| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
a Agriculture and Agri-Food Canada, Lethbridge Research Centre, 5403 1st Ave. S. Lethbridge, AB, T1J 4B1, Canada
* Corresponding author (changc{at}agr.gc.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: Abbreviations: GPS, global positioning system • OC, organic carbon.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Conventionally, the change in the total mass of a given element of soil per unit area is estimated as the difference between the total amounts in a fixed thickness of surface soil at two sampling times. Sometimes the thicknesses of the surface soil layers sampled were determined by profile morphology (e.g., pedogenic A horizons), but often they were fixed. Various researchers (Nye and Greenland, 1964; Jenkinson, 1971) recognized the shortcomings of conventional soil sampling methods, and proposed that rather than basing comparisons on a fixed soil depth or volume, the sampling depths should be adjusted so that comparisons are based on a fixed soil mass. Nye and Greenland (1964) recognized that comparisons between forest and cultivated sites should be based on the same dry soil mass per unit land area (fixed-mass basis) rather than on a fixed soil depth.
Ellert et al. (2001) also advanced a method to assess soil C changes based on a fixed soil mass, and detailed how the sampling might be done to resolve temporal changes while reducing the influence of spatial variations. Gifford and Roderick (2003) proposed a method based on "cumulative mass coordinates," which they claimed was simpler and more accurate than previous fixed-mass methods and would not require explicit measurements of soil core volume or bulk density. It appears, however, that the cumulative coordinate approach is merely another option to calculate C storage within a fixed soil mass. Like previous ones, this option involves interpolation among measured variables to estimate soil C in a "target" or fixed mass. If the cross-sectional area of the soil sample must be uniform with depth, and since soil sample dry mass must be measured accurately, the only additional variable required to estimate sample bulk density is depth. Neglecting estimates of soil sampling depth in some soils, however, will result in a great uncertainty due to the heterogeneous nature of soil with depth.
For a soil system without appreciable net inputs or outputs of material, relative to surface soil mass, any of these fixed-mass methods would be satisfactory. Despite some variations among these methods, the basic principle is the same: determine the amount of constituent contained in a fixed soil mass. When soil bulk density or soil surface elevation is changed by swelling or compaction, the fixed-mass methods should adequately assess net changes in soil constituents. Typically, the net input of organic matter derived from the residues' of situ net primary production is small relative to surface soil mass, and small contributions to soil mass often are neglected in fixed-mass methods. For a more precise estimate of change, more rigorous approaches to mass balance have been described elsewhere (Raats and Klute, 1968; Shurtz, 2003).
Neither fixed-depth nor fixed-mass methods are appropriate for soils with appreciable net inputs or outputs, relative to surface soil mass, of materials such as organic wastes (including livestock manure, sewage sludge, and forestry and food processing byproducts), inorganic wastes, and fills (e.g., drilling mud, dredged sediment) and eroded or deposited soil. In such instances, net inputs or losses after several years may become appreciable relative to soil surface mass. Temporal comparisons among constituents within a fixed-depth and a fixed-mass would either over- or underestimate the changes.
Since neither fixed-depth nor fixed-mass methods are applicable to sites with appreciable changes in soil mass, the objective of this study is to develop a soil sampling method that would be more broadly applicable. Skene (1966) had also pointed out the shortcomings of the fixed depth method for measuring soil C and N changes, and proposed that different soil depths be sampled to account for changes in soil bulk density. In response to this proposal, Henzell et al. (1967) suggested that a surveyor's instrument might be used to directly measure the required difference in soil depth, but they did not explain how. In this paper we describe a method to estimate temporal changes based on sampling, to a fixed subsurface elevation, soil cores that are carefully paired in space.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
 | [1] |
is soil bulk density, v is soil volume, t is time or the sampling date, and i is the depth interval. Assuming soil constituents follow the law of mass conservation, the temporal increase or decrease in total amount of a given element is
 | [2] |
M is the change in the mass of element between two sampling times, Mt1 is the amount initially present (at sampling time t = 1), A is the cumulative additions between sampling times, and R is the cumulative removals or losses. Equation [2] usually applies to elements or constituents that may be transformed to forms that are lost from the soil (e.g., organic matter to CO2 that diffuses to the atmosphere or protein to NO3 that is taken up by plant roots and exported with the harvested plant material). Generally, the additions, A, are more easily determined than the removals or losses, R. Often the recently added elements are more susceptible to removal than those present initially. Therefore, a simpler method is commonly used to determine the change in total amount:
 | [3] |
M, the change in element mass per unit area to a given soil depth. The given depth is the sum of the thicknesses of the depth increments or layers on each particular sampling date.
Methods to estimate temporal changes differ according to the exact depth on which comparisons are based. Sometimes the depth is based on soil profile morphology, and genetic horizons are compared regardless of soil depth. Most often, the soil depth is fixed so that identical soil volumes are compared, regardless of bulk density changes or mass additions. The fixed-mass approach adjusts actual sampling depths (Nye and Greenland, 1964; Jenkinson, 1971) or calculates (via various interpolations) constituent amounts in a soil mass other than that actually sampled (Ellert and Bettany, 1995; Gifford and Roderick, 2003) so that identical soil masses are compared, regardless of soil volume. Since both fixed-depth and fixed-mass approaches may be erroneous, caused by either over- or undersampling (elaborated below) when temporal changes in soil mass are appreciable, we propose elevation-based (to mean sea level) sampling as a more broadly applicable approach. At both sampling times, soil cores are collected to the depth required to reach an identical subsurface elevation where the soil bulk density and volume are beyond the influence of anthropogenic and natural processes, regardless of the thickness of soil overlying the fixed subsurface elevation. By sampling to the depth required to reach a fixed subsurface elevation with high-resolution survey equipment or a global positioning system (GPS), we account for the effects of changes in both soil bulk density and soil mass (attributable to material additions and removals). The fixed-depth method has been widely used worldwide. For example, for soil C accounting, the Intergovernmental Panel on Climate Change (1997) recommended that soil C storage be expressed as the mass of organic C per unit land area to a depth of 30 cm. It has been recognized, however, that such approaches based on fixed sampling depths are valid only when both and v remain unchanged with time. This is most unlikely, however, because and v of shallow soil are very dynamic under natural conditions. Thus, our proposed method might be a viable alternative.
Possible scenarios for temporal dynamics of soil surface elevation and volume are illustrated in Fig. 1 . The initial soil profile is designated as 0. The three possible scenarios of changes in soil profile with time are: (I) no change, (II) an increase in surface elevation, and (III) a decrease in surface elevation. Only in rare instances or short time intervals would Scenario I be encountered. The temporal changes could be inferred directly from analytical concentrations, and approaches based on fixed soil depth, fixed soil mass, and fixed subsurface elevation are identical. Actual soil changes would probably fall under the other two scenarios. Possible causes for Scenario II include deposition of sediments eroded from elsewhere in the landscape, application of amendments such as animal manure, agricultural liming materials, sewage sludge, and other industrial wastes like drilling mud and dredged sediments, and tillage operations. Possible causes for Scenario III include soil erosion, soil compaction (increased bulk density), soil exports with agricultural commodities (e.g., sod farms, tree farms, root crops), and rapid mineral dissolution (possibly caused by acid mine drainage). When comparisons of soil constituents between two sampling times are based on a fixed depth increment, x, the soil will be undersampled for Scenario II and oversampled for Scenario III. To compare the gain or loss of a given element with time, the soil should be sampled from the soil surface to w depth interval for Scenario II and to u depth interval for Scenario III.
|
Another source of error in all methods to estimate temporal changes in soil element mass per unit area is the spatial variability. The error can be minimized, however, by spatial pairing of soil samples collected at successive dates to reduce the contribution of spatial variations to error in the estimation of temporal changes (Ellert et al., 2001). Therefore, as in other method, the elevation-based approach requires spatial pairing of soil samples collected at successive dates. If samples are not closely paired in space, the spatial variability will obscure any temporal changes in soil thickness. The spatial coordinates of the sampling sites and the absolute position of the fixed subsurface elevation can be easily determined either with traditional survey methods or with newer, high-precision GPS methods that require considerably less time. This will easily facilitate spatial pairing of soil sampling.
A Case Study—Temporal Change in Soil Carbon under Heavy Manure Applications
The proposed method is newly developed. No specific experiments were designed to verify this method; however, we used data collected from a long-term experiment with a large amount of animal manure applied to illustrate the importance of the newly proposed soil sampling method.
The research site is located at the Agriculture and Agri-Food Canada Research Centre in Lethbridge, AB, Canada. The area is flat and gently slopes from northwest toward southeast. In 1973, an experiment was initiated to determine the effects of repeated annual application of cattle feedlot manure on the productivity of unirrigated and irrigated soils. The soil is a Calcareous Dark Brown Chernozem (Calcic Haplustoll in the U.S. Soil Taxonomy) with a clay loam texture, developed on glacio-lacustrine parent material. The plots were cropped annually under barley (Hordeum vulgare L.) from 1974 to 1995 and from 1999 to 2005. From 1996 to 1998, canola (Brassica rapus L.) or corn (Zea mays L.) for irrigated and canola or triticale (Triticosecale L.) for unirrigated blocks were grown. All aboveground biomass was removed during harvesting except about 10 to 20 cm of stubble. Details of the research site and the effect of long-term manure amendments on soil chemistry, fertility, P accumulation and balance, and physical properties have been reported (Sommerfeldt and Chang, 1985; Chang et al., 1991, 1993; Chang and Janzen, 1996; Whalen and Chang, 2001; Hao et al., 2003, 2004). Some of the initial and 2003 soil physical and chemical properties are shown in Table 1. The methods used for analysis can be found in those studies.
|
When the experiment was established, recommended manure rates were 30 Mg ha–1 yr–1 (wet weight) for unirrigated soils and 60 Mg ha–1 yr–1 (wet weight) for irrigated soils (Alberta Agriculture, 1980). In areas of intensive livestock production, however, such as the county of Lethbridge in Alberta, actual manure applications often exceed recommended rates because the area of cropland available within an economic hauling distance is limited. Thus, the high manure application rates used in this study may represent actual practices in many areas. The manure came from an open, unpaved commercial cattle feedlot. It did not contain bedding, and was stored for 1 to 2 yr before application. The feedlot cattle manure during 30 yr had an average total C content of 195.6 ± 12.1 g C kg–1, total N of 16.6 ± 0.8 g N kg–1, and total P of 6.4 ± 0.3 g P kg–1 (dry-weight basis). The average water content of the manure was 865 ± 10 g water kg–1 dry manure. Since the feedlot was unpaved, the manure contained appreciable ash, attributable to soil removed from the feedlot during cleaning. Manure application tended to decrease soil sand contents, because the manure contained more silt- and clay-sized mineral particles than the soils receiving the amendment (Gao and Chang, 1996).
Manure and Soil Analysis
Manure and soil samples have been collected periodically since the experiment was established in 1973. Eight manure samples were collected annually during the time of application. Soils were sampled in the fall of 1973 and 2003 to a fixed depth of 150 cm with layers of 0 to 15, 15 to 30, 30 to 60, 60 to 90, 90 to 120, and 120 to 150 cm. Initially the surface soil (0–15 cm) had a clay loam texture (39% clay and 39% sand) and had an organic carbon (OC) content of 1.54 ± 0.022 g C kg–1 (Table 1). Soil bulk density, typically to a depth of 60 cm, was determined in 1973, 1984, 1998, 2000, and 2003.
Cumulative Organic Carbon Applied with Manure, and Temporal Change in Soil Carbon
The quantities of OC applied annually through cattle feedlot manure were calculated based on the amount of applied manure dry weight per area and OC concentration in the manure each year. The cumulative manure OC applied was the sum of OC applied during the years of the study period.
The amount of stored soil organic C was estimated for the initial sampling date in 1973 (t = 1) and for the subsequent sampling date in 2003 (t = 2) as specified in Eq. [3]. The soil volume was based on the thickness of soil above a fixed subsurface elevation, and a unit area (typically the cross-sectional area of the soil core sampled). The temporal change in soil organic C storage to the fixed subsurface elevation was estimated by difference, as given in Eq. [2]. Ideally, surface elevations at each sampling point would have been measured in 1973 and 1973 soil samples would have been collected to the fixed subsurface elevation for each sampling point, and then the procedure repeated in 2003. Since this was not done, we performed a retrospective analysis of the proper soil depths required to reach a fixed subsurface elevation, based on a surface elevation survey conducted in 2005.
Surface Elevation of the Experimental Area
Soil surface elevations and coordinates at the four corners of each replicate plot, plus two points within each plot were determined in 2005 using a real-time kinematic GPS (System 1200 & SmartStation, Leica Geosystems AG, Heerbrugg, Switzerland). The accuracy of the GPS measurements was ±1 cm or better for latitude, longitude, and elevation measurements. To generate the 2005 contour map (Fig. 2
), the elevations were interpolated and smoothed using the LOESS method based on linear regression (first degree polynomial) locally weighted according to the tricube function (Cleveland, 1979; Cleveland et al., 1988).
|
|
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The mean annual rates of manure application, on a dry-weight basis, varied among years by <1% (Table 2). The cumulative amounts of manure dry matter applied from 1973 to 2002 ranged from 483 to 2719 Mg ha–1, depending on application rate (Table 2). In 2003, after 30 yr of manuring at a rate of 180 Mg ha–1 yr–1, the soil bulk densities were 0.79, 1.12, and 1.37 g cm–3 for the 0- to 15-, 15- to 30-, and 30- to 60-cm layers (Table 1). To illustrate the potential effect of manure application on soil surface elevation, we used a mean bulk density of 0.96 g cm–3 for the 0- to 30-cm soil layer, and made the oversimplifying assumption that the applied manure did not decompose or move laterally or vertically. For the 180 Mg ha–1 application rate, the cumulative dry matter input of 2719 Mg ha–1 (Table 2), reaching a soil bulk density of 0.96 g cm–3, produced an increase in surface elevation of 28.3 cm. Of course, since the manure decomposed and dissipated, the increase in surface elevation would be less than this maximum hypothetical value. How much less, however, is not immediately apparent, but we suggest it might be measured using high-precision GPS.
|
|
The contour map indicated general landscape patterns, and was not appreciably influenced by localized increases in elevation associated with heavy manure application (Fig. 2). The contour map indicated that natural variations in surface elevations on the unirrigated half of the experimental field (>100 cm) were much greater than those of the irrigated half. Consequently, the 2005 elevations of the unmanured plots on the unirrigated portion were unreliable estimators of 1973 elevations for adjacent plots that had been heavily manured. On the irrigated portion, the soil surface generally sloped down from north to south (67-cm difference) and the changes in surface elevation were much gentler in the east–west direction (14-cm difference). Consequently, for the irrigated portion of the experimental area, we assumed that 2005 surface elevations of unmanured plots would provide reasonable estimates of 1973 surface elevations of heavily manured plots immediately east or west of the plots with zero manure rates.
Elevation changes due to manure application were estimated by comparing the 2005 surface elevation of manure-amended plots with control plots, i.e., the difference between the averages of surface elevation of 180 Mg ha–1 yr–1 treated plots and that of control plots along east–west transects. The estimated increase in elevation after 32 yr annual applications of 180 Mg ha–1 yr–1 was about 24 cm. If this increase was linear over time, in 2003 after 30 applications the elevation increase would have been 22.5 cm, which is 5.8 cm less than the previous estimate (28.3 cm) assuming manure did not decompose.. Therefore, with manure decomposition and other losses the estimated increase in surface elevation of 24 cm would not be out of line (Fig. 4). The soil bulk density and OC content results indicated that long-term manuring had minimal effect on soil at the depth below 30 cm of original profile (Fig. 4). Therefore, a fixed depth of 60 cm was used for this study. The fixed 0- to 60-cm layer contained 7950 Mg soil ha–1 in 1973, but by 2003 this had decreased to 6975 Mg ha–1 because bulk density decreased. The fixed-mass approach implies that the 0- to 60-cm soil layer in 1973 should be compared with a 0- to 66.9-cm soil layer in 2003 (both containing 7950 Mg soil ha–1; Fig. 4). The total soil mass in 2003 was 10359 Mg ha–1 from 0- to 84-cm depth by the fixed-elevation method (Fig. 4). The difference in soil mass between 1973 and 2003 by the fixed-elevation method was 2409 Mg ha–1, which was close to the total cumulative amount of 2719 Mg ha–1 of manure dry matter added between the two sampling times. Neither fixed depth nor fixed mass account for this soil mass gain. The difference between the estimated soil mass increase and cumulative amount of manure applied was due to the decomposition of organic matter in the cattle manure.
For OC storage, a 60-cm soil layer in 1973 must be compared with an 84-cm soil layer after 30 yr of manuring at 180 Mg ha–1 yr–1 (we assumed the effects of change in elevation due to the last 2 yr manuring on total OC was small because the OC content of soil at the bottom depth was small). The total OC increase from 1973 to 2003 was 337 Mg ha–1 (Fig. 4) or 63.3% of the total applied OC (532 Mg ha–1). Without accounting for the surface elevation increase, the fixed-depth sampling method underestimated the OC increase by 23 Mg ha–1. For the fixed-mass method, the OC increase was 320 Mg ha–1, which is about 16 Mg ha–1 less than the increase for the fixed subsurface elevation. Regardless of whether temporal OC changes are estimated on the basis of fixed soil depth or fixed soil mass, however, they would be underestimated because, after 30 yr of manuring at 180 Mg ha–1 yr–1, 2719 Mg ha–1 of dry matter had been applied to the soil (Table 2). The extent of these underestimations, relative to the approach based on fixed subsurface elevations, depends on the C contents and bulk densities of the subsurface layers. Underestimates would be much greater when the soil sampling depth was shallower or at sites with deep soil profiles and higher C contents at depth. Fortunately for resolving the fate of applied organic wastes, the constituent dynamics in the surface layers probably will dominate estimates of temporal change. This study clearly demonstrated the merits of calculating C storage (could be applicable to other soil constituents) in soil layers extending from the surface to a fixed subsurface elevation to better resolve temporal changes in systems with appreciable changes in soil mass. Subsurface elevations may vary with the contour of the landscape at the initial sampling time, but at each soil sampling location, subsurface elevations should be fixed across sampling times. The subsurface elevation should be below the layers where soil bulk density is influenced by natural processes and anthropogenic activity.
In conclusion, we suggest that detailed elevation data and the coordinates of the soil sampling locations should be recorded when new field experiments are established, especially when substantial net additions to or removals from the soil are expected. This information will be useful to pair temporal comparisons near the sample location, and to adjust sampling depths for changes in soil mass and bulk density. To account for soil C in the context of climate change, soil sampling should include determination of latitude, longitude, and elevation for each sampling point. With the availability of inexpensive and accurate GPS now or in the near future, this should be a realistic option for soil C accounting. The precise elevation measurements that are crucial for our elevation-based method require GPS equipment with subcentimeter accuracy. Currently such equipment still is relatively expensive and complicated to operate; however, both problems will diminish with the continued development of GPS.
Received for publication January 13, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. B. Wuest Correction of Bulk Density and Sampling Method Biases Using Soil Mass per Unit Area Soil Sci. Soc. Am. J., January 21, 2009; 73(1): 312 - 316. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Tian, T. C. Granato, A. E. Cox, R. I. Pietz, C. R. Carlson Jr., and Z. Abedin Soil Carbon Sequestration Resulting from Long-Term Application of Biosolids for Land Reclamation J. Environ. Qual., January 13, 2009; 38(1): 61 - 74. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
