarding the environmental impacts we interpreted the
three statistical parameters (M, CVand skewness) as expected,
unknown and dread risk, according to recent risk research
(Gigerenzer and Fiedler, 2003, unpublished data). This risk
definition is based on the following assumptions: firstly, M is a
measure of the central tendency and thus refers to the
environmental impact that can be expected in an average
among all orchards on a particular farm. Lowering the mean
impact reduces the expected environmental risk. Secondly, the
CV (standard deviation divided by M) shows the variation
level of the single data. The higher the variation level, the less
is known about how much a single result differs from the
expected M. This is one reason to interpret the CV as the
unknown risk. The other reason is that the level of variation is
controllable only to a certain extent, because of unexpected
events in the orchards (e.g. rain comes sooner than forecasted).
Thirdly, if a variable is not normally distributed, a positive
skewness informs us about an asymmetry with a longer tail to
the right, showing that values above M show higher deviations
from M than values below M (the opposite is true for a negative
skewness). Exceedingly high impacts on certain orchards bear
a selective environmental risk. Normally distributed impacts
indicate that a fruit-grower mastered the natural balance in
the orchards and was not forced to take emergency action. For
the dread risk, the management influence was considered to be
high if less than 50% of farms demonstrated distributions
which did not significantly deviate from the normal
distribution. The argument being that a small number of
314
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
farms obviously possess the potential to avoid positively
skewed distribution (dread risk), whereas the majority still
failed to do so. In this study the expression ‘dread risk’ for a
distribution that is skewed towards the unfavorable direction
must be viewed relatively, because the amount and type of
inputs per hectare of pesticides and fertilizers are limited by
the guidelines of integrated fruit-growing.
Generally, statistical risk assessment sets high impact cases
in relation to low impact cases, rather than focusing only on the
worst cases (Brachinger and Weber, 1997; Scholz and Tietje,
2002). As Mouron and Scholz (2005) have shown, statistical
risk assessment is highly relevant for perennial tree crop
systems, since a fruit-grower weighs up the potential gains and
losses among all orchards of a particular farm, having in mind
not only the present but also previous and future seasons.
2.3. Study location and farm-structure
The data used in this study are from a 4-year survey
conducted from 1997 to 2000 on 445 apple orchards of 12
fruit farms in Switzerland. Initially there were 25 farms in
the survey, 12 were selected because of highly homogeneous
boundary conditions concerning production constraints,
farm-structure and education of the farm manager (Zürcher
et al., 2003).
The farms are located in the flatter regions of Eastern and
Central Switzerland. The apple orchards are all on sites, well
qualified for fruit growing according to the land assessment
system for fruit crops in Switzerland (FAW, 1998). The
altitudes of the orchards are similar with an average of 477 m
above sea level (minimum = 390 m, maximum = 610 m,
S.D. = 72 m). The climate data for the period 1961–1990
(MeteoSwiss, 2000; station in Güttingen) underline the good
conditions for apple growing as well: the annual mean
temperature is 8.5 8C with the highest and lowest monthly
mean temperature of "0.3 8C (January) and 17.6 8C (July).
Frost damage in the apple orchards is seldom, occurring about
once in every 15 years. The mean annual precipitation is
916 mm with the highest and lowest monthly mean
precipitation of 58 and 104 mm. Rainfall is well distributed
throughout the growing season, so that hardly any irrigation is
needed for apple-growing. On the other hand, the regular
rainfall requires a consistent application of fungicides
especially in the spring, mainly to control scab (Venturia
inaequalis) and powdery mildew (Podosphaera leucotricha).
Among pests, annual regulation of codling moth (Cydia
pomonella) and rosy apple aphid (Dysaphis plantaginea) was
practiced. Other pests, such as summer fruit tortrix moth
(Adoxophyes orana) and apple sawfly (Hoplocampa testudinea) did not exceed the threshold every year and not in all
orchards. Especially the pests that occur irregularly require a
situation adjusted application of insecticides (Höhn et al.,
2000).
The farms were managed according to the guidelines of
the Swiss program of integrated farm management (IPSuisse, 2003), which aims to achieve profits while respecting
nature and personal health (Boller et al., 1997). The fruitgrowers were already well introduced to the integrated
farming system before our study commenced. They had
access to a local extension service, which recommended
pesticide application according to pest threshold and taking
beneficial organisms into account. Additionally all farm
managers have diplomas in farming resulting from a 3-year
apprenticeship, master courses or internships at several other
farms. They all sold their apples to regional wholesalers and
were subject to the same quality-price conditions.
The 12 farms are specialized full-time family operations
where the main farm income is fruit production. Fresh apple
(Malus domestica Borkh.) was the major fruit type covering
80% of the orchard area per farm (M = 7.2 ha). Table 1
presents the structure of the farms and their apple orchards.
In this study an orchard is defined as a block of apple trees of
one cultivar, planted in the same year and trained in the same
way. The average area of an orchard was 0.4 ha. The
following apple cultivars were planted: golden Delicious
(19%), Jonagold (11%), Idared (11%), Maigold (9%), Elstar
(8%), Boskoop (8%), Arlet (6%), Gravensteiner (6%), Cox
Orange (4%), Gala (4%) and 13 other cultivars with a total of
less than 3%. Sixty-five percent of the orchards were
protected by hail nets; the other orchards had some insurance
against hail damage. All farms had their orchards close by
and could reach them within a distance of 2 km.
2.4. Data recording for LCA
The survey was carried out in cooperation with the Swiss
Federal Research Station for fruit-growing, Viticulture and
horticulture in Wädenswil, Switzerland. The fruit farmers
Table 1
Structure of the 12 fruit farms under study, based on the year 2000
Age of farm manager (years)
Years in the position as farm manager (years)
Permanent labor per farm (N)
Seasonal workers per farm (N)
Acreage per farm with apple orchards (ha)
Acreage of an individual apple orchard (ha)a
Density of trees (trees ha"1)a
Age of apple orchards (years since planting)a
a
Mean
Standard deviation
Minimum
Maximum
43.9
18.4
2.0
7.6
7.2
0.4
2340.0
8.0
10.1
11.5
0.7
7.7
3.9
0.2
602.0
3.1
28.0
4.0
1.0
1.0
1.7
0.1
1502.0
4.0
66.0
48.0
3.0
28.0
14.4
1.4
3984.0
15.0
Only apple orchards in the productive phase (4–15 years old).
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
were instructed to record data for full cost accounting using
a commercial software package (ASA-Agrar #, 2003).
Using this software all farmers kept a diary about all
activities in the individual fruit orchards. Each activity had
to be linked to the actual physical and monetary inputs of
labor force, machinery, equipment, pesticides, fertilizers and
materials. Also the harvested apples in the different quality
categories were recorded. The physical data of the farm
diaries were used for the impact assessment of this study.
Each year the completeness of the recorded data was
checked on each farm. Together with the farm manager the
data was also checked for plausibility. As a result of those
checks, the original total of 499 annual data sets of
individual apple orchards was reduced to 445, covering a
total of 165 ha. There are no missing data for the 445
orchards analyzed in this study. All orchards included in the
study had been in the productive phase, with a minimum age
of four and maximum age of 15 years.
3. Results
3.1. Management influence on inputs and outputs
Table 2 shows the statistical parameters of the inputs and
outputs for the apple-growing process of the 12 farms. The
management influence, represented by mean values, is for all
315
inputs and outputs considerable, with a ratio of maximum
and minimum of factor 2 at the least. Particularly the average
input per farm differs for pesticides at least by factor 5 (e.g.
fungicides: lowest M = 9.2; highest M = 52.8 kg ha"1 active
matter), for fertilizers by factor 10 (e.g. N-fertilizer: lowest
M = 10.5; highest M = 110.3 kg ha"1 nitrogen), for machinery (e.g. diesel: lowest M = 161.9; highest M =
375.0 kg ha"1) and receipts (lowest M = 8.7; highest M =
23.1 k$ ha"1) by factor 2.
The management influence, represented by the coefficient of variance (CV), is for most inputs around factor 5
(Table 2). Machinery shows the lowest average CV by not
exceeding 28%, while for pesticides the average CV is
between 30 to 99%, for fertilizers between 56 to 200% and
for the outputs at 50%. Thus, the data suggest that the
variation level for inputs and outputs is influenced by the
farm management in a significant way.
The management influence, represented by the portion of
farms with normally distributed values, differs among the
variables. Whereas values for machinery and outputs are
normally distributed on most farms (67–100%), for
pesticides and fertilizers the likelihood for positively
skewed distributions is higher. This is indicated by less
than 50% of farms with normal distributions (Table 2).
Remarkable is that while the inputs of fungicides and
herbicides are normally distributed on half of the farms, this
portion is lower for insecticides (17%). Insecticides tend
Table 2
Inputs and outputs per hectare of the apple-growing process; average values of 12 fruit farms, Switzerland, 1997-2000
Mean
Inputs
Pesticides (kg active matter)
Fungicide
Insekticide
Herbicide
Other plant
treatment products
Fertilizers
N-fertiliser (kg N)
Ca- and Mg-fertiliser
(kg Ca, Mg)
K-fertiliser (kg K2O)
P-fertiliser (kg P2O5)
Machinery
Diesel (kg)
Tractor (kg)b
Equipment (kg)b
Buildings (m2) c
Outputs
Total receipts ($ ha"1)
Yield (t ha"1)
Minimum
Maximum
23.5
1.2
1.7
3.1
9.2
0.4
0.9
0.7
52.8
2.5
4.7
7.8
62.0
51.7
10.5
0.0
47.2
3.8
CV (%)
Farms with normally
distributed values (%) a
Skewness
Minimum (%)
Maximum (%)
30
54
43
99
7
33
11
15
55
92
86
250
50
17
50
8
0.4
0.8
0.0
1.2
110.3
157.2
56
119
24
37
100
332
17
9
0.0
1.0
0.0
0.0
102.2
17.4
82
200
16
83
131
424
10
0
0.3
1.7
231.7
19.8
48.5
0.2
161.9
13.0
30.9
0.2
375.0
30.3
73.3
0.3
27
28
28
21
9
15
11
6
46
48
52
44
92
100
75
67
0.3
0.4
0.3
0.3
13463.4
31.4
8671.6
19.1
23114.8
46.2
49
50
24
21
74
73
83
83
0.5
0.4
Water for irrigation and compost are unlisted, since these inputs were only used by two farms to a low extent.
a
Normal distribution was tested by Kolmogorov-Smirnov test, P > 0.05.
b
For life cycle assessment each type of tractor or equipment was attributed with a certain kg h"1 (according to the weight and expected life-hours) which was
multiplied by the hours the tractor or equipment was used.
c
For life cycle assessment m2 was adjusted to an expected life time of 80 years for the buildings; buildings are for shelter of tractors and equipment.
316
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
Table 3
Area-related life cycle impact assessment profile of 12 fruit farms, Switzerland, 1997–2000
Farms with normally
distributed values (%) a
Skewness
28
33
100
75
0.0
"0.1
12
37
92
0.0
41
24
65
25
0.5
0.8
56
26
123
17
0.2
1.0
1.1
3
1
13
92
0.2
3.2
2.2
5.5
73
41
133
58
0.2
25.4
17.9
40.7
23
12
35
75
0.3
Impact categories
Mean
Minimum
Maximum
CV (%)
Minimum (%)
Energy use (GJ eq. ha"1)
Global warming potential
for 100 years (t CO2 eq. ha"1)
Ozone formation
(kg C2H4 eq. ha"1)
Aquatic ecotoxicity
(kg Zn eq. ha"1)
Terrestrial ecotoxicity
(kg Zn eq. ha"1)
Aquatic eutrophication
(kg PO4 eq. ha"1)
Terrestrial eutrophication
(kg PO4 eq. ha"1)
Acidification
(kg SO2 eq. ha"1)
37.6
2.6
23.1
1.6
54.5
3.8
21
22
9
13
17.2
10.6
26.5
23
4.7
2.0
9.3
0.5
0.1
1.0
Maximum (%)
eq.: equivalents.
a
Normal distribution was tested by Kolmogorov-Smirnov test, P > 0.05.
therefore to be applied in an unbalanced way more frequently
than fungicides and herbicides. A reason for this result could
be the irregular occurrence of some pests whereas the disease
situation tends to be steady in the region under study (see
above). Nevertheless, the results suggest a potential towards
more balanced pesticide inputs for most farms. Among the
fertilizers the skewed input on most farms reflects to a certain
extent the situation adjusted application of nutrients. For
nitrogen (N), the annual applications must be adjusted to the
fruit loadings and other indicators in order to achieve a
favorable fruit quality. However, there is still potential to
achieve a more balanced nitrogen input, as 17% of the farms
have demonstrated. Other nutrients (P, K, Ca and Mg) can be
stored in the soil over years and do not necessarily require
annual applications. Thus, seen over a period of 4 years, a
skewed distribution for these nutrients seems to be normal.
3.2. Management influence on area-related
environmental impacts
Table 3 presents the statistical parameters of the impact
assessment profile. The mean values, corresponding to the
expected risk, show for all impact categories a ratio of at
least factor 2. Except for aquatic eutrophication the lowest
and highest means differ by only 10%. In particular, the
impact categories that show a ratio between the farms of
between factors 2 and 3 are energy use (lowest
M = 23.1 GJ eq.1 ha"1; highest M = 54.5 GJ eq. ha"1),
global warming potential (lowest M = 1.6 t CO2 eq. ha"1;
highest M = 3.8 t CO2 eq. ha"1), ozone formation (lowest
M = 10.6 kg C2H4 eq. ha"1; highest M = 26.5 kg C2H4 eq.
ha"1), terrestrial eutrophication (lowest M = 2.2 kg
PO4 eq. ha"1; highest M = 5.5 kg PO4 eq. ha"1) and acidification (lowest M = 17.9 kg SO2 eq. ha"1; highest
1
eq.: equivalents.
M = 40.7 kg SO2 eq. ha"1). Thus, on a per hectare basis,
the management influence to reduce the environmental
impact of these five categories is at factor 2. For terrestrial and
aquatic ecotoxicity, the area-related management influence is
even bigger in the region between factors 4 and 8.
The unknown risk, indicated by the CV, varies considerably among the farms. The management influence to reduce
the CV is at factors 3 to 4 for all impact categories, except for
aquatic eutrophication, where the potential is at factor 13.
Although these management influences can be considered as
high, the absolute influence on the environment through
reducing the variation level might be low, because for most
categories the average of the CV is lower than 23%. Higher
average CV’s occurred for terrestrial eutrophication (73%),
terrestrial ecotoxicity (56%) and aquatic ecotoxicity (41%).
These three categories had the lowest numbers of farms
demonstrating normal distributions and therefore show a high
likelihood for positive skewed distributions.
3.3. Management influence on receipts-related
environmental impacts
Table 4 shows the statistical parameters for the impact
categories with the functional unit ‘total receipts’. Concerning the expected risk, the receipt-related management
influence was in general higher than the area-related
influence. The lowest receipts-related management influence to reduce the expected risk was found for the category
terrestrial ecotoxicity with factor 4 (lowest M = 0.01 g Zn
eq. $"1; highest M = 0.04 g Zn eq. $"1),
the highest
influence occurred for ozone formation with factor 7 (lowest
M = 0.28 g C2H4 eq. $"1; highest M= 2.15 g C2H4 eq. $"1).
The management influence for the receipts-related
unknown risk is at factor 3 (Table 4), the same effect of
management as for the area-related impacts (Table 3). In
contrast to the area-related impacts, the receipts-related
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
317
Table 4
Receipts-related life cycle impact assessment profile of 12 fruit farms, Switzerland, 1997–2000
Impact categories
Energy use (MJ eq. $"1)
Global warming potential
for 100 years
(kg CO2 eq. $"1)
Ozone formation
(g C2H4 eq. $"1)
Aquatic ecotoxicity
(g Zn eq. $"1)
Terrestrial ecotoxicity
(g Zn eq. $"1)
Aquatic eutrophication
(g PO4 eq. $"1)
Terrestrial eutrophication
(g PO4 eq. $"1)
Acidification (g SO2 eq. $"1)
Mean
Minimum
Maximum
CV (%)
Minimum (%)
Maximum (%)
Skewness
Farms with normally
distributed values (%)
a
1.90
0.13
0.69
0.05
4.23
0.24
62
61
32
31
92
94
17
25
1.7
1.6
0.87
0.28
2.15
61
31
102
17
1.7
0.24
0.07
0.44
83
50
136
8
2.1
0.03
0.01
0.04
95
58
143
25
2.1
0.05
0.02
0.11
71
26
98
33
2.2
0.20
0.06
0.35
65
33
109
17
1.8
1.55
0.52
2.83
63
34
98
25
1.8
eq.: equivalents.
a
Normal distribution was tested by Kolmogorov-Smirnov test, P > 0.05.
impacts show a higher average level of variation (CV of 60–
90%). Thus, the influence of the farm management on the
receipts-related variation level might have bigger implications on the environment than for the area-related impacts.
Concerning the receipts-related dread risk, the management influence is high, indicated by only a small proportion
of farms (8–33%) with normally distributed impacts
(Table 4). Most farms show significantly right skewed
impacts in all environmental categories. Generally, a higher
positive skewness resulted if impacts were measured per
receipts instead of per hectare. This suggests that the level of
inputs did in most cases not determine the level of receipts.
The biggest difference was found for energy use, where
100% of the farms showed normally distributed values
measured per hectare (Table 3), whereas only 17% of the
farms did so if measured per receipts (Table 4).
3.4. Correlations between environmental impact
categories
A principal component analysis (PCA) was performed to
group the eight impact categories according to their
correlation. A three component solution turned out to be
adequate. The first principal component accounted for 57%,
the second for 22%, and the third for 11% of the variance.
Principal components loadings after a varimax rotation are
shown in Table 5. Five impact categories related to energy
inputs loaded high on the first component (energy use,
greenhouse potential, ozone formation, terrestrial eutrophication and acidification). The high correlation of these five
impacts is of interest for environmental management because
if one of those impacts is kept low, the other four impacts will
be low as well. The correlation of terrestrial eutrophication
with energy use is caused by N-fertilizers which act on both of
these impact categories: Terrestrial eutrophication is influenced by N losses to air (mainly NH3) and the manufacturing
of N-fertilizers requires high energy consumption. On the
second component the two impact categories aquatic and
terrestrial ecotoxicity loaded high, indicating that these two
impacts act in parallel. Both are related to pesticide inputs. On
the third component only aquatic eutrophication loaded high.
This impact category is dominated by P-compounds (Fig. 1)
through P-losses to water. P-fertilizers need only one
application per year or even less, while N-fertilization
requires several applications over a season. Thus, there is now
correlation between P- and N-fertilization. This background
of fertilization practice gives reason to consider a three
component solution as adequate despite the eigenvalue 0.9 of
the third component being a little lower than the common
threshold of eigenvalue 1.0.
Over all, the results indicate that the three environmental
issues energy, ecotoxicity and aquatic eutrophication can be
influenced independently of each other to a high degree by
Table 5
Principal component analysis (PCA): rotated component matrix with eight
area-related impact categories; 445 apple orchards, Switzerland, 1997–2000
Component
1
Impact categories
Energy use (GJ eq. ha"1)
Global warming potential
for 100 years (t CO2 eq. ha"1)
Ozone formation (kg C2H4 eq. ha"1)
Aquatic ecotoxicity (kg Zn eq. ha"1)
Terrestrial ecotoxicity (kg Zn eq. ha"1)
Aquatic eutrophication (kg PO4 eq. ha"1)
Terrestrial eutrophication (kg PO4 eq. ha"1)
Acidification (kg SO2 eq. ha"1)
Total variance explained
Initial eigenvalues
Variance explained (% of variance)
2
3
0.95
0.95
"0.03
"0.01
0.06
0.20
0.94
0.00
0.07
0.19
0.90
0.94
"0.04
0.93
0.93
0.05
0.13
0.13
"0.01
0.07
0.00
0.98
0.16
0.16
4.58
57.19
1.76
22.00
0.89
11.17
N = 445; loadings exceeding 0.8 are in bold print; eq.: equivalents.
318
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
managing the domain of energy consuming inputs as well as
the applications of pesticides and fertilizers.
3.5. Input-impact map
An input-impact map provides an overview for environmental management, showing which inputs influence which
impacts the most. Based on the results of the PCA presented
above, it is possible to concentrate on three principal
components in environmental management of apple-growing. From each of the three principle components (Table 5)
one highly loading impact category was picked to represent
them, namely: energy use, aquatic ecotoxicity and aquatic
eutrophication. Fig. 1 shows the input-impact-map by
illustrating the level of statistically significant correlations
of the three impact categories with the inputs. Energy use,
representing the first component, shows significant correlations with eight of the total 13 inputs. The highest correlation
of energy use was found with diesel (r = 0.78) and with the
inputs that are connected with the use of diesel, namely:
equipment (r = 0.66), tractors (r = 0.61) and buildings for
machinery shelter (r = 0.59). Hail protection is also highly
energy related (r = 0.59). In addition the inputs of Ca- and
Mg-fertilizer (r = 0.39), N-fertilizer (r = 0.39) and other
plant treatment products (r = 0.30) contribute also to energy
use. In contrast K-fertilizer is neither correlated to energy
use, nor to aquatic ecotoxicity or aquatic eutrophication.
Some of the energy related inputs contribute also to the
impact variable aquatic eutrophication, such as N-fertilizer
and to small extent Ca- and Mg-fertilizer, equipment,
buildings and diesel. For aquatic eutrophication P-fertilizer
(r = 0.78) and N-fertilizer (r = 0.44) are the most related
inputs. Insecticides (r = 0.68) and fungicides (r = 0.48), and
to a minor extent herbicides and P-fertilizer show the highest
correlations with the impact category aquatic ecotoxicity.
3.6. Area- and receipts-related energy use
The trend in average energy use of the farms remains
whether the functional unit is ‘one hectare’ or ‘total receipts’,
as Fig. 2 illustrates. This means that farms with a low energy
use per hectare (i.e. farm numbers 6 and 2) may demonstrate at
the same time low energy use per ‘total receipts’. Apparently,
for those farms it was possible to achieve relatively high
receipts with low (area-related) energy intensity, compared to
other farms of the sample. On the other hand, farms with
relatively high energy input per hectare (i.e. farm numbers 9
and 11) were not able to increase the receipts in the same
proportion, so that those farms end up with high energy use
per output (receipts) and thus with a low energy efficiency.
These results indicate that favorable environmental management has the potential to decrease the energy input and at the
same time to increase the receipts.
Fig. 2 provides also the portions of the different inputs
contributing to energy use. In the average the energy use is
mainly due to energy carriers, i.e. diesel and electricity (33%),
machinery (20%) and hail protection (18%). Mineral
fertilizer, pesticides and buildings contribute each at 10%.
These percentages do not change whether the area- or receiptrelated FU is considered, due to mathematical reasons.2
3.7. Correlation between mean and CV of impact
categories
For the three impact categories energy use, aquatic
ecotoxicity and aquatic eutrophication the correlation
between mean and CV was calculated for the 12 farms.
The results differ depending on the functional unit. In the
case of the receipts-related LCA, the correlation between
mean and CV was found to be positive and statistically
significant, F(1, 10) = 11.24, P < 0.01 for energy use
(Fig. 3i) as well as for aquatic eutrophication (Fig. 3iii),
F(1, 10) = 14.50, P < 0.01, whereas for aquatic ecotoxicity
(Fig. 3ii) this correlation was not significant F(1, 10) = 1.54,
P > 0.05. Thus, farms with a low variation level of receiptsrelated energy use or aquatic eutrophication were likely to
show at the same time low mean values. The linear
relationship between mean and CV can be considered as
high, with r = 0.73 for energy use and r = 0.77 for aquatic
eutrophication. These results indicate that variation control
can be used as a management mechanism to achieve ecoefficiency for energy use and aquatic eutrophication, but this
mechanism does not work in the case of ecotoxicity.
Fig. 1. Area-related input-impact correlation: Pearson correlations of production inputs with three impact categories for 445 apple orchards, Switzerland, 1997–2000; correlations are significant at the 95% significance level
(two-tailed). eq.: equivalents.
2
Changing the FU means changing the denominator, whereas the
numerator remains the same. Thus, the proportion of numerators does
not change independent from the FU.
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
319
Fig. 2. Average use of non-renewable energy resources per hectare (left
scale) and per total receipts (right scale) of life cycle assessment for applegrowing on 12 fruit farms, Switzerland, 1997–2000. eq.: equivalents.
For the functional unit area, the life cycle assessment
showed neither for energy use (P = 0.994) nor for aquatic
eutrophication (P = 0.098) nor for aquatic eutrophication
(P = 0.190) a statistically significant correlation between
mean and CV. Thus, a positive interdependency of expected
and unknown risk was not observed on a per hectare basis, as
it was for receipts-related LCA.
Interesting to note that the farms with the lowest energy use
per receipts (i.e. farm numbers 1 and 2) also achieved lowest
means for aquatic ecotoxicity and aquatic eutrophication
(Fig. 3). This suggests that favorable receipts-related ecoefficiency can be reached on one farm for all three impact
categories. Successful environmental management optimizes
the input domains of energy, pesticides and fertilizers without
a trade-off between these input domains. Thus, attaining high
(receipt-related) eco-efficiencies for all three impact categories are not necessarily contradictory management goals.
4. Discussion
4.1. Expected environmental impacts of integrated
apple-growing compared with arable crops
If we take the mean value as an indicator of the expected
impact, apple-growing can be represented by 37.6 GJ eq. ha"1
for energy use, 4.7 kg Zn eq. ha"1 for aquatic ecotoxicity
Fig. 3. Linear correlation between mean and coefficient of variance for (i)
energy use, (ii) aquatic ecotoxicity, and (iii) aquatic eutrophication per total
receipts of 12 fruit farms, Switzerland, 1997–2000; r is the Pearson correlation
coefficient, P the significance level (two-tailed). eq.: equivalents.
and 1.0 kg PO4 eq. ha"1 for aquatic eutrophication. An LCA
study of Nemecek et al. (2002) on 18 different field crops with
integrated farming practice in Switzerland applied the same
LCA methodology as in this study. The results suggest that
potatoes, sugar beets and carrots cause similar expected
impacts to the ones we found for apple-growing, regarding
energy use and aquatic ecotoxicity. Not only the mean values
but also the percentage of the input-related impacts is similar,
caused by the dominance of energy carriers and machinery
320
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
for the impact category energy use and the dominance of
pesticides for aquatic ecotoxicity. Regarding aquatic eutrophication apple-growing shows by far (at least by factor 10) a
lower meanvalue than all arable crops, mainly as a result of low
P-fertilizer needs in apple-growing.
Because apple trees are perennial, a comparison to arable
farming at a crop rotation level is reasonable. An LCA study in
Switzerland of a crop rotation including potatoes, winter
wheat, grain maize spring barley, grass-clover and forage
catch crop (Nemecek et al., 2001) reported an energy use of
24.6 GJ eq. ha"1, an aquatic ecotoxicity of 2,2 kg Zn
eq. ha"1, and an aquatic eutrophication of 1.5 kg
PO4 eq. ha"1 per year on the average over the whole rotation
of the integrated faming system. Thus the area-related energy
use is about 50% higher for apple-growing in comparison to
the arable crop rotation system, whereas the aquatic
eutrophication is lower. Despite the input of pesticides in
apple-production being a factor of 10 higher (i.e. for applegrowing 29.5 kg ha"1, for arable crop rotation 3.3. kg ha"1
active matter), the aquatic ecotoxicity is only by a factor 2
higher for apple-growing compared to the arable crop rotation
system. The relatively high aquatic ecotoxicity in arable crop
rotation is mainly caused by the input of heavy metals in liquid
and solid manures which are not used in apple-growing.
4.2. Management influence to reduce expected
environmental impacts
The empirical results show that homogeneous boundary
conditions of a farm group, i.e. geographical, technical and
educational background, do not imply homogeneous
environmental impacts. There is a considerable potential
for apple-growers to improve environmental management,
indicated by differences of mean values for impact
categories between factors 2 and 3 on a per hectare basis
and between factors 4 and 7 if calculated per ‘total receipts’
for the examined sample of fruit farms. This management
influence to reduce environmental impacts within a welldefined farming system comes close to the suggested goal of
factor 4 (Weizsäcker et al., 1997) or even factor 10
(Schmidt-Bleek, 1994) which have been proposed as
sustainable target levels.
The management influence has also been investigated in
other studies when comparing different farming systems
(e.g. integrated versus organic). Reganold et al. (2001) for
instance reported a factor 4 between organic and integrated
apple-growing systems in Washington State according to an
environmental impact rating per hectare. For dairy farming
Haas et al. (2001) found on a per hectare basis differences
between intensive and organic farms in Germany corresponding to factor 4 for eutrophication, factor 3 for energy
use; in contrast, the two farming systems did not differ
regarding global warming or acidification. For arable crop
rotation on farms in the UK, however, Bailey et al. (2003)
reported an average difference between conventional and
integrated farms of only 8% for the energy use per hectare.
Thus, the differences between the farming systems, mainly
due to different production technologies and constraints,
might not be greater than the differences within a single
farming system, mainly caused by varying management
abilities.
Large differences between farms within the same farming
system were reported by an LCA study of Rossier and
Gaillard (2004) on Swiss farms. For arable crops the ratio
between the farms for energy use, aquatic ecotoxicity and
eutrophication was factors 5, 6 and 4, respectively, which
comes close to the magnitude we found for apple-growing.
In any case, the authors could not explain the differences in
environmental performance by parameters such as farm
area, age of farm manager, or other external parameters.
They concluded that the management abilities of the farmers
could have been crucial. In Section 4.3 we provide some
practical implications to improve the environmental
management on fruit farms, based on results of the PCA
and the statistical risk assessment.
4.3. Implications to improve environmental
management
With the control of just three key impact categories, a
total of eight impact categories will come under control,
according to the results of the PCA. Additionally the PCA
shows that the three key impact categories can be managed
independently from each other. In particular, a farm manager
should focus on controlling energy use, aquatic ecotoxicity
and aquatic eutrophication. Keeping energy use low is most
effective because a low use of energy correlates with low
levels of four other impact categories (ozone formation,
global warming potential, terrestrial eutrophication and
acidification). Furthermore, the management of aquatic
ecotoxicity parallels management of terrestrial ecotoxicity.
To attain a low level of energy use, an apple-grower should
keep the input of diesel fuel low. This could be accomplished
by using small and fuel efficient engines and by reducing the
input of machinery hours per hectare, for instance, through
optimized organization during the apple picking period.
Similarly important is to keep the number and size (weight)
of the machinery as low as possible but at the same time to
increase machinery life span as much as possible. This might
be reached by an optimal relation between machinery and
area under cultivation. Further the area with hail protection
constructions should be kept as small as possible by orchards
with hail insurance, as long as there are no negative
consequences on the receipts. In order to attain a low
ecotoxicity and low eutrophication the ecological balance in
the orchards should be optimized by a sound but low input of
pesticides and nutrients avoiding negative side effects on
beneficial organisms and plant robustness against diseases.
Such an ecological balance might also be a reason that pests
which occur irregularly in a region can be kept under the
threshold in most cases and thus skewed distributions of
pesticides (especially of insecticides) can be avoided. But
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
this ecological balance might not be easy to achieve and to
hold over the years, as indicated by the low proportion of
farms with normally distributed pesticide inputs. It might
also be difficult to re-establish an ecological balance in an
orchard, because a damaged orchard often takes more than
one season to recover.
Even the expert fruit-grower often does not know for
certain whether a specific input will bring the desired effect
or not. This is because of the uncertainty of rainfall and
temperature and the different susceptibility of orchards to
pests and diseases. As Mouron et al. (in press) point out, a
distributional thinking is adequate to deal with this kind of
risk situation that is typical for orchard farming. A
distributional thinking has high management implication
because it helps to keep in mind not only each orchard as a
specific case but also the whole of all orchards that compose a
fruit farm. In this context the statistical risk assessment
suggests an interesting mechanism, which can help farmers
to implement distributional thinking in their management. It
is based on the significant correlation between the mean and
the level of variability (CV) of receipts-related energy use
and aquatic eutrophication. We suggest that, in practice,
farms may hope to achieve low energy use and low
eutrophication as long as they control the level of variability
of these impact categories in relation to the receipts. The
above mentioned ecological balance might be a key issue to
do so, but the causes that lead to such a control of the
variability among the orchards could be an interesting subject
of further research. Since no such correlation between mean
and CV was found for aquatic ecotoxicity, pesticide inputs
cannot be reduced by a low variation level. But we would like
to point out that farms with low energy use and low aquatic
eutrophication per receipts can also be farms with low
pesticide inputs, as was the case for farms no. 1 and 2.
321
(integrated apple-growing) and under homogeneous
boundary conditions such as production technology,
climate, education, and market access. The high ratio
of around factor 6 for key impact categories of LCA imply
that the promotion of environmentally sound applegrowing is not only a question of choosing a farming
system (e.g. organic versus integrated farming) but that
an understanding of the system specific management
influence is crucial. The study demonstrates that the
methods of PCA and statistical risk assessment complement each other in such an understanding. Whereas
PCA detected environmental domains (area-related energy
use, aquatic ecotoxicity and aquatic eutrophication) which
can be managed independently of each other, statistical
risk assessment provided the insight that for some
domains (receipts-related energy use and aquatic eutrophication) the expected risk (M) can be controlled by
keeping the level of variability (CV) low. Both methods
demonstrate the importance of differentiating between
impact categories and between functional units (arearelated, receipts-related) for managerial purposes at the
farm level.
Because some farms with low energy inputs per receipts
also attained low pesticide and fertilizer inputs per receipts,
we suggest that a favorable environmental performance of a
farm is mainly based on the effective use of technologies,
including proper handling and timing, and probably most
important is a realistic distributional thinking that allows the
farmer to allocate the inputs to the different orchards in
relation to the expected receipts. Thus, we think that the
endeavor of farm data recording for individual orchards,
analyzed by a statistical risk assessment as proposed in this
study, could be relevant to favoring environmentally sound
orchard farming.
4.4. Recommendation of how to use PCA in LCA studies
Acknowledgements
For managerial purposes at the farm level, we recommend for LCA applications initially to include as many
impact categories as available and then, in a second step to
reduce the complexity by grouping the impact categories
according to their correlations by performing a PCA. We
recommend in a third step to single out one highly loading
impact category for each principal component. The chosen
impact categories should be managerially and cognitively
accessible to the farmers (Werner and Scholz, 2002). In the
present study we initially used eight impact categories.
Further impact categories such as biodiversity, soil fertility
or landscape are subject of ongoing research projects and are
recommended to include in future studies.
5. Conclusions
This study examined the variability of environmental
impacts between fruit farms of the same farming system
First of all, the authors would like to thank the fruitgrowers for their co-operation over the whole span of the
study. We also acknowledge the support of the Swiss
Federal Research Station for fruit-growing, Viticulture
and horticulture in Wädenswil, particularly Dante Carint
for assistance in programming the data base, Heinrich
Höhn and Werner Siegfried for their recommendations on
assessing pesticide impacts. For comments on statistical
analysis we thank Michael Siegrist and Ralph Hansmann.
We appreciate comments from Gérard Gaillard, Thomas
Köllner and Claudia Binder, as well as two anonymous
reviewers on an earlier draft of this paper. We wish to
thank Francis Hesford for proof reading of the article. The
study was financially supported by the Swiss Association
of Fruit Growers, Zug; the Swiss Federal Office for
Agriculture, Bern; the Swiss Center for Agricultural
Extension, Lindau and Lausanne; and COOP supermarket,
Basel.
322
P. Mouron et al. / Agriculture, Ecosystems and Environment 114 (2006) 311–322
References
Andersson, K., 2000. LCA of food products and production systems. J. LCA
5 (4), 239–248.
ASA-Agrar #, 2003. Software for cost accounting of perennial crops. At
http://www.asaagrar.com/.
Audsley, A., Alber, S., Clift, R., Cowell, S., Crettaz, R., Gaillard, G.,
Hausheer, J., Jolliet, O., Kleijin, R., Mortensen, B., Pearce, D., Roger,
E., Teulon, H., Weidema, B., van Zeijts, H., 1997. Harmonisation of life
cycle assessment for agriculture. Final report, Concerted Action AIR3CT94-2028, European Commission DG VI, Brussels, Belgium.
Bailey, A.P., Basford, W.D., Penlington, N., Park, J.R., Keatinge, J.D.H.,
Rehman, T., Tranter, R.B., Yates, C.M., 2003. A comparison of energy
use in conventional and integrated arable farming systems in the UK.
Agric. Ecosyst. Environ. 97, 241–253.
Bertschinger, L., Mouron, P., Dolega, E., Höhn, H., Holliger, E., Husistein,
A., Schmid, A., Siegfried, W., Widmer, A., Zürcher, M., Weibel, F.,
2004. Ecological apple production: a comparison of organic and integrated apple-growing. In: Bertschinger, L., Anderson, J.D. (Eds.),
Proceedings of the XXVI International Horticultural Congress: Sustainability of Horticultural Systems, ISHS, Acta Hortic. 638, 321–332.
Boller, E.F., Malavolta, C., Jorg, E. (Eds.), 1997. Guidelines for integrated
production of arable crops in Europe. IOBC Technical Guidline 111, 1st
ed. IOBC/WPRS Bull. 20 (5).
Brachinger, H.W., Weber, M., 1997. Risk as a primitive: a survey of
measures of perceived risk. Oper. Res. Spectrum 19 (4), 235–294.
Brentrup, F., 2003. Life cycle assessment to evaluate the environmental
impact of arable crop production. Ph.D. Thesis. Göttingen, p. 187.
Dinkel, F., Pohl, C., Ros, M., Waldeck, B., 1996. Ökobilanz stärkehaltiger
Kunststoffe – Band I. Bundesamt für Umwelt, Wald und Landschaft
(BUWAL), Bern, Switzerland Schriftenreihe Umwelt 271/I, p. 188.
FAL, 2002. Ökobilanzen – Beitrag zu einer nachhaltigen Landwirtschaft.
Agroscope FAL Reckenholz, Zürich, Schriftenreihe der FAL, vol. 38, p.
37.
FAW, 1998. Die Bewertung der Obstkultur, Flugschrift Nr. 61. Agroscope
FAW, Wädenswil, Switzerland, p. 21.
Frischknecht, R., Bollens, U., Bosshart, S., Ciot, M., Ciseri, L., Doka, G.,
Dones, R., Gantner, U., Hischier, R., Martin, A., 1996. Ökoinventare
von Energiesystemen. Grundlagen für den ökologischen Vergleich von
Energiesystemen und den Einbezug von Energiesystemen in Ökobilanzen für die Schweiz. Auflage No. 3, Gruppe Energie – Stoffe – Umwelt
(ESU), Eidgenössische Technische Hochschule Zürich und Sektion
Ganzheitliche Systemanalysen, Paul Scherrer Institut Villigen/Würenlingen, Schweiz.
Gaillard, G., Hausheer, J., 1997. Ökobilanz des Weizenanbaus: Vergleich
der Intensiven, der Integrierten und der Biologischen Produktion.
Kongressband, VdLUFA, Darmstadt, Germany, pp. 447–450.
Gigerenzer, G., Fiedler, K., 2003. Minds in environments. Max Plank
Institute for Human Development, Berlin. Unpublished, available at
sekgigerenzer@mpib-berlin.mpg.de.
Haas, G., Wetterich, F., Geier, U., 2000. Life cycle assessment framework in
agriculture on the farm level. J. LCA 5 (6), 345–348.
Haas, G., Wetterich, F., Köpke, U., 2001. Comparing intensive, extensified
and organic grassland farming in southern Germany by process life
cycle assessment. Agric. Ecosyst. Environ. 83, 43–53.
Heijungs, R., Guinée, J.B., Huppes, G., Lankreijer, R.M., Udo de Haes,
H.A., Wegener Sleeswijk, A., 1992. Environmental life cycle assessment of products, Guide-October 1992. Leiden.
Höhn, H., Höppli, H.U., Graf, B., 2000. Astrobenuntersuchungen 2000/2001:
auf und ab bei Schildläusen. Obst- und Weinbau 137 (6/01), 141–145.
Houghton, J.T., Meira Filho, L.G., Callander, B.A., Harris, H., Kattenberg,
A., Maskell, A., 1995. Climate change 1995. In: The Science of Climate
Change, Cambridge University Press, United Kingdom.
IP-Suisse, 2003. Gesamtbetriebliche Anforderungen. Schweizerische Vereinigung integriert produzierender Bauern, Zollikofen, Switzerland, at
http://www.ipsuisse.ch/.
Jolliet, O., Crettaz, P., 1997. Critical surface-time 95—a life cycle impact
assessment methodology including fate and exposure. In: ETH Lausanne, Institute of Soil and Water Management, Lausanne, Switzerland.
Keating, B.A., McCown, R.L., 2001. Advances in farming systems analysis
and intervention. Agric. Systems 70, 555–579.
LCA Nordic, 1995. Technical Report No. 10 and Special Report No. 1–2,
TemaNord1995:503,NordicCouncilofMinisters,Copenhagen,Denmark.
Margni, M., Jolliet, O., Rossier, D., Crettaz, P., 2002. Life cycle impact
assessment of pesti-cides on human health and ecosystems. Agric.
Ecosyst. Environ. 93 (1/3), 379–392.
MeteoSwiss, 2000. Klimaatlas der Schweiz. Verlag des Bundesamtes für
Landestopographie, Bern, Switzerland.
Milà i Canals, L., 2003. Contributions to LCA Methodology for Agricultural Systems. Site-dependency and soil degradation impact assessment.
Ph.D. Thesis, p. 251.
Mouron, P., Scholz, R.W., 2005. Income risk management of integrated
apple orchard systems: a full cost analysis of Swiss fruit farms. In:
Mouron, P. (Ed.), Ecological-Economic Life Cycle Management of
Perennial Tree Crop Systems: The Case of Swiss Fruit Farms. Dissertation No. 15899, Swiss Federal Institute of Technology ETH,
Zürich, pp. 23–52, at http://e-collection.ethbib.ethz.ch/.
Mouron, P., Scholz, R.W., Nemecek, T., Weber, O., in press. Life cycle
management on Swiss fruit farms: relating environmental and income
indicators for apple-growing. Ecol. Econ.
Nemecek, T., Frick, C., Dubois, D., Gaillard, G., 2001. Comparing farming
systems at crop rotation level by LCA. In: Geerken, T., Mattson, B.,
Olsson, P., Johansson, E. (Eds.), Proceedings of the International Conference on LCA in Foods, Gothenburg, Gothenburg. SIK, VITO, pp. 65–69.
Nemecek, T., Kufrin, P., Menzi, M., Hebeisen, T., Charles, R., 2002.
Ökobilanzen verschiedener Anbauverfahren wichtiger Ackerkulturen.
114. VDLUFA-Kongress, VDUFA-Schriftenreihe 58, pp. 564–573.
Nemecek, T., 2003. SALCA Templates, Version 1.31. Beschreibung der
Mustersysteme ‘‘SALCA-Betrieb’’ und ‘‘SALCA-Kultur’’. August
2003. Internal Report, Agroscope FAL Reckenholz, Zürich, p. 35.
Nemecek, T., Heil, A., Erzinger, S., Zimmermann, A., 2003. SALCA –
Swiss Agricultural Life Cycle Assessment Database, Umweltinventare
für die Landwirtschaft, Version 031a, May 2003. Report of Agroscope
FAL Reckenholz Zürich and Agroscope FAT Tänikon, Switzerland.
Nemecek, T., Gaillard, G., Zimmermann, A., 2004. Referenzwerte für Ökobilanzen von Landwirtschaftsbetrieben. Agrarforschung 11 (8), 324–329.
Reganold, J.P., Glover, J.D., Andrews, P.K., Hinman, H.R., 2001. Sustainability of three apple production systems. Nature 410, 926–930.
Rossier, D., Gaillard, G., 2004. Ökobilanzierung des Landwirtschaftsbetriebes – Methode und Anwendung in 50 Landwirtschaftsbetrieben.
Agroscope FAL Reckenholz, Zürich, Schriftenreihe der FAL 53, p. 142.
Schaltegger, S., Burrit, R., 2000. Contemporary Environmental Accounting:
Issues, Concepts and Practice. Greenleaf, Sheffield, p. 462.
Schmidt-Bleek, F., 1994. Wieviel Umwelt braucht der Mensch? Birkhäuser,
Berlin, p. 302.
Scholz, R.W., Tietje, O., 2002. Embedded Case Study Methods. Sage
Publications, p. 392.
Scholz, R.W., Wiek, A., 2005. Operational eco-efficiency – comparing
companies’ environmental investments in different domains. J. Ind.
Ecol. 9 (4), 155–170.
Stadig, M., 1997. Life cycle assessment of apple production – case studies
for Sweden, New Zealand and France. SIK Report 630. The Swedish
Inst. For Food and Biotechnology, Gothenburg.
van der Werf, H.M.G., Petit, J., 2002. Evaluation of the environmental
impact of agriculture at the farm level: a comparison and analysis of 12
indicator-based methods. Agric. Ecosyst. Environ. 93, 131–145.
Weizsäcker, E.U., Lovins, A.B., Lovins, L.H., 1997. Faktor vier. Knaur,
Berlin, p. 352.
Werner, F., Scholz, R.W., 2002. Ambiguities in decision-oriented life cycle
inventories: the role of mental models. J. LCA 7 (6), 330–338.
Zürcher, M., Mouron, P., Carint, D., 2003. Erträge und Produktionskosten
im modernen Tafelkernobst-Anbau. Obst- und Weinbau 15, 6–9.