Effects of changes in the soil environment associated with heavy precipitation on soil greenhouse gas fluxes in a Siberian larch forest near Yakutsk

Abstract

A future increase in heavy precipitation events is predicted in boreal regions. An irrigation experiment was conducted in Taiga forest in eastern Siberia to evaluate the effect of heavy precipitation on greenhouse gas ([GHG] CO₂, CH₄, and N₂O) fluxes in the soil. The GHG fluxes on the soil surface were measured using a closed-chamber method and GHG production rates in the mineral soil were estimated using the concentration–gradient method based on Fick’s law. Irrigation water (20 mm day⁻¹) was applied continuously for 6 days (120 mm in total; the same amount as summer precipitation in this region). Greenhouse gas production rates in the organic layer (O-layer) were defined as the difference between the GHG fluxes and the GHG production rates in the mineral soil. Carbon dioxide flux was measured both in root-intact (R _s ) and trenched plots (R _mw ). The root respiration rate (R _r ) was calculated as the difference between R _s and R _mw . Considering the root distribution in the soil, we regarded the CO₂ production rate in the mineral soil to be the microbial respiration rate in the mineral soil (R _mm ) and microbial respiration rate in the O-layer (R _mo ) as the difference between R _mw and R _mm . Irrigation increased both soil temperature and moisture in the irrigated plot. The R _s , CH₄ flux and N₂O flux during the irrigation period were higher in the irrigated plot than that in the non-irrigated plot (P < 0.05; mean R _s  ± standard deviation [SD] (mg C m⁻² h⁻¹) were 171 ± 20 and 109 ± 11, mean CH₄ flux ± SD (μg C m⁻² h⁻¹) were −5.4 ± 4.1 and −14.0 ± 6.5, and mean N₂O flux ± SD (μg N m⁻² h⁻¹) were 1.6 ± 1.6 and 0.2 ± 1.1, respectively). Soil moisture had a positive effect on R _mm and CH₄ production rate in the O-layer, a negative effect on R _r , and did not affect R _mo , the CH₄ production rate in the mineral soil or the N₂O production rates in both the O-layer and the mineral soil. Soil temperature had a positive effect on R _r and R _mo . The increment of global warming potential of the soil mainly resulted from an increase in microbial respiration rates. Future changes in precipitation patterns in this region would accelerate decomposition of the soil organic matter.

Key words:

• Fick’s law
• greenhouse gas
• irrigation
• Siberian Taiga
• trenching method

Introduction

Forests in Russia contain 381 Pg of carbon (C), which accounts for 54% of the total C stock in forests in the northern hemisphere (Goodale et al. 2002), and play an important role as a C sink. Taiga forest near Yakutsk in eastern Siberia lying on permafrost keeps the soil temperature low (Eugster et al. 2000), and the decomposition rate *of soil organic matter is very slow (Rodionow et al. 2006). In addition, the amount of precipitation in this region is very low (Japan Meteorological Agency 2004).

The Intergovernmental Panel on Climate Change (IPCC 2007) predicts that an increase in air temperature, in the amount of precipitation and in heavy precipitation events will occur in the permafrost region as a result of global warming caused by an increase in the concentration of greenhouse gases (GHGs). In Siberia, heavy precipitation events exceeding 20 mm day⁻¹ significantly increased during 1936–1994 (Easterling et al. 2000). An increase in air temperature will accelerate the growth of vegetation and increase the C storage of biomass and litter (Sirotenko and Abashina 2008). In contrast, increased soil temperature accompanied by thawing of permafrost will accelerate the decomposition rate of soil organic matter, enhancing global warming (Rodionow et al. 2006). It is still unknown what effects a change in precipitation pattern will have on the plant growth rate and the decomposition rate of soil organic matter in this region.

Many field studies have shown that precipitation or irrigation on dry soil increased the rate of soil respiration owing to an increase in soil moisture (e.g. Millard et al. 2008). Oberbauer et al. (1992) reported that precipitation events that raised the water table were found to strongly reduce soil respiration in the permafrost region in Alaska. However, very few studies have attempted to divide the effect of precipitation on soil respiration into microbial respiration and root respiration, which are the components of soil respiration (Bond-Lamberty et al. 2004). Dividing soil respiration is important to evaluate the effect of heavy precipitation on plant physiology and the decomposition rate of soil organic matter. Previous studies observed large increases in root and microbial respiration as a result of adding water to dry soil (e.g. Borken et al. 2003; Millard et al. 2008). However, the response of root and microbial respiration to heavy precipitation in the permafrost region is not clear.

In the permafrost region, heavy precipitation would affect not only the CO₂ flux from the soil, but also the exchange of the GHGs CH₄ and N₂O. In general, natural oxic soil absorbs CH₄ (Potter et al. 1996). Striegl et al. (1992) reported that precipitation on dry soil increased CH₄ consumption by the soil. Liu et al. (2008) reported that CH₄ fluxes showed little difference between irrigated and non-irrigated soils in an Inner Mongolian steppe soil. van Huissteden et al. (2008) reported that CH₄ consumption in a forest near Yakutsk remained stable even when the water table fluctuated from 9 to 27 cm. This indicates that methanotrophs occur in the uppermost soil and fluctuations in soil moisture do not affect CH₄ consumption in the forests of this region.

Very few studies have reported the effect of precipitation on N₂O emission from boreal forests. Rodionow et al. (2006) reported that emission of N₂O is negligible from natural Siberian soils. Forest soils near Yakutsk emit or absorb little N₂O (Morishita et al. 2007; Takakai et al. 2008). Du et al. (2006) reported that N₂O emission was stimulated by precipitation events in natural grassland in Inner Mongolia.

The purpose of the present study was to evaluate the effect of heavy precipitation on GHG fluxes from soil in Taiga forest near Yakutsk by conducting an irrigation experiment. In the present study, we measured GHG production rates both in the O-layer and in the mineral soil, and microbial respiration and root respiration separately to evaluate the effect of precipitation on the global warming potential of the soil in this ecosystem in detail.

Materials and methods

Site description

The present study was conducted during July 2004 in a 180-year-old larch forest on sandy loam soil underlain by permafrost (Typic Haploturbels; Soil Survey Staff 2010) in the Spasskaya-Pad experimental forest (62°15′N, 129°37′E) of the Institute for Biological Problems of Cryolithozone, near Yakutsk, Russia. The site is located on a gently sloped (1–2° decline toward north) terrace approximately 200 m in elevation situated on the left bank of the Lena River. The mean annual temperature is −10.0 °C and the mean annual precipitation is approximately 237 mm (Japan Meteorological Agency 2004). The amount of precipitation in 2004 preceding the experiment was lower than the mean precipitation for 30 years (Fig. 1), indicating that the forest floor was relatively dry during the experiment. The forest is mainly occupied by larch trees (Larix cajanderi). The forest floor is covered by shrubby vegetation, including Vaccinium vitis-idaea and Arctous erythrocarpa, and by a thick 10 cm organic layer (O-layer).

The amount of precipitation from 17 June to 26 July 2003 was recorded using a rain gauge (Young, Traverse, Michigan, USA) placed at the top of a 32-m high scaffolding tower built by Ohta et al. (2001).

Layout of the experimental plots

An irrigated plot and a non-irrigated plot of the same size (14 m × 14 m) were established. The non-irrigated plot was established in an area more than 50 m apart from the irrigated plot and was not affected by the irrigation (Fig. 2). Four subplots, measuring 0.16 m², were set up in spring 2003 in each irrigated and non-irrigated plot. Around each subplot, the outside edges of the 0.16 m² were trenched. The trenches were approximately 30 cm deep into the soil, which is deep enough to prevent the invasion of roots of Siberian larch; we estimated from our soil profile examination that approximately 97% of fine roots (<2 mm in diameter) are distributed within the −10 to 26 cm depth of the soil (Table 1; Ono 2003). The trenches were lined with plastic sheets before backfilling to prevent root growth. The above-ground parts of all vegetation were removed from the trenched plots.

Irrigation was carried out between 17.00 and 21.00 hours from 17 to 22 July 2004. Irrigation was accomplished using a rubber hose from a height of approximately 1.5 m. Irrigation of 20 mm per day was conducted every day and a total of 120 mm of irrigation water was applied. The amount of irrigation was approximately equal to the average amount of natural precipitation, including heavy precipitation events, during the summer season (June–September). Lena River water was used for the irrigation and the temperature of the irrigated water was approximately 15°C.

Figure 1 Seasonal variation in air temperature and precipitation in Yakutsk. Open and closed circles indicate the average air temperature for each month over 30 years and in 2004, respectively. Open and closed bars indicate the average precipitation for each month over 30 years and the amount of precipitation for each month in 2004, respectively.

Figure 2 Location map of the study site and the layout of the measurement plots. Microbial respiration in the whole soil was measured in the trenched plots. Fluxes and concentrations of greenhouse gases in soil air were measured near the trenched plots.

Table 1 Soil physical and chemical properties at the study site

Layer^†	Depth (cm)	Bulk density (Mg m^3)	Porosity (m^3 m^−3)	D/D0 ^‡	NO3-N conc.(mgN kg soil^−1)	NH4-N conc.(gN kg soil^−1)	Length of fine roots§ (cm cm^−3)	Soil texture
O	0∼10	0.41	–	–	0.228	0.020	8.0	–
A	∼25	1.10	0.55	0.11	0.023	0.005	0.6	SCL
B1	∼36	1.55	0.41	0.03	0.140	0.006	0.4	SL
B21	∼50	1.63	0.40	0.07	0.094	0.005	0.1	LS
B22Ca	∼72	1.52	0.37	0.06	0.101	0.018	0.2	SL
B23	∼78	1.65	0.39	0.11	–	–	0.0	LS
B3Ca	∼91	1.86	0.34	0.01	–	–	0.0	SL
^†O, O-layer (thickness of the O-layer was 10 cm); A and B, Mineral soil layers (with organic matter and illuviation, respectively); Ca, accumulation of carbonate. ^‡ D/D 0, gas tortuosity factor. ^§Length of fine roots (<2 mm in diameter) was calculated after Ono (2003). SCL, sandy clay loam; SL, sandy loam, LS, loamy sand.

Field measurements

The CO₂ fluxes from the surface of the O-layer (R _s ) were measured using a closed-chamber method (Kusa et al. 2008; Morishita et al. 2005, 2006, 2007; Takakai et al. 2008) from 1 to 23 July 2004. To measure R _s , including root and microbial respiration, four stainless steel cylinders, approximately 20 cm in diameter and 25 cm height, were installed at a depth of 4 cm into the soil both in the irrigated and non-irrigated plots at least 24 h before measurement. The above-ground parts of all plants inside the cylinders were removed carefully to exclude plant respiration. Before beginning the measurements, an air sample inside the cylinder was transferred to a Tedlar bag (OMI Odorair Service, Shiga, Japan) as a 0-min sample, and an acrylic lid with silicone gasket and an inflatable plastic bag to adjust the air pressure inside the chamber was placed on the cylinder immediately. Six minutes later, a 500 mL air sample from inside the chamber was taken out. Each chamber remained open for more than 10 min after completion of the CO₂ flux measurement to allow for a complete exchange of air inside the chamber. Gas samples were then taken to measure CH₄ and N₂O fluxes on the surface of the O-layer. At 0, 30 and 60 min after the top of the chamber was again closed with a lid, a 20 mL gas sample from inside the chamber was taken into a 10 mL vacuum vial and the vial was sealed with a butyl rubber stopper (SVF-10; Nichiden-Rika, Kobe, Japan).

One stainless steel cylinder was installed into the middle of each trenched plot in both the irrigated and non-irrigated plots in four replicates to measure the rate of microbial respiration in the whole soil (R _mw ). The measurements were carried out on the same day as the gas sampling. The procedure for measuring R _mw was the same as that used to measure CO₂.

Stainless steel pipes, 9 mm in diameter, were installed at depths of 10, 20, 30 and 50 cm (four pipes at each depth) out of the trenched plot in both the irrigated and non-irrigated plots in the spring of 2003 to collect air samples from the soil. The pipes were sealed by three-way cocks to allow the gas concentration in the pipes to equilibrate with the soil air. From each of the pipes installed at each depth and each treatment, air samples (50 mL) were taken into a Tedlar bag during the chamber measurements. All samples taken from the same depth and treatment were mixed into a Tedlar bag. Air samples of 50 mL from each of the Tedlar bags were diluted with CO₂-free air for CO₂ analysis, and 20 mL was transferred to 10 mL glass bottles for CH₄ and N₂O analysis.

These measurements were conducted between 10.00 and 18.00 hours to avoid diurnal effects on fluxes. We did not carry out measurements within 12 h of irrigation to avoid effects of the physical movement of water in the soil.

During the chamber measurements, the air and soil temperature and the soil moisture around each chamber were recorded manually. The air temperature around each chamber was measured with a digital thermometer to calculate GHG fluxes. Two replicate soil temperature measurements at a depth of 10 cm near each chamber, which is the boundary between the O-layer and the mineral soil, were also taken using digital thermometers. Two replicate soil moisture measurements as volumetric water content at a soil depth of 0–10 cm (O-layer) from each chamber base were taken with a calibrated time domain reflectometer (TDR) sensor (TRIME-como; Tohoku Electronic Industrial Company, Sendai, Japan).

Soil temperature and moisture were recorded automatically approximately 5 m away from each chamber in both the irrigated and non-irrigated plots (Lopez et al. 2007). Soil temperature was measured using calibrated thermistors at depths of 10, 20, 30, 40 and 60 cm (T ₁₀, T ₂₀, T ₃₀, T ₄₀, and T ₆₀, respectively; 104ET, Ishizuka Denshi, Tokyo, Japan). Because of some sensor troubles with a thermistor at the 20 cm depth, we excluded the T ₂₀ data from the analysis. The thermistors were calibrated using an ice-water bath with a precision of 0.02°C at 0°C; the overall probe accuracy for the temperature range of −20 to 30°C was calculated to be better than ±0.09°C. T ₁₀ was recalibrated using soil temperature measured manually around the chambers. Soil moisture was measured using the frequency domain reflectometry (FDR) method at depths of 0–10, 10–20, 20–30, 30–40 and 50–60 cm (W₁₀, W₂₀, W₃₀, W₄₀ and W₆₀, respectively; EnviroSMART, Sentek, South Australia, Australia). The FDR sensors were calibrated separately for depths of 0–10 cm and for depths below 10 cm (mineral soil layer). For this purpose, we collected 11 in situ soil samples of each soil depth with various moisture conditions and determined the volumetric water content gravimetrically to construct a calibration curve. W₁₀ was recalibrated using soil moisture measured manually around the chambers. Soil temperature was recorded every 30 s and stored every 1 h on average, whereas soil moisture was recorded every 1 h. The soil temperature and moisture recorded at the time of gas sampling were used for the statistical analyses.

Physical and chemical conditions of the soil and water samples

To analyze the physical properties of the soil and water samples, three soil cores 100 cm³ in size were sampled from each layer in 2003. With these undisturbed cores, we measured the volume of the air-filled pore space (AFPS) using a three phase meter (Model DIK-1110; Daiki Rika Kogyo Co., Saitama, Japan). The gas tortuosity factor (D/D ₀) of the undisturbed soil cores was measured and calculated using the method developed by Osozawa and Kubota (1987) under steady-state conditions using O₂ as the diffusing gas. The cores were oven-dried (105°C) for 48 h to obtain bulk density and water-filled pore space (WFPS). The porosity was calculated as the sum of WFPS and AFPS. The length of the fine roots (<2 mm in diameter) in each layer was calculated using a theoretical equation after Ono (2003) as follows:

where R is the length of the fine roots in a unit volume of each layer (cm cm⁻³; Table 1), n is the number of the cut ends of fine roots appearing on each layer in the soil profile and S is the area of each layer in the soil profile (cm⁻²).

Fresh soil samples were sieved (2 mm) and used for the chemical analysis. The concentrations of NO₃ ⁻-N in a 1:20 for the O-layer and a 1:10 for the mineral soil : deionized water mixture were measured using ion chromatography (TOA DIC Analyzer ICA-2000; TOA DKK, Tokyo, Japan). The concentrations of NH₄ ⁺-N in a 1:20 for the O-layer and a 1:10 for the mineral soil : 1 mol L⁻¹ KCl solution mixture were measured using colorimetry (the indophenol-blue method) with a UV-VIS spectrophotometer (UV mini 1240; Shimadzu, Kyoto, Japan).

Irrigated water was sampled directly from the rubber hose that was used to apply the irrigation water. Rain water was sampled using eight rain gutters (0.12 m² each) that had been placed around the irrigated and non-irrigated plots.

The concentrations of NH₄ ⁺-N and NO₃ ⁻-N in the irrigation and rain water were also measured. The contents of NH₄ ⁺-N and NO₃ ⁻-N in the soil were calculated by multiplying the bulk density by the NH₄ ⁺-N or NO₃ ⁻-N concentration of each layer of the soil. The contents of NH₄ ⁺-N and NO₃ ⁻-N in the irrigation and rain water were calculated by multiplying the amount of irrigation or rain water by the NH₄ ⁺-N or NO₃ ⁻-N concentrations of the water.

Analysis of the CO₂, CH₄ and N₂O concentrations

The CO₂ concentration was determined with a portable infrared CO₂ gas analyzer (ZFP9GC11; Fuji Electric Systems, Tokyo, Japan). The CH₄ and N₂O concentrations were analyzed using gas chromatography equipped with a flame ionization detector and an electron capture detector, respectively (GC-8A and GC-14B; Shimadzu).

Estimation of soil-gas diffusivity and calculation of flux, respiration and production rates of greenhouse gases

Greenhouse gas fluxes, measured using a closed-chamber method, were calculated as follows:

where F _c is the flux of GHGs obtained using the closed-chamber method (μg C or N m⁻² h⁻¹), ρ is the gas density of GHGs (CO₂-C, CH₄-C, 0.536 × 10³ g m⁻³; N₂O-N, 1.25 × 10³ g m⁻³), V/A is the height of the chamber (m), dc/dt is the ratio of change in the gas concentration (c) inside the chamber per unit time (t) during the sampling period (m³ m⁻³ h⁻¹) and T is the air temperature (ºC).

We estimated the flux of each GHG from the mineral soil to the O-layer using a soil gradient method (Kusa et al. 2008). The soil-gas diffusivities were estimated using the soil AFPS (Fig. 3). Estimated diffusivities were corrected for temperature effects using D_T/D₂₅ = [(273 + T)/298]^1.75 (Reid et al. 1977). Greenhouse gas flux from the mineral soil to the O-layer was determined from Fick’s law as:

where F _d is the flux obtained from Fick’s law (μg C or N cm⁻² s⁻¹), D is the soil gas coefficient (cm² s⁻¹) at the surface of the mineral soil calculated using the gas tortuosity factor (D/D ₀) and the molecular diffusivity of each GHG in air (D_0CO2  = 0.138, D_0CH4  = 0.191, D_{0N2 O} = 0.143) and dc/dz is the ratio of change in the gas concentration (c) along the soil depth (z) (μg C or N cm⁻³ cm⁻¹). We made quadratic functions of depth (10, 20, 30 and 50 cm) to each GHG concentration to calculate dc/dz at the 10-cm depth, the boundary between the O-layer and the mineral soil, as the slope of the tangent to the quadratic functions at the 10-cm depth (Takle et al. 2004). We did not use the GHG concentration data at the 0-cm depth, ambient air, to make the quadratic functions because there is always some variation in the gas concentrations owing to the turbulent mixing of ambient air with to some extent the topmost O-layer (Pihlatie et al. 2007).

Figure 3 Effect of soil air-filled pore space on the measured gas tortuosity factor (diffusivity relative to that for still air). The different symbols distinguish measurements on soil cores from different layers.

The root respiration rate (R _r ) of each measurement was calculated as the difference between the CO₂ flux from the surface of the O-layer in the root-intact (R _s ) and trenched plots (the microbial respiration rate in the whole soil [R _mw ]) of each measurement. One negative value of R _r was eliminated. We regarded the estimated GHG fluxes from the mineral soil into the O-layer as the GHG production rates in the mineral soil, assuming no movement of GHGs in the permafrost. The GHG production rates in the O-layer were calculated as the difference between the GHG fluxes from the surface of the O-layer and the production rates of GHG in the mineral soil to avoid problems caused by the turbulent mixing of ambient air and the unevenly distributed O-layer constituents (Davidson and Trumbore 1995; Pihlatie et al. 2007; Schwendenmann and Veldkamp 2006). For CO₂, we regarded the CO₂ production rates in the mineral soil as the microbial respiration rates in the mineral soil (R _mm ) assuming that almost all R _r was occurring in the O-layer because in the most part the fine roots were distributed in the O-layer (Table 1). The microbial respiration rate in the O-layer (R _mo ) was calculated as the difference between R _mw and R _mm .

The following equations were applied to the measured temperature and CO₂ flux or production rate data (F _CO2 ) to calculate the normalized CO₂ flux or the production rate (F _CO2,b) corresponding to the base temperature (T _b, 5°C in the present study):

where a and b are fitted constants and T is the soil temperature at the 10-cm depth (Takakai et al. 2008).

Negative fluxes and production rates of CH₄ and N₂O represent downward movement or net consumption of the gases.

To evaluate the relative significance of CO₂, CH₄ and N₂O, their global warming potentials ([GWP] mg CO₂-eq m⁻² h⁻¹) expressed in CO₂-equivalent values were calculated by multiplying the CO₂, CH₄ and N₂O fluxes by 44/12, 16/12 × 25 and 44/28 × 298, respectively (100-year time horizon; IPCC 2007).

Statistical analyses

Two-way anovas or paired t-tests were carried out to compare the average GHG fluxes, GHG production, soil temperature or soil moisture obtained from the irrigated and non-irrigated plots before and during the irrigation period.

Linear or non-linear regression analyses were carried out to examine the relationship between the measured GHG flux and production, soil temperature and moisture. Spearman’s correlation analyses were carried out to examine the relationship between W₂₀ and other variables because of the non-normal distribution of W₂₀.

A multiple regression analysis was used to evaluate the contributions of R _r , R _mo , R _mm , CH₄ production in the O-layer and the mineral soil, and N₂O production in the O-layer and the mineral soil to GWP.

Significance for all statistical analyses was accepted at P = 0.05. All statistical analyses were carried out using the statistical package R, version 2.9.1 (R Development Core Team 2009).

Results

Physical and chemical properties of the soil

The soil physical and chemical properties are shown in Table 1. The dry bulk density in the O-layer was 0.41 and the density in the mineral soil ranged from 1.1 to 1.9 Mg m⁻³, and increased as the depth increased. The porosity in the mineral soil ranged from 0.34 to 0.55 m³ m⁻³ and D/D ₀ in the mineral soil before the irrigation experiment ranged from 0.01 to 0.11 m² s⁻¹, and both decreased with an increase in depth. The NO₃ ⁻-N and NH₄ ⁺-N concentrations in each layer ranged from 0.02 to 0.23 mg kg soil⁻¹ and from 0.00 to 0.02 g kg soil⁻¹, respectively, and both were highest in the O-layer, whereas there was no trend in the mineral soil. The NH₄ ⁺-N contents in the O-layer and in the mineral soil (1–62 cm) were 816 and 9,160 mg m⁻², respectively, and the NO₃ ⁻-N contents were 9 and 82 mg m⁻², respectively. The estimated length of fine roots in each layer ranged from 0 to 80 m L⁻¹ and 86% of the fine roots were distributed in the O-layer.

Precipitation, soil temperature and soil moisture

A total of 11.4 mm of precipitation occurred during the 30 days before the irrigation period. The amount of precipitation from 17 June to 27 July was 19.8 mm.

The soil temperature at depths of 10, 30, 40 and 60 cm (T ₁₀, T ₃₀, T ₄₀ and T ₆₀, respectively) during the flux measurements ranged from −0.5 to 7.6 °C in the irrigated plot and from −0.0 to 6.9 °C in the non-irrigated plot (Fig. 4a,b). Before the irrigation period, the average soil temperature at each depth in the irrigated plot was significantly lower than that in the non-irrigated plot (Table 2; P < 0.01). During the irrigation period, the average T ₁₀ in the irrigated plot was significantly higher than that in the non-irrigated plot (P < 0.01), the average T ₆₀ in the irrigated plot was significantly lower than that in the non-irrigated plot (P < 0.001), and the average T ₃₀ and T ₄₀ did not differ between the irrigated and non-irrigated plots.

Soil moisture in the O-layer (W₁₀) during the flux measurements ranged from 0.11 to 0.14 m³ m⁻³ in the irrigated plot and from 0.10 to 0.13 m³ m⁻³ in the non-irrigated plot (Fig. 4c,d). Soil moisture at depths of 10–20, 20–30, 30–40 and 50–60 cm (W₂₀, W₃₀, W₄₀, and W₆₀, respectively) ranged from 0.13 to 0.31 m³ m⁻³ in the irrigated plot and from 0.10 to 0.16 m³ m⁻³ in the non-irrigated plot. Soil moisture increased with an increase in depth in both plots. Although the average soil moisture at each depth in the irrigated plot was significantly higher than that in the non-irrigated plot, both before and during the irrigation period (Table 2; P < 0.01), soil moisture in the irrigated plot tended to increase with an increase in irrigation (Fig. 4D). W₂₀ in the irrigated plot was significantly higher during the irrigation period than before the irrigation period and W₂₀ in the non-irrigated plot was significantly lower during the irrigation period than before the irrigation period (both at P < 0.001).

Figure 4 (a,b) Temporal variation in soil temperature at 10, 30, 40 and 60 cm depths (T10, T30, T40, and T60, respectively) and (c,d) soil moisture at 0–10, 10–20, 20–30, 30–40 and 50–60 cm depths (W10, W20, W30, W40, and W60, respectively) of the non-irrigated (a,c) and irrigated plots (b,d). Arrows indicate the irrigation events for the irrigated plot. Open bars represent the precipitation amount.

Table 2 Average greenhouse gas fluxes, production, global warming potentials and soil temperature and moisture during each period of measurement

	Before irrigation period		During irrigation period
	Non-irrigated plot	Irrigated plot	Non-irrigated plot	Irrigated plot
R s  (mg C m^−2h^−1)	116 ± 21^a	106 ± 32^a	109 ± 11^a	171 ± 20^b
R r  (mg C m^−2 h^−1)	36 ± 11^a	45 ± 23^a	45 ± 8^a	49 ± 14^a
R mw  (mg C m^−2 h^−1)	80 ± 15^b	67 ± 15^a	63 ± 6^a	122 ± 10^b
R mo  (mg C m^−2 h^−1)	64 ± 13^b	44 ± 16^a	54 ± 5^a	97 ± 8^b
R mm  (mg C m^−2 h^−1)	17 ± 3^a	24 ± 4^b	9 ± 3^a	25 ± 2^b
CH4(w) (μg C m^−2 h^−1)	−8.3 ± 3.4^a	−8.3 ± 3.5^a	−14.0 ± 6.5^a	−5.4 ± 4.1^b
CH4(o) (μg C m^−2 h^−1)	3.8 ± 5.4^a	3.6 ± 4.3^a	−5.67 ± 6.3^a	6.1 ± 4.0^b
CH4(m) (μg C m^−2 h^−1)	−12.0 ± 3.3^a	−11.9 ± 2.4^a	−8.4 ± 1.9^a	−11.5 ± 2.4^a
N2O(w) (μg N m^−2 h^−1)	1.21 ± 1.56^a	0.13 ± 0.63^a	0.21 ± 1.12^a	1.58 ± 1.64^b
N2O(o) (μg N m^−2 h^−1)	1.06 ± 1.52^a	0.06 ± 0.63^a	0.32 ± 1.11^a	1.44 ± 1.71^a
N2O(m)  (μg N m^−2 h^−1)	0.15 ± 0.22^a	0.07 ± 0.19^a	−0.11 ± 0.26^a	0.14 ± 0.26^a
T 10 (°C)	5.53 ± 1.04^b	5.06 ± 1.43^a	6.92 ± 0.60^a	7.63 ± 0.28^b
T 20 (°C)	7.62 ± 1.68^b	4.92 ± 1.18^a	7.76 ± 0.46^b	6.77 ± 0.46^a
T 30 (°C)	4.61 ± 1.09^b	2.97 ± 0.89^a	4.84 ± 0.21^a	5.09 ± 0.52^a
T 40 (°C)	3.12 ± 0.90^b	1.60 ± 0.76^a	3.54 ± 0.19^a	3.69 ± 0.52^a
T 60 (°C)	1.22 ± 0.79^b	−0.08 ± 0.30^a	1.93 ± 0.07^b	1.18 ± 0.49^a
W10 (m^3 m^−3)	0.12 ± 0.01^a	0.13 ± 0.01^b	0.10 ± 0.00^a	0.12 ± 0.00^b
W20 (m^3 m^−3)	0.12 ± 0.00^a	0.24 ± 0.01^b	0.11 ± 0.00^a	0.26 ± 0.00^b
W30 (m^3 m^−3)	0.15 ± 0.00^a	0.24 ± 0.00^b	0.14 ± 0.00^a	0.26 ± 0.01^b
W40 (m^3 m^−3)	0.15 ± 0.00^a	0.21 ± 0.01^b	0.15 ± 0.00^a	0.24 ± 0.01^b
W60 (m^3 m^−3)	0.13 ± 0.02^a	0.22 ± 0.03^b	0.16 ± 0.02^a	0.28 ± 0.02^b
Within the group of greenhouse gas fluxes, production and soil temperature and moisture, values with different letters are significantly different at P < 0.05. R s, soil respiration rate; R r  , root respiration rate; R mw , R mo , R mm , microbial respiration rate from the whole soil, O-layer and mineral soil, respectively; CH4(w), CH4(o), CH4(m), CH4 production rates in the whole soil, O-layer and mineral soil, respectively; N2O(w), N2O(o), N2O(m), N2O production rates in the whole soil, O-layer and mineral soil, respectively; T i, soil temperature at i cm depth; Wi, soil moisture from (i−10) to i cm depth.

Nitrogen supply by irrigation and precipitation

The NH₄ ⁺-N and NO₃ ⁻-N concentrations in the irrigation water were 0.09 and 0.10 mg L⁻¹ and the concentrations in the rain water were 0.14 and 0.005, respectively. The NH₄ ⁺-N and NO₃ ⁻-N supply to the soil by irrigation were 11 and 12 mg m⁻² and by rain water were 2.8 and 0.1, respectively. A total of 13.8 mg m⁻² of NH₄ ⁺-N and 12.1 of NO₃ ⁻-N were added to the soil over the measurement period.

Greenhouse gas flux from the surface of the O-layer

The CO₂ flux from the surface of the O-layer in the root-intact plot (R _s ) ranged from 59 to 200 mg C m⁻² h⁻¹ in the irrigated plot and from 70 to 142 in the non-irrigated plot, and both fluctuated with variation in the soil temperature (Figs 4a,b,5a). Before the irrigation period the average R _s (mean ± standard deviation [SD]) in the irrigated plot (106 ± 32 mg C m⁻² h⁻¹; Table 2) did not differ from that in the non-irrigated plot (116 ± 21). During the irrigation period, the average R _s in the irrigated plot (171 ± 20) was significantly higher than that in the non-irrigated plot (109 ± 11; P < 0.001). The CO₂ flux from the surface of the O-layer in the trenched plot (microbial respiration rate in whole soil; R _mw ) ranged from 50 to 132 mg C m⁻² h⁻¹ in the irrigated plot and from 54 to 109 in the non-irrigated plot, and accounted for 42–108% (mean ± SD; 67 ± 12%) of R _s (Fig. 5b). This percentage was high at the beginning of the measurement period and decreased with time. Before the irrigation period, the average R _mw (mean ± SD) in the irrigated plot (67 ± 15 mg C m⁻² h⁻¹; Table 2) was significantly lower than that in the non-irrigated plot (80 ± 15; P < 0.05). During the irrigation period, the average R _mw in the irrigated plot (122 ± 10) was significantly higher than that in the non-irrigated plot (63 ± 6; P < 0.001).

The ranges of the root respiration rate (R _r ), calculated from the difference between R _s and R _mw , were 26–90 and 12–53 mg C m⁻² h⁻¹ in the irrigated and non-irrigated plots, respectively, accounting for 17–58% (mean ± SD; 35 ± 9%) of R _s (Fig. 5c). This percentage was low at the beginning of the measurement period and increased with an increase in soil temperature in the non-irrigated plot, whereas irrigation depressed the rate in the irrigated plot. The average R _r in the irrigated plot did not differ from that in the non-irrigated plot before or during the irrigation period (Table 2).

Figure 5 Temporal variation in (a) CO2 fluxes from the surface of the O-layer (Rs), (b) microbial respiration from the surface of the O-layer (Rmw) and (c) root respiration (Rr). Open symbols indicate data from the non-irrigated plot and closed symbols indicate data from the irrigated plot. Vertical bars denote the standard error of each parameter. Arrows indicate the irrigation events for the irrigated plot.

The CH₄ flux from the surface of the O-layer ranged from −15 to −0 μg C m⁻² h⁻¹ in the irrigated plot and from −22 to −1 in the non-irrigated plot (Fig. 6a). Before the irrigation period, the average CH₄ flux (mean ± SD) in the irrigated plot (−8.3 ± 3.4 μg C m⁻² h⁻¹; Table 2) did not differ from that in the non-irrigated plot (−8.3 ± 3.5). During the irrigation period, the CH₄ flux in the irrigated plot (−5.4 ± 4.1) was significantly higher than that in the non-irrigated plot (−14.0 ± 6.5; P < 0.001).

The N₂O flux from the surface of the O-layer ranged from −0.9 to 3.9 μg N m⁻² h⁻¹ in the irrigated plot and from −0.7 to 3.8 in the non-irrigated plot (Fig. 6b). Before the irrigation period, the average N₂O flux (mean ± SD) from the irrigated plot (0.13 ± 0.63 μg N m⁻² h⁻¹; Table 2) did not differ from that in the non-irrigated plot (1.21 ± 1.56 μg N m⁻² h⁻¹). During the irrigation period, the N₂O flux from the irrigated plot (1.58 ± 1.64 μg N m⁻² h⁻¹) was significantly higher than that from the non-irrigated plot (0.21 ± 1.12 μg N m⁻² h⁻¹; P < 0.05).

Greenhouse gas concentration in the soil

The ranges of CO₂ concentration of soil air at depths of 10, 20, 30 and 50 cm were 720–4,740 p.p.m.v in the irrigated plot and 900–3,600 p.p.m.v in the non-irrigated plot (Fig. 7a,b). The CO₂ concentrations increased with an increase in depth in both plots.

The ranges of CH₄ concentration of soil air at depths of 10, 20, 30 and 50 cm were 0.30–2.07 p.p.m.v in the irrigated plot and 0.34–2.16 p.p.m.v in the non-irrigated plot (Fig. 7c,d). The CH₄ concentration decreased with an increase in depth in both plots.

The ranges of N₂O concentration of soil air at depths of 10, 20, 30 and 50 cm were 0.30–0.33 p.p.m.v in the irrigated plot and 0.29–0.33 p.p.m.v in the non-irrigated plot (Fig. 7e,f). The N₂O concentration was independent of the depth of soil in both plots.

Figure 6 Temporal variation in (a) CH4 fluxes from the surface of the O-layer and (b) N2O fluxes from the surface of the O-layer of the non-irrigated (open symbols) and irrigated plots (closed symbols). Positive values indicate net emission and negative values indicate net consumption. Vertical bars denote the standard error of each parameter. Arrows indicate the irrigation events for the irrigated plot.

Figure 7 Temporal variation in (a,b) CO2, (c,d) CH4 and (e,f) N2O concentrations (conc.) in soil air at 10, 20, 30 and 50 cm depths of the non-irrigated (a,c,e) and irrigated plots (b,d,f). Data from 10, 20, 30 and 50 cm depths are indicated by open circles, closed circles, open triangles and closed triangles, respectively. Arrows indicate the irrigation events for the irrigated plot.

Greenhouse gas production rate

The ranges of the microbial respiration rate in the O-layer (R _mo ) were 29–105 and 40–80 mg C m⁻² h⁻¹ in the irrigated and non-irrigated plots, respectively, and accounted for 26–62% (mean ± SD; 50 ± 9%) of R _s (Fig. 8a). This percentage was high at the beginning of the measurement period and decreased with an increase in soil temperature in the non-irrigated plot, whereas irrigation enhanced the rate in the irrigated plot. Before the irrigation period, the average R _mo (mean ± SD) in the irrigated plot (44 ± 16 mg C m⁻² h⁻¹; Table 2) was significantly lower than that in the non-irrigated plot (64 ± 13; P < 0.001). During the irrigation period, the average R _mo in the irrigated plot (97 ± 8 mg C m⁻² h⁻¹) was significantly higher than that in the non-irrigated plot (54 ± 5; P < 0.001). The microbial respiration rate in the mineral soil (R _mm ) ranged from 18 to 29 mg C m⁻² h⁻¹ in the irrigated plot and from 5 to 22 mg C m⁻² h⁻¹ in the non-irrigated plot, and accounted for 5–49% (mean ± SD; 16 ± 9%) of R _s (Fig. 8a). This percentage was high at the beginning of the measurement period and decreased with time. The average R _mm values before and during the irrigation period (mean ± SD) in the irrigated plot were 24 ± 4 and 25 ± 2 mg C m⁻² h⁻¹ and both values were significantly higher than the values recorded in the non-irrigated plot (17 ± 3 and 9 ± 3 mg C m⁻² h⁻¹, respectively; P < 0.01; Table 2).

Figure 8 Temporal variation in (a) microbial respiration rate in the O-layer and the mineral soil (Rmo are indicated by circles and Rmm are indicated by triangles, (b) CH4 production rates in the O-layer and the mineral soil (circles and triangles, respectively) and (c) N2O production rates in the O-layer and the mineral soil (circles and triangles, respectively) of the non-irrigated (open symbols) and irrigated plots (closed symbols). Positive values indicate net emission and negative values indicate net consumption. Arrows indicate the irrigation events for the irrigated plot.

Table 3 Average global warming potentials of root respiration (Rr), microbial respiration in the O-layer (Rmo), microbial respiration in the mineral soil (Rmm), and CH4 and N2O production of each part of the soil in each period (expressed in CO2-equivalent; mg CO2-eq m−2 h−1) from the study site

	Before irrigation		During irrigation
	Non-irrigated plot (%)	Irrigated plot (%)	Non-irrigated plot (%)	Irrigated plot (%)
R r	131^a (31)	166^a (43)	167^a (42)	180^a (29)
R mo	233^b (55)	163^a (42)	198^a (50)	355^b (56)
R mm	61^a (14)	86^b (22)	34^a (8)	93^b (15)
CH4 production in the O-layer	0.125^a (0.03)	0.122^a (0.03)	−0.188^a (−0.05)	0.202^b (0.03)
CH4 production in mineral soil	−0.401^a (−0.09)	−0.397^a (−0.10)	−0.280^a (−0.07)	−0.383^a (−0.06)
N2O production in the O-layer	0.496^a (0.12)	0.029^a (0.01)	0.150^a (0.04)	0.674^a (0.11)
N2O production in mineral soil	0.069^a (0.02)	0.031^a (0.01)	−0.051^a (−0.01)	0.064^a (0.01)
Net GWP	425^a (100)	390^a (100)	398^a (100)	628^b (100)
Within the group of greenhouse gas fluxes, production and soil temperature and moisture, values with different letters are significantly different at P < 0.05. Positive values indicate net production and negative values indicate net consumption by the soil. GWP, global warming potential.

The CH₄ production rate in the O-layer ranged from −4 to 11 μg C m⁻² h⁻¹ in the irrigated plot and from −13 to 12 in the non-irrigated plot (Fig. 8b). Before the irrigation period, the average CH₄ production rate in the O-layer (mean ± SD) in the irrigated plot (3.6 ± 4.3 μg C m⁻² h⁻¹; Table 2) did not differ from that in the non-irrigated plot (3.8 ± 5.4). During the irrigation period, the CH₄ production rate in the irrigated plot (6.1 ± 4.0) was significantly higher than that in the non-irrigated plot (−5.7 ± 6.3; P < 0.05). The CH₄ production rate in the mineral soil ranged from −17 to −8 μg C m⁻² h⁻¹ in the irrigated plot and from −17 to −5 in the non-irrigated plot (Fig. 8b). The average CH₄ production rate in the mineral soil in the irrigated plot did not differ from that in the non-irrigated plot before and during the irrigation period (Table 2).

Figure 9 (a,b) CO2 fluxes from the surface of the O-layer (Rs), (d,e) respiration from the surface of the O-layer (Rmw), (g,h) root respiration (Rr), (j,k) microbial respiration rate in the O-layer (Rmo) and (c,f,i,l) normalized Rs, Rmw, Rr and Rmo plotted against soil temperature at a depth of 10 cm (a,d,g,j) and moisture at the 0–10 cm depth (b,c,e,f,h,i,k,l). Open and closed circles indicate data from the non-irrigated and irrigated plots before irrigation and open and closed triangles indicate data from non-irrigated and irrigated plots during irrigation. †P < 0.1; *P ≤ 0.05; **P < 0.01; ***P < 0.001. Error bars indicate the standard error of each parameter.

The N₂O production rate in the O-layer ranged from −1.0 to 3.8 μg N m⁻² h⁻¹ in the irrigated plot and from −1.7 to 3.6 μg N m⁻² h⁻¹ in the non-irrigated plot. The N₂O production rate in the mineral soil ranged from −0.2 to 0.6 μg N m⁻² h⁻¹ in the irrigated plot and from −0.6 to 0.5 in the non-irrigated plot (Fig. 8c). Average N₂O production rates in both the O-layer and the mineral soil in the irrigated plot did not differ from the rates in the non-irrigated plot before and during the irrigation period (Table 2).

Global warming potentials

Before the irrigation period, the GWP of the soil (mean ± SD) in the irrigated plot (390 ± 117 mg CO₂-eq m⁻² h⁻¹) did not differ from that in the non-irrigated plot (425 ± 78). During the irrigation period, the GWP in the irrigated plot (628 ± 75) was significantly higher than that in the non-irrigated plot (398 ± 39; P < 0.001; Table 3). Carbon dioxide production accounted for more than 99% of the GWP in both the irrigated and non-irrigated plots both before and during the irrigation period. The GWP increment resulting from irrigation was mainly caused by an increase in microbial respiration. R _mo made the greatest contribution to GWP in both the irrigated and non-irrigated plots both before and during the irrigation period. The GWP by only CH₄ and N₂O production during the irrigation period in the non-irrigated plot was negative (−0.20 mg CO₂-eq m⁻² h⁻¹) because CH₄ consumption by the soil offset the GWP by N₂O production. However, the GWP in the irrigated plot was positive (0.56 mg CO₂-eq m⁻² h⁻¹), mainly owing to an increase in GWP by N₂O production, which could not be offset by CH₄ consumption.

Controlling factors for greenhouse gas fluxes

Non-linear regressions revealed a significant positive correlation between T ₁₀ and R _s , R _r , R _mw and R _mo (r = 0.87, 0.53, 0.68 and 0.72, respectively, P < 0.01; Fig. 9a,d,g,j). No significant correlation between T ₁₀ and R _mm was recorded (r = 0.11; Fig. 12a). Normalized R _mw and R _mm values were significantly and positively correlated with W₁₀ and W₂₀, respectively, whereas normalized R _r was significantly and negatively correlated with W₁₀ (r = 0.64, 0.76 and −0.38, respectively, P ≤ 0.05; Figs 9f,i 12c). Normalized R _mo had a weak positive correlation with W₁₀ (r = 0.36, P = 0.07; Fig. 9l). Both the CH₄ flux from the surface of the O-layer and the CH₄ production rate in the O-layer had significant positive correlations with W₁₀ (r = 0.60 and 0.62, respectively, P < 0.001; Fig. 10b,d). The CH₄ production rate in the mineral soil was not correlated with either T ₁₀ or W₂₀ (Fig. 12d,e). Each N₂O flux or N₂O production rate did not significantly correlate with either soil temperature or moisture (Figs 11 12f,g). The N₂O flux from the surface of the O-layer had a significant positive correlation with R _mw (r = 0.45, P < 0.05; Fig. 13).

We carried out multiple regression analyses to evaluate the contributions of R _r , R _mo , R _mm , CH₄ production in the O-layer and the mineral soil, and N₂O production in the O-layer and the mineral soil to GWP (Table 4). The results showed that R _mo had the largest effect on the GWP followed by R _r and R _mm . This indicates that microbial activity plays an important role in controlling GWP of the soil in this region. The production rates of CH₄ and N₂O had little effect on the GWP.

Discussion

Effects of irrigation on soil temperature and moisture

Both soil temperature at a depth of 10 cm (temperature of the boundary between the O-layer and the mineral soil [T ₁₀]) and soil moisture in the O-layer (W₁₀) increased with irrigation (Fig. 4b,d). The increase in T ₁₀ results from the higher temperature of the irrigation water (approximately 15°C) than T ₁₀ in the irrigated plot before the irrigation period (4.5°C on average). The temperature of the raindrops should be in the balance between latent heat by evaporation of raindrops and sensible heat from the atmosphere to raindrops (Ogura 1999). Anderson et al. (1998) verified that measured raindrop temperature and wet-bulb temperature were almost equal. The mean temperature of raindrops in July in Yakutsk is calculated to be 14°C from mean air temperature (18.7°C) and mean relative humidity (60%) in July in Yakutsk (Japan Meteorological Agency 2004). These values indicate that an increase in soil temperature as a result of heavy precipitation would occur in this region. Permafrost creates a strong heat sink in the summer that keeps soil temperature low (Eugster et al. 2000) and therefore soil temperature would increase greatly with heavy precipitation events in this region. The very low amount of precipitation in one event (2.2 mm in maximum) might be the reason why the precipitation during the experimental period did not affect the soil temperature (Fig. 4).

Effects of irrigation on CO₂ flux and CO₂ production rate

The increase in R _mo as a result of irrigation was affected by the soil temperature in our study (Fig. 9j); this result is consistent with a previous study (Boone et al. 1998). In contrast, the normalized R _mo value had only a weak positive effect on soil moisture (Fig. 9l). These results indicate that the increase in R _mo was mainly caused by the increase in soil temperature as a result of irrigation. In general, the microbial respiration rate increases immediately after irrigation in dry conditions, whereas the microbial respiration rate did not vary or decrease with irrigation in wet conditions (Howard and Howard 1993). The weak relationship between soil moisture and normalized R _mo indicated that the O-layer at this site was relatively dry, but not severely dry. Waldrop and Zak (2006) mentioned that there was a trend that soil amended with NO₃ ⁻-N (7 mg kg soil⁻¹) could enhance the soil C decomposition rate 1.5-fold. Comparing the supply of NH₄ ⁺-N and NO₃ ⁻-N to the soil by irrigation (11 and 12 mg N m⁻², respectively) with the NH₄ ⁺-N and NO₃ ⁻-N contents in the O-layer (826 and 9 mg N m⁻², respectively), NH₄ ⁺-N supply could be negligible, whereas the NO₃ ⁻-N supply was equivalent to 1.3-fold the NO₃ ⁻-N content in the O-layer. However, an increase in NO₃ ⁻-N concentrations in the O-layer as a result of irrigation could not exceed 0.3 mg kg soil⁻¹ (Table 1). Therefore, it appears that the NO₃ ⁻-N amendment of soil by the irrigation water barely affected the R _mo .

There was no significant correlation between the soil temperature and R _mm (Fig. 12a). In contrast, both R _mm and normalized R _mm were significantly and positively affected by soil moisture (Fig. 12b,c). W₂₀ in the irrigated plot was significantly higher during the irrigation period than before the irrigation period, and W₂₀ in the non-irrigated plot was significantly lower during the irrigation period than before the irrigation period (both at P < 0.001). This indicates that the change in W₂₀ could be the reason for the lower R _mm during the irrigation period in the non-irrigated plot compared with that in the irrigated plot, even if there were no differences between R _mm in the non-irrigated and irrigated plots before the irrigation period. The NH₄ ⁺-N and NO₃ ⁻-N contents in the mineral soil (9,160 and 82 mg N m⁻², respectively) were much higher than those in the O-layer (826 and 9 mg N m⁻², respectively), so that the effect of inflow of N from the O-layer into the mineral soil as a result of irrigation to R _mm was negligible.

As a result, the positive correlation between soil temperature and R _mw (Fig. 9d) was derived from the positive correlation between soil temperature and R _mo . In contrast, the positive correlation between soil moisture and normalized R _mw (Fig. 9f) was derived mainly from the positive correlation between soil moisture and R _mm . These results indicated that a future increase in heavy precipitation events in summer, resulting in increases in soil temperature and moisture, could accelerate the decomposition of soil C in the O-layer as well as in the mineral soil.

Figure 10 (a,b) CH4 fluxes from the surface of the O-layer and (c,d) the CH4 production rate in the O-layer plotted against soil temperature at a depth of 10 cm (a,c) and moisture at the 0–10 cm depth (b,d). Open and closed circles indicate data from non-irrigated and irrigated plots before irrigation and open and closed triangles indicate data from non-irrigated and irrigated plots during irrigation. ***P < 0.001. Error bars indicate the standard error of each parameter.

Figure 11 (a,b) N2O fluxes from the surface of the O-layer and (c,d) the N2O production rate in the O-layer plotted against soil temperature at a depth of 10 cm (a,c) and moisture at the 0–10 cm depth (b,d). Open and closed circles indicate data from non-irrigated and irrigated plots before irrigation and open and closed triangles indicate data from non-irrigated and irrigated plots during irrigation. †P < 0.1. Error bars indicate the standard error of each parameter.

Figure 12 (a,b) Microbial respiration rate in mineral soil and (c,d) CH4 and (e,f) N2O production rates in mineral soil plotted against soil temperature at a depth of 10 cm (a,c,e) and moisture at the 0–10 cm depth (b,d,f). Open and closed circles indicate data from non-irrigated and irrigated plots before irrigation and open and closed triangles indicate data from non-irrigated and irrigated plots during irrigation. ***P < 0.001. The coefficient values written in italics were calculated using Spearman’s correlation analysis.

Irrigation did not increase the root respiration rate (R _r ) in our study, and a negative correlation between normalized R _r and soil moisture and a positive correlation between R _r and soil temperature were observed (Figs 5c,9g,i). A very few studies reported the effect of irrigation or precipitation accompanied by an increase in soil moisture on R _r . R _r was depressed by surplus water owing to reduced soil oxygen diffusion (Bouma and Bryla 2000). A positive relationship between soil temperature and R _r has been reported in many studies (Atkin et al. 2000; Pregitzer et al. 2000). The increase in the percentage of R _r to R _s with soil temperature in the non-irrigated plot in our study strongly supported a positive correlation between soil temperature and R _r . These results indicate that the negative effect of the increase in soil moisture as a result of irrigation on R _r offset the positive effect of the increase in soil temperature on R _r in this ecosystem. It is suggested that the soil in active layer investigated was too wet not to increase R _r accompanied by an increase in soil temperature because of a shortage of oxygen during the irrigation period.

The R _s value recorded in the present study was positively correlated with soil temperature, which increased with an increase in irrigation (Fig. 9a). A positive relationship between soil temperature and R _s has been reported in many studies (Boone et al. 1998; Davidson et al. 1998; Xu and Qi 2001). In contrast, normalized R _s was not affected by an increase in soil moisture as a result of irrigation (Fig. 9c). We concluded that an increase in soil moisture as a result of irrigation in the present study accelerated R _mw , but depressed R _r . Consequently, the soil moisture appeared to have no influence on R _s . This indicates that the permafrost in this region has the potential to supply an adequate amount of water to plants (Sugimoto et al. 2002), but would not have the potential to maintain a high soil moisture, which would depress decomposition of the soil organic matter.

Figure 13 CO2 fluxes from the surface of the O-layer (Rs) plotted against N2O fluxes from the surface of the O-layer. Open and closed circles indicate data from non-irrigated and irrigated plots before irrigation and open and closed triangles indicate data from non-irrigated and irrigated plots during irrigation. *P < 0.05. Vertical and horizontal bars denote the standard error of the Rs and N2O fluxes, respectively.

Table 4 Standardized partial regression coefficients obtained from a multiple regression analysis to evaluate the contributions of root respiration rate (Rr), microbial respiration rate from the O-layer (Rmo), microbial respiration rate from the mineral soil (Rmm), CH4 production in the O-layer and the mineral soil, and N2O production in the O-layer and the mineral soil to global warming potentials

Standardized partial regression coefficients
R r	0.474
R mo	0.695
R mm	0.212
CH4(o)	0.002
CH4(m)	0.001
N2O(o)	0.005
N2O(m)	0.001
CH4(o), CH4(m), CH4 production rate in the O-layer and mineral soil, respectively; N2O(o), N2O(m), N2O production rate in the O-layer and mineral soil, respectively.

Effects of irrigation on CH₄ flux and CH₄ production rate

The CH₄ flux from the surface of the O-layer in the present study (from −2 to −0 μg C m⁻² h⁻¹) ranged over a higher rate than values reported from forests in northern Europe (from −143 to −1 μg C m⁻² h⁻¹; Smith et al. 2000). Outside the permafrost region in Russia, Nakano et al. (2004) reported a CH₄ flux ranging from −210 to −69 μg C m⁻² h⁻¹ in western Siberia and Morishita et al. (2005) reported a CH₄ flux of −63 μg C m⁻² h⁻¹ near Khabarovsk. These fluxes are much lower than the fluxes recorded in the present study. In contrast, in a permafrost region of Russia, Morishita et al. (2006) reported that the CH₄ flux ranged from −3.4 to −1.6 μg C m⁻² h⁻¹ in central Siberia and Takakai et al. (2008) recorded a range from −10 to 3.4 μg C m⁻² h⁻¹ near Yakutsk; these results are similar to our results. Flessa et al. (2008) mentioned that the lower CH₄ consumption in soils with continuous permafrost can be explained by factors that hamper the diffusion of atmospheric CH₄ into the soil. However, van Huissteden et al. (2008) observed CH₄ fluxes ranging from −301 to −278 μg C m⁻² h⁻¹ in a forest only 1 km from our study site. The reason for this variation is unknown.

The CH₄ production rate in the mineral soil was lower than that in the O-layer in the present study. This result is similar to results reported in previous studies (Dong et al. 1998; Saari et al. 1998).

The CH₄ production rate in the mineral soil did not differ between the irrigated and non-irrigated plots, despite a difference in soil moisture in these plots (Figs 8b,12e). The CH₄ production rate in soil is the net result of simultaneously occurring production and consumption of CH₄ within the soil (Butterbach-Bahl and Papen 2002; Yavitt et al. 1995). This indicates that the potential for both gross CH₄ production and consumption rates in the mineral soil did not change with the increase in soil moisture resulting from irrigation.

In contrast, the increase in the CH₄ production rate in the O-layer as a result of irrigation was affected by the soil moisture in our study (Fig. 10d). In the irrigated plot, the CH₄ production rates in the O-layer were positive during the irrigation period. This indicates that the increase in CH₄ production in the O-layer as a result of irrigation in our study is caused by a change in the balance between gross CH₄ production and consumption rates in the O-layer as a result of an increase in soil moisture. The CH₄ production rates in the O-layer were estimated from two types of fluxes measured with different methods, so these rates include various assumptions. However, there is a possibility that a significantly higher CH₄ flux from the surface of the O-layer in the irrigated plot compared with that in the non-irrigated plot during the irrigation period (Fig. 6a) resulted from a change in the CH₄ production rate in the O-layer, because the CH₄ production rate in the mineral soil was not affected by the irrigation. The effect of the addition of NH₄ ⁺ in the irrigation water should not be significant because the NH₄ ⁺ content in the O-layer was much higher than that added by the irrigation (826 and 11 mg N m⁻², respectively). In general, a negative correlation between soil moisture and CH₄ flux is observed under dry conditions and a positive correlation between them is observed under wet conditions (Curry 2007; Striegl et al. 1992). In our study, soil dryness was not a limiting factor for CH₄ consumption.

Effects of irrigation on N₂O flux and N₂O production rate

The N₂O flux from the surface of the O-layer (from −1.7 to 3.9 μg N m⁻² h⁻¹) ranged over a lower rate than previously reported values in boreal forests (from −4.9 to 22.3; Dalal and Allen 2008). Previous studies conducted near Yakutsk also reported very low values (from −2.1 to 4.6; Morishita et al. 2007; Takakai et al. 2008). These low values show characteristics of an N-limited region (Hatano et al. 2001; Schulze et al. 1995). The much larger N₂O production in the O-layer than in the mineral soil observed in the present study was consistent with previous studies (Saari et al. 1997).

A number of studies have reported a positive correlation between soil moisture and temperature and N₂O flux from the surface of the O-layer in natural forests (e.g. Dong et al. 1998; Fest et al. 2009). Although the N₂O flux during the irrigation period in our study was higher in the irrigated plot than in the non-irrigated plot, the N₂O fluxes did not correlate with either soil moisture or temperature (Figs 11,12e,f). Only a weak positive correlation between the N₂O flux from the surface of the O-layer and soil temperature was found (P = 0.08). The N₂O production rates in the O-layer and the mineral soil did not differ between the irrigated and non-irrigated plots both before and during the irrigation period and did not correlate significantly with either soil moisture or soil temperature. In contrast, the N₂O flux from the surface of the O-layer was positively correlated with R _mw (P < 0.05; Fig. 13). This indicates that the main cause of an increase in the N₂O flux from the surface of the O-layer as a result of irrigation was not a direct effect of either soil temperature or moisture, but rather an indirect effect of microbial activities. Comparing NH₄ ⁺-N and NO₃ ⁻-N supply to the soil by irrigation (11 and 12 mg N m⁻², respectively) with the NH₄ ⁺-N and NO₃ ⁻-N contents in the O-layer, which produced most of the N₂O for the whole soil (826 and 9 mg N m⁻², respectively), NH₄ ⁺-N supply was negligible, whereas NO₃ ⁻-N supply was comparable with the NO₃ ⁻-N content in the O-layer. This indicates that NO₃ ⁻-N supply by irrigation might stimulate N₂O emission via an increase in denitrification in the irrigated plot. Liu et al. (2008) could not detect variation in the N₂O flux from the surface of the O-layer after irrigation owing to the lack of substrate for denitrification process on grassland in Inner Mongolia. The weak relationship between N₂O flux from the surface of the O-layer and both soil temperature and moisture in the present study might be caused by a deficiency of N. Du et al. (2006) also observed no significant linear relationship between both soil temperature and moisture and diurnal N₂O flux from the surface of the O-layer in grassland in Inner Mongolia. However, these researchers mentioned that it was the distribution of effective precipitation, rather than the precipitation intensity, which influenced seasonal and inter-annual variations in N₂O flux. Even in the ecosystem investigated in the present study, observations over a longer period are needed to examine the effects of precipitation on N₂O flux from the surface of the O-layer.

Conclusions

In the investigated Taiga forest near Yakutsk, which is characterized by low soil temperature and dry weather, heavy precipitation events would increase both soil temperature and moisture. The increased soil temperature and moisture would accelerate microbial activities, which play an important role in controlling the GWP in the soil, resulting in decomposition of soil organic matter in both the O-layer and the mineral soil. As a result, the release of inorganic N into the soil together with an increase in N₂O emission, which has a huge impact on the GWP of the soil, would be expected. Continuous study on the effect of changes in precipitation patterns on the decomposition of soil organic matter accompanied by N mineralization is crucial in predicting the future GWP of soil in this region. In addition, an increase in the gross CH₄ production rate might decrease the potential of the soil to act as a CH₄ sink. Further study is required to investigate changes in CH₄ production in the O-layer caused by heavy rain events.

Acknowledgments

We are deeply grateful to the staff members of the Institute for Biological Problems of Cryolithozone, Siberian Branch, Russian Academy of Science, for their support during the field research. This study was funded by the RR2002 Program of the Japanese Ministry of Education, Culture, Sports, Science and Technology.