| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Agriculture & Agri-Food Canada Research Centre, P.O. Box 3000, Lethbridge, AB, Canada T1J 4B1
* Corresponding author (ellert{at}em.agr.ca)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Recent strategizing to curb emissions of CO2 to the atmosphere has generated interest in the prospects of managing soil C to promote sequestration as an interim measure to offset CO2 emissions from human activities (Sampson and Scholes et al., 2000). Such strategies imply that recent increases in sequestered soil C might acquire economic value above that associated with plant productivity and ecosystem integrity. To verify the efficacy of any scheme to sequester soil C, however, reliable methods are essential to measure whether temporal changes in C storage have actually occurred (Post et al., 1999).
From the outset, it was recognized that CO2 mitigation strategies based on terrestrial C sequestration must involve vast areas managed so that atmospheric CO2 removal exceeded plant and soil C decomposed back to CO2. In an early calculation, Dyson (1977) estimated that anthropogenic C emissions of 4.5 Pg yr-1 (i.e., 4.5 billion Mg C) could be offset by planting trees (rapidly growing with net C fixation of 6.7 Mg C ha-1 yr-1) on a massive scale (6.7 x 108 ha). Recent estimates suggest that potential rates of net soil C sequestration over more extensive areas of cropland might be on the order of 0.2 to 0.5 Mg C ha-1 yr-1 (Sampson and Scholes et al., 2000). Thus, such small increases in soil C, relative to amounts already present, must extend over vast areas to influence appreciably atmospheric CO2.
Plant and soil sampling to measure directly terrestrial C sequestration can represent only a tiny proportion of the vast land area required. Consequently, some form of scaling-up is required to estimate C sequestration over large areas (Burke, 2000). Soil-based approaches typically integrate various pieces of information, such as: (i) temporal changes in soil C at single points, (ii) spatial variations in soil C distribution and associated cycling processes within landscapes, (iii) geographic data on key variables such as, land use, plant cover, soil properties, and climatic regime. Here we deal with the first piece of information, while recognizing that spatial variations within landscapes, and among regions, biomes, and areas of contrasting land use also must be considered to estimate soil C sequestration for any meaningful land area (Janzen et al., 2001). Point measurements of temporal changes in terrestrial C storage are prerequisite for understanding net land-atmosphere C exchange over more extensive areas.
A major obstacle to measuring temporal C changes is the large quantity of C in most topsoils, relative to annual plant C inputs and CO2 outputs (Ellert et al., 2000). Consequently, any imbalance between inputs and outputs must be large enough or persist long enough to produce detectable changes in soil C storage. It also follows that methods to determine temporal changes must cope with soil variability if such changes are to be resolved against a relatively large background of topsoil C present originally (Post et al., 1999). In response to these obstacles, we proposed a high-resolution method which measures temporal changes in soil C storage by comparing the quantities in soil cores collected from a 4 by 7 m microsite at two sampling times separated by 4 to 8 yr (Ellert et al., 2000). Volumetric soil samples are collected, but the technique depends on subsequent adjustment of soil thickness so that equivalent soil masses are compared. Previously we observed that traditional corrections for bulk density differences often violated the law of mass conservation and led to erroneous comparisons (Ellert and Bettany, 1995). Our objective in this paper is to evaluate the proposed technique by measuring the recovery of a known amount of coal C added to the soil.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
A randomized complete block design was used with three replications and two treatments (none and coal), and each plot was sampled before and after treatment application (precoal and postcoal). The plot dimensions were 4 by 7 m to represent the microsite described in Ellert et al. (2000), and buffer areas between plots were 1.5 to 2.2 m wide. Within each plot, twelve soil coring points were identified at the intersections of a 1-m grid, with six cores collected before and six collected after we applied a precisely known amount of C in the form of coal (Fig. 1) . We selected coal dust as the C source, because it was a conveniently accessible form of organic C that resisted decay under the low temperatures and short duration of this study. Thus, rather than imposing some management treatment, and waiting years or decades for net C storage to change, we used a single addition of recalcitrant C and a short sampling interval. Since all added coal C remained in the microsite, it was unnecessary to calculate net C addition as the difference between unreliable estimates of plant C inputs and CO2 outputs over several years. The coal dust, obtained from a mine near Edmonton, Canada, was screened to pass 1-mm apertures, thoroughly mixed, and weighed into several lots of 700 g each. Coal was applied at 3.64 Mg C ha-1, roughly the amount that might be gained over 10 yr by soil C-conserving practices (Sampson and Scholes et al., 2000).
|
8 cm, using a tandem offset disk to incorporate the crop residues and homogenize the topsoil. Then the soil was repacked by several passes, in alternating directions, with a packer consisting of closely spaced cast-iron wheels. A truck-mounted hydraulic coring device was used to extract six initial, precoal cores at predetermined locations in each plot. Each core was 6.7 cm in diameter, and was sectioned into 10-cm increments to a depth of 50 cm. The six cores were kept separate to assess within plot variability. Electromagnetic markers were placed at the bottom of selected core holes to pinpoint sampling locations and enable close interspersing of initial and subsequent cores (Fig. 1). To prevent topsoil burial, core holes were refilled with soil cores from outside of the plots. Coal dust was applied to three replicate plots after the initial soil cores had been collected. To ensure uniform application, each 700-g lot was transferred to a jar with a perforated lid and sprinkled over an area of 1 m2 delineated by a plywood template. The contrast between black coal dust and light-colored crop residue helped to achieve a uniform distribution. To incorporate the coal dust, the entire study area was tilled to a depth of 13 cm, using a rotary tiller, followed by repacking as described above. The rotary tiller thoroughly incorporated the coal dust to prevent removal by wind, and caused minimal lateral movement of soil beyond the plot borders. After repacking, we collected a subsequent set of cores at six locations, interspersed with the initial cores, within each plot (Fig. 1). Altogether, 360 soil samples were collected, as follows: two treatments, coal and none; three replications; six cores per plot; five soil layers per core, in 10-cm increments to 50 cm; two sampling times, initial (precoal) and subsequent (postcoal).
Sample Preparation and Analyses
Immediately after soil collection, we recorded the field-moist weight of each sample, oven-dried a subsample at 105°C to determine gravimetric moisture content, and air-dried the rest (20°C). We calculated soil density as the oven-dry sample mass divided by the volume sampled. Since the soils were free of stones, corrections for rock fragments were unnecessary. To prepare the soils for analysis, each entire sample, including plant residues, was crushed to pass 2-mm in a perforated drum grinder (capacity 5.424 L). Thus all above and belowground plant residues that had been incorporated into the soil were regarded as soil C. After collecting in a tray below the perforated drum, the soil was thoroughly mixed and a representative subsample (8 mL) was retained for fine grinding. To further reduce particle size (100 mesh sieve; 0.15-mm apertures), subsamples were enclosed, along with two tumbling bars, in stainless steel canisters (capacity 0.253 L, no perforations), and tumbled on a roller bed for 16 h. Each of our roller beds held 104 sample canisters, and was similar in design to that described by Smith and Um (1990).
Total soil C and N were analyzed by combustion-gas chromatography using a CN analyzer (NA-1500, CE Instruments, Milan Italy) after the samples had been weighed into Sn capsules (Baccanti et al., 1993). Organic C was analyzed similarly, except soils were weighed into Ag capsules and acidified to eliminate carbonate as CO2. The Ag capsules resisted corrosion by 6 M HCl, which was added carefully in 20-µL aliquots to prevent soil loss by effervescence, eventually reaching a final volume of 140 µL. After acidification, samples were dried for 24 h at 60°C to drive-off water and HCl, and then the capsules were crimped closed and analyzed (Baccanti et al., 1993). Soil organic C detected by the CN analyzer was expressed as a percentage of sample mass originally weighed into the Ag capsule. Thus, weight changes caused by acidification (weights typically increased in proportion to carbonate contents, as the chloride salts formed were heavier than the carbonates eliminated) had no bearing on the organic C measurements. This analytical method for soil organic C is similar to that described by Nieuwenhuize et al. (1994).
Statistical Analyses
Statistical analyses were performed for three soil constituents (organic C, total C, and total N) expressed in two ways (cumulative amounts, in Mg ha-1, per layer of fixed soil volume and per layer of equivalent soil mass) for five successive layers. Since the amounts of each constituent were calculated cumulatively for increasingly thick sampling depths, separate statistics were calculated for each successive depth. Analyses also were performed for soil mass in a fixed volume to 50 cm, and for soil thickness to attain an equivalent mass of 7000 Mg soil ha-1. A mixed linear statistical model was used to analyze each variable using a split-plot design, with coal treatment (coal vs. none) as the main plot factor and sampling time (initial vs. subsequent) as the subplot factor.
The ‘Mixed’ procedure in SAS was used to perform all statistical analyses, because it permitted the fitting of various error structures, and it correctly handled all comparisons among interaction means in the split-plot analysis (SAS Institute, 1999). Contrast statements were used to separate significant interactions into spatial comparisons within sampling times, and temporal comparisons within coal treatments. Although the experiment had been laid out as a randomized complete block design, the main-plot portion was analyzed as a completely randomized design after it was established that the block effect was not significant. This increased the error degrees of freedom for both the main-plot and contrasts. Within and between plot variability was calculated for each coal treatment, sampling time, and cumulative depth.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Several cores were required to estimate precisely the mean for each plot, because variability within plots was high relative to that among plots. Variability among plot means increased appreciably when even one of the six cores was randomly eliminated from the data. Since the minimum detectable change in soil C increases with variability, sufficient samples should be collected to precisely estimate plot means. At our site, six cores adequately encompassed within-plot variability, and paired temporal comparisons (shown later) improved the detectability of soil C change by reducing the influence of among-plot variability. At sites with greater among- than within-plot variability, detection limits may be lowered by using larger plots (and increasing the number of cores representing each plot), by increasing replication at the plot level, and perhaps by blocking to account for spatial trends.
Organic Carbon in Successive Soil Layers of Fixed Volume
Quantities of soil organic C stored in a fixed sample volume were calculated as products of C concentration, soil thickness, and density for successive layers under contrasting treatments (coal and none) at two sampling times (initial and subsequent) (Fig. 3a)
. Coal application increased soil C storage, but only in the two upper-most layers, because coal incorporation did not extend below 20 cm. Much more subtle than the effect of coal in the top two layers was the tendency for greater C storage at initial relative to subsequent sampling times (all treatments and layers, except where masked by added coal in the top two layers; Fig. 3a). Slightly greater C in fixed soil volumes at the initial sampling time was largely attributable to greater bulk densities.
|
Organic Carbon in Successive Soil Layers of Equivalent Mass
Unambiguous comparisons among soil C storage require consideration of the masses of soil involved (Ellert and Bettany, 1995; Yang and Wander, 1999; Ellert et al., 2000). Without details on soil erosion or deposition, comparisons among soils of stored C (or other constituents) should be based on an equivalent soil mass, otherwise it is unclear whether calculated differences represent actual C changes or haphazard differences in soil mass (as defined by soil density and depth at sampling). Variations in soil mass per volume obscure changes in soil C storage, because C mass depends on soil mass. Rather than providing clarity, traditional ‘corrections for bulk density’ obscure soil C comparisons, because ‘corrected’ values reflect differences in both C content and soil mass. Assuming that soil mass, rather than volume is conserved, proper corrections for bulk density require estimates of C stored in an equivalent soil mass. Even seemingly minor differences in bulk density may distort calculated changes based on comparisons of C in a fixed volume containing variable masses of soil.
To account for unequal soil masses or densities, we calculated organic C stored in successive layers of 1500 Mg soil ha-1 to a cumulative soil mass of 6000 Mg ha-1, plus a fifth layer of 1000 Mg ha-1 to approximately reach 50 cm. The calculation arithmetically resorts the vertical distribution of soil and C to estimate quantities in layers of equivalent soil mass (Fig. 4) . Soil C stored in a layer of equivalent mass includes C in the sampled layer, minus that in soil used to reach equivalent mass in overlying layers, and plus that in soil from lower layers required to attain equivalent mass. For example, the second layer of equivalent soil mass in the coal-amended plots at the subsequent sampling time extends, on average, from 12.3 to 24.4 cm, and includes soil from both the 10- to 20- and 20- to 30-cm volumes that were sampled (Fig. 4).
|
Qualitatively, the distribution of organic C storage in four successive layers of equivalent mass is roughly comparable with that for the fixed sample volume (compare Fig. 3a and b). Again, the most obvious treatment effect is an increase in soil C storage at the subsequent sampling time in the two uppermost layers of coal-amended plots (Fig. 3b). Quantitatively, however, the first two layers of equivalent mass contained more soil and C than corresponding layers of fixed volume, while the opposite was true for the deepest layer. Apart from the effect of coal on C in the upper two layers, there were no consistent differences between C storage in layers of equivalent soil mass at the initial and subsequent sampling time.
Spatial Comparisons to Assess Carbon Recovery
Recovery of organic C added to soils as coal was measured by comparing soil organic C storage in soils with and without added coal. The experiment enabled two such comparisons: a spatial comparison between coal and none treatments at the subsequent sampling time; and a temporal comparison between initial and subsequent samples from the coal treatment. In addition to these, comparisons between coal and none treatments at the initial sampling time, and between initial and subsequent samples from the none treatment (theoretically, both should yield differences of zero) served as further checks to assess variability without coal. Statistical analyses indicated that the treatment by time interactions were significant for soil organic C. Specifically, spatial differences were significant only at the subsequent sampling, whereas temporal differences were significant only in the coal plots.
Even before coal was applied, plots assigned to the coal treatment appeared to contain slightly more C than those of the none treatment (Table 1). Although not statistically significant, mean spatial differences between coal and none plots at the initial sampling were appreciable, ranging from 0.17 to 1.88 Mg C ha-1, and amounting to 5 to 52% of the coal C added subsequently. This lateral and vertical variability hampered the detection of the 3.64 Mg C ha-1 added as coal dust, an amount appreciably less than the mean spatial differences between coal and none plots below the thinnest layer at the subsequent sampling time. At this time, variability for the two thickest layers increased to the extent that differences between coal and none plots (though larger than those for the thinner layers) were not significant at P > 0.10 (Table 1). Thus the spatial comparisons were imprecise and overestimated the coal C actually applied.
|
|
40 Mg ha-1 less soil compared with the none plots, a difference established in the first layer that persisted throughout the entire sampling depth.
Temporal Comparisons to Assess Carbon Recovery
Temporal differences in cumulative organic C storage were smaller and more precise than spatial differences (Table 1). Deviations from zero were smaller for temporal comparisons of the none treatment than for spatial comparisons at the initial sampling time. Similarly, deviations from added coal C (3.64 Mg ha-1) were smaller for temporal comparisons of the coal treatment than for spatial comparisons at the subsequent sampling time. With temporal comparisons, the effect of spatial variability was diminished by the split-plot analysis with sampling time as the subplot factor (i.e., temporal comparisons, essentially, were paired within plots). Even with this improved sensitivity, temporal differences for the none treatment remained insignificant (P > 0.27) and small (Table 1). The absolute value of the largest mean difference amounted to <15% of added C in the coal treatment (Fig. 5)
and <1% of cumulative organic C stored to a soil depth of 50 cm.
|
Recovery of Added Coal as Soil Nitrogen, Organic Carbon, and Total Carbon
The 700 g m-2 of added coal amounted to 3.64 Mg C ha-1 (standard error was 0.03 for n = 18), and 0.0559 Mg N ha-1 or 55.9 kg ha-1 (standard error was 0.9 kg ha-1 for n = 18). The procedure used to eliminate soil carbonate did not affect the C content of coal, indicating that the coal contained negligible inorganic C. The coal C input represented that which might be sequestered during 10 yr of soil C conservation, whereas the N input represented that which might be added as fertilizer in a single year. Added coal C was small relative to background levels in the soil (added C <6% of organic C and <2% of total C to the depth required to attain 7000 Mg soil ha-1) (Fig. 6)
.
Since the C/N ratio of the coal was 65, added coal N amounted to an even smaller proportion of soil N (added N <1% of soil N in 7000 Mg soil ha-1) than the proportions of organic and total soil C added as coal. Consequently, coal addition had no discernible effect on soil N storage, even in the surface layers (Fig. 6).
Recovery of added coal (i.e., the proportion of added C detected by soil sampling and analysis) was incomplete unless temporal comparisons of soil C storage were based on an equivalent soil mass (Fig. 5). To resolve small changes in C against a much larger background of indigenous soil C (Fig. 6), it was imperative to correct for differences in soil mass. Considering the entire 50-cm layer sampled for the coal plots, the temporal difference in soil mass was only 182 Mg ha-1. Even at the minimum mean organic C content (4.4 g kg-1), however, this mass difference still amounted to 0.8 Mg C ha-1 or more than 20% of the added coal C. Although the thickness difference required to attain an equivalent soil mass of 7000 Mg ha-1 was only 1.2 cm, such small adjustments were crucial to the quantitative recovery of the added coal C.
Variability in cumulative soil C increased, roughly in proportion to mean C storage, as increasing amounts of soil were considered (Fig. 6). Thus with increasing soil thickness or mass, variability tended to obscure temporal differences associated with coal addition. Despite this, mean quantities of both organic C and total C were greatest for the coal plots at the subsequent sampling time (Fig. 6). The recoveries of added coal as total soil C were more variable than recoveries as organic C (Fig. 5), because the quantities of total soil C were 2 to 3.5 fold those of organic C, depending on the layers considered (Fig. 6). In coal-amended plots, total C storage was significantly greater at subsequent compared with initial sampling for the first two layers (P < 0.01 for the 0–1500 Mg soil ha-1; P < 0.05 for the 0–3000 Mg soil ha-1), but differences were insignificant (P > 0.15) for greater soil masses.
The amount of added coal C was too small to resolve against the relatively large quantities of total C already present. From these data, it follows that a given change in soil C will become increasingly difficult to detect as soil C storage increases, and as soil depth or mass increases. Thus, it likely will be more difficult to detect a given C change in a C-rich than in a C-poor soil. Similarly, the C change will be more difficult to detect when it is distributed over a soil depth of 1 m, compared with a depth of 0.5 m. Although added C in this study was confined to the surface layers (0–20 cm), it remained detectable as organic soil C over the entire sampling depth. This suggests that the proposed method may be applicable to situations where changes in C storage involve layers of 50 cm or more.
Value Selected for Equivalent Soil Mass
The basis for defining the equivalent soil mass as 1500 Mg ha-1 was somewhat arbitrary, although the resulting layers roughly corresponded with the five volumetric layers that were sampled (Fig. 4). To assess the effect of the value selected for equivalent mass, a nonstatistical comparison was done for values of 1000, 1500, or 2000 Mg soil ha-1. The first value involved extensive reshuffling of soil and associated C among seven layers compared with the five volumetric layers sampled originally (Fig. 7)
. For the calculation based on an equivalent mass of 2000 Mg soil ha-1 (three layers, plus a fourth layer of 1000 Mg soil ha-1 to correspond roughly to the mass sampled), it was assumed that the soil cores had been sectioned into the following increments: 0–20, 20–40, and 40–50 cm. To combine the first four layers sampled into two 20-cm depth increments, bulk density was calculated as the arithmetic mean, while C concentrations were calculated as means weighted according to bulk densities.
|
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This experiment showed that our microsite approach is useful to quantify small temporal changes in soil C storage, such as those attainable in programs implemented to sequester soil C. The approach may be used both in monitoring programs where changes in C storage are measured at specific sites in farm fields, and in research programs where conventional approaches may lack the sensitivity required to resolve relatively small temporal changes in large, heterogeneous plots. When applied to research experiments with appropriate randomization and replication of rival ‘C-sequestering’ treatments, the approach may be useful to quantify the relative efficacy of the treatments. The approach may also be applicable to landscape studies seeking, for example, the relationship between topographic position and soil C sequestration. Of course, other geomorphological techniques will be required for sites with appreciable soil redistribution during the measurment interval. Used wisely, the microsite approach provides a firm foundation for building better systems to measure, monitor and verify soil C sequestration.
|
ACKNOWLEDGMENTS |
|---|
Received for publication June 5, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. E. Russell, J. W. Raich, O. J. Valverde-Barrantes, and R. F. Fisher Tree Species Effects on Soil Properties in Experimental Plantations in Tropical Moist Forest Soil Sci. Soc. Am. J., June 29, 2007; 71(4): 1389 - 1397. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. H. Waiser, C. L. S. Morgan, D. J. Brown, and C. T. Hallmark In Situ Characterization of Soil Clay Content with Visible Near-Infrared Diffuse Reflectance Spectroscopy Soil Sci. Soc. Am. J., March 12, 2007; 71(2): 389 - 396. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. A. Omonode, A. Gal, D. E. Stott, T. S. Abney, and T. J. Vyn Short-term Versus Continuous Chisel and No-till Effects on Soil Carbon and Nitrogen Soil Sci. Soc. Am. J., February 2, 2006; 70(2): 419 - 425. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Journal of Environmental Quality |
||||
