The Three Gorges Dam accelerates acidication of upland soils

Large hydroelectric dams have provided extensive benets in supporting human socio-economic development but their environmental footprint has come under increasing scrutiny. Lack of continued long-term monitoring of the environment remains a key constraint to the understanding of the correlation between dam operation and the environmental changes. Here we assess the impact of the Three Gorges Dam on upland soils in the reservoir area. By tracking the changing of pH value, a general indicator of soil health, over a period spanning nearly three decades before and after the dam construction, we observe continuous pH decline evidencing ongoing soil acidication. In addition to the intensive proton supply from applied fertilizers, the presence of the megaproject is believed to be a major driver of the accelerated soil acidication. Further analysis reveals that the shifts of microclimatic conditions as a result of the dam leads to the enhanced decomposition of organic species and the concomitant loss of exchangeable base cations, which contribute additional protons to soil and reduce soil acid-buffering capacity, respectively. Our ndings identify the ecological impacts of the world’s largest dam and suggest the urgency to initiate environmental protection measures for improved sustainability.

despite some parameters such as soil fertility being included in TGPEEMS. To the best of our knowledge, the undetermined consequences of the dam construction on soil quality in the uplands in the TGD reservoir area, has been overlooked, largely because of the lack of analysis and interpretation of longterm in-situ data collected before and after TGD construction.
Agricultural lands in the TGD reservoir area, with only 0.07 ha/person, are predominantly located in the surrounding mountainous region 7 . Due to soil erosion and the widespread use of fertilizers with high application rates, there is a critical need to assess if soil quality in the uplands is being degraded. Soil acidity, represented by soil pH, is a widely applied metric for soil quality, and its change affects biogeochemical cycles, and, subsequently, soil fertility and soil ecosystem health 23

. A survey in 2017
indicates that the pH of TGD upland soils (75 samples from altitudes between 200 and 1700 m) ranged from 3.66 to 7.77. Importantly, the pH of 49% of the sampling sites was < 5.5 which is often used as the threshold value of healthy and sustainable soils 24 . Such data suggest that soil acidi cation in the TGD uplands has entered an alarming stage, and there is an urgent need for rehabilitation.
In this context, we have performed a systematic study of upland soil to extract insight into the acidi cation dynamics and better inform the policy makers. Although the experimental watershed and the control watershed ( Supplementary Fig. 1) have similar geographical characteristics, crop types and fertilizer management methods, the soil pH in the experimental watershed decreased relatively faster than that in the control watershed. Speci cally, the average soil pH in the experimental watershed declined from 6.56 in 2003 to 5.62 in 2018. This represents a difference in pH (DpH) = −0.94 over the 16-year period. This equates to an 8.7-fold increase in acidity on the basis of the concentration of protons, and is likely to have a substantial impact on plant growth 25 . However, soil pH in the control watershed exhibited a DpH = -0.32, equivalent to a 2.1-fold increase in acidity between 2003 and 2018 (Fig. 1a). Historically, before dam construction, the average soil pH of the TGRA was around 6.80 (1990−2002). The average value from 2003 to 2018, spanning pre-and post-dam construction was 6.11. Importantly, our analysis reveals that the decrease in soil pH after dam construction is faster than that before (Fig. 1b).
We also monitored the latitude dependence of soil. Over the past 16 years, changes in soil pH for the different altitude categories in the experimental watershed are as follows: a decrease from 7.11 to 5.95 (DpH = −1.16) for low (200-600 m) altitudes; 6.43 to 5.51 (DpH = −0.92) for intermediate (600−900 m) altitudes; and 6.13 to 5.31 (DpH = −0.82) for high (900−1200 m) altitudes. Therefore, the reduction in soil pH was most pronounced at low altitudes (Fig. 1c).
To determine why the reduction in soil pH was accelerated after construction of the TGD, we conducted a controlled experiment to assess the change in soil pH under simulated microclimate conditions before and after dam construction. The changed soil pH was the difference between the original soil pH and the average of soil pH provided by the three most recent samples. Our results show that soil pH decreased both before (DpH = 1.08) and after (DpH = 1.17) dam constructions at low altitude (200 m). Similarly, the decrease in pH after dam construction (DpH = -1.09) at intermediate altitude (700 m) was also slightly greater than that before (DpH = -1.04) (Fig. 1d&e).
To explore the reasons accelerating soil acidi cation, we tested the contributions of acid deposition 26 , soil nitri cation and NH 4 + fertilizer nitri cation 27 , organic matter decomposition 28 and exchangeable base cations loss 28 to changes in soil pH because they are known as key factors driving soil acidi cation 25,29 . Firstly, the contribution of acid deposition on decreasing soil pH can be excluded. This is because the average measured rainwater pH in the experimental area over the sampling period was 6.77 (Supplementary Table 1), suggesting that the acidi cation due to rainfall was negligible.
Then for insight into the soil nitri cation (actually soil nitri cation based on two substrates: soil itself and NH 4 + fertilizer in the soil), we determined the nitrifying microorganisms in the experimental watershed because nitri cation in soil is driven by microorganisms 30  To quantify the contribution of NH 4 + fertilizer nitri cation to soil acidity, we investigated the effect of fertilizer applications on changes in soil pH based on the simulated 'NH 4 + fertilizer' and 'no NH 4 + fertilizer' treatments. Results show that the DDpH contributions (i.e., the differences between the DpH values with and without fertilizers) of fertilizers after dam construction (+0.03 at low altitude and +0.07 at intermediate altitude) were similar to those before the dam construction (+0.05 at low altitude and +0.06 at intermediate altitude) (Fig. 1d&e). This implies that the contributions of fertilizers before and after TGD construction are similar. These simulated results further suggest that the accelerated decline in soil pH in the uplands of the TGD watershed is attributable to the dam construction.  Fig. 2a). Signi cant logarithmic or linear relationships between soil pH and the abundances of SOM-decomposers were observed ( Supplementary Fig. 2b&c), implying that the decline in soil pH was directly associated with SOM decomposition.
Finally, we focused on the effects of exchangeable base cations on shifts in soil pH. Given the pH is <6.2, the acidi cation process is buffered by base cations, such as exchangeable Ca 2+ and Mg 2+ 38 . The depletion of base cations (Ca 2+ , Mg 2+ and Na + ) occurred in the experimental watershed after dam construction ( Supplementary Fig. 3), resulting in a decline of acid-buffering capacity. There were good linear relationships between the soil pH values and the exchangeable Ca 2+ , Mg 2+ and Na + concentrations in soils (Fig. 2). These results indicate that the decline in soil pH was partly because of the decreased acid-buffering capacity. To explain why the abundances of nitrifying microorganisms and SOM-decomposers in upland soils increased with time after the TGD construction, we investigated their relations with environmental factors. The change in environmental conditions, and especially local microclimate conditions, was one of the most important factors driving the change in microbial populations 39,40 . It has been previously reported that the microclimate (evaporation, relative humidity, soil temperature and precipitation) in the experimental watershed changed after TGD construction 6 . Thus, these four microclimatic factors were selected to conduct redundancy analyses (RDA) with nitrifying microorganisms and SOM-decomposers ( Supplementary Fig. 4). Our results demonstrate, unambiguously, that the change in the microclimate for the experimental watershed has a substantial promoting effect on the nitrifying microorganisms and SOM-decomposers.
Why did the major base cations decrease continuously year by year? We investigated changes in the soil particle size after the construction of the TGD (Fig. 3). After a long period of erosion, soil particle size changed in the experimental watershed, with the average proportions of silt and clay (from 2004 to 2018) decreasing from 41.5% to 32.4% and 17.3% to 15.7%, respectively, whereas the average proportion of sand increased from 41.5% to 52.0% (Fig. 3). There was a signi cant negative correlation between soil sand content and exchangeable base cations, and a signi cant positive correlation between soil clay or silt content, and exchangeable base cations (p < 0.05) (Supplementary Table 2). Here, it is likely that the reduction in soil clay content in the experimental watershed reduced the content of exchangeable base cations since the adsorbed cations are carried away with the eroded soil particles 41,42 . On the other hand, precipitation directly caused base cations to migrate with water, since they are solubilized in the more acidic runoff. Fig. 3 The change in soil particles size composition in the experimental watershed from 2005 to 2018.
Change in soil particle size composition refers to the difference between particle size composition in a given year and that in the rst year (2004).
Partial least squares path modeling (PLS-PM) was conducted to integrate the time series in processes and changes in soil pH (Fig. 4a). Dam-induced water level uctuation was the dominant factor indirectly affecting the change in microclimate (path coe cient = -0.80). For the nitrifying microorganisms and the SOM-decomposers, the change in the local microclimate had negative effects. The losses of exchangeable base cations were directly due to the shifts of local microclimate conditions. The decrease in soil pH was directly affected by nitrifying microorganisms (path coe cient = -0.28), SOM-decomposers (path coe cient = -0.15), soil exchangeable base cations (path coe cient = 0.98) and fertilizer application (path coe cient = -0.31). It is apparent from this analysis that the water level uctuation caused by the construction of the TGD indirectly contributed to the changes in soil pH in the experimental watershed. An increase in water level results in a lower soil pH (DpH = 0.69, the difference in the average soil pH spanning before and after damming). This study indicates that large dam construction may indirectly disturb upland soil quality in the dam watershed through facilitating H + accumulation and reducing soil pH buffering capacity. The main processes causing soil acidi cation have been revealed by our work (Fig. 4b). (1)  Erosion decreases the soil clay content and leads to the loss of soil exchangeable base cations, resulting in a decline in soil acid-buffering capacity.
Mitigation of soil acidi cation for improving soil ecosystem health for sustainability is urgently needed in the uplands of TGRA. This urgency underscores the need for a soil protection plan to alleviate soil acidi cation using easy-operation, low-cost and effective technologies. Ecological agricultural engineering that manifests long-term dual bene ts -reductions of nutrient and soil losses, and improvements to soil quality, is encouraged 43 . For a long term perspective, we appeal to a national policy to support the continued running of the TGPEEMS, and the inclusion of soil quality and soil health parameters in the TGPEEMS for a comprehensive understanding and assessment of the environmental consequences of TGP. To maintain the long-term sustainability of the uplands of TGRA, a targeted land use policy promoting a reduction of steep slope farming and conversion to forest and grassland is encouraged. Our ndings, although based on the TGRA, are useful for policy making and regular management for the upland soil ecosystems of other large reservoirs worldwide. Figure 3 The change in soil particles size composition in the experimental watershed from 2005 to 2018. Change in soil particle size composition refers to the difference between particle size composition in a given year and that in the rst year (2004).