Biogeosciences Day and night warming have different effect on root lifespan

Roots are key components of C cycling in terrestrial ecosystems and play an important role in the regulation of response of terrestrial ecosystems to global climate warming, which is predicted to occur with greater warming magnitudes at night than during daytime across different regions on the Earth. However, there has been no detailed study to investigate the effect of asymmetrical warming on root dynamics at the level of terrestrial ecosystems. To understand the effects of day and night warming on root lifespan in the semiarid temperate steppe in northern China, a field study with a full factorial design including control, day warming, night warming and continuous warming was conducted using modified rhizotron technique during three growing seasons in 2007–2009. Our results show that day, night and continuous warming had different effects on longevity of roots born in spring, summer and autumn, and that day warming significantly prolonged overall lifespan for the roots born in the three growing seasons, while night warning had no effect on overall lifespan. Day and night warming had different effects on root non-structural carbohydrate content, suggesting that allocation of photoassimilate may account for the differential responses of root lifespan to day and night warming. These results differ from other processes associated with ecosystems C cycle such as total ecosystem productivity, net ecosystem productivity and soil respiration. Our findings highlight that it is essential to incorporate the differential effects of day and night warming on root dynamics into simulating and predicting the responses and feedbacks of terrestrial ecosystems C cycling to global warming.


Introduction
Model projection and field experimental observations have demonstrated that global warming can have significant impact on the structure and function of terrestrial ecosystems in general, and carbon (C) cycle of terrestrial ecosystems in particular (Ciais et al., 2005;Luo, 2007;Piao et al., 2008;Wan et al., 2009;Bai et al., 2010).Climate warming is expected to result in unequal changes in daily minimum and maximum temperatures, with the greater trends of night warming than day warming (Eastering et al., 1997;Alward et al., 1999;Zhou et al., 2007).It has been reported that increased daily minimum and maximum temperatures have different effects on biomass and yield (Dhakhwa et al., 1998;Alward et al., 1999;Peng et al., 2004;Lobell, 2007;Lobell and Ortiz-Monasterio, 2007) as well as C cycling (Wan et al., 2009;Xia et al., 2009) in terrestrial ecosystems.However, it remains unclear whether asymmetrical warming may have different effects on root growth, mortality and turnover, which plays critical roles in regulating nutrient and C cycling between above-and belowground.
Photosynthesis takes place predominantly during day time, while respiration becomes dominant at night time.Day warming would stimulate photosynthesis, while night warming would stimulate respiration.Therefore, day and night warming may have different effects on C absorption and release.Recent studies revealed that day and night warming have different effects on total ecosystem productivity, net ecosystem productivity and soil respiration in semi-arid temperate steppe of northern China (Xia et al., 2009;Wan et al., 2009).The differential response of semi-arid steppe to day and night warning may result from stimulation of plant respiration by night warming, thus altering C allocation due to compensatory response of leaf photosynthesis (Wan et Published by Copernicus Publications on behalf of the European Geosciences Union.al., 2009).Numerous studies have reported that terrestrial ecosystems are highly sensitive to night warming (Alward et al., 1999;Peng et al., 2004;Volder et al., 2004;Lobell, 2007;Beier et al., 2008).Root growth is closely dependent on soil temperature (Pregitzer et al., 2000).Our previous studies showed that root dynamics were also sensitive to continuous warming in the temperate steppe (Bai et al., 2010).Given that root growth, mortality and turnover are closely associated with plant photosynthesis, respiration and belowground C allocation, and that asymmetrically continuous warming has differential impacts on plant photosynthesis and respiration, we hypothesize that root dynamics exhibit differential response to day and night warming.
Roots are an important source of soil organic C and plays a key role in the regulation of ecosystem C cycle (Loya et al., 2004;Sullivan et al., 2007).In addition, roots, as key component of ecosystems C cycling, are also closely related to processes associated with ecosystem C cycling (Farrar and Jones, 2000;Norby and Jackson, 2000).Our recent studies demonstrated that increases in root production and reduced mortality in response to increased temperature and precipitation were significantly correlated with total ecosystem productivity, net ecosystem productivity and soil respiration in the typical grassland in Inner Mongolia (Bai et al., 2010).These findings indicate that root dynamics can have important impacts on ecosystems C cycle under climate change.The responses of root dynamics to warming and precipitation differ from other ecosystem processes such as total ecosystem productivity, net ecosystem productivity and soil respiration.For instance, an increase in temperature and precipitation has an additive effect on gross ecosystem productivity, net ecosystem productivity, ecosystem respiration and soil respiration in Inner Mongolia typical grassland (Niu et al., 2008;Liu et al., 2009).In contrast, there were significant interactive effects of increased temperature and precipitation on cumulative root production and mortality (Bai et al., 2010).These findings indicate that different C processes in ecosystems may vary in response to climate change.Therefore, understanding whether response of root dynamics to asymmetric warming differs from other ecosystem C cycling processes would shed important light on C cycleclimate change feedback.
Root lifespan is an important parameter that determines acquisition of water and nutrients, and affects soil organic C pool.However, to the best our knowledge, there has been no report to evaluate the response of root lifespan to day and night warming.In the present study, we investigated the effects of day and night warming on root lifespan in semiarid temperate steppe in northern China.More specifically, we addressed the following two questions: (1) Does root lifespan have a different response to day and night warming?(2) Is the response of root lifespan to day and night warming comparable to other C cycle process in ecosystems such as gross ecosystem productivity, net ecosystem productivity, ecosystem respiration and soil respiration?
2 Materials and methods

Study site
The study was conducted in a semiarid temperate steppe in Duolun County (42 • 02 N, 116 • 17 E, 1324 m above sea level) in Inner Mongolia, China.Long-term (1953Long-term ( -2007) ) mean annual rainfall is approximately 383 mm with 90 % of the precipitation falling from May to October.Mean annual temperature is 2.18 • C, with the minimum and maximum temperatures ranging from −17.58 • C in January to 18.98 • C in July.The sandy soil of the study site is classified as chestnut according to the Chinese classification, or Haplic Calcisols according to the FAO classification.Mean soil bulk density is 1.31 g cm −3 and pH is 7.7.Soil organic C and total N contents are 16.1 g kg −1 and 1.48 g kg −1 , respectively.The plant community at our experimental site is dominated by Stipa krylovii and Artemisia frigida.

Experimental design
We used a complete random block design with 4 treatments, including control (C), day warming (06:00 a.m.-06:00 p.m.; D), night warming (06:00 p.m.-06:00 a.m.; N), continuous (24 h; W) warming and replicated each treatment 6 times (Wan et al., 2009).Each plot was 3×4 m in size with a 3m distance between any two adjacent plots.The warming in the warmed plots was achieved by continuously heating with MSR-2420 infrared radiators (Kalglo Electronics Inc, Bethlehem, PA, USA) suspended 2.25 m above the ground.To simulate the shading effects of the infrared radiator, one "dummy" heater with the same shape and size as the infrared heater was suspended 2.25 m above ground in each control plot.All the heaters under the warming treatments were set at a radiation output of about 1600 watts.The warming treatment commenced on 23 April 2006.The detailed description of warming treatment can be found in previous publications (Wan et al., 2009;Xia et al., 2009).

Measurement of root dynamics
Root production and mortality were measured by protocols described previously (Bai et al., 2008(Bai et al., , 2010)).Briefly, on 30 August 2006, one glass root-window was installed in each plot.The glass window (0.4-cm thick) of 35 cm in height and 20 cm in length was installed vertically into the soil.On each glass window, a 35×20 cm panel (with 5-cm distance to the bottom and 5 cm to the right and left sides) was separated into three 10×10 cm squares by carving the glass.In order to minimize the impacts of light on root growth, the upper edge of the glass window was installed under the soil surface and a piece of dark iron (20 cm in length, 1.5 cm in breadth and 0.5 mm in thickness) was covered on the top of the glass.A hole was dug in each plot with a vertical profile for the glass window to be installed.The glass window was put tightly to the trench and fixed with one iron stick at each Biogeosciences, 9, 375-384, 2012 www.biogeosciences.net/9/375/2012/side.After the root-windows were inserted, soil was backfilled as tightly as originally.The soil was closely attached to the glass throughout the whole study period.
A digital camera was used to monitor root growth.Observations began on 30 June 2007, which was about 10 months after the installation of glass windows, and lasted until 19 October 2009 with sampling intervals of 15 d.On each sampling date, the soil on one side of the glass window was removed and the screen was cleaned with tissue paper.One digital picture was taken for each of the three 10×10 cm numbered squares.Removed soil was carefully backfilled again after pictures were taken as described previously (Bai et al., 2008(Bai et al., , 2010)).
Appearance and disappearance of roots were analyzed by the software of Mapinfo Professional (5.0; Pitney Bowes Mapinfo Corporation, New York, US) as described by Bai et al. (2008Bai et al. ( , 2010)).For the initially collected images, each root was assigned with an identification number and distinguished it as living or dead based on its colors as described in our previous paper (Bai et al., 2008).For the following image sets, the tracings from the previous date were compared with the new images, thus allowing previously existing roots to be identified.Newly emerged roots were also identified and numbered.Roots that had disappeared at subsequent images were assumed to be dead and decomposed.Complete records were kept for all roots, even for those that were classified as dead.

Measurements of soil temperature and soil moisture
Soil temperatures at the depth of 10 cm were recorded automatically with a Datalogger (STM-01 Soil Temperature Measurement System, Henan Electronic Institute, Zhengzhou, China).Temperature measurements were taken every 10 min and the average of the six measurements within 1 h was stored as the hourly means.Soil moisture (0-10 cm) was measured weekly using a Diviner-2000 Portable Soil Moisture Probe (Sentek, Balmain, Australia).

Measurements of soluble sugars, starch and ecosystem parameters
Root samples that were obtained from the root in-growth measurement in 0-30 cm soils by the end of growing seasons in late October were used to determine soluble sugars and starch following protocols described by Wan et al. (2009).Soluble sugar and starch concentrations in roots were determined as described by Ranwala and Miller (2008).Root nonstructural carbohydrate was taken as the sum of root soluble sugars and starch (Pregitzer et al., 2000).Aboveground net primary productivity (ANPP) was determined during the peak at the end of August in each year.Belowground primary productivity (BNPP) in 0-30 cm soils was measured by the end of growing seasons in late October in each year.Gross ecosystem productivity (GEP), net ecosystem productivity (NEP), ecosystem respiration (ER), soil respiration (SR), and soil inorganic N were measured as described previously (Wan et al., 2009;Xia et al., 2009).

Data analysis and statistics
Root lifespan was calculated as the date on which roots were observed as black or disappeared minus the date on which roots were initially observed on the window.The date of root appearance or disappearance was estimated as the date midway between the sampling periods because they might have occurred on any day during the approx.15-d sampling interval between two consecutive observations (Hooker et al., 2000;López et al., 2001;Anderson et al., 2003).
We selected total of 10 887 new roots germinated in spring (1-15 and 15-30 May), summer (1-16 and 16-31 July) and autumn (31 August-15 September, 15 September-1 October) of 2008 and 3207 new roots germinated in spring (18 April-2 May, 2-17 May), summer (19 June-3 July, 3-19 July) and autumn (19 August-4 September, 4 September-19 October) of 2009 to analyze the effects of different treatments on their survival rates and longevity.To analyze the effects of day warming and night warming on their survival rates and longevity, we selected a total of 20 007 new roots germinated in control, day warming, night warming, and continuous warming plots during the growing seasons of 2007-2009.We calculated their mean longevity through a survival curve using the Kaplan-Meier method with SPSS (12.0) software, and compared the root survival rates by Log-rank test.
A stratified Cox proportional hazards regression was used to estimate the effects of treatment on root lifespan as described by Gill et al. (2002).In the Cox model, the hazard for an individual root at time t is computed based upon the combination of a non-parametric baseline (h0) function and an exponential function of k covariates: The Cox model was used to determine β coefficient for each model covariate and test the null hypothesis of β = 0 with a chi-square statistic.A negative β coefficient indicates a decreased risk of mortality with an increase in the covariate values.A risk ratio (expβ) was used to calculate the percent change in the risk of mortality with a one unit change in the covariate was calculated as well (Wells and Eissenstat, 2001;Gill et al., 2002).
A full factorial design with day and night warming was used in this study.The three-way ANOVAs were used to examine effects of year, day warming, night warming, and their possible interactions on soil temperature and soil moisture.Two-way ANOVAs were used to examine effects of day and night warming on root longevity, root nonstructural carbohydrate content (RNC), ANPP, and BNPP.Regression with correction for autocorrelation and stepwise multiple linear regression analyses were used to examine the relationships between root longevity and soil temperature, soil moisture, www.biogeosciences.net/9/375/2012/Biogeosciences, 9, 375-384, 2012

Soil temperature and moisture
Soil temperature at 10-cm depth varied among the four growing seasons (2006)(2007)(2008)(2009).The experimental design was a 2-factorial one with day and nigh warming as main factors.
Both day (P <0.0001) and night (P <0.0001) warming significantly elevated daily mean soil temperature across the four growing seasons (Table 1; Fig. 1a).Day and night warming increased daily mean soil temperature by 0.38 and 0.50 • C, respectively.No interactive effect between day and night warming (P = 0.467) on soil temperature was observed.

Effect of day warming and night warming on root longevity
Roots initiated in the three growing seasons (spring, summer and autumn) of 2008 and 2009 in the control and warming plots exhibited different survival rates and longevities, and responded to day, night and continuous warming differently (Fig. 2; Table 2).Log-rank test revealed that day warming significantly increased survival rate (Fig. 2a, d) and longevity for the roots initiated in spring (P <0.05;Table 2); while both night and continuous warming had no effect on survival rate (Fig. 2a, d) and longevity for the roots initiated in spring of 2008 and 2009 (P >0.05;Table 2).In contrast, survival rate (Fig. 2b) and longevity (P <0.001;Table 2) of the roots initiated in summer of 2008 were significantly enhanced by continuous warming, while survival rate (Fig. 2e) and longevity (P =0.001;Table 2) of the roots initiated in summer of 2009 were significantly enhanced by day warming.Moreover night warming had no impacts on survival rate and longevity of the roots initiated in summer of 2008 and 2009 (P >0.05; Fig. 2b, e; Table 2).Day warming significantly reduced survival rate (Fig. 2c) and longevity of the roots born in autumn of 2008 (P = 0.006; Table 2), while night warming increased survival rate (Fig. 2c) and longevity of the roots initiated in autumn of 2008 (P = 0.0101; Table 2).Both survival rate and longevity for the roots initiated in autumn of 2008 were not affected by continuous warming (P >0.05;Table 2).Unlike the roots born in autumn of 2008, day warming prolonged root longevity in 2009 (P = 0.001), while night and continuous warming had no effect on rot longevity in 2009 (P >0.05; Fig. 2f; Table 2).Note that values for longevity of roots born in the three seasons in 2009 were significantly shorter than those in 2008 (P <0.001; Fig. 2).
To obtain the overall response of root survival rate and longevity to day, night and continuous warming, we pooled the survival data for roots initiated in the three growing seasons of 2007-2009 and calculated the overall root longevity.In comparison with the overall root longevity in the control plots (118 d), the overall root longevity in the plots of day and continuous warming was significantly different.For instance, the mean overall root longevity in day and continuous warming plots was 7 and 34 d longer than in control plots, respectively (Fig. 3).The results of the Cox proportional hazards regression (Table 3), the mortal risk for roots in day warming and continuous warming plots were marginally (P = 0.08) and significantly reduced (P <0.001), respectively.Furthermore, day warming had significant impact on the overall root longevity (P = 0.032), but no effect (P = 0.187) of night warming on the overall root longevity was observed as analyzed by two-way ANOVAs.There was no interactive effect (P = 0.139) between day and night warming on the overall root longevity (Fig. 3).Further analyses revealed that the overall root longevity in the day warmed subplots was 19.8 % higher (P <0.05) than in the un-warmed, control subplots, whereas no significant change in the overall root longevity (P >0.05) was observed in the night warmed subplots.These results indicate that day warming has a significant effect on the overall root longevity.

Effect of day warming and night warming on root nonstructural carbohydrate content (RNC)
Day warming had significant effect on root nonstructural carbohydrate content (P <0.001), but no effect (P =0.4889) of night warming on root nonstructural carbohydrate content was observed as analyzed by two-way ANOVAs.Day warming increased root nonstructural carbohydrate content by 42.0 %.There were no interactive effects (P = 0.2195) of day and night warming on root nonstructural carbohydrate content (Fig. 4).

Warming-induced changes in root longevity varied with root nonstructural carbohydrate content (RNC) and BNPP/ANPP
Our results demonstrated that the overall root longevity was affected by day warming, but not night warming (Fig. 3).We therefore examined the correlation between root longevity and nonstructural carbohydrate in roots treated with day warming.Across the 24 plots, the day warming-induced changes in root longevity showed positive linear dependence on the day warming-induced changes in RNC (r 2 = 0.69;  P = 0.0407; Fig. 6a).Neither day nor night warming affected ANPP and BNPP (Fig. 5).In addition, no interactive effects (P = 0.6074; P = 0.8030) of day and night warming on ANPP and BNPP were observed (Fig. 5a, b).In contrast, there was an increasing trend of the ratio of BNPP/ANPP under conditions of day and continuous warming (Fig. 5c).Across the 24 plots, the day warming-induced changes in root longevity was positively dependent on changes in the ratio of BNPP/ANPP induced by day warming (r 2 = 0.85, P = 0.0095; Fig. 6b).Stepwise multiple regression analyses revealed that 41.06 % of the spatial variation in root longevity can be explained by root nonstructural carbohydrate content (partial r 2 = 0.2181, P = 0.0214) and soil temperature (partial r 2 = 0.1925, P = 0.0319).In contrast, soil moisture, ANPP, BNPP, BNPP/ANPP, GEP, ER, NEP, SR and soil inorganic N were not significantly correlated with root longevity (P >0.05).

Discussion
Several studies have demonstrated that root dynamics in the ecosystems are sensitive to continuous warming (Majdi and Ohrvik, 2004;Wan et al., 2004;Bai et al., 2010).However, there has been no study to evaluate whether day, night and continuous warming have different effects on root dynamics.In the present study, we provide direct experimental evidence demonstrating that day and night warming had different effect on root lifespan in the semi-arid grassland in northern China.More specifically, we found that roots initiated in different seasons differed in their response to day, night and continuous warming as evidenced by the three warming regimes had different effect on their root survival rate and longevity (Fig. 2, Table 2).In addition to seasonal variations of root longevity in response to the warming treatments, we found that the response of root lifespan to the warming  2).To the best knowledge of authors, there has been no report to compare longevity of roots born in different years in the literature.Our results revealed that longevity of roots born in 2008 was longer than those born in 2009.In our experiments, since the majority of roots born in 2009 were still alive by the end of experiment in October of 2009, the calculated longevity value for roots born in 2009 was underestimated because only those roots with short lifespan were included.Therefore, root longevity data obtained by monitoring root growth for longer periods and larger numbers of roots using rhizotron would be more accurate and reliable.In this context, a general trend that day warming, but not night warming, significantly increased the overall root lifespan was found when data for root survival rate in the three growing seasons of 2007-2009 were pooled as a whole.Our results showed that both day and continuous warming significantly increased the overall root lifespan, while night warming had no effect on the overall root lifespan (Fig. 3).No consistent results have been reported on the effect of continuous warming on root production and mortality in the literature.For instance, it has been reported that root mortality is increased by continuous warming (Majdi and Ohrvik, 2004;Wan et al., 2004), while Johnson et al. (2006) demonstrated that continuous warming has no effect on root production and mortality in Douglas-fir.Gill and Jackson (2000) investigated the effect of mean annual temperatures on root turnover using a W. M. Bai et al.: Root lifespan and asymmetric warming global dataset and found that the higher mean annual temperatures stimulate root turnover in grasslands.In contrast, our previous study showed that continuous warming had negative effects on root production and mortality in the same steppe in northern China (Bai et al., 2010).The different responses of root production to temperature among different ecosystems may largely be explained by the relative contributions between the direct and indirect (via changing soil moisture) effects of warming on ecosystem processes (Shaver et al., 2000;Wan et al., 2005;Bai et al., 2010).In the ecosystems where water availability is relatively abundant, the stimulatory effects of warming are likely to be stronger than the inhibitory effects, thus leading to an enhanced root production and mortality (Gill and Jackson, 2000;Majdi and Ohrvik, 2004;Wan et al., 2004).However, in the arid and semiarid ecosystems as in our study, water availability is a major limiting factor.Therefore, the inhibitory effects are likely to be stronger than the stimulatory effects of warming such that root productivity and mortality would be suppressed (Bai et al., 2010).Our findings that day warming especially continuous warming, significantly increased soil temperature and concurrently decreased soil moisture (Table 1, Fig. 1; Xia et al., 2010) are in line with this proposition.As a consequence, this would reduce root death hazard (Table 3) and prolong root lifespan.
There have been several studies examining the effect of day and night warming on ecosystem productivity, species composition and soil respiration in different ecosystems in general and in semiarid grassland in particular (Alward et al., 1999;Peng et al., 2004;Xia et al., 2009;Wan et al., 2009).However, there has been no detailed study to compare the effect of day warming on root dynamics with that of night warming in the literature so far.In the present study, we studied the response of root dynamics to day and night warming in a semiarid steppe.One important finding is that day and night warming had different effect on root lifespan such that day warming prolonged the overall root lifespan by 19.8 %, while night warming had no impact on the overall root lifespan (Fig. 3).The differential response of root lifespan to day and night warming could be explained by the allocation of ecosystem C. It has been documented that root lifespan is closely related to C supply from shoot to root (Farrar and Jones, 2000;Norby and Jackson, 2000).A decrease in C allocation to root has been suggested to shorten root lifespan (Eissenstat and Yanai, 1997;Hogberg et al., 2001).Our previous studies showed that root lifespan was negatively correlated with aboveground biomass such that an increase in C allocation to aboveground shortened root lifespan (Bai et al., 2008).In the semiarid temperate steppe ecosystem, day warming, but not night warming, has been shown to decrease soil moisture (Xia et al., 2010), thus exacerbating water stress.Our previous studies on the same experimental sites revealed that the direct effect of warming on root production and mortality was less than its indirect effect (drought effect) (Bai et al., 2010).Therefore, plants have to allocate more proportional C to roots for more efficiently acquiring water resources under warming conditions.In this context, Wan et al. (2009) reported that leaf sugar and starch depletion is reduced by day warming.This would facilitate allocation of more carbohydrates to roots.Our results that day warming increased RNC in roots (Fig. 4) are in line with their explanation.Moreover, the observations that a positive dependence of root lifespan on the day warminginduced changes in root nonstructural carbohydrate content and BNPP/ANPP (Fig. 6) provide strong evidence in support of regulation of root lifespan by C allocation in response to asymmetric warming.Our result that root nonstructural carbohydrate content accounted for 21.8 % of the spatial variation in root longevity is also consistent with this argument.
In the present study, we found that day warming, but not night warming, significantly affected the overall root lifespan in the semi arid steppe in northern China.This finding is in contrast to those results that night-warming mainly influences the processes associated with ecosystem C exchange such as GEP, NEP, ER and SR in the same experimental platform (Wan et al., 2009;Xia et al., 2009).For example, Xia et al. (2010) found that night warming significantly increases soil respiration, while day warming has no effect on soil respiration.Wan et al. (2009) reported that day and night warming significantly decrease and increase gross ecosystem productivity and net ecosystem productivity, whereas day and night warming have no any impact on ecosystem respiration.Our previous studies also demonstrated that there were interactive effects of continuous warming and precipitation on root production and mortality, and that these effects were significantly related to gross ecosystem productivity, ecosystem respiration and soil respiration (Bai et al., 2010).However, in the present study, we did not observe any correlation between the day and night warming induced-changes in root longevity and GEP, NEP, ER and SR.These results may imply that water is a key factor in the mediation of ecological processes associated with C cycle in the semi-arid steppe.Research findings in this study and two previous studies (Wan et al., 2009, Xia et al., 2009) in the same experimental platform suggest independence of above-and belowground plant and ecosystem C processes in response to climate warming, highlighting the importance of taking into account the differential response of root lifespan to day and night warming for projection and model simulation of climate-C feedback.

Conclusions
We provide the first direct experimental evidence demonstrating that the overall root lifespan in the semiarid temperate steppe in northern China was sensitive to asymmetrical warming such that day warming, but not night warming, significantly prolonged the overall root lifespan.The findings that day and night warming had differential effects on root lifespan in the semiarid temperate steppe could be largely explained by changes in allocation of photoassimilate.Our results revealed that the response of root dynamics to day and night warming differed from other ecosystem processes associated with C cycling.Therefore, these novel findings highlight the necessity of incorporation of the differential effects of day and night warming into the simulating and predicting the responses and feedbacks of terrestrial ecosystem C cycling to global warming.

Table 1 .
Results (P -values) of three-way ANOVAs on the effects of day (D) and night (N) warming and their interactions on soil temperature and soil moisture over the four experimental years(2006- 2009).

Table 2 .
Longevity of roots born in spring, summer and autumn in different treatments during the growing seasons of 2008 and 2009 (mean ± standard error).The Kaplan-Meier method was used to estimate mean lifespan.Longevity differences between different treatments were compared using Log-rank test.In each vertical row, the values with same superscript letters are not significantly different from each other at P = 0.05.

Table 3 .
Analyses of different warming treatments (day warming, night warming and continuous warming) on the root hazard ratio by the Cox proportional hazards regression.Hazard ratios for categorical covariates are the risk of death relative to a reference level, given for each covariate in the left-hand column.A negative parameter indicates that increases in the covariate would result in decreases in the risk of mortality, and the opposite trend reflects positive parameter values.The values for 95 % confidence intervals are upper and lower limits and are not symmetrical about the hazard ratio.P <0.05 is considered to be significant.