Radial Growth of Picea schrenkiana Influenced by Increasing Temperature in the Tianshan Mountains
ABSTRACT
There is a tendency of warming and wetting in northwestern China in recent decades. However, less is known about whether or not tree growth shows an increasing trend. In this study, we developed three tree-ring width chronologies of Picea schrenkiana in the northern Tianshan Mountains to assess changes in the radial growth of P. schrenkiana and to discuss the stability of the relationships between tree growth and climate. Three chronologies all showed that the tree-ring index has declined significantly since the 1960s. At two western sites, the growth of P. schrenkiana was negatively affected by the summer temperature of the previous year. At the eastern site, early summer temperature-induced drought of the current year was the key factor affecting tree growth. The result of moving-window correlations was consistent with correlation analyses. Spatial correlation analyses revealed that variations in tree-ring width could respond to a wide range of temperature changes in northwestern China, especially in the past half century. We expect that climate warming hinders the radial growth of P. schrenkiana in the northern Tianshan Mountains. Our study also helps to clarify the characteristics of tree growth in northwestern China under the influence of westerlies.
INTRODUCTION
Tree-ring width has been widely used to investigate long-term climate variations (Liang et al. 2008; Yang et al. 2014; Keyimu et al. 2021). However, correlations between climatic factors and tree-ring widths are not always stable (Coppola et al. 2012; Du et al. 2020). Global warming can affect a variety of ecological processes worldwide (Cohen et al. 2016; Crabbe et al. 2016), and some studies have reported the contribution of 20th Century global warming to these unstable relationships between climatic factors and tree-ring widths (Cao et al. 2019; Cai et al. 2020; Keyimu et al. 2020). Therefore, assessing the stability of the relationship between tree-ring widths and the climate is important in the context of global warming.
A large number of studies (Wang et al. 2018, 2019, 2020) have pointed out that with the continuous warming of the climate, forests in the northern and middle latitudes have experienced extensive growth declines. The mean temperature and annual total precipitation have begun to increase significantly in northwestern China in recent decades (Yao et al. 2016). This change has been thought to have profoundly affected physiological and ecological processes of the mountain forest ecosystem in northwestern China (Li et al. 2015). A number of studies have demonstrated that many climatic variables, such as precipitation (Zhang et al. 2013, 2017) and drought (Xu et al. 2015; Zhang et al. 2016), affected tree growth in the Tianshan Mountains. However, uncertainty about the radial growth trend of trees under climate change in the Tianshan Mountains hinders a comprehensive understanding of the characteristics of tree growth in northwestern China and requires further investigation.
The purpose of this study was to (1) investigate the trend of radial tree growth in different locations in the northern Tianshan Mountains in the past half century, (2) explore the key climatic factors affecting tree radial growth at different study sites, and (3) determine whether the relationships between tree growth and climatic factors are stable. Our hypothesis is that recent warming in northwestern China may change the characteristics of radial tree growth. Our results and findings may provide evidence that aids in understanding the response of P. schrenkiana to current regional climate change and also provide a reference for future forest management and cultivation.
MATERIAL AND METHODS
Study Areas and Climate Data
The study area is located in the northern Tianshan Mountains of Xinjiang, China. The TianShan Mountains cover a large proportion of Central Asia (Yu et al. 2020). The regional climate is strongly shaped by the central Asian westerly circulation (Xu et al. 2014) and the ranges in the western Tianshan Mountains have a relatively moist climate, whereas the ranges in the east show a pronounced continental climate (Aizen et al. 1997; Huang et al. 2013).
For our study, climate data were obtained from Wenquan (WQ), Xinyuan (XY), and Urumqi (UM) meteorological stations near our three sampling sites (60km, 40km, and 38km away, respectively). Climate data including monthly temperature and total precipitation from 1959 to 2012 were downloaded from the National Meteorological Information Center (http://data.cma.cn/). The respective annual mean temperatures of the three meteorological records from 1959 to 2012 are 3.93°C, 8.92°C, and 7.15°C, and their mean annual total precipitation is 233.23 mm, 504.75 mm, and 266.52 mm. July was the hottest month for all three meteorological stations. July was the month with the highest monthly precipitation for WQ, and May was the month with the highest monthly precipitation for XY and UM (Figure 2).
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
Tree-Ring Sampling and Chronology Development
Schrenk spruce (P. schrenkiana) is a dominant tree species and also one of the most important zonal vegetation types in the Tianshan Mountains. P. schrenkiana is a tree species with high water requirements and low drought resistance. This species naturally grows in shady areas at 1500–2800 m above sea level. The stands are mostly a pure forest, with canopy densities of 60%–80% (Zhu et al. 2021). The tree-ring cores of living P. schrenkiana were collected at three sites (abbreviated as GZG, KD and TC, respectively) from the Tianshan Mountains. Two sampling sites (Guozigou, GZG; and Kuerdening, KD) are located in the western region of the Tianshan Mountains and one (Tianchi, TC) in the middle region (Figure 1), and the sampling sites are not disturbed by human activity. To minimize non-climatic influences on tree growth, only trees that had no obvious injury or disease were sampled. A minimum of 20 trees were selected at each site, and for each tree, two core samples were collected by extracting cores from opposite sides of each tree at a height of ca. 1.3 m with a 5-mm-diameter increment borer. In total, 142 increment cores were extracted from 71 P. schrenkiana trees in the three sample sites.
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
The tree-ring samples were processed following standard dendrochronological practices (Stokes and Smiley 1968). In the laboratory, the cores were fixed in core mounts, sanded, crossdated and then ring widths were measured to the nearest 0.001 mm by a LINTAB 6 tree-ring measurement station. The crossdating was verified using the program COFECHA (Holmes 1983), and three chronologies were obtained by using the program ARSTAN (Cook and Kairiukstis 2013) to remove the biological growth trends. All measured series were detrended using smoothing splines (with a 67% variance cut-off) to remove anomalous growth trends (Fritts 1976). The detrended data from individual tree cores were processed to produce a standard chronology using a bi-weight robust mean to minimize the influence of outliers, extreme values, or biases in the tree-ring indices. The quality of the chronology was assessed using the following statistical parameters: the mean sensitivity (MS), a measure of the relative change in tree-ring width from a given year to the next year in a given series; standard deviation (SD), a measure of the dispersion of a set of data from their mean; signal-to-noise ratio (SNR), a measure of the common variance in a chronology scaled by a measure of the total variance in the chronology; expressed population signal (EPS), a function of the series replication and the mean inter-series correlation; and the subsample signal strength (SSS) (Wigley et al. 1984), a measure that is applied to determine the minimum number of trees that should be used to obtain a reliable estimation of the mean chronology (Cook 1985).
Growth–Climate Response
A Pearson correlation analysis was employed to study the relationships between tree-ring chronologies and the abovementioned climatic elements over the period of 1960–2012 (53 years). An extended period, from April of the previous year to September of the current year, was chosen to analyze the growth–climate relationships because tree growth may be influenced by climate factors in the current year and those in the previous year (Fritts 1976). To understand the temporal variation and stability of dendroclimatic correlations, a 30-year-window moving correlation analysis was also used in this paper. We used seasonal climate data that were computed by averaging the monthly values (summer: June–August; autumn: September–November; winter: December–February; spring: March–May) in moving correlations. Analyses were performed in the R version 4.1.1 (R Development Core Team 2018) using the packages dplR (Bunn 2008) and treeclim (Zang and Biondi 2015). To explore the impact of regional climatic elements on tree growth in the last century, a spatial correlation analysis was used to study the relationships between the three tree-ring width chronologies and CRU TS4.04 land temperatures over the periods 1901–1960 and 1961–2011. Spatial correlation analysis was done through the http://climexp.knmi.nl website.
RESULTS
Climate Trends and Characteristics of Tree-Ring Width Chronologies
Both warming and wetting trends have been observed in the study area since the 1960s (Figure 3). The climate data from three climate stations show significant increasing trends of annual mean temperature (+0.20°C/decade in WQ, +0.42°C/decade in XY, and +0.22°C/decade in UM). The increasing trends of the annual total precipitation also reach a significant level (+15.44mm/decade in WQ, +20.31mm/decade in XY, and +24.12mm/decade in UM).
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
Table 1 shows the standard parameters of our P. schrenkiana chronologies. The three chronologies showed inter-series correlations (Rbar) ranging between 0.334 and 0.656, with the highest correlations observed in TC, implying a higher common signal among the individual trees of the TC chronology. Compared with GZG and KD, TC had higher MS, SNR and EPS values, whereas KD has a higher SD than GZG and TC (Table 1). We evaluated the reliability of the three tree-ring width chronologies by the SSS. The GZG, KD, and TC chronologies became reliable in 1855, 1680, and 1894, respectively, based on the SSS value, which was >0.85 and included a minimum of nine, six, and three tree-ring sampling cores (Wigley et al. 1984). We also calculated the relationships between the three chronologies for the period 1959–2012. During this period, the KD and TC chronologies were the most strongly correlated (r = 0.61, p < 0.01), followed by the KD–GZG (r = 0.54, p < 0.01). But TC-GZG correlation is only 0.23, which may be related to the long distance between the two sampling sites. In addition, the linear fitting results of three chronologies indicate that the three chronologies all have a significant downward trend during the period 1960–2012. Among them, the downward trend of the KD chronology is the most significant (R2 = 0.479) (Figure 4).
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
Radial Growth–Climate Relationship
The correlation analysis between three tree-ring chronologies and climatic factors (Figure 5) indicates that during 1960–2012, radial tree growth at the GZG site showed significant negative correlation with the mean temperatures in April, May, June, July, and August of the previous year and in April of the current year. Growth at the KD site showed significant negative correlation with the mean temperatures during the growing season in the previous year and the current year. Growth was most strongly and significantly related to mean July temperature in the previous year (r = –0.626). No significant correlations were found between the GZG and KD chronologies and precipitation over the 1960–2012 period. For the TC site, growth was significantly related to mean May and June temperature in the current year and growth also showed significant positive correlation with precipitation in the previous June.
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
The results of the 30-year window moving correlation (Figure 6) show that the correlations between GZG and KD chronologies and summer temperature in the previous year were all negative and significant throughout the time period. Correlation between the TC chronology and summer temperature in current year began to be significant in the 1970s. No significant correlations were found between the GZG chronology and precipitation. Positive and significant correlations were found between the KD chronology and summer and autumn precipitation in previous year, indicating the importance of water availability. Significant correlations between the TC chronology and spring precipitation in the current year became weak and non-significant from the late 1970s onwards.
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
Spatial correlations were calculated between the three tree-ring chronologies and the April–September averaged CRU TS4.04 land temperatures in previous year. The GZG and KD chronologies were negatively correlated with the regional temperature in the area west of the sampling site from 1901 to 1960. These correlations became more significant from 1961 to 2011. The TC chronology was only negatively correlated with the regional temperature in the area west of the sampling site from 1961 to 2011 (Figure 7).
Citation: Tree-Ring Research 78, 2; 10.3959/TRR2021-13
DISCUSSION
For the three chronologies, the growth of P. schrenkiana responded negatively to the mean temperature during the common period in most of the months examined (Figure 5), indicating that temperature limited radial tree growth. Tree growth at GZG and KD sites showed that prior-year climate strongly affected tree-ring formation in the subsequent year. The negative response to prior-year temperature was also observed in the Mediterranean basin (Tegel et al. 2014; Fyllas et al. 2017), and southeastern Europe (Panayotov et al. 2010). The negative and significant correlation obtained between growth and temperature during the growing season in previous year was attributed to the fact that rising temperatures can accelerate respiration by trees (Adams et al. 2009; McDowell et al. 2011; Hartmann et al. 2013), leading to a reduction in carbon storage in the trees and thus affecting radial tree growth in the following year (Liang et al. 2011). The results are consistent with previous studies in southeastern Qinghai-Tibet Plateau (Liang et al. 2010), Changbai Mountain (Yu and Liu 2020), and Xinglong Mountain (Fang et al. 2010). Rising growing season temperatures in current year could negatively impact tree growth through several mechanisms. One plausible reason could be high temperatures during the growing season can accelerate tree transpiration and cause water stress in trees (Li et al. 2006). In addition, the water availability constraints on photosynthesis will be amplified by drought-induced stomatal closure (Adams et al. 2009). Our analyses indicates that radial tree growth in the northern Tianshan Mountains is influenced by summer moisture. Temperature has an obvious lag-effect on radial tree growth in the western Tianshan Mountains.
Many studies have found unstable climate responses in tree-ring width chronologies (Coppola et al. 2012; DeSoto et al. 2012; Begovic et al. 2020). In this study, tree growth in GZG and KD was steadily affected by summer temperature in previous year during the study period. But tree growth in TC was only significantly affected by the summer temperature in current year since the 1970s (Figure 6). The total annual precipitation in the Tianshan Mountains shows a significant decline from west to east, which means that the eastern region of the Tianshan Mountains suffers relatively higher water-pressure deficit than the western region (Qi et al. 2015). Higher temperatures lead to greater evaporation of soil moisture and thus to increased moisture deficit in arid regions (Suarez et al. 2015). Thus, the negative effect of summer temperature in current year on the tree growth in TC has increased. In this study, the results demonstrate that the relationships between tree-ring widths and precipitation varied with climate warming (Figure 6). The warming effect would alter climate factors and influence the sensitivity of tree growth. Rising temperatures may have changed the role of precipitation in tree growth (Wang and Yang 2021). These results indicate that moisture availability during the growing season was still the main limiting factor for radial tree growth under the trend of warming and wetting in northwestern China.
The spatial correlation results obtained from the long-term data series also emphasize the important role of temperature and confirmed the strengthening impact of temperature on tree growth in the northern Tianshan Mountains in the past half century. The climate signals reflected in tree rings have become more prominent since the 1960s, especially the GZG and KD chronologies in the western part of the study area (Figure 7). In addition, tree growth in our study area was mainly related to the climate in the westerly-dominated region, facilitating the possibility of reconstructing the climate controlled by westerlies through the tree-ring index values of trees growing in the northern Tianshan Mountains. However, the lack of a sufficient sample size may affect the study of the large-scale growth characteristics of trees in the northern Tianshan Mountains. Thus, further effort is needed to obtain additional tree-ring data to confirm and refine results from our study.
CONCLUSIONS
Despite a warming and wetting tendency in western China, we found that the growth of P. schrenkiana was limited by moisture and showed a declining trend in recent decades. The relationships between the temperature and radial tree growth are very stable. Our results support the hypothesis that rising temperatures have caused a decrease in the tree-ring width index. The results obtained from the regional long-term tree growth and climate data reveal the relationship between the radial growth of P. schrenkiana and the climatic factors controlled by the system of westerlies in the western Tianshan Mountains. Under the current trend of climate change, temperature will continue to hinder the radial growth of P. schrenkiana in the northern Tianshan Mountains and may further lead to declines in forest productivity and carbon storage in P. schrenkiana forests. Therefore, in the context of increased climate warming and consequences to Tianshan forests, special attention should be paid to the restrictions on tree growth caused by temperature-induced drought in the Tianshan Mountains.
Temperature and total precipitation during 1959–2012, obtained using records from the WQ, XY, and UM meteorological stations in Northwest China.
Locations of the three tree-ring sampling sites and the meteorological stations in Xinjiang, China.
Annual climate trends at the WQ, XY, and UM (1959–2012, dashed lines = annual data, solid lines = regression) climate stations in the Tianshan Mountains. R represents the correlation coefficient over the study period. Significance: * p < 0.05, ** p < 0.01.
Three standard tree-ring width chronologies (dashed lines) and the linear fitting results (solid line) during the period 1960–2012.
Correlation coefficients calculated between three tree-ring width chronologies and the climatic variables (mean temperature and precipitation) from April of the previous year to September of the current year over the 1960–2012 period. The dashed lines indicate the significance levels of 95%. P indicates the previous year.
Moving correlations calculated between the three tree-ring width chronologies and seasonal temperature (T) and precipitation (P) of previous summer (prev.sum), previous autumn (prev.aut), previous winter (prev.win), current spring(curr.spr), and current summer (curr.sum) using 30-year windows. Significant correlations at the 95% significance level are denoted by asterisks.
Spatial correlations between the three tree-ring width chronologies and the April–September mean CRU TS4.04 land temperatures of the previous year obtained through the http://climexp.knmi.nl website over the periods 1901–1960 and 1961–2011. The black dots indicate the sampling sites. The colored regions indicate correlations at significance levels above 0.05.
Contributor Notes