Agriculture, Ecosystems and Environment 120 (2007) 153–165 www.elsevier.com/locate/agee
Soil loss due to harvesting of various crop types in contrasting agro-ecological environments G. Ruysschaert 1, J. Poesen *, G. Verstraeten 2, G. Govers Physical and Regional Geography Research Group, K.U. Leuven, Geo-Institute, Celestijnenlaan 200E, 3001 Heverlee, Belgium Received 13 February 2006; received in revised form 20 August 2006; accepted 21 August 2006 Available online 30 October 2006
Abstract Soil erosion studies on cropland usually only consider water, wind and tillage erosion. However, significant amounts of soil are also lost from the field during the harvest of crops such as sugar beet (Beta vulgaris L.), potato (Solanum tuberosum L.), chicory roots (Cichorium intybus L.), cassava (Manihot spp.) and sweet potato (Ipomoea batatas (L.) Lam). During the harvest soil adhering to the crop, loose soil or soil clods and rock fragments are exported from the field together with these crops. This soil erosion process is referred to as ‘soil losses due to crop harvesting’ (SLCH). Most of the studies investigated SLCH variability and its controlling factors for one crop type in similar agro-ecological environments and for comparable harvesting techniques. In this study, a compilation of SLCH studies was made in order to investigate the effect of crop type, agricultural systems, ecological conditions and harvesting technique on SLCH variability. SLCH rates ranged from few to tens of Mg ha1 harvest1 and SLCH was highly variable both in space and time. Comparison of four studies on SLCH for sugar beet revealed that harvesting technique and soil moisture content at harvesting time can be equally important for SLCH variability. The occurrence of soil clods harvested with the crop explained why SLCH was significantly larger for mechanically harvested potato in Belgium compared to manually harvested potato in China. SLCH values for manually harvested sugar beet, potato, cassava and sweet potato in China and Uganda were in general smaller than SLCH values for mechanically harvested sugar beet, potato and witloof chicory roots measured in Belgium and France. However, SLCH may also vary significantly within Europe due to differences in harvesting techniques. Soil moisture content at harvesting time was besides harvesting technique one of the key factors controlling SLCH variability. There were no systematic differences in SLCH between crop types, although the soil–crop contact area– crop mass ratio could explain more than 40% of the means from several SLCH studies. # 2006 Elsevier B.V. All rights reserved. Keywords: Soil erosion; Soil loss; Crop harvest; SLCH; Sugar beet; Potato; Cassava; Sweet potato; Chicory
1. Introduction Most soil erosion research on cropland focuses on soil redistribution caused by water, wind or tillage (e.g., mouldboard ploughing) and neglects the fact that considerable masses of soil may also be lost from arable land during the harvest of crops such as sugar beet (Beta vulgaris L.), potato (Solanum tuberosum L.), carrot (Daucus carota L.), chicory roots (Cichorium intybus L.), leek (Allium * Corresponding author. Tel.: +32 16 326425; fax: +32 16 322980. E-mail address:
[email protected] (J. Poesen). 1 Post-doctoral researcher of the Research Fund K.U. Leuven. 2 Fund for Scientific Research-Flanders, Belgium. 0167-8809/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2006.08.012
porrum L.), sweet potato (Ipomoea batatas (L.) Lam) and cassava (Manihot spp.). Soil adhering to the crop, loose soil or soil clods and rock fragments are exported from the field together with these harvested crops to external locations such as headlands, farmsteads and crop processing factories. This soil erosion process is referred to as ‘soil loss due to crop harvesting’ or ‘SLCH’ (Ruysschaert et al., 2004). Mean SLCH values for sugar beet calculated from soil tare data measured in sugar factories were 6 Mg ha1 harvest1 for The Netherlands, 14 Mg ha1 harvest1 for France, 9 Mg ha1 harvest1 for Belgium and 5 Mg ha1 harvest1 for Germany for the period 1978–2000 (Ruysschaert et al., 2005). Average SLCH values for potato, measured in field
154
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
studies were 3 Mg ha1 harvest1 in Belgium (Ruysschaert et al., 2006a) and 7 Mg ha1 harvest1 in Germany (Auerswald et al., 2006). Maximum soil losses caused by crop harvesting may rise to tens of Mg per hectare and per harvest (e.g., Poesen et al., 2001). Although SLCH values may thus be from the same order of magnitude as water and tillage erosion values, few studies have incorporated SLCH as a soil erosion process. Some of these studies measured SLCH at field plot scale and aimed at assessing SLCH variability and the importance of controlling factors. These controlling factors were divided by Ruysschaert et al. (2004) into four categories, i.e., soil (e.g., soil texture, soil moisture content at harvest), crop characteristics (e.g., crop shape), agronomic practices (e.g., plant density and crop yield) and harvesting technique. Other studies estimated SLCH variability based on soil tare data measured in crop-processing factories. From the latter studies it is impossible to investigate the effects of harvesting conditions on SLCH. All studies, most from western Europe, investigated SLCH for one or at maximum two crop types and are only representative for similar ecological conditions, a specific agricultural system and comparable harvesting techniques. The overall objectives of this study are therefore (1) to compile all available data on SLCH and (2) to estimate the effect of crop type, agricultural system, ecological conditions and harvesting technique on SLCH variability in addition to the effect of factors investigated in the individual studies (e.g., soil texture and soil moisture content at harvest). The results of this study allow one to assess the importance of SLCH in a range of contrasting agro-ecological environments.
2. Materials and methods 2.1. SLCH terminology SLCH can either be expressed as mass of oven-dry soil per unit of net crop mass or on an area-unit basis. Therefore, Ruysschaert et al. (2004) distinguished between massspecific SLCH (SLCHspec) and crop-specific SLCH (SLCHcrop), i.e.: SLCHspec ðMg Mg1 Þ ¼
M ds þ M rf M crop
(1)
where Mds is the mass of oven-dry soil (Mg), Mrf the mass of rock fragments (Mg), and Mcrop is the net crop mass (Mg), i.e., mass of clean roots or tubers: 1
SLCHcrop ðMg ha
1
harvest Þ ¼ SLCHspec M cy
(2)
where Mcy is the net crop yield (Mg ha1 harvest1). 2.2. Data and data analysis A literature study allowed compiling an overview table of SLCHcrop rates. SLCH values were either obtained from
direct field measurements or derived from soil tare data collected at crop processing factories. The field studies with sufficient information on the harvesting conditions were used to assess the importance of soil texture, soil moisture content at harvest, average crop mass, harvesting technique (manual versus mechanized harvesting), crop type and seedbed type (mounds or ridges versus flat) by means of linear regression analyses or ANOVA linear models (SAS Institute Inc., 1999) with continuous and dummy variables. Firstly, an analysis of SLCH for sugar beet was made, secondly studies of SLCH for potato were compared and finally, field studies of other crop types were added in order to assess the effect of crop type on SLCH variability. Smaller roots or tubers are expected to yield larger adhering SLCHspec values (i.e., soil adhering to the crop and excluding soil clods), as they have larger soil–crop contact area–crop mass ratios. For sugar beet, this ratio was defined by Vermeulen (2001) as the specific soil–beet contact area and is generalized here as the specific soil–crop contact area (Ss). As it can be expected that adhering soil losses are linearly related to the contact area of the crop with the soil, adhering SLCHspec should increase linearly with increasing specific soil–crop contact area. Therefore, the effect of crop type on adhering SLCHspec was investigated by estimating the specific soil–crop contact area (Ss) for each crop type and performing linear regression analysis. Ss for potato and sweet potato was derived from Ruysschaert et al. (2006a): the average Ss value measured was 1.16 cm2 g1. For all other crops, crop density was assumed to be equal to 1 g cm3, so that crop mass could be assessed by crop volume. Ss was then calculated by estimating the largest diameter and by assuming that the crop is cone-shaped (sugar beet, cassava, inulin chicory) or has a shape in between a cone and cylinder (carrot, black salsify, witloof chicory). If mean sugar beet mass was known, the soil–beet contact area was estimated with an equation proposed by Koch (1996): Soilbeet contact area ðcm2 Þ ¼ 86:58 þ 0:49ðM crop =pÞ 9:06 105 ðM crop =pÞ2
(3)
where Mcrop/p is the mean sugar beet mass (g).
3. Results 3.1. Overview of spatial and temporal SLCH variability for several crop types grown in contrasting agroecological environments An overview of all SLCHcrop (Mg ha1 harvest1) values calculated from data reported in the literature is provided in Table 1. Distinction is made between crop and data types, i.e., data based on field measurements and soil tare data provided by factories. Where possible, the temporal (daily,
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
155
Table 1 Overview of calculated soil losses due to crop harvesting (SLCHcrop) for various crops grown in different countries (based on Ruysschaert et al., 2005 and updated according to Ruysschaert, 2005) Number
Country/region
Sugar beet Field data 1 Belgium 2 France 3 Belgium (H) 4 China (H) Factory data 5 Belgium 6 Belgium 7 Belgium 8 Belgium 9 Belgium 10 Belgium 11 The Netherlands 12 The Netherlands 13 The Netherlands 14 The Netherlands 15 The Netherlands 16 The Netherlands 17 The Netherlands 18 The Netherlands 19 France 20 FRG 21 GDR 22 Germany 23 Germany 24 Germany/Bavaria 25 Turkey 26 27 28 29 30 31 32 33 34 35
Belgium Denmark France Germany UK Italy The Netherlands Northern Spain Central Belgium Germany/Bavaria
SLCHcrop (Mg ha1 harvest1), mean (min–max)
M.P.
n
Source
3.6 14.0 13.4 1.0
2002–2004 1984–1986 2001–2002 2002
611 82 48 14
Ruysschaert et al. (2006b) Duval (1988) and Poesen et al. (1999) Ruysschaert (2005) Li et al. (2006)
8.7 (4.4–19.5a/c/all)(1–100i) 8.8 (4.4–19.5a/c/all) 9.3 (4.7–19.4a/c/all) 8.5 (3.0–24.5d/r/s) 8.6 (1.2–18.8w/r/all) 8.3 (4.1–15.6w/c/all) 6.2 (3.4–13.4a/c/all) 5.9 (3.4–9.8a/c/all) 4.7 (0.1–10.3d/c/s) 3.5 (0.1–15.5w/r/all) 3.3 (2.1–5.3w/c/all) 5.2 (0.6–20.8a/r/all) 5.0 (2.0–8.0a/c/all) 4.6 (0.0–10.9a/c/all) 13.8 (7.7–20.5a/c/all) 6.9 (3.7–11.1a/c/all) 5.0 (2.0–9.5a/c/all) 3.7 (2.2–5.5a/c/all) 5.2 (2.2–10.7a/c/all) 6.0 (2.9–9.1a/r/all)* 3.8* (a/c/all)
1968–1996 1968–2000 1978–2000 1993–1995 1990–1996 1990–1996 1972–2001 1978–2000 1984–2004 2004 2004 1984–2004 1984–2004 1949–2004 1978–2000 1977–1989 1959–1989 1990–2000 1978–2000 1983–1985 1989–2000
29 33 23 373 918 90 30 23 280 453 13 860 21 56 23 13 31 11 23 12 n.a.
13.3* (a/c/all) 10.4* (a/c/all) 16.9* (a/c/all) 8.0* (a/c/all) 4.7* (a/c/all) 5.3* (a/c/all) 9.3* (a/c/all) 5.6* (a/c/all) 5.0y 15.0*
1981–1991 1981–1991 1981–1991 1981–1991 1981–1991 1981–1991 1981–1991 1981–1991 1956–1987 n.a.
n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
Poesen et al. (2001) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert (2005) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Ruysschaert et al. (2005) Auerswald and Schmidt (1986) Oruc¸ and Gu¨ngo¨r (2000) and Oztas et al. (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Anonymous (1994) and FAO (2002) Vanden Berghe and Gulinck (1987) Maier and Schwertmann (1981)
1985 1985–1986
4 1
Belotserkovsky and Larionov (1988) Belotserkovsky and Larionov (1988)
(0.2–21.4) (1.0–13.4) (0.2–3.0) (1.8–3.4)*
2002–2003 1996–2002 2002 1985
51 56 30 6
Ruysschaert et al. (2006a) Auerswald et al. (2006) Li et al. (2006) Belotserkovsky and Larionov (1988)
2.2 (0.0–45.2i) 0.6 (0.1–1.1s)*
1999–2001 1985–1986
1151 14
Ruysschaert et al. (2006c) Belotserkovsky and Larionov (1988)
(0.7–30.1) (2.0–44.3) (5.6–25.7) (0.2–1.9)
Mean sugar beet
7.4
Fodder beet Field data 36 Russia 37 Russia
2.3* (1.9–2.6) 3.5 (s)*
Mean fodder beet
2.9
Potato Field data 38 Belgium 39 Germany 40 China (H) 41 Russia
3.2 6.7 1.2 2.5
Factory data 42 Belgium 43 Russia Mean potato
2.7
156
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
Table 1 (Continued ) SLCHcrop (Mg ha1 harvest1), mean (min–max)
M.P.
n
Source
1990–1996
7
Poesen et al. (2001)
11.9 (1.7–70.5)
1996–1997
43
Poesen et al. (2001)
Cassava Field data 46 Uganda (H)
3.4 (0.4–25.8)
2002–2003
149
Isabirye et al. (in press)
Sweet potato Field data 47 Uganda (H)
0.1 (0.0–0.2)
2002
20
Isabirye et al. (in press)
Number
Country/region
Inulin chicory Factory data 44 Belgium Witloof chicory Field data 45 Belgium
8.1 (3.2–12.7a)
Carrot Factory data 48 Belgium 49 Russia
15.8 (0.5–65.5i) 1.3 (0.7–1.7s)*
1995–1996 2001–2002 1985–1986
225 2
Soenens (1997) and Van Esch (2003) Belotserkovsky and Larionov (1988)
Black salsify Factory data 50 Belgium/The Netherlands 51 Belgium/The Netherlands
6.8 (3.6–19.0i) 10.8 (1.4–28.4i)
1995 2001–2002
77 95
Soenens (1997) Ruysschaert (2005)
1986
1
Belotserkovsky and Larionov (1988)
Radish Factory data 52 Russia
1.7 (s) *
M.P., measurement period (year); n, number of observations; H, harvest by hand instead of by machine for the other studies; FRG, former West Germany; GDR, former East Germany; y, SLCHy (Mg ha1 year1) instead of SLCHcrop; *SLCHcrop = mass of oven-dry soil + mass of soil moisture instead of mass of oven-dry soil only; i, minimum and maximum values for individual deliveries; a, based on annual averages; d, based on daily averages; w, based on weekly averages; c, based on averages for the country; r, based on regional averages; s, based on a selective number of data; all, based on all factory data; n.a., not available.
weekly or annual) and spatial scale (regional or national) of data based on soil tare values from factories is indicated. SLCHcrop values varied from few Mg to tens of Mg ha1 harvest1, with a maximum observed of 100 Mg ha1 harvest1 for a sugar beet delivery to a factory in Belgium (Poesen et al., 2001). Considerable variability in SLCH values was observed at various temporal and spatial scales. Differences in SLCHcrop values for manually harvested sugar beet in Belgium, within a field plot, were at maximum 14.2 Mg ha1 harvest1. This was for a field plot with highly variable sand content of the soil (Ruysschaert, 2005). Total SLCHcrop values (i.e., adhering soil and soil clods) for a potato field (ca. 1.7 ha large), sampled several times during the harvesting season of 2002, ranged between 1.0 and 4.9 Mg ha1 harvest1. Total soil losses could be divided in adhering soil and loose soil (soil clods). The adhering soil loss component appeared to vary mainly over time, while the loose soil loss component showed mainly within field plot differences (Ruysschaert et al., 2006a). Spatial differences in weekly average SLCH values (i.e., average of SLCH values for all sugar beet deliveries to the district factory in a given week) for 40 Dutch sugar beet districts were on average 6.5 Mg ha1 harvest1 in 2004, which was larger than the temporal variation within that
season per district, i.e., on average 4.5 Mg ha1 harvest1. The opposite could be concluded from annual average SLCH values (i.e., average of the SLCH values for all sugar beet deliveries to the district factory in a given harvesting season) for the 1984–2004 period; the temporal variability per district was on average 7.7 Mg ha1 harvest1, which is larger than the yearly differences between the districts, i.e., on average 5.6 Mg ha1 harvest1 (Ruysschaert, 2005). An overview of the harvesting conditions of the field studies listed in Table 1, is provided in Table 2. This table distinguishes between mechanized harvesting and harvesting by hand and forms the basis for the analyses of Sections 3.2, 3.3 and 3.4. 3.2. SLCH for manually and mechanically harvested sugar beet grown in contrasting agro-ecological environments Four experiments on soil losses due to sugar beet harvesting could be compared; two on manual harvesting and two on mechanized harvesting (Table 2). The study of Duval (1988) on mechanically harvested sugar beets in the French sugar beet region (northern France) (Tables 1 and 2, no. 2) is discussed by Ruysschaert et al.
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
157
Table 2 Overview of field studies in which soil losses during crop harvest (SLCH) have been measured Mechanized harvest Crop type Study number* Source Country n SLCHspec (Mg Mg1) SLCHcrop (Mg ha1 harvest1) GMC (g g1) %Clay %Sand Soil type
Sugar beet 1 Ruysschaert et al. (2006b) Belgium 611 0.047 (0.009–0.460) 3.6 (0.7–30.1) 0.21 (0.08–0.35) 15 (7–36) 28 (11–65) Haplic Luvisols/eutric Cambisols/eutric Regosols (i)
Sugar beet 2 Duval (1988) France 82 0.255 (0.037–0.806) 14.0 (2.0–44.3) 0.19 (0.06–0.29) n.a. (10–30) n.a. Calcaric/calcic/haplic Luvisols (ii)
Mcrop/p (kg) Mcy (Mg ha1) Seedbed
0.9 (0.4–1.6) 79 (38–104) Flat
n.a. n.a.** Flat
Potato 38 Ruysschaert et al. (2006a) Belgium 51 0.069 (0.008–0.565) 3.2 (0.2–21.4) 0.15 (0.06–0.24) 10 (2–20) 40 (11–86) Haplic Luvisols/eutric Cambisols/plaggic Anthrosols (i) 0.1 (0.06–0.2) 48 (26–79) Ridge
Witloof chicory 45 Poesen et al. (2001) Belgium 43 0.270 (0.019–1.03) 11.9 (1.7–70.5) 0.21 (0.12–0.34) 10 (0–49) 48 (9–86) Haplic Luvisols/eutric Cambisols/eutric Regosols (i) 0.21 (0.08–0.36) 44 (19–87) Flat
Manual harvest Crop type Study number Source
Sugar beet 3 Ruysschaert (2005)
Sugar beet 4 Li et al. (2006)
Potato 40 Li et al. (2006)
Country n SLCHspec (Mg Mg1) SLCHcrop (Mg ha1 harvest1) GMC (g g1) %Clay %Sand Soil type
Belgium 48 0.18 (0.07–0.35) 13.4 (5.6–25.7)
China 14 0.014 (0.005–0.029) 1.0 (0.2–1.9)
China 30 0.032 (0.008–0.065) 1.2 (0.2–3.0)
0.22 (0.14–0.28) 17 (12–21) 15 (9–47) Haplic Luvisols/eutric Cambisols/eutric Regosols (i)
0.16 (0.10–0.24) 30 (24–38) 15 (6–25) Haplic Chernozems/haplic Phaeozems/haplic Kastanozems/calcic Cambisols (iii) 0.70 (0.40–1.02) 64 (42–91) Ridge
0.15 (0.07–0.24) 27 (16–39) 25 (7–52) Haplic Chernozems/haplic Phaeozems/haplic Kastanozems/calcic Cambisols (iii) 0.14 (0.95–0.29) 36 (18–78) Ridge
Mcrop/p (kg) Mcy (Mg ha1) Seedbed
1.0 (0.6–2.1) 77 (44–114) Flat
Cassava 46 Isabirye et al. (in press) Uganda 149 0.021 (0.003–0.161) 3.4 (0.4–25.8)
Sweet potato 47 Isabirye et al. (in press) Uganda 20 0.003 (0.002–0.007) 0.1 (0.0–0.2)
0.13 (0.04–0.35) 23 (9–39) 67 (56–80) Rhodi lixic ferralsols
0.09 (0.04–0.11) 23 (23–26) 68 (66–68) Rhodi lixic ferralsols
n.a. 161 Flat
n.a. 28 Mound
Distinction is made between manually and mechanically harvested crops. Besides the mean values, observed minimum and maximum values are indicated between brackets. *See Table 1; **a sugar beet yield of 55 Mg ha1 was assumed for calculation of SLCHcrop. (i) Belgian soil map (1:20 000); committee for mapping soils and vegetation in Belgium and FAO et al. (1998); (ii) Duval (1988) and FAO et al. (1998); (iii) FAO (1974); n, number of observations; GMC, gravimetric soil moisture content at harvest; Mcrop/p, mean root or tuber mass; Mcy, net crop yield; n.a., not available.
(2004). The most important predictor variable was gravimetric soil moisture content (GMC) during the harvest, which was exponentially and positively related to SLCHspec (R2 = 0.50). Other factors that could explain part of the variability were %clay, soil organic matter content and diameter of the beet crown. The second data set on mechanically harvested sugar beets (Tables 1 and 2, no. 1) is described by Ruysschaert et al. (2006b) and is based on soil losses occurring during the harvest of beets grown for sugar beet variety trials, representatively distributed over the Belgian sugar beet area and organised by the sugar beet research institute (KBIVB-IRBAB). GMC was also the best predictor variable in this study and, in accordance with Duval (1988), positively and exponentially related to SLCHspec. Other factors determining SLCHspec were %clay, mean beet mass (Mcrop/p) and plant density (PD).
Sugar beet was manually harvested in Belgium during an experiment investigating topography-induced variability of SLCH (Ruysschaert, 2005; Tables 1 and 2, no. 3). Researchers harvested sugar beet following a standard procedure. In this study, only 18% of the variability could be explained by a positive and linear relationship with GMC. Soil organic matter content, mean beet mass and plant density were other factors determining SLCHspec. Li et al. (2006) measured soil losses caused by the harvest of sugar beets in northeast China (Tables 1 and 2, no. 4). In this study, the sugar beets were manually harvested by farmers (owners of the field plots) as they are used to do it in practice. The coefficient of determination was largest for positive power (R2 = 0.50) and exponential (R2 = 0.47) regression equations between GMC and SLCHspec. Good relationships were also obtained with soil texture, mean root mass, plant density and crop yield.
158
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
The correlation with %clay was, in contrast with the studies by Duval (1988) and Ruysschaert et al. (2006b; no. 1), negative and could be attributed to co-linearity between the independent variables.
From all four studies, it could be concluded that GMC is a key factor explaining SLCH variability. In Fig. 1(a), SLCHcrop is plotted against GMC. Large differences in SLCHcrop and SLCHspec values between the four studies exist for similar soil
Fig. 1. Illustration of the studies on soil losses due to crop harvesting (SLCH) for sugar beet described in Table 2. (a) Scatter plot between crop-specific SLCH (SLCHcrop) and gravimetric soil moisture content during the harvest (GMC) and the exponential curves fitted through the data. (b) Scatter plot between the natural logarithm of mass-specific SLCH (SLCHspec; Mg Mg1) and GMC. Bold lines represent the results of the linear regression analysis with GMC and dummy variables for each study as independent variables (Eq. (4); R2 = 0.84), while the regular lines indicate the regression lines between ln(SLCHspec) and GMC fitted for each study separately.
G. Ruysschaert et al. / Agriculture, Ecosystems and Environment 120 (2007) 153–165
moisture contents. The linear regression equation for the natural logarithm of SLCHspec (ln(SLCHspec)) with GMC and study type, reflecting the overall effect of soil, agronomic practices and harvesting technique related factors, as independent variables yielded the following result: lnðSLCHspec Þ ¼ 5:42 þ 10:43 GMC 0:64 dumHC þ 1:38 dumHB þ 1:83 dumMF ; p < 0:0001;
R2 -adj: ¼ 0:84 (4)
where GMC is the gravimetric soil moisture content at harvesting time (g g1), dumHC = 1 for sugar beet harvested by hand in China (Li et al., 2006; no. 4), dumHB = 1 for sugar beet harvested by hand in Belgium (Ruysschaert, 2005; no. 3) and dumMF = 1 for mechanized harvested sugar beet in France (Duval, 1988; no. 2). In all other cases the dummy variables are zero. All parameters were significant ( p < 0.0001). Fig. 1(b) illustrates this regression equation by means of bold lines, while regular (thin) lines are results of the regression analysis between ln(SLCHspec) and GMC for all four studies separately. The regular and bold lines are in good agreement for mechanized sugar beet harvesting in Belgium and France and manual harvesting in China. This means that the relative effect of GMC on ln(SLCHspec) is equal in these studies but that there is a constant difference in ln(SLCHspec) values for similar soil moisture contents. An exception is the study on manually harvested sugar beets in Belgium. The regression is biased by SLCHspec values at the lowest soil moisture contents. Hard and dry soil clods were attached to the rootlets and were not broken by the standard cleaning procedure applied. This is in contrast with mechanized harvesting during which rootlets are broken when the beets are uplifted. Omitting the data for the soil moisture contents
