Increasing aridity over the past 223 years in the Nepal Himalaya inferred from a tree-ring δ 18 O chronology

Abstract

A tree-ring δ¹⁸O chronology of Abies spectabilis from the Nepal Himalaya was established to study hydroclimate in the summer monsoon season over the past 223 years (ad 1778–2000). Response function analysis with ambient climatic records revealed that tree-ring δ¹⁸O was primarily controlled by the amount of precipitation and relative humidity during the monsoon season (June–September). Since tree-ring δ¹⁸O was simultaneously correlated with temperature, drought history in the monsoon season was reconstructed by calibrating against the Palmer Drought Severity Index (PDSI). Our reconstruction that accounts for 33.7% of the PDSI variance shows a decreasing trend of precipitation/moisture over the past two centuries, and reduction of monsoon activity can be found across different proxy records from the Himalaya and Tibet. Spatial correlation analysis with global sea surface temperatures suggests that the tropical oceans play a role in modulating hydroclimate in the Nepal Himalaya. Although the dynamic mechanisms of the weakening trend of monsoon intensity still remain to be analyzed, rising sea surface temperatures over the tropical Pacific and Indian Ocean could be responsible for the reduction of summer monsoon.

Keywords

climate reconstruction dendrochronology Nepal oxygen isotope ratios Palmer Drought Severity Index summer monsoon

Introduction

The potential impacts resulting from hydrologic changes in monsoon Asia are significant since more than half of the world population depends on the recurrence of seasonal monsoon rainfall. Recent analyses based on instrumental data indicate an overall weakening trend of global land monsoon precipitation (Wang and Ding, 2006; Zhou et al., 2008b), with a decreasing trend found in South Asia. However, the meteorological records used in these studies are limited to the latter half of the 20th century. Therefore, proxy climate records extending back to the pre-instrumental era are required for placing the recent reduction of monsoon intensity in a long-term context with natural variations.

An increasing number of studies indicate that tree rings from the Himalaya provide comprehensive pictures of past climate variability over the past several centuries (e.g. Hughes, 1992; Yadav et al., 2004). However, ring widths used in the previous studies are mainly governed by pre-monsoon (spring) climate, and therefore reconstruction of the actual monsoon variability is rare. In addition, the standardization procedure that is used to remove non-climatic variability retained in ring-width measurements often results in loss of climate variations on centennial and longer timescales (Cook et al., 1995), although this can be largely overcome by proper treatment such as the regional curve standardization applied to a large number of ring-width series (e.g. Esper et al., 2002).

Recent progress in isotope dendroclimatology in the lower latitudes revealed that δ¹⁸O of tree cellulose is a promising proxy to reconstruct past precipitation and relative humidity, with or without visible annual rings (Anchukaitis and Evans, 2010; Anchukaitis et al., 2008b; Evans and Schrag, 2004; Managave et al., 2010; Poussart and Schrag, 2005; Poussart et al., 2004). This is because the oxygen isotope ratios of tree cellulose are mainly controlled by two factors, i.e. δ¹⁸O of source water (precipitation) and relative humidity (Roden et al., 2000). With regard to the former, a negative correlation with the amount of precipitation is observed in the lower latitudes, while a positive correlation is seen with temperature in the higher latitudes (Araguás-Araguás et al., 1998; Dansgaard, 1964). It is therefore expected in the lower latitudes that δ¹⁸O of tree cellulose shows a negative correlation with the amount of precipitation. On the other hand, a negative correlation is also observed between relative humidity and δ¹⁸O of tree cellulose (e.g. Evans and Schrag, 2004; Managave et al., 2010). The overall negative associations with relative humidity and the amount of precipitation in the pluvial monsoon regions indicate that tree-ring δ¹⁸O would be used as a pure proxy of past hydroclimate (wet–dry conditions).

Our preliminary investigations based on the same samples used in this study for the last 50 years reveled that tree-ring δ¹⁸O was much more sensitive to rainfall fluctuations than other tree-ring parameters such as width and density typically used in dendroclimatology (Sano et al., 2010). Here we present a robust 223 yr tree-ring δ¹⁸O chronology of Abies spectabilis, as a complementary proxy to ring width mostly used for climate reconstructions in the Himalaya. Our δ¹⁸O chronology that preserves low-frequency signals is used to develop a reconstruction of the Palmer Drought Severity Index (PDSI), a metric for moisture availability, for the monsoon season in the Nepal Himalaya, and describe its relation to other proxy records.

Theory

Trees take up soil water, which originates from precipitation, through roots without any isotopic fractionation (White et al., 1985), and then transport it to leaves through xylem. The oxygen isotopic composition in leaf water is governed by transpiration through stomata, which leads to a preferential loss of lighter isotopes and consequent enrichment in δ¹⁸O. This effect can be expressed as the modified Craig–Gordon equation:

δ^{18} O_{L} = δ^{18} O_{X} + ε * + ε_{k} + (δ^{18} O_{A} - ε_{k} - δ^{18} O_{X}) e_{a} / e_{i}

(1)

where δ¹⁸O_L, δ¹⁸O_X, and δ¹⁸O_A are the δ¹⁸O of leaf water, xylem water, and atmospheric water vapor, ϵ^* and ϵ_k are equilibrium and kinetic isotopic fractionation factors, and e_a and e_i are ambient and intercellular water vapor pressures. The equilibrium fractionation is due to the change of phase from liquid to vapor, and varies slightly with temperature; ϵ^* = 9.8‰ at 20°C (Majoube, 1971). The kinetic fractionation results from the diffusion of vapor into unsaturated air; ϵ_k ≈ 26.5‰ (Farquhar et al., 1989). Relative humidity (h) can be substituted for the vapor pressure ratio (e_a/e_i) (Dongmann et al., 1974), since leaf interstitial space is saturated with water vapor and leaf temperature is assumed to follow the ambient temperature. By assuming that atmospheric water vapor is isotopically equilibrated with precipitation, δ¹⁸O_A can be represented as δ¹⁸O_P – ϵ^*. δ¹⁸O_X is considered equal to δ¹⁸O of precipitation (δ¹⁸O_P). As a whole Eq. (1) can be simplified as:

δ^{18} O_{L} = δ^{18} O_{P} + (ε * + ε_{k}) (1 - h)

(2)

Biochemical fractionation associated with synthesis of sucrose in the leaf and the extent to which carbon-bound oxygen undergoes exchange with xylem water during cellulose synthesis affect the δ¹⁸O of cellulose. These processes can be summarized as follows (Roden et al., 2000; Sternberg 2009):

δ^{18} O_{C} = f (δ^{18} O_{X} + ε_{o}) + (1 - f) (δ^{18} O_{L} + ε_{o})

(3)

where δ¹⁸O_C is the δ¹⁸O of cellulose, f is the fraction of oxygen that exchanged with the xylem water (precipitation) during cellulose synthesis, ϵ_o is the biological fractionation factor between leaf water and sucrose during photosynthesis, and between the xylem water and the exchanged oxygen in the carbohydrate during cellulose synthesis. Values of f and ϵ_o are estimated to be 0.42 (Roden et al., 2000) and 27‰ (DeNiro and Epstein, 1981; Yakir and DeNiro, 1990), respectively.

Combining Eqs (2) and (3) gives an equation in terms of δ¹⁸O_C:

δ^{18} O_{C} = f (δ^{18} O_{P} + ε_{o}) + (1 - f) [δ^{18} O_{P} + (ε * + ε_{k}) (1 - h) + ε_{o}]

(4)

This model suggests that the δ¹⁸O of cellulose is mainly controlled by the δ¹⁸O of precipitation and relative humidity, since the other factors do not change significantly.

Materials and methods

Tree-ring data

Tree-ring samples were collected from A. spectabilis growing close to the treeline at 3850 m a.s.l. in Humla District, western Nepal (Figure 1). Firs in the site are considered to grow from March through September on the basis of climatic responses of their ring width and densities previously analyzed (Sano et al., 2005). Paired increment cores were bored at breast height (1.3 m above ground) from each of 23 trees for a total of 46 core samples.

Figure 1.

The study region showing the sampling site (29°51′N and 81°56′E), the Jumla and Mukteshwar meteorological stations, and ice-core sites of Dasuopu (Zhao and Moore, 2006) and Everest (Kaspari et al., 2007). Tree-ring (Grießinger et al., 2011) and lake sediment (Chu et al., 2011) sites in Tibet are shown in the upper map. The area of the four-gridbox average of the PDSI data (Dai et al., 2004) used in this study is enclosed by the solid line.

Cross-dating was performed by visually matching ring-width variations for all cores to determine the absolute year of each ring, and then was statistically checked by the software COFECHA (Holmes, 1983). The mean segment length of the dated series is 226.0 years with a standard deviation of 33.6 years. The mean inter-series correlation of standardized ring-width series for the common period of 1825–2000 is 0.25. Of the 40 cores precisely dated, ten cores from five trees were selected for isotope analysis. More specifically, samples that show ‘normal’ ring size, series length, and inter-series correlations among ring-width series were selected, in order to compare the strength of climatic signals between ring-width and δ¹⁸O chronologies without bias (Sano et al., 2010). The mean segment length of the selected five trees is 217 years with a standard deviation of 10.5 years, and the mean inter-series correlation of the ring-width indices for the selected five trees is 0.29.

Every annual ring of those ten cores (five tress) was split with a scalpel, and the resultant shavings were pooled for each year to obtain representative stable isotope records at the site. The δ¹⁸O chronology spans the time period from ad 1778 to 2000 (223 years), and is comprised of two to four trees during the 1778–1801 period and of all the five trees from 1802 to 2000 (see Figure 5).

To validate the pooling method as well as to compare the strength of common variations between ring-width and isotope series, single-tree measurements for the period of 1951–1980 were also conducted for two other trees, the ring widths (and corresponding indices) of which were less correlated with each other (R = 0.23 and −0.15, respectively). For evaluating coherence of these data, we calculated the Rbar (Cook and Kairiukstis, 1990), which is the mean inter-series correlation, and the expressed population signal or EPS (Wigley et al., 1984), which indicates how well the chronology estimates a theoretically infinite population.

After the samples were finely milled using a ball mill, cellulose was extracted by a modified method based on the protocols of Brendel et al. (2000) and Anchukaitis et al. (2008a). Oxygen isotope ratios (¹⁸O/¹⁶O) of cellulose samples were determined by an isotope ratio mass spectrometer (ThermoQuest Delta Plus) interfaced with a pyrolysis-type elemental analyzer (ThermoQuest TC/EA). The ¹⁸O/¹⁶O ratios were expressed as δ¹⁸O (‰) deviation relative to the VSMOW international standard. The standard deviation derived from repeatedly measured standard material was 0.31‰.

Climate analyses

We employed linear correlation analyses between the tree-ring δ¹⁸O and monthly climate data with a 19-month window (prior March through current September) to identify the climatic response. For this purpose, the rainfall records (ad 1957–1996) of the Jumla station (Figure 1; 70 km south of the site) nearest to the sampling site were utilized. While it is desirable to use the temperature and relative humidity data from the same station, the 18-year observations of the temperature and the lack of relative humidity data cannot be utilized for a meaningful correlation analysis, and therefore the temperature (ad 1898–1991) and relative humidity records (ad 1933–1991) from Mukteshwar (200 km west of the site) were used instead. In order to examine relative effects of each climatic factor on the δ¹⁸O variations, correlations were calculated for the period of ad 1957–1991, being common to all the factors. Then we computed correlations using the entire period available for each observation.

The mean annual precipitation in the study region is 796 mm, 66% of which falls in the summer monsoon months of June through September. Rainfall in the pre-monsoon (March–April) and post-monsoon (October–November) seasons together contributes 24% of annual precipitation. The mean annual temperature is 13.5°C with a maximum of 18.7°C in June and a minimum of 6.2°C in January. The mean annual relative humidity is 63.9 % with a maximum of 92.2% in August and a minimum of 50.1% in December.

We also correlated the δ¹⁸O chronology with the 2.5°×2.5° gridded global Palmer Drought Severity Index (PDSI) series produced by Dai et al. (2004). The PDSI is a drought metric based upon a water balance model for which temperature, precipitation and soil characteristics are considered (Palmer, 1965), whereby positive and negative values of the PDSI correspond respectively to wet and dry conditions of soil moisture. While one would expect that the single gridpoint containing the study site provides the most representative data for the region, our preliminary analysis reveled that records from the neighboring gridpoints located in India showed higher correlations with our δ¹⁸O chronology. This is because instrumental station data are considerably limited in Nepal, as compared with those in India. In addition, precipitation variability often differs among stations even within a short distance because of the complex topography, indicating that the single gridpoint of the PDSI cannot be assumed to represent the local climate of the sampling site. We therefore generated an average from the four gridpoints (Figure 1), in which a dense network of meteorological stations has been developed in India, in order to obtain more reliable records for this analysis.

Results and discussion

Coherence of tree-ring δ¹⁸O

As shown in Figure 2, δ¹⁸O variations for the period of 1951–1980 were strongly correlated between the pooled and single-tree series (Rbar = 0.80; the EPS = 0.92, N = 3), even though their ring-width-index variations were weakly matched (Rbar = 0.27; EPS = 0.53, N = 3). Since we selected two trees for the single-tree measurement on the basis of a weak correlation of their ring widths between each other, the Rbar of 0.27 was inevitably lower than Rbars computed for other trees that were not utilized for the present study. Nevertheless, the Rbar of 0.80 for the δ¹⁸O was apparently higher than that of 0.32 calculated using ring-width indices of all the 20 trees. It therefore seems reasonable to conclude that coherence of δ¹⁸O variations is higher than that of ring-width variations, and the pooling method leads further to maximizing the signal. The lower correlation of the ring width could stem from the fact that radial growth is usually affected by endogenous disturbance pulses, such as competition with neighboring trees (Fritts, 1976).

Figure 2.

Plots of (a) ring-width index and (b) tree-ring δ¹⁸O for the same trees (two single-tree series and one five-tree pooled series).

The EPS of 0.92 calculated using three time series for the δ¹⁸O attains the generally accepted threshold value of 0.85 or greater. Though we cannot calculate EPS values before ad 1950 because of the lack of sample replication, our pooled chronology that is comprised of two to five trees would attain EPS > 0.85 over the entire period of 1778–2000 if we applied the Rbar of 0.80 to these sample numbers (EPS = 0.89 for two trees, 0.95 for five trees). Therefore our chronology is considered robust over the entire period of 1778–2000.

Climate signals in tree-ring δ¹⁸O

Responses of the tree-ring δ¹⁸O to monthly rainfall, temperature and relative humidity for the common period of ad 1957–1991 are shown in Figure 3a–c. The δ¹⁸O was correlated negatively with rainfall in June, August and September, positively with temperature in June, and negatively with relative humidity in June–September (P < 0.05). To simplify climate signals in the chronology, we also correlated the δ¹⁸O with June–September (monsoon) rainfall, temperature and relative humidity. It turned out that the δ¹⁸O was correlated negatively with monsoon rainfall (R = −0.60, P < 0.001), positively with temperature (R = 0.35, p < 0.05), and negatively with relative humidity (R = −0.62, P < 0.001). On the other hand, relatively higher correlations for the common period were seen with the PDSI in June, July, August and September, with a highest correlation found in the June–September PDSI (Figure 3d; R = −0.73, P < 0.001). Climatic responses using the entire period available for each observation (Figure 3e–h) are broadly consistent with those calculated for the common period, indicating that the relationships between climate and tree-ring δ¹⁸O are temporally stable. It is therefore quite certain that the PDSI, rather than rainfall, temperature or relative humidity, is an appropriate single parameter to explain the δ¹⁸O variance.

Figure 3.

Responses of tree-ring δ¹⁸O to (a) Jumla precipitation, (b) Mukteshwar temperature and (c) relative humidity, and (d) the PDSI records for the common period of ad 1957–1991. Right column (e–h) is same as left one, but for the entire period available for each observation.

As explained in the previous section, δ¹⁸O of tree cellulose depends mainly on two factors, i.e. δ¹⁸O of precipitation and relative humidity (e.g. Ramesh et al., 1986; Robertson et al., 2001; Roden et al., 2000; Saurer et al., 1997). Regarding the former, a negative correlation between δ¹⁸O of precipitation and the amount of precipitation is observed in tropical regions including the Himalaya, which is termed as the ‘amount effect’ (Araguás-Araguás et al., 1998; Dansgaard, 1964; Yadava and Ramesh, 2005). The significant negative correlation of the tree-ring δ¹⁸O with monsoon rainfall indicates that our sampled trees certainly record δ¹⁸O of precipitation. This is further confirmed by a significant correlation (R = 0.49; P < 0.01) between weighted mean δ¹⁸O of June–September precipitation at New Delhi and δ¹⁸O of the tree rings (Figure 4). It is clear, however, that the correlation of 0.49 is not high enough to explain variations in tree-ring δ¹⁸O solely by the δ¹⁸O of precipitation (and hence the amount of precipitation). This is because relative humidity also affects tree-ring δ¹⁸O, as seen with the significant negative correlation between them. More specifically, lower atmospheric humidity causes higher vapor pressure gradient between leaf interstitial space and the atmosphere, resulting in a preferential loss of lighter isotopes and consequent enrichment in δ¹⁸O of leaf water. When relative humidity and precipitation are used as predictors in multiple regression analysis, the tree-ring δ¹⁸O is shown to be highly significant (Multiple R = 0.72; P < 0.001). It therefore seems reasonable to conclude that the tree-ring δ¹⁸O is largely governed by these two climatic factors.

Figure 4.

Plots of inter-annual variations in weighted mean δ¹⁸O of June–September precipitation at New Delhi (Global Network of Isotopes in Precipitation, http://www-naweb.iaea.org/napc/ih/GNIP/IHS_GNIP.html) and tree-ring δ¹⁸O. The gaps represent missing data.

The relatively weak but significant positive correlation with temperatures in the monsoon season indicates one minor effect on the tree-ring δ¹⁸O. Our interpretation of the temperature signal in the δ¹⁸O chronology is that higher temperatures stimulate evaporation of soil water, resulting in a preferential loss of lighter isotopes in soil water, which in turn is taken up by trees. Since all the responses of tree-ring δ¹⁸O to the climatic factors (precipitation, relative humidity, and temperature) are related to dry–wet conditions, the PDSI is considered to be the most useful single variable to explain the variance of our δ¹⁸O chronology.

Reconstruction of the PDSI

Based on the results of response analyses, we finally identified that the monsoon season during June–September optimized the PDSI signal in the chronology (N = 104; R = −0.58; P = 1.0 × 10^-10). Hence, the June–September PDSI was used to develop a linear regression for statistical reconstruction. In addition to the linear regression, the scaling method was also applied to our δ¹⁸O chronology for PDSI reconstruction. ‘Scaling’ refers to the equalization of the mean and the standard deviation of a proxy time series to those of an instrumental record for an overlapped period (Esper et al., 2005). While the variance of a reconstructed time series derived from the least squares linear regression is inevitably less than that of an instrumental record, the ‘scaling’ procedure retains the same variance of the instrumental record in the reconstructed record, at the expense of inflated error variance.

The regression model using the entire PDSI data set of 1897–2000 was statistically significant (Table 1). We further scrutinized the integrity of the model by split-period calibration-verification tests used in dendroclimatology (Cook and Kairiukstis, 1990; Fritts, 1976). As shown in Table 1, two most rigorous verification tests, i.e. the reduction of error (RE) and the coefficient of efficiency (CE) were positive, as well as all the other tests were shown to be statistically significant. Therefore, these results sufficiently proved the validity of the regression model. Based on the model that accounts for 33.7% of the actual PDSI variance, summer monsoon droughts were reconstructed back to ad 1778 (Figure 5).

Table 1.

Calibration and verification statistics for the common period of 1897–2000.

Calibration period	r	R ²	Verification period	r	RE	CE	Sign test (+/−)
Full (1897–2000)	0.581**	0.337	–	–	–	–	–
Early half (1897–1948)	0.613**	0.377	Late half (1949–2000)	0.655**	0.156^a	0.110^a	42/10**
Late half (1949–2000)	0.655**	0.429	Early half (1897–1948)	0.613**	0.120^a	0.083^a	36/16*

RE, CE>0.

p<0.01, **p<0.001.

Figure 5.

Plots of (a) the actual and reconstructed June–September PDSI for the overlapped period of ad 1897–2000, (b) the reconstructed PDSI with sample depth, and (c) regressed/scaled reconstructions over the entire period of ad 1778–2000. The thick curves represent the 30-year cubic spline.

Amplitudes of the reconstructed PDSI measured from the driest (1986–1995) to wettest (1794–1803) decades were 2.78 and 4.78, respectively, for regression and scaling methods (Figure 5c). It is quite certain that the regression and scaling approaches significantly affect the absolute PDSI amplitude. Though we cannot conclude which method is appropriate to capture the past PDSI, the range of these changes must be kept in mind when directly comparing absolute values among various reconstructions.

The reconstructed June–September PDSI retains low-frequency variations (defined as components greater than 50 years), in which a decreasing trend over the past two centuries was most apparent (Figure 6a). The linear regression of the reconstructed PDSI was statistically significant (N = 223; R = −0.32; P = 8.3 × 10^-7), and the decreasing rates were −0.68 and −1.17 PDSI per 100 years, respectively, for the regression and scaling reconstructions. It should be duly noted that potential biases related to aging of trees have been reported (Esper et al., 2010; Nakatsuka, 2007; Treydte et al., 2006). More specifically, Treydte et al. (2006) identified an age-related declining trend for δ¹⁸O of juniper by analyzing biologically younger and older tree rings covering the same time period, but this had little effect on their millennium-long climate reconstruction. Treydte et al. (2006) recognized younger rings by systematically wider rings, so that the earliest 40 years (before ad 1820) of our chronology may also reflect an age-related trend. Interestingly, our chronology shows an overall increasing trend of δ¹⁸O values, while all aging effects identified in the previous studies (Esper et al., 2010; Nakatsuka, 2007; Treydte et al., 2006) are seen as declining trends. Since the declining trends are found even in different species and locations, with relatively high and constant sample replication in the case of Esper et al. (2010), aging effect does not seem to show increasing but declining δ¹⁸O. Therefore, if we applied the declining effect to our chronology, δ¹⁸O values before c. 1820 would be lower than those obtained in this study, and hence higher the PDSI in the reconstruction. However, it is equally plausible that aging effect does not significantly appear in our δ¹⁸O chronology. In fact, Young et al. (2011) have recently reported that age-related trends of 28 pine trees are not seen with tree rings whose cambial ages exceed c. 50 years. Whichever the case, it seems reasonable to conclude that the increasing (decreasing) trend of the δ¹⁸O chronology (the reconstructed PDSI) originates from climatic signals, and this is consistent with other proxy records from the Himalaya and Tibet, as will be discussed below.

Figure 6.

Tree-ring δ¹⁸O from (a) western Nepal (this study) and (b) Tibet (Grießinger et al., 2011), (c) ice-core δD from Mt Everest (Kaspari et al., 2007), (d) snow accumulation from Dasuopu (Zhao and Moore, 2006), and (e) verve thickness from Tibet (Chu et al., 2011). The vertical axes in (a–c) are inverted here owing to the negative association of δ¹⁸O and δD with the amount of precipitation. All time series are individually normalized to have mean 0 and variance 1, and smoothed using a low-pass filter (frequencies greater than 50 years retained). Note that original records of (b) and (c) cover the periods of the last 842 years and the last millennium, respectively.

Comparison with other proxy records

The moisture decrease appearing in our chronology is consistent with reduction in summer monsoon rainfall instrumentally observed over Northeast India since the late 19th century (Rupa Kumar et al., 1992), and over the western Himalaya since the 1960s (Basistha et al., 2008). As shown in Figure 6, decrease in summer monsoon rainfall is also found in other proxy records, i.e. (b) δ¹⁸O of tree rings from Tibet (Grießinger et al., 2011), (c) δD of an ice core from Mt Everest (Kaspari et al., 2007), (d) snow accumulation of an ice core from Dasuopu (Zhao and Moore, 2006), (e) verve thickness of lake sediment in Tibet (Chu et al., 2011). Though multidecadal variations do not match among these records because of different localities and environmental sensitivities depending on proxy material being analyzed, the overall trends toward arid conditions indicate that summer monsoon has weakened over a wide area of the Himalaya and Tibet for at least the last couple of centuries. In contrast, Treydte et al. (2006) reported that the 20th century was the wettest period over the past millennium in northern Pakistan. Since their study area is not dominated by monsoon but by westerly synoptic fronts throughout the year, with maximum precipitation appearing in winter and spring, it is not surprising that their record shows the opposite trend compared with the monsoon-related reconstructions. Interestingly, these consistent trends toward dry/wet conditions are concurrent with global warming, and thus may indicate some causal relationship between them.

Our reconstructed PDSI shows a negative correlation with sea surface temperature (SST) over most of the tropical Pacific and Indian Ocean, in particular the central Pacific (Figure 7), suggesting that the tropical oceans play a role in modulating hydroclimate in the Himalaya. In fact, a modeling study indicates that a weakening trend of monsoon precipitation found in northern India and the eastern Tibetan Plateau over the latter half of the 20th century can result from an increase in SST over the tropical Pacific and Indian Ocean (Zhou et al., 2008a). When coupled with the fact that a consistent warming over the last two centuries was found in a coral-based reconstruction of large-scale tropical SSTs (Wilson et al., 2006), the SST forcing can explain the overall decrease trend in our reconstructed PDSI. It should be noted, however, that other proxy records from lower altitudes indicate a strengthening of the monsoon winds (Anderson et al., 2002) and precipitation (Wang et al., 2005). Kaspari et al. (2007) indicate that north–south differences of monsoon intensity are attributed to a southward shift in the overall position of the monsoon trough. The present study suggests that elevated SST over the tropical oceans may be responsible for the increased north–south differences over the couple of centuries.

Figure 7.

Spatial correlation field comparing the reconstructed PDSI with SSTs (ERSST v3., Smith et al., 2008) for the period of ad 1901–2000. This figure was generated using KNMI Climate Explorer (http://climexp.knmi.nl/).

Footnotes

Acknowledgements

We are grateful to Drs LG Thompson, GWK Moore, S Kaspari, J Grießinger and G Chu for generously providing us with their proxy records, and to Mr Nagabushana for his tireless help during the course of this study. MS thanks the Director, PRL for permission to be associated with the Physical Research Laboratory.

This study was funded by the Sasakawa scientific research grant (No. 21-528) from the Japan Science Society. The funding for analyses was from ISRO-GBP, Govt. of India.

References

Anchukaitis

Evans

(2010) Tropical cloud forest climate variability and the demise of the Monteverde golden toad. Proceedings of the National Academy of Sciences 107: 5036–5040.

Anchukaitis

Evans

Lange

. (2008a) Consequences of a rapid cellulose extraction technique for oxygen isotope and radiocarbon analyses. Analytical Chemistry 80: 2035–2041.

Anchukaitis

Evans

Wheelwright

(2008b) Stable isotope chronology and climate signal calibration in neotropical montane cloud forest trees. Journal of Geophysical Research 113: G03030, DOI: 10.1029/2007JG00061310.1029/2007JG000613.

Anderson

Overpeck

Gupta

(2002) Increase in the Asian southwest monsoon during the past four centuries. Science 297: 596–599.

Araguás-Araguás

Froehlich

Rozanski

(1998) Stable isotope composition of precipitation over southeast Asia. Journal of Geophysical Research 103: 28,721–28,742.

Basistha

Arya

Goel

(2008) Analysis of historical changes in rainfall in the Indian Himalayas. International Journal of Climatology doi: 10.1002/joc.170610.1002/joc.1706.

Brendel

Iannetta

PPM

Stewart

(2000) A rapid and simple method to isolate pure alpha-cellulose. Phytochemical Analysis 11: 7–10.

Chu

Sun

Yang

. (2011) Evidence for decreasing South Asian summer monsoon in the past 160 years from varved sediment in Lake Xinluhai, Tibetan Plateau. Journal of Geophysical Research 116: D02116, doi: 10.1029/2010JD01445410.1029/2010JD014454.

Cook

Kairiukstis

(1990) Methods of Dendrochronology. Dordrecht: Kluwer.

10.

Cook

Briffa

Meko

. (1995) The ‘segment length curse’ in long tree-ring chronology development for palaeoclimatic studies. The Holocene 5: 229–237.

11.

Dai

Trenberth

Qian

(2004) A global dataset of Palmer Drought Sensitivity Index for 1870–2002: Relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology 5: 1117–1130.

12.

Dansgaard

(1964) Stable isotopes in precipitation. Tellus 16: 436–468.

13.

DeNiro

Epstein

(1981) Isotopic composition of cellulose from aquatic organisms. Geochimica et Cosmochimica Acta 45: 1885–1894.

14.

Dongmann

Nürnberg

Förstel

. (1974) On the enrichment of H₂¹⁸O in the leaves of transpiring plants. Radiation and Environmental Biophysics 11: 41–52.

15.

Esper

Cook

Schweingruber

(2002) Low-frequency signals in long tree-ring chronologies for reconstructing past temperature variability. Science 295: 2250–2253.

16.

Esper

Frank

Battipaglia

. (2010) Low-frequency noise in δ¹³C and δ¹⁸O tree ring data: A case study of Pinus uncinata in the Spanish Pyrenees. Global Biogeochemical Cycles 24: GB4018, doi:10.1029/2010GB00377210.1029/2010GB003772.

17.

Esper

Frank

Wilson

RJS

. (2005) Effect of scaling and regression on reconstructed temperature amplitude for the millennium. Geophysical Research Letters 32: L07711, doi:10.1029/2004GL02123610.1029/2004GL021236.

18.

Evans

Schrag

(2004) A stable isotope-based approach to tropical dendroclimatology. Geochimica et Cosmochimica Acta 68: 3295–3305.

19.

Farquhar

Hubick

Condon

. (1989) Carbon isotope fractionation and plant water-use efficiency. In: Rundel

Ehleringer

Nagy

(eds) Stable Isotopes in Ecological Research. New York: Springer-Verlag, pp. 21–40.

20.

Fritts

(1976) Tree Rings and Climate. New York: Academic Press.

21.

Grießinger

Bräuning

Helle

. (2011) Late Holocene Asian summer monsoon variability reflected by δ¹⁸O in tree-rings from Tibetan junipers. Geophysical Research Letters 38: L03701, doi:10.1029/2010GL04598810.1029/2010GL045988.

22.

Holmes

(1983) Computer-assisted quality control in tree-ring dating and measurement. Tree-ring Bulletin 43: 69–78.

23.

Hughes

(1992) Dendroclimatic evidence from the western Himalaya. In: Bradley

Jones

(eds) Climate Since A.D. 1500. London: Routledge, pp. 415–431.

24.

Kaspari

Mayewski

Kang

. (2007) Reduction in northward incursions of the South Asian monsoon since ~ 1400 AD inferred from a Mt. Everest ice core. Geophysical Research Letters 34: L16701, doi:10.1029/2007GL03044010.1029/2007GL030440.

25.

Majoube

(1971) Fractionnement en oxygen-18 et en deuterium entre l’eau et sa vapeur. Journal of de Chimie Physique et de Physico-Chimie Biologique 68: 1423–1436.

26.

Managave

Sheshshayee

Borgaonkar

. (2010) Past break-monsoon conditions detectable by high resolution intra-annual δ¹⁸O analysis of teak rings. Geophysical Research Letters 37: L05702, doi:10.1029/2009GL04117210.1029/2009GL041172.

27.

Nakatsuka

(2007) Reconstruction paleo-climate using tree-ring oxygen isotopic ratios. Low Temperature Science 65: 49–56 (in Japanese, with English abstract).

28.

Palmer

(1965) Meteorological Drought. Washington DC: U.S. Department of Commerce, Weather Bureau Research Paper No. 45, 58 pp.

29.

Poussart

Schrag

(2005) Seasonally resolved stable isotope chronologies from northern Thailand deciduous trees. Earth and Planetary Science Letters 235: 752–765.

30.

Poussart

Evans

Schrag

(2004) Resolving seasonality in tropical trees: Multi-decade, high-resolution oxygen and carbon isotope records from Indonesia and Thailand. Earth and Planetary Science Letters 218: 301–316.

31.

Ramesh

Bhattacharya

Gopalan

(1986) Climatic correlations in the stable isotope records of silver fir (Abies pindrow) trees from Kashmir, India. Earth and Planetary Science Letters 79: 66–74.

32.

Robertson

Waterhouse

Barker

. (2001) Oxygen isotope ratios of oak in east England: Implications for reconstructing the isotopic composition of precipitation. Earth and Planetary Science Letters 191: 21–31.

33.

Roden

Lin

Ehleringer

(2000) A mechanistic model for interpretation of hydrogen and oxygen isotope ratios in tree-ring cellulose. Geochimica et Cosmochimica Acta 64: 21–35.

34.

Rupa Kumar

Pant

Parthasarathy

. (1992) Spatial and subseasonal patterns of the long-term trends of Indian summer monsoon rainfall. International Journal of Climatology 12: 257–268.

35.

Sano

Furuta

Kobayashi

. (2005) Temperature variations since the mid-18th century for western Nepal, as reconstructed from tree-ring width and density of Abies spectabilis. Dendrochronologia 23: 83–92.

36.

Sano

Sheshshayee

Managave

. (2010) Climatic potential of δ¹⁸O of Abies spectabilis from the Nepal Himalaya. Dendrochronologia 28: 93–98.

37.

Saurer

Borella

Leuenberger

(1997) δ¹⁸O of tree rings of beech (Fagus silvatica) as a record of δ¹⁸O of the growing season precipitation. Tellus 49B: 80–92.

38.

Smith

Reynolds

Peterson

. (2008) Improvements to NOAA’s historical merged land–ocean surface temperature analysis (1880–2006). Journal of Climate 21: 2283–2296.

39.

Sternberg

LSL

(2009) Oxygen stable isotope ratios of tree-ring cellulose: The next phase of understanding. New Phytologist 181: 553–562.

40.

Treydte

Schleser

Helle

. (2006) The twentieth century was the wettest period in northern Pakistan over the past millennium. Nature 440: 1179–1182.

41.

Wang

Ding

(2006) Changes in global precipitation over the past 56 years. Geophysical Research Letters 33: L06711, doi:10.1029/2005GL02534710.1029/2005GL025347.

42.

Wang

Cheng

Edwards

. (2005) The Holocene Asian monsoon: Links to solar changes and North Atlantic climate. Science 308: 854–857.

43.

White

JWC

Cook

Lawrence

. (1985) The D/H ratios of sap in trees: Implications for water sources and tree ring D/H ratios. Geochimica et Cosmochimica Acta 49: 237–246.

44.

Wigley

TML

Briffa

Jones

(1984) On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology 23: 201–213.

45.

Wilson

Tudhope

Brohan

. (2006) Two-hundred-fifty years of reconstructed and modeled tropical temperatures. Journal of Geophysical Research 111: C10007, doi:10010.11029/12005JC00318810010.11029/12005JC003188.

46.

Yadav

Park

W-K

Singh

. (2004) Do the western Himalayas defy global warming? Geophysical Research Letters 31: L17201, doi:10.1029/2004GL02020110.1029/2004GL020201.

47.

Yadava

Ramesh

(2005) Monsoon reconstruction from radiocarbon dated tropical Indian speleothems. The Holocene 15: 48–59.

48.

Yakir

DeNiro

(1990) Oxygen and hydrogen isotope fractionation during cellulose metabolism in Lemna gibba L. Plant Physiology 93: 325–332.

49.

Young

GHF

Demmler

Gunnarson

. (2011) Age trends in tree ring growth and isotopic archives: A case study of Pinus sylvestris L. from northwestern Norway. Global Biogeochemical Cycles 25: GB2020, doi:10.1029/2010GB00391310.1029/2010GB003913.

50.

Zhao

Moore

GWK

(2006) Reduction in Himalayan snow accumulation and weakening of the trade winds over the Pacific since the 1840s. Geophysical Research Letters 33: L17709, doi:10.1029/2006GL02733910.1029/2006GL027339.

51.

Zhou

. (2008a) Ocean forcing to changes in global monsoon precipitation over the recent half-century. Journal of Climate 21: 3833–3852.

52.

Zhou

Zhang

(2008b) Changes in global land monsoon area and total rainfall accumulation over the last half century. Geophysical Research Letters 35: L16707, doi:10.1029/2008GL03488110.1029/2008GL034881.

Increasing aridity over the past 223 years in the Nepal Himalaya inferred from a tree-ring δ 18 O chronology

Abstract

Keywords

Introduction

Theory

Materials and methods

Tree-ring data

Climate analyses

Results and discussion

Coherence of tree-ring δ18O

Climate signals in tree-ring δ18O

Reconstruction of the PDSI

Comparison with other proxy records

Footnotes

Acknowledgements

References

Coherence of tree-ring δ¹⁸O

Climate signals in tree-ring δ¹⁸O