European dietary patterns and their associated land use: variation
between and within countries
Henri de Ruitera,b, Thomas Kastnerb, Sanderine Nonhebela
a
Center for Energy and Environmental Sciences, University of Groningen, P.O. Box 221, 9700 AE,
Groningen, The Netherlands;
b
Institute of Social Ecology Vienna, Alpen-Adria Universität Klagenfurt, Wien, Graz,
Schottenfeldgasse 29, 1070 Vienna, Austria
henri.deruiter@hutton.ac.uk
thomas.kastner@aau.at
s.nonhebel@rug.nl
Corresponding author
Henri de Ruiter
Information and Computing Sciences Group, James Hutton Institute, Craigiebuckler, Aberdeen AB15
8QH, Scotland UK
henri.deruiter@hutton.ac.uk
+44 (0) 844 928 5428
Acknowledgements
TK acknowledges funding from the European Research Council Grant ERC-263522 (LUISE).
1.
Introduction
In the last decade, research on the environmental impact of food consumption patterns has
increased. This is caused by the rising awareness that the consumption of food has major
environmental consequences. For instance, it is estimated that 20-30% of the environmental impacts of
final household consumption is due to our food (Tukker et al. 2011). Land use has been one of the
main indicators of environmental impact, as there is an increasing debate about whether there will be
enough suitable land for the production of food. It is estimated that 38% of total arable land is already
in use for the production of food and it is expected that this will increase in the upcoming decades
(FAO, 2011). Therefore, the relation between food consumption patterns and land use has become
increasingly important as a topic for research. Land use is calculated for past and current situations for
different regions (e.g. Geeraert, 2012; Zhen et al. 2010) and for future projections (e.g. Wirsenius et al.
2010).
There are large differences in the use of land among individual food items (e.g. Gerbens-Leenes et
al. 2002). For example, the production of meat in particular is inherently inefficient compared to plantbased food items, so the role of animal products in the use of land has been a prominent focus of
research (e.g. Baroni et al. 2007; Nijdam et al. 2012). Identifying variations in the land use
requirements of different food items leads to proposals for more sustainable diets.
However, diets of people are more than the sum of individual food items. They are complex
combinations of different food items, influenced by cultural and regional preferences. As a result, it
might prove difficult to exclude individual land-intensive food items from a specific diet. Moreover,
the land use impact of food items depends also on the productivity of a specific agricultural system.
To investigate the influence of dietary patterns on land use, the combination of different dietary
patterns and associated agricultural systems are analyzed here, to investigate sources of variation
between European countries. This analysis can produce insights into the relation between dietary
patterns and land use. The analysis is carried out by calculating the land use of the dietary patterns of
16 European countries. The food consumption data are obtained at the household level and thus
represent a different scale than studies which use FAO data for their analyses. This finer spatial
resolution of data for food consumption also allows the identification of sources of variation within
countries. For the current analysis, education of the household head is investigated as a possible socioeconomic variable associated with variation in land use. Identifying variations in land use of dietary
patterns between, and within, countries leads to a better understanding of the relation between dietary
patterns, the agricultural system and land use.
1.1
Variation in land use between different dietary patterns
There are differences in dietary patterns between and within European countries, resulting mostly
from their different cultural norms (Naska et al. 2006). An early study on land use and dietary patterns,
conducted by Gerbens-Leenes et al. (2002), developed a methodology for the assessment of land
required for food and showed, in combination with household data, the land use for the Dutch
consumption pattern in 1990. A follow-up study assessed the relative land use for 14 European
countries, compared to the Dutch consumption pattern and land use in 1990 (Gerbens-Leenes &
Nonhebel, 2005). This study showed large differences in land use for the existing dietary patterns in
Europe.
The total land use related to food consumption is determined by both consumption (type and
amount of food products consumed) and production (high yields lead to a lower use of land) of food.
The studies conducted by Gerbens-Leenes et al. (2002) and Gerbens-Leenes & Nonhebel (2005) used
one production system (the Dutch system) for the calculation of the use of land; they were designed to
show the relevance of specific dietary patterns for the use of land. A recent advancement in the
methodology made it possible to also include other national production systems. Kastner et al. (2012)
use an updated methodology and FAO data from 1961-2007 for a consistent comparison of land use in
different subcontinents at different points in time. They show that variations in per-capita food supply
are large and linked to income levels, especially qualitatively. On the other hand, variations in land use
were less pronounced, and not clearly linked to income levels. This latter finding can be explained by
large differences in output per unit land (production system). This emphasizes the importance of the
production system in analyses about land use. We use the methodology from Kastner et al. (2012) to
investigate whether differences in agricultural systems are a possible source of variation for the land
use of European dietary patterns.
A challenging issue in comparative studies on land use for food is the difference in collection
methodology for food consumption data, since different definitions and methods exist for the
recording of food consumption. To compensate for this, most comparative studies use food supply
data obtained from the FAO food balance sheets (FBS). Because these data are harmonized as much as
possible, using FAO data for inter-country comparisons is a common practice. Nevertheless, these data
represent the national supply level, and information about consumption within the household is lost.
On the other hand, analyses from single-country studies which use household data cannot be used for
comparisons with other countries, as the methodology for the collection of household food data differs
substantially between countries. Hence they cannot be used for the purpose of the current paper.
In this paper, we use a standardized and post-harmonized databank, developed in the Data Food
Networking Project (DAFNE). This dataset allows the assessment of land use associated with dietary
patterns on the household level, without risking flaws because of differences in the collection methods.
Minimizing the effects of differences in collection methods is essential for the purpose of our analysis,
to exclude methodological differences as a source of variation. The DAFNE project exploits food and
socio-demographic data collected in the household budget surveys (HBS), aiming at the development
of a cost-effective food databank that allows monitoring of food availability both within and between
European populations (Trichopoulou et al. 2003). Currently, 24 European countries are included in the
database (DAFNE, 2005). The initiative is intended to be used as a cost-effective nutrition monitoring
tool in a European context. The databank allows for comparison between countries and within
countries. This latter fact makes it also possible to investigate the influences of socio-economic
variables on land use. Since it is known that dietary patterns are also influenced by socio-economic
factors, this represents a major advantage compared to the use of FAO national supply data. For
instance, Hulshof et al. (2003) show that people with a high socio-economic status (SES) in The
Netherlands consume more vegetables, cheese and alcohol, whereas people with a low SES consume
more potatoes, meat and meat products, visible fats and coffee. Education is studied as a possible
source of variation in this paper, since epidemiological studies show that it represents the strongest and
most important social predictor in Western countries for a healthy diet (Johansson et al. 1999; Roos et
al. 1996). People with a higher level of education generally have a healthier diet, characterized by a
higher consumption of fruit and vegetables. For that reason, education is chosen as most likely source
of variation for dietary patterns within European countries.
Thus, this analysis uses the method by Kastner et al. (2012) and data obtained from the DAFNE
databank to assess the land use related to the dietary patterns in Europe. We are able to incorporate
both the influence of differences in consumption patterns as well as differences in the agricultural
systems on total land use. The paper assesses both the cropland use of the European dietary patterns,
as well as the effect of education on this land use. In doing so, this paper contributes to the body of
knowledge on the relation between dietary patterns and land use. This is accomplished by using food
consumption data on the household level instead of using data on a national supply level.
2.
Methodology
2.1
Collection of household budget survey data
The used food consumption data are freely available at the Dafnesoft website (DAFNE, 2005).
The methodology for the collection of these data consists of several steps and is described in detail by
Trichopoulou et al. (2003). A brief overview is given in this section. First, the data are collected from
several European countries through household budget surveys (HBS). These surveys collect data on
food availability at the household level. Hereafter, the raw HBS data are incorporated in a central
database. The next step involves the post-harmonization of the food, demographic and socio-economic
information. The final step comprises the calculation of the average, daily individual food availability.
The average daily individual food availability is expressed in grams per day. The data obtained from
this databank can be used for comparisons between and within countries.
For comparisons within countries, food availability in the DAFNE databank can be obtained for
different socioeconomic variables: education, occupation, locality and household composition.
Education is chosen as socioeconomic indicator in the current study. The variable differentiates three
categories in the data bank: ‘primary education / illiterate’, ‘secondary education’ and ‘higher
education (college / university)’. Though differences in the national educational systems do exist, data
on the education of the head of the household could be classified under these three comparable
categories (Trichopoulou, 2008). The three categories are described in this paper as ‘elementary
education’, ‘secondary education’ and ‘higher education’. The food availability data is grouped
according to the highest education of the head of the household.
Food availability data for 24 European countries are available, but the data for the United
Kingdom do not include the indicator education, and the UK is therefore excluded. For some
countries, surveys from several years are available; the surveys closest to the year 2000 are chosen.
The included surveys range from 1988 to 2000. Table A.1 lists the included surveys and the sample
information. Germany only has data available for the two highest educational levels: secondary and
higher education. Luxembourg is excluded because of its small size and because yield data are only
available in an aggregate with Belgium.
2.2
Methodological issues about household food consumption
In the HBS and in this paper, household food consumption is conceived as household food
availability, as opposed to household food intake. Food availability is a useful measure for our
analysis, as the availability of food represents a larger amount of food than actual intake (including
intake and household level waste). Consequently, availability of food has a higher environmental
impact than food intake. Moreover, to our knowledge, there is no standardized and harmonized
databank for food intake. Household food availability can be considered as ‘all food that enters the
household’. For instance, food can also be consumed outside the household. The DAFNE initiative
also focuses on the consumption outside the house. However, the majority of the European countries
included in the DAFNE data list only expenses concerning out-of-home consumption (Trichopoulou et
al. 2005). This is less accurate compared to quantity data. Another issue is whether meals of guests are
included in the household consumption and whether food produced at home is included in the
household food availability. Table A.1 lists the information about the inclusion of guest meals and the
capturing of the seasonal variability. Table A.1 also presents the HBS included in this study and their
main sample information. More information about the samples is available on the website (DAFNE,
2005). Waste throughout the food chain is not considered in the DAFNE dataset, as the collection
method is aimed at the consumption of food within the household.
2.3
Collection of land use data
Country-specific data for the land use for food are obtained from Kastner et al. (2012). The exact
calculation of these land uses is explained in Kastner et al. (2012) and briefly described here. The land
use for specific food items depends on the consumption of these items (information obtained from
FBS) and on the yields of the corresponding crops (obtained from the FAO). Kastner et al. (2012) use
country-specific yields to assess cropland use in a given nation. However, international trade poses a
limit to this approach: food consumed in one country can originate from land within the country or
from imports from outside. To account for this, Kastner et al. (2012) split the amount of a crop
available within a country into shares of domestic production and imports. The land use for the
domestic production is calculated using the domestic yields. For the import share, the authors
performed a calculation to assess the average world market yield. The total land use is computed by
adding up the domestic production and import land requirements. The calculated land use involves the
demand for cropland, grassland and pastures are not included.
Kastner et al. (2012) use the assumption that a calorie from animal origin needs three times more
land than a ‘vegetal’ calorie (non-animal calorie). This assumption is based on a calculation comparing
total feed input from cropland into the livestock sector to total livestock output. We were able to
combine land use data and food consumption data for 16 of the 23 countries present on the DAFNE
website.
2.4
Converting HBS data to land use data
Figure 1 shows the steps used in this study. DAFNE data are obtained at the highest aggregation
level, comprising 22 food categories (g/day/cap). Two food categories are investigated at a lower
level: non-alcoholic beverages and total added lipids. The first category is examined at a lower level to
exclude the availability data of mineral water, as no cropland is required for mineral water. The latter
category is obtained at a lower level to discriminate between lipids from vegetal and animal origin.
The land use data obtained from Kastner et al. (2012) are for the agricultural commodities used in
the FAO food balance sheets. The food consumption data, however, are for food categories and not
expressed in agricultural commodities. This requires an extra step to convert food consumption data to
agricultural commodities, using equivalent factors. These equivalent factors are listed in table A.2. To
give an example, one DAFNE category is ‘bread and rolls’ and it is assumed that 0.67 of this category
consists of cereals. As table A.2 shows, a DAFNE category is only allocated according to the main
FAO category. That is, the other 0.33 of the ‘bread and rolls’ category is not allocated to another FAO
category. This is done because it is assumed that the other part consists mainly of water (this is the
case for ‘Bakery products’; Bread and rolls’; and ‘Soft drinks’). Cheese is converted to milk
equivalents, using the FAO extraction rate of 0.12 (FAO, 2001a).
The next step in the calculation involves the conversion of quantities (in kg) to calories. This is
necessary because the land use data are expressed in m2/kcal. FAO conversion factors are used for this
step (FAO, 2001b). The conversion factors are on a lower aggregation level than the FAO categories.
That is, conversion factors are available for food items (e.g. apples), but not for food categories (e.g.
fruits). Therefore, assumptions are made to obtain a category conversion factor (see table A.3).
The final step is the calculation of calories to land use. The land use data are obtained for each
country from Kastner et al. (2012). These values are used to calculate the land use per kcal for each
food category. By dividing the land use calculated by Kastner et al. with the caloric supply (obtained
from the FBS), land use (m2/kcal) for each food category is obtained. For instance, the land use
calculated for cereals in Austria (1999) equals 242 m2 per year and per capita. The amount of calories
supplied by 242 m2 equals 905 kcal/day/capita. This implies that 0.27 m2/capita was needed in 1999 to
provide a daily kilocalorie year-round. This value is multiplied with the caloric availability (DAFNE)
used in this study. The land use (m2/kcal) differs per country and per year, as yields differ per country
and per year. Appendix B lists the land use (m2/kcal) for each category in each year.
3.
Results
3.1.
Variation in land use between different European dietary patterns
Figure 2 presents the land use for the 16 European dietary patterns, broken down by educational
level. Large differences are seen between the countries. Malta has the largest land use (3,227 m2 for
the elementary education group), while Ireland has the lowest land use (1,476 m2 for the elementary
education group); the difference between the highest and lowest land use is thus a factor 2.2. In
general, countries in Southern Europe have a much higher land use than countries in other parts of
Europe. A major reason, not related to dietary patterns, is the production system of the European
countries. To give an illustration, the agricultural system in Germany needed 0.22 m2 in 1998 to
produce one daily kilocalorie of cereals year-round. In contrast, the agricultural system of Spain
needed 0.54 m2 to produce this daily kilocalorie in 1998. This difference is caused by the higher yields
in Germany. Therefore, although the caloric availability in Germany is higher than in Spain, the total
land use of the German diet is lower than for Spain. This emphasizes that the influence of the local
agricultural system is very significant.
Another major reason for the differences in land use between European countries is the total
caloric food availability. Generally, countries in Southern Europe have a large caloric availability
(though the Iberian Peninsula seems to be an exception). Malta and Greece have, because of their high
caloric availability and because of their production system, the highest land use.
To exclude the effect of the total caloric availability on the calculated land use, the land uses of the
16 European dietary patterns are calculated to a 2,500 kcal diet (figure 3). Two scenarios are presented
in figure 3, with the Austrian situation as reference (=100%). The first scenario is the calculation of
the land use for a 2,500 kcal diet with country-specific yields. This scenario still includes the effects of
the country-specific agricultural productivities and different composition of the diets. Countries with a
percentage higher than 100% have a lower agricultural productivity compared to Austria. For instance,
Spain and Portugal have an agricultural system with comparatively low land productivity.
To further isolate the effect of the composition of the diet, the relative land use is calculated with
one agricultural system and associated import ratio (Austrian system). This second scenario is also
presented in figure 3 and shows the relative land use compared to the Austrian land use. Finland and
France have the highest land use because of their dietary composition. This is caused by the high
proportion of animal products in their diet (46% of total caloric availability). On the other hand,
Hungary has the lowest relative land use. This is caused by the high proportion of cereals and low
proportion of animal products in the diet.
Overall, figure 3 shows that the relative land use is more affected by the differences in agricultural
productivities than by differences in the composition of the diets. It can therefore be concluded that
within Europe, the influence of the agricultural system is more pronounced than the effect of different
compositions of the diets.
3.2
Contribution of different food categories to total land use
Animal products contribute very substantially to the land use in every European country. On
average, the share of animal products in total land use is more than 60%. France has the highest share
of animal products in the total land use (more than 70%), while Italy has the lowest share (below
60%).
Vegetable oils have a considerable share in the Southern countries. Vegetable oils represent the
category with the most visible differences between the regions. Whereas the share of vegetable oils in
the total land use is larger than 12% in Southern Europe, this share is around 7% in the other countries.
Less visible differences are present for sugar & sweeteners (Southern countries consume less sugar,
but as the land use for sugar is small, this does not contribute much to land use differences) and
cereals. Eastern countries especially, such as Hungary and Poland, have a larger share of cereals in the
total use of land. However, we have to take into account a possible time effect, as the year of the
survey is ± 10 years earlier than most other surveys (1991 and 1988, respectively). It could be that the
high cereals and low animal products share in these countries is a reflection of a European
consumption pattern from a decade before the year 2000.
Figure 2 also shows the caloric availability of the different food categories. It shows that cereals in
general have a large share in the caloric availability, but that this share is much smaller in the land use
picture. This is caused by the high efficiency of cereals production. The opposite is true for animal
products. Animal products have a larger share in the total land use than in the caloric availability. This
emphasizes the inefficiency of animal products.
3.3
Variation within countries: the influence of educational level
Figure 2 clearly shows that people with an elementary education have higher caloric food
availability and consequently a larger land use. This is true for all countries, except for Portugal. The
difference between people with a secondary education and people with a higher education is less
marked. As the production system is the same within countries, the differences can be attributed to
dietary factors. Two dietary factors influence total land use: total caloric availability and composition
of the diet. The caloric availability is substantially higher for the elementary education group. On
average, people with an elementary education consume 2,809 kcal per day, while the other groups
consume 2,380 kcal and 2,302 kcal, respectively. However, when the caloric shares are examined, few
differences are observed. People with an elementary education seem to consume more vegetable oil
and less animal products. This is caused by the lower milk (and milk products) consumption of the
elementary education group. The caloric shares indicate that people with a higher education consume
more fruits and less starchy roots. There is no observable difference for the cereals category.
To separate the effect of caloric availability and composition of the diet, we calculated the relative
land use for the different education groups, based on a 2,500 kcal diet. Across all the countries, the
relative cropland use (for a 2,500 kcal diet) is on average 2,096 m2 for the elementary education group;
2,116 m2 for the secondary education group; and 2,141 m2 for the higher education group. The
variation for individual countries is large, as both caloric availability and yields differ, but it could hint
at a trend, namely, that people with a higher education consume more ‘costly’ calories (in terms of
land use per kcal). This is demonstrated by the fact that 14 out of 16 countries (Germany is excluded
because it has no ‘elementary education’ category) have a higher relative land use for the higher
education group compared to the elementary education group. Although the differences are not very
large (the highest difference is 9%), it is a consistent finding in the current analysis. This could be
explained by the higher share of animal products (i.e. milk) in the total land use for the higher
education group.
4.
Discussion
In the past decades, the analysis of dietary patterns, in addition to studying individual dietary
components has emerged as a method of exploring the relations between diet and disease (Hu, 2004).
This shifting focus is caused by the increasing recognition that the intake of various dietary factors is
highly correlated, and many interactions between components of a diet and disease risk may exist. The
same development has taken place for the relation between dietary patterns and land use in the past
decade: shifting away from individual food items to the analysis of complete dietary patterns. The
analysis presented in this paper is based on one point in time and represents a ‘snap-shot’ of the
situation around the year 2000. Using this snap-shot, sources of variation can be observed and
investigated. The objective of this paper is therefore not to obtain a perfect, quantitative picture of the
cropland required for food; rather, the focus is on sources of variation and relative differences between
the land use related to the European dietary patterns.
4.1
Variation in land use between European dietary patterns
The variation in land use related to European dietary patterns is large, with a factor 2.2 difference
between the lowest land use and the highest land use between countries. The current analysis shows
that two factors are responsible for the main differences in the land use of the European dietary
patterns: caloric availability and the productivity of the agricultural system. The influence of
differences in the composition of the diet between countries is minor, suggesting that the dietary
patterns in Europe are quite similar and that, on this level of aggregation, the differences between
countries do not lead to large variation in land use. It is not suggested that the composition of the diet
has no effect at all on the land use for food. A radical switch to a vegetarian diet would drastically
decrease the land use. Moreover, Kastner et al. (2012), show that dietary change is an important
contributor to the impact of diets worldwide, especially in (Eastern) Asia. The current analysis does
not discriminate between the land use of, for instance, different livestock products. Bovine meat needs
in general more land than poultry or pork meat, however much of this difference is explained by the
need for grazing lands which were not part of our analysis. As the total land use is greatly influenced
by consumption of animal products, the differences in land use of specific animal products may
increase the differences between the land use of different dietary patterns. On the other hand,
increasing the level of detail would add another level of complexity, which would lead to more
unwanted variation caused by methodological differences. For the resolution of dietary patterns
considered in our analysis, the composition of the diet is not a major source of variation between
national European dietary patterns.
Gerbens-Leenes & Nonhebel (2005) also calculate the relative differences for the land use related
to food of different European countries. In their analysis, the difference between the highest relative
land use and the lowest relative land use is 37%. In the current analysis, the difference in relative land
use for a 2,500 kcal diet is 25% (figure 3). Without this caloric standardization, the difference
increases to 56%. This suggests that, although the study by Gerbens-Leenes & Nonhebel (2005) is on
a national food supply level, the relative differences in land use are of the same order of magnitude,
although it might also suggest that differences between countries are higher when using household
data for the calculation of land use related to food consumption. When the individual countries are
examined, France has the highest land use in both analyses, but the order in the countries differs
substantially between the two analyses. One of the reasons is that Gerbens-Leenes & Nonhebel (2005)
do not correct for the caloric availability. If figure 3 is constructed without correction for caloric
availability, the order of the countries is more similar to Gerbens-Leenes & Nonhebel (2005).
If we compare the land use of the current analysis to the land use calculated with FAO data by
Kastner et al. (2012), the FAO land use calculations are higher than the calculations in this paper. FAO
calculations for the year 2000 show an average land use of 2,632 m2 for Eastern Europe (1,750 m2 for
secondary education group in this paper); 1,909 m2 for Northern Europe (1,834 m2); 3,108 m2 for
Southern Europe (2,546 m2); and 2,024 m2 for Western Europe (1615 m2). Although the regions in the
Kastner et al. (2012) study involve more countries, this comparison suggests that land use calculated
with FAO data leads to higher estimates of land use due to the dietary patterns in Europe. This is in
line with other studies which show that FAO data are consistently higher for food consumption than
HBS data (Elmadfa, 2009; Naska et al. 2009; Pomerleau et al. 2003; Serra-Majem et al. 2003).
Therefore, the analysis of land use for food on the household levels is a valuable addition to existing
studies on a national food supply level. After all, household food consumption data is a more specific
reflection of the actual per capita food consumption than FAO data. Food supply on the national level
can display large national trends in the supply of food per capita, whereas household food availability
is better able to show the actual dietary patterns in a country. The assessment of land use related to
food consumption on a household level is, therefore, more specific than on the national supply level,
and represents a further advancement in the assessment of the effect of the dietary patterns on land
use.
The single largest contributor to the total land use is the consumption of animal products. In each
country, the share of animal products in the total land use is higher than 50%. As the total share of
animal products in the caloric availability is in the order of 30-45%, this emphasizes the high land use
of animal products. Vegetable oils are consumed in larger quantities in the Southern countries, while
the consumption of sugar & sweeteners is lower in these countries.
The differences in the caloric shares of the food categories are relatively small and have a minor
effect on the total land use. Despite this small effect, the differences in land use related to the dietary
patterns are prominent. This is caused by the different agricultural systems. A highly productive
agricultural system, combined with suitable local circumstances, leads to high yields and thus to low
land use for food. Especially the countries in Southern Europe have lower productive agricultural
systems. The more arid climate of these countries is one of the obvious reasons for the low
productivities of these countries. Naska et al. (2006) suggest that the dietary patterns of Europe are
narrowing towards a more uniform pattern. However, in an environmental context, it is likely that the
differences in land use will continue to exist, as local climate circumstances will be playing a major
role in the productivities of the agricultural systems.
4.2
Variation in land use within countries: influence of educational level
This paper shows that people with an elementary education have a larger land use than people with
a secondary or higher education. The main reason for this result is that people with a lower education
have a higher caloric availability. The differences in the composition of the diets have a very small
effect and suggest that people with a higher education consume more ‘costly’ calories. This effect
slightly counteracts the larger land use of the lower education group. As a result, it can be concluded
that the largest part of the variation in land use within countries is caused by the caloric availability.
People with a lower education could consume more food, but they could also waste more food.
Results from the literature for the relation between education (or SES) and caloric intake are
mixed. Hulshof et al. (2003) show a significant effect for men between a ‘high’ and ‘low’
socioeconomic status (high SES is associated with lower caloric availability) in The Netherlands; no
effect was found for women. A review by Darmon & Drewnowski (2008) shows inconsistent results
for this correlation. A systematic review conducted by Giskes et al. (2009) for European countries,
shows the same result, no consistent association between socio-economic status and energy intakes,
despite the fact that the prevalence of obesity is higher among socioeconomic disadvantaged groups.
This is not the case for the current results; almost all countries show a higher caloric availability for
the elementary education group. One of the reasons could be that in the current study, education was
the sociological factor, while other definitions of SES sometimes include occupation, income or other
factors.
Education is chosen in this paper as the major socio-economic variable, since research indicates
that education is the most consistent predictor of a healthy diet. Common indicators for a healthy diet
consist of specific measures, such as the amount of fruits and vegetables, fibre, fat as percentage of
total energy intake, (e.g. Johansson et al. 1999). It may be that education is a good predictor for a
healthy diet, but not as an indicator for an environmental-friendly diet. For instance, fruits and
vegetables barely show up in the current land use pictures, as their caloric content is very low.
Therefore, choosing other socio-economic variables as sources of variation may lead to other results.
We used the same level of aggregation for food consumption within countries. Thus, we did not
discriminate between different animal products. Although we did not study this in detail, it seems that
especially higher educated people consume relatively less pork (DAFNE, 2005). An interesting
observation is that Southern countries have a diet characterized by the high amounts of vegetable oils,
especially olive oil. Olive oil is generally considered as part of a healthy diet. In the current analysis,
the high availability of vegetable oils contributes to the higher land use for the Southern countries. The
other European countries have considerably higher caloric shares of sugar & sweeteners, which are not
necessarily part of a healthy diet. However, the high share of sugar does not translate in a large use of
land, as sugar is very efficiently produced. Therefore, there might be trade-offs between some food
categories in obtaining both a healthy and environmental-friendly diet.
4.3
Conclusions and implications
The calculated cropland use related to European dietary patterns ranges from ± 1,500 m2 per capita
to ± 3,000 m2 per capita. This paper shows that agricultural systems and total caloric availabilities are
the largest sources of variation in the land use of European dietary patterns. The composition of the
diet between countries is a much smaller source of variation, at least at this level of aggregation. Using
educational level as a proxy for differences within countries, it is shown that caloric availability is the
major source of variation in land use within countries and that people with low education need more
land to sustain their dietary patterns.
Using household food consumption data leads to lower land use estimates associated with dietary
patterns than FAO data. As household food consumption data are a better reflection of dietary
patterns, the use of these data is a valuable addition to FAO data analyses.
The current analysis emphasizes the relevance of an efficient agricultural system and the total
availability of calories. Moreover, our results suggest that people with a higher education consume less
calories, and hence have a lower demand for land.
References
Baroni, L., Cenci, L., Tettamanti, M., Berati, M., 2007. Evaluating the environmental impact of
various dietary patterns combined with different food production systems. Eur J Clin Nutr, 61,
279-286
DAFNE, D. F. N. 2005. The pan-european food data bank based on household budget surveys.
Retrieved 08/03, 2012, from http://www.nut.uoa.gr/Dafnesoftweb/
DAFNE IV., 2003. European Food Availability Databank based on Household Budget Surveys –
Annex 3: Comparable categories of socio-demographic data. Retrieved 08/23, 2012, from
http://ec.europa.eu/health/ph_projects/2002/monitoring/monitoring_2002_04_en.htm
Darmon, N., & Drewnowski, A., 2008. Does social class predict diet quality? Am J Clin Nutri., 87(5),
1107-1117.
Drewnowski, A., & Specter, S. E., 2004. Poverty and obesity: The role of energy density and energy
costs.. Am J Clin Nutri., 79(1), 6-16.
Elmadfa et al. (2009). European nutrition and health report 2009. Basel: European Commission,
Health and Consumer Protection.
FAO., 2001a. III. Procedures for Preparation of Food Balance Sheets – 3. Livestock Sector. In: Food
balance sheets - A handbook –Rome: Food and Agricultural Organization of the United Nations.
FAO., 2001b. Food balance sheets - A handbook - annex I. Rome: Food and Agricultural Organization
of the United Nations.
FAO., 2011. The state of the world's land and water resources for food and agriculture (SOLAW) managing systems at risk. Rome; London: Food and Agricultural Organization of the United
Nations; Earthscan.
Geeraert, F., 2012. Sustainability and dietary change: the implications of Swedish food consumption
patterns 1960-2006. Int J Consum Stud
Gerbens-Leenes, P. W., Nonhebel, S., & Ivens, W. P. M. F., 2002. A method to determine land
requirements relating to food consumption patterns. Agri.,, Ecosyst. and Environ., 90(1), 47-58.
Gerbens-Leenes & Nonhebel, S., 2005. Food and land use. The influence of consumption patterns on
the use of agricultural resources. Appetite, 45: 24-31
Giskes, K., Avendano, M., Brug, J., Kunst, E.A., 2009. A systematic review of studies on
socioeconomic inequalities in dietary intakes associated with weight gain and overweight/obesity
conducted among European adults. Obes Rev., 11: 413-429
Hu, F.B., 2002. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin
Lipidol 13, 3–9
Hulshof, K. F. A. M., Brussaard, J. H., Kruizinga, A. G., Telman, J., & Löwik, H., 2003. Socioeconomic status, dietary intake and 10 y trends: The dutch national food consumption survey.
Eur J Clin Nutr, 57, 128-137.
Irala-Estévez, J. D., Groth, M., Johansson, L., Oltersdorf, U., Prättälä, R., & Martínez-González, M.
A., 2000. A systematic review of socio-economic differences in food habits in europe:
Consumption of fruit and vegetables. Eur J Clin Nutr, 54(9), 706-714.
Johansson, L., Thelle, D.S., Solvoll, K., Bjorneboe, G.A., Drevon, C.A., 1999. Healthy dietary habits
in relation to social determinants and lifestyle factors. Br J Nutr, 81, 211-220.
Kastner, T., Rivas, M. J. I., Koch, W., & Nonhebel, S., 2012. Global changes in diets and the
consequences for land requirements for food. PNAS, 109(18), 6868-6872.
Naska, A., Fouskasis, D., Oikonomou, E., Almeida, M.D.V., Berg, M.A., Gedrich, K., Moreiras, O.,
Nelson, M., Trygg, K., Turrini, A., Remaut, A.M., Volatier, J.L., Trichopoulou, A., DAFNE
participants., 2006. Dietary patterns and their socio-demographic determinants in 10 European
countries: data from the DAFNE databank. Eur J Clin Nutr, 60: 181-190.
Naska, A., Berg, M. A., Cuadrado, C., Freisling, H., Gedrich, K., Gregorič, M., et al., 2009. Food
balance sheet and household budget survey dietary data and mortality patterns in Europe. Br J
Nutr, 102, 166-171.
Nijdam, D., Rood, T., Westhoek, H., 2012. The price of protein: Review of land use and carbon
footprints from life cycle assessments of animal food products and their substitutes. Food Policy,
37(6): 760-770.
Pomerleau J., Lock, K., McKee, M., 2003. Discrepancies between ecological and individual data on
fruit and vegetable consumption in fifteen countries. Br J Nutr, 89(6): 827-834.
Roos, E., Prättaäla, R., Lahelma, E., Kleemola, P. & Pietinen P., 1996. Modern and healthy?:
socioeconomic differences in the quality of diet. Eur J Clin Nutr 50, 753–760.
Serra-Majem, L., MacLean, D., Ribas, L., Brulé, D., Sekula, W., Prattala, R., et al., 2003. Comparative
analysis of nutrition data from national, household, and individual levels: Results from a WHOCINDI collaborative project in canada, finland, poland, and spain. Journal of Epidemiology and
Community Health, 57(1), 74-80.
Trichopoulou, A., & Naska, A., 2003. European food availability databank based on household budget
surveys. Eur J Pub Health, 13(3 Supplement), 24-28.
Trichopoulou, A., Naska, A., & Oikonomou, E., 2005. The DAFNE databank: The past and future of
monitoring the dietary habits of europeans. J Publ Health, 13(2), 69-72.
Trichopoulou A., 2008. Final Implementation Report of the DAFNE V Project. Agreement No
SPC.2003117.
Tukker, A., Goldbohm, R.A., de Koning, A., Verheijden, M., Kleijn, R., Wolf, O., Pérez-Domínquez,
I., Rueda-Cantuche, J.M., 2011. Environmental impacts of changes to healthier diets in Europe.
Ecol Ec, 70: 1776-1788.
Zhen, L., Cao, S., Cheng, S., Xie, G., Wei, Y., Liu, X., Li, F., 2010. Arable land requirements based
on food consumption patterns in rural Guyuan District, Western China. Ecol Ec, 69(7): 14431453.
Appendix A
-
Equivalence factors and Caloric conversion factors
Country
Year
Sample size
Austria
Belgium
Cyprus
Finland
France
Germany
Greece
Hungary
Ireland
Italy
Malta
Norway
Poland
Portugal
Spain
Sweden
1999
1999
1996-1997
1998
1991
1998
1998
1991
1999
1996
2000
1996-1998
1988
2000
1998-1999
1996
7,098
3,745
3,308
4,359
12,680
6,258
11,813
7,644
22,740
2,586
22,740
29,664
10,024
14,644
2,026
1
Meals guest
included
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
No
No
No
Yes
Yes
Yes
Yes
Recording period
14 days
1 month; 3 month for bulk1
14 days
14 days
7 days
One quarter
14 days
2 months
14 days
10 days
21 days
10 days
3 months
14 days
7 days
14 days
1-month intensive recording + 3 months retrospective recording for bulk expenses
Table A.1. Survey characteristics (DAFNE, 2005).
DAFNE Levels
FAO Category
Equivalence factor
Bakery products
Cereals
0.67
Beverages, Alcoholic
Alcoholic Beverages
1
Bread and Rolls
Cereals
0.67
Cereals and Products
Cereals
1
Cheese
Milk
8.33
Eggs
Animal products
1
Fish and Seafood
Animal products
1
Flour
Cereals
1
Fruits
Fruits
1
Juices
Fruits
1
Meat and products
Animal products
1
Milk
Milk
1
Milk products
Milk
1
Nuts
Oil crops
1
Pasta
Cereals
1
Potatoes
Starchy Roots
1
Pulses
Pulses
1
Soft drinks
Sugar & Sweeteners
0.1
Stimulants
Stimulants
1
Total added lipids animal
Animal products
1
Total added lipids vegetal
Vegetable Oils
1
Vegetables
Vegetables
1
Table A.2. DAFNE levels (DAFNE, 2005), corresponding FAO categories and equivalence factors (FAO,
2001a).
FAO Category
Cereals
Animal products
Milk
Vegetable oils
Fruits
Oilcrops
Vegetables
Pulses
Starchy roots
Sugar & sweeteners
Stimulants
Alcoholic beverages
Caloric value (kcal/kg)
3340
2480
610
8840
510
5800
270
3410
670
3870
480
585
FAO conversion factor used
Wheat
⅓ Beef preparations + ⅓ Pig meat + ⅓ Poultry meat
Cow milk, whole fresh
Olive Oil
⅓ Bananas + ⅓ Apples + ⅓ Fruit, nes, fresh
Groundnuts prepared
⅓ Onion + ⅓ Tomato + ⅓ Carrot
Beans, dry
Potatoes
Sugar Refined
½ Tea + ½ Coffee roasted
½ Beer Barley + ½ Wine
Table A.3. Caloric conversion factors used throughout the study. Based on FAO (2001b).
Appendix B
-
Relative cropland use
The data are based on the cropland use calculated by Kastner et al. (2012) and FAO consumption data (in
kilocalories). The calculation is performed by dividing the total land use with the total caloric supply according
to the FAO. This is done for each of the FAO food categories. The results differ per country and are shown in the
tables. The year used for the calculation is the same year as the survey data (see table 2 in the main article).
Category
Austria
Belgium
Cyprus
Finland
France
Germany
Greece
Hungary
Alc. Bev.
Cereals
Fruits
Oil crops
Pulses
St. Roots
Stimulants
Sugar &
Sweeteners
Veg. Oils
Vegetables
Animal
Products
0.36
0.27
0.46
0.65
0.57
0.17
8.77
0.20
0.38
0.40
0.56
0.72
0.63
0.14
10.27
0.21
0.40
0.38
0.45
0.71
0.55
0.13
8.66
0.24
0.41
0.50
0.59
0.66
1.08
0.25
10.20
0.20
0.73
0.22
0.45
0.76
0.62
0.18
7.34
0.22
0.28
0.22
0.55
0.52
0.37
0.14
3.39
0.20
0.39
0.56
0.81
1.55
0.69
0.25
4.79
0.20
0.39
0.29
0.72
0.50
0.53
0.37
8.33
0.22
0.44
0.58
1.19
0.51
0.47
1.23
0.49
0.44
1.28
0.57
0.78
1.61
0.46
0.89
1.32
0.43
0.72
1.09
0.70
0.65
1.78
0.54
1.07
1.35
Category
Ireland
Italy
Malta
Norway
Poland
Portugal
Spain
Sweden
Alc. Bev.
Cereals
Fruits
Oil crops
Pulses
St. Roots
Stimulants
Sugar &
Sweeteners
Veg. Oils
Vegetables
Animal
Products
0.20
0.32
0.69
0.64
0.35
0.19
3.21
0.21
0.56
0.45
0.95
1.19
0.53
0.23
10.59
0.22
0.29
0.48
0.68
0.95
0.41
0.27
1.40
0.20
0.38
0.37
0.60
0.99
0.58
0.20
5.11
0.20
0.27
0.52
0.84
0.76
0.52
0.29
8.31
0.21
0.62
0.52
1.10
1.52
0.90
0.35
2.72
0.22
0.75
0.54
0.63
1.28
1.64
0.21
7.60
0.20
0.32
0.26
0.73
0.69
0.44
0.17
10.83
0.21
0.48
0.59
1.04
0.73
0.67
1.71
0.61
0.88
1.45
0.51
0.80
1.37
0.40
0.74
1.31
1.06
0.75
1.97
1.04
0.65
2.16
0.43
0.66
1.24
Table B.1 Cropland use (m2/year) to provide 1 kcal/day. Relative cropland use for different food
categories and countries.
1. Survey-specific quantity data
aggregated into FAO categories
2. Quantities converted to calories
using FAO conversion factors
3. Calories calculated to cropland
use using country-specific data
(Kastner et al. (2012))
Figure 1. Scheme showing the
calculation procedure for this study
Elementary Secondary
4000
3500
3000
2500
2000
1500
1000
500
0
Higher
Elementary
Secondary
Higher
Malta
Elementary Secondary
4000
3500
3000
2500
2000
1500
1000
500
0
4000
3500
3000
2500
2000
1500
1000
500
0
4000
3500
3000
2500
2000
1500
1000
500
0
4000
3500
3000
2500
2000
1500
1000
500
0
4000
3500
3000
2500
2000
1500
1000
500
0
Higher
Caloric availability (kcal/day)
Calories
Caloric availability (kcal/day)
Calories
Land
4000
3500
3000
2500
2000
1500
1000
500
0
Caloric availability (kcal/day)
Calories
Land
Calories
Land
Calories
Land
Calories
4000
3500
3000
2500
2000
1500
1000
500
0
Calroic availability (kcal/day)
Elementary Secondary
Calories
Elementary Secondary
Land
Elementary Secondary
Land
Hungary
Calories
Land
Calories
France
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Cropland use (m2/year)
Belgium
Calories
Land
Higher
Calories
Higher
Land
Calories
Land
Calories
Higher
Land
Secondary
Land
Secondary
Calories
Italy
Secondary
Land
Elementary
Calories
Greece
Calories
Elementary
Land
Calories
Finland
Land
Calories
Land
Calories
Elementary
Land
Higher
Calories
Higher
Land
Calories
Land
Calories
Land
Calories
Land
Cropland use (m2/year)
Higher
Land
Elementary Secondary
Calories
Elementary Secondary
Land
Calories
Land
Calories
Land
Cropland use (m2/year)
Elementary Secondary
Land
Calories
Land
Calories
Land
Cropland use (m2/year)
Austria
Cyprus
Higher
Germany
Higher
Ireland
Higher
Norway
Higher
Elementary
Sweden
Elementary Secondary
Higher
Caloric availability (kcal/day)
Calories
Land
Calories
Land
4000
3500
3000
2500
2000
1500
1000
500
0
Calories
Higher
Elementary Secondary
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Calories
Land
Secondary
Higher
Alc. Beverages + Stimulants
4000
3500
3000
2500
2000
1500
1000
500
0
Land
Calories
Land
Calories
Land
Calories
Calories
Land
4000
3500
3000
2500
2000
1500
1000
500
0
Elementary Secondary
Cropland use (m2/year)
Spain
Sugar & sweeteners
Vegetable oils
Pulses + Starchy Roots
Fruits & Vegetables
Animal products
Cereals
Oilcrops
Figure 2. Cropland use associated with dietary patterns of European countries, for different education groups.
Caloric availability (kcal/day)
Portugal
4000
3500
3000
2500
2000
1500
1000
500
0
Land
Cropland use (m2/year)
Poland
Relative cropland use (Austria = 100%)
Cropland use for 2,500 kcal diet
250%
200%
150%
100%
50%
0%
Country-specific productivities
Austrian productivities
Figure 3. Relative cropland use for a diet with 2,500 kcal, calculated for the secondary education groups. The Austrian
situation is the reference situation. ‘Country-specific productivities’ is the calculation with country-specific yields. Thus,
countries with percentages >100% have a lower land productivities compared to the Austrian system or a more landintensive diet. ‘Austrian productivities’ is the calculation with Austrian yields for all countries. The effect of different
productivities is therefore excluded, and the differences can be explained by the variations in the composition of the diets.
Countries with percentages >100% have a more land-intensive diet compared to Austria.