| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Soil Science, 51 Campus Drive, Univ. of Saskatchewan, Saskatoon, SK, Canada, S7N 5A8
b Dep. of Water and Environment, Alterra, P.O. Box 47, 6700 AA Wageningen, the Netherlands
c Dep. of Agronomy and Range Science, Univ. of California-Davis, Davis, CA 95616
* Corresponding author: (walley{at}sask.usask.ca)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CV, coefficient of variation • HCFS, high catchment footslopes • HCSH, high catchment shoulders • LCFS, low catchment footslopes • LCSH, low catchment shoulders • SNAI, soil N availability index • NMIN, cumulative N released during a 2-wk aerobic incubation • N0, potentially mineralizable N estimated using a 16-wk aerobic incubation • NO3AEM, NO3 sorbed on anion-exchange membranes • NKCl, N extracted with hot 2 M KCl • NHYDR, N hydrolyzed with hot 2 M KCl • SOC, soil organic C
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Over the years, numerous SNAIs, based on either biological or chemical principles (Bremner, 1965; Keeney, 1982), have been proposed. Stanford and Smith (1972) developed a biologically based, long-term incubation method whereby potentially mineralizable N can be estimated using one-pool (Stanford and Smith, 1972), two-pool (Molina et al., 1983), or incremental models (Ellert and Bettany, 1988). Incubation methods are time-consuming by nature, and thus more recent research has focused on the development of more rapid chemical extraction methods, such as the use of hot KCl, for estimating potentially available N (e.g., Gianello and Bremner, 1986; Smith and Li, 1993; Jalil et al., 1996; Curtin and Wen, 1999). Anion-exchange membranes also have been used to estimate soil N availability (Qian and Schoenau, 1995; Ziadi et al., 1999).
Because many of the methods used to estimate N availability measure, in part, the release of N from some component of the soil organic matter pool, various indices have been found to be closely related to total soil organic matter levels (Keeney, 1982; Wang et al., 2001). The quantity and quality of soil organic matter is known to vary within fields, with the most pronounced variability usually occurring on hummocky or rolling terrain (Gregorich and Anderson, 1985; Verity and Anderson, 1990). This variability has been attributed to loss and redistribution of topsoil (Gregorich and Anderson, 1985; Pennock and de Jong, 1990; McConkey et al., 1997) and moisture (Verity and Anderson, 1990) from upper to lower slope positions. It follows that labile fractions of soil organic matter, and thus potentially available N, are likely to vary according to topography. The goal of variable rate fertilizer N application is to correctly predict the ability of a soil to supply N and thereby adjust fertilizer N application rates to reflect the variability in the N requirements of the growing crop.
Interest in identifying suitable techniques for estimating soil N availability for precision farming applications led us to the objective of this study; to assess the relationships between several SNAIs and the yield and total N accumulation of unfertilized wheat grown on a glacial till landscape in the semi-arid region of the Northern Great Plains of America.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A single 300-m north-south transect consisting of 100 sampling points at 3-m intervals was selected. The sampling points and surrounding area were surveyed using a total station laser theodolite (Model Set 5, Sokkisha Co. Ltd., Japan). Elevation data were used to derive a digital elevation model of the surface using 3 by 3 m grid cells following the procedures outlined by Pennock et al. (1987)( 1992) and sample locations were systematically classified as shoulders or footslopes. Shoulders with catchment areas of 9 m2 or less were classified as LCSH whereas shoulders with catchment areas >9 m2 were classified as HCSH. A threshold value of 45 m2 was used in footslopes to distinguish between LCFS and HCFS.
Soil samples were collected on 4 through 5 May 1998. Soil cores, 6.5 cm in diameter, were taken to a depth of 60 cm in three increments (0–15, 15–30, and 30–60 cm). Samples were stored at 4°C. Data relating to the 0- to 15-cm and the 0- to 60-cm soil depths are presented here. Hard red spring wheat (Triticum aestivum var. AC Barrie) was seeded on 15 May 1998 at a rate of 90 kg ha-1 and fertilized with NH4H2PO4, supplying P at a rate of 12 kg ha-1 and N at a rate of 5.5 kg ha-1. No additional fertilizer N was applied. A second strip of wheat fertilized with an additional 70 kg N ha-1 was seeded immediately adjacent to the non-N strip; however, the following discussion is limited to the data from the non-N strip. At crop maturity on 3 Sept. 1998 a 1-m2 yield sample was hand harvested at each sampling point, dried to constant weight at 60°C, weighed, and threshed. Grain and straw yield, and N accumulation were determined.
Soil Nitrogen Availability Indices
We used two biological and three chemical SNAIs including: (i) NMIN, (ii) N0, (iii) NO3AEM, (iv) NKCl, and (v) NHYDR. The aerobic incubation (i.e., NMIN and N0) was carried out using the methodology and leaching apparatus as described by Campbell et al. (1993). Briefly, 100 g of oven-dry equivalent field-moist soil was mixed with 100 g of acid-washed quartz sand for each of the 100 soils (0- to 15-cm depth). The soils were then packed into 30-cm long leaching tubes (4.45-cm i.d.). The prepared soils were preleached under vacuum with 200 mL of 0.01 M CaCl2 followed by 10 mL of a N-free nutrient solution. Leachate was analyzed for mineral N. The final moisture content of each sample was adjusted to 22.5% by weight using deionized water. Samples were similarly extracted every 2 wk during the first 12 wk and again at 16 wk. Leachates were analyzed for NO3 and NH4 using a Technicon Auto Analyzer II (Labtronics Inc., Tarrytown, NY).
Cumulative mineral N data from the 16-wk incubation was used to calculate N0. Both N0 and the rate constant (k) were estimated by means of a nonlinear regression-iteration program, assuming a one-pool model and first-order kinetics as described by Campbell et al. (1993).
Anion-exchange membranes were used to estimate NO3AEM for the three depths in the laboratory, according to Qian and Schoenau (1995). Because NO3AEM is a function of time (2 wk) and the surface area of the anion-exchange membrane (10 cm2), the units of this measurement are expressed as a supply rate of micrograms of NO3 per 10 cm-2 per 2 wk.
A modified method of Gianello and Bremner (1986) was used to determine NKCl. Briefly, 3 g of air-dried soil, ground to pass a 2-mm sieve, was heated in 20 mL of 2 M KCl in a stoppered tube for 4 h at 100°C. Once cooled to room temperature, the solution was decanted and filtered through a Whatman #2 paper (Whatman International Ltd., Maidstone, England), and analyzed using a Technicon Autoanalyzer II System (Labtronics, Tarrytown, NY).
An estimate of NHYDR was calculated by subtracting the quantity of inorganic N extracted with 2 M KCl at room temperature from the amount in the heated extract (Wang et al., 2001).
Standard Soil and Plant Analysis
Soil inorganic N (NO3 and NH4) was determined using standard procedures (Keeney and Nelson, 1982). Gravimetric soil moisture content for each soil depth was determined in the spring prior to seeding. Soil samples (30 g) were oven dried for 24 h at 105°C and the moisture content determined. Total N in the straw and grain, and total soil N and C were determined by dry combustion using a LECO CNS-2000 analyzer (LECO Instruments, Ltd., St. Joseph, MI). Total soil organic C (SOC) was determined using a LECO Carbon Determinator CR-12 (LECO Instruments, Ltd., St. Joseph, MI).
Statistical Analysis
Descriptive statistics of all SNAIs and standard analyses were collected for the different landform complexes, and outliers (defined as values falling outside the average plus or minus three times the standard deviation) were removed. Differences of SNAIs and standard analyses between landform complexes were tested for significance using a one-way ANOVA. Pearson correlations and coefficients of determination between the SNAIs and standard analyses versus crop yield and aboveground N accumulation were calculated and tested for significance.
To test the added value of the SNAIs in explaining variation in N accumulation, compared with the standard analyses, a forward stepwise regression (
enter = 0.15, tolerance = 0.01) was carried out using Systat (SPSS, 1998) for every landform complex and across the entire transect. For every SNAI, the regression was performed with crop N-accumulation as the dependent variable, and all standard analyses (total organic C and N, mineral N and soil moisture for 0- to 15- and 0- to 60-cm depth of A horizon) and the SNAI as independent variables.
Experimental variograms were calculated to quantify spatial variability of the SNAIs and the standard analyses across the transect (Goovaerts, 1997). Cross-variograms between crop N accumulation and these variables were calculated to quantify spatial cross-correlations. Both variograms and cross-variograms were modeled using Variowin (Pannatier, 1996). Parameters of the modeled (cross-)variogram such as nugget (i.e., spatial variance at distances close to 0), sill (i.e., spatial variance at distances beyond spatial correlation), range (i.e., the range of spatial correlation), and relative nugget effect (i.e., the relative size of the nugget as compared with the sill) were used for characterization of spatial correlation (Goovaerts, 1997).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
25% higher in footslopes as compared with shoulders. In contrast, a significant landform pattern was not detected for N0. Relatively high variability in N0 estimates may have obscured any real landform relationships.
|
|
With the exception of N0, SNAIs were significantly correlated with both grain yield and total N accumulation when the entire transect was considered (Table 5). Although N0 was weakly correlated to grain yield (r = 0.208), no significant correlation with N accumulation existed. At the transect level, NKCl and NHYDR had the strongest correlations whereas correlation coefficients were weaker for NO3AEM and NMIN. However, although highly significant (probability levels of 0.01 or greater), all correlation coefficients for the SNAIs were relatively low, with none explaining more than 39% of the observed variability. Interestingly, correlation coefficients associated with relative elevation (Table 4) were similar in magnitude to those associated with many of the SNAIs (Table 5).
|
Coefficients of Determination (r2) for Soil Properties and Crop Nitrogen Accumulation
The SNAIs explained only a small proportion of the variability associated with crop N accumulation (Fig. 1). For the 0- to 60-cm depth, NKCl had the strongest relationship with crop N accumulation, explaining as much as 39% of the variability, although this relationship was only observed in the shoulder segments. For the 0- to 15-cm depth, NHYDR explained 20% of the variability on LCFS and HCFS but failed to explain more than 10% of the variability on LCSH and HCSH. In contrast, when the 0- to 60-cm soil depth was considered, NHYDR explained 20% of the variability on shoulders, and only 10% on footslopes. The variability explained by both N0 and NO3AEM was 10% or less, irrespective of landform position.
Some of the basic soil properties explained as much or more of the variability associated with total N accumulated by the crop than did the SNAIs (Fig. 1). Total SOC (0- to 60-cm depth) explained as much as 53% of the variability in crop N accumulation on LCSH and 29% on HCSH. Total SOC was a poor predictor, however, on footslope segments. Similarly, total soil N (0- to 15- and 0- to 60-cm soil depth) explained 20% of the variability on the HCFS and HCSH segments. These values declined to 10% or lower for the LCSH and LCFS segments. Mineral N (0–15 cm) explained 36% of the variability in crop N accumulation only on the LCSH complex and explained <10% on other landform segments. A higher coefficient of determination was observed for A horizon depth (r2 = 0.43), but only on HCSH segments. Soil moisture was a poor predictor of crop N accumulation at this site.
The stepwise regression analysis of basic soil properties and SNAIs on N accumulation suggest that including the SNAIs in the regression analysis failed to improve the relationship beyond that described by the basic soil properties (i.e., total organic C and N, mineral N, soil moisture, and depth of A horizon) in shoulder segments (Table 6). As a consequence, none of the SNAIs were included in the regression models. Relative elevation similarly failed to improve the relationships (data not shown). For the HCFS, only NMIN improved the relationship and was retained in the regression. Of the SNAIs, NKCl (0- to 15- and 0- to 60-cm soil depth), NHYDR (0- to 15- and 0- to 60-cm soil depth), and NMIN were included in the regression model for the LCFS complex but none of these models had R2 values exceeding 0.33.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Others have reported strong spatial landscape patterns in semi-arid landscapes for various N-cycling processes including denitrification, biological N2 fixation, and leaching (Pennock et al., 1992; Stevenson et al., 1995; Farrell et al., 1996; Stevenson and van Kessel, 1997) and generally have attributed these patterns to topographically controlled moisture redistribution. Several studies, including ours, have shown a clear relationship between quantitatively defined landform segments and spring soil moisture, with spring soil moisture decreasing downslope (e.g., Zebarth and de Jong, 1989; Beckie et al., 1997; Pennock and Corre, 2001).
The relationship between spring inorganic N and topography is less clear. We detected lower levels of inorganic N in the lower landscape segments, particularly when the entire 0- to 60-cm depth was evaluated (Table 2). Although some researchers have reported higher levels of inorganic N in lower landscape positions (Stevenson and van Kessel, 1996), others have failed to detect consistent, temporally stable patterns of inorganic N (van Kessel et al., 1993; Beckie et al., 1997). The differences in the reported observations are not entirely unexpected because soil N availability is related to a host of N-cycling processes and these processes are, in turn, influenced differently by many (and in some cases, independent) controls. Thus, our observation that lower levels of inorganic N were observed in lower landscape positions likely reflects the occurrence of N losses such as leaching or denitrification in these positions. However, in a different year (with a different moisture regime) or at a different site, the moisture levels may be more conducive to net gains of inorganic N (via mineralization) in these positions. Thus, because N-cycling processes are necessarily tied to moisture availability, and are sensitive to absolute levels of available moisture, the extent to which one process or another dominates the net outcome of the N-cycling process is difficult to predict. This is particularly true for semi-arid landscapes where spring soil moisture is largely controlled by snowmelt recharge.
A strong impact of topography on grain yield is commonly observed in semi-arid regions (Stevenson and Van Kessel, 1996; McConkey et al., 1997; van Groenigen et al., 2000) but also has been reported in regions with higher precipitation (Kravchenko and Bullock, 2000). Although higher grain yields often are observed in lower positions, excess rainfall sometimes leads to an inversion of this pattern, because of flooding of the lower areas combined with poor drainage (Lindstrom et al., 1986).
We observed a significant relationship between relative elevation and both grain yield and crop N accumulation with yields and N accumulation both increasing downslope and with increasing catchment areas (Table 4). Differences in relative elevation at the scale encountered at this site are not the causal factor influencing crop yields or N accumulation, per se. Rather, relative elevation stands proxy for an unidentified factor or factors that in turn control crop growth and N accumulation. It is likely that moisture redistribution is one of these factors. By definition, the contribution of growing season precipitation and subsequent moisture redistribution is not known prior to seeding; a time when fertilizer N recommendations and decisions generally are made. Thus, if precipitation and subsequent moisture redistribution is a component of the variability described by elevation, models predicting yield response to fertilizer N should include this factor. Interestingly, when we included relative elevation in a stepwise regression with the basic soil properties and both yield and N accumulation, relative elevation failed to improve the relationship already described (data not presented). Kravchenko and Bullock (2000) similarly observed that although the cumulative effect of topographical features explained between 6 to 50% of the yield variability of corn (Zea mays L.) and soybean [Glycine max (L.) Merr.], the impact of including this parameter with basic soil properties (organic matter, cation-exchange capacity, P, and K) in stepwise regressions was variable. For example, for some fields, topography in combination with basic soil properties explained as much as 78% of the yield variability whereas for other fields, only 10% of the variability was described. They reported that the greatest effect of topography occurred during extreme weather conditions (too wet or too dry) and at locations with extreme topography, which further suggests that moisture redistribution may be of considerable importance. Moreover, although the spatial variability of yields was strongly related to topography, these authors and others (e.g., Timlin et al., 1998) noted that the intra-annual differences in the weather during study periods often result in even larger temporal differences in grain yield. As the vagaries of the weather remain difficult to predict, in particular the precipitation pattern in semi-arid regions, predicting with some degree of confidence both the fraction of N that will be mineralized across the landscape and the crop N demand remains a challenge. It remains to be seen if the contribution of predicted precipitation and subsequent moisture redistribution can be described adequately, given the multitude of direct and indirect responses (i.e., both plant and microbial) that moisture redistribution may elicit.
All the SNAIs, with the exception of the N0, were significantly correlated with total grain yield and total N accumulated in unfertilized wheat when evaluated across the entire transect (Table 5). In addition, the modeled cross-variograms all show a clear, positive spatial correlation between the SNAIs and N accumulation by the crop (Fig. 2; Table 7). The variograms of NO3AEM at both depths showed nugget effects higher than 70%, and the cross variograms of these SNAIs were the only ones with nugget effects. This may indicate that, although NO3AEM is significantly correlated with N accumulation and yield (Table 3), measurement errors were relatively large for this SNAI, in our hands. This might also explain why NO3AEM is not included in the stepwise regression models (Table 6).
Jalil et al. (1996) similarly reported a significant correlation (r2 = 0.79, P < 0.001) between NKCl from 38 different soils and N accumulation by canola (Brassica campestris L.) grown in a controlled-environment chamber. In the same study, N0 was correlated with NKCl, and therefore it is likely that N0 also was correlated with plant N accumulation. Smith and Li (1993) similarly reported far stronger correlations between NKCl and plant N accumulation (r2 = 0.64–0.85, P < 0.001) than those that we observed (maximum r2 = 0.39, P > 0.001); however, the controlled conditions used in these experiments were presumably closer to ideal than those encountered in the field and likely contributed to the strength of their observed relationships. It is noteworthy that the modeled variograms of N accumulation and NKCl (0–15 cm) are both exponential, and are almost identical in nugget effect and range (Table 7). Combined with a cross-variogram showing a clear spatial correlation (Fig. 2), this suggests that the two variables are indeed linked, albeit in a weak fashion due to other influences. However, N0 shows a very short range in both the modeled variogram and cross-variogram (Table 7; Fig. 2), indicating a much weaker correlation with N accumulation.
Soil Nitrogen Availability Indices as Soil Testing Tools
Differences in the degree to which relationships between SNAIs and crop growth are expressed underscores the need to assess the predictive ability of any soil testing diagnostic tool under field conditions. For example, the rate of soil N mineralization depends on soil water availability and thus prolonged water deficiencies in shoulders in semi-arid environments could slow the rate of decomposition in these positions relative to others (Schimel et al., 1991). However, SNAIs typically are determined under optimal moisture conditions in a laboratory and do not reflect inherent soil moisture differences that normally would be encountered in the field. As a consequence, the analytical procedures themselves negate the impact that the varied soil moisture content may have on the release of N, even though ample evidence exists to support the notion that soil moisture shows a high spatial and temporal variability across landscapes (Stevenson et al., 1995; Pennock and Corre, 2001). Therefore, if we are to improve the predictive ability of SNAIs, models must include the impact of soil moisture on N-mineralization and immobilization processes. Unfortunately, the impact of soil moisture on any soil N-cycling process may be very difficult to predict because of the inherent complexity of the N-cycle. For example, footslopes and shoulders are fixed positions in a landscape and thus the difference in the hydrological features at the two landscape positions could, in the long term, lead to permanent adaptations of the microbial community (Robertson et al., 1990; Parkin, 1993). If such fundamental shifts occurred over the long term, it is reasonable to presume that a short term change in the moisture content of the soil in any given landscape position could elicit different N-cycling responses, depending on the dominant microbial population in that particular area within the landscape.
The overall purpose of using the various SNAIs is to estimate the size of the potentially mineralizable-N pool. To date, many of the studies in which the suitability of various methods for estimating the potentially mineralizable-N pool have been examined, have compared various soil testing procedures to each other. Recently, the validity of various references that have been used as benchmarks, with a particular reference to N0 as originally proposed by Stanford and Smith (1972), has been questioned (Wang et al., 2001). Wang et al. (2001) argue that N0 reflects net N mineralization and does not adequately account for N losses or immobilization. Therefore, gross N mineralization, as determined by 15N isotope dilution during a 2-wk incubation period, was proposed as the unequivocal reference for N mineralization ability of a soil. It is not clear why these authors used an unusually long 2-wk incubation for their study given that the validity of the 15N dilution technique rests on the assumption that 15N assimilated by microorganisms does not remineralize during the incubation period. Hart et al. (1994) recommended a 1- to 3-d incubation to prevent significant violations of this assumption. Irrespective of the technique employed, an estimate of gross N mineralization determined under optimum controlled conditions would indeed reflect the maximum potential of the soil to mineralize N. However, because soil N tests are often used as tools for predicting possible crop N deficiencies and determining N fertilizer requirements, it is not clear if gross N mineralization potential is of agronomic relevance, particularly for crops grown in semi-arid environments. Indeed, one might argue that net N mineralization, which indirectly includes a measure of immobilization and N losses, may be a more relevant measure of N availability in an agronomic context.
Correlations Between Soil Nitrogen Availability Indices, Yield, and Total Plant Nitrogen at the Landform Complex Level
Many soil properties, including SNAIs, differed significantly between landscape positions (Tables 2 and 3). Moreover, although significant correlations were detected between these properties and measured crop growth parameters such as yield and N accumulation when the entire transect was considered in the analysis (Tables 4 and 5), similar relationships were not detected within individual landscape segments. Although failure to detect significant relationships within landscape segments may have occurred as a consequence of the reduced number of observations used in the analysis, it is also possible that this observation reflects a real breakdown in the relationships. Assuming the latter, it is clear that we have not yet fully identified the factor or factors controlled by topography that, in turn, control the variables we measured (i.e., grain yield, N accumulation, SNAIs, etc.).
The failure to detect significant relationships between many soil properties and grain yield or N accumulation within landform segments is of particular relevance in terms of soil sampling strategies. Fields that are under site-specific fertilizer management practices are often grid sampled (Robert et al., 1999). Using a grid design, sampling locations do not reflect landform position. However, our study, and others (Pennock et al., 1992; Stevenson et al., 1995; Stevenson and van Kessel, 1997) demonstrate that in semi-arid landscapes the dominant controls on soil and plant processes are related to differences in topography as well as the size of the catchment area. As a consequence, failure to stratify soil sampling by landscape position ignores this important control. Methods such as those developed by van Groenigen et al. (2000) could aid in optimizing sampling schemes for such stratified landscapes.
The lack of detectable correlations between various measures of available N and both grain yield or crop N accumulation within landscape segments suggests that N-cycling processes and plant productivity may be controlled by different sets of factors (Turner et al., 1997). Although the various SNAIs presumably provide a measure of the size of the soil N pool that is potentially available to the growing plant, these estimates typically were inconsistent (or only poorly related) with the N that the plant actually utilized. Total plant N accumulation (or yield) reflects the demand of N over the season, which may well be entirely independent on the rate of soil mineralization. Indeed, our data suggest that in most instances the control that topography had on plant growth and N accumulation was independent of the control that topography had on the various SNAIs because within landform segments (i.e., when the impact of topography is removed) significant relationships between these variables were no longer detected. Significant correlations between relative elevation and both grain yield and N accumulation provide further evidence that unidentified factors controlled by topography exert a considerable control on the productivity potential in different landscape positions and these factors may, in turn, exert independent controls on the variables of interest. For example, it is conceivable that long-term and persistent moisture limitations in upperslope positions limited organic matter inputs and thus reduced the size of the N pool measured using NKCl. However, grain yields may similarly be depressed in upperslope positions but the limitation may not have been related to the size of the N pool but rather some other unmeasured variable (e.g., phosphate) that was similarly controlled by landscape position. In semi-arid regions, water remains the key factor that controls the demand for, and availability of, soil N although other nonnutritional factors such as weed and disease pressure also may play a role (Stevenson and Van Kessel, 1996) and could obscure real correlations between SNAI and yield or total plant N. It is important to note that the degree to which these nonnutritional factors may influence the availability and demand for N may well depend on the relative productivity of a given field. Pennock and Corre (2001), working within the same field at a location immediately adjacent to our study, reported high soil organic matter levels and soil redistribution patterns that were indicative of a relatively high quality soil as compared with others in the same geographic region.
A high degree of spatial variability in soil properties at a small scale would also preclude any significant correlation between SNAIs and basic soil properties with yield and crop N accumulation. In a detailed study by Robertson et al. (1997), the variability of soil resources (moisture, total C and N, inorganic N) and soil microbiological attributes (N-mineralization potential among others) was determined across a 48-ha agricultural field in southwest Michigan. A total of 600 samples or an average of 12.5 per hectare were taken. Using a multiple-regression analysis, remarkably little correlation between productivity of soybean and soil properties could be observed. Spatial variability was high for all soil, plant, and microbial properties with CV's that ranged from 40 to 90%—levels similar to those observed in our study. Although in our study the sampling intensity was much higher (equivalent to 2500 samples per hectare), the ranges of the variogram models for most of the parameters tested in both studies were remarkably similar. For example, the range for N0 in our study was 21 m (Table 7), whereas in Michigan, the range was 28.4 m (Robertson et al., 1997). Because a high degree of variability for all soil resources appears to be the norm rather than the exception, we should not expect a high degree of correlation between any single measure of N availability and either crop growth or N accumulation. It follows that we should not expect these diagnostic tools to predict, with any high degree of accuracy, the real N requirements of a crop in any given year on a point-by-point basis within a field (as is the expectation with grid sampling).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Although significant relationships between crop yield and N accumulation and the various SNAIs evaluated were detected, these relationships were at best weak and typically were detected only when data from across the entire transect were analyzed. These relationships often were not detected within landscape segments, suggesting that the relationships existed largely as a consequence of overriding topographical controls. These observations suggest that grid sampling as a means of assessing fertilizer N requirements remains ill-advised for glacial till semi-arid landscapes.
Of the SNAIs evaluated, plant N accumulation and grain yield appeared most closely correlated with NKCl. However, these relationships were never strong, with NKCl explaining, at best, 39% of the variability associated with plant N accumulation.
These results however, do not invalidate the use of SNAIs for soil test applications, per se. Indeed, several SNAIs (and in particular, NKCl) show promise in terms of describing the size of the potentially available soil N pool. However, knowing the size of the N0 pool across the landscape may be of little significance if we cannot identify more of the factors that control the contribution that this pool makes to crop productivity. Clearly, estimating soil N availability is only of value if these estimates improve our ability to predict crop growth and N demand. In our study, stepwise multiple regression analysis indicated that basic soil properties such as total soil N or total soil C in combination with the thickness of the A horizon were as powerful or more powerful than the SNAIs in explaining yield and N accumulation variability. Although this may bode well for the use of basic soil properties as diagnostic tools, these properties typically failed to explain more than 50% of the yield and grain variability. As Kravchenko and Bullock (2000) concluded, yield variability is caused by a host of factors—the challenge is to identify measurable factors that, in combination, describe an agronomically useful portion of crop variability.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 24, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Sharifi, B. J. Zebarth, D. L. Burton, C. A. Grant, and J. M. Cooper Evaluation of Some Indices of Potentially Mineralizable Nitrogen in Soil Soil Sci. Soc. Am. J., June 8, 2007; 71(4): 1233 - 1239. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. T. Yates, B. C. Si, R. E. Farrell, and D. J. Pennock Wavelet Spectra of Nitrous Oxide Emission from Hummocky Terrain during Spring Snowmelt Soil Sci. Soc. Am. J., May 23, 2006; 70(4): 1110 - 1120. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. L. Mulvaney, S. A. Khan, and T. R. Ellsworth Need for a Soil-Based Approach in Managing Nitrogen Fertilizers for Profitable Corn Production Soil Sci. Soc. Am. J., December 2, 2005; 70(1): 172 - 182. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. W. Johnson, P. S. J. Verburg, and J. A. Arnone SOIL EXTRACTION, ION EXCHANGE RESIN, AND ION EXCHANGE MEMBRANE MEASURES OF SOIL MINERAL NITROGEN DURING INCUBATION OF A TALLGRASS PRAIRIE SOIL Soil Sci. Soc. Am. J., January 1, 2005; 69(1): 260 - 265. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
