Prominent role of sulfate reduction in robust sulfur retention in subtropical soil

Sulfur budgets in catchments indicated that about 80% of the deposited sulfur was retained in the subtropical soil, it alleviates the historical acidication caused by elevated deposition. The strong sulfur retention was attributed to the reversible sulfate adsorption in previous studies. Here we report that sulfate reduction is a prominent yet thus far overlooked mechanism for sulfur retention, based upon the comprehensive evidence of soil sulfur storage and multi-isotope within entire soil prole along a hydrological continuum in a typical subtropical catchment of China. Using a dual isotopic mass balance model, we determined that annual ux of reduction accounted for approximately 38% of sulfur retention, which was close to the proportion of reduced species in soil. Consequently, the release of sulfur legacy would be less serious with the decreasing sulfur deposition in China, compared to the projections only considering adsorption.


Introduction
Atmospheric sulfur (S) deposition has caused widespread soil and surface water acidi cation, and led the loss of base cation and nutrient, and the mobilization of harmful metals [1][2][3] . Sulfate (SO 4 2-) retention in ecosystems has been implicated as a counter effect to the environmental impacts derived from S deposition [1][2][3] . It was poor in temperate forests of Europe and North America, manifesting as almost all the deposited sulfur (S) leaching out even with moderate atmospheric input (Fig. 1a). However, signi cant S retentions (as large as 160 kg S ha -1 yr -1 , accounting about 80% of the total S deposition) have been widely reported in subtropical cathments, even under chronically heavy S deposition (Fig. 1).
The mechanisms for sulfur retention remains unclear in subtropical forest catchments. Sulfate adsorption has usually been reported as the main mechanism for SO 4 2retention in forest catchments 4-7 , and used as the only factor for modeling and predicting the acidi cation processes 8,9 . While a large reservoir of adsorbed SO 4 2-was observed due to the strong adsorption capacity, this could not explain the long-term S sink in the deeper soil and before-stream zone in subtropical regions 10,11 . In addition, S incorporation into soil organic matter might not be the reasonable cause to the strong SO 4 2retention due to the limited organic matter even in the upper soil in subtropical forests, though some studies have emphasized the importance of S incorporation into soil organic matter 5,12 . Sulfate reduction usually played a minor role or ignored in forested catchments dominated by aerobic soil 5,6,12,13 , but could be an additional mechanism for attenuating SO 4 2in catchments with a larger area of wetland [14][15][16] . Its contribution to S retention in subtropical must be con rmed.
The S retention mechanisms were di cult to distinguish and quantify merely based on the traditional ux/budget approach. Sulfur isotope is a useful tool to identify the sulfate reduction in ecosystems 14-17 , due to the distinctive isotope fractionation effects ( 34 ε) between the sulfate reduction (as large as 72‰) and the other processes (~0‰) 18 . Furthermore, the isotope approach combined with mass balance could be used to estimate uxes of SO 4 2turnover processes 15 . In this study, we determined the hydrological SO 4 2uxes, soil S storage, and δ 34 S and δ 18 O within the entire soil pro le along a hydrological continuum in a typical subtropical catchment of China (TSP, Fig. 1b & 2). Sulfate reduction was ascertained by the isotopic signal, and the mechanisms for S retention were quanti ed as using a model involving mass and isotope balance. The results showed that SO 4 2adsorption, reduction, and net incorporation into organic matter, accounting 46%, 30%, and 4% of the total deposited S, respectively.
Potential environmental hazards of the legacy S in subtropical soil were further proposed.

Results And Discussion
Large S retention in the subtropical soil TSP catchment had a S retention average of 141 kg S ha − 1 yr − 1 during 2001 to 2017, with 177 ± 41.4 kg S ha − 1 yr − 1 input from the throughfall, 36.2 ± 7.20 kg S ha − 1 yr − 1 export by stream water (Fig. 2c), and limited S-gas emissions 19 . The chronic S retention in this catchment is attributed to the large S storage in the soil. Soil from 0 to 215 cm deep acts as a large S store of ~ 14,300 kg S ha − 1 (Fig. 3a), which is equivalent to the cumulative S retention in ~ 100 years according to the annual retention ux from 2001 to 2017. Corresponding to stronger S retention, the S reservoir is larger than temperate forest soil even with high S application 13 .
At the hillslope, the S store in the soil was dominated by adsorbed SO 4 2− and total reduced S with proportions of 35% and 34%, followed by organic S of 23% (Fig. 3a), it is similar to the S species composition in the entire catchment which mainly composed of hillslope topography (96% of the total area). The reduced S, as trivial species in previous studies, however, have extraordinary large contribution to the total S in the soil in this catchment. The proportion gradually increased with the soil depth, exceeding 70% at the deepest layer (Fig. 3a). In addition, the S store in reduced species had larger proportion in plots at groundwater discharge zone (49%) than those at hillslope (34%). These implies SO 4 2− reduction may be an important mechanism for S retention, especially in the deeper soil and the water logged zone before stream.
The large S retention, accounted to ~ 80% of the S deposition within TSP catchment, was widely found in subtropical ( Fig. 1). Subtropical China has one of the highest levels of atmospheric S deposition in the world 20 . The annual S deposition in subtropical forest (145 million ha, accounting for 61% of the forested land in China and 45% of the total subtropical forests in the world) was averaged at 37.0 kg S ha − 1 yr − 1 from 2005 to 2015 (Fig. 1b), which is used to represent the historical S deposition. According to the retention ratio, the total S retention was 4.29 Tg S yr − 1 , and the S store in soil was calculated as ~ 429 Tg in the entire subtropical forest in China. There may be 150, 146, and 98.7 Tg S stored in the subtropical forest soil as adsorbed SO 4 2− , reduced S, and organic S, consecutively, when on the basis of the S species composition in TSP catchment with typical soil across subtropical China.  Fig. 3b). This corresponded to the trend with high sulfate reduction rates (Trend A) rather than that with equilibrium isotope exchange (Trend B), indicating rapid sulfate reduction in this catchment. The slope shifts from low values in hillslope (0.36, R 2 = 0.94) to higher values in groundwater discharge zone (0.49, R 2 = 0.91), conforming to the stronger anaerobic condition and more reduced S store in the groundwater discharge zone.

Mechanisms for S retention quanti ed with modelling
A model was developed based upon mass and isotope balance estimated the co-occurrence of three Sretention processes in soils in TSP catchment, including SO 4 2− adsorption, reduction, and net incorporation into organic matter, accounting 46%, 30%, and 4% of the total deposited S, respectively ( Fig. 4). This was based on the assumption that the 34 S fractionation effect during the reduction ( 34 ε r ) is 16‰ (according to the Rayleigh distillation equation, Supplementary Text 4). The S uxes calculated by the model is sensitive to the 34 ε r , and the contributions of adsorption and reduction to total S deposition varied from 37-53% and 23 to 38%, respectively, with 34 ε r in the range of 12‰-20‰. The uncertainties of S uxes from incorporation, adsorption, and reduction were 20%, 49%, and 35% with synthesis consideration of the impact factors.
As the main retention mechanism, sulfate adsorption, having a total ux of ~ 81.2 kg S ha − 1 yr − 1 in this catchment, can explain the increasing adsorbed SO 4 2− store in the soil (359, 611, and 3,600 kg S ha − 1 for 0-50 cm soil in 2004 10 , 2011 21 , and 2017, Fig. 3a, respectively). Generally, the old and deeply weathered subtropical soils with low pH (3.7-4.1in TSP catchment) and large Fe/Al oxides have large adsorption capacity 4,6,9,22 . In addition, sulfate adsorption is concentration-related 4,6,8 , whereby high dissolved sulfate concentration in the soil (Fig. 2c)  adsorption was quanti ed using isotopic model, and further clarifying the contribution to S retention. The model results emphasized the importance of sulfate reduction on S retention within TSP catchment, combined with the evidence of S store and isotopic signals. The strong sulfate reduction in the subtropical zone appears to be promoted by su cient anaerobic conditions due to both the hot and humid climate. And the abundant Fe-oxides (or Fe 2+ ) in the soil 22 can x the sul de, and form iron sul de minerals, may subsequently convert to pyrite. In the biogeochemistry of S, subtropical soil might be an important region to capture the reactive S except the extensively studied salt marshes, peatlands, and marine sediments [23][24][25] .
It is common to treat the entire forest catchment as one soil type (well-drained forest soils) rather than along the whole soil pro le and hydrological gradients 13 . The variations in 3-dimensional retention uxes account for differences in soil-related depth and water ow path. They are explicitly demonstrated in this study, showing the importance of the deeper soil and the waterlogged zone in S retention, and where sulfate reduction performed prominently. The S ux decreased by 42.3 and 77.1 kg S ha − 1 yr − 1 in the upland 40 cm and deeper soil in aerated hillslope, respectively. This attributed to 20% and 44% of the total S deposition in the entire forested catchment of TSP (Fig. 4). The residual 16% of S retention occurred in the waterlogged groundwater discharge zone, although which only accounts for 4% of the total basin area (Fig. 4). The overlook of these regions fed the limited awareness of the role played by sulfate reduction on S retention. Potential consequences of large S retention S retention is of crucial importance for resisting cationic nutrient transport and aluminium mobilization [1][2][3]26 . Although China recived chronically elevated atmospheric S deposition with the increase of SO 2 from 6.8 in the 1980s to 34 Tg yr − 1 in 2006 27 , the robust S retention in subtropical China (~ 80% of total S deposition) alleviated the acidi cation in the past decades 11 . Relatively light surface water acidi cation in China was observed as a result, even under heavy deposition compared to that in temperate forests 28 .
However, the S retention in soil ultimately would act as a delayed risk of concern in environment and human health 29 . Potential hazard depends on the stability of the legacy S in response to the changes in environmental conditions. The consistency or discrepancy of contributions between calculated uxes and reservoirs indicated the difference in stability of S species. For instance, the annual ux of reduction was estimated to be 38% of the retention, using a dual isotopic mass balance model, it was close to the proportion of reduced species in soil. However, the proportion of adsorbed S store in the soil (35%) was signi cantly lower than the ratio of sulfate adsorption ux to total S retention (58%). This indicates that the reduced S reservoir is relatively stable, and the historical reduction was well accumulated in soil pro le. In contrast, the adsorption of sulfate is highly reversible, and the previously adsorbed S in soil might be released in some periods. Numerous studies have documented that desorption of adsorbed S would occur in light of continuous decline in atmospheric S deposition, and it reported to delay the recovery of soil and water acidi cation in America and Europe 1,3,9,30

Study site, sampling and chemical analyses
This study was conducted in Tieshanping (TSP), a 4.6 ha headwater catchment located in the center of the subtropical China (Fig. 1b). It has a typical and widespread soil type (Haplic Acrisol) and vegetation type (Pinus massoniana). A relatively steep northeast-facing hillslope and a hydrologically connected terraced groundwater discharge zone were selected to dominant landscape in this study (Fig. 2).  where Q is water ux, F i , F n , F a , F r , and F o represent annual S uxes of input (atmospheric deposition for the upper layer, and leaching from the rst layer for the deeper layer), net mineralization, adsorption, reduction, and output, respectively. where total S is dominated by organic forms, Fig. 3a). δ 34 S o denotes the δ 34 S of SO 4 2in the soil water. 34 ε r refers to the isotope effect (ε = light k/ heavy k -1, reported in ‰, with k being the rate constant) by reduction. The 34 ε r was assumed as 16‰, with a range of 12-20‰ in this study, according to the

Declarations
Data availability The data that support the nding of this study are available from the corresponding author (D.L.) upon request.