Crop Science 42:804-815 (2002)
© 2002 Crop Science Society of America
CROP ECOLOGY, MANAGEMENT & QUALITY
Spatial Variability of Soybean Quality Data as a Function of Field Topography
I. Spatial Data Analysis
A. N. Kravchenko and
D. G. Bullock*
Dep. of Crop Sciences, 1102 S. Goodwin Ave., Univ. of Illinois, Urbana, IL 61801
* Corresponding author (dbullock{at}uiuc.edu)
 |
ABSTRACT
|
|---|
Analysis and quantitative characterization of the spatial variability of soybean [Glycine max (L.) Merr.] protein and oil concentrations is an important task for site-specific soybean management. The objectives of this study were to spatially characterize the variability of soybean protein and oil concentrations in five fields of central Illinois, and to determine the influence of field topographical features, such as elevation range, terrain slope, and terrain curvature, on soybean protein and oil concentrations and their distributions within the studied fields. More than 200 samples for soybean quality analysis were collected from each of the three studied fields in fall of 1998, and from each of the two fields in fall of 1999. Dense elevation measurements were taken on each of the fields and topographical features, such as slope and curvature, were derived from the elevation data using the geographic information system (GIS). Both protein and oil concentrations were spatially correlated in all the studied fields as indicated by variograms. Field topography strongly affected soybean quality; however, its influence depended on the weather conditions during the growing seasons. Higher protein was observed at higher elevation sites, as well as at sites with higher slopes and convex curvatures during growing seasons with sufficient or excessive precipitation, while lower proteins were observed at such topographical conditions when the growing season was dry. Ranges of spatial cross-correlation between protein concentrations and elevation were related to the ranges and changes in shape of the elevation variograms, suggesting that spatial variability of field elevation can be used as an indicator of the spatial distribution of high protein or oil soybeans.
Abbreviations: GIS, geographic information system
|
INTRODUCTION
|
|---|
SOYBEAN IS AN IMPORTANT SOURCE FOR SATISFYING the world demand for protein and vegetable oil (Smith and Huyser, 1987). Soybean grain quality, as measured by protein and oil concentrations, is determined by both genetics and environment; however, to the best of our knowledge, their relative contribution has not been reported in the literature. Currently there is insufficient information to predict grain quality based upon cultivar or growing conditions.
In the past, plant breeders have traditionally emphasized breeding high yielding soybean rather than increasing the protein or oil content of soybean grain (Burton, 1985; Wilcox, 1985). This has changed substantially and a growing interest has been expressed by the food industry for obtaining soybean with improved seed qualities (Wilson, 1991). For example, efforts have been extended to breeding soybean for higher protein (Burton, 1985, 1987, 1991; Wilcox and Cavins, 1995), higher oil (Wilcox, 1985), higher contents of S-containing amino acids, such as methionine and cysteine (Sexton et al., 1998), as well as on modifying soybean oil fatty acid composition, including oleic (Rahman et al., 1996), palmitic, and linolenic acid contents (Rebetzke et al., 1998; Stoj
in et al., 1998a, 1998b). It should be noted, however, that soybean quality characteristics are not independent, and thus specific breeding objectives are required. For example, protein concentration is often inversely correlated with both oil concentration and grain yield (Burton, 1985; Helms and Orf, 1998; Wilcox, 1998). Hence, breeding for high protein will result in lower yielding soybean with lower oil concentration.
While genetic improvement is possible and desirable, it is also important to recognize that soybean grain yield and seed quality are strongly affected by environmental factors such as temperature, light, water availability, and soil physical and chemical properties (Gibson and Mullen, 1996; Raper and Kramer, 1987; Spilker et al., 1981). Environmental factors affecting soybean growth are often spatially variable and the variability extends from microscales to field and watershed scales (Warrick et al., 1986). Fortunately, the majority of soil properties are often distributed within fields in a manner that is amenable to geostatistical description and analysis (Warrick et al., 1986).
Variability of environmental and soil factors contributes to crop performance variability. For example, variability in soil water content was found to be similar to variability in crop yield in an irrigation experiment in Utah (Or and Hanks, 1992). Miller et al. (1988) observed spatial correlations between wheat harvest indices and soil clay content. Spatial structure in within-field crop yield variability was observed by Jaynes and Colvin (1997) for 6 yr of corn and soybean yields in Iowa and by Miller et al. (1988) for wheat grain yield in California. Mulla (1993) observed spatial structure in wheat yield distributions in eastern Washington. Similarly, Eghball and Varvel (1997) have shown that soybean grain yield has both spatial and temporal variability and that the strong influence of environment will affect how the spatial variability for grain yield is expressed in any given year.
Topography is a particularly attractive variable in describing and predicting spatial variability of crop yields for precision agriculture management. It is a soil formation factor that defines distribution of soil moisture, organic matter, nutrients, soil textural composition, and other soil properties affecting plant growth within a field (Changere and Lal, 1997; Gessler et al., 2000; Hanna et al., 1982; Moore et al., 1993). It also affects within field temperature and humidity variations. Hence, topography can be regarded as a compound parameter that reflects combined influence of various yield-affecting factors and interactions. A valuable advantage of topographical data is that they are easy to obtain compared with time- and labor-consuming measurements of soil properties. Topography has substantial influence on spatial variability of crop yields. Kravchenko and Bullock (2000) found that in various fields, topographical variables, such as elevation, terrain slope, and curvature, explained from 6 to 54% of corn and soybean yield variability. Timlin et al. (1998) found surface curvature to be a useful parameter for describing yield, topography, and weather relationships.
Since spatial structure has been detected in distributions of soybean yields, we hypothesize that soybean protein and oil contents are also distributed within a field according to a certain spatial structure. However, we know of no information in the literature on within-field spatial variability of soybean protein and oil contents. We also hypothesize that within-field variability in topography and soil properties may affect spatial variability of soybean protein and oil contents. The objectives of this study were (i) to determine the extent of spatial variability of soybean protein and oil concentration data on an agricultural field scale, and (ii) to determine the influence of field topographical features, such as elevation range, terrain slope, and terrain curvature, on soybean protein and oil concentrations and their spatial distributions within the studied fields.
|
MATERIALS AND METHODS
|
|---|
Soybean protein and oil concentrations were studied using data from five agricultural fields located in eastern Illinois. Soils of the fields are classified as Haplaquolls, Argiudolls, and Endoaquolls with silt loam and silty clay loam surface textures (Table 1). The fields vary in size from 10 to 36 ha (Table 1). A summary of the field topographical features along with dominant field soils is presented in Table 1. The differences between the highest and the lowest elevations of the fields used in the study varied from 5.2 to 9.1 m. The maximum terrain slope was equal to 5.6°, terrain curvature values ranged from -0.45 to 0.43 x 10-2 m. Each of the fields has been in a corn (Zea mays L.) and soybean rotation for at least 30 yr. The crops are grown in a conventional manner utilizing drilled rows. Cultivars are indicated in Table 1.
Soybean seed samples were collected in the fall of 1998 from the DL98, RF98, and WL198 fields and in the fall of 1999 from the KN99 and WL299 fields. The number of seed samples collected from unique positions at each field is presented in Table 1 and sample locations for each of the fields are shown in Fig. 1 . Each sample was collected by removing 
20 contiguous whole plants just prior to a combine pass. The sample locations were geo-referenced later using survey grid GPS (Leica 500 RTK). The plant samples were later hand threshed to separate the grain from the stover. Oil and protein concentrations were measured via near-infrared reflectance method for protein and oil determination in soybeans (American Association of Cereal Chemists, 1998).
View larger version (69K):
[in this window]
[in a new window]
|
Fig. 1. Soybean seed sampling locations along with transects for analyzing relationships between soybean protein or oil concentrations and topography, and elevation maps for the (a) DL98, (b) RF98, (c) WL198, (d) KN99, and (e) WL299 fields.
|
|
A Leica 500 RTK GPS system (Leica, Geosystem, Heerbrugg, Switzerland) or a SOKKIA SET 5 total station (SOKKIA Corp., Overland Park, KS) was used to measure elevation at set points in each field in order to produce digital terrain maps. Elevation measurements for the WL198 and WL299 fields were taken on a semiregular grid with a distance between measurements of 10 m. For the DL98, DL99, RF98, and RF99 fields, the distance between the elevation measurements varied from 2 to 60 m, depending on the complexity of the terrain. Measurements on level parts of the field were taken at greater distances, while marked depressions and elevations were sampled more intensely. Elevation measurements at the KN99 field were taken at each soybean seed sampling location. Topographical features considered in the study included (i) elevation range, calculated as a difference between the point elevation and minimum field elevation and measured in meters, (ii) slope, defined as the first derivative of the terrain surface and measured in degrees, and (iii) curvature, defined as the second derivative of the terrain surface and measured in 10-2 m. ArcView GIS Spatial Analyst (Environmental Systems Research Institute, 1996) was used to convert elevation measurements into cell-based maps. Slope and curvature maps were derived from the elevation maps on the same cell basis. More details on topographical data used in the study can be found in Kravchenko and Bullock (2000).
Data analysis consisted of two parts: a whole field data analysis, and a transect data analysis. For the first part, the whole field data sets were studied by means of statistics and geostatistics. For the transect data analysis, the GIS was utilized to create transects, one across each field, and the relationships between soybean protein and oil concentrations and topography were studied for each of the transects separately. Transect lines were selected so that they were situated approximately in the middle of the field, encompassed areas with diverse field topography, and covered a large number of soybean seed sampling points. After the appropriate transect line was drawn through the field, all the sampling points located within a 10-m distance from the line were included in the transect. For each of the sampling points, elevation range, curvature, and slope were obtained from the map cell in which the point was situated. Locations of the transects are shown in Fig. 1.
Daily temperature and precipitation data were collected at on-site weather stations on the WL198 and WL299 fields during the growing seasons of 1998 and 1999. For the rest of the fields, the weather data were obtained from the nearest state weather stations (Midwestern Climate Center, Champaign, IL). The distances between the DL98, RF98, and KN99 fields and the weather stations were 13 km, 7 km, and 10 km, respectively.
Geostatistical Data Analysis
Geostatistics was used to analyze spatial variability of soybean quality data (Journel and Huijbregts, 1978). Spatial structure of soybean protein and oil concentrations [
(h)] was characterized by sample variograms calculated from the experimental data using the following equation:
 | [1] |
where xi and xi + h are sampling locations separated by a distance h, Z(xi) and Z(xi + h) are measured values of the variable Z at the corresponding locations, and n is the total number of sample pairs for the distance h. In this study, the sample variograms were fitted with spherical and exponential variogram models defined by Eq. [2] and [3], respectively (Deutsch and Journel, 1998):
 | [2] |
and
 | [3] |
where a is the variogram range, c0 is the nugget component of the variogram, and c is the positive variance contribution or sill. Variogram range defines the distance beyond which spatial correlation between the samples ceases to exist and, hence, it can be used as an indicator of the appropriate cell size for a field grid in site-specific management (Han et al., 1994). The nugget value is interpreted as a random component in the data spatial variability. The ratio of the nugget, c0, to the nugget plus sill, c0 + c, characterizes the importance of the random component in the whole field spatial variability of the data.
Adequacy of the chosen variogram models was tested using the cross-validation technique. For cross-validation, each data point from the experimental data set is eliminated in turn and kriging is used to estimate its value based on the remaining data. Cross-validation criteria, such as the coefficient of determination between measured and estimated values, mean error, sum of squared errors, and reduced kriging error, are calculated based on the experimental data and the kriging estimates (Myers, 1991). During cross-validation, the variogram model parameters are modified, then, kriging with the new variogram parameters is used to estimate the data values, and the cross-variogram criteria are calculated based on the kriging estimates. The variogram model parameters are further adjusted and the process is repeated until the optimal values of the cross-validation criteria are obtained.
Cross-variograms were used to analyze spatial aspects of the relationships between oil or protein concentrations and topography:
 | [4] |
where Subscripts 1 and 2 are related to two different variables, for example, protein or oil concentration and elevation range. The cross-variogram value increases with distance if the variables are positively correlated and decreases with distance if the variables are negatively correlated. The cross-variogram sill for the standardized variables reflects the magnitude of the correlation (Goovaerts, 1997), and its range defines the distance within which the correlation between the variables exists.
The cross-correlogram is another characteristic of the spatial correlation between two variables that has been found particularly useful for comparing spatial variability patterns of different variables (Stein et al., 1997). It is defined as:
 | [5] |
where m1-h and m2+h are the means of Variable 1 and Variable 2 separated by the distance h, and 
21-h and 22+h are the corresponding variances. The cross-correlogram defines the correlation existing between the values of Variable 1 and values of Variable 2 separated by the distance h. At zero distance the cross-correlogram is equal to a Pearson correlation coefficient. Cross-correlogram values at h 
0 depend on the direction of h, and 
12(h) is not equal to 12(-h), because different sets of Variable 1 and Variable 2 values are involved in cross-correlogram calculations in opposite directions. In this study there was no physical basis for assuming that these differences might be caused by anything but random fluctuations, hence, we used an average of 12(h) and 12(-h) values (Goovaerts, 1997). Since a cross-correlogram is essentially a correlation coefficient between sets of Variable 1 and Variable 2 data with data pairs separated by a certain distance, its statistical significance is determined as that of the Pearson correlation coefficient based on the number of data pairs involved in the cross-correlogram calculation at each particular distance.
Geostatistical analysis was performed using the geostatistical software package GSLIB (Deutsch and Journel, 1998) and GS+ Geostatistics for the Environmental Sciences package (Gamma Design Software, Plainwell, MI).
|
RESULTS AND DISCUSSION
|
|---|
Statistical and Geostatistical Data Analyses
Statistical summaries of the soybean protein and oil concentrations are presented in Table 2. The mean protein values were 420 g kg-1 with the minimum protein value of 337 g kg-1 at the WL198 field in 1998 and the maximum observed protein value of 487 g kg-1 at the WL299 site in 1999. The mean oil values were 185 g kg-1 with the highest oil concentration of 211 g kg-1 observed at the KN99 field in 1999 and the lowest of 138 g kg-1 at the WL198 site in 1998. Protein values from the fields sampled in 1999 (KN99, WL299) were higher than those from the fields sampled in 1998 (DL98, RF98, WL198). Oil concentrations were also greater in 1999 than in 1998, although the difference was smaller than that of protein (Table 2).
View this table:
[in this window]
[in a new window]
|
Table 2. Mean and variance for soybean protein and oil concentrations along with parameters of the best fitted variogram model, cross-validation criteria including reduced kriging error (RKE), and coefficient of determination between measured and estimated values (r2), and the number of closest samples used for kriging estimation (N).
|
|
Soybeans of the same cultivar (‘FS 331STS’) grown on the DL98 and RF98 fields in 1998 had similar mean protein and oil concentrations, however, considerable within field variability of protein and oil concentrations was observed in both fields as indicated by large ranges of protein and oil values and high variances (Table 2). For the soybeans of the same cultivar (‘Pioneer 93B51’) grown at the WL198 and KN99 fields in the two different years, not only were wide ranges of protein and oil concentrations observed within the fields, but the mean protein and oil concentration values were substantially different for the two fields (Table 2). These observations indicate importance of environmental factor variations occurring within a field and of year-to-year weather differences for variability of soybean protein and oil concentrations.
Sample variograms had a well defined spatial structure with a sample variogram increasing consistently from its nugget value at the smallest lag distance and reaching the sill within a clearly identifiable range (Fig. 2 and 3)
. Variogram model parameters for both protein and oil concentrations varied widely from field to field (Table 2). There is no information available in the literature on spatial variability of soybean protein and oil concentrations that the authors are aware of. However, it is well known that spatial variability of many other plant characteristics, including crop grain yields (Yang et al., 1998) and nutrient uptake (Borges and Mallarino, 1997), differs from field to field. Hence, the observed variability of variogram parameters among fields was an expected result produced most likely by the differences in spatial variability of field topography and soil properties. No trends in the variogram parameter values related to either the year or the studied variable were observed. Variogram ranges were relatively large, exceeding 70 m for protein and oil concentrations of most of the fields. Both protein and oil concentrations had a relatively high nugget/sill fraction [c0/(c0 + c)] (Table 2), thus indicating that for most of the protein and oil data, a large portion of the data variability existed at distances shorter than the sampling interval used. Large influence of the short range variability component explains relatively poor representation of the data spatial variability by the variogram models as reflected in cross-validation criteria. Relatively high errors between actual data and the cross-validation estimates and low coefficients of determination (r2) were obtained during the cross validation procedure (Table 2).
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 2. Sample variograms and variogram models for soybean protein concentration data of the DL98, RF98, WL198, KN99, and WL299 fields.
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 3. Sample variograms and variogram models for soybean oil concentration data of the DL98, RF998, WL198, KN99, and WL299 fields.
|
|
Factors Affecting Soybean Quality Parameters
Having observed spatial variability in field distributions of soybean protein and oil concentrations, we examined the relationship between soybean growth and topography by analyzing correlation coefficients between topographical attributes (i.e., elevation range, terrain slope, and terrain surface curvature) and soybean protein or oil concentrations. Elevation range was significantly positively correlated (P = 0.05) with protein in three of the studied fields, slope was positively correlated with protein concentrations at the DL98 and WL198 fields, and curvature was positively correlated with protein concentration at the KN99 field and negatively correlated with protein at the WL299 field (Table 3). Oil concentration was less affected by topography than protein concentration. Only at the WL299 field was there a significant correlation between elevation and oil concentration, and curvature and oil concentration. Oil concentration and slope were negatively correlated at the DL98 field. Although correlation coefficients calculated for the whole field data supplied general indications on how protein and oil concentrations were affected by topography, they alone were not sufficient for detailed analysis of the relationships between soybean quality parameters and topography. The correlation coefficients calculated based on the whole field data, in fact, showed a weaker relationship between the studied variables than that observed via analyzing transect data. Similar notice of poor usefulness of correlation coefficients calculated based on pooled whole field data for analyzing the factors affecting crop yields was also made by Miller et al. (1988).
A more detailed and informative view of topography in relation to soybean protein and oil distributions was obtained from the transect data. Figure 4
shows protein data plotted vs. elevation range and terrain curvature for the transects at the DL98 (Fig. 4a), RF98 (Fig. 4b), and WL299 (Fig. 4c) fields. All topographic variables (i.e., elevation range, curvature, and slope) clearly affected protein distribution in most of the studied fields. However, their effects varied from field to field and were different even in different parts of the same field. Moreover, depending on the weather conditions, topographical features produced opposite effects on protein concentrations of different fields. For example, protein distribution along the transect in the DL98 field was similar to the elevation range profile, with higher proteins generally observed at uphill positions and mounds, and lower protein values observed in depressions and downhill positions. Even relatively small mounds (<0.5 m above the surrounding grounds) had noticeably higher protein values than the surrounding depressions (Fig. 4a). A similar effect of elevation on protein (higher protein values at topographically higher locations) was observed for parts of the WL198 and KN99 fields (not shown). On the other hand, an opposite relationship existed at the WL299 field (Fig. 4c), where lower protein concentrations were observed at sites with higher elevation and curvature and higher protein concentrations were found at lower located sites with concave curvatures.
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 4. Protein concentration values plotted along with elevation ranges and curvatures for transects drawn in the north-south direction in the central parts of the DL98, RF98, and WL299 fields.
|
|
In some fields the correspondence between protein concentration and curvature or slope was more prominent than the effect of the elevation range. For example, at the RF98 field (Fig. 4b), higher protein concentrations corresponded to high positive curvatures, or convex terrain surfaces, lower protein concentrations were observed at sites with negative curvatures, concave surfaces, or depressions [r2 for protein and curvature was equal to 0.321 (P = 0.05)]. This trend was also noticeable for the transect from the DL98 field [r2 for protein and curvature was equal to 0.487 (P = 0.01)]. Slope was the main factor affecting protein distribution across the WL198 field [Fig. 5
; r2 for protein and slope was equal to 0.174 (P = 0.01)]. Sites with higher slopes had higher protein concentrations, while low slope sites had lower protein. Analysis of oil distributions along the transects did not indicate any noticeable relationships between oil concentrations and topographical parameters for most of the studied fields.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 5. Protein concentration values plotted along with terrain slope for the transect drawn in the north-south direction in the central part of the WL198 field.
|
|
The differences in topographical effects on protein and oil concentrations observed in different fields can be explained in part by the summer weather patterns. Analysis of the weather data reveals that the KN99 field in 1999 and the WL198 and DL98 fields in 1998 received relatively large amounts of precipitation in the beginning of summer. The cumulative precipitation in the first 2 wk of June was equal to 21 cm at the KN99 field, and to 15.5 cm at the WL198 and DL98 fields. The cumulative precipitation at the WL299 field during 1–14 June 1999 was <2 cm. To compare, the long-term average (1961–1990) precipitation for the whole month of June for the studied fields is 10 cm (Midwestern Climate Center, Champaign, IL). The maximum amounts of precipitation that fell during a 3-d period in the first 2 wk of June were as high as 16.6 cm at the KN99 and 10 cm at the WL198 and DL98 fields. These large rainfall events resulted in temporary ponding and other drainage problems in lower located areas and poorly drained depressions, and presumably affected plant growth and soybean quality at those locations. This may explain the positive correlation between protein concentration and elevation for these three fields (KN99, WL198, and DL98), positive correlations with slope at the DL98 and WL198 fields, the positive correlation with curvature at the KN99 field, as well as positive relationships between protein concentration and elevation, slope, and curvature observed on the transects of these fields. On the other hand, dry conditions at the WL299 field strongly affected plant growth and soybean quality on the higher located sites, resulting in negative correlation between protein concentration and curvature (Table 3) and lower protein concentration values at higher located sites observed on the WL299 transect. The influence of precipitation on how soybean quality and topography were related supports the hypothesis that soybean quality and topography correlations are complex functions of the field water balance. The water balance of each site depends on the amounts of water supplied by precipitation and on the water redistribution in the field following precipitation, the latter largely defined by the field topography and soil hydraulic properties. At present, only field topography is adequately measured in the studied fields. A study that combines analysis of soybean protein and oil concentration data with monitoring soil moisture conditions at the soybean sample collection sites, as well as analysis of soil texture, hydraulic properties, organic matter content, and soil nutrient contents will be necessary to determine the mechanisms regulating soybean protein and oil variability on a field scale.
Spatial aspects of the protein and oil relationships with topography were further analyzed using cross-variograms and cross-correlograms of the whole field data. These are presented along with variograms for elevation range on Fig. 6, 7, 8, and 9
, for the DL98, WL198, KN99, and WL299 fields, respectively. Protein and elevation cross-variograms and cross-correlograms calculated on the basis of the whole field data are shown for the DL98, WL198, and KN99 fields, and the oil and elevation cross-variogram for the WL299 field. For the most of the data, the maximum cross-correlogram values were observed at zero distances. As the distance between the elevation measurement points and protein or oil sampling sites increased, the cross-correlogram values gradually decreased until becoming insignificant (P = 0.05) at a certain distance specific to each field. In some fields, such as DL98 and KN99, the cross-correlograms became significant again but with a negative sign at much larger distances. These negative correlations at large distances result from the correlograms being calculated using elevation data from hill tops and protein or oil data from depressions and vice versa. For the RF98 field, the protein and elevation and oil and elevation cross-correlogram values were statistically insignificant and erratic (not shown). The observed cross-correlograms indicate that the influence of elevation on protein or oil concentrations was not limited to individual sites with high and low elevation, but encompassed the whole areas with either high or low elevation prevailing. That is, higher protein values were observed not only at sites with the highest elevation, but also within a 100- to 150-m distance of these sites. Likewise, low protein values were seen not only at the sites with the lowest elevation but also in the area of 100- to 150-m size surrounding these sites. The distance at which significant correlations between protein or oil and elevation ceased to exist was similar either to the cross-variogram range (DL98 and KN99 fields) or to the range of the first of the cross-variogram nested structures (WL299 field) (Fig. 6–9). The fact that significant correlations between soybean protein and oil concentrations and elevation existed at relatively large distances supports the possibility of differential soybean harvesting for either high protein or high oil, based on the field topography.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 6. Experimental cross-correlogram for soybean protein concentration vs. elevation along with the cross-correlogram significance limit (P = 0.05), experimental cross-variogram for protein and elevation, and variogram for elevation range of the DL98 field.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 7. Experimental cross-correlogram for soybean protein concentration vs. elevation along with the cross-correlogram significance limit (P = 0.05), experimental cross-variogram for protein and elevation, and variogram for elevation range of the WL198 field.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 8. Experimental cross-correlogram for soybean protein concentration vs. elevation along with the cross-correlogram significance limit (P = 0.05), experimental cross-variogram for protein and elevation, and variogram for elevation range of the KN99 field.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 9. Experimental cross-correlogram for soybean oil concentration vs. elevation along with the cross-correlogram significance limit (P = 0.05), experimental cross-variogram for oil and elevation, and variogram for elevation range of the WL299 field.
|
|
The relationships between soybean protein or oil concentrations and elevation of each particular field as well as the potential for differential soybean harvesting largely depend on the field topography. Statistically significant across comparatively large distances, cross-correlograms were observed in the fields where relatively large areas were occupied by both depressions and summits. It was expected that the water availability conditions of these two land forms were relatively homogeneous and represented the two extreme cases of water availability in the field. Such topography was most prominent at the DL98, WL198, KN99, and WL299 fields. The larger and the more homogeneous the summit and depression parts of the field, the larger significant correlation ranges could be expected in the field. Elevation variograms provided approximate indications of how large this range could be for a given field. For the DL98 field, the significant cross-correlogram range was similar to the range of the first nested structure in the elevation variogram (Fig. 6). In the KN99 field, the range of significant cross-correlogram values coincided with the elevation variogram range (Fig. 8). In the WL299 field, the cross-correlogram range corresponded to changes in the elevation variogram slopes (Fig. 9). In the WL198 field, the change in the elevation variogram slope also coincided with a decline in the cross-correlogram significance (Fig. 7).
In the RF98 field, most of the field was represented by a hill's summit and a shoulder slope with only insignificant portions of the field occupied by a depression. Lack of diversity in land forms of this field caused insignificant correlations between protein or oil concentrations and elevation. In this field, the small scale relief played a more noticeable role in providing plants with water supply, as was reflected in the protein and curvature transect of the RF98 field (Fig. 4b).
|
CONCLUSIONS
|
|---|
Soybean protein and oil concentrations were distributed within the studied fields with well-defined spatial structures. Although short range variations of the protein and oil variograms represented a relatively large part of total data variability for most of the studied fields, well-defined spatial structure with sample variograms increasing consistently from the nugget values at the smallest lag distance and reaching the sills within a clearly identifiable range was observed in all of the fields.
Field topography substantially affected soybean protein and oil concentrations, and observed topographical influences were hypothesized to be a result of differences in amount and availability of soil water across the studied fields. Topographical effects on protein and oil concentration distributions of every field were explained using information on prevailing field topography and weather conditions during the growing seasons. Geostatistical analysis of the protein, oil, and topography spatial variability using cross-variograms and cross-correlograms revealed that the spatial structure in the protein and oil concentration distributions was related to the spatial structure of elevation. Particularly, the range of significant cross-correlation between protein concentrations and elevation was similar to either the range of the elevation variogram or to the distance at which there was a change in the spatial structure of elevation represented by a change in the slope of the elevation variogram.
Received for publication March 6, 2001.
|
REFERENCES
|
|---|
- American Association of Cereal Chemists. 1998. Approved methods of AACC. AACC, St. Paul, MN.
- Borges, R., and A.P. Mallarino. 1997. Field-scale variability of phosphorus and potassium uptake by no-till corn and soybean. Soil Sci. Soc. Am. J. 61:846–853.[Abstract/Free Full Text]
- Burton, J.W. 1985. Breeding soybeans for improved protein quantity and quality. p. 361–367. In R. Shibles (ed.) Proc. World Soybean Research Conference, 3rd, Ames, IA. 12–17 Aug. 1984. Westview Press, Boulder, CO.
- Burton, J.W. 1987. Quantitative genetics: results relevant to soybean breeding. p. 211–247. In J.R. Wilcox (ed.) Soybeans: Improvement, production, and users. 2nd ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.
- Burton, J.W. 1991. Development of high-yielding high-protein soybean germplasm. p. 118–122. In R.F. Wilson (ed.) Designing value-added soybeans for markets of the future. AOCS, Champaign, IL.
- Changere, A., and R. Lal. 1997. Slope position and erosional effects on soil properties and corn production on a Miamian soil in Central Ohio. J. Sustainable Agric. 11:5–21.
- Deutsch, C.V., and A.G. Journel. 1998. Geostatistical software library and user's guide. Oxford University Press, New York.
- Eghball, B., and G.E. Varvel. 1997. Fractal analysis of temporal yield variability of crop sequences: Implications for site-specific management. Agron. J. 89:851–855.[Abstract/Free Full Text]
- Environmental Systems Research Institute. 1996. ArcView spatial analyst. ESRI, Redlands, CA.
- Gessler, P.E., O.A. Chadwick, F. Chamran, L. Althouse, and K. Holms. 2000. Modeling soil-landscape and ecosystem properties using terrain attributes. Soil Sci. Soc. Am. J. 64:2046–2056.[Abstract/Free Full Text]
- Gibson, L.R., and R.E. Mullen. 1996. Soybean seed quality reduction by high day and night temperature. Crop Sci. 36:1615–1619.[Abstract/Free Full Text]
- Goovaerts, P. 1997. Geostatistics for natural resources evaluation. Oxford University Press, New York.
- Han, S., J.W. Hummel, C.E. Goering, and M.D. Cahn. 1994. Cell size selection for site-specific crop management. Trans. ASAE 37:19–26.
- Hanna, A.Y., P.W. Harlan, and D.T. Lewis. 1982. Soil available water as influenced by landscape position and aspect. Agron. J. 74:999–1004.[Abstract/Free Full Text]
- Helms, T.C., and J.H. Orf. 1998. Protein, oil, and yield of soybean lines selected for increased protein. Crop Sci. 38:707–711.[Abstract/Free Full Text]
- Jaynes, D.B., and T.S. Colvin. 1997. Spatiotemporal variability of corn and soybean yield. Agron. J. 89:30–37.[Abstract/Free Full Text]
- Journel, A.G., and C. Huijbregts. 1978. Mining geostatistics. Academic Press, New York.
- Kravchenko, A.N., and D.G. Bullock. 2000. Correlation of grain yield with topography and soil properties. Agron. J. 92:75–83.[Abstract/Free Full Text]
- Miller, M.P., M.J. Singer, and D.R. Nielsen. 1988. Spatial variability of wheat yield and soil properties on complex hills. Soil Sci. Soc. Am. J. 52:1133–1141.[Abstract/Free Full Text]
- Moore, I.D., P.E. Gessler, and G.A. Nielson. 1993. Soil attribute prediction using terrain analysis. Soil Sci. Soc. Am. J. 57:443–452.[Abstract/Free Full Text]
- Mulla, D.J. 1993. Mapping and managing spatial patterns in soil fertility and crop yield. p. 15–26. In P.C. Robert et al. (ed.) Soil specific crop management. ASA, CSSA, and SSSA, Madison, WI.
- Myers, D.E. 1991. Interpolation and estimation with spatially located data. Chemom. Intell. Lab. Syst. 11:209–228.
- Or, D., and R.J. Hanks. 1992. Soil water and crop yield spatial variability induced by irrigation nonuniformity. Soil Sci. Soc. Am. J. 56: 226–233.[Abstract/Free Full Text]
- Rahman, S.M., Y. Takagi, and T. Kinoshita. 1996. Genetic control of high oleic acid content in the seed oil of two soybean mutants. Crop Sci. 36:1125–1128.[Abstract/Free Full Text]
- Raper, C.D., and P.J. Kramer. 1987. Stress physiology. p. 590–641. In J.R. Wilcox (ed.) Soybeans: Improvement, production, and users. 2nd ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison,WI.
- Rebetzke, G.J., J.W. Burton, T.E. Carter, and R.F. Wilson. 1998. Changes in agronomic and seed characteristics with selection for reduced palmitic acid content in soybean. Crop Sci. 38:297–302.[Abstract/Free Full Text]
- Sexton, P.J., S.L. Naeve, N.C. Paek, and R. Shibles. 1998. Sulfur availability, cotyledon nitrogen:sulfur ratio, and relative abundance of seed storage proteins of soybean. Crop Sci. 38:983–986.[Abstract/Free Full Text]
- Smith, K.J., and W. Huyser. 1987. World distribution and significance of soybean. p. 1–22. In J.R. Wilcox (ed.) Soybeans: Improvement, production, and users. 2nd ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.
- Spilker, D.A., A.F. Schmitthenner, and C.W. Ellett. 1981. Effects of humidity, temperature, fertility, and cultivar on the reduction of soybean seed quality by Phomopsis sp. Phytopathology 71:1027– 1029.
- Stein, A., J. Brouwer, and J. Bouma. 1997. Methods of comparing spatial variability patterns of millet yield and soil data. Soil Sci. Soc. Am. J. 61:861–870.[Abstract/Free Full Text]
- Stojin, D., G.R. Ablett, B.M. Luzzi, and J.W. Tanner. 1998a. Use of gene substitution to quantify partial dominance in low palmitic acid soybean. Crop Sci. 38:1437–1441.[Abstract/Free Full Text]
- Stojin, D., B.M. Luzzi, G.R. Ablett, and J.W. Tanner. 1998b. Inheritance of low linolenic acid level in the soybean line RG10. Crop Sci. 38:1441–1444.[Abstract/Free Full Text]
- Timlin, D.J., Ya. Pachepsky, V.A. Snyder, and R.B. Bryant. 1998. Spatial and temporal variability of corn grain yield on a hillslope. Soil Sci. Soc. Am. J. 62:764–773.[Abstract/Free Full Text]
- Warrick, A.W., D.E. Myers, and D.R. Nielsen. 1986. Geostatistical methods applied to soil science. p. 53–82. In Klute et al. (ed.) Methods of soil analysis. Part 1. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
- Wilcox, J.R. 1985. Breeding soybean for improved oil quantity and quality. p. 380–386. In R. Shibles (ed.) Proc. World Soybean Research Conference, 3rd, Bangkok, Thailand. 20–26 Feb. 1994. Westview Press, Boulder, CO.
- Wilcox, J.R. 1998. Increasing seed protein in soybean with eight cycles of recurrent selection. Crop Sci. 38:1536–1540.[Abstract/Free Full Text]
- Wilcox, J.R., and J.F. Cavins. 1995. Backcrossing high seed protein to a soybean cultivar. Crop Sci. 35:1036–1041.[Abstract/Free Full Text]
- Wilson, R.F. 1991. Advances in the genetic alteration of soybean oil composition. p. 38–52. In R.F. Wilson (ed.) Designing value-added soybeans for markets of the future. American Oil Chemists' Society, Champaign, IL.
- Yang, C., C.L. Petersen, G.J. Shropshire, and T. Otawa. 1998. Spatial variability of field topography and wheat yield in the Palouse region of the Pacific Northwest. Trans. ASAE 41:17–27.
This article has been cited by other articles:
|
|
|
|
 
D. S. Bullock, N. Kitchen, and D. G. Bullock
Multidisciplinary Teams: A Necessity for Research in Precision Agriculture Systems
Crop Sci.,
September 1, 2007;
47(5):
1765 - 1769.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. B. Tarr, K. J. Moore, and P. M. Dixon
Spectral Reflectance as a Covariate for Estimating Pasture Productivity and Composition
Crop Sci.,
May 6, 2005;
45(3):
996 - 1003.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
J. Iqbal, J. J. Read, A. J. Thomasson, and J. N. Jenkins
Relationships between Soil-Landscape and Dryland Cotton Lint Yield
Soil Sci. Soc. Am. J.,
May 6, 2005;
69(3):
872 - 882.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
B. C. Si and R. E. Farrell
Scale-Dependent Relationship between Wheat Yield and Topographic Indices: A Wavelet Approach
Soil Sci. Soc. Am. J.,
March 1, 2004;
68(2):
577 - 587.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
F. Avendano, F. J. Pierce, O. Schabenberger, and H. Melakeberhan
The Spatial Distribution of Soybean Cyst Nematode in Relation to Soil Texture and Soil Map Unit
Agron. J.,
January 1, 2004;
96(1):
181 - 194.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. N. Kravchenko, K. D. Thelen, D. G. Bullock, and N. R. Miller
Relationship among Crop Grain Yield, Topography, and Soil Electrical Conductivity Studied with Cross-Correlograms
Agron. J.,
September 1, 2003;
95(5):
1132 - 1139.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. N. Kravchenko and D. G. Bullock
Spatial Variability of Soybean Quality Data as a Function of Field Topography: II. A Proposed Technique for Calculating the Size of the Area for Differential Soybean Harvest
Crop Sci.,
May 1, 2002;
42(3):
816 - 821.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
