This is the accepted version of the article: Yang, Kaijun; Peng, Changhui; Peñuelas, Josep; [et al.]. «Immediate and carry- over effects of increased soil frost on soil respiration and microbial activity in a spruce forest». Soil Biology and Biochemistry, Vol. 135 (Aug. 2019), p. 51-59. DOI 10.1016/j.soilbio.2019.04.012 This version is avaible at https://ddd.uab.cat/record/217001 under the terms of the license 1 Immediate and carry-over effects of increased soil frost on soil respiration 2 and microbial activity in a spruce forest 3 Kaijun Yang1, 2, Changhui Peng3, 4, Josep Peñuelas2,5, Paul Kardol6, Zhijie Li1, 7, Li Zhang1, 4 Xiangyin Ni1, Kai Yue1, Bo Tan1, Rui Yin8, Zhenfeng Xu1* 5 1. Institute of Ecology and Forest, Sichuan Agricultural University, Chengdu, China 6 2. Global Ecology Unit CREAF-CSIC-UAB, CSIC, Barcelona, Catalonia, Spain 7 3. Department of Biological Science, Institute of Environment Sciences, University of Quebec 8 at Montreal, Montreal, Canada; 9 4. Center for Ecological Forecasting and Global Change, College of Forestry, Northwest 10 Agriculture & Forest University, Yangling, Shaanxi, China; 11 5. CREAF, Barcelona, Spain 12 6. Department of Forest Ecology and Management, Swedish University of Agricultural 13 Science, 90183, Umeå, Sweden 14 7. Forschungszentrum Jülich GmbH, Agrosphere (IBG-3), Jülich, Germany 15 8. Helmholtz-Centre for Environmental Research-UFZ, Department of Community Ecology, 16 Theodor-Lieser-Strasse 4, 06110 Halle (Saale), Germany 17 18 Correspondence: Dr. Zhenfeng Xu, Email: xuzf@sicau.edu.cn, Tel: +86 28 86290957, 19 No.211, Huimin Road, Wenjiang, Chengdu, Sichuan, 611130 China 20 Key words: Snow exclusion, soil enzyme, microbial biomass, nitrogen availability, soil 21 aggregate, fine root 22 Running title: Frost effect on soil carbon cycling 1 23 Abstract 24 Increased soil frost associated with winter climate change could have immediate and carry-over 25 effects on biological processes in high-altitude forest soils, but the nature of these processes 26 remain poorly understood. We conducted a snow-exclusion experiment to investigate the 27 immediate and cross-seasonal effects of increased soil frost on soil CO2 efflux and biological 28 activity in a subalpine spruce forest on the eastern Tibetan Plateau, China. The increased frost 29 reduced soil CO2 efflux by ~15 and ~19% in the winters of 2015/2016 and 2016/2017, 30 respectively. Increased frost tended to decrease soil basal respiration, the amount of microbial 31 phospholipid fatty acids and the activities of enzymes involved in soil carbon cycling during 32 the winters. Winter soil nitrogen availabilities were higher in the snow-exclusion treatment than 33 in the control plots. However, these effects did not carry over to the following growing season. 34 Our results suggest that increased frost reduces winter soil respiration by direct environmental 35 effects (e.g. soil temperature) and indirect biological processes (e.g. microbial biomass and 36 activity), whereas increased frost did not have any cross-seasonal effects. These findings 37 underscore the ecological importance of seasonal snowpack and microbe-associated carbon 38 processes in subalpine forests where winter snowfall is decreasing substantially. 39 40 41 42 43 44 45 1. Introduction 2 46 Seasonal snow cover is a major control of biogeochemical cycling in cold environments 47 (Jusselme et al., 2016). Many snowy areas at high latitudes and altitudes have experienced 48 substantial climate change in recent decades, and this trend is predicted to continue in this 49 century (IPCC, 2013). Climate-induced changes are particularly rapid in high latitude and 50 alpine ecosystems, where rising temperatures have profound effects on winter conditions, such 51 as snowfall, soil frost and extreme climatic events (Liu et al., 2012; IPCC, 2013). Winter 52 precipitation in these regions is more likely to occur in the form of rain rather than snow due to 53 winter warming (Wang et al., 2016). The lack of insulating snow cover could consequently 54 increase soil frost (Groffman et al., 2001a; Bokhorst et al., 2013), which could in turn have 55 complex and large impacts on soil microbiological and biochemical processes in cold forests. 56 Winter biological processes and their controls are not as well understood as growing- 57 season processes, despite the importance of winter warming and biological activity in cold 58 systems (Sanders-DeMott and Templer, 2017). Recent studies have found that soil biological 59 processes are sensitive to warming-induced changes in winter conditions, especially snow cover 60 and soil frost (Haei and Laudon, 2015; De Long et al. 2016; Li et al., 2016). Snow removal has 61 negative or neutral influences on winter soil respiration in boreal and temperate forests 62 (Groffman et al., 2006; Aanderud et al., 2013; Bokhorst et al., 2013), but changes in snow cover 63 can also alter biological processes in snow-free periods (Muhr et al., 2009; Wubs et al., 2018). 64 Snow exclusion can suppress soil respiration in the snow-free season in high-latitude 65 ecosystems (Öquist and Laudon, 2008; Zhao et al. 2017). To our knowledge, however, soil 66 biological responses to changing soil frost have rarely been investigated in both snow-covered 67 and snow-free periods in the same experiment. A better understanding of the impacts of 68 intensified soil frost on the biotic and abiotic controls over the dynamics of soil C in both snow- 69 covered and snow-free seasons is thus essential for accurately modeling and predicting potential 70 C feedbacks in a warmer world. 71 Altered soil frost may directly and indirectly affect soil C cycling, such as by affecting soil 3 72 temperature and moisture (Aanderud et al., 2013; Song et al., 2017), soil microbial biomass and 73 activity (Monson et al., 2006b; Sorensen et al., 2016) and substrate quality and quantity (Brooks 74 et al., 2004; Steinweg et al., 2008; Comerford et al., 2013). The direction and magnitude of 75 biological responses to increased frost may be determined by the combined effect of these 76 processes. Diverse techniques have provided insight in recent years into the impacts of winter 77 climate change on soil C cycling as the importance of winter processes has increased (Li et al., 78 2016a). Most field-manipulation studies have focused mainly on high-latitude systems, 79 including peatlands and boreal forests (Sanders-DeMott and Templer, 2017). Soil biological 80 responses from low-latitude cold systems with unique winter conditions, such as Tibetan 81 subalpine forests, however, remain unknown. 82 The Tibetan Plateau, the Earth’s ‘Third Pole’, has warmed substantially, especially in 83 winter (Chen et al., 2013). Winter snowfall has decreased at a rate of 0.6 mm y-1 in recent 84 decades (Wang et al., 2016; Xu et al., 2017). Seasonal snowpack in this region has unique 85 characteristics, such as shorter duration and shallower depth relative to high latitudes. Winter 86 soil temperature is also near the physical melting point and is sensitive to changes in snow cover 87 (Li et al., 2017). The subalpine forests of southwestern China contain a large amount of soil 88 organic C (Zhang et al., 2013), but most studies of global-change biology have only focused on 89 responses during the growing season (e.g. Xu et al., 2012; 2015; Yin et al., 2013), even though 90 warming is extremely pronounced and microbial activity is unexpectedly high during winter 91 (Wang et al., 2016; Wang et al. 2012; Tan et al., 2014). Future soil frost will also likely affect 92 the biological and environmental controls of soil C cycling in these forests, but the underlying 93 mechanisms of such processes remain unknown. 94 We conducted a snow-manipulation experiment to investigate the immediate and carry- 95 over effects of increased winter frost on soil C cycling in a spruce forest on the eastern Tibetan 96 Plateau. Specifically, we hypothesized that (1) more intense frost in the soil as a result of exclusion 97 of snow would decrease microbial activity and soil respiration in winter; (2) frost-reduced biological 4 98 processes would carry over into the subsequent snow-free growing season. 99 2. Materials and methods 100 2.1 Site description 101 The field manipulation experiment was conducted in a dragon spruce (Picea asperata 102 Mast.) stand at the Long-term Research Station of Alpine Forest Ecosystems of Sichuan 103 Agricultural University on the eastern Tibetan Plateau of China (31°15′N, 102°53′E; 3021 m 104 a.s.l.). The mean annual precipitation and temperature are 850 mm and 3.0 °C, respectively. 105 Snow generally begins to accumulate in late November and melts in late March the following 106 year. The soil is classified as a Cambic Umbrisol (IUSS Working Group WRB, 2007). The soil 107 (0-15 cm) contains 88.5 g kg-1 organic C and 5.4 g kg-1 nitrogen (N) and has a pH of 6.4 (Li et 108 al., 2017). 109 2.2 Experimental design 110 Winter snowfall was excluded using shelters to intensify soil frost. Shelters are considered 111 to be a useful tool for studying the responses of soil processes to winter climate change because 112 they can effectively reduce snow cover and minimize the changes in other unwanted 113 environmental conditions (Li et al., 2016a). In early November 2015, six wooden roofs were 114 set up in the spruce forest to prevent the accumulation of snow on the ground. One control plot 115 was established in the vicinity of each roof. The roofs were 2 m in height with a ground area of 116 3 × 3 m. The snow manipulation began in mid-November and ended in late March the following 117 year. 118 2.3 Soil sampling 119 Soil samples were collected from the topsoil (0-15 cm) in the frost period (FP, late January), 5 120 early thawing period (ETP, early April) and the middle of the growing season (MGS, mid- 121 August) in the year of 2016 and 2017, respectively. Three cores (5 cm in diameter, 15 cm in 122 depth) were collected in each plot at each sampling. The three cores from each plot were 123 combined to form one composite sample. Each composite sample was passed through a 2-mm 124 sieve, and any visible living plant material was manually removed. The sieved soil was used 125 for biochemical analysis. 126 2.4 Soil CO2 efflux 127 Two PVC collars (20 cm in diameter, 12 cm in height) were permanently installed in each 128 plot for measuring soil respiration. Soil CO2 efflux was measured using a portable infrared gas 129 analyzer (Li-8100, Li-Cor Inc., Lincoln, USA) between 10:00 and 14:00 (Beijing time, China 130 Standard Time) approximately every two weeks during the experimental period. The 131 measurements were repeated twice for each collar to minimize measurement errors. Soil 132 temperature and moisture at a depth of 5 cm were also measured near each collar using 133 thermocouple and Theta probes, respectively, connected to the LI-8100. Small red flags were 134 attached to the PVC collars in the control plots to minimize disturbance during the period of 135 snow cover. During the winters of 2015/2016 (four times) and 2016/2017 (once), the surface 136 snow was removed carefully from the top of the collars when the snowpack was thicker than 137 the height of the collar. We then waited 5 min to allow the system to equilibrate before 138 measuring the CO2 efflux. The removed snow was gently backfilled after the measurements. 139 We measured CO2 efflux under the natural snowpack (~10 cm) in the winter of 2015/2016 in 140 the same forest stand adjacent to the snow manipulation site. CO2 efflux was also measured 141 after removing the snow within and around the collars. CO2 efflux did not differ significantly 142 before and after snow removal (unpublished data). Snow removal therefore likely negligibly 143 affected the quantification of immediate CO2 efflux, at least within an interval of a few minutes. 6 144 2.5 Microclimate, extractable N and microbial respiration 145 Air temperature 2 m above the ground in the forest stand was measured using Thermochron 146 iButton DS1923–F5 Recorders (Maxim Dallas Semiconductor Corp., USA) every 2 h during 147 the experimental period. Meanwhile, soil temperatures 5 cm below the surface were recorded 148 in the snow-exclusion and control plots, respectively. Snow depth in the control plots was 149 measured by a metal ruler approximately every two weeks during winter. 150 Soil extractable N (nitrate, NO3--N, and ammonium, NH4+-N) was extracted with 2 M KCl 151 (1:5 soil:solution). The extracts were shaken for 1 h and filtered with a filter paper. The 152 concentrations of NO3−-N and NH4+-N in the extracts were determined by colorimetry (Li et al., 153 2017). 154 The rate of soil microbial respiration was estimated using alkali absorption (Anderson et 155 al., 1982). Soil samples (50 g) were incubated in 1-L jars at 20 °C for 2 weeks. Empty jars 156 without soil were used as controls. The CO2 produced was captured with 0.5 M NaOH in a 157 beaker suspended inside each jar. The NaOH solution was removed and titrated with 0.25 M 158 HCL solution to determine the amount of CO2 produced. Microbial respiration was reported as 159 mg CO2 kg-1 soil d-1. 160 2.6 Aggregate fraction and fine-root biomass 161 Aggregates were isolated as described by Kristiansen et al. (2006). Two soil cores from 162 each plot were collected from the 0-15 cm layer using an auger 10 cm in diameter in the early 163 thawing periods of 2015/2016 and 2016/2017 winters. Soil samples were air-dried to optimal 164 moisture (~10-15%) that would allow limited mechanical stress to maximize brittle failure 165 along natural planes of weakness, and the samples were then gently manually crumbled to <8 166 mm. The recovered samples were transferred to a nest of sieves (2 and 0.25 mm) and shaken at 167 100 min-1 for 2 min. All visible roots and stones were removed, and aggregates >2 mm (large 168 macroaggregates) were collected. The same procedure was used for the material retained on the 7 169 0.25 mm sieve, isolating an aggregate size class 0.25-2 mm (small macroaggregates). The 170 remaining material passing through the 0.25 mm sieve was identified as aggregate class <0.25 171 mm (microaggregates). 172 Two soil cores were collected from each plot using an auger (15 cm long and 10 cm in 173 diameter) in the ETPs of 2015/2016 and 2016/2017 winters. Root samples were washed in the 174 laboratory on sieves (mesh size 0.1 mm) and dried to constant weight at 65 °C. Fine roots (<2 175 mm in diameter) were separated into live and dead components based on their color and 176 mechanical consistency. 177 2.7 Assays of soil phospholipid fatty acids and enzymes 178 Microbial biomass was estimated as the total extractable phospholipid fatty acids (PLFAs) 179 with a modified method described by White et al., (1996). Lipids from 2 g of fresh soil were 180 extracted in a chloroform-methanol-phosphate buffer mixture (1:2:0.8). The phospholipids in 181 the extracts were transformed by alkaline methanolysis into fatty acid methyl esters (FAMEs), 182 which were identified by gas chromatography/mass spectrometry (GC/MS-QP2010 Series, 183 Shimadzu, Japan). Fatty acids were quantified by comparisons of the peak areas from the 184 sample with the peak areas of internal standards at 19:0 (nonadecanoic methyl ester) of the 185 known concentration. The areas were used to estimate the abundance of PLFA markers, which 186 were expressed as nmole g-1 dry soil. 187 We assessed the activities of four enzymes involved in soil C cycling: two hydrolytic 188 enzymes, β-glucosidase (BG) that catalyzes one of the later steps of cellulose degradation and 189 β-N-acetyl-glucosaminidase (NAG) involved in the breakdown of chitin and fungal cell walls, 190 and two oxidases, polyphenol oxidase (PPO) that breaks down recalcitrant polymers such as 191 lignin and humic compounds and peroxidase (POD), a nonspecific enzyme that oxidizes and 192 depolymerizes lignin. The activities were measured using assay techniques described by Allison 193 and Jastrow (2006). Substrate solutions were 5 mM pNP-β-glucopyranoside for BG, 50 mM 8 194 pyrogallol and 50 mM EDTA for PPO, 2 mM pNP-β-N-acetylglucosaminide for NAG and 5 195 mM L-DOPA and 10 μL of 0.3% H2O2 for POD. Activities were determined using a microplate 196 spectrophotometer and expressed as μmol of substrate produced or consumed h-1 g-1 dry soil. 197 Data analysis 198 A repeated-measures ANOVA was used to test the effects of treatment, sampling date and 199 their interactions on all response variables. A Bonferroni post hoc test was used to examine the 200 treatment effect on the variables on a given sampling date when the interaction of treatment and 201 sampling date was significant (P < 0.05), and a paired t-test was used when the interaction was 202 not significant. All data were tested for the assumptions of an ANOVA before analysis. 203 Heterogeneous data were ln-transformed before analysis. An exponential regression model was 204 used to describe the relationship between CO2 efflux and soil temperature during specific 205 periods (winter, growing season and entire year). All data from two winters or growing seasons 206 were used for the analyses due to the limited number of measurements. Winter was defined as 207 the period between the first day in autumn and the last day in spring when soil temperature was 208 continually below 5 °C for 5 d in the control plots. The temperature sensitivity (Q10) of soil 209 respiration was estimated using van’t Hoff equation (Van’s Hoff, 1898). R = α × eβ × T, Where R 210 is the soil respiration rate (μmol m-2 s-1), T is the soil temperature at 5 cm (°C), α and β are 211 parameters. The Q10 values are calculated as: Q10 = e10×β. All statistical tests were performed 212 using the Software Statistical Package for the Social Sciences (SPSS) version 17.0 (IBM SPSS 213 Statistics Inc., Chicago, IL, USA). 214 215 3. Results 216 3.1 Treatment effect on winter soil conditions 217 The mean and minimum air temperatures were -2.1 and -14.1 °C during the winter of 9 218 2015/2016 and -0.9 °C and -6.4 °C during the winter of 2016/2017, respectively (Figure 1 a). 219 The maximum snow depth was 40 cm in the winter of 2015/2016 but only 23 cm in the winter 220 of 2016/2017. The mean air temperature in the winter of 2015/2016 (-2.1 °C) was comparable 221 to the seven-year average of -2.4 °C for 2010-2016. The mean air temperature in the winter of 222 2016/2017 (-0.9 °C), however, was the highest in the last seven winters and 1.5 °C higher than 223 the mean. 224 The snow-exclusion treatment successfully created a more intense frost regime in both 225 winters (Figure 1 a). The minimum daily mean soil temperatures were -2.2 °C (2015/2016) and 226 -2.4 °C (2016/2017) in the snow-exclusion plots but were only -0.5 °C (2015/2016) and -1.3 °C 227 (2016/2017) in the control plots. The numbers of days with differences in soil temperature 228 ≥0.5 °C between the treatment and control plots were 42 and 56 for the winters of 2015/2016 229 and 2016/2017, respectively. Such differences were mainly during mid- and late winter when 230 snow cover was >10 cm. Soil temperature fluctuated more in the treatment than the control 231 plots. Volumetric soil moisture was similar between the treatment and control plots across the 232 two years (F=3.364, P = 0.116, Figure 1 b). 233 3.2 Soil CO2 efflux 234 The snow exclusion lowered CO2 efflux early in the winter of 2015/2016 and in mid-winter 235 of 2016/2017. The snow exclusion reduced CO2 efflux by averages of 15% and 19% in the 236 winters of 2015/2016 and 2016/2017, respectively, and these reductions were statistically 237 significant (F = 11.13, P < 0.01 for the winter of 2015/2016; F = 9.143, P < 0.05 for the winter 238 of 2016/2017). The snow-exclusion manipulation, however, did not affect CO2 efflux during 239 the snow-free growing seasons (F = 1.065, P = 0.323 for 2016; F = 1.354, P = 0.305 for 2017). 240 Mean CO2 efflux differed marginally between the frost regimes in the winter of 2015/2016 (t = 241 2.006, P = 0.076; Table 1), but differed significantly between the regimes in the winter of 242 2016/2017 (t = 3.909, P < 0.01). Mean CO2 efflux nevertheless did not differ significantly 243 between the treatment and control plots in either growing season (t = -1.584, P = 0.335 for 2016; 10 244 t = -0.285, P = 0.465 for 2017). 245 Soil CO2 efflux increased exponentially with soil temperature throughout the study period 246 (Figure 3a-c). Soil temperature explained 82-83% of the variation in CO2 effluxes during the 247 growing seasons (Figure 3b) but explained only 52-53% of the variations in the winters (Figure 248 3a). Soil temperature explained 90-91% of the variance in CO2 effluxes when the data for the 249 two years were pooled (Figure 3c). The temperature sensitivity (Q10) of the CO2 efflux was 250 23.3, 3.2 and 4.4 in the treatment plots and 22.6, 3.3 and 4.7 in the control plots for winter, 251 growing season and the entire year, respectively. Q10 did not differ significantly between the 252 frost regimes for each period modeled (all P > 0.05). 253 3.3 Soil PLFAs and microbial respiration 254 The intensified frost tended to decrease the soil PLFAs biomarkers. PLFA content was 255 lower in the snow-exclusion treatment in than the control plots in the FPs of 2015/2016 (t = - 256 2.072, P < 0.05) and 2016/2017 (t = -3.686, P < 0.05; Figure 4) but did not differ significantly 257 in the MGSs of 2016 (t = 1.368, P = 0.245) or 2017 (t = 0.035, P = 0.895). 258 Microbial activity, measured as basal respiration without roots, was estimated by 259 determining CO2 emission. The intensified frost tended to decrease soil microbial respiration 260 in the winter. The snow-exclusion treatment negatively affected soil microbial respiration in 261 the FP of 2015/2016 (t = -0.918, P < 0.05; Figure 5) and in the ETP of 2016/2017 (t = -5.821, 262 P < 0.01) but had no effect in the MGSs of 2016 and 2017 (both P > 0.05) 263 3.4 Soil enzymes 264 The activities of the soil enzymes varied significantly with sampling date (all P < 0.01, 265 Figure 6a-d). The snow-exclusion treatment tended to inhibit activities in the winter. Activity 266 was significantly lower in the treatment than in the control plots for BG in the ETP of 2015/2016 267 (t = -1.975, P < 0.05; Figure 6a) and for PPO in the ETP of 2016/2017 (t = -2.643, P < 0.05; 11 268 Figure 6b). The intensified frost decreased POD activity in the FPs of 2016 and 2017 (both P 269 < 0.05, Figure 6c) and decreased NAG activity in the ETPs of 2016 and 2017 (both P < 0.05, 270 Figure 6d) but did not significantly affect the activities of the enzymes in the MGSs of 2016 or 271 2017. 272 3.5 Soil extractable N 273 Frost treatment, sampling date and their interaction all significantly affected soil NH 4+-N 274 concentration (all P < 0.05, Figure 7a). The snow-exclusion treatment increased NH4+-N 275 concentrations in the FP and ETP of 2015/2016 (all P < 0.01) but not in the winter of 2016/2017 276 (both P > 0.05). Likewise, the intensified frost increased NO3--N concentrations in both winters 277 (F = 16.575, P < 0.01; Figure 7b). NO3--N concentrations were significantly higher in the 278 treatment than the control plots in the ETP of 2015/2016 (t = 2.309, P < 0.05) and in the FP of 279 2016/2017 (t = 5.017, P < 0.01). Neither NH4+-N nor NO3--N concentration, however, differed 280 between the frost regimes in the MGS of 2016 and 2017 (both P > 0.05). 281 3.6 Aggregate fraction and fine-root biomass 282 The relative distribution of the aggregate-size classes of the bulk soil was in the order small 283 macroaggregates (0.25-2 mm) > large macroaggregates (>2 mm) > microaggregates (<0.25 mm) 284 irrespective of frost regime (F = 221.75, P < 0.001; Table 2). The snow-exclusion treatment 285 did not affect the distribution of aggregates in the size classes (F = 0.159, P = 0.897), the live 286 fine-root biomass (F = 0.202, P = 0.663; Table 3) or the dead fine-root biomass (F = 0.171, P 287 = 0.688) in the ETP. 288 289 4. Discussion 290 We investigated the impact of intensified soil frost on soil C cycling in a Tibetan subalpine 291 spruce forest using a field experiment manipulating frost. Our main objective was to determine 12 292 whether differences in winter frost conditions induced immediate and carry-over effects on soil 293 CO2 efflux. In line with the first hypothesis, snow exclusion resulted in more intensive soil frost, 294 which decreased soil respiration in the winter season. However, contrary to what we expected, 295 increased winter frost did not have carry-over effects on soil respiration in the subsequent 296 growing season. Several potential mechanisms have been tested to account for the underlying 297 the changes of CO2 efflux during the snow-covered and snow-free seasons. 298 Firstly, frost manipulation could decrease winter CO2 efflux, in part, by the direct effect of 299 temperature, because snow exclusion during the winter decreased soil temperature. A slight 300 reduction in soil temperature over the lower range is especially important to CO2 production, 301 because the temperature sensitivity of microbial processes is extremely high at low 302 temperatures (Davidson and Janssens, 2006; Schütt et al., 2014). Soil temperature in our study 303 accounted for only 52-53% of the variance in winter CO2 efflux. CO2 efflux was very sensitive 304 to small changes in soil temperature throughout the winter, particularly near 0 °C. Q10 was much 305 higher in the winter than in the growing season regardless of frost regime. A several studies 306 have described a surprisingly higher temperature sensitivities for winter soil respiration in 307 temperate and boreal forests (Monson et al., 2006a; Muhr et al., 2009; Wang et al., 2010). Q10, 308 however, was estimated over a narrow temperature span (~5-6 °C), and the temperature-CO2 309 relationship was as weak as in temperate forests (Schindlbacher et al., 2007; Wang et al., 2010; 310 Schindlbacher et al., 2014). 311 Other factors co-varying with soil temperature may also likely regulate winter CO2 efflux. 312 The intensified frost may have produced a stronger ‘freezing drought’, which would likely limit 313 microbial activity and the extracellular diffusion of substrates (Rivkina et al., 2000). Soil 314 moisture did not differ significantly between the treatment and control plots during the winters, 315 suggesting that soil moisture was not likely responsible for the decreased winter CO2 efflux. 316 The lack of significant differences in CO2 efflux during the snow-free seasons was likewise 317 partially attributed to the lack of significant differences in both soil temperature and moisture 13 318 between the treatment and control plots during the growing season. 319 Secondly, root activity is extremely low during dormant seasons, and winter soil 320 respiration in cold ecosystems is primarily derived from microbial decomposition (Muhr et al., 321 2009; Wang et al., 2010). Winter soil respiration is thus largely determined by the biomass and 322 activity of soil microbes (Lipson et al., 2002; Moorhead et al., 2014). Soil microbes are very 323 susceptible to soil frost (Monson et al., 2006b; Aanderud et al., 2013), which can kill a 324 substantial proportion of the organisms by the rupture of cell membranes by ice crystals 325 (Sulkava and Huhta, 2003; Jusselme et al., 2016; Gavazov et al., 2017). We also found that the 326 increased frost significantly reduced soil microbial PLFAs, implying a lowered potential for 327 the microbial community to metabolize soil C in the winters. A significant decline in the cross- 328 winter microbial PLFAs also implied that seasonal frost would kill the soil microbes in the 329 spruce forest irrespective of frost manipulation. Our measurements of microbial basal 330 respiration, excluding plant roots, also indicated a similar decline with winter frost, consistent 331 with in situ soil CO2 efflux. Frost-induced decreases in winter soil respiration may thus largely 332 be attributed to the lower microbial biomass and activity. Soil PLFAs and basal respiration in 333 the middle of the growing season nevertheless did not differ significantly between frost regimes. 334 These observations may partially account for the neutral effect of increased frost on CO2 efflux 335 during the snow-free growing seasons. 336 Soil enzymes play very important roles in the cycling of soil C and nutrients. Little 337 attention has been paid to enzymatic activities in studies of winter climate change, despite the 338 importance of soil enzymes in soil C cycling. A recent study found that enzymatic activities 339 were negatively correlated with the intensity of soil frost in mixed-hardwood forests (Sorensen 340 et al., 2016), and another experiment also found that snow removal decreased the activity of 341 soil invertase in an alpine spruce-fir forest (Tan et al., 2014). We assayed the activities of four 342 enzymes involved in soil C cycling to further assess the functional capacity of soil. The 343 intensified frost tended to reduce soil enzymatic activities. Soil enzymatic activities are strongly 14 344 temperature-dependent (Tabatabai, 1982), so a decrease in soil temperature caused by snow- 345 exclusion may, to some extent, reduce soil enzymatic activities directly. The lower activities 346 may also partly be attributed to the smaller population size of the microbes, which are an 347 important source of enzyme synthesis. Soil enzymes, as proximate agents of the decomposition 348 of soil organic C, can break down plant and microbial cell walls and catabolize macromolecules 349 into soluble substrates for microbial assimilation (Sinsabaugh et al., 2008). Frost-induced 350 decreases in enzymatic activities may thus constrain this decomposition, which could also 351 partly account for the lower winter CO2 efflux. Conversely, intensified frost did not affect 352 activities in the snow-free growing seasons, which may account for the lack of significant 353 responses during the subsequent growing season. 354 Thirdly, soil frost may also have affected the decomposition of soil C in winter by altering 355 nutrient availability. Intensified frost can increase the mortality of roots and microbes (Henry, 356 2007; Repo et al., 2014; Blume-Werry et al., 2016), which are important substrates for soil 357 microbial metabolism during winter (Schimel et al., 2004). Dead roots and microbes are also 358 main N sources during winter in cold systems (Chapin III et al., 1988; Tierney et al., 2001). In 359 an earlier study we observed that soil at -5 °C could release considerable extractable N in the 360 soils of this spruce forest, possibly due to the effect of freezing on microbial mortality (Xu et 361 al., 2014). The snow-exclusion treatment in the present study stimulated the production of soil 362 extractable N in the two winters, likely due mainly to the increased microbial mortality. An 363 increase in N availability but a decrease in soil PLFAs throughout the winter, irrespective of 364 the frost regime, may also support this conclusion. Live and dead fine-root biomass also notably 365 did not differ significantly between frost regimes later in the winter, further suggesting that the 366 increased N availability was mainly attributed to microbial mortality rather than root injury. An 367 increase in N availability coincided with a decrease in CO2 efflux, implying that the cycling of 368 soil C could be decoupled from N availability during winter under intensified frost. 369 In addition to microbial and root mortality, substrate availability could have been affected 15 370 by the physically disruptive effects of frost on soil aggregates (Chai et al., 2014). Freezing can 371 break down macroaggregates into microaggregates (Oztas and Fayetorbay, 2003). 372 Microaggregates with a larger surface area have more contact points, which can potentially 373 increase the amount of substrate decomposed by microorganisms (Grogan et al., 2004). Snow 374 removal increased the fraction of microaggregates in a northern hardwood forest, implying that 375 soil substrate could become more accessible to soil microorganisms (Steinweg et al., 2008). 376 Our observations, however, did not provide further evidence that more intense frost could 377 disrupt aggregates in the soil of this Tibetan spruce forest. The intensified frost did not affect 378 the distribution of aggregates among the size classes, suggesting that frost-associated changes 379 to aggregates may not importantly affect soil respiration in the spruce forest during the winter 380 and growing season. 381 Lastly, the flux of CO2 derived from decaying litter accounts for a considerable part of 382 total soil respiration during winter (Uchida et al., 2005). CO2 flux derived from aboveground 383 litter accounts for an average of 14.2% of total soil respiration in this spruce-forest stand (Xiong 384 et al., 2015). In a previous study we also found that the mass loss of spruce needles over the 385 winter constituted 18.3-28.8% of the net loss rates for the entire year (Xu et al., 2016). The lack 386 of snow cover at this experimental site decreased the temperature of the surface soil by an 387 average of 1.4 °C during the winter (Li et al., 2017), implying that litter decomposition was 388 most likely inhibited by the lower temperatures. A growing number of studies have documented 389 that thick snow covers can provide relatively stable conditions for biological activity, favoring 390 the decomposition of plant litter (Christenson et al., 2010; Bokhorst et al., 2013; Saccone et al., 391 2013). The rates of decomposition of litter from subalpine tree species in this area similarly 392 decrease with decreasing snow depth (Ni et al., 2014; He et al., 2015). The lower rate of litter 393 decomposition due to the lack of snow cover may therefore also have contributed to the lower 394 winter soil respiration in the snow-exclusion plots, but further supporting evidence is needed. 395 16 396 5. Implications 397 The climate on the Tibetan Plateau has changed considerably in recent decades, especially 398 in winter (Wang et al., 2016). Winter snowfall has tended to decrease substantially due to strong 399 winter warming (Xu et al., 2017). The decrease or absence of insulating snow cover associated 400 with climate change may thus increase the duration and intensity of soil frost in the future in 401 this special region. The importance of soil frost, large storage of soil C and sensitivity of snow 402 cover to winter warming indicate that understanding the potential effects of projected frost 403 increases on soil C cycling in the subalpine forests of western China is essential. To our 404 knowledge, our study is the first to identify the effects of changes in soil frost on soil C cycling 405 in a Tibetan forest. Our results generally indicate that more intense soil frost decreases winter 406 soil respiration and biological activities. Winter soil CO2 emission was lower in the snow- 407 exclusion than in the control plots during the two winters of the study. Intensified soil frost, 408 however, did not affect soil CO2 efflux and biological activities during the subsequent growing 409 season, suggesting that a short-term change in snow cover does not produce large carry-over 410 effects in snow-free periods. If the observed effects apply to natural conditions, intensified soil 411 frost would decrease the amount of soil C released to the atmosphere from subalpine forests 412 during winter, but additional supporting evidence is needed. 413 This study was conducted during two contrasting winters (cold winter and thick snow 414 cover in 2015/2016 and mild winter and thin snow cover in 2016/2017) so offered a good 415 opportunity for determining the effect of the lack of snow cover in winters with different 416 weather on soil C cycling in the Tibetan spruce forest. The decrease in soil respiration due to 417 frost in the first winter occurred early but then disappeared, suggesting that soil biological 418 processes may begin to acclimate to the frost late in the winter. Soil respiration early in the mild 419 winter of 2016/2017 did not differ significantly between the treatment and control plots, mainly 420 due to the absence of an insulating snow cover. CO2 effluxes, however, were lower in the 421 treatment plots after the formation of a steady snow cover (>10 cm). In addition to variable 17 422 snowfall, extreme winter events (e.g. warm weather and snow storms) may become more 423 frequent and likely under scenarios of future climate, indicating the complexity and uncertainty 424 of winter climate change in this specific region. The comparably strong climate change and 425 variable winter snowfall on the Tibetan Plateau bring great challenges and opportunities for 426 studying winter climate change and its impacts on the structure and function of Tibetan 427 ecosystems. Long-term monitoring is strongly needed for exploring the natural winter 428 variations and underlying mechanisms of the observed phenomena to help models for providing 429 more realistic predictions of future winter conditions. 430 The frost intensity due to the lack of snow cover was low at our experimental forest site, 431 unlike in temperate and boreal forests (e.g. Groffman et al., 2006; Muhr et al., 2009; Sorensen 432 et al., 2016), but the difference in temperature minima was nearly 2 °C, likely due to the site- 433 specific characteristics, such as winter snowfall, air temperature, properties of soil heat transfer 434 and albedo. Such soil frost, however, had large impacts on soil respiration, microbial PLAFs, 435 enzymatic activity and N availability, suggesting that Tibetan forest soils will be sensitive to 436 changing soil frost in the future. The direction and magnitude of the response of soil respiration 437 to intensified soil frost may largely depend on the interaction between less snowfall and warmer 438 temperature in winter. Winter warming may offset the negative effects induced by frost to some 439 extent. Seasonal snow cover in cold regions plays a key role in decoupling soil from cold winter 440 weather, but soil temperature is often insensitive to a small change in air temperature. Changes 441 in snow cover will thus likely have a stronger influence on soil biological processes than winter 442 warming itself (Groffman et al., 2001a), and long-term changes in soil frost may also have 443 carry-over effects on soil C dynamics during the subsequent growing season in cold systems 444 (Zhao et al., 2017). More research is warranted to integrate potential factors and separate their 445 relative importance for a better understanding and ability to predict potential C feedbacks in 446 snowy regions under a warmer future. 447 18 448 449 6. Conclusions 450 This study explored the immediate and carry-over effects of intensified soil frost on soil C 451 cycling in a subalpine spruce forest on the Tibetan Plateau of China. Our results suggested that 452 the lack of snow cover increased the intensity of soil frost, which decreased soil respiration in 453 the winters but did not affect it during the subsequent growing seasons. Frost decreased 454 microbial biomass and activities in the winters but not in the snow-free growing seasons. More 455 intense soil frost did not affect the size distribution of soil aggregates or the fine-root biomass. 456 Predicted soil frost due to winter climate change may thus decrease winter soil respiration by 457 direct environmental effects (e.g. soil temperatures) and indirect biological processes (e.g. 458 microbial biomass and activities) in the subalpine forests on the Tibetan Plateau. Intensified 459 soil frost did not cause cross-seasonal effects on soil CO2 efflux or biological activities in the 460 subsequent growing seasons. Our observations underscore the ecological importance of 461 seasonal snowpack and microbe-associated C processes in subalpine forest soils. These findings 462 improve our understanding of the response of soil C dynamics to winter climate change in this 463 region experiencing large decreases in winter snowfall. 464 465 466 Acknowledgements 467 We are very grateful to Fan Yang, Liyan Zhuang, Yu Tan and Rui Cao for their great support 468 with field and lab. We also thank Martin Schädler, Prof. Dr. Wanqin Yang and Pro. Dr. Fuzhong 469 Wu for their helpful suggestions on the statistical analysis. We also thank the experimental 470 platform－Long-term Research Station of Alpine Forest Ecosystems. This work was supported 471 by the National Natural Science Foundation of China (31570601 and 31700542) and the 19 472 National Key Research and Development Program of China (2016YFC0502505 and 473 2017YFC0505003). Kaijun Yang appreciates the financial support from the China Scholarship 474 Council (CSC). 475 476 477 Competing interests 478 The authors declare no competing financial interests. 479 480 481 482 483 484 References 485 Aanderud, Z.T., Jones, S.E., Schoolmaster Jr, D.R., Fierer, N., Lennon, J.T., 2013. Sensitivity 486 of soil respiration and microbial communities to altered snowfall. Soil Biology & 487 Biochemistry 57, 217–227. 488 Anderson, J.P.E., 1982. Soil Respiration. In: Page, A.L., Miller, R.H., Keeney, D.R. (Eds.), 489 Methods of Soil Analysis, Part 2: Chemical and Microbiological Properties. American 490 Society of Agronomy-Soil Science Society of America, Madison, pp. 831-871. 491 Allison, S.D., Jastrow, J.D., 2006. Activities of extracellular enzymes in physically isolated 492 fractions of restored grassland soils. Soil Biology & Biochemistry 38, 3245–3256. 493 Blume-Werry, G., Kreyling, J., Laudon, H., Milbau, A., 2016. Short-term climate change 494 manipulation effects do not scale up to long-term legacies: effects of an absent snow cover 20 495 on boreal forest plants. Journal of Ecology 104, 1638–1648. 496 Bokhorst, S., Metcalfe, D.B., Wardle, D.A., 2013. Reduction in snow depth negatively affects 497 decomposers but impact on decomposition rates is substrate dependent. Soil Biology & 498 Biochemistry 62, 157–164. 499 Brooks, P.D., Mcknight, D., Elder, K., 2004. Carbon limitation of soil respiration under winter 500 snowpacks: potential feedbacks between growing season and winter carbon fluxes. Global 501 Change Biology 11, 231–238. 502 Chai, Y.J., Zeng, X.B., S, Z.E., Bai, L.Y., Su, S.M., Huang, T., 2014. Effects of freeze-thaw on 503 aggregate stability and the organic carbon and nitrogen enrichment ratios in aggregate 504 fractions. Soil Use and Management 30, 507–516. 505 Chapin III, F.S., Fetcher, N., Kielland, K., Everett, K.R., Linkins, A.E., 1988. Productivity and 506 nutrient cycling of Alaskan tundra: enhancement by flowing soil water. Ecology 69, 693– 507 702. 508 Chen, H., Zhu, Q.A., Peng, C.H., Wu, N., Wang Y.F., Fang, X.Q., Gao, Y.H., Zhu, D., Yang, 509 G., Tian, J.Q., Kang, X., Piao, S.L., Ouyang, H., Xiang, W.H., Luo, Z.B., Jiang, H., Song, 510 X.Z., Zhang, Y., Yu, G.R., Zhao, X.Q., Gong, P., Yao, T.D., Wu, J.H., 2013. The impacts 511 of climate change and human activities on biogeochemical cycles on the Qinghai-Tibetan 512 Plateau. Global Change Biology 19, 2940–2955. 513 Christenson, L.M., Mitchell, M.J., Groffman, P.M., Lovett, G.M., 2010. Winter climate change 514 implications for decomposition in northeastern forests: comparisons of sugar maple litter 515 with herbivore fecal inputs. Global Change Biology 16, 2589–2601. 516 Comerford, D.P., Schaberg, P.G., Templer, P.H., Socci, A.M., Campbell, J.L., Wallin K.F., 517 2013. Influence of experimental snow removal on root and canopy physiology of sugar 518 maple trees in a northern hardwood forest. Oecologia 171, 261–269. 519 Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomposition and 520 feedbacks to climate change. Nature 440, 165–173. 521 De Long, J.R., Laudon, H, Blume-Werry, G, Kardol, P. 2016. Nematode community resistant 21 522 to deep soil frost in boreal forest soils. Pedobiologia 59, 243–251. 523 Du, E., Zhou, Z., Li, P., Jiang, L., Hu, X., Fang, J., 2013. Winter soil respiration during soil- 524 freezing process in a boreal forest in Northeast China. Journal of Plant Ecology 6, 800 525 349–357. 526 Gavazov, K., Ingrisch, J., Hasibeder, R., Mills, R.T.E., Buttler, A., Gleixner, G., Pumpanen, J., 527 Bahn, M., 2017. Winter ecology of a subalpine grassland: Effects of snow removal on soil 528 respiration, microbial structure and function. Science of the Total Environment 590, 316- 529 324. 530 Groffman, P.M., Driscoll, C.T., Fahey, T.J., Hardy, J.P., Fitzhugh, R.D., Tierney, G.L., 2001a. 531 Colder soils in a warmer world: A snow manipulation study in a northern hardwood forest 532 ecosystem. Biogeochemistry 56, 135–150. 533 Groffman, P.M., Driscoll, C.T., Fahey, T.J., Hardy, J.P., Fitzhugh, R.D., Tierney, G.L., 2001b. 534 Effects of mild winter freezing on soil nitrogen and carbon dynamics in a northern 535 hardwood forest. Biogeochemistry 56, 191–213. 536 Groffman, P.M. , Hardy, J.P., Driscoll, C.T., Fahey, T.J., 2006. Snow depth, soil freezing, and 537 fluxes of carbon dioxide, nitrous oxide and methane in a northern hardwood forest. Global 538 Change Biology 12, 1748–1760. 539 Grogan, P., Michelsen, A., Ambus, P., Jonasson, S., 2004. Freeze-thaw regime effects on 540 carbon and nitrogen dynamics in sub-arctic heath tundra mesocosms. Soil Biology & 541 Biochemistry 36, 641–654. 542 Haei, M., Laudon, H., 2015. Carbon dynamics and changing winter conditions: a review of 543 current understanding and future research directions. Biogeosciences Discussion 12, 544 15763–15808. 545 Henry, H.A.L., 2007. Soil freeze-thaw cycle experiments: trends, methodological weaknesses 546 and suggested improvements. Soil Biology & Biochemistry 39, 977–986 547 He, W., Wu, F.Z., Zhang, D.J., Yang, W.Q., Tan, B., Zhao, Y.Y., Wu, Q.Q., 2015. The effects 548 of forest gaps on cellulose degradation in the foliar litter of two shrub species in an alpine 22 549 fir forest. Plant and Soil 393, 109–122. 550 IPCC (2013) Climate Change 2013: The Physical Science Basis. Contribution of Working 551 Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. 552 Cambridge University Press, Cambridge, UK and New York, NY. 553 Janssens, I.A., Pilegaard, K., 2003. Large seasonal changes in Q10 of soil respiration in a beech 554 forest. Global Change Biology 9, 911–918. 555 Jusselme, M.D., Saccone, P., Zinger, L., Faure, M., Le Roux X., Guillaumaud, N., Bernard,L., 556 Clement, J. C., Poly, F., 2016. Variations in snow depth modify N-related soil microbial 557 abundances and functioning during winter in subalpine grassland. Soil Biology & 558 Biochemistry 92, 27–37. 559 Kristiansen, S.M., Schjnning, P., Thomsen, I.K., Olesen, J.E., Kristensen, K., Christensen, B.T., 560 2006. Similarity of differently sized macro-aggregates in arable soils of different texture. 561 Geoderma 137, 147–154. 562 Lipson, D.A., Schadt, C.W., Schmidt, S.K., 2002. Changes in soil microbial community 563 structure and function in an alpine dry meadow following spring snow melt. Microbial 564 Ecology 43, 307–314. 565 Liu, J.P., Curry, J. A., Wang, H.J., Song, M.R., Horton, R. M., 2012. Impact of declining Arctic 566 sea ice on winter snowfall. Proceedings of the National Academy of Sciences 109, 6781- 567 6783. 568 Li, W.B., Wu, J.B., Bai, E., Jin, C.J., Wang, A.Z., Yuan, F.H., Guan, D.X., 2016a. Response of 569 terrestrial carbon dynamics to snow cover change: A meta-analysis of experimental 570 manipulation (II). Soil Biology & Biochemistry 103, 388–393. 571 Li, Z.J., Yang, W.Q., Yue, K., Justine, M.F., He, R.Y., Yang, K.J., Zhuang, L.Y., Wu, F.Z., Tan, 572 B., Zhang, L., Xu, Z.F., 2017. Effects of snow absence on winter soil nitrogen dynamics in a 573 subalpine spruce forest of southwestern China. Geoderma, 307, 107–113. 574 Monson, R.K., Lipson, D.L., Burns, S.P., Turnipseed, A.A., Delany, A.C., Williams, M.W., 575 Schmidt, S.K., 2006b. Winter forest soil respiration controlled by climate and microbial 23 576 community composition. Nature 439, 711–714. 577 Moorhead, D., Lashermes, G., Recous, S., Bertrand, I., 2014. Interacting microbe and litter 578 quality controls on litter decomposition: a modeling analysis. PLOS ONE 9: e108769. 579 Muhr, J., Borken, W., Matzner, E., 2009. Effects of soil frost on soil respiration and its 580 radiocarbon signature in a Norway spruce forest soil. Global Change Biology 15, 782– 581 793. 582 Ni, X.Y., Yang, W.Q., Li, H., Xu, L.Y., He, J., Tan, B., Wu, F.Z., 2014. The responses of early 583 foliar litter humification to reduced snow cover during winter in an alpine forest. Canadian 584 Journal of Soil Science 94, 453–461. 585 Oquist, M.G., Laudon, H., 2008. Winter soil frost conditions in boreal forests control growing 586 season soil CO2 concentration and its atmospheric exchange. Global Change Biology 14, 587 2839–2847. 588 Oztas, T., Fayetorbay, F., 2003. Effect of freezing and thawing processes on soil aggregate 589 stability. Catena 52, 1–8. 590 Repo, T., Sirki, S., Heinonen, J., Lavigne, A., Roitto, M., Koljonen, E., Koljonen, E., Sutinen, 591 S., Finer, L., 2014. Effects of frozen soil on growth and longevity of fine roots of Norway 592 spruce. Forest Ecology and Management 313, 112–122. 593 Rivkina, E.M., Friedmann, E.I., McKay, C.P., Gilichinsky, D.A., 2000. Metabolic activity of 594 permafrost bacteria below the freezing point. Applied and Environmental Microbiology 595 66, 3230–3233. 596 Saccone, P., Morin, S., Baptist, F., Bonneville, J.M., Colace, M.P., Domine, F., Faure, M., 597 Geremia, R., Lochet, J., Poly, F., Lavorel, S., Clement, J.C., 2013. The effects of snowpack 598 properties and plant strategies on litter decomposition during winter in subalpine meadows. 599 Plant and Soil 363, 215–229. 600 Sanders-DeMott, R., Templer, P.H., 2017. What about winter? Integrating the missing season 601 into climate change experiments in seasonally snow covered ecosystems. Methods in 602 Ecology and Evolution 8, 1183–1191. 24 603 Schimel, J.P., Bilbrough, C., Welker, J. A., 2004. Increased snow depth affects microbial 604 activity and nitrogen mineralization in two Arctic tundra communities. Soil Biology & 605 Biochemistry 36, 217–227. 606 Schindlbacher, A., Jandl, R., Schindlbacher, S., 2014. Natural variations in snow cover do not 607 affect the annual soil CO2 efflux from a mid-elevation temperate forest. Global Change 608 Biology 20, 622–632. 609 Schindlbacher, A., Zechmeister-Boltenstern, S., Glatzel, G., Jandl, R., 2007. Winter soil 610 respiration from an Austrian mountain forest. Agricultural and Forest Meteorology 146, 611 205–215. 612 Schutt, M., Borken, W., Spott, O., Stange, C.F., Matzner, E., 2014. Temperature sensitivity of 613 C and N mineralization in temperate forest soils at low temperatures. Soil Biology & 614 Biochemistry 69, 320–327. 615 Sinsabaugh, R.L., Lauber, C.L., Weintraub, M.N., Ahmed, B., Allison, S.D., Crenshaw, C., 616 Contosta, A.R., Cusack, D., Frey, S., Gallo M.E., 2008. Stoichiometry of soil enzyme 617 activity at global scale. Ecology Letters 11, 1252–1264. 618 Song, Y., Zou, Y.C., Wang, G.P., Yu, X.F., 2017. Altered soil carbon and nitrogen cycles due 619 to the freeze-thaw effect: A meta-analysis. Soil Biology & Biochemistry 109, 35–49. 620 Sorensen, P.O., Templer, P.H., Finzi, A.C., 2016. Contrasting effects of winter snowpack and 621 soil frost on growing season microbial biomass and enzyme activity in two mixed- 622 hardwood forests. Biogeochemistry 128, 141–154. 623 Steinweg, J.M., Fisk, M.C., McAlexander, B., Groffman, P.M., Hardy, J.P., 2008. Experimental 624 snowpack reduction alters organic matter and net N mineralization potential of soil 625 macroaggregates in a northern hardwood forest. Biology and Fertility of Soils 45, 1–10 626 Sulkava, P., Huhta, V., 2003. Effects of hard frost and freeze–thaw cycles on decomposer 627 communities and N mineralisation in boreal forest soil. Applied Soil Ecology 22, 225– 628 239. 629 Tan, B., Wu, F.Z., Yang, W.Q., He, X.H., 2014. Snow removal alters soil microbial biomass 25 630 and enzyme activity in a Tibetan alpine forest. Applied Soil Ecology 76, 34–41. 631 Tierney, G.L., Fahey, T.J., Groffman, P.M., Hardy, J.P., Fitzhugh, R.D., Driscoll, C.T., 2001. 632 Soil freezing alters fine root dynamics in a northern hardwood forest. Biogeochemistry 56, 633 175–190. 634 Uchidaa, M., Mo, W., Nakatsubo, T., Tsuchiya, Y., Horikoshi, T., Koizumi, H., 2005. 635 Microbial activity and litter decomposition under snow cover in a cool-temperate broad- 636 leaved deciduous forest. Agricultural and Forest Meteorology 134, 102-109. 637 Van’s Hoff, J.H., 1898. Lectures on Theoretical and Physical Chemistry. Part 1: Chemical 638 Dynamics. Edward Arnold, London, pp. 224–229. 639 Wang, A., Wu, F.Z., Yang, W.Q., Wu, Z.C., Wang, X.X., Tan, B., 2012. Abundance and 640 composition dynamics of ammonia-oxidizing archaea in alpine fir forest on the eastern 641 Tibetan Plateau of China. Canadian Journal of Microbiology 58, 572-580. 642 Wang, C., Han, Y., Chen, J., Wang, X., Zhang, Q., Bond-Lamberty, B., 2013. Seasonality of 643 soil CO2 efflux in a temperate forest: Biophysical effects of snowpack and spring freeze–929 644 thaw cycles. Agricultural and Forest Meteorology 177, 83–92. 645 Wang, J., Zhang, M.J., Wang, S.J., Ren, Z.G., Che, Y.J., Qiang, F., Qu, D.Y., 2016. Decrease in 646 snowfall/rainfall ratio in the Tibetan Plateau from 1961 to 2013. Journal of Geograrphy 647 Sciences 26, 1277–1288. 648 Wang, W., Peng, S.S., Wang, T., Fang, J.Y., 2010. Winter soil CO2 efflux and its contribution 649 to annual soil respiration in different ecosystems of a forest-steppe ecotone, north China. 650 Soil Biology & Biochemistry 42, 451–458. 651 White, D.C., Stair, J.O., Ringelberg, D.B., 1996. Quantitative comparisons of in situ microbial 652 biodiversity by signature biomarker analysis. Journal of Industrial Microbiology 17, 185– 653 196. 654 Wubs, E.R.J., Woodin, S.J., Stutter, M.I., Wipf, S., Sommerkorn, M., Wal, R.V.D., 2018. Two 655 decades of altered snow cover does not affect soil microbial ability to catabolize carbon 656 compounds in an oceanic alpine heath. Soil Biology & Biochemistry 124, 101–104. 26 657 Xiong, L., Xu, Z.F., Yang, W.Q., Yin, R., Tang, S.S., Wang, B., Xu, L.Y., Chang, C.H., 2015. 658 Aboveground litter contribution to soil respiration in a subalpine dragon spruce plantation 659 of western Siehuan. Acta Ecologica Sinica, 35, 4678–4686, in Chinese with an English 660 abstract. 661 Xu, M., Qi, Y., 2001. Spatial and seasonal variations of Q10 determined by soil respiration 662 measurements at a Sierra Nevadan forest. Global Biogeochemitry Cycles 15, 687–696. 663 Xu, W.F., Ma, L.J., Ma, M.N., Zhang, H.C., Yuan, W.P., 2017. Spatial–temporal variability of 664 snow cover and depth in the Qinghai-Tibetan Plateau. Journal of Climate 30, 1521–1533. 665 Xu, Z.F., Liu, Q., Yin, H,J., 2014. Effects of temperature on soil net nitrogen mineralisation in 666 two contrasting forests on the eastern Tibetan Plateau, China. Soil Research 52, 562–567. 667 Xu, Z.F., Pu, X.Z., Yin, H.J., Zhao, C.Z., Liu, Q., Wu, F.Z., 2012. Warming effects on the early 668 decomposition of three litter types, Eastern Tibetan Plateau, China. European Journal of 669 Soil Science 63, 360–367. 670 Xu, Z.F., Wan, C.A., Xiong, P., Tang, Z., Hu, R., Cao, G., Liu, Q., 2010. Initial responses of 671 soil CO2 efflux and C, N pools to experimental warming in two contrasting forest 672 ecosystems, Eastern Tibetan Plateau, China. Plant and Soil 336, 183–195. 673 Xu, Z.F., Zhao, C.Z., Yin, H.J., Liu, Q., 2015. Warming and forest management interactively 674 affect the decomposition of subalpine forests on the eastern Tibetan Plateau: A four-year 675 experiment. Geoderma 239, 223-228. 676 Xu, Z.F., Zhu, J.X., Wu, F.Z., Liu, Y., Tan, B., Yang, W.Q., 2016. Effects of litter quality and 677 climate change along an elevational gradient on litter decomposition of subalpine forests, 678 Eastern Tibetan Plateau, China. Journal of Forestry Research 27, 505–511. 679 Yin, H.J., Li, Y.F., Xiao, J., Xu, Z.F., Cheng, X.Y., Liu, Q., 2013. Enhanced root exudation 680 stimulates soil nitrogen transformations in a subalpine coniferous forest under 681 experimental warming. Global Change Biology 19, 2158–2167. 682 Zhang, Y.D., Gu, F.X., Liu, S.R., Liu,Y.C., Li, C., 2013. Variations of carbon stock with forest 683 types in subalpine region of southwestern China. Forest Ecology and Management 300, 27 684 88–95. 685 Zhao, J.B., Peichl, M., Nilsson, M.B., 2017. Long-term enhanced winter soil frost alters 686 growing season CO2 fluxes through its impact on vegetation development in a boreal 687 peatland. Global Change Biology 23, 3139–3153. 688 689 690 691 692 693 694 695 Table 1 Mean soil CO2 efflux (μmol m-2 s-1, means ± SEs) during the winter and growing 696 season. Year Period Treatment Control 2016 Winter 0.47 ± 0.08 0.55 ± 0.08 Growing season 2.07 ± 0.28 2.19 ± 0.29 2017 Winter 0.43 ± 0.06 0.53 ± 0.07 Growing season 2.23 ± 0.26 2.25 ± 0.30 697 698 699 Table 2. Relative distribution of the aggregate size classes (%, means ± SEs) in the snow- 700 exclusion treatment and control plots in the early thawing period of 2016 and 2017. 28 Year Size Treatment Control 2016 >2 mm 34.9 ± 5.7 32.9 ± 6.9 0.25-2 mm 59.9 ± 4.7 58.6 ± 4.2 <0.25 mm 5.2 ± 2.0 8.5 ± 2.8 2017 >2 mm 35.3 ± 6.4 36.7 ± 4.5 0.25-2 mm 53.9 ± 3.8 50.7 ± 2.2 <0.25 mm 10.8 ± 3.6 12.6 ± 2.8 701 702 Table 3. Content of live and dead fine roots (g m-2, means ± SEs) up to a depth of 15 cm in the 703 snow-exclusion treatment and control plots in the early thawing period of 2016 and 2017. Year Fine root pool Treatment Control 2016 Live 243.7 ± 45.6 251.7 ± 35.6 Dead 92.8 ± 14.4 79.4 ± 26.3 2017 Live 221.2 ± 38.5 215.6 ± 28.9 Dead 94.3 ± 22.6 88.1 ± 16.4 704 705 29 706 707 708 709 710 711 712 713 Figure legends 714 Figure 1. Air temperature, soil temperature and snow depth (a) and soil moisture (b) in the 715 snow-exclusion treatment and control plots during the experimental period. 716 Figure 2. Soil CO2 efflux (means ± SEs) in the snow-exclusion treatment and control plots 717 during the experimental period. Significant differences between the control and treatment on a 718 given date are indicated by asterisks (P < 0.05). 719 Figure 3. Exponential relationships between soil CO2 efflux and soil temperature (means ± SEs) 720 in the treatment and control plots for winter, growing season and entire year, respectively. 721 Figure 4. Total soil phospholipid fatty acids (means ± SEs) in the snow-exclusion treatment 722 and control plots. Significant differences between the control and treatment on a given date are 723 indicated by asterisks (P < 0.05). FP, frost period; ETP, early thawing period; MGS, middle of 724 the growing season. 30 725 Figure 5. Soil microbial respiration (means ± SEs) in the snow-exclusion treatment and control 726 plots. Significant differences between the control and treatment on a given date are indicated 727 by asterisks (P < 0.05). FP, frost period; ETP, early thawing period; MGS, middle of the growing 728 season. 729 Figure 6. Activities of (a) β-glucosidase, (b) polyphenol oxidase, (c) peroxidase and (d) β-N- 730 acetyl-glucosaminidase (means ± SEs) in the snow-exclusion treatment and control plots. 731 Significant differences between the control and treatment on a given date are indicated by 732 asterisks (P < 0.05). FP, frost period; ETP, early thawing period; MGS, middle of the growing 733 season. 734 Figure 7. Soil ammonium (a) and nitrate (b) concentrations (means ± SEs) in the snow- 735 exclusion treatment and control plots. Significant differences between the control and treatment 736 on a given date are indicated by asterisks (P < 0.05). FP, frost period; ETP, early thawing period; 737 MGS, middle of the growing season. 738 739 740 741 742 743 31 744 745 746 747 748 749 750 Figure 1 32 751 752 753 754 Figure 2 33 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 Figure 3 34 772 4 Growing season Soil CO2 efflux ( mol m s ) -2 -1 3 2 1 Control: R2=0.83, P<0.001 Treatment: R2=0.82, P<0.001 0 -3 0 3 6 9 12 15 773 Soil temperature (℃) 4 Entire year Soil CO2 efflux ( mol m s ) -2 -1 3 2 1 Control: R2=0.90, P<0.001 Treatment: R2=0.91, P<0.001 0 -3 0 3 6 9 12 15 774 Soil temperature (℃) 775 776 Figure 4 35 777 778 779 Figure 5 780 781 782 36 783 784 785 786 787 Figure 6 788 789 790 791 Figure 7 37 792 793 38