Abstract
Rivers originating from the Andes Mountains are vital water sources for agricultural and societal needs in South America, yet are increasingly threatened by climate change. Here we reconstruct streamflow of the Negro River (Rió Negro) in northern Patagonia over the past 827 years using tree-ring records from the south-central Andes foothills. This reconstruction reveals an unprecedented decline in river flow in recent decades. Moisture from the El Niño–Southern Oscillation and the Southern Annular Mode provides a key water source, but temperature rise has increased atmospheric moisture demand and reduced availability. Antarctic amplification has further disrupted circulation patterns and accelerated warming, intensifying regional aridity. Since the 20th century, the Negro River has experienced a sustained flow reduction of approximately 10% per decade. Projections under various emission scenarios indicate continued decline, highlighting the urgent need for adaptive water management strategies to mitigate escalating water scarcity risks in the region.
Similar content being viewed by others
Introduction
It is indisputable that the intensifying impacts of climate change are profoundly altering the spatial and temporal distribution of water resources, particularly in desert ecosystems1,2,3,4,5. Patagonia, located at the southern tip of South America, is characterized by its semi-arid and desert-like environment. which faces unique hydrological challenges due to geographic isolation and complex climate patterns6,7. The region’s water resources are primarily derived from rivers originating in the south-central Andes Mountains; thus, any variations in hydrological dynamics, particularly streamflow fluctuations, directly affect consumption in urban centers, the hydroelectric power generation and agro-pastoral activity downstream8. In recent decades, the accelerating process of aridification in the south-central Andes region, including Patagonia has contributed to the degradation of forests and grasslands, posing serious threats to both socio-economic development and ecological health9,10,11,12,13. As streamflow represents a crucial water resource directly utilized by human societies and ecosystems, understanding its long-term evolution and response to climate change has become increasingly important14,15,16,17. Projections indicate that ongoing global warming will exacerbate water supply pressures in the future18,19,20, presenting a serious challenge for natural resource regulators and environmental policymakers.
Long-term hydrological reconstructions are essential for understanding historical water availability, offering valuable insights into natural variability and the impacts of climate change21,22,23. Globally, streamflow reconstructions utilizing proxies such as tree rings have proven effective in identifying droughts and floods over centuries, providing a historical baseline for evaluating current and future water resources16,17,24,25,26,27. Efforts previously have been made to reconstruct the past hydroclimatic variability in the Greater Andes region7,8,10,28,29,30,31,32. There is evidence of variations in streamflow across the Andes (18°–55°S) during the instrumental era33, and efforts have been made to increase knowledge about the variability of streamflow in sectors of northern Patagonia10,12,31,32,34,35. However, new streamflow reconstructions for the mainstem of the Patagonian rivers are needed to increase both temporally and spatially in order to understand the magnitude of the hydrological variability processes in the south-central Andes Mountains range in the context of the ongoing climate change.
The rivers on the eastern slopes of the Andes irrigate a larger area of farmland that is crucial for global food supply security. In this sense, we focused on the Negro River (Río Negro), which is formed by the confluence of the Neuquén and Limay tributary rivers originating in the Andes Mountains (Fig. 1a). At ~630 km in length, it is the longest river in Patagonia. Given the context of warming and aridification, sharp declines in the observed streamflow of the Negro River (Fig. S1), make it imperative to investigate the natural variability and drivers of streamflow in a long-term context and the potential role these may play in future water availability. Especially, Patagonia climate variations are of more than regional interest: studies of relict tree stumps in lakes have suggested that medieval droughts in Patagonia may have coincided with epic droughts in North America through synchronous reorientation of mid-latitude storm tracks in the Northern Hemisphere and Southern Hemisphere36. Our reconstruction highlights critical periods of hydrological stress, including sharp declines, and will link these periods to known climate events, offering a broader perspective on the river’s vulnerability to future climate changes.
a Topography of the Negro River basin, including the locations of the hydrological station and sampling sites. The control basin for the hydrological station (Primera Angostura) is provided by the Global Runoff Data Center (GRDC)52. b Spatial correlation patterns between observed annual streamflow in the Negro River and runoff from the G-RUN dataset83, considering the period from 1928 to 2017. The Negro River basin is highlighted and dots indicate the 95% significance level, same as below. c Schematic diagram of the large-scale spatial location and main circulations. The topographic base-map for (a) and (c) is sourced from Natural Earth. d A photograph of sampling sites for Araucaria araucana trees. e Reconstructed October–January streamflow in the Negro River during 1197–2023, after bias correction using Quantile Mapping (QM) method54,55,56,57. An 11-year low-pass filtered curve was applied to highlight the low-frequency variability. The horizontal middle line presents the mean and the horizontal dashed lines represent the mean ±1.5 standard deviations (SDs). The gray shadow represents ± root mean square error (RMSE) while circles represent extreme years. Also provided simultaneously are the running 45-year slope and the decline severity (calculated by multiplying the mean slope by the duration; slopes must be below zero for more than 11 consecutive years)59 for the reconstructed streamflow. The width of the bars indicates the start and end dates of each decline period.
Previous studies suggest that the inter-decadal variability of Patagonia’s hydroclimate can be partially attributed to large-scale ocean-atmosphere circulations, particularly related to El Niño-Southern Oscillation (ENSO) and Southern Annular Mode (SAM)7,8,37,38,39,40,41,42,43,44,45. These are also possible driving mechanisms for multi-continental droughts, meaning that related investigations should not be limited to the local scale17,39,46. Given the proximity of the two regions, it is crucial to investigate how rapid warming in Antarctica may alter the hydroclimatic processes of the south-central Andes47,48,49,50,51, as such changes could represent historically rare occurrences. In this study, we aim to: (1) develop a reliable streamflow reconstruction for the mainstem of the Negro River in Patagonia and identify periods of water crises; (2) identify and quantify the primary hydroclimatic drivers of streamflow variability, assessing their relative contributions; (3) explore potential large-scale driving mechanisms, particularly the role of Antarctic amplification; and (4) project future risks to the Negro River under warming scenarios.
Results and discussions
Recent decline is the most severe based on an 827-year streamflow reconstruction
In the upper Negro River basin, we constructed a robust composite tree-ring chronology for Araucaria araucana using new samples and records from the International Tree-Ring Data Bank (ITRDB), extending back to 1197 CE (Figs. 1, and S2 and Table S1). The composite chronology is closely related to the Negro streamflow from the Global Runoff Data Center (GRDC)52, which reflects the hydrological variations in the south-central Andes and Patagonia (Figs. 1b and S1). Thus, during the austral summer (October to January), when the response is strongest, a direct linear regression model was established between the chronology and streamflow from 1928 to 2017, which can explain 44.6% of the observed streamflow variance (Figs. S3 and S4).
Our streamflow reconstruction was statistically validated using a split test scheme (Table S2), which yielded effective statistics, indicating robust reconstruction capability. Furthermore, the reconstruction’s long-term coherence is supported by its significant spatial correlation with hydroclimate fields from the Paleo Hydrodynamics Data Assimilation product (PHYDA)53 and the South American Drought Atlas (SADA)8 (Fig. S5). Given that reconstruction often involves over-estimation or under-estimation, we choose a bias correction procedure using Quantile Mapping (QM) method54,55,56,57 to reduce these systematic differences (Figs. 1e and S6). The Negro streamflow reconstruction covers 827 years (1197–2023), with a mean of ~910 m3·s-1 and a standard deviation of 351.849 m3·s-1 (Fig. 1e). It includes 11 extended pluvial conditions (1242–1271, 1319–1348, 1353–1373, 1397–1409, 1468–1488, 1539–1572, 1576–1602, 1611–1622, 1632–1652, 1660–1712, 1910–1958) and 14 extended drought conditions (1197–1218, 1223–1241, 1272–1295, 1306–1318, 1374–1396, 1410–1424, 1429–1458, 1489–1538, 1741–1754, 1763–1794, 1798–1828, 1836–1880, 1887–1897, 1968–2023), accompanied by a slight increase in the drought rate, especially severe events that occur once every 30 to 100 years (Table S3 and Fig. S7).
More notably, the first half of the 20th century stands out as one of the wettest periods in the reconstruction. Resource management strategies based on, or partly based on, this data may overestimate the baseline amount of water that can be expected from the natural system, not unlike the situation with the Colorado River compact in the Southwestern United States58. This issue is of particular concern when combined with the rapid aridification and streamflow decline observed in the Negro River following this pluvial period. Therefore, we focus not only on drought (low streamflow) but, more critically, on episodes of pronounced streamflow decline from a high baseline. The severity and persistence of these declines, which challenge water systems adapted to plenty, are our primary concern. Based on a method59 constructed using running slopes and durations, ten periods of severe streamflow decline have been identified (Fig. 1e). Comprising them, the recent event has been confirmed as the most s severe, exhibiting the lowest combined values for both the decline rate and its duration. Although the choice of running window may alter the judgment of the decline severity, the fact that the recent decline is unprecedented over the past eight centuries remains unchanged (Figs. 1e and S8). While the mid-to-late 13th century event exhibits a steeper rate of decline over a shorter period (Fig. 1e), the modern event is unparalleled in its persistence. Consequently, although the recent decline has not been as rapid as that recorded in the 13th century, its prolonged nature has resulted in a more substantial cumulative water loss, posing a more severe threat to long-term water security. In particular, the Greater Andes region showed a similar decline in water variables, which is extremely rare across different time scales9,10,11,12. Overall, the severe water crisis in the Greater Andes region is alarming due to its historical rarity, especially considering the dependence of major urban centers in Patagonia on water supplies from the south-central Andes water system.
The combined effects of climate changes on streamflow variations
Understanding the effects of evolving climate patterns on river hydrology is essential for predicting future water security4,5, especially as our reconstructed streamflow baseline clearly illustrates the exceptional nature of recent declines. These extreme declines necessitate the detection and attribution of the combined effects of key climate changes on hydrological processes. Here, climate variables considered in this study include precipitation60, Palmer Drought Severity Index (PDSI)61, Standardized Precipitation Evapotranspiration Index (SPEI)62, soil moisture63,64, maximum temperature60, average temperature60, minimum temperature60, and potential evapotranspiration60, which encompass water supply and dissipation, as well as their composite effects.
It is evident that precipitation, as the main source of water recharge in summer, is significantly positively correlated with reconstructed Negro streamflow27,65,66,67 (Fig. 2a), with consistent responses to wet and dry signals7,8,68,69,70 (Fig. 2b–d). Conversely, increased atmospheric water demand due to rising temperatures leads to higher evapotranspiration, further reducing water availability during hot periods17,30,71,72 (Fig. 2e–h). Geographically, the south-central Andes is the most critical region for water supply, encompassing most of the Negro River basin (Fig. 2a–d). This is because the Andes Mountains block and lift moisture from the Pacific Ocean, causing it to condense and form precipitation69. In contrast, the narrow channel of the lower Negro River, which traverses through the desertified northern Patagonia, delivers little, especially as to precipitation (Fig. 2a). Warming and evapotranspiration signals, which cover a larger area covering the entire basin, intensify the inhibition of the convergence process, indicating that global warming is likely one of the driving forces behind the Negro streamflow decline (Fig. 2e–h). Significant correlations support the above findings, implying a strong link between regional hydrological processes and climate changes (Fig. 2i).
Spatial correlation patterns of the reconstructed streamflow with simultaneous gridded climatic data during the common period from 1982 to 2017, including precipitation (PRE)60 (a), Palmer Drought Severity Index (PDSI)61 (b), Standardized Precipitation Evapotranspiration Index (SPEI)62 (c), soil moisture 0–10 cm (SM)63,64 (d), maximum temperature (TMX)60 (e), mean temperature (TMP)60 (f), minimum temperature (TMN)60 (g), and potential evapotranspiration (PET)60 (h). i Correlation between climate variables within the basin and reconstructed streamflow. The dashed and dotted lines indicate the 95% and 99% significance levels, respectively. j A box-and-whisker plot shows contributions from climate changes based on Multivariate Linear Regression (MLR) and Hierarchical Partitioning (HP) methods16,73,74,75, using 1000 Monte Carlo tests121. The box indicates the 25–75% range of 1000 tests, the whisker line indicates the 1.5 interquartile range (IQR), the horizontal line indicates the mean, and extra dots indicate outliers. Corresponding mean contribution values are also provided.
Further quantitative analysis shows that these eight climate variables in the basin jointly contributed to 64.9% of the reconstructed streamflow variation using the Multiple Linear Regression (MLR) method16,73,74 during the common period, and showed different characteristics after decomposition by the Hierarchical Partitioning (HP) method75 (Fig. 2j). Among these, PDSI and soil moisture contribute the most, averaging 13.7% and 12.8%, respectively, and are relatively stable due to the higher box in Fig. 2j. Next in order are mean temperature, evapotranspiration, precipitation, SPEI, maximum temperature, and minimum temperature, contributing 8.9%, 7.4%, 6.7%, 6.0%, 5.3%, and 4.1%, respectively. The above results are similar in observed Negro streamflow (Fig. S9).
Overall, as the systematic integration of precipitation and temperature—the two most direct climate variables76—both PDSI and soil moisture, and even SPEI, all exhibit strong associative ability for the Negro streamflow variability, fully reflecting regional dry and wet signals. It should be noted that, in the context of global warming, the effect of evapotranspiration due to higher temperature has caught up with precipitation, and the contribution from the mean temperature is not only stable but also has the highest floor level, meaning that warming is playing an ever-increasing role in causing severe droughts72. Therefore, the combined effects of climate changes on water resources can no longer be ignored, particularly with regard to synergistic water-heat regulation, aiming to provide universal significance for early warning of water security in the south-central Andes region in the context of global warming.
Circulation-driven mechanisms and Antarctic amplification modulation
Given the importance of moisture delivery and temperature for Negro streamflow variations, it is necessary to further analyze where water and heat come from and the potential causes of the recent decline. Observed and reconstructed high flows in the Negro River correspond to high sea surface temperature (SST) and low sea level pressure (SLP) anomalies in the tropical east-central Pacific (Fig. S10). Meanwhile, a similar pattern occurs at mid- to high-latitudes in the Southern Hemisphere, along with an extremely high SLP anomaly in the Antarctic. Thus, the critical moderating role of the ENSO and SAM in hydrological variations across the south-central Andes is highlighted. In particular, severe declines were associated with La Niña and positive SAM years, during which periods the values deviated downward by –110.785 m3·s-1 and –134.752 m3·s-1, respectively (p < 0.05) (Fig. 3a, b).
Surface temperature anomalies based on the Paleo Hydrodynamics Data Assimilation (PHYDA) product53, corresponding to the decline periods before the Second Industrial Revolution (a) and the recent most severe decline period (b). Anomalies are calculated based on the entire reconstructed period. The insets show results of the Superposed Epoch Analysis (SEA) testing123 the response of reconstructed streamflow to 40 La Niña events124 and 60 positive Southern Annular Mode (SAM) events125. The dashed line indicates the 95% significance level. c Spatial patterns of precipitation anomalies (shading, mm·day-1) & total water vapor transport anomalies (total quv, vectors, where uq and vq are multiplied by 1000, kg·m-1·s-1) associated with La Niña events in the Community Earth System Model-Last Millennium Ensemble (CESM-LME) simulations77,78. d Spatial patterns of precipitation anomalies (shading, mm·day-1) & sea level pressure (SLP) anomalies (contours, Pa) associated with positive SAM events in the CESM-LME simulations. e Large-scale surface temperature (shading, °C) and SLP (contours, Pa) fields derived by regression using standardized spatial average surface temperatures around the Antarctic Peninsula (60°–90°S, 30°–110°W) in the CESM-LME simulations. f Warming trends since the 20th century based on Berkeley Earth Surface Temperature Anomaly Field97. The insets in (c–f) highlight relevant information about the study area.
Results from the Community Earth System Model-Last Millennium Ensemble (CESM-LME)77,78 further support the notion that ENSO and SAM regulate a portion of the water supply at centennial and greater scales. During La Niña years, cold anomalies in the tropical east-central Pacific SSTs lead to an increase in the SLP, which amplified the SLP gradient with the Indo-Pacific warm pool37,39,40,41,79 (Fig. S11a). The resulting enhancing of the east-west pressure gradient in the equatorial Pacific implies a stronger Pacific Walker Circulation (PWC)80. Water vapor transport anomalies with eastward wind patterns prevents water vapor from the Pacific Ocean to reach the south-central Andes, thus contributing to intensified drought conditions (Fig. 3c). And the positive SAM phase can similarly excite La Niña-like SSTs (Fig. S10b). On the other hand, the Antarctic low-pressure anomaly indicated by the positive SAM phase assists in the austral westerlies to drift southward, along with the Antarctic Peninsula’s cold temperatures to the Greater Andes region and trigger La Niña, which inhibit for the recharge of the regional water resources42,43,81 (Figs. 3d and S11b).
It is particularly noteworthy that this unprecedented recent decline occurred under La Niña-type SST conditions similar to those during past events, and even more significantly, against the backdrop of the unprecedented global warming since the 20th century (Fig. 3a, b). Persistent La Niña-induced precipitation deficits along the west coast of South America and the Andes Mountains have exacerbated the rapid decline in the Negro streamflow40,82. Concurrently, low pressures associated with the ongoing Antarctic warming since the 20th century are not only pulling the westerlies southward, reducing water vapor supply48,81, but also causing high temperatures in the south-central Andes, greatly intensifying regional evapotranspiration71, further exacerbating the water crisis (Fig. 3e). Even Antarctic warming is slightly reducing the east-central SSTs, exacerbating La Niña to some extent50 (Fig. 3e). In particular, the recent rate of warming in Antarctica is more than three times the global average rate47,48,49,50,51 (Fig. 3f). Such intense Antarctic amplification deserves special attention because it is profoundly altering atmospheric circulation and south-central Andean hydroclimatic dynamics in unprecedented ways. The simulated streamflow from CESM-LME also shows a recent downward trend, indicating that global warming is profoundly altering hydrological processes in this region (Fig. S12). Given the dependence of the major Patagonian urban centers on the water supply originated from rivers in south-central Andes, it is imperative to prioritize the protection and management of forest and river systems to mitigate the adverse effects of this unprecedented decline in the face of climate change.
Future streamflow projections and water resource management
Utilizing the Global runoff dataset (G-RUN)83, which exhibits strong performance in both observed and reconstructed streamflows (Fig. S13), we employ a neural network technique84,85,86 for the downscaling model training to achieve robust validation (Fig. S14a). This downscaling model was then applied to the Coupled Model Intercomparison Project Phase 6 (CMIP6)87 ensemble to project future streamflow variations for the Negro River (Figs. 4a and S14b). Not to be overlooked is the fact that the reconstructed and observed streamflows in the Negro River declined by a staggering –10.33% and –11.34% per decade, respectively, during the observed period (1928–2017). Even with the rebound in the late 19th century, after once again experiencing the most severe decline in nearly a century, the streamflow of the Negro River is currently only ~50% of the historical period. Projections for the SSP2-4.5 scenario still indicate continued streamflow declines of –3.78% per decade in the context of future warming, even reaching –7.57% per decade for the more severe SSP5-8.5 scenario. The magnitude of future warming would suggest that Antarctic amplification is still causing the south-central Andes water crisis, even under conditions of enhanced El Niño conditions (Fig. 4b). Although future warming of SSTs in the central Pacific Ocean will facilitate water vapor transport to the Greater Andes region to some extent, more intense warming due to the Antarctic amplification effect may impede this process. Under ongoing global warming, the Antarctic amplification is expected to exacerbate the weakening and southward displacement of the West Wind Drift, thereby reducing moisture supply to the south-central Andes48,81,88,89,90. Concurrently, it will trigger significant warming in Greater Andes region71,91,92. These combined effects are anticipated to result in widespread warming and substantially reduced precipitation in the primary water source region, the south-central Andes. Consequently, it will be challenging to mitigate river disruptions within desert ecosystems. In particular, water resources in the south-central Andes and Patagonia are at risk of depletion under the SSP5-8.5 scenario of high greenhouse gas emissions and extreme warming.
a The percentage changes in reconstructed, observed, and future projected streamflows in the Negro River, compared to the entire reconstruction period. Future projections include SSP2-4.5 and SSP5-8.5 scenarios for CMIP6 data87 downscaled by a neural network technique84,85,86. The respective trends and the 45-year running slopes are given. b Future projections of surface temperature trends (°C·decade-1) in the CMIP6 ensemble. The insets highlight precipitation trend (mm·day-1·decade-1) about the study area. Schematics of societies and ecosystems in Patagonia in rising (c) and declining (d) streamflow. Oceans, mountains, forests, rivers, agro-pastoral crossroads, urban centers, and major weather patterns are depicted in the schematics.
Given the projected trends toward more extreme hydroclimatic conditions, effective water resource management must ensure the availability of water for rivers in Patagonia’s desert ecosystems within the framework of existing understanding (Fig. 4c, d). In the case of the Negro River, for example, which is anticipated to experience significant streamflow reductions, it is crucial to enhance drought preparedness through improved water conservation measures, efficient irrigation practices, and the development of alternative water sources17,18,20. It is essential to implement sustainable land management practices to combat desertification processes and strengthen the resilience of ecosystems that depend on the water provided by the Andean basins7,13,16,93,94. In addition, water resources management and protection are complicated by uncertainty of projections and rapid environmental changes without precedent under the Antarctic amplification effects. Effective stakeholder collaboration, continuous monitoring, and adaptive management are critical for mitigating risks from extreme hydrological changes. This will enhance the long-term resilience of both natural and human systems under international cooperation frameworks.
Conclusion
Using tree-ring samples from A. araucana located in the foothills of the south-central Andes, we have developed a robust composite tree-ring width chronology spanning the last eight centuries. Employing linear regression and QM bias correction methods, we reconstructed austral summer streamflow variations of the Negro River since 1197 CE. Our results indicate that the recent streamflow decline is unprecedented over the past 827 years and is strongly influenced by regional climate changes, particularly the synergistic effects of precipitation and temperature on regional hydrological signals. Antarctic amplification is leading to ENSO- and SAM-driven precipitation to reduce and significant warming in the south-central Andes, thereby exacerbating the water crisis. Future projections suggest that Antarctic amplification will continue to cause streamflow declines, even under enhanced El Niño conditions. These findings provide a long-term perspective on water resource research in the Greater Andes region, particularly Patagonia. In light of climate warming and the resultant drought and water crises, we recommend the implementation of early warning systems and comprehensive water resource management plans for this region.
Materials and methods
Tree-ring sampling information
In March 2024, tree-ring samples were collected from three sites in the Andes Mountains of northwestern Patagonia, in the upper reaches of the Negro River (Fig. 1a). These open forests on sandy or shallow rocky soils featured A. araucana trees, at sites PPA (38.87°S, 70.59°W, 1662 m a.s.l.), PPH (38.68°S, 70.84°W, 1628 m a.s.l.), and ALA (38.88°S, 71.14°W, 1279 m a.s.l.). Following International Tree Ring Data Bank (ITRDB) guidelines, two cores per tree were extracted using 10 mm increment borers at ~1.3 m above ground to minimize non-climatic influences. Additionally, tree-ring records from three nearby ITRDB sites—PP (38.88°S, 70.62°W, 1620 m a.s.l., No. arge015), RH (39.40°S, 70.80°W, 1380 m a.s.l., No. arge004), and CH (38.10°S, 70.85°W, 1650 m a.s.l., No. arge006)—were utilized35.
Primary hydroclimatic data
The primary hydroclimatic data used in this study are as follows:
1) Observed monthly streamflow data: Streamflow data were obtained from the GRDC, under the World Meteorological Organization (WMO)52. Data were collected for the Primera Angostura station (40.43°S, 63.67°W, 320 m a.s.l.) in the Negro River basin. This station is critical as the only control station in the mainstem, and it records natural streamflow. Post-processing involved excluding years with significant missing values and interpolating minor gaps (less than 1%). The data span from 1928 to 2018 for Primera Angostura.
2) Observed gridded monthly hydroclimatic data: This includes precipitation, maximum, mean, and minimum temperatures, potential evaporation, and the PDSI, sourced from the Climate Research Unit (CRU TS4.07) at the University of East Anglia, UK60,61. The SPEI dataset is provided by the Consejo Superior de Investigaciones Científicas (CSIC)62. Global monthly gridded surface runoff dataset (G-RUN) is obtained from the Institute for Atmospheric and Climate Science of the Eidgenössische Technische Hochschule Zürich (IAC ETH Zurich, Switzerland)83. These datasets cover 1928–2018, with a 0.5° × 0.5° resolution. Soil moisture data (0–10 cm depth) is from NASA’s FEWS NET Land Data Assimilation System (FLDAS) spans 1982–2018 at a 0.1° × 0.1° resolution63,64. Additionally, the Hadley Center of the UK Met Office provided the global monthly gridded SST dataset (HadISST1) with a 1° × 1° resolution95 and SLP dataset (HadSLP2r) with a 5° × 5° resolution96. An extra dataset of global surface temperatures is also used, which comes from the Berkeley Earth Surface Temperature Project, spanning 1850–2023 at a 1° × 1° resolution because it has data for both oceans and land in the same spatio-temporal framework97.
3) Paleo-hydroclimatic reconstructed data: Data primarily derive from the Paleo Hydrodynamics Data Assimilation product (PHYDA)53 and the South American Drought Atlas (SADA)8. PHYDA integrates 2978 global paleo-records with climate models through a data assimilation method to estimate hydroclimatic and atmospheric-oceanic conditions over the past 2000 years, with resolutions of ~2° for land and atmospheric clusters and ~1° for oceanic clusters. SADA reconstructs PDSI for South America (12°–56°S) using 286 tree-ring chronologies through point-by-point regression (PPR) method over the past 600 years. Both datasets have been reliably validated in capturing regional hydroclimatic changes, including in Patagonia7,17,39,98. It is worth noting that our reconstruction and PHYDA both use some tree-ring chronologies from the PAGES2k Network99, but there is no overlap with SADA. Overall, there are some differences between the three in their reconstructed goals, methods, and techniques. Meanwhile, our reconstruction also incorporates new chronologies, giving rise to distinctive features.
4) Paleo-hydroclimate simulation data: These are sourced from the CESM-LME, utilizing CESM1.1 runs77,78. This model integrates atmospheric, oceanic, land surface, sea ice, and land ice components. CESM-LME simulates Earth system dynamics over the last millennium, accounting for initial conditions, external forcings, and essential physical processes. The ensemble includes 13 all-forcing members, covering variables such as precipitation, surface temperature, wind speed, and specific humidity. Despite potential variability across runs, CESM-LME provides valuable insights into long-term interactions based on core physical principles, validated for revealing driving mechanisms, including in Patagonia16,17,77,78,100,101,102,103,104,105. The streamflow variable in CESM-LME has low correlation with our reconstruction, indicating that internal variabilities within the climate system have a profound influence on Negro streamflow changes17,106 (Fig. S12). However, recent trends show a downward trend, indicating that global warming is deeply altering the hydrological processes in the region.
Chronology development and reconstruction
The newly collected samples from northwestern Patagonia were pre-treated by drying, sanding, and polishing. Tree-ring widths were then measured with an accuracy of 0.001 mm using CooRecorder 9.4 software107. Tree-ring records from ITRDB were cross-dated using COFECHA software, being problematic cores excluded108. A composite tree-ring width chronology was developed from 140 cores of 72 trees using ARSTAN software109. A negative exponential function was applied to remove the effects of non-environmental factors over growth110. The standardized chronology was chosen to retain low-frequency signals, with the Expressed Population Signal exceeding 0.8, and extending the chronology back to 1197 CE25,111 (Fig. S2 & Table S1). Furthermore, a significant correlation between the old and new chronologies (r = 0.472, n = 778, covering 1197–1974) confirmed both the accuracy of cross-dating and the consistency of tree growth patterns.
For the Negro River, streamflow data from the Primera Angostura station was analyzed with our composite chronology. The optimal reconstruction correlation was found for austral summer (October to January) (r = 0.668, p < 0.001) (Fig. S3). A linear regression model between the composite chronology and October–January streamflow was calculated for 1928–2017 (R2 = 44.6%, Radj2 = 44.0%, n = 90, F = 70.793, p < 0.001) (Fig. S4). It is worth noting that more complex methods, such as nested principal component regression16,17,112, were not employed. Instead, a straightforward composite chronology and linear regression model were used, as the chronologies from the ITRDB were truncated in 1974 CE, limiting coverage of the full observed period. The linear regression model equation is given by:
where STR is the October–January streamflow and Wi is the tree-ring width index for year i. The split test method divided the entire period into the first 2/3 (1928–1987) and the last 1/3 (1988–2017) for further calibration and validation, including sign test, reduction of error, and coefficient of efficiency113 (Table S2). The correlation coefficient for first differences (difference between current year and previous year) of observed with reconstructed streamflow also supports a significant inter-annual comparison, excluding linear trends (r = 0.579, p < 0.001). Root mean square error (RMSE) is used to detect the uncertainty of reconstruction and observation17,24, and an RMSE (312.453 m3·s-1) less than one SD (351.849 m3·s-1) indicates that the uncertainty is acceptable.
Bias correction
Since the probability distribution of tree-ring reconstructions usually differs from observed data, particularly regarding over-estimation or under-estimation, we used the Quantile Mapping (QM) method to correct these biases54,55,56. Here, we have implemented QM using the bicus toolkit in Python57 (https://ibicus.readthedocs.io/en/latest/getting_started/overview.html). The core principle of QM is to adjust the quantiles of the original reconstruction by establishing a mapping relationship between the cumulative distribution functions of the reconstruction and the observation. The bias at a given quantile in the reconstruction is estimated by calculating the difference between the local regression and the quantile obtained from the observations. These different bias estimates for each quantile can then be added to the reconstructed quantiles. The bias correction significantly improved the capture of extremes in our reconstruction. Both the 1:1 plot and the quantile-quantile (QQ) plot show a more concentrated distribution that better fits the observed data (Fig. S6a, b). Specifically, the standard deviation (SD) during the observed period increased from 257.877 m3·s-1 to 385.596 m3·s-1, closely matching the observed streamflow standard deviation of 386.226 m3·s-1. Furthermore, throughout the entire reconstruction period (1197–2023), the correlation between the pre- and post-calibration reconstructions was as high as 0.992, but the SD increased from 234.615 m3·s-1 to 351.849 m3·s-1, indicating an enhanced ability to capture variability and extremes over the long term without changing the inherent characteristics of the reconstruction (Fig. S6c).
Definition of drought and decline severity
Generally speaking, drought refers to a prolonged water shortage that results in significantly lower levels of available water resources than the long-term average, and may have widely negative impacts on the ecological environment and socio-economic development72,114. There is no universally accepted or uniform method for defining drought, as it depends on the data, methods, and specific criteria used59,115. Of particular importance is the decline severity, which refers to the intensity of the dynamic process of systematic and sustained reduction in streamflow, leading to a continuous decline in available water resources116,117. A sustained reduction in streamflow is the most sensitive early warning sign that a water system is nearing or undergoing severe deterioration. The decline severity, combined with its rate and duration, can be used to assess the risk of accelerated stress and deep crisis, which is more predictive than directly monitoring low levels.
In this study, we first define extended pluvial or drought conditions based on 11-year low-pass filters for periods above or below the mean for more than 10 consecutive years17,105. Second, we use quantile regression to examine the temporal evolution of streamflow at different quantiles, with a particular focus on drying trends56. Last but not least, we are interested in the occurrence and extent of streamflow declines. We used an improved method to calculate the severity of the decline, taking into account both rate and duration59. The specific steps are: (1) Calculate the running 45-year slopes of the reconstruction; (2) Identify periods where slopes remained below zero for at least 11 years as declining periods; (3) Calculate the severity as the multiplication of the average slope and the duration of each period. Here, we select 45 years as the slope calculation window for two reasons. First, the 45-year window is half of the observed period and aligns with half a century, allowing for a comprehensive view of the evolving characteristics of interannual, decadal, and centennial scales, as well as the observed scale. Second, the median window length can avoid excessive anomalies, over-smoothing, and missing data at both ends. We have also calculated the results based on the running 61-year slopes for cross-validation, and the results are similar (Fig. S8). Ultimately, our assessment of the most “severe” event prioritizes the degree of sustained reduction, which is a function of both the magnitude of the decline and its duration.
Impacts and contributions from regional climate changes
Considering that the time spans of the observed gridded climate datasets vary and that early data are often inaccurate60,61,62,63,64,118, we choose the common time period (1982–2017) of all datasets for analysis. Pearson correlation analysis and spatial response patterns are used to determine the extent and scope of the impact of climate changes on Negro streamflow7,16,17,29,70,98,119. The Multivariate Linear Regression (MLR) method is used to reveal the contribution of climate variables to Negro streamflow variability16,73,74. The variance explained by the constructed MLR model is the total contribution of climate changes. However, due to co-linearity and inter-correlation among climate variables, it is not possible to separate the individual contributions of each climate variable by directly using the ratio of regression coefficients in MLR120. Here, we use the Hierarchical Partitioning (HP) method to solve the above problems and obtain more reliable and interpretable individual contribution assessments75. The specific steps are: (1) Standardize all independent variables to eliminate the influence of units of measurement; (2) Generate 1000 randomly arranged sequences of independent variables (Monte Carlo test)121; (3) Construct a stepwise regression model for each permutation and record the incremental value of the explained variance when each variable is added; (4) Calculate the arithmetic mean of the variance increments for each variable in all permutations, which is the individual explanatory variable. Based on the MLR and HP methods, we can evenly distribute the shared variance of related variables through random sorting and use 1000 Monte Carlo tests to ensure the stability of the results. In addition, the selection of the time window undoubtedly influences the quantitative contributions. However, given the current data selection and availability, an increase in duration and quantity does not necessarily guarantee greater accuracy. While extending the dataset could potentially alter the numerical values, it is unlikely to impact the qualitative outcomes significantly. Consequently, our findings remain robust and pertinent.
Large-scale teleconnection mechanisms
Large-scale atmospheric circulation is often one of the main mechanisms driving regional hydroclimatic changes8,16,39. Therefore, ENSO and SAM are taken into consideration because they are close to the south-central Andes and closely related to the Negro streamflow (Figs. 1c and S9). Spatial response and anomaly patterns are used to reveal potential teleconnection links70,122. We also use the Superposed Epoch Analysis (SEA) method123 to test the effects of reconstructed La Niña and positive SAM events124,125, corresponding to drought conditions, on streamflow variations. Events and positive or negative phases are determined by the years above or below the mean ± 1.5 SDs. That year is defined as the base year (year 0), and the four years before and six years after are also calculated. Results are considered significant only when they exceed the 5–95% range of 1000 Monte Carlo results. In the CESM-LME simulation, the ENSO index is calculated from the regional mean SST in the Niño 3.4 region (5°N–5°S, 170°W–120°W), which helps to reveal common signals of climate change in the Pacific Ocean126,127, and the SAM index is defined as the difference between the standardized latitudinal mean SLP at 40°S and 70°S, which helps to track the large-scale circulation signals in the Southern Hemisphere and Antarctic Oscillation44,128. The differences in the positive and negative phases of ENSO and SAM indices in large-scale climate space fields are used to reveal their driving mechanisms16,107,129,130. In addition, the emergence of Antarctic amplification and its potential effects are explored through regression and trend analyses50,122.
Future projections of a deep learning method for CMIP6 downscaling
For future streamflow projections, we use an ensemble of 24 models from the CMIP687 (Table S4) to obtain continuous runoff and surface temperature data for the period 1850–2100. Each model has been transformed to a 0.5° × 0.5° grid using bilinear interpolation and all models were averaged24,131. Given different units and spatio-temporal heterogeneity, we apply a deep learning method, specifically neural networks, to downscale the CMIP6 runoff data84,85,86.
The deep learning method is rigorously trained and validated using observed constraints to ensure robust future projections. We utilized the 0.5° × 0.5° gridded monthly runoff product (G-RUN)83 as our primary observed benchmark dataset. In the downscaling step, which is based on the assumption that CMIP6 is consistent with the observed hydrological convergence process, the procedures involve: (1) Spatially matching G-RUN grid cells with streamflow records from target hydrological stations; (2) Training the conventional neural network84 to establish the transfer function translating G-RUN runoff inputs to local streamflow outputs; (3) Chronologically splitting data into training (first 2/3) and independent test sets (last 1/3) for validation; (4) Constructing the most effective downscaling model, which has been well-trained and proven to be reliable, based on deep learning and applying it to the CMIP6 data. Currently, deep learning models are increasingly used in hydroclimatic downscaling fields for their ability to capture complex nonlinear spatio-temporal patterns85,86.
Here, the reason we choose the G-RUN gridded dataset is that it provides spatially and temporally continuous runoff data, and the time span and data accuracy are relatively excellent17,83,132 (Fig. S13). In other words, the G-RUN dataset provides a basis in reality for gridded data constrained to basin or site data, which can then be used to train a reliable deep learning model for downscaling CMIP6 data. In addition, observed rather than reconstructed streamflow is used because both G-RUN and CMIP6 only cover the past century, and downscaling using observed streamflow would be more accurate. The QM-corrected streamflow reconstruction can also be directly compared with observation, providing a baseline for assessing current and future streamflow changes relative to long-term natural variability. The training results of the deep learning model using the G-RUN gridded dataset and observed streamflow for downscaling are excellent, with correlation coefficients of over 0.9 during both the training and validation periods (Fig. S14).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Observed streamflow data from hydrological stations can be obtained at the Global Runoff Data Center (GRDC) (https://portal.grdc.bafg.de/). The tree-ring width data can be found in the International Tree Ring Data Bank (ITRDB) (https://www.ncei.noaa.gov/products/paleoclimatology/tree-ring). The reconstruction result can be downloaded on the National Tibetan Plateau / Third Pole Environment Data Center (https://doi.org/10.11888/Terre.tpdc.302985). The observed gridded hydroclimatic data can be obtained from the Koninklijk Netherlands Meteorologisch Instituut (KNMI) (https://climexp.knmi.nl/). The Paleo Hydrodynamics Data Assimilation product (PHYDA) and the South American Drought Atlas (SADA) are publicly available from the Zenodo data repository (https://zenodo.org/records/1198817) and the Center for Climate and Resilience Research (CR)2 (http://www.cr2.cl/datos-dendro-sada/), respectively. The Community Earth System Model-Last Millennium Ensemble (CESM-LME) database is provided by the Earth system grid (https://www.earthsystemgrid.org/dataset/ucar.cgd.ccsm4.CESM_CAM5_LME.html). All data from Coupled Model Intercomparison Project Phase 6 (CMIP6) simulations used in our analyses are freely available from the Earth System Grid Federation (ESGF) (https://esgf-node.llnl.gov/search/cmip6/). The topographic base-map is sourced from Natural Earth (https://www.naturalearthdata.com/).
References
Haddeland, I. et al. Global water resources affected by human interventions and climate change. Proc. Natl Acad. Sci. USA 111, 3251–3256 (2014).
Martínez-Valderrama, J., Guirado, E. & Maestre, F. T. Desertifying deserts. Nat. Sustain. 3, 572–575 (2020).
Wu, X. F. et al. Attribution and risk projections of hydrological drought over water-scarce Central Asia. Earth Future 13, e2024EF005243 (2025).
Swain, D. L. et al. Hydroclimate volatility on a warming Earth. Nat. Rev. Earth Environ. 6, 35–50 (2025).
Pokhrel, Y. et al. Global terrestrial water storage and drought severity under climate change. Nat. Clim. Change 11, 226–233 (2021).
Gaitán, J. J. et al. Plant species richness and shrub cover attenuate drought effects on ecosystem functioning across Patagonian rangelands. Biol. Lett. 10, 20140673 (2014).
Hadad, M. A., González-Reyes, A., Roig, F. A., Matskovsky, V. & Cherubini, P. Tree-ring-based hydroclimatic reconstruction for the northwest Argentine Patagonia since 1055 CE and its teleconnection to large-scale atmospheric circulation. Glob. Planet. Change 202, 103496 (2021).
Morales, M. S. et al. Six hundred years of South American tree rings reveal an increase in severe hydroclimatic events since mid-20th century. Proc. Natl Acad. Sci. USA 117, 16816–16823 (2020).
Neukom, R. et al. Facing unprecedented drying of the Central Andes? Precipitation variability over the period AD 1000-2100. Environ. Res. Lett. 10, 084017 (2015).
Fernández, A. et al. Dendrohydrology and water resources management in south-central Chile: lessons from the Rio Imperial streamflow reconstruction. Hydrol. Earth Syst. Sci. 22, 2921–2935 (2018).
González-Reyes, A. et al. Recent multispecies tree-growth decline reveals a severe aridity change in Mediterranean Chile. Environ. Res. Lett. 19, 064046 (2024).
Muñoz, A. A. et al. Streamflow variability in the Chilean Temperate-Mediterranean climate transition (35°S-42°S) during the last 400 years inferred from tree-ring records. Clim. Dyn. 47, 4051–4066 (2016).
Feron, S. et al. South America is becoming warmer, drier, and more flammable. Commun. Earth Environ. 5, 501 (2024).
Milly, P. C. D. et al. Climate change - Stationarity is dead: Whither water management?. Science 319, 573–574 (2008).
Di Baldassarre, G. et al. Sociohydrology: Scientific challenges in addressing the sustainable development goals. Water Resour. Res. 55, 6327–6355 (2019).
Chen, F. et al. Southeast Asian ecological dependency on Tibetan Plateau streamflow over the last millennium. Nat. Geosci. 16, 1151–1158 (2023).
Chen, F. et al. Role of Pacific Ocean climate in regulating runoff in the source areas of water transfer projects on the Pacific Rim. npj Clim. Atmos. Sci. 7, 153 (2024).
Natalia, P., Silvia, F., Silvina, S. & Miguel, P. Climate change in northern Patagonia: critical decrease in water resources. Theor. Appl. Climatol. 140, 807–822 (2020).
Zhang, Y. et al. Future global streamflow declines are probably more severe than previously estimated. Nat. Water 1, 261–271 (2023).
Rivera, J., Robo, M., Bianchi, E. & Mulleady, C. Impact of climate change on the streamflow in northern Patagonia. J. Water Clim. Change 15, 2074–2091 (2024).
Cook, E. R., Woodhouse, C. A., Eakin, C. M., Meko, D. M. & Stahle, D. W. Long-term aridity changes in the western United States. Science 306, 1015–1018 (2004).
Cook, E. R. et al. Asian monsoon failure and megadrought during the last millennium. Science 328, 486–489 (2010).
Williams, A. P. et al. Large contribution from anthropogenic warming to an emerging North American megadrought. Science 368, 318–318 (2020).
Rao, M. P. et al. Seven centuries of reconstructed Brahmaputra River discharge demonstrate underestimated high discharge and flood hazard frequency. Nat. Commun. 11, 6017 (2020).
Khan, N., Nguyen, H. T. T., Galelli, S. & Cherubini, P. Increasing drought risks over the past four centuries amidst projected flood intensification in the Kabul River Basin (Afghanistan and Pakistan)-evidence from tree rings. Geophys. Res. Lett. 49, e2022GL100703 (2022).
Viorica, N. et al. The first tree-ring reconstruction of streamflow variability over the last ~250 years in the Lower Danube. J. Hydrol. 617, 129150 (2023).
Zhao, X. E., Fang, K. Y., Chen, F., Martín, H. & Roig, F. A. Reconstructed Jing River streamflow from western China: A 399-year perspective for hydrological changes in the Loess Plateau. J. Hydrol. 621, 129573 (2023).
Boninsegna, J. A. et al. Dendroclimatological reconstructions in South America: a review. Paleogeogr. Paleoclimatol. Paleoecol. 281, 210–228 (2009).
Hadad, M. A. et al. Frost record in tree rings linked to atmospheric circulation in northern Check tor. Patagonia. Paleogeogr. Paleoclimatol. Paleoecol. 524, 201–211 (2019).
Gallardo, V. B., Hadad, M. A., Roig, F. A., Gatica, G. & Chen, F. Spatio-temporal linkage variations between NDVI and tree rings on the leeward side of the northern Patagonian Andes. For. Ecol. Manag. 553, 121593 (2024).
Urrutia, R. B. et al. Multicentury tree ring reconstruction of annual streamflow for the Maule River watershed in south central Chile. Water Resour. Res. 47, W06527 (2011).
Lara, A., Villalba, R. & Urrutia, R. A 400-year tree-ring record of the Puelo River summer-fall streamflow in the Valdivian Rainforest eco-region, Chile. Clim. Change 86, 331–356 (2008).
Masiokas, M. H. et al. Streamflow variations across the Andes (18°-55°S) during the instrumental era. Sci. Rep. 9, 17879 (2019).
Mundo, I. A. et al. Multi-century tree-ring based reconstruction of the Neuquén River streamflow, northern Patagonia, Argentina. Clim. Past 8, 815–829 (2012).
Holmes, R. L., Stockton, C. W. & LaMarche, V. C. Jr Extension of river flow records in Argentina from long tree-ring chronologies. Water Resour. Bull. 15, 1081–1085 (1979).
Stine, S. Extreme and persistent drought in California and Patagonia during mediaeval time. Nature 369, 546–549 (1994).
Tedeschi, R. G., Grimm, A. M. & Cavalcanti, I. F. A. Influence of Central and East ENSO on precipitation and its extreme events in South America during austral autumn and winter. Int. J. Climatol. 36, 4797–4814 (2016).
Agosta, E. A., Hurtado, S. I. & Martin, P. B. Easterlies”-induced precipitation in eastern Patagonia: seasonal influences of ENSO’S FLAVOURS and SAM. Int. J. Climatol. 40, 5464–5484 (2020).
Steiger, N. J., Smerdon, J. E., Seager, R., Williams, A. P. & Varuolo-Clarke, A. M. ENSO-driven coupled megadroughts in North and South America over the last millennium. Nat. Geosci. 14, 739–744 (2021).
Rehbein, A. & Ambrizzi, T. ENSO teleconnections pathways in South America. Clim. Dyn. 61, 1277–1292 (2023).
Liu, Y. J., Man, W. M., Zhou, T. J. & Zuo, M. Global multiproxy ENSO reconstruction over the past millennium. J. Geophys. Res. Atmos. 129, e2023JD040491 (2024).
Gillett, N. P., Kell, T. D. & Jones, P. D. Regional climate impacts of the Southern Annular Mode. Geophys. Res. Lett. 33, L23704 (2006).
Fogt, R. L. & Marshall, G. J. The Southern Annular Mode: Variability, trends, and climate impacts across the Southern Hemisphere. Wiley Interdiscip. Rev. Clim. Change 11, e652 (2020).
Dätwyler, C. et al. Teleconnection stationarity, variability and trends of the Southern Annular Mode (SAM) during the last millennium. Clim. Dyn. 51, 2321–2339 (2018).
Lyu, Z. Q. et al. South American monsoon intensification during the last millennium driven by joint Pacific and Atlantic forcing. Sci. Adv. 10, eado9543 (2024).
Torbenson, M. C. A. et al. Disruption of drought teleconnections between ENSO-influenced regions around 1700 CE. Geophys. Res. Lett. 52, e2025GL115600 (2025).
Roland, T. P. et al. Sustained greening of the Antarctic Peninsula observed from satellites. Nat. Geosci. 17, 1121–1126 (2024).
Clem, K. R. et al. Record warming at the South Pole during the past three decades. Nat. Clim. Change 10, 762–770 (2020).
Aitken, A. R. A. et al. Repeated large-scale retreat and advance of Totten Glacier indicated by inland bed erosion. Nature 533, 385–389 (2016).
Wu, Q. G. et al. Pacific sub-decadal sea surface temperature variations contributed to recent Aantarctic Sea ice decline trend. Nat. Commun. 16, 3386 (2025).
Morioka, Y. et al. Antarctic sea ice multidecadal variability triggered by Southern Annular Mode and deep convection. Commun. Earth Environ. 5, 633 (2024).
GRDC. Report 46 (Aug 2015): Twelfth Meeting of the GRDC Steering Committee (2015).
Steiger, N. J., Smerdon, J. E., Cook, E. R. & Cook, B. I. Data descriptor: a reconstruction of global hydroclimate and dynamical variables over the Common Era. Sci. Data 5, 180086 (2018).
Robeson, S. M., Maxwell, J. T. & Ficklin, D. L. Bias correction of paleoclimatic reconstructions: a new look at 1,200+years of upper Colorado River flow. Geophys. Res. Lett. 47, e2019GL086689 (2020).
Maxwell, J. T. et al. 1,100-year reconstruction of baseflow for the Santee River, South Carolina, USA reveals connection to the North Atlantic Subtropical High. Geophys. Res. Lett. 49, e2022GL100742 (2022).
Maxwell, J. T. et al. Recent increases in tropical cyclone precipitation extremes over the US east coast. Proc. Natl Acad. Sci. USA 118, e2105636118 (2021).
Spuler, F. R., Wessel, J. B., Comyn-Platt, E., Varndell, J. & Cagnazzo, C. ibicus: a new open-source Python package and comprehensive interface for statistical bias adjustment and evaluation in climate modelling (v1.0.1). Geosci. Model Dev. 17, 1249–1269 (2024).
Meko, D. M., Woodhouse, C. A. & Winitsky, A. G. Tree-ring perspectives on the Colorado River: looking back and moving forward. J. Am. Water Resour. Assoc. 58, 604–621 (2022).
Mccabe, G. J., Wolock, D. M. & Gangopadhyay, S. Past and projected future droughts in the upper Colorado River Basin. Geophys. Res. Lett. 51, e2023GL107978 (2024).
Harris, I., Jones, P. D., Osborn, T. J. & Lister, D. H. Updated high-resolution grids of monthly climatic observations - the CRU TS3.10 dataset. Int. J. Climatol. 34, 623–642 (2014).
van der Schrier, G., Barichivich, J., Briffa, K. R. & Jones, P. D. A scPDSI-based global data set of dry and wet spells for 1901-2009. J. Geophys. Res. Atmos. 118, 4025–4048 (2013).
Beguería, S., Vicente-Serrano, S. M., Reig, F. & Latorre, B. Standardized precipitation evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration models, tools, datasets and drought monitoring. Int. J. Climatol. 34, 3001–3023 (2014).
Kumar, S. V. et al. Land Information System - An interoperable framework for high resolution land surface modeling. Environ. Model. Softw. 21, 1402–1415 (2006).
Peters-Lidard, C. D. et al. High-performance Earth system modeling with NASA/GSFC’s Land Information System. Innov. Syst. Softw. Eng. 3, 157–165 (2007).
Zhao, X. E., Hu, M., Chen, F., Wang, S. J. & Hou, T. Y. Impact of Pacific Walker Circulation driven by Interdecadal Pacific Oscillation on precipitation variability in the North China Monsoon Marginal Region over the past five centuries. Glob. Planet. Change 253, 104913 (2025).
Jackson, R. B. et al. Water in a changing world. Ecol. Appl. 11, 1027–1045 (2001).
Cao, H. H. et al. Tree-ring insights into past and future streamflow variations in Beijing, Northern China. Water Resour. Res. 61, e2024WR038084 (2025).
Gangopadhyay, S., Woodhouse, C. A., McCabe, G. J., Routson, C. C. & Meko, D. M. Tree rings reveal unmatched 2nd century drought in the Colorado River Basin. Geophys. Res. Lett. 49, e2022GL098781 (2022).
Hou, T. Y. et al. Tree radial growth and vegetation dynamics in northern China and northern Patagonia regulated by ocean-atmosphere circulation. For. Ecol. Manag. 593, 122913 (2025).
Yue, W. P. et al. Late Ming Dynasty weak monsoon induced a harmonized megadrought across north-to-south China. Commun. Earth Environ. 5, 439 (2024).
Carroll, R. W. et al. Declining groundwater storage expected to amplify mountain streamflow reductions in a warmer world. Nat. Water 2, 419–433 (2024).
Gebrechorkos, S. H. et al. Warming accelerates global drought severity. Nature 642, 628–635 (2025).
Wang, S. J., Man, W. M., Chen, F., Zuo, M. & Tang, W. H. The relative contributions of internal variability and external forcing to Pacific Walker Circulation over the last millennium. J. Clim. 38, 219–233 (2024).
Zuo, M., Zhou, T. J. & Man, W. M. Understanding surface temperature changes over the Tibetan Plateau in the last millennium from a modeling perspective. Clim. Dyn. 621, 5483–5499 (2024).
Chevan, A. & Sutherland, M. Hierarchical partitioning. Am. Stat. 45, 90–96 (1991).
Mukherjee, S., Mishra, A. & Trenberth, K. E. Climate change and drought: a perspective on drought indices. Curr. Clim. Change Rep. 4, 145–163 (2018).
Hurrell, J. W. et al. The Community Earth System Model: a framework for collaborative research. Bull. Am. Meteorol. Soc. 94, 1339–1360 (2013).
Otto-Bliesner, B. L. et al. Climate variability and change since 850 CE: an ensemble approach with the Community Earth System Model. Bull. Am. Meteorol. Soc. 97, 735–754 (2016).
Grimm, A. M. & Tedeschi, R. G. ENSO and extreme rainfall events in South America. J. Clim. 22, 1589–1609 (2009).
Wu, M. N. et al. A very likely weakening of Pacific Walker Circulation in constrained near-future projections. Nat. Commun. 12, 6502 (2021).
Zhang, Z. R. et al. Evidence for large-scale climate forcing of dense shelf water variability in the Ross Sea. Nat. Commun. 15, 8190 (2024).
Sun, Q. H. et al. Possible increased frequency of ENSO-related dry and wet conditions over some major watersheds in a warming climate. Bull. Am. Meteorol. Soc. 101, E409–E426 (2020).
Ghiggi, G., Humphrey, V., Seneviratne, S. I. & Gudmundsson, L. G-RUN ENSEMBLE: a multi-forcing observation-based global runoff reanalysis. Water Resour. Res. 57, e2020WR028787 (2021).
D’Odorico, P., Revelli, R. & Ridolfi, L. On the use of neural networks for dendroclimatic reconstructions. Geophys. Res. Lett. 27, 791–794 (2000).
Baño-Medina, J., Manzanas, R. & Gutiérrez, J. M. Configuration and intercomparison of deep learning neural models for statistical downscaling. Geosci. Model Dev. 13, 2109–2124 (2020).
Baño-Medina, J. Understanding deep learning decisions in statistical downscaling models. In CI2020: Proceedings of the 10th International Conference on Climate Informatics, Association for Computing Machinery, 79–85 (New York, 2021).
Eyring, V. et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev. 9, 1937–1958 (2016).
Holland, P. R., Bracegirdle, T. J., Dutrieux, P., Jenkins, A. & Steig, E. J. West Antarctic ice loss influenced by internal climate variability and anthropogenic forcing. Nat. Geosci. 12, 718–724 (2019).
Li, Z. & Ding, Q. H. A global poleward shift of atmospheric rivers. Sci. Adv. 10, eadq0604 (2024).
Abram, N. J. et al. Emerging evidence of abrupt changes in the Antarctic environment. Nature 644, 621–633 (2025).
Blau, M. T., Kad, P., Turton, J. V. & Ha, K. J. Uneven global retreat of persistent mountain snow cover alongside mountain warming from ERA5-land. npj Clim. Atmos. Sci. 7, 278 (2024).
Collazo, S. et al. Influence of large-scale circulation and local feedbacks on extreme summer heat in Argentina in 2022/23. Commun. Earth Environ. 5, 231 (2024).
Li, J. B. et al. Deciphering human contributions to Yellow River flow reductions and downstream drying using centuries-long tree ring records. Geophys. Res. Lett. 46, 898–905 (2019).
de la Fuente, A., Meruane, C. & Suárez, F. Long-term spatiotemporal variability in high Andean wetlands in northern Chile. Sci. Total Environ. 756, 143830 (2021).
Rayner, N. A. et al. Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. Atmos. 108, 4407 (2003).
Allan, R. & Ansell, T. A new globally complete monthly historical gridded mean sea level pressure dataset (HadSLP2): 1850-2004. J. Clim. 19, 5816–5842 (2006).
Rohde, R. A. & Hausfather, Z. The Berkeley Earth land/ocean temperature record. Earth Syst. Sci. Data 12, 3469–3479 (2020).
Chen, F. et al. Coupled Pacific Rim megadroughts contributed to the fall of the Ming Dynasty’s capital in 1644 CE. Sci. Bull. 69, 3106–3114 (2024).
Consortium, P. k. A global multiproxy database for temperature reconstructions of the Common Era. Sci. Data 4, 170088 (2017).
Blackmon, M. et al. The community climate system model. Bull. Am. Meteorol. Soc. 82, 2357–2376 (2001).
Davis, N. A. et al. Climate, variability, and climate sensitivity of “middle atmosphere” chemistry configurations of the Community Earth System Model version 2, Whole Atmosphere Community Climate Model version 6 (CESM2(WACCM6)). J. Adv. Model. Earth Syst. 15, e2022MS003579 (2023).
Stevenson, S., Otto-Bliesner, B., Fasullo, J. & Brady, E. El Nino like” hydroclimate responses to last millennium volcanic eruptions. J. Clim. 29, 2907–2921 (2016).
Roldán-Gómez, P. J., González-Rouco, J. F., Smerdon, J. E. & García-Pereira, F. Model and proxy evidence for coordinated changes in the hydroclimate of distant regions over the Last Millennium. Clim. Past 19, 2361–2387 (2023).
Hu, M. et al. Strong coupling between soil moisture and temperature intensifies warming and humidification on the Tibetan Plateau: Evidence from 200-year tree ring records. Paleogeogr. Paleoclimatol. Paleoecol. 644, 112206 (2024).
Wang, S. J. et al. Past, present and future changes in the annual streamflow of the Lancang-Mekong River and their driving mechanisms. Sci. Total Environ. 947, 174707 (2024).
Wang, S. J., Wang, F., Yang, B., Qin, C. & Dannenberg, M. P. Recent and future climate extremes in the northeastern Qinghai-Tibet Plateau under anthropogenic forcing. Geophys. Res. Lett. 52, e2025GL115814 (2025).
Wang, S. J. et al. Reconstructed springtime (March-June) precipitation tracked by tree rings dating back to 1760 CE in the Qinling-Bashan mountainous area. Paleogeogr. Paleoclimatol. Paleoecol. 604, 111211 (2022).
Cook, E. R. A Time Series Analysis Approach to Tree Ting Standardization (University of Arizona, 1985).
Holmes, R. L. Computer assisted quality control in tree-ring dating and measurement. Tree-Ring Bull. 43, 69–78 (1983).
Fritts, H. C. Tree-Rings and Climate (Academic Press, 1976).
Wigley, T. M. L., Briffa, K. R. & Jones, P. D. On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. J. Appl. Meteorol. Climatol. 23, 201–213 (1984).
Wang, J. L. et al. Internal and external forcing of multidecadal Atlantic climate variability over the past 1,200 years. Nat. Geosci. 10, 512–517 (2017).
Cook, E. R., Meko, D. M., Stahle, D. W. & Cleaveland, M. K. Drought reconstructions for the continental United States. J. Clim. 12, 1145–1162 (1999).
Lian, X. et al. Multifaceted characteristics of dryland aridity changes in a warming world. Nat. Rev. Earth Environ. 2, 232–250 (2021).
Ault, T. R. On the essentials of drought in a changing climate. Science 368, 256–260 (2020).
Peng, K. et al. Decomposing the effects of changes in catchment characteristics on runoff into chain transmission effects of climate change and human activities using an improved Budyko framework. Earth Future 13, e2025EF006041 (2025).
Zhang, Y. Q. et al. Future global streamflow declines are probably more severe than previously estimated. Nat. Water 1, 261–271 (2023).
Menemenlis, S., Vecchi, G. A., Yang, W. C., Fueglistaler, S. & Raghuraman, S. P. Consequential differences in satellite-era sea surface temperature trends across datasets. Nat. Clim. Change 15, 897–903 (2025).
Hu, M. et al. Increasing evidence of an anthropogenic signal in drought variations on the river source areas of southeastern Tibetan Plateau. J. Hydrol. 660, 133508 (2025).
García-Palacios, P., Gross, N., Gaitán, J. & Maestre, F. T. Climate mediates the biodiversity-ecosystem stability relationship globally. Proc. Natl Acad. Sci. USA 115, 8400–8405 (2018).
Adams, J. B., Mann, M. E. & Ammann, C. M. Proxy evidence for an El Nino-like response to volcanic forcing. Nature 426, 274–278 (2003).
Wang, S. J. et al. Greening of Eurasia’s center driven by low-latitude climate warming. For. Ecosyst. 13, 100330 (2025).
Haurwitz, M. W. & Brier, G. W. A critique of the superposed epoch analysis method: its application to solar–weather relations. Mon. Weather Rev. 109, 2074–2079 (1981).
Emile-Geay, J., Cobb, K. M., Mann, M. E. & Wittenberg, A. T. Estimating central equatorial Pacific SST variability over the past millennium. Part II: reconstructions and implications. J. Clim. 26, 2329–2352 (2013).
Abram, N. J. et al. Evolution of the Southern Annular Mode during the past millennium. Nat. Clim. Change 4, 564–569 (2014).
Ropelewski, C. F. & Jones, P. D. An extension of the Tahiti–Darwin Southern Oscillation Index. Mon. Weather Rev. 115, 2161–2165 (1987).
Chen, F. et al. Central Asian river streamflows have not continued to increase during the recent warming hiatus. Atmos. Res. 246, 105124 (2020).
Nan, S. L. & Li, J. P. The relationship between summer precipitation in the Yangtze River valley and the previous Southern Hemisphere Annular Mode. Geophys. Res. Lett. 30, 2266 (2003).
Hu, M. et al. Climate change increases the instability of the water supply for hydropower stations on the Tibetan Plateau. Environ. Earth Sci. 18, 114040 (2023).
Lu, Z. Y. et al. Increased frequency of multi-year El Nino-Southern Oscillation events across the Holocene. Nat. Geosci. 18, 337–343 (2025).
Xu, Z. F., Han, Y., Tam, C. Y., Yang, Z. L. & Fu, C. B. Bias-corrected CMIP6 global dataset for dynamical downscaling of the historical and future climate (1979-2100). Sci. Data 8, 293 (2021).
Yi, S., Saemian, P. & Sneeuw, N. Estimating runoff from pan-Arctic drainage basins for 2002-2019 using an improved runoff-storage relationship. Remote Sens. Environ. 298, 113816 (2023).
Acknowledgements
This work was supported by the National Natural Science Foundation of China (NSFC project No. 32061123008) and Excellent Research Group Program for Tibetan Plateau Earth System (No. 42588201). This work was also supported by the Agencia Nacional de Promoción Científica y Tecnológica of Argentina (PICT-2018-1056 to MAH), a Cooperation International Project between CONICET and NSFC-2019. We acknowledge the efforts of the Global Runoff Data Center (GRDC) for organizing observed streamflow data and the tree-ring scholars for contributing tree-ring records to the International Tree Ring Database (ITRDB). We also extend our thanks to the editors and reviewers for their valuable comments and suggestions.
Author information
Authors and Affiliations
Contributions
F.C. conceived and designed the original study. S.J.W. completed the chronology development, data analysis, and drafting of the manuscript. M.H. and X.E.Z. assisted with data analysis and visualization. F.C., M.A.H., F.A.R., D.M.M. and M.C.A.T. made critical modifications and supervised the language of the manuscript. F.C., M.A.H., F.A.R., S.P. and H.L.Z. conducted the sample collection. Y.X., Y.P.C., W.P.Y. and H.H.C. assisted with sample processing and manuscript preparation. All authors contributed to interpreting the results and discussions and revised the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Communications Earth and Environment thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Soumaya Belmecheri, Alireza Bahadori and Aliénor Lavergne. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Wang, S., Hu, M., Chen, F. et al. Recent south-central Andes water crisis driven by Antarctic amplification is unprecedented over the last eight centuries. Commun Earth Environ 6, 937 (2025). https://doi.org/10.1038/s43247-025-02858-1
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s43247-025-02858-1
