Deglaciation in the subtropical Andes has led to a peak in sediment delivery

Mountain glaciers are powerful agents of erosion and supply substantial sediment and solutes to downstream environments, influencing geomorphology, ecosystems, water resources, and hydropower. As climate warming drives glacier thinning and retreat, their erosive capacity is expected to decline, yet theory and observations suggest a transient increase in sediment export—termed peak sediment—can occur as subglacial hydrological connectivity expands and mobilizes long-stored sediments. The timing and evolution of peak sediment are hard to evaluate due to unknowns in glacier thermal regimes, subglacial sediment storage, and hydro-sedimentological connectivity, as well as the delayed, non-linear catchment response to climate forcing. This study investigates whether subtropical Andean catchments have already passed their peak sediment by analyzing six decades of suspended sediment concentration (SSC) from 11 rivers (27–35°S), quantifying multidecadal changes in SSC and its relationship to modeled glacier ice melt, and inferring implications for future sediment flux.

Prior work documents the strong role of glaciers in mountain erosion and sediment delivery and the expectation that glacier mass loss reduces basal shear stress and sliding velocities, decreasing long-term erosion. However, hydrological reorganization during deglaciation can transiently increase sediment export as subglacial drainage extends and accesses stored till, analogous to peak water. Peak sediment magnitude and duration depend on climate variability, glacier thermal regime, bed topography, and sediment stores, and may persist beyond peak water and over century timescales. Observational and modeling studies from the Andes, Alps, Arctic, and Himalaya report contemporary increases in sediment export linked to warming, while broader projections suggest regional to global peak sediment may end between the late 21st and 22nd centuries, though basin-scale analyses indicate heterogeneity with some catchments near current maxima and declines later this century. In the subtropical Andes, recent increases in turbidity have been linked to enhanced connectivity of sub- and proglacial sediment sources under warm, dry conditions, with limited roles for land-use change, GLOFs, or permafrost-related slope instability.

Study region and data: Eleven subtropical Andean river basins between 27–35°S with varying glacier coverage were analyzed. Gauging stations are tens to ~100 km downstream of glacier termini. Daily suspended sediment concentration (SSC; mg l−1) from Chile’s DGA (1964–2017) were quality controlled, with single-day gaps linearly filled. Monthly means were computed when >10 daily measurements existed; otherwise, infilling used adjacent years’ same-month values or adjacent months within year. Annual means required ≥10 qualifying months; warm-season (October–March) means required ≥4 qualifying months. Emphasis was on warm season due to activation of subglacial drainage and higher glacial sediment signal. A gauge with anomalous values (Río Illapel en las Burras) was excluded. Suspended sediment yield (SSY) was computed by SSC × concurrent streamflow for supplementary analyses. Comparisons between 1964–1981 and 2000–2017 used gauges with ≥6 years in each period (9 gauges annual; 11 warm season). Glacier runoff: Daily glacier ice melt for all glaciers in the Maipo basin (1955–2020) was taken from TOPKAPI-ETH, a distributed glacio-hydrological model constrained by geodetic mass balance, snow cover, and local hydro-meteorological data. Outputs included annual glacier area/volume and daily rainfall, snowmelt, and ice melt (from on-glacier and off-glacier sources). Annual total ice melt (km³ yr−1) represents glacial streamflow magnitude; specific ice melt (m yr−1) normalizes by glacierized area to indicate climatic forcing strength and potential subglacial connectivity. Regime shift detection: Sequential Regime Shift Detection (Rodionov, 2004) assessed temporal changes in SSC (and other series), with target significance level 0.1, cut-off length 15 years, and outlier weighting factor h = 4. The method computes a regime shift index and applies sequential unequal-variance two-tailed t-tests to identify step changes in mean values, robust to outliers via weighting. Sensitivity tests with alternative parameters yielded similar results. Extreme turbidity events (ETE) and ice melt–sediment linkage: ETE were defined as days with SSC above the 85th percentile during the warm season. The Aconcagua basin, with the fewest gaps and highest correlation with regional reference turbidity, was used to build a consistent ETE time series (results cross-checked using Maipo). To minimize rainstorm influence, a melt-driven ETE series excluding days with precipitation was also constructed (91% of ETE), yielding similar results. Modeled annual ice melt from the adjacent Maipo basin was used as a proxy for proglacial streamflow; basins are adjacent with similar hypsometries and glacier properties, and ice melt in Maipo correlates strongly with Juncal Norte glacier melt in Aconcagua (R² = 0.86). The interannual relationship between ice melt and ETE frequency was analyzed via linear regression (ETE vs ice melt) and correlation coefficient r, computed in 7-year sliding windows and summarized with 9-year moving averages. The slope reflects sensitivity of ETE frequency to ice melt (erodibility, channel geometry, and access to new sediment sources), and the intercept serves as a proxy for sediment availability when r is positive. A power-law alternative produced analogous conclusions. Turbidity records (hourly, since 1990) for Maipo provided corroboration but were not used for long-term trend analysis. Climatic context: ERA5-Land monthly temperature and precipitation anomalies over 27–35°S, 69–72°W characterized multidecadal climate variability, including dry/warm conditions in the 2010s and below-average precipitation in the 1970s. The Pacific Decadal Oscillation (PDO) context and cryospheric mass-balance memory were considered for interpreting connectivity changes.

- Three SSC regimes were identified in glacierized basins with the most complete records (Aconcagua, Mapocho, Maipo): (I) high SSC from late 1960s to mid-1970s, (II) low SSC from late 1970s to 2000s, and (III) renewed high SSC in the 2010s. The magnitude of regime III is about half that of regime I, despite both being high-SSC phases. - Across the broader set of basins, the higher SSC magnitude during regime I (1970s) is evident primarily in glacierized catchments; non-glacierized basins did not show the same change, indicating a cryospheric origin of the change. - In the Maipo basin, both total and specific ice melt were greater during the high-SSC regimes (I and III) than during regime II, consistent with enhanced subglacial connectivity in high-SSC phases. Regime I had greater total ice melt but lower specific melt than regime III, consistent with larger glacierized area and slightly colder/wetter climate in the 1970s. Differences were not statistically significant at 95% but were coherent with independent climatic and glaciological context. - The interannual relationship between ice melt and ETE frequency shifted across regimes: r was significantly positive in regimes I and III and significantly negative in regime II. Slopes were similarly positive and significantly steeper in I and III than the negative slopes in II, indicating ETE frequency was more sensitive to ice melt in high-SSC phases. The intercept—used as a proxy for sediment availability when r is positive—was significantly higher in regime I and decreased through regime II to regime III, with a slight recent recovery. This pattern points to reduced availability of easily mobilized glacial sediment compared to the 1970s. - Combining the declining intercept (sediment availability) with decreasing regional glacial runoff implies that most glacierized subtropical Andean catchments have already passed the maximum of peak sediment; future sediment export will likely decline gradually. - Basin-scale analysis shows a long-term decrease in SSC with decreasing glacier volume. An exception is the high-elevation, colder Pulido basin, which shows increased SSC despite glacierization, suggesting that cold-based, high-elevation glaciers may not yet have reached their peak sediment and may have lower peak magnitude due to reduced sliding and smaller sediment stores. - Multidecadal variability in the ice melt–ETE link reflects both regional climate variability (e.g., PDO-related precipitation swings) and cryospheric memory effects (e.g., multi-year development/closure of moulins and subglacial conduits), which modulate hydraulic connectivity and thus sediment evacuation. - While not the primary focus, suspended sediment mass flux (SSY) had its highest values in the 1970s–1980s, consistent with SSC-based inferences of an earlier peak.

The study’s central question—whether subtropical Andean catchments have passed peak sediment—was addressed by combining long-term SSC records with modeled glacier ice melt and examining their evolving relationship. High SSC phases coincide with periods of stronger glacial melt and enhanced subglacial connectivity, confirming cryospheric control on sediment export. The significant reduction in the regression intercept from the 1970s to the 2010s, interpreted as diminished sediment availability, together with declining glacial runoff, supports the conclusion that most glacierized basins in this region have already passed the peak in sediment delivery. The exception of colder, higher-elevation basins (e.g., Pulido) highlights spatial heterogeneity governed by glacier thermal regime and elevation, with some basins likely yet to reach peak sediment and possibly experiencing a lower peak magnitude. The work underscores the importance of multidecadal climate variability and glacio-hydrological memory in modulating connectivity and sediment evacuation, implying that peak timing and shape are not solely monotonic functions of warming but are filtered by basin characteristics and climate oscillations. These findings have direct implications for reservoir management, water supply reliability, hydropower operations, and coastal ecosystems that depend on sediment and nutrient delivery.

Using uniquely long SSC records (since the 1960s) from 11 subtropical Andean rivers and modeled glacier melt, the study identifies alternating SSC regimes and shows that the 1970s exhibited the highest sediment concentrations, exceeding the recent decade’s high values. The evolving ice melt–ETE relationship indicates reduced sediment availability today relative to the 1970s, consistent with progressive exhaustion of glacially stored sediment. Coupled with declining glacial runoff, this suggests that peak sediment has already passed in most glacierized subtropical Andean catchments, while some high-elevation, cold-based glacier basins may not yet have peaked. The magnitude and timing of peak sediment are strongly conditioned by topography, climate variability, and hydro-sedimentological connectivity. Future research should prioritize: direct characterization of glacier thermal regimes and subglacial sediment stores; expanded monitoring in high-elevation, cold-based glacier basins; integration of multidecadal climate variability (e.g., PDO) and cryospheric memory into predictive models; improved proximal measurements of proglacial sediment and discharge to reduce far-field uncertainties; and coupled modeling of sediment production, storage, and evacuation to refine projections of post-peak decline.

Key uncertainties arise from limited knowledge of glacier thermal regimes, subglacial sediment storage, and basin-scale hydro-sedimentological connectivity. Many SSC gauges are far downstream from glacier termini, necessitating reliance on proxies (e.g., ETE frequency, regression intercepts) that may be influenced by non-glacial events, though precipitation-filtered analyses suggest a minor role for rain-driven extremes. Only one basin (Maipo) had detailed glacier ice melt modeling for the full period, and inter-basin transfer of melt metrics assumes similarity supported by correlations but not direct measurement. Some inferred differences between regimes (e.g., in total and specific melt) were not statistically significant at the 95% level. SSC data required gap-filling and aggregation rules that could smooth extremes. Suspended sediment mass flux was not analyzed in depth in the main text. Secondary sediment sources (permafrost degradation, paraglacial landslides, and a single GLOF in the early 1980s) may contribute episodically. Unmonitored basins introduce regional extrapolation uncertainty.