Forest buffers soil temperature and postpones soil thaw as indicated by a three-year large-scale soil temperature monitoring in the forest-steppe ecotone in Inner Asia

Global and Planetary Change 104 (2013) 1–6

Contents lists available at SciVerse ScienceDirect

Global and Planetary Change

j ourna l homepage: www.e lsev ie r .com/ locate /g lop lacha

Forest buffers soil temperature and postpones soil thaw as indicated by a three-yearlarge-scale soil temperature monitoring in the forest-steppe ecotone in Inner Asia

Guozheng Hu a, Hongyan Liu a,⁎, Oleg A. Anenkhonov b, Andrey Yu. Korolyuk c,Denis V. Sandanov b, Dali Guo d

a Department of Ecology, College of Urban and Environmental Sciences, Peking University, Beijing 100871, Chinab Institute of General and Experimental Biology, Russian Academy of Sciences, Siberian Branch, Sakhyanovoi Str. 6, Ulan Ude 670047, Russiac Central Siberian Botanical Garden, Russian Academy of Sciences, Siberian Branch, Zolotodolinskaya Str. 101, Novosibirsk 630090, Russiad Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China

⁎ Corresponding author.E-mail address: [email protected] (H. Liu).

0921-8181/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.gloplacha.2013.02.002

a b s t r a c t

Article history:Received 12 October 2012Accepted 7 February 2013Available online 16 February 2013

Keywords:Soil temperatureSoil freeze and thawForest-steppe ecotoneSouthern SiberiaNorthern China

Increases in soil temperature affect both soil biotic process and the soil hydrological cycle in the mid andhigh latitudes of the Northern Hemisphere. Using a three-year automatic record of soil temperature inpaired forest and steppe plots along a temperature gradient from northern China to southern Siberia inRussia, collected from 2008–2011, we investigated how vegetation cover has impacted soil temperatureat a regional scale, with focus on soil freezing/thawing timing. We found that there was a buffering effectof forests on soil temperature, as indicated by cooler soil temperatures in the warm season and warmersoil temperatures in the cold season. Forest soil thawed about 15 days later than steppe soil at the samesite. At a regional scale, the onset date of soil freezing showed significant positive correlation with cold sea-son soil temperature in both forest and steppe. The onset date of soil thawing was significantly negativelycorrelated with cold seasonmean daily soil temperature (DST) in the steppe, but showed no significant cor-relation in the forest. In terms of space-for-time-substitution, it could be implied that forest might face anincreased likelihood of drought by impeding snowmelt infiltration, under the present warming trend insoil temperature.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Over the last three decades, the global land surface temperaturehas increased by 0.32±0.09 °C per decade, with the largest increasein the Northern Hemisphere (IPCC, 2007). A corresponding increasingtrend in soil temperature has been reported in mid and high latitudesof the Northern Hemisphere, e.g. northeastern China and northernChina (Wang et al., 2009), northern and northwestern U.S. (Hu andFeng, 2003), and the Netherlands (Jacobs et al., 2011). This warmingtrend in soil temperature has occurred most strongly during winterand spring (Hu and Feng, 2003; Mellander et al., 2005).

The changing soil temperature has had direct impacts on both soilbiotic process and soil hydrological cycle in the mid and high latitudesof the Northern Hemisphere: for example, soil respiration (Schlesingerand Andrews, 2000) and soil carbon decomposition (Davidson andJanssens, 2006; von Lützow and Kögel-Knabner, 2009). Soil freezing/thawing processes determine soil water availability in these snow-dominated regions. The timing of soil thawing impacts the amount ofsnowmelt infiltration (Barnett et al., 2005). Changes in soil hydrological

rights reserved.

processes during the snowmelt periodmay partially explain forestmor-tality (Huntington, 2006; Sheffield and Wood, 2008; McDowell et al.,2011). A site-level study in northern Mongolia has shown a clear differ-ence in soil temperature between the forest edge and open steppe,suggesting that vegetation cover plays a role in soil temperature regula-tion (Dulamsuren and Hauck, 2008). However, this kind of site-levelstudy is insufficient to model soil temperature dynamics under climatewarming. To make a future prediction, we need a large-scale studyalong a temperature gradient to enable a space-for-time-substitution,and the forest-steppe ecotone in Inner Asia with a latitudinal span ofmore than 10° and site-level forest-steppe coexistence that exactlymeet the requirement of this study (Liu et al., 2012).

In this study, soil temperature was monitored for paired samplesin the forest-steppe ecotone in Inner Asia along a temperature gradi-ent to distinguish the effect of vegetation cover on soil temperaturedynamics, which has received little attention in previous studies.We hypothesized that forest would buffer soil temperature and influ-ence on the timing of soil freezing and thawing processes. Besidesquantitatively comparing the soil thermal properties between forestand steppe, we focus on the relationships between onset dates offreezing/thawing and mean daily soil temperature in the cold seasonin order to explore the possible feedback between vegetation and soildynamics under the current climate warming.

Table 1Climate, altitude, dominating tree species and time-serial of each site.

Sites MAT(°C)

MAP(mm)

Altitude(m a.s.l.)

Dominating treespecies

Data time Interval(h)

Forest Steppe

ISH −4.5 387.0 1003 L. sibirica 2008.8.23–2010.6.24, 2010.10.16–2011.10.19 2008.8.23–2009.8.31, 2010.10.16–2011.10.19 1KAM −4.1 569.1 399 P. sylvestris 2010.8.17–2011.10.6 0.5KOM −3.5 385.5 1013 B. platyphylla 2008.8.23–2009.8.30, 2009.9.6–2010.10.13, 2010.10.20–

2011.10.182008.8.23–2009.8.10 0.5

SHE −3.5 536.2 1006 L. sibirica 2010.8.17–2011.10.6 0.5CHE −3.2 501.9 521 L. sibirica 2010.8.14–2011.10.5 2010.8.14–2011.10.1 0.5EEGN −3.1 382.3 709 B. platyphylla 2008.10.5–2009.7.16, 2009.10.28–2010.9.27,

2010.10.24–2011.10.92008.10.5–2009.7.16, 2009.10.28–2010.9.27,2010.10.24–2011.10.9

0.5

KUP −3.0 480.7 961 L. sibirica 2010.8.15–2011.5.30 2010.8.15–2011.10.6 0.5AMG −2.8 385.5 811 P. sylvestris 2008.8.23–2010.6.27, 2010.10.27–2011.10.18 2008.8.23–2010.6.27, 2010.10.27–2011.10.18 1DLE −2.2 383.3 917 B. platyphylla 2008.10.7–2009.7.19, 2009.10.30–2010.9.26,

2010.10.25–2011.10.102008.10.7–2009.7.19, 2009.10.30–2010.9.26,2010.10.25–2011.10.10

0.5

BQK 2.4 383.3 1607 B. platyphylla 2008.10.28–2009.9.11, 2009.9.13–2010.8.16,2010.10.27–2011.9.27

2008.10.28–2009.9.11, 2009.9.13–2010.8.16,2010.10.27–2011.10.13

0.5

HLH 3.5 368.9 999 B. platyphylla 2008.10.9–2009.7.20, 2009.10.31–2010.9.25,2010.10.26-2011.10.12

2009.10.31–2010.9.25, 2010.10.26–2011.10.12 0.5

2 G. Hu et al. / Global and Planetary Change 104 (2013) 1–6

2. Materials and methods

2.1. Study area

The sample area comprised four sites in the northern AltaiMountainregion, three sites in the Trans-Baikal region in Russia and four sites onthe edge of the Inner Mongolian Plateau in China (Fig. 1). All sites aredominated by typical aspect-dependent vegetation patterns with for-ests on the northern hill slopes and steppe on the southern slopes(Peshkova, 1985; Wallis de Vries et al., 1996; Dulamsuren et al., 2005;Chytrý et al., 2008). Microclimatic studies within similar forest-steppelandscapes were carried out in northern Mongolia (Beresneva, 1980,1983). During the growth period northern slopes receive much lesswarmth than southern ones, about 410–450 W/m2, and 600–700 W/m2

respectively (Beresneva, 1983). Also, albedo in coniferous forests is only55% of that in the grasslands (Beresneva, 1980). Paired steppe sampleand forest samples were selected to be as close as possible at each site.Forest samples were dominated by Larix sibirica, Larix gmelinii, Pinussylvestris and Betula platyphylla in this area. All sites were distributedalong a gradient of mean annual temperature ranging from −4.5 to3.5 °C (CRU TS 3.0, 1950–2006), but with similar mean annual precipi-tation totals of around 400 mm (Fig. 1; Table 1).

2.2. Soil temperature record

Soil temperatures at 10 cm depth were recorded hourly or halfhourly at each site in the interior of the forest (away from gaps)and steppe stands, respectively, using temperature data loggers(HOBO, Onset Computer Corporation, MA, USA), which were settledcarefully under the mineral soil block and covered with leaf litter.Loggers were installed in the soil in 2008, except for those in thenorthern Altai Mountains which were installed in 2010. Malfunction

Fig. 1. Site distribution, with mean annual temperature

of battery and possible wrong setting of the instrument before mea-surement also led to data hiatus for either forest or steppe sites. Inthe transition from summer to autumn of each year there is a shortperiod when the loggers were retrieved to allow data downloadingand battery checking and replacement, before reinstalling in thesoil. Therefore, there are some gaps in the data from each plot (Fig. 2).

2.3. Statistical analysis

Mean, maximum and minimum daily soil temperatures (DSTs) werecalculated using the original data. The onset date of soil thawing was de-fined as the daywhen theminimumDSTfinally exceeded 0 °C during theearly season. The onset date of soil freezing was calculated as the daywhen maximum DST finally decreased to below 0 °C in the late season.

To compare soil temperature between forest and steppe, a pairedt-test was performed to mean DST on each site between these two vege-tation covers in the soil warm season (from June to September) and soilcold season (from December to March) over all years of monitoring. Thedifference between the paired values is assumed to be normally distribut-ed, and the null hypothesis that the expectation is zero is tested by t-test.So paired t-test would present the differences which are influenced bytwo vegetation covers better by either to increase the statistical power,or to reduce the effects of confounders than in an ordinary unpaired test-ing situation. To show the influence of winter warming on soil freezingand thawing processes, linear regression was performed to analyze thecorrelation between the mean DST in the soil cold season and the onsetdate of soil freezing/thawing in both the forest and steppe.

Owing to gaps in the data, the analysis to DST only used intact datawhich contained no gap. And the analysis of soil freezing/thawing dateonly useddata out of gaps, too. Thedata gapswould reduce available sam-ple size. Analysis of quality would, however, paid no price to the datagaps.

and precipitation from 1950 to 2006 (CRU TS 3.0).

Fig. 2.Mean daily soil temperature (°C) of each plot was represented by the black line in the steppe and the red line in forest. Box plots show mean (square), median (middle line),and the 75% (top of box), 25% (bottom of box), 90% (top bar) and 10% (bottom bar) of the mean DST for June to September and December to March. Significance levels of the pairedt-tests are indicated by stars in the box plots (two stars for pb0.05, three stars for pb0.01). (For interpretation of the references to color in this figure legend, the reader is referred tothe web version of this article.)

3G. Hu et al. / Global and Planetary Change 104 (2013) 1–6

3. Results

3.1. Differences in soil temperature between forest and steppe

The study region had a distinct soil warm season (from June toSeptember) and soil cold season (from December to March) (Fig. 2).The mean DST in the steppe was distinctly higher than that in the

forest during the warm season, as the differences were significantlypositive between steppe and forest at all sites in the paired t-test(Fig. 2). The differences in mean DST were 4.00±0.39 °C to 7.94±0.20 °C between steppe and forest in the warm season. The meanDST values during the warm season were 16.05±0.27 °C to 20.32±0.23 °C in the steppe and 8.11±0.20 °C to 13.93±0.22 °C in theforest.

4 G. Hu et al. / Global and Planetary Change 104 (2013) 1–6

By contrast, an opposite temperature pattern was observed duringthe cold season. The differences in mean DST were significantly neg-ative between steppe and forest at all sites except the site HLH, asshown by the paired t-test (Fig. 2). The difference in mean DST was3.38±0.23 °C between steppe and forest in the cold season at HLH.The mean DST during the cold season was −8.53±0.27 °C in thesteppe and −11.91±0.22 °C in the forest at the site HLH. At othersites, the differences in mean DST were −5.20±0.17 °C to −0.17±0.07 °C between steppe and forest during the cold season. The meanDST of the cold season was −12.80±0.23 °C to −9.14±0.53 °C inthe steppe and −10.59±0.17 °C to −6.24±0.12 °C in the forest.

3.2. Onset date of soil freezing and thawing in the forest and steppe

Soil thawed half a month earlier in the steppe than in the forest:the difference in onset date of soil thawing was −15±3 days(Fig. 3a, 18 pairs, pb0.001) between the steppe and the forest in thepaired t-test. Meanwhile, soil thawed on Julian Day 107±3 in thesteppe and Julian Day 122±3 in the forest. However, soil froze inthe steppe (Julian Day 310±3) on nearly the same day as that inthe forest (Julian Day 307±3). The difference in onset date of soilfreezing between steppe and forest in the paired t-test was only3±1 days (Fig. 3b, 14 pairs, pb0.05).

Forest soil thawing responded insensitively to the observed soiltemperature, as the onset date of soil thawing had no significant cor-relation with mean DST of the cold season in the forest (p=0.90,Fig. 3c). Steppe soil, however, thawed earlier in warmer winters, asshown by the onset date of first soil thawing which was significantlynegatively correlated with mean DST of the cold season in the steppe(pb0.05, Fig. 3d). Onset date of soil freezing was significantly posi-tively correlated with mean DST of the cold season in both forest(pb0.05, Fig. 3e) and steppe (pb0.05, Fig. 3f). Also, the onset dateof soil freezing in the forest appeared to have a faster response tomean DST of the cold season, as indicated by the steeper slope of lin-ear regression in the forest than in the steppe (Fig. 3e, f).

3.3. Interannual difference of soil freezing and thawing

The general patterns of soil freezing and thawing processes wererevealed by the results obtained with the complete data set from2008 to 2011, while the data for each year showed several patterns.Only the onset date of soil thawing in the forest showed no significantcorrelation with mean DST in the cold season, when the whole dataset from 2008 to 2011 was used for linear regression. Similarly, nosignificant correlation was found between onset date of soil thawingin the forest and soil temperature when the linear regression wasperformed with data for each year. A data gap in the cold seasonfrom 2009 to 2010 resulted in the poor outcome of the statisticalanalysis (green triangles indicate 2010 in Fig. 3). In the steppe, thegeneral pattern of the soil thawing process was confirmed by the con-sistent trend between data from 2009 and 2011 (as shown by red cir-cles and blue squares in Fig. 3d). However, the general pattern of thefreezing timing seemed less robust at an annual scale, because onlyone year of data (2010 in the forest, 2008 in the steppe) showed a sig-nificant and consistent trend.

Fig. 3. Box plots showing mean (square), median (middle line), and the 75% (top of box),25% (bottom of box), 90% (top bar) and 10% (bottom bar) percentiles of soil thawing date(a) and freezing date (b). p-Values are represented by stars in the box plots (two stars forpb0.05, three stars for pb0.01). Also shown are the correlations between cold seasonmeanDST and (c) thawing date in forest; (d) thawingdate in steppe; (e) freezing date in for-est; and (f) freezing date in steppe. The black line represents the correlation with the totaldata set, labeled by the corresponding linear regression equation and R2. The red circles rep-resents the cold season from 2008 to 2009. The green triangles represents the cold seasonfrom 2009 to 2010. The blue squares represents the cold season from 2010 to 2011. Thethick line represents correlation with pb0.05, and the thin line represents correlation withpb0.1. (For interpretation of the references to color in this figure legend, the reader is re-ferred to the web version of this article.)

Fig. 4. The feedback between forest and the soil temperature and soil hydrological pro-cess. Red arrows show positive effects; blue arrows show negative effects; and the dot-ted line showed no significant effect. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

5G. Hu et al. / Global and Planetary Change 104 (2013) 1–6

4. Discussion

4.1. Buffer effect of forest on soil temperature

Our results showed that forest buffered the soil temperature byreducing soil temperature during the warm season and raising it dur-ing the cold season. However, topographic effects cannot be ignored.In paired samples, the steppe on sunny slopes received more radia-tion than that of the forest (Beresneva, 1983). In our study region,the differences in mean DST between steppe and forest were of oppo-site signs in the cold season and warm season at all sites except HLH.This opposite trend was accounted for more by vegetation cover thanby topography. Forest soil received less solar radiation, due to theshadier aspect combined with higher albedo of the canopy(Henderson-Sellers and Wilson, 1983; Liang et al., 2003). Althoughthe steppe benefitted from receiving more solar radiation because ofits distribution on sunny slope aspects, the forest soil was warmerduring the cold season at most sites due to thick litter and slowlymelting snow (Wieringa, 1992), which both provide efficient thermalinsulation. This buffering effect of forest on soil temperature mightalter soil processes such as soil respiration, which further contributesto forest ecosystem dynamics (Schlesinger and Andrews, 2000;Davidson and Janssens, 2006; von Lützow and Kögel-Knabner, 2009).

4.2. Soil freezing and thawing timing in the forest and steppe

The timing of soil thawing was clearly delayed in the forest rela-tive to that in steppe, while soil freezing occurred at nearly thesame time in the paired plots. The decrease in solar radiation couldhave caused the timings of soil freezing to be very close in the steppeand the forest, when snow cover had not formed. Meanwhile, thermalinsulation, especially snow cover, caused the soil to thaw nearly15 days later in the forest than in the steppe. A similar pattern inthe soil thawing process was found on the boreal plain of northern Al-berta, Canada, with soil (0–40 cm depth) thawing earlier in harvestedsites than in forested sites (Whitson et al., 2005).

This difference in timing of soil thawing was associated with snowmelt. In Minnesota, forest harvesting resulted in greater and earliersnowmelt peak runoff (Verry et al., 1983), while snow melted earlierat sites with more southern aspects than those with more northernaspects in a montane meadow near the Rocky Mountain BiologicalLaboratory in Colorado, USA (Roy et al., 2004). Because snow provideseffective thermal insulation, and harvested forest has similar vegeta-tion cover characteristics to those of steppe, it could be inferred thatthe later snow melt in the forest led to a delayed soil thawing.

Although we have observed a difference in soil thawing betweenforest and steppe, the large interannual differences suggested thatwe need a longer observation for a better understanding of the soilfreezing/thawing processes.

4.3. Enhanced drought threatens forest

Based on the above results, we hypothesized that drought riskwould be greater in the forest. Firstly, soil in the forest experienced alonger cold season drought than that in the steppe. This was inducedby the longer duration of loss in available water, because of the similaronset of soil freezing in both environments but a half-month delay insoil thawing in forest. Secondly, forest soil might receive less snowmeltinfiltration under climate warming, with feedback between the forest,soil temperature and soil hydrological processes shown in Fig. 4. Awarming trend in observed soil temperature has been reported in midand high latitudes of the Northern Hemisphere (corresponding to theincrease of soil temperature in Fig. 4), e.g. in northeast China andnorth China (Wang et al., 2009), the U.S. (Hu and Feng, 2003) andnorthern Sweden (Mellander et al., 2005). In addition, a trend towardsearlier snow melt has been reported in many areas (Dettinger and

Cayan, 1995; Stone et al., 2002; Stewart et al., 2004; Chapin et al.,2005). Since the frozen soil impedes infiltration of snowmelt (Kaneand Chacho, 1990; Hardy et al., 2001; Barnett et al., 2005; Iwata et al.,2010), the insensitive response of thawing time of the forest to soil tem-perature change would enhance drought risk by extending the effect ofimpeded snowmelt infiltration under a warming trend in soil tempera-ture (as illustrated by the increasing lag in Fig. 4). Steppe would face amuch less serious drought risk or even better soil water conditions (aweaker increasing trend or even decrease of lagged days in Fig. 4), asits soil would thaw earlier if thewinter becomeswarmer. The enhancedseverity and duration of drought risk would be an important climaticfactor causing forest dieback (Allen, 2009). Further, the buffer effect offorest towards soil temperature would decrease following any forestdieback.

The feedback model described in Fig. 4, however, remains a con-ceptual one, and there are still many uncertainties, for example, themechanism of weak relationship between soil temperature and soilthawing date. Such factors as the heat produced by different microbesmight account for the different responses between forest and steppesoils (Khvorostyanov et al., 2008).

5. Conclusions

In this study, we found that forest had a significant buffering effecton soil temperature, e.g. the forest reduced soil temperature duringthe warm season while raising it during the cold season. Forestcover could lead to delayed soil thawing in the forest when comparedwith corresponding steppe sites. Under the trend of soil warming, thesoil freezing process would be delayed in both the forest and steppe,while the soil thawing process would be advanced in the steppe butremain stable in the forest. This implies that the stable thawingtime of the forest would enhance drought risk by extending the peri-od of impeded snowmelt infiltration under a warming trend in soiltemperature, which could constitute an important climatic factorcausing forest dieback in this snow-dominated area. Further observa-tions of hydrological processes are now required.

Acknowledgments

This study was granted by the National Natural Science Founda-tion of China (No. 41071124, 41011120251) and Russian Foundationof Basic Research (No. 13-04-91180). We thank Natalya Badmaevaand Yuke Zhang for the assistance in the field work, and Dave Chan-dler for editing the text.

6 G. Hu et al. / Global and Planetary Change 104 (2013) 1–6

Reference

Allen, C., 2009. Climate-induced forest dieback: an escalating global phenomenon.Unasylva 231/232, 43–49.

Barnett, T.P., Adam, J.C., Lettenmaier, D.P., 2005. Potential impacts of a warming climateon water availability in snow-dominated regions. Nature 438, 303–309.

Beresneva, I.A., 1980. Teplovoi balans podstilayushchei poverkhnosti territoriiMongol'skoi Narodnoi Respubliki (The warming balance at reflecting surfaces inMongolian People's Republic). Proceedings of Voeikov Geophysical Observatory441, 117–126 (in Russian).

Beresneva, I.A., 1983. Klimat (Climate). In: Lavrenko, E.M., Sokolov, V.E. (Eds.), Gornayalesostep' Vostochnogo Khangaya: Biologicheskye resursy I prirodnye usloviyaMongol'skoi Narodnoi Respubliki. (Mountain Forest-Steppe in the EasternKhangai Ridge: Biological Resources and Natural Conditions of Mongolian People'sRepublic). Nauka, Moscow, pp. 32–39 (in Russian).

Chapin III, F.S., Sturm, M., Serreze, M.C., McFadden, J.P., Key, J.R., Lloyd, A.H., McGuire,A.D., Rupp, T.S., Lynch, A.H., Schimel, J.P., Beringer, J., Chapman, W.L., Epstein,H.E., Euskirchen, E.S., Hinzman, L.D., Jia, G., Ping, C.-L., Tape, K.D., Thompson,C.D.C., Walker, D.A., Welker, J.M., 2005. Role of land-surface changes in Arctic sum-mer warming. Science 310, 657–660.

Chytrý, M., Danihelka, J., Ermakov, N., Hájek, M., Hájková, P., Kočí, M., Kubešová, S.,Lustyk, P., Otýpková, Z., Popov, D., Roleček, J., Řezníčková, M., Šmarda, P.,Valachovič, M., 2008. Diversity of forest vegetation across a strong gradient of cli-matic continentality: Western Sayan Mountains, southern Siberia. Plant Ecology196, 61–83.

Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomposi-tion and feedbacks to climate change. Nature 440, 165–173.

Dettinger, M.D., Cayan, D.R., 1995. Large-scale atmospheric forcing of recent trends to-ward early snowmelt runoff in California. Journal of Climate 8, 606–623.

Dulamsuren, C., Hauck, M., 2008. Spatial and seasonal variation of climate on steppeslopes of the northern Mongolian mountain taiga. Grassland Science 54, 217–230.

Dulamsuren, C., Hauck, M., Muhlenberg, M., 2005. Vegetation at the taiga forest-steppeborderline in the western Khentey Mountains, northern Mongolia. AnnalesBotanici Fennici 42, 411–426.

Hardy, J.P., et al., 2001. Snow depth manipulation and its influence on soil frost andwater dynamics in a northern hardwood forest. Biogeochemistry 56, 151–174.

Henderson-Sellers, A., Wilson, M., 1983. Surface albedo data for climatic modeling. Re-views of Geophysics 21, 1743–1778.

Hu, Q., Feng, S., 2003. A daily soil temperature dataset and soil temperature climatologyof the contiguous United States. Journal of Applied Meteorology 42, 1139–1156.

Huntington, T.G., 2006. Evidence for intensification of the global water cycle: reviewand synthesis. Journal of Hydrology 319, 83–95.

Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L.(Eds.), 2007. Contribution of Working Group I to the Fourth Assessment Report ofthe Intergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, UK.

Iwata, Y., Hayashi, M., Suzuki, S., Hirota, T., Hasegawa, S., 2010. Effects of snow cover onsoil freezing, water movement, and snowmelt infiltration: a paired plot experi-ment. Water Resources Research 46, W09504.

Jacobs, A.F.G., Heusinkveld, B.G., Holtslag, A.A.M., 2011. Long-term record and analysisof soil temperatures and soil heat fluxes in a grassland area, The Netherlands. Ag-ricultural and Forest Meteorology 151, 774–780.

Kane, D., Chacho, E., 1990. Frozen ground effects on infiltration and runoff. In: Ryan,W.L., Crissmann, R.D. (Eds.), Cold Regions Hydrology and Hydraulics. ASCE, Techni-cal Council on Cold Regions Engineering Monograph, New York, pp. 259–300.

Khvorostyanov, D.V., Ciais, P., Krinner, G., Zimov, S.A., 2008. Vulnerability of eastSiberia's frozen carbon stores to future warming. Geophysical Research Letters35, L10703. http://dx.doi.org/10.1029/2008GL033639.

Liang, S., Shuey, C.J., Russ, A.L., Fang, H., Chen, M., Walthall, C.L., Daughtry, C.S.T., Hunt,R., 2003. Narrowband to broadband conversions of land surface albedo: II. Valida-tion. Remote Sensing of Environment 84, 25–41.

Liu, H., He, S., Anenkhonov, O., Hu, G., Sandanov, D., Badmaeva, N., 2012. Topography-controlled soil water content and the coexistence of forest and steppe in northernChina. Physical Geography 33, 561–573.

McDowell, N.G., Beerling, D.J., Breshears, D.D., Fisher, R.A., Raffa, K.F., Stitt, M., 2011. Theinterdependence of mechanisms underlying climate-driven vegetation mortality.Trends in Ecology & Evolution 26, 523–532.

Mellander, P.E., Laudon, H., Bishop, K., 2005. Modelling variability of snow depths andsoil temperatures in Scots pine stands. Agricultural and Forest Meteorology 133,109–118.

Peshkova, G.A., 1985. Rastitel'nost' Sibiri: Predbaikal'e i Zabaikal'e (Vegetation ofSiberia: Cis- and Transbaikal Region). Nauka, Novosibirsk (in Russian).

Roy, B.A., Güsewell, S., Harte, J., 2004. Response of plant pathogens and herbivores to awarming experiment. Ecology 85, 2570–2581.

Schlesinger, W.H., Andrews, J.A., 2000. Soil respiration and the global carbon cycle. Bio-geochemistry 48, 7–20.

Sheffield, J., Wood, E.F., 2008. Global trends and variability in soil moisture and droughtcharacteristics, 1950–2000, from observation-driven simulations of the terrestrialhydrologic cycle. Journal of Climate 21, 432–458.

Stewart, I.T., Cayan, D.R., Dettinger, M.D., 2004. Changes in snowmelt runoff timing inwestern North America under a business as usual' climate change scenario. Climat-ic Change 62, 217–232.

Stone, R.S., Dutton, E.G., Harris, J.M., Longenecker, D., 2002. Earlier spring snowmelt innorthern Alaska as an indicator of climate change. Journal of Geophysical Research107 (D10). http://dx.doi.org/10.1029/2000JD000286.

Verry, E.S., Lewis, J.R., Brooks, K.N., 1983. Aspen clearcutting increases snowmelt andstorm flow peaks in North Central Minnesota. Journal of the American Water Re-sources Association 19, 59–67.

von Lützow, M., Kögel-Knabner, I., 2009. Temperature sensitivity of soil organic matterdecomposition—what do we know? Biology and Fertility of Soils 46, 1–15.

Wallis de Vries, M., Manibazar, N., Dügerlham, S., 1996. The vegetation of the forest-steppe region of Hustain Nuruu, Mongolia. Plant Ecology 122, 111–127.

Wang, X., Guo, W., Zhong, Z., Cui, X., 2009. Long term trends of soil moisture and tem-perature change in East China in relationship with climate background. AdvancedEarth Sciences 24, 181–191 (in Chinese with English abstract).

Whitson, I., Chanasyk, D., Prepas, E., 2005. Effect of forest harvest on soil temperatureand water storage and movement patterns on boreal plain hillslopes. Journal of En-vironmental Engineering and Science 4, 429–439.

Wieringa, J., 1992. Updating the Davenport roughness classification. Journal of WindEngineering and Industrial Aerodynamics 41, 357–368.