284 M. Hoffmann et al. Agriculture, Ecosystems and Environment 80 2000 277–290
from 1865 until 1900 was assumed to be the same con- sidering that the change in ammonia emissions due to a
change in the way manure was handled was relatively small and NO
x
emissions from internal-combustion engines were still very low.
2.3. Calculation of average leaching and gross load Eighty-one leaching estimates were calculated for
each decade three crops, three soils and nine regions. Average leaching rates for each of the nine regions
were calculated by 1 weighting the leaching esti- mates according to the distribution percentage of
area of each of the three soil types for each crop class. The same soil distribution was used as in previous
study Johnsson and Hoffmann, 1998; 2 multiplying average leaching rates for each of the crops by acreage
of each crop; and 3 summing up the three gross loads from each crop and dividing by total acreage of arable
land in each region. Average leaching for the whole country was calculated as the sum of gross load in
each region divided by the total acreage of arable land.
Fig. 4. Input of N from both manure and fertiliser N and output as yield in tonnes A and expressed as per hectare of utilised arable land B.
2.4. Calculation of reduced N retention in lakes An estimate of the effect of draining lakes on N re-
tention was also done. This was calculated by estimat- ing the area of drained lakes and assigning a retention
to this lake area. The average N retention used in this estimate was 104 kg N ha
− 1
of lake area as reported in a study by Arheimer et al. 1997. Since retention
is dependent on nitrate concentration, it was adjusted to change in simulated average concentrations in dis-
charge from arable land, assuming that retention is proportional to concentration if all other factors are
unchanged.
3. Results
3.1. N flows to and from arable land The relation between input and output of N in
Swedish agriculture has undergone dramatic changes during the study period Fig. 4. From 1860 to 1930,
M. Hoffmann et al. Agriculture, Ecosystems and Environment 80 2000 277–290 285
the input of N with manure and commercial fertilis- ers was invariably smaller than the output with the
harvest. To some extent, this deficit was reduced by atmospheric deposition and N fixation. However, even
in the 1950s, N deposition only accounted for 30 of the deposition occurring in southern Sweden in 1985
Swedish Environmental Protection Agency, 1997. Furthermore, the acreage of annual legumes grown in
the 19th century was small BiSOS-N, 1865–1905. Thus, only a minor fraction of the difference between
output and input of N can be accounted for by deposi- tion and fixation. If all N fluxes are considered the N
deficit would have been significantly reduced as clover Trifolium spp. in grassland became more prevalent
towards the end of the 19th century. Nevertheless, the information available strongly indicates that 19th cen-
tury Swedish agriculture was characterised by a net removal of N from arable land. This was particularly
true for areas where cultivation of new land Fig. 3 enabled a considerable exploitation of organic N that
previously had accumulated in the soil.
In the 1950s, increased use of commercial N fer- tilisers created an entirely new N balance in Swedish
agriculture. Within less than a decade, the input of N surpassed the output with the harvest, and in the
mid-1970s this difference between input and output was at its largest. During the past two decades yet an-
other trend could be discerned; the harvest of N as a sum for all crops continues to increase in spite of the
Table 5 Harvests of grain and of N in grain in kg ha
− 1
as average of spring and winter cereals in Sweden Region
a
1865 1875
1885 1895
1905 1915
1925 1935
1945 1955
1965 1975
1985 Harvest of grain kg ha
− 1
Gss 1121
1317 1438
1440 1800
1924 2127
2516 2305
2638 3493
4013 4442
Gmb 1003
1169 1239
1153 1184
1552 1686
2041 1682
2112 2705
3167 3538
Gsk 1095
1243 1333
1275 1271
1413 1441
1750 1443
1569 2348
2742 3134
Gns 1010
1135 1139
1192 1292
1514 1625
2037 1746
2110 2936
3520 4069
Ss 1004
1200 1324
1300 1279
1483 1630
1935 1588
2001 2569
3037 3527
Ssk 1216
1419 1401
1674 1593
1476 1571
1680 1164
1420 1950
2274 2702
Harvest of N in grain kg N ha
− 1
Gss 19
23 25
25 31
34 37
45 40
46 62
70 78
Gmb 18
21 22
20 21
28 30
37 30
37 48
55 62
Gsk 19
21 23
22 22
24 25
31 25
27 41
48 55
Gns 18
20 20
21 22
26 28
37 31
37 52
62 71
Ss 18
21 23
23 22
26 28
34 28
35 45
53 62
Ssk 21
24 24
28 27
25 26
29 20
24 34
40 47
a
The abbreviations in the column are for each of the regions which are illustrated in Fig. 1.
fact that the fertilisation level is practically unchanged. This is partly because of increasing yields due to im-
provement in management practices, and partly be- cause of a smaller area of set-aside.
The trends in N input and output just described refer to the average situation in Sweden. In specific
areas, the temporal changes in N balance were even more pronounced. In particular, it is worth noticing
that the relatively uniform distribution of livestock that prevailed until about 1940 has since been replaced
by more specialised farming resulting in more pro- nounced regional differences. Animal density has de-
creased in central Sweden and increased in southern- most Sweden Fig. 2.
3.2. Leaching estimates 3.2.1. Cereals
It is generally recognised that cereal production with its high input of N can cause considerable N losses to
water. The model calculations in this study provided strong evidence that the agricultural practices prevail-
ing in the 1860s also caused substantial losses data not shown. Although the input of N at that time was
very low, the uptake of N by the crops Table 5 was often limited by weeds, pests, insects and unfavourable
chemical or physical soil conditions. Hence, it was not unusual for large amounts of leachable N to be left
in the soil at harvest. During the more than 100 years
286 M. Hoffmann et al. Agriculture, Ecosystems and Environment 80 2000 277–290
that have passed since the 1860s, both N input and N uptake efficiency have exhibited almost unbroken
upward trends. The model calculations showed that, until the 1920s, the increase in uptake efficiency was
larger than the increase in fertilisation rate and, conse- quently, the leaching of N decreased. This period was
followed by an approximately 50-year-long period of increased fertilisation and leaching, and it was not un-
til the 1970s that the upward trend in N leaching was curbed.
3.2.2. Bare fallow and grass Several experimental studies have demonstrated
that bare fallow causes substantial loss of N from the root zone Addiscott, 1988; Turtola, 1993; Gustafson,
1996. This has two major explanations. First, the amount of water infiltrating the soil increases when
the vegetation is removed. Secondly, the plant uptake of N is negligible if the soil is bare. The model calcu-
lations in the current study indicated that the annual N leaching from bare fallow was on average about
50 kg ha
− 1
per year data not shown. If weeds were prevalent this may have been an overestimate. How-
Fig. 5. Leaching a and gross load b for three assumptions of mineralisation and denitrification rates. A ‘High’: the entire increase in arable land originated from ploughing up of old grassland and a low denitrification rate was assumed; B ‘Normal’: 60 of the increase
in arable land originated from ploughing up of old grassland and a ‘normal’ denitrification rate was assumed; and C ‘Low’: 20 of the increase in arable land originated from ploughing up of old grassland and a high denitrification rate was assumed.
ever, the model calculations did not take into account that, in the past, manure was often spread on bare
fallow and that cultivation for weed control tillage enhances mineralisation and leaching processes.
Calculated leaching rates for grassland ranged from 1 to 11 kg ha
− 1
per year depending on soil type and leaching region. The area-weighted average leaching
of N was 5 kg ha
− 1
per year. 3.3. Gross N load and average leaching
The load during the 19th century increased some- what during the first two decades investigated
1860–1880 and then decreased until the 1930s Fig. 5. Changes in gross load of N were a result
of changes in 1 crop distribution; 2 total acreage of arable land; and 3 calculated leaching estimates.
Although the acreage of arable land was at its largest in the 1930s, gross load of N was at its smallest.
This was mainly due to three factors: 1 as much as 40 of the arable land was under grass Fig. 3;
2 enhanced mineralisation due to recent cultivation had practically ceased; and 3 leaching rate esti-
M. Hoffmann et al. Agriculture, Ecosystems and Environment 80 2000 277–290 287
Fig. 6. Calculated N leaching expressed in kg of leached N per harvested kg of N in foodstuff and fodder per ha of arable land in Sweden.
mates from cereals were at their lowest because the utilisation of N had increased.
From the 1930s until 1970, leaching rates from ce- reals increased by two-thirds; the load from bare fal-
low ceased whereas leaching from grassland remained unchanged.
The net effect of this was an increase in average N leaching rates from arable land of 60 Fig. 5a, while
the increase in gross load was only 30 Fig. 5b. One explanation for the smaller increase in gross load
compared to the increase in average leaching rates was the decrease in area of arable land. Fig. 3 shows that
practically the entire decrease in arable land was due to reduced acreage of grassland. Since leaching rates
from grassland were very low the effect on load was small. Furthermore, the decrease in arable land was
largest in the forest regions Gsk and Ssk, which had low leaching rates. Leaching has increased since the
1930s in all regions except the forest regions Gsk and Ssk. The largest increase in leaching rates occurred in
the two most southerly regions Gss and Gmb. These regions also witnessed the biggest increase in animal
density and input of N from manure Fig. 2. Gross load has decreased somewhat in the two forested areas
Gsk and Ssk.
To evaluate how much the change in acreage of arable land and change in crop distribution affected
the change in gross load to watercourses, the leaching rate estimates for 1865 were applied to all decades.
The result was that the change in load to watercourses followed the development in acreage of arable land,
i.e. gross load to watercourses increased from 1865 to a maximum in about 1920–1930 when the acreage
of arable land was largest, and decreased thereafter. This shows that, since change in leaching for bare
fallow was small, the decrease in leaching rates from cereals had a large influence on calculated gross load to
watercourses. Thus, the leaching estimates for cereals were critical for the final calculation of gross load to
watercourses via leaching. Leaching rates can, besides being expressed in ab-
solute values, also be expressed in terms of efficiency. Relating leaching rates to yields showed that ‘leach-
ing efficiency’ has decreased during the period inves- tigated since yields have increased more than leaching
Fig. 6. Production of a certain amount of food causes less leaching today compared to in the 19th century.
Efficiency of N application by manure and fertiliser was expressed as leached amount of N per applied kg
of N. It has decreased in about the same magnitude as leached amount of N per kg of N harvested in the
field. This illustrates the difficulty of making a sim- ple connection between input of N and leaching for
moderate inputs of N since the significance of miner- alisation for leaching is not taken into account in such
a comparison.
4. Discussion