Soil carbon stocks in particle-size fractions under seasonally irrigated, grazed pasture

Abstract

For 60 years at Winchmore, soils under pasture grazed by sheep received rainfall or rainfall plus irrigation during summer to limit the water deficit. For soil samples collected to different depths during years 27 and 60, carbon (C) concentration was measured in three size fractions, 53–2000 µm (sand), 5–53 µm (silt) and <5 µm (clay). After 27 years of rainfall plus irrigation, C stocks to depth 0.225 m in the sand-, silt- and clay-size fractions averaged 0.9±0.2 (SD, n=3), 1.4±0.2 and 3.5±0.2 kg m⁻², respectively, and 1.2±0.1, 1.6±0.1 and 4.7±0.4 kg m⁻² in plots receiving rainfall alone. After 60 years, to depth 0.25 m, the corresponding values were 1.1±0.1, 1.6±0.1, 3.0±0.1 kg m⁻² (rainfall plus irrigation) and 1.6±0.3, 2.2±0.3, 3.9±0.5 kg m⁻² (rainfall alone) and to depth 1 m, 1.8±0.2, 2.1±0.1, 4.4±0.4 kg m⁻² (rainfall plus irrigation) and 2.8±0.5, 3.3±0.4, 6.0±0.8 kg m⁻² (rainfall alone). Unexpectedly, C in the three size fractions responded similarly to irrigation.

Keywords:

• carbon
• decomposition
• grassland
• irrigation
• organic matter

Introduction

For 60 years at Winchmore, South Island, New Zealand (43°48′S, 171°48′E, 160 masl), stony soils under continuous pasture grazing by sheep have received rainfall (nil irrigation) or rainfall plus irrigation to limit the water deficit (McDowell & Smith 2012). Based on soil samples collected after 60 years, the stock of organic carbon (C) to a depth of 1 m was significantly less (P<0.05) in the irrigated plots (by 32%, on average) than in plots receiving rainfall alone (Kelliher et al. 2012). Partitioning soil C into pools or fractions may assist in understanding the dynamics governing soil C stocks in response to irrigation. For this study, we measured C in three size fractions (53–2000 µm), (5–53 µm) and (<5 µm). Particulate organic matter (POM) includes C in the sand (>53 µm) fraction, considered the most labile portion of soil C. In the silt- and clay-size fractions, the C has been considered increasingly stable. Such interpretations have been reported for particle-size-fraction measurements made on soil samples from long-term field experiments involving cultivation, organic amendment and fertilizer applications (Hassink 1997; Gerzabek et al. 2001; Sleutel et al. 2006). However, Dungait et al. (2012) argued that soil organic matter turnover is governed by more than just size distribution, but also accessibility. In this study, we postulated that C stocks in the irrigated plots at Winchmore would be less than in plots receiving rainfall alone, and C stocks in particle size fractions would respond differently to irrigation depending on the particle size (i.e. POM-C was expected to be most responsive, and C in the clay-size fraction the least responsive). To test the hypothesis, we analysed two sets of archived soil samples, collected from the same plots at Winchmore after 27 and 60 years of irrigation.

Materials and methods

The Winchmore irrigation trial, including the climate and soil water and animal grazing regimes, has been fully described by Kelliher et al. (2012). Briefly, the basis for irrigation was to limit the soil water deficit. When the water content was less than 20%, plots were flood irrigated with 100 mm of water (average of seven irrigations per year). Hereafter, this will be called the 20% irrigation treatment. A nil irrigation treatment received rainfall alone. Soils were sampled (in 2009, 60 years after the trial began) midway between the borders in each plot, a mechanical digger was used to excavate a pit, approximately 1 m wide, 2 m long and 1.5 m deep. Along a pit face, a steel frame (0.4×0.4 m) was used to determine where samples were taken across six depth intervals (0–0.075, 0.075–0.15, 0.15–0.25, 0.25–0.50, 0.50–0.75 and 0.75–1.00 m). In total, approximately 1500 kg of material (stones and soil) was excavated from six pits, including three 20% irrigated plots and three plots receiving rainfall alone. Each sample was weighed, the soil separated from stones by passing through a 2 mm sieve and the stones were weighed again. From each soil sample, approximately 3 kg was set aside for analysis, and the remainder returned with the stones to the pit, which was refilled. A portion of each soil sample was weighed, dried at 105 °C for 24 h and weighed again to determine the water content. Excluding the weights of water and stones and accounting for sampled volume, each sample's bulk density was calculated. In 1976, 27 years after the trial began, the soils had been sampled in the same plots, but using an auger across four depth intervals (0–0.075, 0.075–0.15, 0.15–0.225 and 0.225–0.3 m). These samples had also been passed through a 2 mm sieve, dried, placed in labelled containers and stored until this study. In the laboratory, the processing of both sets of samples included the removal of roots and drying for 16 h in an oven (105 °C). The POM was isolated by sieving (53 µm) after dispersing the soils (20 g sample in 60 mL deionized water) using an ultrasonic probe (65 W for 60 s). To disperse this soil, sonication treatment was as effective as the conventional chemical method using sodium hexametaphosphate (Calgon, data not shown). The <53 µm material was further divided into 5–53 µm and <5 µm fractions by gravitational sedimentation (Gee & Or 2002). While the <5 µm fraction included some fine silt (2–5 µm), hereafter, this will be called a clay-size fraction. The isolated fractions were dried (60 °C), weighed and a subsample of each fraction analysed for total C concentration using a Leco Truspec C/N analyser (LECO Corporation, St Joseph, MI, USA). The amount of C in each fraction, expressed in units of g C kg⁻¹ soil, was calculated from the mass proportion of each fraction and its C concentration.

Soil sample data can be vertically integrated to determine the C stock. Vertical integration was done on a volumetric basis from the surface to depths of 0.075, 0.15, 0.25, 0.50, 0.75 and 1 m. Each calculation was a product of the C concentration, bulk density and depth interval. For comparison, vertical integration was also done on an equivalent mass basis (Ellert et al. 2002). For soil mass equivalence, calculations were products of the C concentration, bulk density and depth for each sampled interval, and summed from the surface to depths 0.75 and 1 m in the nil and 20% irrigated plots, respectively (the mass of soil to these depths was 611 kg m⁻² for all plots, see below). Statistical analyses were done as treatment group comparisons by T test assuming unequal variance and, where means have been reported, error bounds are±standard deviation. For the difference between the two group means, standard errors were calculated using the variances and sample sizes to firstly compute the degrees of freedom and T statistic. Then, the T statistic was multiplied by a root-mean-square combination of the variances and either the actual or larger sample sizes (i.e. n=3 or n > 3) to examine the effect of sample size on the statistical significance.

Results and discussion

The only statistically significant differences in particle size distribution between the nil and 20% irrigation treatments were in the depth 0.75–1 m after 60 years of irrigation, and these differences were not substantial (Table 1). For the uppermost 0.25 m, weighted mean mass fractions were 0.23, 0.48 and 0.29 for the sand-, silt- and clay-size particles, respectively. Below 0.75 m, the sand-size fraction was up to 0.95 of soil mass with as little as 0.03 and 0.02 for the silt- and clay-size fractions, respectively. In the uppermost 0.25 m, there were few stones (on average, only 13 kg m⁻²), but below this depth, stone mass per unit area was 21 times greater (on average, 269 kg m⁻², Table 2). The mean (stone-free) soil mass was not significantly different in nil and 20% irrigated plots from the surface to any of the sampled depths (Table 3). However, from the surface to depth 1 m, there was a substantial difference between mean (stone-free) soil masses in nil and 20% irrigated plots (761 versus 611 kg m⁻²). The soil mass to 0.75 m depth in the nil treatment was equal to that in the uppermost 1 m of the 20% irrigation treatment, enabling data for these depths to be used in equivalent mass comparisons of C stocks.

Table 1  Soil mass fractions (mean±standard deviation, n=3) in three size classes for samples after 27 and 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation).

	Sand (>53 µm)		Silt (5–53 µm)		Clay (<5 µm)
Depth (m)	Nil irrigation	20% irrigation	Nil irrigation	20% irrigation	Nil irrigation	20% irrigation
27 years
0–0.075	0.23±0.02	0.23±0.02	0.47±0.01	0.48±0.01	0.30±0.02	0.29±0.01
0.075–0.15	0.22±0.01	0.23±0.01	0.47±0.01	0.47±0.002	0.31±0.01	0.30±0.01
0.15–0.225	0.23±0.01	0.23±0.02	0.46±0.002	0.46±0.01	0.31±0.01	0.31±0.01
0.225–0.30	0.22±0.01	0.22±0.01	0.46±0.004	0.46±0.004	0.32±0.01	0.32±0.01
60 years
0–0.075	0.25±0.02	0.23±0.01	0.46±0.02	0.46±0.01	0.27±0.004	0.28±0.01
0.075–0.15	0.23±0.01	0.23±0.001	0.46±0.01	0.46±0.002	0.29±0.01	0.26±0.002
0.15–0.25	0.23±0.01	0.24±0.03	0.46±0.002	0.46±0.003	0.31±0.004	0.30±0.03
0.25–0.50	0.38±0.12	0.24±0.01	0.35±0.08	0.43±0.01	0.27±0.03	0.33±0.003
0.50–0.75	0.75±0.20	0.72±0.19	0.14±0.10	0.14±0.10	0.12±0.11	0.14±0.09
0.75–1.00	0.91±0.02	0.95±0.01	0.06±0.01	0.03±0.01	0.04±0.01	0.02±0.005

Table 2  Mass of stones per unit area from sampled depth intervals (mean±standard deviation, n=3) after 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation).

	Stones (kg m^−2)
Depth (m)	Nil irrigation	20% irrigation
0–0.075	2±4	0±0
0.075–0.15	19±11	11±5
0.15–0.25	23±16	22±14
0.25–0.50	262±156	183±36
0.50–0.75	265±110	229±68
0.75–1.00	395±86	277±49

Table 3  Cumulative (stone-free) soil mass per unit area from sampled depth intervals (soil density, mean±standard deviation, n=3) after 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation).

	Soil mass (kg m^−2)
Depth (m)	Nil irrigation	20% irrigation
0.075	68±4	70±4
0.15	158±17	124±14
0.25	268±18	266±11
0.50	464±46	423±62
0.75	610±63	496±96
1.00	761±41	611±93

After 27 years, the only statistically significant differences in C concentration between the nil and 20% irrigation treatments were in the sand-size fraction from depth 0.15–0.225 m, and the clay-size fraction from depths 0–0.075 and 0.15–0.30 m (Table 4). After 60 years, no consistent, statistically significant differences were found. The only statistically significant differences in whole-soil C (total C kg⁻¹ soil) between the nil and 20% irrigation treatments were from depth 0.15 to 0.225 m after 27 years (21.9±0.3 versus 18.7±1.1, Table 4) and, after 60 years, from the surface to depth 0.075 m (44.4±2.7 versus 37.2±1.2). However, when calculated volumetrically to a depth of 0.075 m, there were no statistically significant differences in total C stocks between the nil and 20% irrigation treatments after 27 and 60 years (Table 5).

Table 4  Carbon (C) concentration (mean±standard deviation, n=3) of soil samples in three size fractions after 27 and 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation). Statistical significance of a treatment effect was determined by P values (values < 0.05 in bold) which follow the 20% irrigation standard deviations.

	Soil C content, g kg^−1 soil
	>53 µm		5–53 µm		<5 µm
Depth (m)	Nil irrigation	20% irrigation	Nil irrigation	20% irrigation	Nil irrigation	20% irrigation
27 years
0–0.075	9.4±1.9	7.0±1.6, 0.192	8.5±1.1	9.1±1.2, 0.566	26.2±1.0	17.5±3.3, 0.048
0.075–0.15	4.5±0.9	3.5±0.9, 0.217	7.2±0.4	6.7±1.3, 0.558	19.3±0.6	16.9±1.2, 0.061
0.15–0.225	2.4±0.2	1.9±0.1, 0.022	4.8±0.2	4.0±0.7, 0.178	14.7±0.4	12.8±0.4, 0.005
0.225–0.30	1.6±0.1	1.2±0.2, 0.054	3.6±0.4	2.4±0.6, 0.060	12.2±1.2	8.9±1.4, 0.036
60 years
0–0.075	12.1±1.9	8.8±0.3, 0.093	12.0±1.4	9.1±0.3, 0.068	20.3±0.6	19.3±0.9, 0.211
0.075–0.15	5.4±1.0	4.4±0.3, 0.219	8.4±1.2	7.2±0.3, 0.215	18.5±1.7	13.8±0.5, 0.044
0.15–0.25	2.6±0.6	2.1±0.4, 0.234	5.6±1.1	4.3±0.4, 0.200	8.3±1.7	6.5±0.1, 0.218
0.25–0.50	2.0±0.8	0.8±0.1, 0.118	3.0±0.6	1.5±0.3, 0.034	5.0±1.4	3.0±1.1, 0.122
0.50–0.75	3.0±0.6	2.4±0.7, 0.292	2.1±0.3	1.9±0.2, 0.315	5.2±0.4	4.6±1.0, 0.391
0.75–1.00	2.7±0.2	2.7±1.0, 0.904	1.4±0.4	1.0±0.1, 0.199	4.1±0.5	4.2±0.5, 0.799

Table 5  Soil carbon (C) stocks (mean±standard deviation, n=3) calculated volumetrically by depth in three size fractions after 27 and 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation). Each calculation was a product of the C concentration, bulk density and depth interval. Statistical significance of a treatment effect was determined by P values (values< 0.05 in bold) which follow the 20% irrigation standard deviations.

	Soil C storage (kg C m^−2)
Treatment and years of treatment	>53 µm fraction	5–53 µm fraction	<5 µm fraction	Sum of three fractions
27 years
	0–0.075 m
Nil irrigation	0.6±0.1	0.6±0.1	1.8±0.1	3.0±0.2
20% irrigation	0.5±0.1, 0.238	0.6±0.1, 0.508	1.2±0.3, 0.056	2.4±0.6, 0.192
	0–0.15 m
Nil irrigation	1.0±0.2	1.2±0.2	3.5±0.4	5.8±0.6
20% irrigation	0.7±0.2, 0.052	1.0±0.2, 0.155	2.1±0.3, 0.020	3.8±0.6, 0.016
	0–0.225 m
Nil irrigation	1.2±0.1	1.6±0.1	4.7±0.4	7.6±0.5
20% irrigation	0.9±0.2, 0.051	1.4±0.2, 0.208	3.5±0.2, 0.020	5.8±0.5, 0.013
	0–0.30 m
Nil irrigation	1.4±0.1	2.2±0.2	6.5±0.3	10.1±0.6
20% irrigation	1.0±0.2, 0.047	1.7±0.2, 0.057	4.5±0.2, 0.001	7.2±0.6, 0.004
60 years
	0–0.075 m
Nil irrigation	0.8±0.1	0.8±0.1	1.4±0.1	3.0±0.2
20% irrigation	0.6±0.1, 0.065	0.6±0.1, 0.106	1.3±0.1, 0.789	2.6±0.2, 0.108
	0–0.15 m
Nil irrigation	1.3±0.2	1.6±0.3	3.1±0.4	5.9±0.9
20% irrigation	0.8±0.1, 0.094	1.0±0.1, 0.081	2.1±0.3, 0.051	4.0±0.4, 0.047
	0–0.25 m
Nil irrigation	1.6±0.3	2.2±0.3	3.9±0.5	7.7±1.0
20% irrigation	1.1±0.1, 0.104	1.6±0.1, 0.080	3.0±0.1, 0.081	5.8±0.2, 0.080
	0–0.50 m
Nil irrigation	2.0±0.3	2.8±0.3	4.9±0.6	9.6±0.1
20% irrigation	1.3±0.1, 0.075	1.9±0.1, 0.049	3.5±0.2, 0.037	6.6±0.3, 0.061
	0–0.75 m
Nil irrigation	2.4±0.5	3.1±0.4	5.6±0.8	11.1±0.7
20% irrigation	1.4±0.0, 0.080	2.0±0.2, 0.025	3.9±0.5, 0.049	7.3±0.6, 0.038
	0–1.0 m
Nil irrigation	2.8±0.5	3.3±0.4	6.0±0.8	12.1±1.6
20% irrigation	1.8±0.2, 0.050	2.1±0.1, 0.013	4.4±0.4, 0.033	8.3±0.5, 0.053

Generally, total C stocks assessed at different depths to a maximum of 1 m were significantly less in the 20% irrigation plots. For example, to depth 0.15 m, the treatment effect was 2 kg C m⁻² (5.8 versus 3.8 after 27 years and 5.9 versus 4.9 after 60 years). After 60 years, to depth 1 m, the treatment effect was 3.8 kg C m⁻² (12.1 versus 8.3). When C stocks were calculated on an equivalent mass basis by comparing amounts of C to depth 1 m in the 20% irrigation plots with that to 0.75 m in the nil irrigation plots, the treatment effect reduced to 2.8 kg C m⁻² (11.1 versus 8.3) and it was not statistically significant (Table 6). However, this treatment effect of 2.8 kg C m⁻² is substantial, so an additional calculation was explored. This involved increasing the number of plots (replicates) from three to four, combining the ‘new’ sample size with unchanged values of the two means and standard deviations, and recalculating a standard error for the difference between the two means. This recalculation yielded a standard error of 2.8 kg C m⁻², identical to the mean treatment effect and reduced from 4.5 kg C m⁻² that had been calculated using the ‘actual’ data including a sample size of three. Thus, the difference between the two means would likely have been statistically significant if the number of replicates had been four, instead of three.

Table 6  Soil carbon (C) stocks (mean±standard deviation, n=3) in three size fractions according to equivalent mass calculations after 60 years of rainfall and summer irrigation when soil water content reached 20% (20% irrigation) or rainfall alone (nil irrigation). For soil mass equivalence, calculations were products of the C concentration, bulk density and depth for each sampled interval, and summed from the surface to depths 0.75 and 1 m in the nil and 20% irrigated plots, respectively.

	Soil C storage (kg C m^−2)
Treatment	>53 µm fraction	5–53 µm fraction	<5 µm fraction	Sum of three fractions
Nil irrigation	2.4±0.5	3.1±0.4	5.6±0.8	11.1±1.7
20% irrigation	1.8±0.2	2.1±0.1	4.4±0.4	8.3±0.5

After 27 years, C in the POM, silt and clay fractions represented, on average, 16%, 22% and 61%, respectively, of the total C stock to depth 0.225 m. Carbon stocks in those fractions were correspondingly and significantly less in the 20% irrigation plots by 25%, 12% and 26%. After 60 years, C in the POM, silt and clay fractions represented, on average, 20%, 28% and 52%, respectively, of the total C storage to depth 0.25 m. The C stocks in those fractions were correspondingly and significantly less in the 20% irrigation plots by 31%, 27% and 23%. After 60 years to depths 0.5 and 1 m, the corresponding percentages were broadly similar. Our results showed that C in the clay-size fraction, commonly regarded as the most stable C fraction, was as responsive to irrigation as POM, an observation that does not support our hypothesis. While the results of measuring soil C in sand-, silt- and clay-size fractions suggested such partitioning had not been a driver of responsiveness to irrigation in this study, further research is warranted about irrigation effects on C stocks in grassland soils and the processes determining organic matter turnover under field conditions.

Acknowledgements

Funding to FM Kelliher was provided by the New Zealand Agricultural Greenhouse Gas Research Centre. Additional funding was provided by the Agricultural and Marketing Research and Development Trust (AGMARDT), the New Zealand Ministry for Primary Industries, and Environment Canterbury. We are grateful to Weiwen Qiu for laboratory assistance, Chikako van Koten and Peter West for valuable discussions, and Jiafa Luo, Alec Mackay and anonymous referees for constructive critiques of the manuscript.