KVADRATNET-data

Observed figures for rates of animal manure and mineral fertilizer may be obtained from the KVADRATNET, which is a square grid of 7×7 km established by DAAS in the winter 1986/87 with the purpose of annual estimates on the content of mineral soil N in the spring.

This systematic grid has 830 intersections on land, 590 of these are located on farmland. Each intersection is represented by a 50×50 m area managed by the farmer, who also annually re-ports on manure and fertilizer applications. Heidmann et al. (2001) drew attention to elements of uncertainty associated with the reporting: 1) Application rate, particularly for animal ma-nure, 2) conversion to nutrient rate using standard values, and 3) the simple risk of notes on applications going missing.

The KVADRATNET data consist at present of 3300 observations on winter wheat. We se-lected the intersections points of the most important preceding crops (cereals, sugar beet, rape or peas) and the dominating soil types (JB nr. <9, according to the Danish soil classification system). Data control was omitted in this preliminary analysis, but a few recordings with zero application or missing values were deleted without further verification, indicating that some quality control is needed for more detailed analysis. Some of the deleted recordings may be due to organic farming conditions. The data did not give us the opportunity to distinguish be-tween ordinary wheat for feeding and wheat for bread as the latter may allow a higher N rate to be applied. Bread wheat accounts for <10% of the area with winter wheat, and have a sup-plemental standard N rate of 30 kg N/ha. The remaining data were divided into two data sets:

A) exclusively fertilized with mineral fertilizer N, and B) fertilized with animal manure plus mineral fertilizer N (Table 5.1).

Several changes in agricultural systems have taken place over time (e.g. area with winter wheat, livestock holdings and crop rotation), which adds noise to data and causes consider-able variation. In a crop rotation including winter wheat, an intersection point may be redrawn systematically at intervals of 3-4 years, but as the crop preceding winter wheat more fre-quently has become winter wheat since the middle of the 1990s, which indicates continuous cropping, systematic redrawing has increased. This causes correlated data that may be consid-ered as repeated measurements, and analysis on sub-datasets of intersection point redrawn

≥10 times may be more suitable to estimate changes over time as some random variation may be excluded. The datasets sub-A and sub-B reflect the typologies of winter wheat crops

fertil-prescribed by the statutory order taking into account (i) the soil type, (ii) the effect of previous crop, (iii) the long-term effect of animal manure and (iv) the annual N prognosis. The differ-ence between EONR and RSNR corresponds to -13% and the yield was reduced by 1.9 dt/ha.

However, the standard N rates for each crop are controlled at farm level giving the farmer the opportunity to redistribute fertilizer N between crops and thereby make some individual ad-justments of the N rates. Hence, the actual application N rate may depend on the farmers’

decisions.

5.3.2 KVADRATNET-data

Observed figures for rates of animal manure and mineral fertilizer may be obtained from the KVADRATNET, which is a square grid of 7×7 km established by DAAS in the winter 1986/87 with the purpose of annual estimates on the content of mineral soil N in the spring.

This systematic grid has 830 intersections on land, 590 of these are located on farmland. Each intersection is represented by a 50×50 m area managed by the farmer, who also annually re-ports on manure and fertilizer applications. Heidmann et al. (2001) drew attention to elements of uncertainty associated with the reporting: 1) Application rate, particularly for animal ma-nure, 2) conversion to nutrient rate using standard values, and 3) the simple risk of notes on applications going missing.

The KVADRATNET data consist at present of 3300 observations on winter wheat. We se-lected the intersections points of the most important preceding crops (cereals, sugar beet, rape or peas) and the dominating soil types (JB nr. <9, according to the Danish soil classification system). Data control was omitted in this preliminary analysis, but a few recordings with zero application or missing values were deleted without further verification, indicating that some quality control is needed for more detailed analysis. Some of the deleted recordings may be due to organic farming conditions. The data did not give us the opportunity to distinguish be-tween ordinary wheat for feeding and wheat for bread as the latter may allow a higher N rate to be applied. Bread wheat accounts for <10% of the area with winter wheat, and have a sup-plemental standard N rate of 30 kg N/ha. The remaining data were divided into two data sets:

A) exclusively fertilized with mineral fertilizer N, and B) fertilized with animal manure plus mineral fertilizer N (Table 5.1).

Several changes in agricultural systems have taken place over time (e.g. area with winter wheat, livestock holdings and crop rotation), which adds noise to data and causes consider-able variation. In a crop rotation including winter wheat, an intersection point may be redrawn systematically at intervals of 3-4 years, but as the crop preceding winter wheat more fre-quently has become winter wheat since the middle of the 1990s, which indicates continuous cropping, systematic redrawing has increased. This causes correlated data that may be consid-ered as repeated measurements, and analysis on sub-datasets of intersection point redrawn

≥10 times may be more suitable to estimate changes over time as some random variation may be excluded. The datasets sub-A and sub-B reflect the typologies of winter wheat crops

ized with mineral fertilizer only and with a combination of animal manure and mineral fertil-izer, respectively (Table 5.1).

Table 5.1 Overview of the datasets extracted from the KVADRATNET database for winter wheat.

Dataset A Dataset B

Mineral fertilizer N exclusively

Animal manure plus mineral fertilizer N

No. of observations 1374 1103

No. of intersection points 389 311

Dominating soil types * JB6, JB7 and JB4 JB6, JB7 and JB4

Subdataset ** Dataset sub-A Dataset sub-B

No. of observations 190 190

No. of intersection points 16 17

Dominating soil types * JB7 JB6, JB7

Dominating county Zeeland Jutland

* JB no. according to the Danish soil classification system

** More than nine recordings of individual intersection points

To get the advantage of the repeated measurements we focus on the sub-datasets that both represent sandy loam soils but in two different regions of the country (Table 5.1). We focus on the changes comparing two periods: the past, from the beginning of the 1990s to 2000 (some variation in the starting year used depending on the dataset) versus today, the period 2001-06 (reduced standard N rates corresponding to c. 90% of the standard N rates). The fig-ures of the 10, 25, 50, 75 and 90% fractiles give a general view of the data and we used the median to calculate N rates for different periods.

For intersection points without application of animal manure the mineral fertilizer N rate changed from 182 kg N/ha as an average of the median during 1991-2000 to 166 kg N/ha (corresponding to -9%) as an average of the median during 2003-06 using sub-dataset A (Fig-ure 5.4). Using the full dataset A the reduction in N rate was in the same order, 19 kg N/ha (data not shown), but the changes in the average median value was from 175 to 156 kg N/ha (corresponding to -11%). In Figure 5.4 it is noticed that the values for the 50, 75 and 90%

fractile are very close to each other for 2005/06, indicating a maximum N rate of 170 kg N/ha in practice under the conditions of reduced standard N rates. The maximum N rate during 1991-2000 (before and just after introduction of standard N rates) has been about 205 kg N/ha based on an average of the 90% fractile (Figure 5.4). This indicates a maximum decrease in the N rate of 17% compared with the situation before introduction of reduced standard N rates.

ized with mineral fertilizer only and with a combination of animal manure and mineral fertil-izer, respectively (Table 5.1).

Table 5.1 Overview of the datasets extracted from the KVADRATNET database for winter wheat.

Dataset A Dataset B

Mineral fertilizer N exclusively

Animal manure plus mineral fertilizer N

No. of observations 1374 1103

No. of intersection points 389 311

Dominating soil types * JB6, JB7 and JB4 JB6, JB7 and JB4

Subdataset ** Dataset sub-A Dataset sub-B

No. of observations 190 190

No. of intersection points 16 17

Dominating soil types * JB7 JB6, JB7

Dominating county Zeeland Jutland

* JB no. according to the Danish soil classification system

** More than nine recordings of individual intersection points

To get the advantage of the repeated measurements we focus on the sub-datasets that both represent sandy loam soils but in two different regions of the country (Table 5.1). We focus on the changes comparing two periods: the past, from the beginning of the 1990s to 2000 (some variation in the starting year used depending on the dataset) versus today, the period 2001-06 (reduced standard N rates corresponding to c. 90% of the standard N rates). The fig-ures of the 10, 25, 50, 75 and 90% fractiles give a general view of the data and we used the median to calculate N rates for different periods.

For intersection points without application of animal manure the mineral fertilizer N rate changed from 182 kg N/ha as an average of the median during 1991-2000 to 166 kg N/ha (corresponding to -9%) as an average of the median during 2003-06 using sub-dataset A (Fig-ure 5.4). Using the full dataset A the reduction in N rate was in the same order, 19 kg N/ha (data not shown), but the changes in the average median value was from 175 to 156 kg N/ha (corresponding to -11%). In Figure 5.4 it is noticed that the values for the 50, 75 and 90%

fractile are very close to each other for 2005/06, indicating a maximum N rate of 170 kg N/ha in practice under the conditions of reduced standard N rates. The maximum N rate during 1991-2000 (before and just after introduction of standard N rates) has been about 205 kg N/ha based on an average of the 90% fractile (Figure 5.4). This indicates a maximum decrease in the N rate of 17% compared with the situation before introduction of reduced standard N rates.

Year

1985 1990 1995 2000 2005

Nitrogen rate [kg/ha]

60 80 100 120 140 160 180 200 220 240

Figure 5.4 Fractiles of mineral fertilizer N rate for dataset sub-A (Table 3.1). Full line: 50% fractile, dashed lines: 25 and 75% fractiles, dotted lines: 10 and 90% fractiles.

Using these estimates the yield losses due to reduced standard N rates were estimated using the DAAS sub-set of 115 response curves. The yield was calculated for the application rates of 166 and 182 kg N/ha based on dataset sub-A for each response function. Using these pair-wise estimates the yield difference between the average median N rate during the past and of today was estimated at 1.4 dt/ha (standard deviation 0.87 dt/ha). Using the average median N rates for the full dataset A that has a higher variation than for sub-A, the yield reduction was estimated at 2.0 dt/ha (standard deviation 1.07 dt/ha).

In this way we have answered the question regarding the yield effect of reduced standard N rates where mineral fertilizer was used exclusively. Including animal manure the question is more complicated as there have been several changes in the substitution rate for N in different types of animal manures together with changes in claims regarding application time and method (Mikkelsen et al., 2005).

Using the full dataset B the maximum application rate of animal manure N has been reduced considerably fulfilling the aims of the implemented legislation. Since 1985 a steady reduction in the use of mineral fertilizer N in combination with animal manure has been observed (Fig-ure 5.5).

Based on the repeated measurements in dataset sub-B the average of the median was calcu-lated for two periods (Table 5.2). Converting the change in total-N rate applied in animal ma-nure to mineral N corresponding to an ammonia-N:total-N ratio of 0.65, the overall change in N rate is (22 + 33×0.65) = 43 kg N/ha. A similar figure is obtained using the full dataset B.

The figure is twice the figure for using mineral fertilizer only. However, the change in mineral

Year

1985 1990 1995 2000 2005

Nitrogen rate [kg/ha]

60 80 100 120 140 160 180 200 220 240

Figure 5.4 Fractiles of mineral fertilizer N rate for dataset sub-A (Table 3.1). Full line: 50% fractile, dashed lines: 25 and 75% fractiles, dotted lines: 10 and 90% fractiles.

Using these estimates the yield losses due to reduced standard N rates were estimated using the DAAS sub-set of 115 response curves. The yield was calculated for the application rates of 166 and 182 kg N/ha based on dataset sub-A for each response function. Using these pair-wise estimates the yield difference between the average median N rate during the past and of today was estimated at 1.4 dt/ha (standard deviation 0.87 dt/ha). Using the average median N rates for the full dataset A that has a higher variation than for sub-A, the yield reduction was estimated at 2.0 dt/ha (standard deviation 1.07 dt/ha).

In this way we have answered the question regarding the yield effect of reduced standard N rates where mineral fertilizer was used exclusively. Including animal manure the question is more complicated as there have been several changes in the substitution rate for N in different types of animal manures together with changes in claims regarding application time and method (Mikkelsen et al., 2005).

Using the full dataset B the maximum application rate of animal manure N has been reduced considerably fulfilling the aims of the implemented legislation. Since 1985 a steady reduction in the use of mineral fertilizer N in combination with animal manure has been observed (Fig-ure 5.5).

Based on the repeated measurements in dataset sub-B the average of the median was calcu-lated for two periods (Table 5.2). Converting the change in total-N rate applied in animal ma-nure to mineral N corresponding to an ammonia-N:total-N ratio of 0.65, the overall change in N rate is (22 + 33×0.65) = 43 kg N/ha. A similar figure is obtained using the full dataset B.

The figure is twice the figure for using mineral fertilizer only. However, the change in mineral

fertilizer N rate is in the same order, and the change in animal manure N rate may be related to implementation of DEPA statutory orders fulfilling the Nitrate Directive.

Year

1985 1990 1995 2000 2005

Nitrogen rate [kg/ha]

0 50 100 150 200

Figure 5.5 Fractiles of mineral fertilizer N application rate used in combination with animal manure, dataset B (Table 5.1). Full line: 50% fractile, dashed lines: 25 and 75% fractiles, dotted lines: 10 and 90% fractiles.

Table 5.2 Change in the median application rate of total-N in mineral fertilizer and animal manure at 17 intersection points using animal manure (Dataset sub-B). Based on the median shown in Figure 5.5.

Figures for animal manure require further data control.

Mineral fertilizer Animal manure

1992-2000 102 158

2001-2006 80 125

Difference 22 33

The median N rate for the 17 intersection points using animal manure was 205 kg N/ha (102 + 158×0.65) as an average of 1992-2000. This rate is similar to the 90% fractile of the sub-A data for 16 intersection points receiving mineral fertilizer only, indicating that the N rates in the sub-B dataset have been supra-optimal in the past. The change from supra-optimal to op-timal is included in the estimate of 43 kg N/ha, which therefore is an overestimation of the change from standard N rates to reduced standard N rates for farms using animal manure.

However, this preliminary analysis indicates that data require further control regarding rate and time of animal manures application.

fertilizer N rate is in the same order, and the change in animal manure N rate may be related to implementation of DEPA statutory orders fulfilling the Nitrate Directive.

Year

1985 1990 1995 2000 2005

Nitrogen rate [kg/ha]

0 50 100 150 200

Figure 5.5 Fractiles of mineral fertilizer N application rate used in combination with animal manure, dataset B (Table 5.1). Full line: 50% fractile, dashed lines: 25 and 75% fractiles, dotted lines: 10 and 90% fractiles.

Table 5.2 Change in the median application rate of total-N in mineral fertilizer and animal manure at 17 intersection points using animal manure (Dataset sub-B). Based on the median shown in Figure 5.5.

Figures for animal manure require further data control.

Mineral fertilizer Animal manure

1992-2000 102 158

2001-2006 80 125

Difference 22 33

The median N rate for the 17 intersection points using animal manure was 205 kg N/ha (102 + 158×0.65) as an average of 1992-2000. This rate is similar to the 90% fractile of the sub-A data for 16 intersection points receiving mineral fertilizer only, indicating that the N rates in the sub-B dataset have been supra-optimal in the past. The change from supra-optimal to op-timal is included in the estimate of 43 kg N/ha, which therefore is an overestimation of the change from standard N rates to reduced standard N rates for farms using animal manure.

However, this preliminary analysis indicates that data require further control regarding rate and time of animal manures application.

In document summary Causes of yield stagnation in winter wheat in denmark (Sider 66-70)