Clipping has stronger effects on plant production than does warming in three alpine meadow sites on the Northern Tibetan Plateau

The relative effects of warming and clipping on vegetation growth are not fully understood. Therefore, we compared the relative effects of experimental warming and clipping on the normalised difference vegetation index (NDVI), green NDVI (GNDVI), soil-adjusted vegetation index (SAVI), aboveground biomass (AGB) and gross primary production (GPP) in three alpine meadow sites (A, B and C) on the Northern Tibetan Plateau from 2013 to 2015. There were no obvious effects of experimental warming on the NDVI, GNDVI, SAVI, AGB and GPP at the three sites, which were most likely attributed to experimental warming-induced warming and drying conditions. In contrast, clipping significantly decreased the NDVI, SAVI and AGB by 27.8%, 31.3% and 18.2% at site A, by 27.1%, 31.8% and 27.7% at site B, and by 12.3%, 15.1% and 17.6% at site C, respectively. Clipping also significantly reduced the GNDVI and GPP by 11.1% and 28.2% at site A and by 18.9% and 33.7% at site B, respectively. Clipping marginally decreased the GNDVI by 8.7% (p = 0.060) and GPP (p = 0.082) by 14.4% at site C. Therefore, clipping had stronger effects on vegetation growth than did warming in the three alpine meadow sites on the Tibetan Plateau.

to mimic grazing and agricultural hay harvest in alpine meadows on the Tibetan Plateau 30,31 . Warming and clipping/grazing can indirectly affect moisture regimes, and changing moisture regimes can significantly impact the recent change in the carbon cycle sensitivity to temperature variability 32,33 . Several studies have investigated the responses of aboveground biomass to warming and clipping/grazing under controlled warming and clipping/ grazing conditions, but there are still inconsistent results in alpine meadows on the Tibetan Plateau 21,31,34,35 . Some studies found that experimental warming rather than clipping/grazing had significant effects on AGB across all the observed years in alpine meadows on the Tibetan Plateau 31,35 . In contrast, other studies demonstrated that clipping/grazing rather than experimental warming had significant effects on AGB across all the observed years in alpine meadows on the Tibetan Plateau 34 . These results demonstrate that it is still unclear which one of these two driving factors (warming and clipping/grazing) has a stronger effect on vegetation productivity in alpine meadows on the Tibetan Plateau.
Experimental warming and clipping/grazing have been shown to have lag effects on AGB 34,36 , and in some studies, the effects of experimental warming and clipping/grazing on AGB varied with year in alpine meadows on the Tibetan Plateau 31,35,37,38 . These findings suggest that the relative strengths of experimental warming and clipping/grazing effects on biomass production can vary over time. However, most of these warming and clipping/grazing experiments in alpine meadows on the Tibetan Plateau lasted less than 5 years 31,34,35 . Therefore, it is unclear whether the relative strengths of experimental warming and clipping/grazing effects on vegetation productivity in the short term (<5 years) are different from those in the long term (>5 years) in alpine meadows on the Tibetan Plateau.
Although these previous studies have examined the relative strengths of warming and clipping/grazing effects on AGB and GPP, no studies have investigated the relative responses of the NDVI, GNDVI and SAVI to experimental warming and clipping/grazing under controlled experimental warming and clipping/grazing conditions in alpine meadows on the Tibetan Plateau 31,34,35 . Therefore, in this study, a field warming and clipping experiment was conducted in 2008 at three alpine meadow sites on the Northern Tibetan Plateau. The monthly NDVI, GNDVI, SAVI, GPP and AGB from 2013 to 2015 were measured. The main objective of this study was to compare the relative strengths of experimental warming and clipping effects on the NDVI, GNDVI, SAVI, AGB and GPP in three alpine meadow sites on the Northern Tibetan Plateau.

Results
Effects of warming and clipping on soil temperature (T s ), soil moisture (SM), air temperature (T a ) and vapor pressure deficit (VPD). Across all three growing seasons from 2013 to 2015, there were no significant main effects of clipping and no significant interactive effects of warming and clipping on T s and SM at sites A, B and C (Table S1). Experimental warming significantly increased T s by 1.31 °C, 1.27 °C and 1.17 °C but significantly decreased SM by 0.02 m 3 m −3 , 0.03 m 3 m −3 and 0.04 m 3 m −3 at sites A, B and C, respectively ( Figure S1). Experimental warming significantly increased T a by 1.42 °C, 1.18 °C and 1.24 °C and VPD by 0.12 kPa, 0.10 kPa and 0.08 kPa at sites A, B and C, respectively ( Figure S1). Clipping significantly decreased T a by 0.14 °C at site A but increased T a by 0.25 °C at site C ( Figure S1). Clipping significantly increased VPD by 0.02 kPa at site B.
Effects of warming and clipping on NDVI, GNDVI, SAVI, AGB and GPP. There were significant inter-annual variations in the NDVI, GNDVI, SAVI, AGB and GPP (Table 1 Across all three growing seasons from 2013 to 2015, there were no significant main effects of experimental warming and no significant interactive effects of experimental warming and clipping on the NDVI, GNDVI, SAVI and AGB at sites A, B and C (  Fig. 2). There were also no significant interactive effects of experimental warming and clipping on GPP at sites A and B (Table 1).
However, there were significant main effects of clipping (Table 1 Relationships between the NDVI, GNDVI, SAVI, AGB and GPP and T s , SM, T a and VPD. T s , SM, T a and VPD significantly explained 30%, 52%, 32% and 40% of the variation in the NDVI, respectively (Figs 3 and 4). T s , SM, T a and VPD significantly explained 13%, 33%, 17% and 33% of the variation in the GNDVI, respectively (Figs 3 and 4). T s , SM, T a and VPD significantly explained 29%, 47%, 33% and 42% of the variation in the SAVI, respectively (Figs 3 and 4). T s , SM, T a and VPD significantly explained 21%, 39%, 25% and 38% of the variation in AGB, respectively (Figs 3 and 4). T s , SM, T a and VPD significantly explained 25%, 47%, 30% and 45% of the variation in GPP, respectively (Figs 3 and 4).  38 . Experimental warming increased AGB in a wet alpine meadow site but decreased AGB in a dry alpine meadow site on the Tibetan Plateau 42 . Therefore, water availability regulates the responses of biomass and productivity to warming in alpine meadows on the Tibetan Plateau 39 . This finding may be attributed to the following mechanisms. First, nitrogen addition can increase grassland vegetation growth 43 , and moisture availability can regulate the warming effects on soil nitrogen availability in alpine meadows on the Tibetan Plateau 44 . Second, the minimum value of soil moisture is 11.8% for meadow vegetation growth 45 , and warming-induced soil drying may aggravate the negative effects of low soil moisture on alpine vegetation growth 39 . Third, warming-induced increases in VPD may cause stomatal closure, affect stomatal conductance and intercellular CO 2 concentration, and depress vegetation photosynthesis 46 .
There were inconsistent interannual variations in growing season precipitation and productivity in the current study.  36 . These findings suggest that warming may have legacy effects on biomass production in alpine meadows on the Tibetan Plateau 31,36 . The close relationship between warming duration and warming effects on vegetation biomass and production may be related to the following mechanisms. First, plant photosynthetic capacity is generally positively related to leaf nitrogen content 47,48 , and the positive effect of warming on the leaf nitrogen concentration can decrease with warming duration 49 . Second, soil biota can regulate soil nitrogen availability 44 , which in turn can limit alpine vegetation growth 43 . The effects of warming on the abundance of soil biota are reduced with warming duration 50 .  Figs 1 and 2). Our previous study, which was conducted in the same alpine meadow sites, demonstrated that clipping significantly reduced AGB and GPP in 2012 41 . Similarly, heavy grazing resulted in a decrease in AGB and caused degradation from 2006 to 2010 in an alpine meadow in the Haibei region 35 . Clipping also significantly reduced AGB in an alpine meadow in the Yangtze River Source Region 34 . However, clipping marginally (p < 0.10) increased AGB from 1998 to 2001 in an alpine meadow in the Haibei region 31 . These different responses of vegetation biomass and production to clipping/grazing can be attributed to clipping intensity and frequency. For example, clipping occurred three times during the growing season, and on average, approximately 43%, 44% and 47% of the total maximum aboveground biomass was removed in the alpine meadow sites A, B and C, respectively. Grazing occurred two or three times during the growing season, and an average of approximately 50% of the total maximum aboveground biomass was removed in Wang et al. 35 . An average of approximately 37% of the total maximum aboveground biomass was removed in Peng et al. 34 . In contrast, clipping occurred only once a year and 15% of the total maximum aboveground biomass was removed in Klein et al. 31 .
The effects of clipping on vegetation biomass and production may be attributed to the following mechanisms. First, clipping can cause compensatory growth, and the compensatory growth magnitudes may decrease with clipping intensity 51,52 . Second, clipping can reduce green leaf area and fractional photosynthetically active radiation absorbed by vegetation 51,53 , which in turn can suppress vegetation photosynthesis and production accumulation. Third, clipping may also affect vegetation growth by indirectly influencing environmental temperature and moisture conditions. However, clipping only significantly affected T a by 0.14 °C at site A and by 0.25 °C at site C. Clipping only significantly increased VPD by 0.02 kPa at site B but not sites A and C. Clipping did not affect T s and SM at the three alpine meadow sites (Figs 3 and 4). Moreover, SM had the strongest effects on vegetation indices, AGB and GPP (Figs 3 and 4). These findings suggest that the clipping-induced changes in environmental temperature and moisture conditions most likely have negligible effects on vegetation biomass and production in the three alpine meadow sites on the Northern Tibetan Plateau.
Stronger effects of clipping than experimental warming. Our findings suggested that the response magnitudes of the NDVI, GNDVI, SAVI, AGB and GPP to clipping were stronger than those to experimental warming (Table 1). Our previous study, which was conducted in the same alpine meadow sites, also implied that clipping had stronger effects on GPP and AGB in 2012 than did experimental warming 41 . Similarly, clipping    . Relationships (a) between the normalised difference vegetation index (NDVI) and soil temperature (T s ), (b) between the NDVI and air temperature (T a ), (c) between the green NDVI (GNDVI) and T s , (d) between the GNDVI and T a , (e) between the soil-adjusted vegetation index (SAVI) and T s , (f) between the SAVI and T a , (g) between aboveground biomass (AGB) and T s , (h) between AGB and T a , (i) between gross primary production (GPP) and T s , and (j) between GPP and T a in alpine meadows on the Tibetan Plateau.
SCiENtiFiC REPORTS | 7: 16330 | DOI:10.1038/s41598-017-16645-2 Aboveground plants were clipped to approximately 1 cm in height using scissors, and the clipped aboveground biomass was removed, oven-dried at 65 °C for 48 h and weighed. Microclimate measurements. SM at a depth of 10 cm, T s at a depth of 5 cm, T a and relative humidity (RH) at a height of 15 cm were continuously monitored using weather stations (HOBO weather station, Onset Computer, Bourne, MA, USA) during the whole study period from June to September in 2013-2015. VPD was calculated using measured T a and RH.
There were only 3 out of 122 days when T a in the 'W' plots was lower than that in the 'C' plots in 2014 at site A ( Figure S2). There were 5 and 18 out of 122 days when T a in the 'W + CL' plots was lower than that in the 'CL' plots in 2014 and 2015 at site A, respectively ( Figure S3). There were only 3 and 1 out of 122 days when T a in the 'W' plots was lower than that in the 'C' plots in 2014 and 2015 at site B, respectively ( Figure S2). There were only 1, 6 and 1 out of 122 days when T a in the 'W + CL' plots was lower than that in the 'CL' plots in 2013, 2014 and 2015 at site B, respectively ( Figure S3). There were 17 and 1 out of 122 days when T a in the 'W' plots was lower than that in the 'C' plots in 2013 and 2014 at site C, respectively ( Figure S2). There were only 2, 4 and 4 out of 122 days when T a in the 'W + CL' plots was lower than that in the 'CL' plots in 2013, 2014 and 2015 at site C, respectively ( Figure S3). There were only 3 out of 122 days when T s in the 'W + CL' plots was lower than that in the 'CL' plots in 2014 at site A ( Figure S4). There were only 9 out of 122 days when T s in the 'W' plots was lower than that in the 'C' plots in 2014 at site B ( Figure S5). There were only 11 out of 122 days when T s in the 'W + CL' plots was lower than that in the 'CL' plots in 2014 at site B ( Figure S4). There were only 7 and 5 out of 122 days when T s in the 'W' plots was lower than that in the 'C' plots in 2013 and 2014 at site C, respectively ( Figure S5). There were 25, 5 and 2 out of 122 days when T s in the 'W + CL' plots was lower than that in the 'CL' plots in 2013, 2014 and 2015 at site C, respectively ( Figure S4). However, there were no cases when both T s and T a were lower in the OTCs than that outside the OTCs during the three consecutive growing seasons from 2013 to 2015 at sites B and C. There were only 2 out of 122 days when both T s and T a were lower in the 'W + CL' plots than in the 'CL' plots at site A. These figures indicated that the OTCs were effective in elevating environmental temperature in the three alpine meadow sites in the current study.
Vegetation indices measurements and AGB estimation. Images in a 0.50 m × 0.50 m subplot in the centre of each target plot were taken with a Tetracam Agricultural Digital Camera (ADC, Tetracam Inc., Chatsworth, CA, USA) 39 . The ADC lens was parallel to the surface at a height of approximately 1.00 m with a field of view of 0.80 m × 0.60 m and a spatial resolution of 0.40 mm. To produce the calibration parameters, images of a white Teflon plate provided by the Tetracam manufacturer were also taken by the ADC before and after each batch of target images or when the light conditions changed remarkably. The NDVI, GNDVI and SAVI were calculated with following equations (i.e., equations 1-3) using PixelWrench2 software (included with the ADC): where ρ nir , ρ red and ρ green indicate the near-infrared (760-900 nm), red (630-690 nm) and green (520-600 nm) of the ADC, respectively. A non-destructive method was used to estimate AGB, which was based on the exponential regression equation between AGB and NDVI (i.e., AGB = 10.33e 3.28NDVI ) 39 .

GPP algorithm.
A detailed description of the GPP algorithm can be found in our previous studies 54 . A concise description is only listed for this study. GPP was estimated using the following equations: where APAR is absorbed photosynthetically active radiation by the plant canopy; LUE is light-use efficiency; FPAR is the fraction of plant canopy APAR; PAR is photosynthetically active radiation; LUE max is maximum LUE; T aminscalar is temperature attenuation scalar; VPD scalar is water attenuation scalar. T aminscalar can be calculated using daily minimum T a (T amin ), and VPD scalar can be calculated using daytime mean VPD 55 . FPAR can be calculated using the observed NDVI 39 .
Statistical Analysis. For each site, a repeated-measures analysis of variance was used to estimate the main and interactive effects of experimental warming, clipping and measuring year on the T s , SM, T a , VPD, NDVI, GNDVI, SAVI, AGB and GPP (Table 1, S1). For each year, a repeated-measures analysis of variance was used to estimate the main and interactive effects of experimental warming, clipping and measuring month on the NDVI, GNDVI, SAVI, AGB and GPP (Figs 1 and 2). Linear regressions of the NDVI, GNDVI, SAVI, AGB and GPP with T s , SM, T a , VPD were conducted (Figs 3 and 4). All the statistical analyses were performed using SPSS software (version 16.0; SPSS Inc., Chicago, IL).