1 Introduction

Warm-season droughts are recurrent natural phenomena that have drastic implications for water resources, agriculture productivity, and the stability of the natural ecosystem. A comprehensive understanding of long-term drought variability and their underlying climatic mechanism is of paramount essential for formulating effective strategies in water management and climate change adaptation. In recent decades, there has been an alarming increase in both the severity and frequency of extreme drought events, resulting in substantial economic losses on a global scale (Easterling et al. 2000; Huang et al. 2016; van der Schrier et al. 2013). Several studies hypothesized that drought is one of the most devastating climate disasters, exerting a profound influence on subsistence systems and water resources, leading to societal failures (Gaire et al. 2019; Pederson et al. 2014). Hence, to address the impact of drought on water resources more informedly, long-term drought variability and the potential climatic mechanism that instigates these events become a significant concern (Gaire et al. 2019; IPCC 2007).

The South Asian summer monsoon (SASM) is an integral component of the global climate system (Cook et al. 2010). This phenomenon serves as a dynamic and crucial source of moisture, significantly influencing the Indian subcontinent and the adjacent mountainous regions, including central-northern Pakistan (Betzler et al. 2016; Webster et al. 1998). The SAMS exhibits distinct phases; its active phase is characterized by intense precipitation, resulting in heavy snowfall and floods. In contracts, the break phase is marked by drought, thereby leading to water scarcity issues. The linkage between the atmospheric patterns and the SAMS, particularly the impact of the Atlantic Multi-decadal Oscillation (AMO), El Niño-Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), and solar activity, highlights the relationship between global and regional climate drivers (Ahmad et al. 2020; Kumar et al. 2006; Malik et al. 2017; Sinha et al. 2011, 2007; Webster et al. 1998). Hence, understanding of these linkages is essential for predicting extreme weather events and their socioeconomic implication.

Central-northern Pakistan is located between the majestic peaks of the Himalayan and Hindu Kush in the north and spread over the Indus River basin and Karakoram to the south. The region supports the world’s largest irrigation network, sustaining millions of livelihoods. Drought reconstruction in central-northern Pakistan using natural proxies such as tree-ring is essential for understanding historical climate variability and its impact on the region. In recent decades, tree-ring-based reconstruction studies have been undertaken in northern Pakistan. For example, some previous studies unlock our understanding of the historical streamflow of the River Indus (Chen et al. 2021; Cook et al. 2013; Rao et al. 2018). The precipitation reconstruction, based on tree-ring width and oxygen isotope chronologies, provides a comprehensive understanding of historical precipitation patterns (Khan et al. 2019; Treydte et al. 2006). However, SASM relieves drier conditions (Wang et al. 2021), while the westerlies bring drier and cooler conditions during winter in northern Pakistan (Treydte et al. 2006). Therefore, knowledge of long-term drought variability in the context of SASM and westerlies is crucial.

Some warm-season temperature reconstruction studies have been carried out in northern Pakistan (Asad et al. 2016; Khan et al. 2021; Zafar et al. 2015). In addition, Ahmad et al. (2020) carried out PDSI reconstruction based on tree-ring Cedrus deodara from the Chitral region of Pakistan. However, the study provided valuable information on past drought variability and their teleconnection with AMO, Monsoon Asia Drought Atlas (Cook et al. 2010), and ENSO. Indeed, it is essential to highlight that their study does not address the impact of the SASM and solar activity on regional drought patterns. These factors are the key drivers of climate fluctuation in the region and need special attention. In addition, regional to large-scale comparisons are vital to obtain broader implications of reconstructed scPDSI. Therefore, in the current study, we sought to answer how Abies pindrow responds to scPDSI. We hypothesized that warm-season (April-July) scPDSI could significantly influence the radial growth of Abies pindrow and could be a reliable proxy for the reconstruction of the history of scPDSI. To evaluate our hypothesis, we developed the following objectives: (1) to build a new tree-ring chronology from the understudy region of central-northern Pakistan, (2) to evaluate the relationship between radial growth of Abies pindrow and scPDSI, (3) to reconstruct the long-term scPDSI variability from central-northern Pakistan, and (4) to demonstrate the possible climate mechanism which significantly affects the regional scPDSI.

2 Materials and methods

2.1 Sampling sites and climate

The sampling sites include two stands of Abies pindrow, namely Kot (35° 26′ N, 72° 38′ E, 2430–2734 m a.s.l) and Tuti (35° 36′ N, 72° 59′ E, 2650–2850 m a.s.l) from the Kandia Valley, central-northern Pakistan (Fig. 1 and Table 1). Generally, the lower altitude zone of this region is characterized as drier or arid, whereas the higher altitude zone is moist or humid (Mayer et al. 2010; Winiger et al. 2005). For example, the lower altitude zone (1454 m) of this region received less than the total average precipitation of 137 mm, while more than 720 mm of average precipitation occurred in the high altitude zone (4120 m) (Mayer et al. 2010; Treydte et al. 2006). The monthly climate data of Gilgit station (Fig. 1c) for the period 1955–2017 CE indicate that high precipitation occurred in April (22.7 mm) and May (26.3 mm), and the lowest precipitation was recorded in January (4.8 mm) and December (6.9 mm). The mean monthly maximum temperature (36.1 °C) was observed in July, whereas the mean monthly minimum temperature (− 2.6 °C) was recorded in January.

Fig. 1
figure 1

a Map of Pakistan and its neighboring countries, b tree-ring sampling sites (Kot and Tuti) for Abies pindrow in the Kandia valley of central-northern Pakistan denoted by green triangles, and meteorological station in the Gilgit region marked by red triangles. c Climograph of meteorological station near our study area, for the period 1955–2017 CE. In the left map (a), blue numbers 1, 2, 3, 4, 5, and 6 correspond to the study sites of Ahmad et al. (2020), Khan et al. (2019), Treydte et al. (2006), Singh et al. (2022), Yadav et al. (2017), and Cook et al. (2010), respectively

Full size image
Table 1 Sampling sites informations and statistical features of standardized tree-ring width chronologies from the Kandia Valley of central-northern Pakistan
Full size table

In northern Pakistan, there are two observation stations, namely Gilgit (35° 55′ N, 74° 20′ E, 1454 m a.s.l) and Astore (35° 24′ N, 74° 54′ E, 2000 m a.s.l.). These observation stations are relatively distant (Gilgit = 256 km; Astore = 233 km) from our study area. The observation records from these stations contain some missing data, which we addressed by averaging data from the prior year and subsequent year. No fixed distance for the observation station is universally applicable. However, the observation station located in proximity (10–20 km) to the sampling site is mostly preferred. This approach is more likely to present the microclimate condition experienced by taxa. Due to the considerable distance between our sampled sites and observation stations, we obtained a climate dataset from CRU TS4.07 (averaged over 33.5°–34.5°N, 72.0–73.0°E) via KNMI climate explorer (van der Schrier et al. 2013). The climate data (temperature and precipitation) from CRU TS4.07 and the Gilgit observation station suggested a significant relationship (p > 0.01), whereas no meaningful relationship was observed between CRU TS4.07 and the Astore observation station. The Z-scores analysis of climate data from Gilgit and CRU TS4.07 demonstrated an increase in temperature and rainfall in recent decades, while the data from the Astore observation station showed a decline in rainfall (Fig. 2). A significant decline in rainfall was observed from 1955 to 1972 CE.

Fig. 2
figure 2

Graphical comparison (z-scores) between rainfall and temperature of (a) Gilgit (GLT), (b) CRU TS4.07, and (c) Astore (AST) stations. The green and orange bars represent moist and drought conditions, respectively, and the bold line indicates temperature fluctuation over time

Full size image

2.2 Tree-ring data

Abies pindrow (Royle), also known as west Himalayan fir, is widely distributed from India to central Nepal and Afghanistan to Pakistan. It generally grows from dry to moist temperate regions ranging from 2000 to 3300 m a.s.l (Ahmed 1989; Xiang et al. 2013). Two core samples were extracted from each tree; however, for a few young trees confined on a very steep slope, we obtained only one core. A total of 82 core samples were extracted from 46 living via using the 5-mm-diameter Haglöf Sweden increment borer. The collected samples were air-dried at room temperature, mounted with glue, and polished with progressive sandpaper (320 grit) (Fritts 1976; Speer 2010). After polishing the core samples, clear and distinct rings appeared. The width of the tree ring was measured with a high level of precision (i.e., 0.001 mm accuracy) using the LINTAB tree-ring measuring system. The cross-dating of collected core samples was performed through TSAP software. Of the 82 core samples, 67 were successfully cross-dated (Table 1). A few samples exhibiting anomalous growth, missing rings, and false rings were rejected and not included in our analysis. The quality and coherence of cross-dating were checked with COFECHA software (Holmes 1983) The standard chronology of each site was developed by using ARSTAN software (Cook et al. 1990). The inter-stands correlation of obtained chronologies revealed a high and significant relationship (r = 0.893, n = 399, p < 0.01). Consequently, all series of tree-ring growth were combined to construct a regional chronology (RC). In order to remove non-climatic trends, the negative exponential curve was applied. To assess the mean correlation and strength of the common signal between all tree-ring series, expressed population signal (EPS) and Rbar (inter-series correlation) were calculated with ARSTAN software (Cook and Kairiukstis 1990; Wigley et al. 1984).

2.3 Reconstruction methodology

The Pearson’s correlation coefficient analysis between RC and climate dataset was conducted for the common period 1950–2017 CE. A strong correlation was observed between RC and the CRU TS4.07 climate dataset. Therefore, in this study, we utilized the CRU TS4.07 climate dataset. Correlation coefficients were calculated from the prior year, October, to the current year, September, via DENDROCLIM2002 software (Biondi and Waikul 2004). For reconstruction, we used a simple linear regression model, the RC index as a predictor, and the mean scPDSI of April-July was used as a predictant. The reliability of our reconstruction was determined with split calibration-verification tests through the validation method. The statistical analysis includes R2 (coefficient of multiple determination), RE (reduction of error), CE (coefficient of efficiency), and sign test (Cook and Kairiukstis 1990). The positive value of RE and CE showed the stability and effectiveness of our reconstruction. The Durbin-Watson statics, which reflect the autocorrelation in the residual, showed the reliability of our reconstruction. Based on scPDSI values, the reconstructed scPDSI were characterized as incipient, mild/slight, moderate, severe, and extreme drought (van der Schrier et al. 2013). The “ − 0.5 to 0.5” value of scPDSI is normal, and the negative value of scPDSI indicates drought. For example, the scPDSI value of − 0.5 to − 1.0 is incipient drought, − 1 to − 2 is mild drought, − 2 to − 3 is moderate drought, − 3 to − 4 is severe drought, and <  − 4 indicates extreme drought. The positive values of scPDSI indicate wet periods (van der Schrier et al. 2013). In order to identify the spatial representation of our reconstructed scPDSI with CRU TS4.07 scPDSI and sea surface temperature, spatial correlations were determined for the common period 1955–2017 CE using KNMI climate explorer (https://climexp.knmi.nl). The periodicities cycles in our reconstruction were estimated with multi-taper method spectral analysis (MTM) (Mann and Lees 1996).

3 Results

3.1 Tree-ring chronology and its response to climatic variables

The length of the tree-ring chronology from the Kot site was 1616–2017 CE, while for the Tuti site, it was 1620–2017 CE (Supplementary Fig. S1). To ensure the reliability of RC, the developed RC was truncated prior to 1670 via EPS threshold limit (EPS > 0.85) (Fig. 3). The RC had a high series-inter correlation (0.566–0.610). The variance in the first eigenvector (73.1–74.3%) and standard deviation (0.210–0.213) demonstrated that the ring widths of Abies pindrow are responding to common climate factors (Table 1). Furthermore, the RC reveals extremely narrow ring in specific years, including 1677–1678, 1697–1699, 1705, 1726–1727, 1802, 1815, 1820, 1841, 1853, 1855, 1872, 1892, 1947, 1953, and 2001–2002.

Fig. 3
figure 3

a RC plot of Abies pindrow from the Kandia valley of central-northern Pakistan, (b) sample depth, (c) running expressed population signal (EPS), and, (d) Rbar (mean correlation between ring-width series) statistics

Full size image

The RC exhibits a significant positive correlation (p < 0.05) with precipitation from the prior year, October, and March–May of the current year (Fig. 4a). In contrast, the RC demonstrated a significant negative correlation with minimum temperature (Tmin) during the prior year, October–November, and the current year, April–May. Furthermore, the RC showed a significant negative relationship with the maximum temperature (Tmax) of the prior year, December, and the current year, March–May (Fig. 4b). A notably strong and statistically significant correlation was found between RC and scPDSI. The RC revealed a stronger correlation with the mean monthly scPDSI (r = 0.666, p < 0.001) of warm season (April–July) than precipitation (r = 0.311, p < 0.05), minimum temperature (r =  − 0.338, p < 0.05), and maximum temperature (r =  − 0.306, p < 0.05). This indicates that scPDSI of warm season is the fundamental factor influencing the radial growth of Abies pindrow.

Fig. 4
figure 4

a The correlation coefficient between the RC and total monthly precipitation and mean monthly scPDSI and (b) correlation coefficients between the RC and mean monthly minimum temperature (Tmin) and maximum temperature (Tmax). The correlation coefficients were computed from the prior year October to the current year September over the common period 1950–2017 CE. The black and orange horizontal dashed lines reveal significance level at the 95% and 99% confidence limits (C.L), respectively. The M-M represents the correlation coefficient for the months of March–May, whereas the A–J shows the correlation coefficient for the months of April–July

Full size image

3.2 scPDSI reconstruction

Based on the growth-climate analysis, we selected the mean scPDSI of April–July for the reconstruction via linear regression model. The transfer function was

$${{\text{scPDSI}}}_{4-7}=-9.12+9.42 {\text{RC}}$$
(1)

where scPDSI4-7 is the April–July self-calibrated Palmer drought severity index (scPDSI), and RC is the standard regional chronology of Abies pindrow. The reconstruction model accounts for 44% variance for the actual scPDSI during the common period 1955–2017 CE (n = 67, F = 52.549, p < 0.001). The reconstructed scPDSI fit well on the actual scPDSI curve except for a few extraordinarily high values (Fig. 5a). Durbin-Watson value (Durbin and Watson 1950), which revealed the autocorrelation of the reconstruction, was 2.154. The statistical analysis demonstrated that the actual and reconstructed scPDSI are synchronous at low-frequency variations. The positive value of the RE and CE further confirmed the stability and validity of our climatic reconstruction model (Table 2). The long-term reconstructed scPDSI showed consistency with the actual scPDSI. Therefore, this study is valuable to understand the drought variability over several hundred years in central-northern Pakistan.

Fig. 5
figure 5

a Actual (red) and reconstructed (blue) warm-season (April–July) self-calibrated Palmer drought severity index (scPDSI) over the common calibration period 1950–2017 CE. b Long-term reconstructed warm-season scPDSI from the Kandia valley, central-northern Pakistan. Both the actual and reconstructed scPDSI were smoothed with 11-year low-pass filter. The central black horizontal dotted line represents the mean of long-term scPDSI reconstruction (0.24). The black solid lines show the mean value of ± 1σ (1.50). The lime-dotted lines represent the threshold limit for severe drought or wet condition, and the red-dotted lines reveal threshold lines for extreme drought or wetter condition, followed by standard scPDSI categorization (van der Schrier et al. 2013)

Full size image
Table 2 Statistical characteristics of (a) leave-one-out cross-validation and (b) split calibration–validation methods for the scPDSI reconstruction from the Kandia Valley of central-northern Pakistan
Full size table

3.3 Characteristics of reconstructed scPDSI sequence

The long-term mean of the reconstructed scPDSI was 0.24, with a standard deviation (σ) of 1.50, indicating low-frequency drought variability over the past 348 years (Fig. 5b). The periods 1729–1730, 1752–1753, 1775–1776, 1829–1830, 1905–1908, 1950–1952, 1990–1991, and 1999–2000 showed mild drought. Besides, 1674–1675, 1766–1768, 1780–1783, 1791–1793, 1867–1869, 1878–1879, 1882–1885, 1909–1910, and 1968–1970 revealed mild wet conditions. Moderate drought periods were observed in 1677–1678, 1698–1699, 1727–1728, 1742–1743, 1927–1928, and 1947–1948, while moderate wet periods occurred in 1687–1689, 1709–1710, 1735–1736, 1860–1862, and 1966–1967. Furthermore, extreme drought in central-northern Pakistan occurred in 1705, 1726, and 1815, while the wettest conditions were observed in 1724, 1807, and 2013. Throughout the reconstruction, 1890–1940 was the long-lasting drought period in central-northern Pakistan.

The spatial correlation between the reconstructed and CRU TS4.07 gridded scPDSI from April to July demonstrated a positive field correlation, extending predominantly north–south (Fig. 6a). Meanwhile, the reconstructed scPDSI revealed a negative field correlation with certain regions in Bhutan and India. The possible reason could be the influence of local topography, which leads to the effects of rain shadows and variation in regional rainfall. In addition, short-term climate anomalies and monsoonal patterns have drastic effects on regional moisture variability, resulting a negative association. However, to identify the precise causes of this negative relation, a detailed climatological investigation is needed. The spatial relationship between reconstructed scPDSI and sea surface temperature (SST) can contribute to the influence of SST patterns on regional climate.

Fig. 6
figure 6

a Spatial correlations (p < 0.05) between reconstructed and regional gridded (CRU TS 4.01) warm-season scPDSI during the common period 1901–2017 CE and (b) reconstructed warm-season scPDSI with sea surface temperature (HadISST1) during the common period 1870–2017 CE. The black and lime stars in the left map demonstrate tree-ring sampling site and the capital (Islamabad), respectively

Full size image

Furthermore, a spatial correlation analysis was conducted between the reconstructed scPDSI and SST data (HadISST1) for the period April–July (Fig. 6b). The results revealed a significant positive correlation between our reconstructed scPDSI and the northern Indian Ocean, as well as the western tropical Pacific Ocean. This implies that these regions are important sources of water vapor for central-northern Pakistan. Changes in SST have the ultimate impact on atmospheric circulation, leading to a change in rainfall pattern. For instance, warmer SST might improve moisture transport and result in considerable rainfall in a specific region. This relationship highlighted the connection between distinct ocean and terrestrial climate systems and emphasized the significance of large-scale oceanic factors in shaping regional to large-scale climatic conditions.

The MTM spectral analysis (Mann and Lees 1996) revealed inter-annual cycles (6.8, 3.2, 2.7, 2.5, and 2.3 years) and multi-decadal cycles (11.7, 15.2, 16.2, 17.9, and 128 years) in the reconstructed scPDSI from the Kandia Valley in central-northern Pakistan (Fig. 7).

Fig. 7
figure 7

MTM power spectral analysis for the reconstructed warm-season scPDSI from the Kandia Valley, central-northern Pakistan

Full size image

4 Discussion

4.1 Regional chronology (RC) and its response to climate variables

The regional chronology (RC), developed from tree-ring data of two stands of Abies pindrow, spans from 1620 to 2017 CE. The statistical parameters of the RC, including Rbar between tress, mean sensitivity (MS), standard deviation (SD), and signal-to-noise ratio (SNR), underscored the high dendroclimatic potential of Abies pindrow (Fritts 1976). The characteristics of the RC fall within the range observed in previous dendroclimatic studies of the same species and other species in Pakistan (Ahmad et al. 2020; Ahmed 1989; Chen et al. 2021; Khan et al. 2021; Zafar et al. 2015). The extremely narrow rings recorded in the RC were attributed to drought conditions (Fritts 1976; Shi et al. 2014), as detected during moderate and extremely drought epochs in our reconstruction. The narrow rings recorded in our study may be linked with regional to large-scale climate variability. For instance, the extremely narrow ring (2001) was observed by previous studies in northern Pakistan (Ahmad et al. 2020; Esper et al. 2003; Khan et al. 2013; Zafar et al. 2010). Furthermore, the extremely narrow ring recorded in the year 1705 was consistent with the scPDSI study of the trans-Himalayan region of central Himalaya (Gaire et al. 2019). Similarly, the narrow ring (1947) was recorded in the western Tian Shan Mountain of Kyrgyzstan (Zhang et al. 2017). The resemblance of these narrow rings indicated the influence of large-scale climate variability over central-northern Pakistan.

The radial growth of Abies pindrow showed a significant positive correlation with precipitation from the prior October and March–May of the current year. Meanwhile, the radial growth of Abies pindrow revealed a significant positive correlation with scPDSI from the prior year, October, to the current year, September. In contrast, there is a significant negative correlation with temperature from prior October–December and March–July of the current year. Notably, the radial growth of Abies pindrow showed a substantial positive correlation with scPDSI and precipitation and a negative correlation with temperature during April–July, indicating that scPDSI is the essential climate factor. Similar results have been observed in the studies conducted in Pakistan and neighboring countries (Ahmad et al. 2020; Chen et al. 2016; Gaire et al. 2019; Khan et al. 2019; Singh et al. 2017). Studies are evident that in the high-altitude region, the radial growth of tree initiates in spring and stops at the end of summer (Ahmed et al. 2011; Gutiérrez et al. 2011). However, it is obvious that at the timberline, the radial growth of the tree is sensitive to temperature (Asad et al. 2016), while below the timberline, it is mostly influenced by scPDSI and precipitation, especially during the growing season (Bhandari et al. 2020; Khan et al. 2019). Studies showed that below the timberline in arid-semiarid regions, temperature plays an opposing role in the radial growth of trees as it serves as a key factor in intensifying drought and increasing evapotranspiration (Fang et al. 2012; Ram 2012; Shi et al. 2016). Hence, there is an apparent negative correlation with radial growth. However, below the timberline, the scPDSI reflects the balance between rainfall, soil moisture, and evapotranspiration and exhibits a positive correlation with radial growth of tree. In addition, it is noteworthy that during April–July, trees allocate a significant portion of their energy to radial growth, which necessitates enough water for photosynthesis and the transport of nutrition via vascular system. However, when scPDSI indicates limited moisture content, it reduces the radial growth of tree due to insufficient moisture content and delay photosynthesis, resulting narrow rings formation.

4.2 Drought variability for the Kandia Valley central-northern Pakistan since 1670 CE

The tree-ring reconstructed April–July scPDSI is linked to a scPDSI field situated approximately north of the 30–36° latitude, characterized by a significant southeast-northwest extension. The reconstruction revealed a strong association with central-northern Pakistan. In accordance with the standard classification of drought, our scPDSI reconstruction demonstrated eight mild drought periods and six moderate drought periods (van der Schrier et al. 2013). The notable mild drought period from 1905 to 1908 CE was coherent with a dry period documented in the tree-ring-based drought reconstruction of Tien Shan Mountain, Kazakhstan (Chen et al. 2016). In addition, the drought period spanning from 1920 to 1930 CE was reported from northern China (Liu et al. 2009; Yang et al. 2011; Zhang et al. 2011) and northeastern Mongolia (Pederson et al. 2001). This drought affected a wide area of China, resulting in economic damage and marvelous loss of life (Liang et al. 2006; Shi et al. 2016). Our study shows that the evidence of the drought period occurred in 1920–1930 CE. However, no record of high-intensified drought and economic losses is available from central-northern Pakistan. The most prominent feature observed in our reconstructed scPDSI is the increase of high-magnitude droughts in the early eighteenth century, which persisted to the mid-nineteenth century with some dominant wet phases. Nevertheless, a prolonged increasing drought trend was observed in the late nineteenth century and early twentieth century. The same drought trend has been observed in tree-ring-based drought studies in other countries (Gaire et al. 2019; Sano et al. 2012; Yadav et al. 2017). The drought trend observed in the late eighteenth and early nineteenth centuries is similar to the study of northern Pakistan (Treydte et al. 2006). Persistent evidence of moderate drought (1920–1930 CE) in the early twentieth century was observed in the tree-ring-based drought study of the Chitral region (Ahmad et al. 2020). However, it was not as prominent as in this study. The possible reason for this discrepancy may be differences in the standardized categorization of drought events. The recorded wettest years, such as 1724, 1807, and 2013, coincide with other reconstructions of central Asia, Nepal, and the Chitral region of northern Pakistan (Ahmad et al. 2020; Chen et al. 2019). According to Treydte et al. (2006), twentieth-century warming could increase the moisture-holding capacity of the atmosphere. A similar increasing moisture trend was observed in our reconstruction (Fig. 5), as well as in the anomalies of meteorological station data (Fig. 2).

4.3 Periodicities in the drought reconstruction of central-northern Pakistan

The detected annual periodicity cycles, ranging from 2.3 to 2.7, 3.2, and 6.8 years, indicated potential links to quasi-biennial oscillations and El Niño–Southern Oscillation (ENSO), which are the key drivers of climate variability (Allan et al. 1996; Burgers et al. 2005). The climate of central-northern Pakistan is mutually influenced by the SASM and westerly patterns. Therefore, the observation of high-frequency periodicity cycles in our study demonstrated that drought variability in our study area may be responsive to large-scale ocean–atmosphere–land circulations. The 11.7-year cycle showed a resemblance with the 11-year solar activity cycle, hinting at a potential linkage between local drought changes and solar activity. In addition, the centennial–scale cycle (128 years) recorded in our reconstruction resembles similar findings in the surrounding Tibetan Plateau, emphasizing the complex and interrelated nature of climate patterns (Fang et al. 2010; Peng and Liu 2013).

4.4 Regional to large-scale comparison

The drought formation may be potentially influenced by common factors at the regional or large scale (Borgaonkar et al. 2010; Cai et al. 2015; Gaire et al. 2019; Peng and Liu 2013). Therefore, in order to achieve a more holistic prospect of drought variability, it is essential to consider both regional and large-scale perspectives. On the regional scale, our reconstructed April–July scPDSI showed a significant positive correlation (r = 0.307, p < 0.01) with the reconstructed March–August scPDSI from the Chitral region (Ahmad et al. 2020) (Table 3). However, it is essential to note that some of the dry phases in our reconstructed scPDSI do not align with the study of Chitral (Ahmad et al. 2020) (Fig. 8a). For example, the dry phases, including 1670–1690, 1710–1730, 1830–1850, and 1920–1930 CE, identified in our study and previous studies from Pakistan and neighboring regions (Cook et al. 2010; Khan et al. 2019; Singh et al. 2022; Treydte et al. 2006; Yadav et al. 2017), are not actually dry in the reconstruction of Chitral (Ahmad et al. 2020). Moreover, our reconstruction revealed a significant positive correlation (r = 0.254, p < 0.01) with the June–May precipitation of Khan et al. (2019). However, with the exception of a few phases, most of the dry phases are aligned with our reconstruction. A weak and negative correlation (r =  − 0.150, p < 0.01) was found between our reconstructed scPDSI and October–September reconstructed precipitation from the Karakoram region (Treydte et al. 2006). This negative correlation may be attributed to the utilization of different approaches for climate reconstruction and differences in local climate. For instance, the reconstructed precipitation of Karakoram (Treydte et al. 2006) was based on annually resolved oxygen isotope, while the scPDSI reconstruction of our study is based on annual growth data. Furthermore, variation in regional and topographic features leads to weak correlation. According to Shah et al. (2010), the pre-monsoon season during April–July is a transitional phase from the winter circulation to monsoon circulation in the region. During this phase, westerly waves move northward, and the frequency of western disturbance decreases in comparison to the winter peak months. The weak correlation between our reconstruction and Treydte et al. (2006) could be attributed to this transitional phenomenon. In addition, it is important to note that our sampling site is located about 650 m and 750 m below the elevation range of the sites studied by Ahmad et al. (2020) and Treydte et al. (2006), respectively. This means that our sampling site experienced high evapotranspiration compared to the sites they studied.

Table 3 Correlation between our reconstructed scPDSI and prior studies conducted in Pakistan and the neighboring regions
Full size table
Fig. 8
figure 8

Graphical comparison between the central-northern Pakistan scPDSI reconstruction (this study), a tree-ring-based precipitation and drought studies from Pakistan, and (b) drought reconstruction from nearby regions

Full size image

The comparative analysis of the reconstructed scPDSI from central-northern Pakistan in this study with the northwest Himalaya of India, the northwest Himalaya and Karakoram region of India, and the Monsoon Asia Drought Atlas showed a high degree of consistency at both inter-annual and inter-decadal scales (Fig. 8b). Most of the drought phases identified in our reconstruction align with the findings of these studies. Notably, our reconstructed scPDSI exhibited a strong positive relationship with the June–May drought (Singh et al. 2022), the standardized precipitation index (SPI) of May (SP18-May) (Yadav et al. 2017), and the June–August drought (MADA) (Cook et al. 2010) highlighting the robustness of our reconstruction and elucidating the spatial and regional patterns of hydroclimatic variability (Table 3). After applying an 11-year low-pass filter, the correlation between our reconstruction and the one presented by Yadav et al. (2017) has increased, whereas the correlations with other moisture-sensitive tree-ring records decreased.

4.5 Possible climatic mechanism

Climatic mechanisms play a significant role in shaping the intensity and duration of drought events. Identifying these potential climatic drivers is crucial. To evaluate these potential drivers and their influence on drought variability, a correlation analysis was carried out between our reconstructed scPDSI and various indices, including SASMI, AMO, PDO, NINO, and NAO.

Our reconstruction reveals a significant positive correlation (r = 0.37, p < 0.01) with SASMI (Fig. 9a). The peiods of low SASMI, notably during 1950–1958, 1983–1993, and 1999–2004 CE, were consistent with the dry phases identified in our warm-season scPDSI reconstruction. Similarly, the observed high SASMI from 2005 to the present align with our reconstruction. However, in comparison to SASMI, our reconstruction exhibits a weak linkage with the Indian summer monsoon (ISM). This discrepancy may be attributed to the regional sensitivity of our reconstruction, which aligns more closely with SASMI and appears to be less influenced by the broader dynamics of the ISM (Khan et al. 2019). However, to fully validate and understand this explanation, further research is needed.

Fig. 9
figure 9

a Comparison (Z-scores) between the reconstructed warm-season scPDSI (this study) with South Asia Summer Monsoon index (SASMI) (Li and Zeng 2002) and (b) Atlantic Multi-decadal Oscillation (AMO) (van Oldenborgh et al. 2009). All series were smoothed with 11-year low-pass filter

Full size image

Furthermore, our warm-season scPDSI exhibits a significant positive correlation (r = 0.41, p < 0.01) with AMO (Fig. 9b). On a decadal scale, the drier phases observed in our warm-season scPDSI reconstruction exhibit notable consistency with the negative phases of the AMO, particularly during 1880–1895, 1902–1930, 1946–1950, and 1989–1997 CE. This association implies a potential link between the negative phases of AMO and the occurrence of drier conditions in central-northern Pakistan. A study conducted in the Chitral region of Pakistan indicated a possible linkage between reconstructed drought and SASMI and AMO (Ahmad et al. 2020). However, its essential to note that their analysis was based on low-frequency variability, specifically a 31-year moving average. Additionally, their reconstruction reflects variation aligning with certain phases of AMO and SASMI.

A number of studies highlight the influence of ENSO on regional moisture change, emphasizing temporal variability (Krishna and Rao 2010; Kumar et al. 2007; Sano et al. 2009). Our warm-season scPDSI reconstruction generally aligns with the prominent phases of ENSO, particularly NINO 3.4 (Supplementary Fig. S2). Studies have shown that ENSO is the potential drivers that has a significant influence on hydrological changes associated with AMO (Chen et al. 2019; Palmer et al. 2015; Shi et al. 2018). According to Shi et al. (2014), drought coincided with the warm phase of ENSO, linked to Indian summer monsoon failure. This implies that the recent increasing moisture trend in central-northern Pakistan is potentially associated with ENSO. Additionally, the robust relationship of our reconstruction with specific months of NINO4, NINO3.4, and NINO3 provides additional confirmation (Table 4).

Table 4 Correlation between reconstructed scPDSI (this study) and the climatic drivers
Full size table

The Pacific Decadal Oscillation (PDO) is also an important climate driver influencing atmosphere circulation, sea temperature, and rainfall patterns, thereby impacting drought variability in different regions (de Oliveira-Júnior et al. 2018; Schoennagel et al. 2005). Our reconstruction reveals a noteworthy relationship with specific months of the PDO (Table 4), showing a substantial link between PDO and drought variability in central-northern Pakistan. Despite the North Atlantic Oscillation (NAO) not profoundly affecting the climate of the northern hemisphere, including central-northern Pakistan. Our reconstruction still exhibits a significant correlation with the June month of NAO. This correlation might be attributed to similar atmospheric circulation and rainfall patterns between the northern and southern hemispheres.

The solar modulation function and sunspot number are important factors that have a significant influence on climate dynamics (Hong-yan et al. 2015; Muscheler et al. 2007). Historical long-term records of the solar modulation function and sunspot number show considerable variation (Fig. 10a, b), notably characterized by the Maunder minimum (1645–1715 CE), the Dalton minimum (1790–1820 CE), and the Damon minimum (1900–1920 CE) periods. The prolonged drought periods identified in our reconstruction coincide with reduced solar activity, suggesting a potential link between drought and solar variation. Furthermore, the solar modulation function and sunspot number indicate maximum solar activity during 1760–1800 CE and from 1950 to the present. Notably, solar activity during 1760–1800 CE exhibited a higher intensity than that revealed in recent decades. This potentially reflects the intricate natural variability inherent in solar cycles, characterized by their intermittent nature. Our scPDSI reconstruction demonstrates drought condition during the period of stronger solar activity, contrasting with wetter condition in recent decades (Fig. 10c). These insights enhance our understanding of the intricate relationship between variations in the solar cycle and drought variability in central-northern Pakistan, emphasizing the imperative for future in-depth research on these complex connections.

Fig. 10
figure 10

a Graphical comparison between the solar activity and (b) sunspot number with (c) our scPDSI reconstruction. All series were smoothed with 11-year low-pass filter

Full size image

5 Conclusions

We developed a regional chronology (RC) from two stands of Abies pindrow in the Kandia Valley, central-northern Pakistan. Growth-climate analysis revealed that warm-season (April–July) scPDSI is the fundamental factor that significantly influences the radial growth of Abies pindrow. Using linear regression, we reconstructed a 348-year (1670–2017 CE) long-term warm-season scPDS variability. The reconstructed scPDSI explained a 40% variance in the actual scPDSI during the common period of 1950–2017 CE. Furthermore, the multi-taper method (MTM) spectral analysis identified distinct cycles of inter-annual (6.8, 3.2, 2.7, 2.5, and 2.3 years) and multi-decadal (11.7, 15.2, 16.2, 17.9, and 128 years). The inter-annual cycles revealed possible links between central-northern Pakistan and the El Niño-Southern Oscillation (ENSO), as confirmed by correlation analysis of our scPDSI reconstruction with ENSO and the Pacific Decadal Oscillation (PDO). In addition, our scPDSI reconstruction revealed significant links with the South Asian Summer Monsoon (SASMI), as well as the Atlantic Multidecadal Oscillation (AMO). These findings indicate that SASM and AMO are the key drivers that significantly influencing drought variability in central-northern Pakistan. Furthermore, we also observed that some prolonged drought events are consistent with period of reduced solar activity and sunspot numbers. These findings enhance our understanding of long-term drought variability and its potential links with climate mechanisms. However, to determine the mechanism and impact of these interactions at different timescales, further research is awaited.