Global soil acidification impacts on belowground processes

With continuous nitrogen (N) enrichment and sulfur (S) deposition, soil acidification has accelerated and become a global environmental issue. However, a full understanding of the general pattern of ecosystem belowground processes in response to soil acidification due to the impacting factors remains elusive. We conducted a meta-analysis of soil acidification impacts on belowground functions using 304 observations from 49 independent studies, mainly including soil cations, soil nutrient, respiration, root and microbial biomass. Our results show that acid addition significantly reduced soil pH by 0.24 on average, with less pH decrease in forest than non-forest ecosystems. The response ratio of soil pH was positively correlated with site precipitation and temperature, but negatively with initial soil pH. Soil base cations (Ca2+, Mg2+, Na+) decreased while non-base cations (Al3+, Fe3+) increased with soil acidification. Soil respiration, fine root biomass, microbial biomass carbon and nitrogen were significantly reduced by 14.7%, 19.1%, 9.6% and 12.1%, respectively, under acid addition. These indicate that soil carbon processes are sensitive to soil acidification. Overall, our meta-analysis suggests a strong negative impact of soil acidification on belowground functions, with the potential to suppress soil carbon emission. It also arouses our attention to the toxic effects of soil ions on terrestrial ecosystems.


Introduction
Since the mid 20th century, acid rain has become a serious global environmental problem due to rapid industrial development (Blank 1985, Duan et al 2016. The main sources of acid and acidifying pollutants are sulfur dioxide (SO 2 ), nitrogen oxides (NO x ) and ammonia (NH 3 ) emitted from fossil fuel combustion and agricultural activities (Zhao et al 2009, Yang et al 2012. Though SO 2 and NO x emissions have been reduced in Europe and North America, they are increasing in many developing countries due to coal combustion (Gao et al 2018). Soil acidification is a natural process, which has been accelerated by increases in N and S deposition associated with human activities (Grieve 2001, Kunhikrishnan et al 2016. Human accelerated soil acidification alters biogeochemical cycles and impairs ecosystem function (Stevens et al 2010, Liang et al 2018. Therefore, understanding the general patterns of ecosystem processes with acid deposition across diverse environments will provide valuable knowledge for predicting future ecosystem dynamics under global change. To date, however, there has been no systematic global synthesis of acid deposition impact on ecosystem functions. Acid deposition has complex effects on ecosystems, especially for belowground processes. First, more H + input to soil along with SO 4 2and NO 3 induced by acid deposition may directly affect microbial activities (Kuperman and Edwards 1997). Second, H + from acid deposition will compete with base cations (e.g. K + , Mg 2+ , Ca 2+ ) for replacement, which increase base cation leaching out of soil. This further reduces soil acid buffering capacity (Driscoll et al 2003) and nutrient Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
availability (Likens et al 1996). Third, increased soil acidification with continuous acid deposition has the potential to mobilize and release free Al 3+ and Fe 3+ to soil solution. This accumulation of toxic elements in topsoil may eventually impair root growth and microbial activity (Godbold et al 1988, Kochian 1995, Poschenrieder et al 2008, which will consequently reduces soil respiration. The rate and form of acid deposition, soil type, environmental factor, and ecosystem type all may regulate the effects of acid deposition on soil processes. Normally, acid deposition rate should be a major factor to drive soil acidification (Vanhala et al 1996). Different acid forms, e.g. H 2 SO 4 and HNO 3 , may also contribute to the variable impacts of acid deposition due to their different adsorption mechanism. NO 3 is adsorbed only through electrostatic attraction, while SO 4 2can be specially adsorbed through ligand exchange, especially in variable charge soils (Curtin andSyers 1990, Guadalix andPardo 1991). This special adsorption may lead to a release of hydroxyl ions, which could neutralize a part of the acids and retard soil acidification to some extent. Furthermore, soil type is a significant contributor to regulating soil acidification response. It is expected that soils with different initial pH may go through different acidification buffering phases (Bowman et al 2008). Soil with a lower pH generally experiences greater acidweathering, which makes less sensitive to external acid input (De Vries et al 1989, Zhu 2017. High precipitation accelerates the leaching of soil cations and further aggravates acidification (Lapenis et al 2004, Ling et al 2007. Low temperature possibly depresses litter decomposition (Oulehle et al 2011, Liang et al 2013, leading to litter accumulation and then weakening soil acidification magnitude (Aerts 1997). The influence of these abiotic and biotic factors in combination finally causes different ecosystem response to acid deposition. It is a challenge but essential to quantify the influences of those factors on the soil acidification impacts across different experiments with different application rates and acid agents to soils with varying buffering capacities.
Here, we compiled a global dataset (304 observations) from 49 case studies and performed a metaanalysis to quantify belowground process dynamics in response to experimental acid addition. Specifically, we addressed the following questions: (1) How have various belowground processes respond to acid addition at global scale? (2) What are the main controlling factors for the responses of belowground processes?

Data compilation
Web of Science, Google Scholar and China National Knowledge Infrastructure were searched for peerreviewed publications on experiments dealing with acid deposition and ecosystem function. The searched keywords were: (acid deposition or S deposition or simulated acid rain) AND (soil cations (e.g. Na + , Mg 2+ , Ca 2+ ), soil nutrient (e.g. SOC, STN), soil respiration, fine root biomass, microbial biomass). The following criteria were employed to screen appropriate studies for analysis: (1) only acid addition experiments in the field were included, with the treatment duration lasting at least one growing season; (2) The control and acid addition treatments had to experience the same climate and soil condition; (3) examined variables were required to be clearly described by their means, sample sizes and standard deviation.
To acquire as many observations as possible, we gathered the data at each peak biomass stage of the growing season during all measurement years. If several studies with different vegetation types or environmental conditions (i.e. annual temperature or precipitation) were reported in an article, each study was considered to be independent. Table-form data were directly extracted, while graph-form data were obtained by the Engauge Digitizer software (Free Software Foundation, Inc., Boston, MA, USA). If climate variables could not be obtained from the papers, we used the latitude and longitude of each study to extract these data from a global database (http://worldclim. org/). Soil type data were acquired from the FAO database (http://fao.org/). Finally, a global dataset was established with 49 independent studies from 45 papers (figure S1 is available online at stacks.iop.org/ ERL/14/074003/mmedia). This dataset covered the area with latitude range from 23.15 to 69.75°N and elevation range from 10 to 1200 m. Mean annual temperature varied from −2°C to 21.4°C, and precipitation from 130 to 2400 mm. Ecosystems included forest, grassland and peatland, but we sorted them into two groups for analysis (forest and non-forest). This is due to the lack of data from grassland and peatland ecosystems. Experimental duration spanned 1 to 14 years.

Meta-analysis
As described in previous studies (Hedges et al 1999, Lu et al 2011, we employed meta-analysis techniques to evaluate the impacts of acid addition on ecosystem belowground processes. Effect sizes of acid treatment were calculated as equation (1): where X treatment and X control are the mean values in acid addition and control treatments, respectively.
Effect sizes and their subsequent inferences in meta-analysis may be influenced by how individual observations are weighted (Mueller et al 2012, Ma andChen 2016). According to previous studies (Wu et al 2011, Ma andChen 2016) and our analysis, the weighted method based on variance assigns extreme importance to individual effect sizes, with a result that overall effect is mostly determined by a small number of extreme observations. Thus, we calculated a weighting factor (w) based on the sample size in each experiment as follows ( b p e are coefficient, the random effect of 'study' and sampling error respectively. The possible autocorrelation among observations within each study was explicitly accounted by the random effect of study . Linear, power and quadratic functions were applied to examine the relationships between log(RR) of belowground processes (i.e. soil cations-K + , Na + , Mg 2+ , Ca 2+ , Zn 2+ , Mn 2+ , Al 3+ , Fe 3+ ; soil nutrient-SOC, DOC, STN, NO , 3 -NH 4 + , AP, soil C:N; soil respiration; fine root biomass; microbial biomass) and log(RR) of soil acidification (soil pH versus acid addition level), MAT or MAP. We selected the best bivariate relationships based on the Akaike Information Criterion (AIC). If the difference in AIC among multiple models was less than two, the simple model was selected. If the difference in AIC was larger than two, the model with lower AIC was chosen (Wagenmakers 2003). Furthermore, we found that the residuals of most models did not follow a normal distribution with the Shapiro-Wilk's test. Thus, we applied a nonparametric bootstrap analysis to estimate the effect sizes and their 95% confidence interval (CI) using the 'boot' package (Davison andHinkley 1997, Canty andRipley 2012). In addition, we analyzed the relationships of belowground processes with both acid addition level and soil pH change, and found that the results were consistent between two analyses. To include as much data in analysis as possible, we mainly presented the results with the soil pH analysis in our main text, but put those with the acid level analysis in the supplementary material.
All independent variables were scaled to ensure the comparability between the results. For ease of interpretation, we converted the results of log (RR) as a percentage of belowground process responses to acid treatment (i.e.100 10 1 ). All statistical analyses and figure plotting were performed in R version 3.4.3 (R Core Team 2017).

Effects of acid treatment on belowground processes
At global scale, acid addition significantly reduced soil pH by 0.24 on average (figures 1, 5). Similarly, acid addition caused a significant decrease in soil base cations, such as Na + , Mg 2+ and Ca 2+ . In contrast, acid treatment increased soil Al 3+ and Fe 3+ by 22.7% and 48.6%, respectively (figures 1, 5). Moreover, acid addition further reduced soil carbon processes significantly, with a lower decrease in MBC (9.6%) than fine root biomass (19.1%) and soil respiration (14.7%). For microbial community, bacterial biomass showed a significant decrease (16.4%), whereas fungal biomass was not sensitive (figures 1, 5). Acid addition mostly had no impact on soil nutrient, such as SOC, DOC, STN, NH 4 -N, NO 3 -N, available P or soil C: N.
Across the acid addition gradient, soil pH was significantly reduced when acid addition rate was higher than 5 kmol H + ha −1 yr −1 (figure 2). Below this level, soil pH did not show any significant response. Furthermore, we found similar relationships of belowground processes with acid addition level versus soil pH change. To be concise, here we mainly presented those results with the pH analysis. For soil cations, soil Na + , Mg 2+ and Ca 2+ reduced linearly with decreasing soil pH (figure 3). By contrast, soil Al 3+ and Fe 3+ showed a positive relationship with soil acidification (figure 3). With respect to soil nutrient, soil acidification significantly suppressed soil NO 3 -N and soil C: N, but promoted soil NH 4 -N and available P (figure 3). Soil respiration and microbial biomass carbon displayed a positive relationship with soil acidification ( figure 3). Moreover, soil acidification linearly reduced bacterial biomass, whereas enhanced fungal biomass ( figure 3).
For different forms of acid addition, our results demonstrated that most belowground processes showed similar responses to both addition of H 2 SO 4 +HNO 3 and H 2 SO 4 alone. Moreover, no significant difference of soil NO 3 occurred between these two acid forms. There was a positive response of soil NH 4 + caused by H 2 SO 4 addition, but no response under H 2 SO 4 +HNO 3 addition (figure S3).

Ecosystem type affecting acid addition effects
Between different ecosystems, forests had a lower reduction in soil pH (0.15) than non-forest ecosystems (0.96). In forests, acid addition posed a significant impact on most belowground processes, negatively affecting soil Mg 2+ , Ca 2+ , soil respiration, fine root biomass, MBC, MBN and bacterial biomass but positively influencing soil Fe 3+ , NH 4 -N, available P and fungal biomass (figure S4a). In non-forest ecosystems, acid addition only decreased soil Na + by 29.5% and increased soil Al 3+ by 87.1% (figure S4b). In addition, among different soil types, acrisols, kastanozem, luvisol and podzol showed a significant pH reduction, whereas anthrosol, calcisol and cambisol had no significant response (figure 2).

Environmental factors regulating acid addition effects
The response ratio of soil pH showed a negative relationship with initial soil pH ( figure 4(a)), but a positive relationship with site temperature and precipitation (figures 4(b), (c)). High temperature tended to lessen the response magnitudes of soil Al 3+ , NO 3 -N, NH 4 -N and available P to acid addition, but to enhance that of fine root biomass (figure S5b). More precipitation intensified the effect of acid treatment on soil Mg 2+ and fine root biomass, while it attenuated the impact on soil Al 3+ , NH 4 -N and fungal biomass (figure S5a).

Discussion
Soil itself is a buffer system for external H + input. When H + input exceeds the maximum of soil buffer capacity, it will cause soil acidification. By synthesizing the results from global acid addition experiments, we found that acid addition significantly reduced soil pH by 0.24 unit (figures 1, 5). A similar reduction in pH occurred, with a decrease of 0.26 pH unit, reported in a global synthesis of nitrogen addition experiments (Tian and Niu 2015), but it is lower than a decrease of 0.63 unit reported from Chinese northern grasslands over the last two decades (Yang et al 2012). Soil acidification became significant when the H + addition ) Notably, the data in the graph were presented by pH p H . treatment control rate was more than 5 kmol ha −1 yr −1 , which indicates a threshold level of acid deposition driving acidification (Liao and Jiang (2002) WRB 2015). This may lead to the larger decrease of soil pH in non-forests than forest ecosystems. Acid addition-induced soil acidification further altered soil nutrient availability. The increase of soil available P with soil pH reduction indicates that much H + input by acid addition mainly promotes the release of phosphate from Fe or Al binding compound (Barrow andShaw 1979, Barrow 2017). In addition, we found that soil NH 4 + increased with a decrease in soil pH, likely due to the inhibition of nitrification and plant N uptake ( In line with our expectation, soil acidification caused a significant reduction in fine root biomass, which is mainly due to the following reasons. First, acid deposition accelerated the leaching of soil base cations (Mg 2+ , Ca 2+ ), further reducing their availability (Vanhala et al 1996, Pennanen et al 1998, Chen et al 2013, Chen et al 2015. This may directly reduce plant uptake for these essential elements, resulting in plant nutrient deficiency and then limiting primary productivity (Kochian 1995, Van Den Berg et al 2005, Vanguelova et al 2007, Chen et al 2013. Second, once base cations have been depleted, soil may release and accumulate toxic ions like Al 3+ (Bowman et al 2008). In this study, we indeed found that acid addition led to a significant increase in Al 3+ and Fe 3+ . Increasing Al 3+ can reduce root nutrient absorption, and much absorption of Al 3+ will interfere with plant physiological processes and finally cause Al 3+ toxicity (Ulrich et al 1980). These together could explain the fact that root biomass was significantly reduced by acid deposition (Hahn and Marschner 1998, Li et al 2018). Acid deposition also induced a significant decrease in bacterial biomass, but had no effect on fungal biomass (figures 1, 5). As shown in the summary figure (figure 5), acid addition-induced soil acidification and the ion toxicity (Al 3+ and Fe 3+ ) were two key factors to affect soil microbes (Chen et al 2013). The decline in bacteria biomass and no change in fungal biomass may be attributed to the fact that fungi are less sensitive to acidification and more tolerant to H + and Al 3+ when compared to bacteria (Rousk et al 2009, Aliasgharzad et al 2010, Strickland and Rousk 2010. For instance, due to the difference in cell wall structure between fungi and bacteria (Myrold and Nason 1992), fungi can store excess H + in the vacuoles and then extrude it into the environment (Kuperman and Edwards 1997). For the Al 3+ toxicity under acid deposition, on the one hand, its high concentration may directly decrease microbial biomass due to its toxicity to microbial cells (Pina and Cervantes 1996). On the other hand, higher level of Al 3+ perhaps reduces plant substrate inputs to soil (e.g. dead root, root exudate), further decreasing the conversion efficiency of plant C into microbial biomass C (Pietri and Brookes 2008, Oulehle et al 2018).
Soil respiration, which consists of autotrophic respiration (R a ) and heterotrophic respiration (R h ), is one of the largest carbon effluxes in terrestrial ecosystems (Kuzyakov 2006, Luo andZhou 2006). The C release through R a or R h is associated with belowground C pools, such as root biomass, soil organic C (SOC) and microbial biomass C (MBC) (Zhou et al 2014). Several mechanisms for root and microbial processes may help explain the decline in soil respiration under soil acidification (figure 5). First, more H + likely reduces microbial physiology and biomass, and then depresses soil heterotrophic respiration (Riggs and Hobbie 2016). Second, the nutrient limitation of microbial growth due to soil base cation loss (e.g. Mg 2+ , Ca 2+ ) (Bowman et al 2008, Oulehle et al 2018) and other ion (e.g. H + , Al 3+ ) toxicity (Tian and Niu 2015) tend to reduce soil heterotrophic respiration. Moreover, we also found a negative effect of acid addition on fine root biomass, which should cause a decrease in autotrophic respiration and thus soil respiration (Davidson et al 2006, Liang et al 2013. Acid deposition impacts on belowground processes varied with environmental factors. Soil pH decreased Figure 3. Linear slopes in the relationship between the response ratio of soil pH and soil cations (K + , Na + , Mg 2+ , Ca 2+ , Zn 2+ , Mn 2+ , Al 3+ , Fe 3+ ), soil nutrient (SOC-soil organic C, DOC-dissolved organic C, STN-soil total N, Soil NO 3 -NO 3 -N, Soil NH 4 -NH 4 -N, APavailable P, soil C:N), soil respiration, fine root biomass or microbes (MBC-microbial biomass C, MBN-microbial biomass N, Funfungal biomass, Bac-bacterial biomass). If the 95% CI does not overlap the slope=0, it means a significant response in soil belowground processes caused by acidification. more at sites with high initial pH ( figure 4(a)), which is likely because soils have different acid buffering capacities depending on the initial pH. It is generally accepted that the exchange capacity of soil cation is: trivalent cations (Fe 3+ >Al 3+ )>divalent cations (Mn 2+ > Zn 2+ >Ca 2+ >Mg 2+ )>monovalent cations (K + > Na + ). High valence cations will be replaced to buffer soil acidification only when low valence cations are mostly depleted. Based on the charge equivalent principle, soil with high initial pH is more vulnerable to be acidified. Contrary to our expectation, soil pH reduced less in sites with higher precipitation (figure 4(c)). Although more precipitation is expected to promote soil acidification by accelerating cation leaching, sites with high precipitation had a low initial soil pH in our study, which were less vulnerable to be acidified ( figure 4(c)). In a word, our findings emphasize the interactive effects of acid deposition and environmental factors on soil acidification.
Overall, this study has important implications for soil biogeochemical cycles. We found that acid addition significantly reduced the exchangeable base cations of Na + , Mg 2+ , Ca 2+ in soils. Similar to the result of N-induced soil acidification (Tian and Niu 2015), acid deposition induced soil acidification process goes through different buffering stages. The decline in soil pH is accompanied by a depletion of soil base cations. Once base cations have been depleted, soil reaches toxic levels because of Al 3+ release. A significant increase in soil free Al 3+ has been already detected under acid deposition, suggesting that soil in terrestrial ecosystem has begun to enter Al 3+ buffering stage. All these alert our attention to the danger of the coming soil acidification and its buffering stages of toxic Al 3+ and Fe 3+ .

Conclusion
Our synthesis revealed that belowground processes were sensitive to acid deposition at global scale and across different ecosystems. Global soil acidification is currently in a transition stage from base cation (Ca 2+ , Mg 2+ , K + ) to non-base cation buffering (Al 3+ , Fe 3+ ). This calls our attention to the toxic effects of soil ions on terrestrial ecosystems. Moreover, acid deposition further caused a decline in microbial biomass, fine root biomass and soil respiration, suggesting that the inhibition of soil carbon emission will substantially change soil carbon balance and its feedback to climate change. However, it is difficult to predict the magnitude of soil acidification with acid deposition and its impacts on belowground processes, mainly due to the complex relationships of acid effects with diverse environmental factors (soil properties and climate). Overall, this meta-analysis provides the first global viewpoint on linking belowground processes with soil acidification under acid deposition.