Observed air/soil temperature trends in open land and understory of a subtropical mountain forest, SW China

Authors

  • Guangyong You,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Graduate University of the Chinese Academy of Sciences, Beijing 100049, China
    Search for more papers by this author
  • Dr. Yiping Zhang,

    Corresponding author
    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    • Kunming Section, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, 88 Xuefu Road, Kunming, Yunnan 650223, China.
    Search for more papers by this author
  • Douglas Schaefer,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    Search for more papers by this author
  • Liqing Sha,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    Search for more papers by this author
  • Yuhong Liu,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    Search for more papers by this author
  • Hede Gong,

    1. Faculty of Ecotourism, Southwest Forestry University, Kunming 650224,China
    Search for more papers by this author
  • Zhenghong Tan,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    Search for more papers by this author
  • Zhiyun Lu,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    Search for more papers by this author
  • Chuansheng Wu,

    1. Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Menglun, Mengla, Yunnan 666303, China
    2. Ailaoshan Station for Subtropical Forest Ecosystem Studies, Chinese Ecosystem Research Network, Jingdong, Yunnan 676209 China
    3. National Forest Ecosystem Research Station at Ailaoshan, Jingdong, Yunnan 676209, China
    Search for more papers by this author
  • Youneng Xie

    1. Jingdong Bureau of National Nature Reserve, Jingdong, Yunnan 676209, China
    Search for more papers by this author

Abstract

This study seeks a further understanding on climate trends in a subtropical mountain forest, SW China. Air (Ta) and soil temperature (Ts), both in open land (1983–2010) and under a forest canopy (1986–2010), were investigated. Short-term radiation components were also measured simultaneously both in open land and understory to explore the relationships of microclimatic variables. Correlations of Ta and Ts with sunshine hours (St) and wind speed (Ws) were also analysed as driving factors of the temperature trends.

The results showed that (1) Understory radiation components were greatly reduced by the forest canopy, showing a strong effect of forest canopy on microclimatic variables. Ts_0 in open land was significantly correlated with solar radiation. Wind speed had significant influences on differences between Ta and Ts_0, between open land Ts_0 and understory Ts_0. The long-term data showed that Ts_0 under forest canopy were closely coupled with Ta in open land. (2) Ta had a larger increase than Ts_0 in open land, and temperature increases in winter were greater than in other seasons. Soil temperature at depths under forest canopy had nearly twice the increases of those on open land; we attributed this to the higher relative increase of Ws over St. (3) A slope change in 1998 was detected in the Ts_0 and Ta difference (Ts_0 − Ta) series, suggesting different response of Ts_0 and Ta since that year. Deceleration of St and stability of Ws may have been factors.

This study improves our understanding of warming in a nature reserve where anthropogenic influences are absent. Further studies are needed for the biological and biochemical implications on subtropical mountain forest. Copyright © 2012 Royal Meteorological Society

1. Introduction

Over the past century, global average air temperatures have increased by 0.07 °C per decade (1901 ∼ 2005), with radiative-transfer models projecting a further 1.8–4.0 °C increase by 2100 (IPCC, 2007). In China, air temperature has increased by 0.27 °C per decade from 1961 to 2003 with accelerating trends after 1990 (Liu et al., 2004). However, few studies have examined corresponding trends in soil temperature (Jacobs et al., 2011).

Soil temperature controls the biological and ecological processes through the limitations on plant growth and forest distribution (Körner and Paulsen, 2004). Soil respiration, which has a substantial effect on atmospheric carbon cycling, is sensitive to changes in soil temperature (Bond-Lamberty and Thomson, 2010). As soil temperature may differ from air temperature, exploring trends in both soil and air temperature could improve our understanding of warming and its biological and biochemical consequences.

Few long-term soil temperature, trends have been reported, with most from open land meteorological stations (García-Suárez and Butler, 2006; Subedi and Fullen, 2009). However, those trends may not well represent those from under forest canopies. Previous studies comparing forest and open land temperatures were based on short data series and focused on temperature extremes (Morecroft et al., 1998; Paul et al., 2004; Renaud and Rebetez, 2009). Long-term temperature trends under forest canopies have been examined in one study of temperature extremes under forest cover with different elevations, aspects and forest types (Ferrez et al., 2011). Consequently, comparing temperature trends under forest canopies with those more commonly measured in open lands should improve our understanding of warming in forested areas.

This study took place in a mountain forest area south-west China and addresses the following questions: (1) What are the microclimatic differences between open land and under forest canopy? (2) What is the pattern of air/soil temperature trends both in open land and under forest canopy? (3) When does soil temperature respond differently than air temperature in open land?

2. Study site and methods

2.1. Study site

The Ailaoshan Station for Subtropical Forest Ecosystem Studies (ASSFE, 24°32′N, 101°01′E and 2480 m asl.) is located in Jingdong County, Yunnan Province and the northern part of Ailaoshan Natural Reserve (Figure 1). Subtropical montane evergreen broad-leaved forest (dominated by Castanopsis wattii and Lithocarpus xylocarpus) persists in Ailaoshan Natural Reserve. The strata of this forest include canopy (18–25 m), shrub (1–3 m) and herb layers (<0.5 m) (Qiu and Xie, 1998). Basal area is 91 m2 ha−1, leaf area index (LAI) varies from 4.8 to 6.5 through the year and tree density is 2728 individuals ha−1 (Schaefer et al., 2009). The forest ecosystem is free of management with a stand age > 300 years (Tan et al., 2011). This mountain forest is characterized by shortage of growing season warmth and high solar radiation (Fang et al., 1996; Qiu and Xie, 1998; Gao et al., 2009). The soil is loamy clay, with soil volumetric water content rarely falling below 35% in the upper 50 cm (Gong et al., 2011).

Figure 1.

Location of the study site and the topography of the meteorological stations of open land (open land) and under forest canopy (forest)

In an area without forest cover, a standard meteorological observation station (open land) was established in 1981. The shortest distance between this station and the forest edge was about 200 m. Another meteorological station was established under the forest canopy without disturbing the forest structure. The shortest distance from the under forest canopy meteorological station to the forest edge was about 300 m. Both of these meteorological stations are at the same altitude and similar aspects (Figure 1). In the recent three decades, no tree mortality or gap-forming events have taken place near the understory meteorological station. Stable litterfall shows that the LAI in this study site has been constant at least since 1991 (Liu et al., 2002; Eriksson et al., 2005; Schaefer et al., 2009).

2.2. Meteorological observations

In this study, we utilized temperature observations in open land (1983–2010) and under forest canopy (1986–2010). Air and soil temperature observations, in both settings, followed procedures of the China Meteorological Administration. Air temperature (Ta) was measured with a mercury thermometer inside a standard radiation shelter at 1.5 m height. Soil temperatures (Ts) were measured by bent stem mercury thermometers at depths of 0, 5, 10, 15 and 20 cm (abbreviated as Ts_0, Ts_5, Ts_10, Ts_15 and Ts_20, respectively). Thermometers located at 0 cm were placed on the mineral soil surface in open land and on the organic layer under the forest canopy. Litterfall on the thermometer under forest canopy was periodically removed to keep the surface thermometer exposed to the air. Daily mean temperatures were averaged from measurements at 02:00, 08:00, 14:00 and 20:00 China time. Monthly mean temperature is averaged from those daily mean temperatures. Seasonal temperatures were averaged over three months (Spring is March to May, Summer is June to August, Autumn is September to November and Winter is December to the next February). Throughout the measurement periods, equipment, procedures and sampling locations have not been changed.

For relationship of the Ta and Ts trends, sunshine hours (St) and wind speed (Ws) records in open land meteorological station were used for correlation analysis. St was observed by dark-tube sunshine hour recorder. China Meteorological Administration (2003) provided the detailed descriptions of instrument and measurement. Ws was recorded at 10 m height. Daily Ws were averages from observations at 08:00, 14:00 and 20:00 China time. Monthly mean St and Ws were averaged from those daily values.

2.3. Radiation budgets in open land and under forest canopy

To explore the radiation budgets of open land and under forest canopy, we took advantage of the automatic recording meteorological observation system (located in the open land). Global radiation (Q) and reflective radiation (Qa) were recorded by CM11 (Kipp & Zonen, Delft, The Netherlands). Net radiation (Rn) was measured by CNR1 (Kipp & Zonen, Delft, The Netherlands). Soil heat flux (G) was measured with HFP01 (Hukseflux, Delft, The Netherlands).

We made four (January, April, July and October) short-term micrometeorological campaigns in the under forest canopy area. Each observation continued for 11 or more days (12 d in January, 17 d in April, 11 d in July and 12 d in October). Net radiation (Rn), together with downward/upward short/long wave radiation, was measured by CNR1 (Kipp & Zonen, Delft, The Netherlands), and soil heat flux (G) was measured by HFP01 (Hukseflux, Delft, The Netherlands). Simultaneous measurements showed that the ratios of daily Q to downward short wave radiation were 1.055 (n = 30, R2 = 0.983) in January, 1.057 (n = 30, R2 = 0.986) in April, 1.144 (n = 30, R2 = 0.991) in July and 1.069 (n = 30, R2 = 0.952) in October. Q and Qa under forest canopy were computed by the measured downward/upward short wave radiation and the coefficients of each month. Effective radiation (I) was computed by subtracting Qa and Rn from Q. Downward radiation components were defined as positive, and upward radiation components were defined as negative.

2.4. Trend analysis

Trends in air and soil temperatures were analysed with the Mann–Kendall test (Mann, 1945; Hamed, 2008). Mann–Kendall's tau values and their significances were calculated by the ‘Kendall’package (McLeod, 2009) in the R environment (R Development Core Team, 2010). Positive tau values indicated an increasing trend.

Serial autocorrelation could influence the trend significance detected by Mann–Kendall test (von Storch, 1995). A pre-whitening procedure (MK-TFPW) was applied here to reduce the influence of autocorrelation on the significance of Mann–Kendall test results (Yue et al., 2002). Then, the Mann–Kendall test was applied to pre-whitened temperature series.

Long-term trends are computed by simple linear regression (SLR) y = a1t + a0, where t is the time (season/years), y is the temperature in season/year and a1 is the slope of temperature trend.

To clarify the relationship between soil and air temperature, these temperature series were standardized to dimensionless indices (Equation 1, Equation 2). Using AnClim software (Štěpánek, 2008), ten years low-pass-filter lines were built for the dimensionless time series. To reveal any abrupt changes in the temperature series, we built a soil–air temperature difference series. The Mann–Whitney–Pettit test was conducted on the soil–air temperature difference series (Pettitt, 1979). The Mann–Whitney–Pettit test was calculated with AnClim software (Štěpánek, 2008).

equation image(1)
equation image(2)

Yi is air/soil temperature series. Ȳ is the averaged value of air/soil temperature series.

3. Results

3.1. Microclimatic difference between open land and under forest canopy

Table I lists the daytime and night-time radiation budgets. During daytime, understory Q was < 5% of Q in open land (3.1% in January, 4.1% in April, 3.7% in July and 2.3% in October). Daytime Ta and Ts_0 in open land were much higher than those under forest canopy; however, their differences are small at night, especially for Ta in these two sites. Correlation analysis on daily records showed that Q in open land had significant correlation with Ts_0 (P < 0.05), wind speed had significant correlations with the difference between Ts_0 and Ta (P < 0.001) in open land and Ts_0 difference between open land and under forest canopy had weak correlation with the concurrent wind speed (P < 0.1).

Table I. Comparison of radiation components (MJ m−2 d−1) and temperatures ( °C) between open land and under forest canopy
  Open landForest
  JanuaryAprilJulyOctoberJanuaryAprilJulyOctober
  1. Daily average and standard error were computed by short-term observations in each season (12 d in January, 17 d in April, 11 d in July, 12 d in October). Daytime is the period of 7:00–19:00 local time, and night-time is the rest of the period in the days.

 Q9.76 ± 1.8719.55 ± 1.8310.71 ± 1.378.71 ± 1.740.27 ± 0.060.85 ± 0.100.43 ± 0.170.15 ± 0.02
 Qa− 2.04 ± 0.38− 3.66 ± 0.33− 2.12 ± 0.24− 1.79 ± 0.36− 0.07 ± 0.02− 0.17 ± 0.02− 0.05 ± 0.01− 0.03 ± 0.00
 I− 2.66 ± 0.57− 4.85 ± 0.44− 3.36 ± 1.19− 2.43 ± 0.47− 0.08 ± 0.060.07 ± 0.02− 0.05 ± 0.020.03 ± 0.04
DaytimeRn5.06 ± 0.9910.08 ± 1.505.18 ± 1.075.25 ± 1.190.11 ± 0.040.71 ± 0.080.15 ± 0.050.13 ± 0.04
 G0.48 ± 0.170.85 ± 0.160.89 ± 0.210.72 ± 0.20− 0.11 ± 0.040.13 ± 0.020.11 ± 0.04− 0.07 ± 0.04
 Ta5.14 ± 0.6215.74 ± 0.6916.18 ± 0.4511.18 ± 0.364.02 ± 0.4513.54 ± 0.5214.35 ± 0.249.92 ± 0.48
 Ts_09.48 ± 0.7920.40 ± 0.9620.37 ± 0.6314.51 ± 0.294.33 ± 0.2412.40 ± 0.2812.77 ± 0.1912.97 ± 0.02
 I− 1.21 ± 0.31− 1.75 ± 0.15− 0.75 ± 0.10− 1.22 ± 0.33− 0.12 ± 0.02− 0.02 ± 0.02− 0.05 ± 0.00− 0.12 ± 0.03
 G− 0.70 ± 0.10− 0.94 ± 0.07− 0.71 ± 0.05− 0.77 ± 0.15− 0.23 ± 0.03− 0.10 ± 0.02− 0.09 ± 0.01− 0.21 ± 0.06
Night-timeTa3.46 ± 0.6111.00 ± 0.4414.15 ± 0.218.21 ± 1.133.60 ± 0.3711.69 ± 0.3613.90 ± 0.218.99 ± 0.85
 Ts_05.67 ± 0.5112.72 ± 0.2617.18 ± 0.2710.90 ± 0.774.25 ± 0.2511.73 ± 0.2413.84 ± 0.1912.97 ± 0.02

Long-term averages of Ta and Ts_0, both in the open land and under forest canopy, showed unimodal patterns of annual variation (Figure 2). In the open land, average Ts_0 was 2.3 °C higher than Ta, and Ts_0 was 0.2 °C higher than Ta under forest canopy. Further, Ts_0 in the open land was 2.4 °C higher than that under the forest canopy. Average soil temperatures increased with depth with a rate of 0.17 °C per 10 cm in open land and 0.07 °C per 10 cm under forest canopy.

Figure 2.

The average annual variation of Ta and Ts_0 in open land (Open, data from 1983 to 2010) and under forest canopy (Forest, data from 1986 to 2010). ○, Ta; ▴, Ts_0

3.2. Air/soil temperature trends in both sites

As the time series of Ta and Ts at all depths had low serial autocorrelations, the Mann–Kendall trend test of pre-whitened series showed similar results to the original Mann–Kendall trend test (Table II). Both open land and under forest canopy showed significant positive trends in Ta. Ts at all depths had significant positive trends under forest canopy. In open land, Ts trends of all depths were positive, but trends in Ts_0, Ts_5 and Ts_10 were insignificant and trends in Ts_15, Ts_20 were significant.

Table II. Trend analysis of Ta and Ts at all depths in both open land and under forest canopy with ‘MK test original’, ‘TFPW MK test’ and ‘SLR’ means the tau value, the tau value with pre-whitened transformation and simple linear regression model, respectively
 Open landUnder forest canopy
 MK test Original tauTFPW MK taua1 (SLR) and 95% confidence interval ( °C/decade)MK test Original tauTFPW MK taua1 (SLR) and 95% confidence interval ( °C/decade)
  • ***

    , two sides significance < 0.01;

  • **

    , two sides significance < 0.05;

  • *

    , two sides significance < 0.1.

 Ta0.25*0.23*0.24*0.150.190.20
   (−0.03, 0.52)  (−0.13, 0.52)
 Ts_00.090.070.110.180.220.22
   (−0.34, 0.55)  (−0.05, 0.50)
 Ts_50.020.030.120.26*0.25*0.22*
   (−0.26, 0.469)  (−0.05, 0.50)
Spring
 Ts_10− 0.03− 0.010.070.40***0.41***0.34**
   (−0.30, 0.45)  (0.07, 0.60)
 Ts_150.030.060.160.32**0.31**0.26*
   (−0.22, 0.54)  (−0.01, 0.53)
 Ts_200.080.090.220.24*0.24*0.20
   (−0.20, 0.64)  (−0.07, 0.48)
 Ta0.49***0.50***0.21***0.26*0.55***0.13**
   (0.11, 0.31)  (0.01, 0.25)
 Ts_0− 0.020.080.090.24*0.33**0.13**
   (−0.12, 0.39)  (0.01, 0.25)
 Ts_50.060.110.030.160.220.09
   (−0.04, 0.30)  (−0.06, 0.24)
Summer
 Ts_100.020.04− 0.010.34**0.26*0.16**
   (−0.19, 0.18)  (0.04, 0.28)
 Ts_150.02− 0.040.010.24*0.25*0.12*
   (−0.17, 0.19)  (−0.02, 0.25)
 Ts_200.12− 0.020.050.140.060.08
   (−0.13, 0.24)  (−0.07, 0.22)
 Ta0.31**0.43***0.33**0.180.38***0.23
   (0.08, 0.58)  (−0.07, 0.52)
 Ts_0− 0.12− 0.13− 0.090.220.220.26*
   (−0.42, 0.25)  (−0.02, 0.55)
 Ts_50.040.040.070.33**0.37**0.32**
   (−0.24, 0.38)  (0.07, 0.58)
Autumn
 Ts_10− 0.03− 0.050.050.37**0.38***0.33***
   (−0.22, 0.32)  (0.10, 0.57)
 Ts_150.090.080.080.34**0.220.33**
   (−0.17, 0.34)  (0.08, 0.59)
 Ts_200.080.070.090.43***0.32**0.43***
   (−0.18, 0.37)  (0.12, 0.74)
 Ta0.44***0.59***0.63***0.41***0.43***0.55**
   (0.26, 1.00)  (0.13, 0.97)
 Ts_00.110.030.160.26*0.180.33
   (−0.29, 0.61)  (−0.07, 0.74)
 Ts_50.32**0.32**0.32*0.40***0.27*0.47**
   (0.03, 0.67)  (0.12, 0.82)
Winter
 Ts_100.27**0.200.30*0.37**0.26*0.41**
   (−0.04, 0.64)  (0.07, 0.74)
 Ts_150.23*0.160.33*0.30**0.220.38**
   (−0.03, 0.68)  (0.03, 0.74)
 Ts_200.200.140.33*0.28*0.35**0.46**
   (−0.04, 0.71)  (0.03, 0.89)
 Ta0.48***0.45***0.36***0.39***0.44***0.27***
   (0.18, 0.53)  (0.08, 0.47)
 Ts_00.100.180.070.34**0.200.24**
   (−0.18, 0.31)  (0.05, 0.43)
 Ts_50.220.210013**0.39***0.220.28***
   (0.08, 0.35)  (0.09, 0.46)
Year
 Ts_100.190.190.100.47***0.32**0.31***
   (−0.10, 0.31)  (0.14, 0.48)
 Ts_150.25*0.23*0.150.43***0.29**0.27***
   (−0.05, 0.34)  (0.10, 0.44)
 Ts_200.25*0.24*0.17*0.40***0.28*0.29***
   (−0.03, 0.38)  (0.10, 0.49)

Mean annual Ta increased more rapidly than that of Ts in open land, and the Ta increase was higher in the open land (0.36 °C/decade) than under forest canopy (0.27 °C/decade). However, the annual mean Ts_0 slightly increased in the open land (0.07 °C/decade), which was much less than that under the forest canopy (0.24 °C/decade). Ts increases at all depths under forest canopy were nearly double those in the open land. Both Ta and Ts at all depths showed higher increases in winter than in other seasons (Table II).

3.3. Different response between soil temperature and air temperature

Figure 3 shows the standardized temperature series of open land Ta and Ts_0 in each season and the whole year. Low-pass filtered and standardized Ta series was similar to Ts_0 in spring and winter. In summer and autumn, however, this standardized Ts_0 has responded differently than that of Ta. The Mann–Whitney–Pettit test showed that the year of 1998 was a change point in the soil–air temperature difference series (P < 0.05), which suggests that Ts_0 has responded differently than Ta since that year.

Figure 3.

Standardized time series of Ta and Ts_0 in each season and the whole year (average) of open land. Thin dotted lines were the standardized time series of Ta (open circle) and Ts (open triangle). Bold lines were ten-year low-pass filter of Ta (solid line) and Ts (dashed line). ○, Ta; ▵, Ts_0

Correlation analysis of long-term data sets showed significant positive correlations between St and Ts_0 and between Ws and Ta (Table III). Deceleration of St was seen after 1998 and Ws showed stable significant positive trends both before and after 1998 (Table IV).

Table III. Pearson correlation analysis on the standardized time series of Ta, Ts_0 and their difference (Ts_0 − Ta) with standardized St and Ws
 Standardized variablesOpen landUnder forest canopy
  TaTs_0Ts_0 − TaTaTs_0Ts_0 − Ta
  • ***

    , two sides significance < 0.01;

  • **

    , two sides significance < 0.05;

  • *

    , two sides significance < 0.1.

SpringSt0.434**0.3340.1060.404*0.392*− 0.174
 Ws0.188− 0.030− 0.252− 0.003− 0.080− 0.299
SummerSt0.1410.576***0.506**− 0.045− 0.143− 0.266
 Ws0.1760.042− 0.0310.024− 0.071− 0.269
AutumnSt− 0.0330.1840.2360.0380.011− 0.092
 Ws0.223− 0.213− 0.453**0.2410.2430.029
WinterSt0.424**0.321− 0.1630.2700.173− 0.233
 Ws0.516**− 0.021− 0.704***0.135− 0.072− 0.501**
AverageSt0.1800.392*0.259− 0.0220.0000.116
 Ws0.398*− 0.118− 0.533***0.061− 0.007− 0.295
Table IV. Linear trends of Ta, Ts_0, St and Ws in before-1998 (Bf), after-1998 (Af) and the entire periods (1983–2010)
 St (h·10−1)Ws (m ·s−1·10a−1)Ta( °C·10a−1)Ts_0 ( °C·10a−1)
 Bf/Af1983–2010Bf/Af1983–2010Bf/AfBf/Af
  • ***

    , two sides significance < 0.01;

  • **

    , two sides significance < 0.05;

  • *

    , two sides significance < 0.1.

Spring22.02/4.681.061.20*/0.93***0.50***0.09/− 0.10− 0.39/− 0.45
Summer7.58/− 0.555.990.88**/1.28***0.35**0.03/0.070.44/0.13
Autumn25.54/12.216.251.04***/1.11***0.40***0.34/− 0.100.41/− 0.57
Winter5.88/19.0010.98**1.28***/1.22***0.65***0.26/− 0.19− 0.74/− 0.68
Average15.25/8.836.071.10***/1.14***0.48***0.18/− 0.08− 0.07/− 0.39

4. Discussion

4.1. Microclimates in open land and under forest canopy

This study shows that limited energy reached the ground under the forest canopy. Part of the radiation would have been absorbed by vegetation and transferred to water vapour instead of temperature increase. Therefore, daytime understory Ts_0 was substantially lower than that of open land. During night-time, decrease in understory Ts_0 was greatly lower than that of open land, as a result of forest canopy limiting heat loss (Table I). Consequently, forest canopy has strong effect on the temperature extremes (Renaud and Rebetez, 2009; Ferrez et al., 2011). However, the impact of forest cover on mean temperature could not be entirely explained by the limitation of radiation (Renaud and Rebetez, 2009). Soil temperature below the forest canopy could be more influenced by air temperature and wind speed (Carlson and Groot, 1997; Morecroft et al., 1998; Paul et al., 2004). In this study, nocturnal Ta in open land was close to that of understory (Table I), indicating sufficient heat exchange and a strong effect of wind advection (Renaud and Rebetez, 2009). Owing to high heat capacity of soil, daily understory Ts_0 did not respond as rapidly as did Ta. Therefore, the correlation significance between wind speed and difference of Ts_0 in open land and under forest canopy was low (P < 0.1). However, monthly data showed that understory Ts_0 was closely correlated with Ta in open land with a linear regression coefficient of 1.02 (n = 300, R2 = 0.977) for open land Ta to understory Ts_0. We conclude that Ts_0 in open land is sensitive to Q, and wind speed plays important roles in open land Ta and understory Ts_0.

equals;_

equals;_

4.2. Ta and Ts trends in open land

This study showed that Ta and Ts at all depths have increased more in winter than in other seasons. Strong winter warming has also been observed elsewhere in south-west China (Liu and Chen, 2000; Fan et al., 2010; Qin et al., 2010). As winter St had significant correlation with Ta and Ts_0 (Table III), strong winter warming in this study could be attributed to the stronger increase of winter St (Table IV). This positive trend of St was also observed in rural and mountainous areas of the Yunnan-Guizhou Plateau (Zheng et al., 2010).

Correlation analysis showed that Ta and Ts_0 were more influenced by Ws and St, respectively. The role of turbulence on Ta was confirmed by a significant negative correlation between Ws and the soil–air temperature difference (Table III). Consequently, changes in St and Ws could separately influence trends in Ts_0 and Ta. This study revealed a stable increase in Ws, which agrees with a previous report by Jiang et al. (2010). However, the trend in St has decelerated after 1998 (Table IV). Consequently, Ts_0 has responded differently than Ta since that year (Figure 3). This change point agrees with You et al. (2010) who reported a sharp decrease of St and increase of total cloudiness at the end of the 1990s. As a whole, compared with long-term averages, Ws has increased 4.9% and St has increased 1.6% in these recent 30 years. Therefore, change of Ta (and the closely correlated understory Ts_0) is greater than open land Ts_0.

As soil temperature and air temperature may have been driven separately, no conclusion has been made on the global pattern of trends in air temperature and soil temperature. A higher trend of soil temperature than air temperature has been reported in Northern Ireland (García-Suárez and Butler, 2006). Also, soil temperature in south-west China and East Tibetan Plateau slightly decreased from 1954 to 2001 (Lu et al., 2006), which contradicted with the slightly increase of air temperature (Liu et al., 2004). The increase St in this study only reflects an increase in duration of sunshine hours above the threshold of 120 W m−2; the amplitude of solar radiation could not be observed in our St records. Accordingly, increased St recorded by the dark-tube sunshine hour recorder implies an increased frequency of cloud-free days (Qian et al., 2006), and any increase in aerosols and their influence on solar radiation could not be seen in our St. Trend in soil temperature is considered to be correlated with changes in solar radiation (Jacobs et al., 2011). As a result, the lower trend of Ts_0 compared with Ta in open land could be considered as an indicator of decreased solar radiation, possibly from increased atmospheric aerosols (Alpert and Kishcha, 2008; Alpert et al., 2005; Liu et al., 2004; Qian et al., 2006).

5. Implication

This study reveals an increasing trend of 0.36 °C/decade in Ta, which is double the previous reported trends in our regional temperatures (Liu and Chen, 2000; He and Zhang, 2005; Wan et al., 2009; Fan et al., 2010). A higher temperature trend is consistent with a high sensitivity of mountain areas to climate change (Beniston, 2003; Shrestha and Aryal, 2011). As a result, the distribution of mountain forests, which is considered to be limited by temperature (Fang et al., 1996), could be displaced upward. However, stronger winter warming in this study suggests an enhanced aridity in winter; increased temperature and sunshine hours could strengthen evaporative water demand. Consequently, the impact of changes in temperature and aridity on the ecological and biological perspective of mountain forests needs to be addressed.

Strong canopy effects on the understory micrometeorology were revealed in this study. As a result, Ts at all depths in open land were higher than those under forest canopy (Figure 2). Higher soil temperatures in open land could be important for seed germination and thus forest restoration. Trends of understory Ts at all depths were higher than those of open land. As a result of close relation between soil temperature and soil respiration (Jones et al., 2006), warming effects on soil respiration could be underestimated if predictions are based on meteorological observations in open land.

Acknowledgements

This is a research contribution from the Ailaoshan Station for Subtropical Forest Ecosystem Studies (ASSFE), Chinese Academy of Sciences, the Ailaoshan National Ecosystem Observation Research Network Station and Chinese Ecological Research Network. We appreciate the valuable comments from anonymous referees. Junbin Zhao, Xiamo Liu and Jiafu Wu provide their help in data analysis and figure preparation. A meaningful conversation with Dr Zexin Fan helps us with the trends comparison. The free software R (version 2.12.0) and AnClim (version 5.023) are appreciated for their convenient and time-saving calculation. The local governors, for example Mr. Youyong Luo, are appreciated for their efforts in forest protection. The long-term manual observers, Chengchang Luo, Jinhua Qi, Xin Luo, are appreciated for their continuous and laborious work in observation. This work was supported by National Key Project for Basic Research (No. 2010CB833501-01-07), Yunnan science and technology plan projects (No. 2011FA025), Knowledge Innovation Program of the Chinese Academy of Sciences (No. KZCX2-YW-Q1-05-04), Strategic Priority Research Program of the Chinese Academy of Science (No. XDA05050601) and National Natural Science Foundation of China (No. 31061140359).

Ancillary