The Blocking Effect is the Key Factor for the Sharp Descending of Soil Respiration Between Late Autumn and Early Winter in a Larch Forest

Throughout the year 2013 (from April to November), we measured the CO2 efflux and investigated the soil physical properties weekly to characterize seasonal patterns of soil respiration and to examine blocking effect on winter soil respiration using Open Dynamic Chamber (ODC) method. We first affirmed that the local morning 10:00 o’clock with the smallest sampling errors of 0.5 to 2% was the optimal measurement point. The diurnal variation of soil respiration displayed as a single-crest curve which ascended with the increasing soil temperature in morning to a crest in early afternoon (about 13:30-15:30 o’clock), then descended. There was a significant difference in soil respiration among seasons. Hence, two separate models were used to describe the temperature sensitivity of Rs (Q10) in growing season and non-growing season better. The fitted Q10 value of Rs was 2.1 in time scale of a whole year. The obvious fluctuation of Q10 values between dormancy season (90.0) and growing season (1.6) indicated a high sensitivity of soil respiration to soil temperature in early winter. The maintenance respiration of roots in the dormant period was calculated as 0.11 μmol CO2/m/sec. The annual mean Rs and Ra were 2.53 and 1.12 μmol CO2/m/sec, separately. The proportion of Ra in Rs had a relative smaller range of 14 to 55% in different seasons, higher in summer and lower in winter. The soil respiration decreased sharply in the early winter, especially after the melt of the first snow. In this study, we first defined the blocking effect of a thin ice layer on soil respiration and considered the transformation of soil water from liquid to solid as the reason of soil respiration descending in early winter. This theory may raise more concerns on the physical blocking effect on soil respiration in future researches.


INTRODUCTION
Soil Respiration (R s ) is an ecosystem process that releases CO 2 from soil via root Respiration (R a ), microbial decomposition of litter and soil organic matter (R h ) and fauna respiration.There are two keys for the extraordinary active research on soil respiration in the past decades.One is that it is among the least understood subjects in ecosystem ecology and the other is that it represents the second largest flux of carbon cycling between the atmosphere and terrestrial ecosystems (Luo, 2007).
Soil respiration displays strong temporal variation over time.Because it is the environment factor that changes strongly on a diurnal scale, a close function of soil temperature can explain the diurnal variation in soil respiration in most cases (Bajracharya et al., 2000;Xu and Qi, 2001;Liu et al., 2006).The diurnal variation must be accounted appropriately to avoid errors when point measurements of soil respiration are used to estimate annual soil carbon efflux.Davidson et al. (1998) pointed out that the midmorning efflux closely approximate the 24-h mean efflux.There is no certain time point that is given to determine the diurnal mean of soil respiration.Thus, assuring an optimal sampling point not only can estimate the annual CO 2 efflux more precisely, but also can save the sampling time and labor.
Seasonal variation in soil CO 2 efflux has been observed in almost all ecosystems.On a global scale, soil CO 2 efflux peaks in summer when plants grow fastest in both temperate zones and near-equatorial regions (Raich and Potter, 1995;Raich et al., 2002).The seasonal variation is largely driven by changes in temperature, moisture, photosynthate production and/or their combinations (Raich and Schlesinger, 1992;Raich et al., 2002;Gaumont-Guay et al., 2006a).Wang et al. (2013) reported that in a temperate forest the empirical R s -T s -SWC models explained 65.3 to 94.1% of the variability in the R s data, but the number of the regression terms and their coefficients varied with seasons.It indicates that the models should be fitted to the seasonal data sets separately to explicitly describe the seasonality of R s .
While most of the carbon influx to terrestrial ecosystems can be attributed to photosynthesis only during the spring and summer months, soil respiration occurs throughout a whole year.For this reason, when estimating annual carbon budgets, wintertime soil respiration can be quantitatively considerable.For example, more than half of the carbon sequestered by trees during the growing season can be consumed by winter respiration in forests (Sommerfeld et al., 1993;Winston et al., 1997;Monson et al., 2002).However, winter respiration is highly variable and may be regulated by fluctuations in environmental variables such as soil temperature and moisture (Brooks et al., 1997;Mikan et al., 2002;Schimel and Mikan, 2005;Liptzin et al., 2009).Du et al. (2013) proposed a "freeze-thaw critical point" hypothesis, which stated that the Q 10 value above freeze-thaw critical point was much higher than that below it (16.0 vs. 3.5) and that was probably regulated by the steep change in soil free water availability during the soil-freezing process.They ascribed the sharp decreasing of soil respiration to the shortage of free water caused by subzero soil temperature.We know that except in excessively wet or compacted conditions, CO 2 movement in soil occurs through a continuous network of air-filled pores that connect the surface to the deeper layers of the soil (Hillel, 1998).However, usually in winter, a thin ice layer covers the surface of frozen soil particles and reduces the porosity in soil.This ice layer serves as a diffusion barrier which reduces oxygen supply to the microbes and partly prevents the effusion of CO 2 (Teepe et al., 2001;Koponen et al., 2006).We believed that the subzero soil temperature in early winter was not the exclusive factor conducing the sharp decreasing of soil respiration in early winter.A more reasonable possibility is that the soil frozen which is induced by low soil temperature changes the soil physical properties totally.Most water in soil exists in a solid form of ice, fills most former air-filled pores and blocks the gas diffusion from soil to atmosphere.In this condition, we called this condition "blocking effect" and believed that it was a significant reason for the steep shift of soil respiration in early winter.
Though the diversity of sources of CO 2 efflux, accurate partitioning of observed soil respiration to various source components is a critical step towards mechanistic understanding of soil respiration itself and its responses to environmental changes.Usually root respiration accounts for approximately half of the total soil respiration but varies from 10 to 90% among different studies (Hanson et al., 2000).Root respiration is often conceptually separated into growth respiration and maintenance respiration.Growth respiration produces the energy and building blocks (i.e., metabolic intermediates) for the biosynthesis of structural compounds and maintenance respiration yields the energy used by the normal activities of viable cells.The start and stop of fine root growth are closely related to soil temperature (Barber et al., 1988).When the soil temperature drops to 4°C, the fine roots stop growing (Cheng et al., 2006).In this study, we defined this soil temperature as the "R a dormancy critical point" under which the root start to be dormant with a maintenance respiration, then we estimated the root dormancy respiration.The objectives of this study were to:  Determine an optimal measurement time point and characterize seasonal patterns of soil respiration. Establish separate models to describe the sensitivity of R s better in growing season and nongrowing season. Analyze blocking effect on winter soil respiration and argue its rationality.

Site description:
The study area is locate in the Saihanba National Forest Park (SNFP) of Weichang Manchu and Mongolian Autonomous County.More details about our sample plots can be found in Wang et al. (2014).
In this study, we selected a 16-year-old (in 2013) larch forest (Larix principis-rupprechtii) with ast and density of 3415±128 trees/ha.The average Diameter at Breast Height (DBH) and tree height are 8.65±1.63cm and 8.56±1.15m, respectively.L. principisrupprechtiiis the dominant species whose vertical structure includes the tree layer (>2 m); the shrub layer (0.3-2 m), with composition including Malus baccata, Rosa davurica, Crataegi cuneatae, Rhododendron micranthumand and Potentilla fruticosa; and the floor layer (<0.3 m), consisting of Maianthemum bifolium, Saussurea japonica and Thalictrum aquilegifolium.The gray forest soil with pH values of 6.32 to 6.71 is deeper than 1.5 m.Soil parent materials are eluvium, saprolite and alluvium.
Field measurement: Soil respiration, soil temperature and volumetric water content at 5 cm depth were measured on 3 trenched matrixes (1×1 m) and 3 accompanying untrenched ones.A PVC soil collar, with a height of 5 cm and a diameter of 20 cm, was permanently inserted in the soil at each sampling location with the help of a piece of wood and a hammer.The collar extended 3 cm above the soil surface and inserted 2 cm in soil.R h was measured on the trenched locations and R s was measured on the untrenched ones and R a was calculated by difference of R s and R h .Soil respiration was measured with a Li-8100A-103 survey chamber connected to an analyzer control unit of the Li-8100A Automated Soil CO 2 Flux System (Li-Cor Inc., Lincoln, NE, USA) for data collection and storage.In this study, from the end of April to early November, to determine a suitable measuring point in day, we take measurements every half hour in a selected fine day weekly and compare the diurnal mean value with the point measurements.Simultaneously, soil temperature and moisture at 5 cm depth were measured nearby each collar using an auxiliary soil temperature probe (Omega Engineering Inc., USA) and a Theta probe (Delta-T Devices, Cambridge, UK), respectively.The soil bulk density, capillary porosity and field capacity were determined on the same day.
Data analysis and modeling: R s and R a in growing season and non-growing season: In 2013, L. principis-rupprechtii started to bud in early May and to wither at the end of September.We regarded June to September as growing season and April, May, October and November as non-growing season.The significances of R s and R a between growing season and non-growing season were verified by using the paired t-test.
Effects of soil temperature and soil moisture: After the correlation analysis, R s and T s were correlated significantly (p<0.01),but R s and SWC were not significant correlated (p = 0.212).The correlations of R a and T s , R a and SWC were both significant (p = 0, p = 0).Soil respiration was fitted to model ( 1 (3) where, W (volumetric water content, VWC %) was the soil moisture at 5 cm soil layer and A 0 , k and q were parameters.
The variations of soil bulk density, capillary porosity and field capacity in different seasons also were verified by using the paired t-test.All statistics were calculated using PASW Statistics 18 (IBM, Armonk, NY, USA) with a level of significance of p<0.05.All figures were made using SigmaPlot 12 (Systat Software Inc., San Jose, CA, US).

Diurnal variations of R s and the optimal sampling point:
The diurnal variation of soil respiration in growing season was similar to that in non-growing season.In 24 h, the rate of R s had the minimum around 04:30 o'clock (local time) and it followed the increasing trend of soil temperature to a peak around early afternoon (about 13:30-15:30 o'clock), but then leveled off with slight fluctuations while soil temperature kept increasing in the afternoon.From evening to early morning of the next day, the rate of R s followed the declining trend of soil temperature with little fluctuations (Fig. 1).The diurnal range was normally less than 1 µmol CO 2 /m 2 /sec, or about 15 to 40% of its diurnal mean.The optimal measurement point confirmed by comparing the daily mean of R s with point measurements was local morning 10:00 o'clock, at which the measurements had the smallest sampling errors of 0.5 to 2% (Fig. 1).The entire daytime measurements in growing season tended to overestimate the daily mean rates by 2 to 7% and in non-growing season usually equaled to the daily mean.
Seasonal variations of R s and R a and maintenance respiration of root: The annual mean of R s (from April to November) was 2.53 µmol CO 2 /m 2 /sec with a mean of 3.97 µmol CO 2 /m 2 /sec in growing season and 0.98 µmol CO 2 /m 2 /sec in non-growing season.The seasonal trend of R s ascended from April.The values of R s had a very significant difference (p<0.001) between growing season and non-growing season.R s was 1.19 µmol CO 2 /m 2 /sec in April and the peak value in later July was 5.03 µmol CO 2 /m 2 /sec.Then it descended gradually to a minimum of 0.6 µmol CO 2 /m 2 /sec (Fig. 2).R a also had great difference between growing season and non-growing season (p<0.01).R a had a similar monthly pattern with that of R s .The annual R a averaged 1.12 µmol CO 2 /m 2 /sec.It was 0.21 µmol CO 2 /m 2 /sec and 1.51 µmol CO 2 /m 2 /sec in non-growing and growing season, separately.The obtained root dormancy respiration was 0.11 µmol CO 2 /m 2 /sec, 9.8% of the annual R a .
The proportions of R a in R s had a very significant variation in different seasons (p<0.01).The R a /R s in root dormancy period when the soil temperature dropped to less than 4°C had a minimum of 14%, while in summer reached to a maximum of 55%, averaged 32% annually (Fig. 2).

The sensitivity of R s and R a in different seasons:
Combining the data in growing season and nongrowing season, three exponential models based on soil temperature accounted for 40 to 70% of the variation in CO 2 fluxes (Fig. 3).The fitted Q 10 values of R s were 90, 1.62 and 2.12 in dormancy season, growing season and a whole year, separately.The fundamental soil respiration was 0.57 µmol CO 2 /m 2 /sec when the soil temperature was 0°C.
In root dormancy period, the root respiration was highly sensitive to soil temperature and had a very great Q 10 value of 602.The Q 10 were 2.6 and 3 in growing season and a whole year separately.The ratio of Q 10 in non-growing season and in growing season was a prodigious value of 232 (Fig. 4).The fundamental root respiration was 0.11 µmol CO 2 /m 2 /sec.

The steep decreasing of T s and SWC in early winter:
In early winter when the soil temperature dropped sharply to 0°C, The R s decreased to a minimum of 0.48 µmol CO 2 /m 2 /sec and the free water in soil declined from 30 to 6% (Fig. 5).The previous air-filled pores The annual variations of soil physical properties: At the end of April, the soil moisture was great (about 40%) for the melting snow.Then it kept a comparative wet condition (always more than 20%) in the following rainy season until the autumn when a drought occurred.Two snow event occurred in early winter (10 and 27 October) when the soil temperature was still above 2°C.
The melting snow water provided greater soil water content (30%).With the soil temperature decreased to freezing point and free soil water frozen, the soil moisture dropped gradually to less than 8% (Fig. 6).All the seasonal variations of field moisture capacity, capillary porosity and bulk density were very significant (p<0.01)(Fig. 7).Field moisture capacity had a similar trend with soil water content (Fig. 6 and 7).The capillary porosity in non-growing season was about 15% and in growing season around 30%.In the early winter (the end of October), the soil had a greater water content (34.8%, both in the form of liquid water and solid ice), relatively lower capillary porosity (11.2%) and bulk density (1.83).

DISCUSSION
The temporal variations of R s and R a : In some forests where the leaf area index is great, the soil temperature varied gently for a heavy shade, thus the diurnal variation for soil respiration may not be very apparent (Davidson et al., 2000;Betson et al., 2007).Our studying forest which had a relative great canopy density of 0.8 and a thick litter layer of almost 10 cm, the soil temperature in upper soil layer had a small daily variation of 2 to 3°C in growing season, thus the temperature-dependent soil respiration fluctuated slightly over a day.
We confirmed a comparative precise measuring point by comparing the daily mean values of R s with the measurements taken at 10:00 o'clock (local time) and found that both of them were almost equal.Xu and Qi (2001) also found that the measurements taken between 09:00 and 11:00 o'clock can better represent the daily mean soil respiration than the entire daytime measurements which tend to overestimate the daily mean rates by 4 to 6%.This finding could lightening the working intensity of soil respiration measurement to a very great extent, especially when there are a lot of positions to be measured in one study, on the condition of measurement accuracy guaranteed.If the daily means were estimated from the measurements that were taken in the warmest period of the day, usually around 14:00 o'clock, the estimation of daily and monthly rates of soil respiration both can be substantially biased.Luo et al. (2001) and Luo (2007) found high soil respiration during summer and low respiration in winter, a distinct seasonal pattern.In forest (Epron et al., 1999;Savage and Davidson, 2001;Epron et al., 2004) and cropland (Beyer, 1991) ecosystems, the seasonal patterns of soil respiration also exist.For example, King et al. (2004) indicated that over one year, soil respiration was nearly 0 µmol CO 2 /m 2 /sec in the winter, was about 10 µmol CO 2 /m 2 /sec in the summer in the Duke Forest, North Carolina.From 1997 to 2002, the seasonal pattern repeats, but different peaks in summer and roughs in winter appear to inter-annual variation.Our results of R s rates which varied from 0.43 µmol CO 2 /m 2 /sec in November to 6.5 µmol CO 2 /m 2 /sec in August 2013, had a significant seasonal variation that with a maximum in August which lagged behind the soil temperature peak and lower values in spring and early winter when soil temperature was relative lower.In summer, both of the soil temperature and moisture were great and good for microbe activities and litter decomposition.Thus soil respiration in summer had a higher rate.In early winter the soil temperature were too low to support the proper activities of microbe.The mineralization of organic carbon retarded to a very slow level.The related soil respiration displayed a decreasing trend.
The Q 10 of R s and R a and the dormancy critical point: Estimated at different geographic locations of various ecosystems, the estimated Q 10 values for soil respiration vary widely from little more than 1 (low sensitivity) to more than 10 (high sensitivity) (Wildung et al., 1975;Bouma et al., 1997;Davidson et al., 2000;Reichstein et al., 2003;Gaumont-Guay et al., 2006a).Based on data compiled nearly 20 years ago, Raich and Schlesinger (1992) calculated that a 2.4 for the global median Q 10 value, with a range of 1.3 to 3.3.Lloyd and Taylor (1994) suggested that the effective mass of carbon per unit area leads to variation in Q 10 values in different studies.In European and North American forest ecosystems, Q 10 values range from 2.0 to 6.3 (Davidson et al., 1998;Janssens et al., 2001;Gaumont-Guay et al., 2006b).The corrected respiration from our study followed a similar temperature-respiration response function.The annual mean Q 10 of R s in our studying plots was 2.1, which was in the middle of the Q 10 range of 1.3-3.3 reported by Raich and Schlesinger (1992), smaller than the global median Q 10 value of 2.4.It suggested that the soil respiration was more sensitive to the ascending temperature, depending on the special soil texture and environment characteristics in our studying plots.The fitted Q 10 value of R s was 90 in dormancy season.This greater Q 10 showed that the soil respiration in dormant period was very sensitive to the soil temperature elevating.
At the end of September, the soil temperature dropped to less than 4°C under which the root started to dormant and kept a maintenance respiration (Fig. 6) and most leaves withered and fell to earth.The ratio of R a to R s decreased from 34% at the end of September to 24% in dormant period, almost 30%.Our data suggested that there were different temperature control mechanisms for root respiration at temperatures above and below the R a dormancy critical point.It indicated that root respiration could not be effectively defined by a single temperature equation to accurately predict annual C budget in the cold regions.The Q 10 value (602) of R a in dormant period was likely to overestimate the Q 10 (2.6) above 4°C and underestimate the Q 10 (16.0) in a whole year.The great differences among the Q 10 values at different soil temperatures indicated an obvious existence of the "R a dormancy critical point".
The blocking effect on soil respiration: The results of Koponen et al. (2006) study showed that the respiration rate of completely frozen soils never reached zero and freezing and thawing of boreal soils did not have a strong effect on microbial biomass or community structure.This likely supports a viewpoint that soil respiration is not affected by the blocking effect of frozen soil.However, Teepe et al. (2001) detected two higher N 2 O emissions, respectively in continuous soil freezing and thawing period.They found that microorganisms during continuous soil freezing in an unfrozen water film on the soil matrix can produce gases.Nevertheless, these gases were prevented to release by a layer of frozen water in form of ice which represents a diffusion barrier.Peak emissions during soil thawing were explained by the physical release of trapped gases.Du et al. (2013) also pointed out that soil respiration might be inhibited when the snow depth increases to a critical depth which blocks gas diffusion.The soil was wet after the first snow melt in the early winter.On the one hand, when the soil temperature dropped to ice point, the soil top layer was covered by a thin ice layer which stopped the gas exchange between soil and the atmosphere.On the other hand, the original air-filled pores were filled with ices.Only few gases were allowed to pass through.In our study, the sharp drops of SWC, capillary porosity and bulk density in the early winter indicated the forming of thin ice lay and the increasing blocking effect on the CO 2 release.201004021) very much.Thank our colleagues Jie Liu and WeibinYou, for their enthusiastic helps.At-last, we sincerely thank Chunli Biwho reviewed and revised this manuscript carefully.
): R s = A 0 e kT (1)where, T was soil temperature (°C) at 5 cm soil layer, A 0 and k were parameters.The Q 10 value, a quotient of change in respiration caused by change in temperature by 10°C, was calculated as in Eq. (2):Q 10 = e 10k(2)Considering the integrated effects of SWC and soil temperature, model (3) was used: R a = A 0 e (k+qW) T

Fig. 1 :
Fig. 1: Diurnal variations of R s and T s in the DOY of 237, 252, 293, and 307, respectively.The intersections mean the optimal sampling time points

Fig. 2 : 72 Fig. 3 :
Fig. 2: Seasonal variations of R s (triangle up), R a (square), and the ratio of R a to R s (circle); bars are means±S.E., n = 72