Multi-elemental stoichiometric ratios of atmospheric wet deposition in Chinese terrestrial ecosystems.

Intense human activities have significantly altered the concentrations of atmospheric components that enter ecosystems through wet and dry deposition, thereby affecting elemental cycles. However, atmospheric wet deposition multi-elemental stoichiometric ratios are poorly understood, hindering systematic exploration of atmospheric deposition effects on ecosystems. Monthly precipitation concentrations of six elements-nitrogen (N), phosphorus (P), sulfur (S), potassium (K), calcium (Ca), and magnesium (Mg)-were measured from 2013 to 2021 by the China Wet Deposition Observation Network (ChinaWD). The multi-elemental stoichiometric ratio of atmospheric wet deposition in Chinese terrestrial ecosystems was N: K: Ca: Mg: S: P = 31: 11: 67: 5.5: 28: 1, and there were differences between vegetation zones. Wet deposition N: S and N: Ca ratios exhibited initially increasing then decreasing inter-annual trends, whereas N: P ratios did not exhibit significant trends, with strong interannual variability. Wet deposition of multi-elements was significantly spatially negatively correlated with soil nutrient elements content (except for N), which indicates that wet deposition could facilitate soil nutrient replenishment, especially for nutrient-poor areas. Wet N deposition and N: P ratios were spatially negatively correlated with ecosystem and soil P densities. Meanwhile, wet deposition N: P ratios were all higher than those of ecosystem components (vegetation, soil, litter, and microorganisms) in different vegetation zones. High input of N deposition may reinforce P limitations in part of the ecosystem. The findings of this study establish a foundation for designing multi-elemental control experiments and exploring the ecological effects of atmospheric deposition.

Manipulating Identical Filter Redundancy for Efficient Pruning on Deep and Complicated CNN.

The existence of redundancy in convolutional neural networks (CNNs) enables us to remove some filters/channels with acceptable performance drops. However, the training objective of CNNs usually tends to minimize an accuracy-related loss function without any attention paid to the redundancy, making the redundancy distribute randomly on all the filters, such that removing any of them may trigger information loss and accuracy drop, necessitating a fine-tuning step for recovery. In this article, we propose to manipulate the redundancy during training to facilitate network pruning. To this end, we propose a novel centripetal SGD (C-SGD) to make some filters identical, resulting in ideal redundancy patterns, as such filters become purely redundant due to their duplicates, hence removing them does not harm the network. As shown on CIFAR and ImageNet, C-SGD delivers better performance because the redundancy is better organized, compared to the existing methods. The efficiency also characterizes C-SGD because it is as fast as regular SGD, requires no fine-tuning, and can be conducted simultaneously on all the layers even in very deep CNNs. Besides, C-SGD can improve the accuracy of CNNs by first training a model with the same architecture but wider layers and then squeezing it into the original width.

Atmospheric wet organic nitrogen deposition in China: Insights from the national observation network.

Organic nitrogen (N) is an important component of atmospheric reactive N deposition, and its bioavailability is almost as important as that of inorganic N. Currently, there are limited reports of national observations of organic N deposition; most stations are concentrated in rural and urban areas, with even fewer long-term observations of natural ecosystems in remote areas. Based on the China Wet Deposition Observation Network, this study regularly collected monthly wet deposition samples from 43 typical ecosystems from 2013 to 2021 and measured related N concentrations. The aim was to provide a more comprehensive assessment of the multi-component characteristics of atmospheric wet N deposition and reveal the influencing factors and potential sources of wet dissolved organic N (DON) deposition. The results showed that atmospheric wet deposition fluxes of NO3-, NH4+, DON and dissolved total N (DTN) were 4.68, 5.25, 4.32, and 13.05 kg N ha-1 yr-1, respectively, and that DON accounted for 30 % of DTN deposition (potentially up to 50 % in remote areas). Wet DON deposition was related to anthropogenic emissions (agriculture, biomass burning, and traffic), natural emissions (volatile organic compound emissions from vegetation), and precipitation processes. The wet DON deposition flux was higher in South, Central, and Southwest China, with more precipitation and intensive agricultural activities or more vegetation cover, and lower in Northwest China and Inner Mongolia, with less precipitation and human activities or vegetation cover. DON was the main contributor to DTN deposition in remote areas and was possibly related to natural emissions. In rural and urban areas, DON may have been more influenced by agricultural activities and anthropogenic emissions. This study quantified the long-term spatiotemporal patterns of wet N deposition and provides a reference for future N addition experiments and N cycle studies. Further consideration of DON deposition is required, especially in the context of anthropogenic control of NO2 and NH3.

[Ecosystem principles and main issues in regional ecological restoration and environmental governance in China]. / 中国区域生态恢复和环境治理的生态系统原理及若干学术问题.

Modern ecology is an analytical method and philosophical concept to solve major resource, environmental and ecological issues encountered during global sustainable development. In the long-term development processes, ecology constantly absorbed and integrated knowledge from related disciplines, forms a system of modern ecology and ecosystem science that closely related to climate system, biological system and socio-economic system, and raises ecosystem principles that directly support the practices of regional ecological restoration and environmental governance. The national needs in the new stage have given ecology a new mission. It is urgently needed to summarize and condense the principles of macro-ecosystems and apply them to regional ecological restoration and environmental governance with the aim to promote the high-quality development of society and economy. Against the background of the multiple severe challenges faced by global sustainable development, we comprehensively elaborated the logics and scientific mission of ecosystem science, organized the principle system of ecosystem science related to ecological restoration and environmental governance, and discussed major academic problems in regional ecological restoration and environmental governance of China. Finally, we emphasized that China has several regional macro-ecosystems of global significances. Conducting theoretical and practical research on macro-ecosystem is not only an urgent need for the construction of ecological civilization, but also the forefront of ecosystem science research, which is expected to make new contributions to theory development of ecology, and global ecological and environmental governance.

Impacts of nitrogen fertilizer type and application rate on soil acidification rate under a wheat-maize double cropping system.

Nitrogen (N) fertilizer-induced soil acidification in Chinese croplands is well-known, but insight in the impacts of different N fertilizer management approaches (fertilizer type and rate) on soil acidification rates is very limited. Here, we conducted a field experiment on a moderate acid soil to quantify soil acidification rates in response to N fertilization by different fertilizer types and N rates through monitoring the fate of elements (mainly nutrients) related to H+ production and consumption. Two N fertilizer types (urea and NH4Cl) and three N rates (control, optimized and conventional, 0/120/240 kg N ha-1 for wheat, 0/160/320 kg N ha-1 for maize) were included. Nitrogen addition led to an average H+ production of 4.0, 8.7, 11.4, 29.7 and 52.6 keq ha-1 yr-1, respectively, for the control, optimized urea, conventional urea, optimized NH4Cl and conventional NH4Cl plots. This was accompanied with a decline in soil base saturation of 1-10% and in soil pH of 0.1-0.7 units in the topsoil (0-20 cm). Removal of base cations by crop harvesting and N transformations contributed ~70% and ~20% to the H+ production in the urea treated plots, being ~20% and ~75% in the NH4Cl treated plots, respectively. The large NH4+ input via fertilization in the NH4Cl treated plots strongly enhanced the H+ production induced by N transformations. The low contribution of N transformations to the H+ production in the urea treated plots was due to the limited NO3- leaching, induced by the high N losses to air caused by denitrification. Increased N addition by urea, however, strongly increased H+ production by enhanced plant uptake of base cations, mainly due to a large potassium uptake in straw. Our results highlight the important role of optimizing fertilizer form and N rate as well as straw return to the field in alleviating soil acidification.

Cropland acidification increases risk of yield losses and food insecurity in China.

Distinct cropland acidification has been reported in China due to nitrogen (N) fertilizer overuse. However, the impacts on food production and thereby on food security are largely unknown. Yield losses in the period 1980-2050 were therefore assessed by simulating soil pH changes combined with derived pH-yield relationships for wheat, maize and rice. If the N fertilizer input continues to increase at 1% annually, the predicted average soil pH decline is about one unit and relative yield losses are expected to increase from approximately 4%-24% during 2010-2050. If the N fertilizer increase stops in 2020 (N2020), the expected losses are approximately 16% in 2050, which is comparable to a scenario of 100% crop residue return (100%RR). However, if 30% of the N fertilizer is replaced by manure N (30%MR), the losses reduce to near 5% in 2050. Soil acidification was predicted to reverse and expected losses are only 2.5% in 2050 in a combined scenario of N2020, 100%RR and 30%MR. Our results illustrate the potential food insecurity induced by cropland acidification and address the necessity of mitigation.

Impact of 13-years of nitrogen addition on nitrous oxide and methane fluxes and ecosystem respiration in a temperate grassland.

Nitrogen (N) fertilizer application and atmospheric N deposition will profoundly affect greenhouse gas (GHGs) emissions, especially nitrous oxide (N2O) and methane (CH4) fluxes and ecosystem respiration (Re, i.e. CO2 emissions). However, the impacts of long-term N inputs and the often associated N-induced soil acidification on GHG fluxes in arid and semi-arid ecosystems, especially temperate grasslands, are still uncertain. An in situ experiment was conducted to investigate the effect of long-term (13-years) N addition on N2O and CH4 fluxes and Re from a temperate grassland in Inner Mongolia, northeast China, from April 2017 to October 2018. Soil pH values in the 0-5 cm layer receiving 120 (N120) and 240 (N240) kg N ha-1 decreased from 7.12 to 4.37 and 4.18, respectively, after 13 years of N inputs. Soil CH4 uptake was significantly reduced, but N2O emission was enhanced significantly by N addition. However, N addition had no impact on Re. Structural Equation Modeling indicated that soil NH4+-N content was the dominant control of N2O emissions, but with less effect of the decreasing pH. In contrast, CH4 uptake was generally controlled by soil pH and NO3--N content, and Re by forb biomass. The measured changes in N2O and CH4 fluxes and Re from temperate grassland will have a profoundly impact on climate change.

Responses of soil organic carbon turnover to nitrogen deposition are associated with nitrogen input rates: Derived from soil 14C evidences.

Elevated atmospheric nitrogen (N) deposition has exerted profound influences on ecosystems. Understanding the effects of N deposition on the dynamics of soil organic carbon (SOC) is important in the studies of global carbon cycle. Although many studies have examined the effects of N deposition on SOC turnover using N addition experiments, the effects were reported to be different across studies. Thus, we lack a predictive understanding of how SOC turnover respond to atmospheric N deposition. The inconsistent results could be associated with ecosystem types and N addition rates. This study mainly wants to confirm the argument that the response of SOC turnover to N deposition is related with N input rates. We conducted a field experiment with multiple N addition levels (0, 3, 6, 12, and 24 g N m-2·yr-1) in Inner Mongolia Grassland, China. To better reveal the responses of SOC turnover to N enrichment, this study measured the soil 14C contents, because it can indicate SOC turnover directly. Compared with the control treatment (0 g N m-2·yr-1), N addition inhibits SOC turnover at the addition rate of 3 g N m-2·yr-1, whereas SOC turnover is not affected when N addition rate was 6, 12, and 24 g N m-2·yr-1. Our results suggest that N input rates affect the responses of SOC turnover to N enrichment. Thus, this study can confirm the argument mentioned above. Based on this study, it should be considered in the climate prediction model that varied atmospheric N deposition levels across regions may have different impacts on local SOC turnover. In addition, we also carried out a soil incubation to compare between the results obtained in incubation and that in 14C measurements. Two results are found to be inconsistent with each other. This indicates that soil respiration from incubation experiments could not comprehensively assess the effects of N deposition on SOC turnover.

Cumulative and partially recoverable impacts of nitrogen addition on a temperate steppe.

Atmospheric nitrogen (N) deposition has been shown to decrease biodiversity and change nutrient cycles in terrestrial ecosystems. However, our understanding of ecological responses to chronic N addition and ecological recovery of grassland from N enrichment is limited. Here we present evidence from an 11-year grassland experiment with a range of N addition rates (0, 30, 60, 120, 240, and 480 kg N·ha-1 ·yr-1 ) in Inner Mongolia, China. Chronic N addition led to a reduction in species richness, Shannon diversity index, and soil pH and an increase in aboveground biomass, foliar N, and soil mineral N. High N addition rates (240 and 480 kg N·ha-1 ·yr-1 ) showed significant effects in the first and second years, which stabilized over time. Nitrogen addition at low rates (30 and 60 kg N·ha-1 ·yr-1 ) took longer (e.g., three years or more) to achieve significant effects. The negative impacts of high N addition (480 kg N·ha-1 ·yr-1 ) were reduced and species richness, Shannon diversity index, and soil pH showed a limited but rapid recovery with the cessation of N addition. Our findings suggest serious and cumulative impacts of N addition on plant and soil communities but the potential for partial system recovery over time if N inputs decline or cease.

Enhanced acidification in Chinese croplands as derived from element budgets in the period 1980-2010.

Significant soil pH decrease has been reported in Chinese croplands in response to enhanced chemical fertilizer application and crop yields. However, the temporal and spatial variation of soil acidification rates across Chinese croplands is still unclear. We therefore assessed trends in soil acidification rates across provincial China for the period 1980-2010 by calculating inputs-outputs of major cations and anions in cropland systems. Nitrogen (N) induced proton production increased from 4.7keqH+/ha/yr in 1980 to a peak of 11.0keqH+/ha/yr in 1996 and remained nearly constant after 2000 at a rate of approximately 8.6keqH+/ha/yr. The proton production induced by crop removal increased from 1.2 to 2.3keqH+/ha/yr. The total proton production thus increased from 5.9 to 10.9keqH+/ha/yr in the 30years. As a result, the actual acidification rate, reflected by (base) cation losses, accelerated from 2.3 to 6.2keqH+/ha/yr and the potential acidification rate, reflected by phosphorus accumulation accelerated from 0.2 to 1.3keqH+/ha/yr. The national averaged total acidification rates were thus estimated to increase from 2.6 to 7.6keqH+/ha/yr in the past 30years. The highest soil acidification rate occurred in the Jiangsu Province with a rate of 17.9keqH+/ha/yr, which was due to both high N application rates and high base cation removals by crops and crop residues. The combination of elevated N inputs and decreased N use efficiency (NUE) in response to those N inputs, thus enhancing the nitrate discharge, were the main reasons for the accelerated acidification in Chinese croplands. Considering the expected growth of food demand in the future, and the linkage between grain production and fertilizer N consumption, a further acceleration of soil acidification can thus be expected, unless the N inputs is reduced and/or the NUE is increased substantially.

Modeling soil acidification in typical Chinese cropping systems.

We applied the adapted model VSD+ to assess cropland acidification in four typical Chinese cropping systems (single Maize (M), Wheat-Maize (W-M), Wheat-Rice (W-R) and Rice-Rice (R-R)) on dominant soils in view of its potential threat to grain production. By considering the current situation and possible improvements in field (nutrient) management, five scenarios were designed: i) Business as usual (BAU); ii) No nitrogen (N) fertilizer increase after 2020 (N2020); iii) 100% crop residues return to cropland (100%RR); iv) manure N was applied to replace 30% of chemical N fertilizer (30%MR) and v) Integrated N2020 and 30%MR with 100%RR after 2020 (INMR). Results illustrated that in the investigated calcareous soils, the calcium carbonate buffering system can keep pH at a high level for >150years. In non-calcareous soils, a moderate to strong decline in both base saturation and pH is predicted for the coming decades in the BAU scenario. We predicted that approximately 13% of the considered croplands may suffer from Al toxicity in 2050 following the BAU scenario. The N2020, 100%RR and 30%MR scenarios reduce the acidification rates by 16%, 47% and 99%, respectively, compared to BAU. INMR is the most effective strategy on reducing acidification and leads to no Al toxicity in croplands in 2050. Both improved manure and field management are required to manage acidification in wheat-maize cropping system.

Model-Based Analysis of the Long-Term Effects of Fertilization Management on Cropland Soil Acidification.

Agricultural soil acidification in China is known to be caused by the over-application of nitrogen (N) fertilizers, but the long-term impacts of different fertilization practices on intensive cropland soil acidification are largely unknown. Here, we further developed the soil acidification model VSD+ for intensive agricultural systems and validated it against observed data from three long-term fertilization experiments in China. The model simulated well the changes in soil pH and base saturation over the last 20 years. The validated model was adopted to quantify the contribution of N and base cation (BC) fluxes to soil acidification. The net NO3- leaching and NO4+input accounted for 80% of the proton production under N application, whereas one-third of acid was produced by BC uptake when N was not applied. The simulated long-term (1990-2050) effects of different fertilizations on soil acidification showed that balanced N application combined with manure application avoids reduction of both soil pH and base saturation, while application of calcium nitrate and liming increases these two soil properties. Reducing NH4+ input and NO3- leaching by optimizing N management and increasing BC inputs by manure application thus already seem to be effective approaches to mitigating soil acidification in intensive cropland systems.