Influence of pore size distribution and soil water ... - CSIRO Publishing

CSIRO PUBLISHING Soil Research, 2012, 50, 125–135 http://dx.doi.org/10.1071/SR11112

Influence of pore size distribution and soil water content on nitrous oxide emissions Tony J. van der Weerden A,D, Francis M. Kelliher B,C, and Cecile A. M. de Klein A A

AgResearch, Invermay Agricultural Centre, Mosgiel, New Zealand. AgResearch, Canterbury Agricultural Centre, Lincoln, New Zealand. C Lincoln University, Department of Soil and Physical Sciences, Lincoln, New Zealand. D Corresponding author. Email: [email protected] B

Abstract. Nitrous oxide (N2O) emissions from agricultural soils have been estimated to comprise about two-thirds of the biosphere’s contribution of this potent greenhouse gas. In pasture systems grazed by farmed animals, where substrate is generally available, spatial variation in emissions, in addition to that cause by the patchiness of urine deposition, has been attributed to soil aeration, as governed by gas diffusion. However, this parameter is not readily measured, and the soil’s water-filled pore space (WFPS) has often been used as a proxy, despite gas diffusion in soils depending on the volumetric fractions of water and air. With changing water content, these fractions will reflect the soil’s pore size distribution. The aims of this study were: (i) to determine if the pore size distribution of two pastoral soils explains previously observed differences in N2O emissions under field conditions, and (ii) to assess the most appropriate soil water/gas diffusion metric for estimating N2O emissions. The N2O emissions were measured from intact cores of two soils (one classified as well drained and one as poorly drained) that had been sampled to a depth of 50 mm beneath grazed pasture. Nitrogen (N, 500 kg N/ha) was applied to soil cores as aqueous nitrate solution, and the cores were drained under controlled conditions at a constant temperature. The poorly drained soil had a larger proportion of macropores (23.5 v. 18.7% in the well-drained soil), resulting in more rapid drainage and increased pore continuity, thereby reducing the duration of anaerobicity, and leading to lower N2O emissions. Emissions were related to three soil water proxies including WFPS, volumetric water content (VWC), and matric potential (MP), and to relative diffusion (RD). All parameters showed highly significant relationships with N2O emissions (P < 0.001), with RD, WFPS, VWC, and MP accounting for 59, 72, 88, and 93% of the variability, respectively. As VWC is more readily determined than MP, the former is potentially more suitable for estimating N2O emission from different soils across a range of time and space scales under field conditions. Additional keywords: denitrification, soil pore structure, soil porosity, soil water. Received 16 May 2011, accepted 27 February 2012, published online 28 March 2012

Introduction Agricultural soils are the principle source of nitrous oxide (N2O) emissions to the atmosphere. From pastoral agriculture, most of these emissions can be attributed to urine deposited onto soils by grazing animals. For example, in New Zealand, this accounts for ~80% of the annual national N2O emissions inventory including direct and indirect emissions (de Klein and Ledgard 2005). While both nitrification and denitrification can contribute to direct N2O emissions from grazed pastures in New Zealand, denitrification is considered to be the major contributor (de Klein and van Logtestijn 1994; Müller and Sherlock 2004). Proximal environmental factors that affect the immediate environment of the bacterial cell responsible for denitrification include nitrate (NO3–) concentration, carbon (C) supply rate, oxygen diffusion, and temperature. Beneath a urine patch of grazed pasture, high soil mineral nitrogen (N) content and readily available C concentrations can stimulate denitrification and therefore N2O production. In the absence of Journal compilation CSIRO 2012

C, high NO3– content can inhibit N2O-reductase activity, thereby lowering the N2 : N2O ratio and potentially increasing N2O emissions (Weier et al. 1993). Gases diffuse at a much lower rate through water than air, so rainfall can significantly and rapidly reduce the oxygen diffusion rate in soils, while drainage has an opposite effect. For this reason, N2O production can be stimulated by rainfall, providing other controlling variables have not become limiting, and such episodes can contribute significantly to the annual N2O emissions (Li et al. 1992). The effect of rainfall on the gas diffusion rate in soils depends on rainfall rate and duration as well as the soil’s pore size distribution determining the water storage capacity and drainage rate. Gas diffusion rate in soils is not a variable that can be readily measured, so soil water content is often used as a proxy. Water-filled pore space (WFPS) was first proposed by Linn and Doran (1984). This metric has since been widely used in incubation and field studies investigating N2O emissions from www.publish.csiro.au/journals/sr

126

Soil Research

soils (Dobbie and Smith 2001, 2003; Bateman and Baggs 2005; del Prado et al. 2006; Ciarlo et al. 2007). However, gas diffusion in soils depends on the volumetric fractions of water and air (Millington 1959; Millington and Quirk 1961). Moreover, soils with different bulk densities will have different volumes of air and water at any given value of WFPS, causing variation in relationships between WFPS and N2O emissions (Castellano et al. 2010). Thus, it has been suggested that volumetric water content (VWC) or volumetric gas content (VGC) should be related to N2O emissions (Farquharson and Baldock 2008). These researchers consider that VGC may be more appropriate for relating to nitrification-induced N2O emissions, while VWC is potentially a better descriptor of denitrification-induced N2O emissions. Alternatively, Castellano et al. (2010) and Müller (1996) suggest matric potential (MP), defined as a measure of the energy of water in contact with the soil matrix relative to the energy of free water (Davidson 1993), will be a more consistent predictor of the relative N2O emissions across different soils compared with WFPS because emissions have been found to peak within a very narrow range of MP in contrast to the wide range of WFPS values. It has also been suggested that relative diffusivity (RD), which is defined as oxygen diffusion through soil relative to diffusion in free air, is a more appropriate predictor of N2O emissions, as it relates strongly to NO3– and C supply for denitrification (Petersen et al. 2008). Given the potential errors associated with measuring or assessing each of these metrics, we postulate that VWC will be the most suitable metric to relate with N2O emissions for different soils. The aims of this experiment were: (i) to determine if differences in the pore size distribution of two pastoral soils could explain previously observed differences in N2O emissions from each soil, and (ii) to assess whether VWC, WFPS, MP, or RD is the most appropriate metric for gas diffusion in relation to N2O emissions from these two soils when subjected to a range of water regimes in the laboratory. A solution of NO3– was applied as the N source to determine N2O produced via denitrification. We also applied a novel technique where N2O emissions and water retention curves were determined concurrently from intact cores. To our knowledge, concurrent determination of water retention curves and N2O emissions has not been done previously. Materials and methods Soil cores Undisturbed, intact soil cores were collected on 16 January 2009. Soil cores (120 cores, i.d. 100 mm, height 50 mm, volume 377 cm3) were carefully collected from the surface layer (0–5 cm) of two soil types (60 cores per soil type following a grid pattern within an area 0.9 by 1.5 m) using stainless steel rings. The cores were stored at 48C for 2 days before their preparation for the study. Soil types used for this study were a poorly drained Otokia silt loam (NZ Taxonomy, Fragic Perchgley Pallic soil; USDA Taxonomy, Fragiaquepts; Hewitt 2010) and a moderately well-drained (hereafter termed ‘well-drained’) Wingatui silt loam (NZ Taxonomy, weathered Fluvial Recent soil; USDA Taxonomy, Dystrudepts; Hewitt 2010). Both sites supported a mixed perennial ryegrass (Lolium perenne)–white

T. J. van der Weerden et al.

clover (Trifolium repens) sward rotationally grazed by sheep. The pastures had not been grazed for at least 1 month before soil sampling. Both soils have been extensively studied in previous, seasonal field experiments in New Zealand to determine the N2O emission factor for cow urine (EF3PRP, which relates to emissions from direct deposition of animal waste onto pasture, range, and paddocks; IPCC 1997). The physical nature of these soils is of interest because, in several studies, N2O emissions were greater from the well-drained Wingatui soil than the poorly drained Otokia soil (e.g. Sherlock et al. 2003; de Klein et al. 2004). Additional soil samples were collected for physical and chemical characterisation (Table 1). Experimental design The main treatment of the experimental design was five different levels of MP (–0.5, –1, –3, –5, and –10 kPa), with subtreatments of soil type (well-drained and poorly drained) and N addition (0N and +N). The different levels of MP correspond to maximum pore sizes of 600, 300, 100, 60, and 30 mm, respectively. Pores >30 mm are regarded as macropores. Soil bulk density, moisture release curves, and pore size distribution were determined for each core in addition to VWC, VGC, and WFPS. Cores were prepared for adjustment of MP by trimming the soil surface to remove pasture cover, followed by application of calcium carbonate slurry to the surface and, once set, ‘peeling’ it off to provide an unsmeared, natural soil surface. The bottom of the cores was trimmed evenly and fitted with a fine gauze cloth. For soil physical research, earthworms are normally removed from intact cores using a formaldehyde treatment (Drewry et al. 2000). However, this approach would also affect microbial activity, so could not be employed in this study. Instead, earthworms were extracted from the cores using several approaches including saturation of cores, covering with black plastic, and application of a mild electric shock to the cores for 15 s using a 9 V battery. As intact cores were to remain on tension tables for ~8 weeks, there was a risk of fungal growth on soil core surfaces. Although Table 1.

Physical and chemical properties of poorly drained Otokia and well-drained Wingatui silt loam soils

Property

Otokia silt loam

Wingatui silt loam

Physical Fine sand (%) (60–200 mm) Silt (%) (2–60 mm) Clay (%) ( 0.05). These results suggested that prochloraz did not affect N2O emissions, and therefore it was used as a preventative fungicide to treat cores before commencing the main experiment. The experiment was designed so that N2O emission and moisture release data would be collected concurrently from the same set of soil cores. To our knowledge, concurrent measurements of moisture retention curves and N2O emissions have not been done previously. Our approach is outlined below: 1. Intact cores (120) were divided into five groups of 24 cores (12 cores per soil type), with each group assigned a specific MP at which cores would be treated with or without a nitrate solution (+N and 0N, respectively), and subsequent N2O emissions would be measured until those from +N cores returned to levels measured from 0N cores. 2. All 120 cores were saturated and then stepwise drained to –10 kPa, with each core weighed at each tension for determining moisture release curves. 3. As each group of cores reached its assigned MP for N2O measurements, N treatments were applied to the surface of the +N cores within that group, which were then allowed to equilibrate at the prescribed MP before the first N2O emission measurement. 4. Soil cores within a group were kept at their assigned MP for ~8 weeks, during which N2O emissions were measured on six occasions, until emissions from the +N treatment (replicated eight times) returned to levels measured from 0N cores (replicated four times); 5. On completion of gas emission measurements (at ~8 weeks after application of N), collection of moisture release data continued until all groups of cores had reached the lowest MP (–10 kPa). 6. All soil cores went through the same procedure, with cores at the highest tension (–0.5 kPa) being treated with N and gas-sampled a little earlier than those treated with N and sampled at –10 kPa. However, measurements on all 120 cores were completed within 9 days of each other.

Soil Research

127

Nitrogen was applied to the +N cores as dissolved potassium nitrate at a rate equivalent to 500 kg N/ha. An automated pipette was used to deliver 16 mL of solution evenly to the core surface, equivalent to an application rate of 2 mm, ensuring infiltration into the core was achieved. An equivalent volume of deionised water was applied to 0N cores. Nitrous oxide measurement The N2O emissions were measured on six occasions from each core, until emissions from +N treatments returned to values measured from the 0N treatments. The first sampling occurred 4–5 days after N was applied (depending on soil type and assigned MP), as several days were required for cores to equilibrate at the prescribed tension following N or water application. Emissions were determined by removing the stainless steel rings containing the intact soil cores from the tension table, quickly placing each core into a 750-cm3 plastic container, and sealing with an air-tight lid. Headspace gas sample volumes of either 1 or 2 mL were removed from the container via rubber septa positioned in the lid. Gas samples were collected at 30 and 60 min following closure, while background air samples were used to represent 0 min. Gas samples were injected into 6-mL Exetainer vials (Labco, High Wycombe, UK) containing ambient concentrations of N2O at atmospheric pressure, resulting in a pressurised vial containing 1 or 2 mL of container headspace air and 6 mL of ambient air. This allowed dilution of the sample before analysis via gas chromatography, necessary to accommodate the high headspace concentrations. Following gas sampling, each core was removed from its container and weighed, with any grass growing on the core surface trimmed back. Cores were then returned to the appropriate tension table and re-equilibrated at the required tension; for each sampling, the total duration cores were removed from the tension table was ~80 min. The volume of water remaining in the container following soil core removal was minimal, even for the wettest cores (those held at 0.5 kPa on the tension table). This is because the cores were held at 0 kPa tension while in the container, thus no drainage was imposed. On completion of six gas samplings (~2 months after experiment initiation), the tension placed on each batch of cores was increased, with weights recorded at critical tensions. Following weighing of cores at the lowest MP of –10 kPa, cores were oven-dried at 1058C and re-weighed. Gas samples were analysed for N2O concentrations by gas chromatograph using an SRI 8610 automated gas chromatograph (SRI Instruments, Torrance, CA, USA) (de Klein et al. 2003). The N2O emissions were calculated for each container from the increase in head space N2O concentrations over the sampling time. The hourly N2O emissions (mg N/m2.h) used the same approach as adopted for calculating emissions from field plots (de Klein et al. 2003), as follows: N2 O ¼

dN2 O M V 1000 dT Vm A

ð1Þ

where dN2O is the increase in head space N2O concentrations over time (mL/L); dT is the enclosure period (h); M is the molar weight of N in N2O; Vm is the molar volume of gas at the sampling temperature (L/mol); V is the headspace volume (m3);

128

Soil Research


and A is the area covered (m2), with container headspace height equivalent to 0.042 m. An emissions detection limit was calculated using Eqn 1 with measurements of the GC system precision, as 0.27 mg N/m2.h. Of the 720 emission determinations, only three were below the detection limit and therefore assigned the value of zero. Net cumulative emissions were calculated by trapezoidal integration of emissions measured from individual soil cores over an 8-week period, followed by computation of the difference in average losses from the +N and 0N treatments at each MP. Data analysis The MP (y) and VWC (cm/cm3) data were plotted using a log–log transformation to determine the slope (–b) and air entry potential (ye) for each soil (Campbell 1974; Moldrup et al. 1996). The MP has been assumed equal to soil water potential for comparisons with the data reported by Clapp and Hornberger (1978) on the basis of soil texture, particularly clay content, which has been measured for the two soils studied here. The air entry potential is the value of MP when air enters soil following saturation and drainage (Campbell 1974), determined from the exponential of the intercept of a log–log plot of y and VWC. The RD was not directly measured, but instead modelled by relating oxygen diffusion through soil (DP, m3 soil-air/m soil.s) to diffusion of oxygen in free air (DO, m2 air/s) using the threeporosity-encased (3POE) model (Moldrup et al. 2005): e X100 DP ¼ F2 DO F

ð2Þ

where F is the soil’s total porosity (m3/m3), e is the VGC (air-filled porosity; m3 soil-air/m3 soil), and X100 is defined as a dimensionless tortuosity–continuity parameter at MP of –100 cm H2O ( –10 kPa): 1 log e4100 ð3Þ X100 ¼ 2 þ 100 log eF Several simple modelling methods have been developed for predicting gas diffusivity, as measurements can be difficult. While Allaire et al. (2008) suggest some preliminary measurements of RD can assist with determining which model is most appropriate, we chose the method of Moldrup et al. (2005) for several reasons: (i) it has been developed for intact cores; (ii) it attempts to account for pore size distribution; and (iii) in this study, it can be applied using parameters already

determined from the moisture release curve. The 3POE model uses three points on the moisture release curve to predict DP. Pore continuity, defined as the ratio between RD and e (Gradwell 1961; Petersen et al. 2008), was also determined to investigate its response to declining water content with decreasing MP. For calculating pore continuity, predicted RD values were used. Because N2O emissions were not normally distributed, data were log-transformed. Linear regression relationships were analysed between log N2O and WFPS, VWC, MP, and RD, with different slope and intercepts for each soil type in the fitted model using GENSTAT 12 (Payne et al. 2009). Differences between soils for these relationships were tested for statistical significance using standard regression analysis F-tests. Soil differences in cumulative emission at each MP were compared using analysis of variance of the log-transformed values. Results Water retention curves For the poorly drained Otokia soil, a power function closely fitted the MP and VWC data (R2 = 0.90, P < 0.001; Table 2). The slope (–b) was 8.6, which corresponds with a clay content of 34% according to Clapp and Hornberger (1978); this was nearly identical to the measured clay content of 33% (Table 1). According to the power function, the air entry potential was 6.8 cm, similar to the lowest tension used (5 cm). While still showing a significant relationship, the power function did not fit the well-drained Wingatui soil data so well, particularly when all five tensions (from 5 to 100 cm) were included (R2 = 0.81, P < 0.001; Table 2). For this soil, the b parameter was 11.9, greater than the largest value of 11.4 reported by Clapp and Hornberger (1978), and corresponding with a clay content of 63% (Table 2), greatly exceeding the measured clay content of 37% (Tables 1 and 2). Examination of the log–log plot (not shown) suggested that the data at –0.5 kPa (5 cm) were outliers. Therefore, we excluded these data, and the b parameter became 9.8, which, according to Clapp and Hornberger (1978), is predicted to have a clay content of 34–43%, fitting well with the measured value of 37%. The regression was only slightly improved, at R2 = 0.82, and on this basis, b parameters for the two soils (8.6 and 9.8) are not significantly different (P > 0.05). By excluding –0.5 kPa MP values for the well-drained Wingatui soil data, the air entry potential also changed to 10.1 cm, double the value of the omitted tension (5 cm).

Table 2. Water retention parameters for the Otokia and Wingatui soils, using the Campbell soil-water retention function (Campbell 1974; Moldrup et al. 1996), obtained as the best-fit slopes of the soil-water characteristic curves in a log–log soil water potential (cm H2O)–volumetric water content (v/v) for –5 > y > –100 cm, where –b and ye represent slope and air entry potential, respectively Values are means (s.e.m. in parentheses). One outlier soil core removed from each soil type Soil Otokia Wingatui WingatuiB A

Soil water potential (cm H2O)

–b

R2

ye

n

Mean clay fractionA (%)

Measured clay fraction (%)

–5 > y > –100 –5 > y > –100 –10 > y > –100

8.6 (0.5) 11.9 (1.0) 9.8 (0.9)

0.90 0.81 0.82

6.8 (1.1) 6.7 (1.1) 10.1 (1.2)

35 35 30

34 >63 34–45

33 37 37

Based on representative values reported by Clapp and Hornberger (1978). Excluding y = –5 cm.

B

Pore size distribution and soil water content on N2O emissions

Porosity characteristics Total porosity was similar for the two soils, but the pore size distributions differed (Table 3). This was attributed to the welldrained Wingatui soil having 5% greater clay content and 7% greater bulk density (Table 1). Micropores (30 mm diameter) represented a significantly larger proportion in the poorly drained Otokia soil (24%, P < 0.01), mainly due to a significantly greater proportion of large pores (60–300 mm diameter, P < 0.001; Table 3). Near saturation (MP of –0.5 kPa), VWC was 0.61–0.62 cm3/ 3 cm for both soils (Fig. 1), reflecting their similar porosities at this tension, and the two corresponding values of pore continuity were also not significantly different (Table 4). As MP decreased to –1 kPa and then –3 kPa in the Otokia soil, VWC declined to 0.49 cm3/cm3 with corresponding pore continuity increasing by more than a factor of three, due to the drainage of pores >100 mm diameter. For the Wingatui soil, the corresponding VWC was 0.56 cm3/cm3 and the pore continuity had only doubled, resulting in significantly less pore continuity than the Otokia soil at both –1 and –3 kPa tensions (P < 0.01; Table 4). At field capacity (–10 kPa), VWC for the poorly drained Otokia soil and the well-drained Wingatui soil was, on average, 0.43 and 0.49 cm3/cm3, respectively. Influence of MP on N2O emissions and cumulative losses The N2O emissions decreased over the 62-day-long incubation period (Fig. 2), presumably due to exhaustion of the NO3– supply for denitrifying bacteria. However, soil mineral N content could not be determined at the end of the study because all of the soil was needed for bulk density measurements for each core. Net cumulative emissions were greatest at the highest MP of –0.5 kPa, and indistinguishable between the two soils (Table 4). Pore continuity was also similar for the two soils at this tension (Table 4). As MP decreased, cumulative emissions declined significantly, while pore continuity increased. The decline in cumulative emissions and the increase in pore continuity were slower for the well-drained Wingatui soil. This soil had significantly higher cumulative emissions compared with the poorly drained Otokia soil at MP of –1 and –3 kPa (P < 0.05), while the Otokia soil produced significantly higher cumulative emissions than the Wingatui soil (P < 0.05) at the lowest MP

Soil Research

of –10 kPa. However, the cumulative emissions at –10 kPa were very small compared with those at MP –3 kPa. When focusing on the +N cores on the first sampling occasion, when the emissions were greatest, a similar pattern was observed for the relationship between MP and N2O emissions, with the emission rate from the Otokia soil declining more than from the Wingatui soil when MP decreased from –0.5 to –3 kPa (Fig. 3). Proxy for gas diffusion An analysis of hourly N2O emissions against each of the three proxies for gas diffusion and against relative diffusion was conducted for data collected from the first sampling occasion, this dataset having the highest emissions and largest range of values. The strongest relationship was between log N2O emissions and log MP (R2 = 0.93, P < 0.001, averaged over the two soil types; Fig. 4). There was no significant difference in the log N2O and log MP relationships for the two soils. Back-transforming the data, a two-fold increase in MP corresponded to a seven-fold (s.e. 0.5) increase in N2O emissions. The relationship between log N2O emissions and VWC was also strong, with VWC accounting for 88% of the variance (P < 0.001; Fig. 5). However, the variance associated with VWC was significantly lower than that for MP (P < 0.05). There were significant differences (P = 0.005) between rates of increase for the two soils, with emissions from the poorly drained Otokia and the well-drained Wingatui increasing 85-fold (s.e. 20) and 285-fold (s.e. 99), respectively, with every 0.10 cm3/cm3 increase in VWC. Averaged over the two soil types, a 0.10 cm3/cm3 increase in VWC corresponded to a 126-fold (s.e. 26) increase in N2O emissions. This analysis was conducted with N2O emissions presented on a log scale. On a linear scale, N2O emissions increased markedly above a soilwater content ‘threshold’ of 0.55 cm3/cm3 soil (Fig. 6). There was a weaker relationship between log N2O emissions and WFPS and no significant difference existed between the two soils (R2 = 0.72; P < 0.001, Fig. 7); a 10 percentage point increase in WFPS corresponded to a 16.5-fold (s.e. 3.4) increase in N2O emissions. On a linear scale, N2O emissions increased sharply above a WFPS threshold of 77% (Fig. 8). Although there was a significant negative linear relationship between log N2O emissions and RD (R2 = 0.59; P < 0.001, Fig. 9), it was weaker than the relationships found for VWC and WFPS. A 0.01 point increase in RD resulted in a 4.8-fold (s.e. 0.75) decrease in N2O emissions.

Table 3. Pore size distribution of the Otokia and Wingatui soils Values are means (s.e.m. in parentheses). n.s., Not significant (P > 0.05) Porosity and pore size distribution (% of soil volume)

Matric potential (–kPa)

Total porosity Microporosity (pore size 300 mm

>10 0–10 5–10 3–5 1–3 0.05 for both). For WFPS, the two slopes were not different (P > 0.05), but the y-intercepts were significantly different (P < 0.001). Discussion Influence of porosity and pore size distribution on N2O emissions Differences in the pore size distribution, particularly the volume of pores 60–300 mm in diameter, provides an explanation for the

higher N2O emissions observed from the well-drained Wingatui soil compared with the poorly drained Otokia soil. While the two soils have virtually identical total porosities of 68% v/v (Table 1), the different pore size distributions influenced the pore continuity and therefore gas content when VWC exceeded field capacity (MP –10 kPa). Because the Otokia soil had a greater proportion of large pores (>60 mm diameter; MP 0.05) Matric potential (–kPa) 0.5 1.0 3.0 5.0 10.0

Otokia 0.035 0.061 0.126 0.133 0.182

Pore continuity (ratio RD : n) Wingatui

(0.009) (0.007) (0.007) (0.008) (0.008)

0.028 0.028 0.055 0.081 0.112

(0.005) (0.005) (0.010) (0.014) (0.012)

F prob.

Cumulative net N2O emissions (1000 mg N2O-N/m2) Otokia Wingatui F prob.

n.s. P < 0.01 P < 0.001 P < 0.01 P < 0.001

21964 4922 319 78 46

(1380) (977) (67) (16) (9)

25833 19651 915 155 19

(1518) (4172) (192) (31) (4)

n.s. P < 0.01 P < 0.05 n.s. P < 0.05

Soil Research

N2O emission (µg N2O-N/m2.h × 1000)


N2O emission (µg N/m2.h × 1000)

80

60

40

20

1000.000 y = 21.22 x–2.695 R 2 = 0.901

100.000

10.000 1.000 y = 11.06 x–2.948 R 2 = 0.949

0.100 0.010 0.001 0.1

10.0

1.0

Matric potential (–kPa)

0 0

10

20

30

40

50

60

70

Days after N applied

Fig. 4. Scatter plot of N2O emissions, plotted on a log scale, against matric potential (–kPa), also plotted on a log scale (R2 = 0.93). Otokia soil, * and solid line; Wingatui soil, * and dashed line. Data are limited to emissions measured from +N cores on first sampling occasion only.


Fig. 2. Rate of N2O emissions over 62 days, measured from +N treatment applied to Otokia (*) and Wingatui (*) soils held at –0.5 kPa (solid line) and –1.0 kPa (dashed line) matric potential. Error bars indicate s.e.m. (n = 8 for each soil at each matric potential).

80

N2O emission (µg N/m2. h × 1000)

131

60

40

20

1000.000 100.000 10.000

y = 8 × 10–11e44.43x R 2 = 0.946

1.000 0.100

y = 4 × 10–14e56.51x R 2 = 0.996

0.010 0.001 0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

Volumetric water content (cm3/cm3) 0 0

2

4

6

8

10

Matrix potential (–kPa) Fig. 3. Relationship between N2O emission and matric potential, determined for +N treatments applied to Otokia (*) and Wingatui (*) soils at the first sampling occasion. Error bars indicate s.e.m. (n = 8 for each soil at each matric potential).

significantly different (P > 0.05). This suggested that while the hydraulic properties of these two soils are very similar, they behave differently with respect to N2O production and emission at low MP (–3 kPa). Previous field experiments suggested greater N2O emissions from the Wingatui soil, and combined with the results of this study, it could be surmised that the field conditions commonly included low MP in the soils. While not included as a variable in our study, compaction can reduce pore size distribution and pore continuity (Petersen et al. 2008). Livestock grazing can cause substantial soil surface compaction, the extent of which will depend on the pressure under the animal hoof, soil moisture content, and the

Fig. 5. Scatter plot of N2O emissions, plotted on log scale, against volumetric water content (cm3/cm3) (R2 = 0.88). Otokia soil, * and solid line; Wingatui soil, * and dashed line. Data are limited to emissions measured from +N cores on first sampling occasion only.

soil’s natural resilience to physical pressure. Soil compaction can result in an increase in bulk density, WFPS, and anaerobic fraction (Ball et al. 2008; Uchida et al. 2008), which in association with reduced macropore diameter and pore continuity can lead to increased rates of N2O production via denitrification and subsequent N2O emissions (Oenema et al. 1997; Bhandral et al. 2007; Ball et al. 2008). It is interesting to note that the Wingatui soil showed greater variation in the relationships between each of the proxies for gas diffusion and N2O emissions than the Otokia soil (Figs 4–9). This may be due to a smaller percentage of macropores (>60 mm) (Table 3), which, at equal tensions, could induce anoxic conditions for a longer period, thereby maintaining higher denitrification rates for longer. However, a more likely explanation is that this soil has lower pore continuity and thus

Soil Research


100



132

80

60

40

20

0 0.40

0.45

0.50

0.55

0.60

100

80

60

40

20

0 50

0.65

60

Volumetric water content (cm3 water/cm3 soil)

80

90

100

Fig. 8. Scatter plot of N2O emissions against WFPS plotted. *, Otokia soil; *, Wingatui soil. Data are limited to emissions measured from +N cores on first sampling occasion only.


Fig. 6. Scatter plot of N2O emissions against volumetric water content (cm3/cm3). *, Otokia soil; *, Wingatui soil. Data are limited to emissions measured from +N cores on first sampling occasion only.


70

Water filled pore space (%)

1000.000

1000.000

100.000 y = 2 × 10 e R 2 = 0.874

–10 0.292x

10.000 1.000 0.100 0.010

y = 3 × 10–10e0.264x R 2 = 0.545

0.001 50

60

70

80

90

100

Water filled pore space (%)

100.000 y = 20.394e–146x R 2 = 0.715

10.000 1.000 0.100 0.010 0.001

y = 9.213e–193x R 2 = 0.445

0.00

0.02

0.04

0.06

0.08

0.10

Relative diffusivity

Fig. 7. Scatter plot of N2O emissions, plotted on log scale, against WFPS (%) (R2 = 0.72). Otokia soil, * and solid line; Wingatui soil, * dashed line. Data are limited to emissions measured from +N cores on first sampling occasion only.

Fig. 9. Scatter plot of N2O emissions, plotted on a log scale, against relative diffusivity (DP/DO) (R2 = 0.59). Otokia soil, * and solid line; Wingatui soil, * and dashed line. Data are limited to emissions measured from +N cores on first sampling occasion only.

may have a larger volume of disconnected micropores, which would reduce diffusion of mineral N and thus limit the number of denitrifying microsites. An understanding of the influence of soil moisture hysteresis on the relation between VWC and MP may help explain the variability in relationships between N2O emissions and soil water or gas status found in various studies and may therefore aid the prediction of N2O emissions. In this investigation, the relation between VWC and MP (Fig. 1) was determined by drainage; that is, the suction applied to the soil cores was increased, resulting in smaller pores progressively emptying. This is a standard technique, as it has been found to be more reliable than a wetting curve (Marshall and Holmes 1992). However, it is likely that a wetting curve will be different to a drainage curve. Field soils are either rapidly wetting up during rainfall/irrigation events, or drying down subsequent to these events; soil water content

will vary according to which cycle is occurring, thus influencing N2O production and emissions. Under field conditions, a rapid increase in pore continuity with drying will lead to a release of N2O produced earlier when the soil was wetter, and the N2O had been effectively trapped by the slow rate of diffusion through the wet pore space (Horgan and Ball 2005). This may result in higher N2O emissions at any given soil moisture content than in a wetting cycle. When reporting relationships between N2O emissions and soil water conditions it is therefore important to highlight whether this relationship was observed under a drying or a wetting cycle. This may improve the development of causal N2O and soil water relationships and thus aid their predictive power. In our investigation, we measured N2O emissions from soil cores held at a constant tension following N application; therefore, it is unlikely that the drying cycle released significant amounts of trapped N2O produced earlier under wetter conditions.


Proxy for gas diffusion Matric potential was the best independent variable relating to the measured emissions, with 93% of the variance being explained. This was significantly greater than the 88% of variance explained by VWC, which, in turn, was considerably and significantly greater than 72% and 59% explained by WFPS and RD, respectively. MP has been suggested as a more suitable soil water descriptor than WFPS across different soil types (Castellano et al. 2010). Using large, intact cores, those workers injected a nitrate solution equivalent to 100 kg N/ha into the top 15 cm and measured N2O emissions over a 96-h period. MP showed less variation than WFPS when related to maximum N2O fluxes, which appeared to peak at –4 kPa. In contrast, results from the current experiment and those by Müller (1996) suggest emissions were greatest when soils were near saturation (