Boqian Yan·Jian Yu·Qijing Liu·Lihua Wang·Lile Hu

Abstract Tree-ring width chronologies of Larix chinensis were developed from the northern and southern slopes of the Qinling Mountains in Shaanxi Province,and climatic factors affecting the tree-ring widths of L.chinensis were examined.Correlation analysis showed that similar correlations between tree-ring width chronologies and climatic factors demonstrated that radial growth responded to climate change on both slopes.The radial growth of L.chinensis was mainly limited by temperature,especially the growing season.In contrast,both chronologies were negatively correlated with precipitation in May of the previous year and April of the current year.Spatial climate-correlation analyses with gridded land-surface climate data revealed that our tree-ring width chronologies contained a strong regional temperature signal over much of northcentral and eastern China.Spatial correlation with seasurface temperature fields highlights the influence of the Pacific Ocean,Indian Ocean,and North Atlantic Ocean.Wavelet coherence analysis indicated the existence of some decadal and interannual cycles in the two tree-ring width chronologies.This may suggest the influences of El Nin~o-Southern Oscillation and solar activity on tree growth in the Qinling Mountains.These findings will help us understand the growth response of L.chinensis to climate change in the Qinling region,and they provide critical information for future climate reconstructions based on this species in semi-humid regions.

Keywords Climate response·Dendroclimatic·Tree-ring width·L.chinensis·Qinling Mountains

Introduction

The response of trees and forests to climate change has been attracting greater attention over the past few decades.Impact on the growth and survival of individual tree species will affect whole ecosystems and the essential services they provide to human beings(Bonan 2008).In terms of individual tree growth processes,alpine timberline forests and upper distribution boundaries of tree species are quite sensitive to climate change(Yang et al.2013;Liang et al.2014).

Tree rings represent a record of past climatological and ecological conditions experienced by trees throughout the growth process(Fritts 1976;Orwig and Abrams 1997).Therefore,radial growth,reflected by tree-ring widths,is considered to be closely related to climate at the alpine timberline,and that relationship has been studied extensively throughout the world(Yang et al.2013;Liang et al.2014).Such work is crucial for the evaluation of forest structure,composition,and function(e.g.,spatial distribution and productivity)under future climate regimes(Dai et al.2011;Fang et al.2015).

Numerous dendrochronological studies have been done in recent years on the effects of climate on the radial growth of trees in alpine and subalpine regions(Dang et al.2013;Yang et al.2013).In the subalpine/alpine zone,the radial growth of trees is subject to environmental gradients associated with elevation(Liu et al.2013).Generally,tree growth at the upper treeline is limited by summer temperature,and at lower elevations,water availability is often the limiting climatic factor(Chen et al.2011;Sidor et al.2015). However, previous studies have concluded that precipitation constitutes another major factor limiting radial growth,and precipitation can have greater direct or indirect effects than temperature in some regions(Oberhuber 2004;Liang et al.2014).For example,Liang et al.(2014)found that the growth of Betula utilis at the upper timberline was persistently limited by moisture availability(i.e.,a drought-induced alpine timberline).The physiological mechanisms determining tree-growth response to climatic factors on alpine and subalpine timberlines are very complicated.

The Qinling Mountains are situated in the transitional zone between two macroclimatic regimes(subtropical and warm-temperate zones),thus making it a biologically rich area and sensitive to climatic change(Liu et al.2003,2009;Shi et al.2012;Chen et al.2014).In addition,it is a crucial watershed between the Yangtze River and the Yellow River due to their East-West alignment(Yan and Zheng 2001;Li et al.2012).Meanwhile,since this area is in a pristine state with little anthropogenic disturbance,it is ideal for investigating tree radial growth-climate interactions.There are an increasing number of dendroclimatology studies being conducted in this region(Dang et al.2007;Qin et al.2016).

Larix chinensis is an endemic species in alpine timberlines throughout the Qinling Mountains,especially Taibai Mountain, where the vertical distribution ranges from approximately 2800-3500 m a.s.l(Zhang et al.2005).It is also the only surviving larch in the Qinling Mountains.L.chinensis and its habitat provide ideal conditions for the study of climate-growth relationship. Although several dendrochronological studies have been conducted on L.chinensis forests in the Taibai Mountain(Dai et al.2003;Kang et al.2010;Sun et al.2010),most of those studies focused on the reconstruction of various climatic factors of past centuries(Liu and Shao 2003;Garfin et al.2005).Despite much progress exploring the relationship between tree growth and climate response for the Qinling Mountains using tree-ring data,the spatial coverage of the study sites remains sparse.

In this study, we developed two tree-ring width chronologies for L.chinensis from northern and southern slopes of the Qinling Mountains. We examined the potential of using tree rings of L.chinensis for dendroclimatological study,including the response of radial tree growth to climate changes.To further clarify the effect of climate on the L.chinensis in this region,we used spatial climate-correlation analyses to investigate linkages between the radial growth of L.chinensis and sea-surface temperatures.

Materials and methods

Study area

Study sites were located in the Taibai Mountain National Nature Reserve (107°22′-107°52′N, 33°49′-34°06′E;1060-3771 m a.s.l.)in the central area of the Qinling Mountains, Shaanxi Province in northwestern China(Fig.1).The Qinling Mountains are in the transitional region between the warm temperate and subtropical zones.This area is a physiographic barrier that serves as a dividing line between northern and southern China.The total area of Taibai Mountain is 56,325 ha;it is the first peak in the eastern Tibetan Plateau with an altitude of 3767 m a.s.l.

The Qinling Mountains are located at the southern edge of the warm temperate climate zone and are characterized by an inland monsoon climate:summers are warm and humid because of the monsoon from the southwest and the Pacific,and winters are cold and dry under the influence of northerly and northwesterly air currents. The nearest meteorological station is located in Wugong county(Figs.1,2;34°15′N,128°13′E;447.8 m a.s.l.).The annual mean temperature from 1959 to 2015,was 13.32°C.The coldest month on average was January(mean temperature of-0.45°C),and the warmest month on average was July(mean temperature of 26.22°C).Mean annual precipitation is 596.89 mm,of which about 51%occurs between July and September.

Vegetation on the southern slopes of the Qinling Mountains was similar to the northern slopes;the only difference was that the lower elevations supported deciduous broad-leaved forests (Zhao et al. 2014). Taibai Mountain has distinct climatic variation along altitudinal gradients and clear vertical zonation in vegetation,especially in southern slopes.It represents the major forest types in northwestern China;southern slopes are characterized by five distinct vertical vegetation zones,consisting of Quercus variabilis forests(750-1300 m a.s.l.),mixed conifer and hardwood forests(1300-2300 m a.s.l.),birch forests (2300-2650 m a.s.l.), coniferous forests(2650-3400 m a.s.l.), and subalpine meadows (above 3400 m a.s.l.;Zhao et al.2014).

Fig.1 Location map of sampling sites and meteorological stations

Fig.2 Mean monthly precipitation(bars),mean maximum temperature(line with triangles),mean minimum temperature(line with solid circles)and mean temperature(line with hollow circles)from the nearby Wugong meteorological stations from 1959 to 2015

At the upper limit of the timberline in Taibai Mountain,L.chinensis is the dominant tree species,which develops in pure forest.Above the treeline are alpine shrub communities consisting of Rhododendron purdomii, Rh. Clementinae,and Spiraea alpine.Herbaceous communities are dominated by small statured species, including Allium prattii， Gentiana apiata， Codonopsis tsinlingensis, and Anemone taipaiensis.L.chinensis forests in lower elevations are intermixed with Abies fargesii and Betula utilis.

Sampling sites

In August 2009,increment cores were collected at four sites located at the upper limit of the northern(site code NU)and southern slopes(site code SU),and the lower limit of the northern(site code NL)and southern slopes(site code SL)of Taibai Mountain(Table 1).These are natural forests with no artificial disturbance affecting tree growth.To minimize noise caused by non-climatic factors,only healthy trees were included.In general,one core was extracted at breast height(1.3 m above ground)from each tree,using an increment borer.A total of 166 increment cores were collected.Cores were brought to the laboratory and air-dried prior to mounting and sanding;all samples were prepared following standard dendrochronological techniques(Stokes and Smiley 1968).Cores with abnormal ring features,such as missing rings or indistinct boundaries that are typically difficult to crossdate,were excluded;thus,a total of 65 cores were retained for final analysis.Table 1 provides details regarding the sites and sampled trees.

Chronology development

Tree-ring widths were measured to 0.01 mm using the LINTABTm6.0 measuring system.The accuracy of width measurement and cross-dating was verified using COFECHA(Holmes 1983).To mitigate potential trend distortion problems in traditionally detrended chronologies(Melvin and Briffa 2008),we used a signal-free method to detrendtree-ring series using the RCSigFree program(Melvin and Briffa 2008).Age-related trends were removed from the raw tree-ring series,using negative exponential curves or linear regression curves of any slope.The ratio method was used to calculate tree-ring indices,and the age-dependent spline was selected to stabilize the variance caused by core numbers.The stabilized signal-free chronology was used for the subsequent analysis(Fig.3).The expressed population signal(EPS)was used to evaluate the reliability of the tree-ring chronologies(Wigley et al.1984).

Table 1 Stand conditions of sampling sites of L.chinensis in Taibai Mountain

Climate data

Because of the complex mountainous terrain,the alpine climate at the sampling site is quite different from that of the nearest meteorological stations,all of which are located far from the study sites.To emphasize tree growth-climate relationships,we calculated monthly mean temperature(Tm) and precipitation (Pm) from gridded datasets of CRU2.1(Climate Research Unit,UK)encompassing the area(33.5°-34.5°N;107.5°-108.5°E)for further analysis.The CRU2.1 gridded data sets were downloaded from the KNMI climate explorer(http://climexp.knmi/nl).

Statistical analysis

Pearson's correlation analysis was used to identify the climate signals in the tree-ring width indices.Considering that tree growth could be influenced by climate conditions in the previous and current years(Fritts 1976),correlation analysis was performed for an 18-month interval from April of the previous growth year to September of the current growth year from 1960 to 2008.To assess the common growth forces between individual sites,a principal component analyses (PCA) was applied over the common period of the chronologies.

The multitaper method(MTM)of spectral analysis was used to examine the variability of the PC#1 of tree-ringwidth in the frequency domain(Mann and Lees 1996).We employed the wavelet coherence method to evaluate similarities between two chronologies because it can reveal local correspondence at different time-frequency domains.The 5%significance level against red noise is shown as a thick contour in the appropriate figures.In addition,we also used the KNMI climate explorer(http://climexp.knmi.nl)to generate correlation fields with synoptic-scale climate parameters,such as the updated 0.5×0.5 gridded meteorological dataset of CRU and the sea-surface temperature(SST)dataset of HadISST1.These maps allowed us to evaluate the representativeness of regional chronologies at regional scales.

Fig.3 Plot of new tree-ring width chronologies and the sample depth of L.chinensis in the northern and southern slopes in Taibai Mountain,and the PC#1 of tree-ring widths.The red curve superimposed on the tree-ring indices is an 10-year FFT low pass filter

Results

Chronological characteristics

The cross-matching results indicated there were no absent or false rings in any of the crossdated cores.Two tree-ring width chronologies covering 199 years(1810-2008)and 200 years(1809-2008)were developed for the northern and southern slopes of Taibai Mountain, respectively.Comparison of the two chronologies showed that they matched well with each other and had similar statistical characteristics(data not shown).Furthermore,the expressed population signal(EPS)of L.chinensis chronologies on northern and southern slopes were consistently greater than 0.85 during the period 1898-2008 and 1847-2008,respectively, which is generally considered an acceptable threshold for reliable chronologies(Wigley et al.1984).The two chronologies were significantly correlated with each other(r=0.418,p ＜0.01,n=199).According to the PCA,PC#1RWaccounted for 70.9%of the total variance.

Multi-taper method(MTM)spectral analysis revealed some significant low- and high-frequency cycles at PC#1RW. Low-frequency peaks were found at 11.7-12.0 years.Other significant peaks were found at 2.5-2.6, 3.4-3.5, 5.3-5.7, 5.9, 6.1, 7.2-7.5, and 7.8-8.9 years. Wavelet coherency analysis results are shown in Fig.4.The coherence of the two chronologies has varied in time across most spectral bands.However,the periods of greatest coherency occurred on long temporal scales(20-35 and 45-65 years).In addition,significant high-frequency coherence was observed between the two tree-ring width chronologies on long temporal scales(2-8 years)throughout 1810-1890 and 1910-2000.

Fig.4 Wavelet coherence between the two tree-ring width chronologies.The 95%significance level against red noise is shown as a black contour.Arrows indicate the phase of the coherence,where right is in phase and left is antiphase

Tree growth-climate relationships

Correlation coefficients between the tree-ring width chronology and the relevant climate factors for the period 1960-2008 are shown in Fig.5. Tree growth-climate response patterns were very similar for the two slopes and generally the regional chronologies displayed higher correlations with temperature than the individual site chronologies. In particular, tree growth was positively correlated with the period April-September of the previous year,as well as February-March and May-July mean temperatures of the current year for the northern slopes(p ＜0.05).Tree growth also had a strong positive relationship to May-July mean temperature of the previous and current year in the southern slopes.On northern slopes,we observed a significant negative correlation between treering width chronology and precipitation of the previous April-May and the current March-April period;a similar relationship was also found on the southern slopes.

Seasonally averaged climate factors are often more representative of long-term climatic conditions than individual months;thus,to investigate common growth forces,we also screened PC#1RWand the individual tree-ring width chronologies in the correlation analyses with the seasonal combinations of climate factors from the previous April to the current September.The highest correlation was found between PC#1RWand May-July mean temperature of the current year(r=0.646,p ＜0.001).Meanwhile,we also found similar correlations for individual tree-ring chronologies with temperature(p ＜0.01).These results suggest that the growing season temperatures is the most critical limiting factor on tree growth in this area.

Spatial climate correlation analyses with gridded land surface climate data revealed that our tree-ring width chronologies contained a strong regional temperaturesignal over much of north-central and eastern China(Fig.6).Over the common period,1960-2008,significant positive correlations between the PC#1RWand the SST in the Pacific Ocean,the Indian Ocean,and the North Atlantic Ocean were evident (Fig.7a). Similarly, we observed correlations between PC#1RWand SST for the common period,1901-2008,although the correlations were lower(Fig.7b).

Fig.5 Correlation analysis of tree-ring width indices with mean temperature(Tm)and precipitation(Pm)in different slopes.The dashed lines indicate a significance level of 0.05

Discussion

Tree growth-climate relationships

Radial tree growth in cold and moist sites at high latitudes in the Northern Hemisphere has shown a strong correlation with temperature variability over large areas of Asia(Li et al.2011;Yu et al.2011;Zhu et al.2017).In this study,we also found that temperature played a more important role than precipitation in limiting the radial growth of L.chinensis,which is consistent with earlier findings(Chen et al.2011;Qin et al.2016).Temperature,particularly during the growing season, is a pronounced growthlimiting factor for L.chinensis at the timberline in the Qinling Mountains.At high altitudes,the limited responses to precipitation also suggest that forest trees are not subjected to much water-deficit stress.

Fig.6 Spatial correlations between the gridded mean(a),maximum(b),and minimum(c)temperature(May-July)from CRU and the PC#1RW over their overlapping periods(1960-2009)

Fig.7 Spatial correlations between the PC#1 of tree-ring widths and the gridded sea surface temperature(SST)dataset of HadISST1 over their overlapping periods from 1960-2008(a)and 1901-2008(b),respectively

The radial growth of L.chinensis was significantly and positively correlated (r=0.646, p ＜0.001) with the growing season mean temperature(May-July).A similar relationship between tree growth and summer mean temperature has been reported in the Qinling and Hengduan Mountains(Li et al.2011;Dang et al.2013;Chen et al.2014).Chen et al.(2014)studied the dendroclimatology of Abies faxoniana at the timberline in the western Qinling Mountains,located about 373 km west of our study site.They found that the tree-ring width chronology of A.faxoniana mainly responded to February-July mean temperature variability. Qin et al. (2016) also found that temperature,rather than precipitation,was a fundamental constraining climate factor for tree growth toward the upper tree lines of Taibai Mountain.

In the subalpine/alpine zone,a shorter growing season near the timberline is concentrated in July and August;higher summer temperatures promote earlier snow-melting and more rapid soil warming,thereby lengthening the snow-free growing season.In addition,higher temperatures during the growing season would enhance leaf area and the photosynthetic rate of trees. In contrast, low summer temperature at the timberline may limit cambial and fine root activity,which could also cause spring frost injuries of needles and shoots or even damage cambial tissues.

In contrast to our findings,Dang et al.(2007)found that summer precipitation significantly affects A.fargesii radial growth in the high-elevation areas in the Qinling Mountains.This discrepancy may be due to different altitudes of the sampling positions in the two investigations;previous studies were conducted in lower positions(2670-2760 m a.s.l.),while our sampling position was close to the upper timberline in the Qinling Mountains.Although relatively small changes in elevation do not affect tree growth by itself,it is a site factor influencing mean annual precipitation and temperature that do have physiological effects on tree growth(Fritts 1976).

Many studies have demonstrated the indirect effect of changes in altitude on tree growth in many regions of the world(Yu et al.2011;Sidor et al.2015).Above the timberline,where overall growth conditions are harsh,the annual mean temperature is generally lower than-2°C,strong winds and severe weather conditions(Dai et al.2001),thus the climate-growth relationships are different at the finer topographic scales.In addition,the shorter growth seasons near the alpine timberline on Taibai Mountain,concentrated in July and August,means that temperatures below freezing last from the previous October to the current April.Previous April temperature increases with the melting winter snow and rising soil temperatures,which profoundly alleviate spring drought and increases tree-ring width.

The effect of melting snow on plants is equivalent to precipitation,which can help explain the significant negative correlations between radial growth and precipitation in April of the current year.Lower temperatures and ample precipitation in April at the upper limit could restrict tree growth by decreasing the rate of photosynthesis and accumulation of plant nutrients(Fritts 1976).Furthermore,the significantly negative association between precipitation in the current April and tree growth may reflect responses to mechanical damage from sleet,ice,or snow.

Warm conditions in September may provide conditions that facilitate carbon storage,promote mycorrhizal root growth by maintaining soils above freezing,and favor maturation of needles,shoots,and buds against winter stressors and accelerate subsequent tree growth at the beginning of the next growing season.A stronger positive correlation between September temperature and radial growth has also been reported for P. koraiensis in Changbai Mountain(Yu et al.2011)and P.cembra in western Austria(Oberhuber 2004).

Linkages with the large-scale climate as illustrated in Fig.3,PC#1RWshows periods of reduced and enhanced growth over the 199-year period,which indicates decadal scale changes in the environment.On a decadal timescale,a 10-year FFT smoothing of the tree-ring width revealed that relatively major below-average growth periods(continuously below average for more than 10 years)occurred in 1813-1824,1844-1859,1878-1890,1909-1926,and 1963-1979. Major high-growth periods (continuously above average for more than 10 years) occurred in 1860-1877.Interpreting these in terms of the local climate might be premature,but the observed relationships with SST offer the hope of linking tree growth in the study to larger dynamics of the climate system.

The MTM spectral analysis revealed the existence of several important cycles in PC#1RW, which may help explain the tree-ring width variations in the study region.The 2.5-7.2-year peak cycle falls within the overall bandwidth of natural climate oscillations,such as the El Nin~o-Southern Oscillation(ENSO),suggesting that ENSO possibly exerted influence on local tree growth.In addition,the 11.7-12.0-year cycles might suggest the influence of solar effects, such as sunspot activity. Chen et al.(2015a,b)also found a possible connection of the radial growth of subtropical tree species with regional climate and SST variations in southeast China.

These high-frequency cycles also suggest the tree growth may have strong associations with large-scale ocean atmosphere-land circulation systems.As illustrated in Fig.7,there is a notable positive correlation with the Pacific Ocean,Indian Ocean,and North Atlantic Ocean SSTs.Additionally,positive correlations between the PC#1 of tree-ring width and SSTs in the Nino 3.4 region from the eastern equatorial Pacific Ocean indicate a possible connection between the growth of L.chinensis and ENSO.This is the same general pattern described for A.faxoniana by Chen et al.(2014)for the LLG site in the western Qinling Mountains.Anomalous warmer temperatures of the surface waters during El Nino years appear to be beneficial to tree radial growth by affecting the local climate regime over the sampling sites (Chen et al.2014,2015a,b).In summary,these results indicated that ENSO contributed to tree growth in this area.

To further establish whether the tree radial growth exhibited links with large-scale atmospheric circulation,composite maps of 500-hPa vector wind composite anomaly for the ring width of the 10 highest and 10 lowest value years during the period 1948-2012 were created.As shown in Fig.8,during the ring width of the lowest value years,the southeast monsoon is strong,warm,and a wet air mass control in the study area. Precipitation and soil moisture are increased,resulting in reduced mean temperatures in May-July,which inhibits tree-ring growth:this is consistent with high-level transport of warm and wet air from the Pacific Ocean by the strong Asian summer monsoon.

However,during the ring width of the highest value years,the southeast monsoon is weakened,the study area has a dry westerly flow,resulting in increased mean temperatures in May-July,which leads to wider tree-rings.As the dividing line between north China and south China,the climate of the Qinling Mountains may also have been affected by the Westerlies.In the eastern Qinling Mountains,Chen et al.(2015a,b)found that the Westerlies and ENSO might have had an effect on the Qinling Mountains.Liu et al.(2013)also reported that the variation of tree growth agreed well with other reconstructed temperature series from nearby and remote regions,suggesting that L.chinensis could respond to broad-scale climate variability.Significant correlations with the eastern equatorial Pacific Ocean SSTs support such a connection with the ENSO.However,the mechanism on how these two circulation systems(the Westerlies and Asian summer monsoon)interact and how they control tree growth of central China at various timescales requires further investigation.

Conclusion

Based on tree-ring width data of L.chinensis from northern and southern slopes in the Qinling Mountains of Shaanxi Province,we developed two tree-ring width chronologies.The use of two chronologies for dendroclimatological studies in the Qinling Mountains may lead to reconstruction of some climate factors for at least the past two centuries. The correlation analysis showed that similar correlations between tree-ring width chronologies and climatic factors demonstrated a common radial growth response of trees on the two slopes to climate change.The radial growth of L.chinensis was mainly limited by temperature,especially during the growing season.In contrast,both chronologies were negatively correlated with precipitation in May of the previous year and April of the current year.

However,in studies to this point,the calibration with climate data is not sufficiently strong to pass certain basic tests of reconstruction fidelity(not shown).The PCA and spatial correlation analyses with gridded land-surface climate data also revealed that May-July temperatures were the most important common influence on tree growth in the study area,which covers much of north-central and eastern China.

Spatial correlation with sea-surface temperature fields highlights the influence of the Pacific Ocean,Indian Ocean,and North Atlantic Ocean.Wavelet coherence analysis indicated the existence of some decadal and inter-annual cycles in the two tree-ring width chronologies.Power spectrum revealed the existence of significant frequency cycles of variability at 2.5-2.6,3.4-3.5,5.3-5.7,5.9,6.1,7.2-7.5, 7.8-8.9, and 11.7-12.0 years, which may be linked to large-scale atmospheric-oceanic variability,such as the El Nin~o-Southern Oscillation(ENSO)and solar activity.This may suggest the influences of ENSO and solar activity on tree growth in the Qinling Mountains.In conclusion,these findings will help us understand the growth response of this tree species to climate change in the Qinling region and provide critical information for future climate reconstructions using this species in semihumid regions.

AcknowledgementsThe authors express their thanks to Fei Wang,Henglu Yu,Jianming Guo,and You Du for their help with the sampling and measuring of tree cores.They also thank the reviewers for their diligence in providing useful comments that improved this work.