Enhanced atmospheric phosphorus deposition in Asia and Europe in the past two decades

Yuepeng Pn , Bowen Liu , Jing Co , Jin Liu , Shili Tin , Enzi Du

a State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry (LAPC), Institute of Atmospheric Physics, Chinese Academy of Sciences,Beijing, China

b University of Chinese Academy of Sciences, Beijing, China

c State Key Laboratory of Earth Surface Processes and Resource Ecology, Faculty of Geographical Science, Beijing Normal University, Beijing, China

d School of Natural Resources, Faculty of Geographical Science, Beijing Normal University, Beijing, China

Keywords:Atmospheric phosphorus deposition Wet deposition Dry deposition Bulk deposition Air pollution

A B S T R A C T There is increasing interest in understanding atmospheric phosphorus (P) deposition and its impacts on plant productivity and carbon sinks in ecosystems. However, the global pattern of P deposition remains poorly understood,primarily due to the sparseness of data in Asia. In this study, the authors compiled 396 published observations of atmospheric P deposition from 1959 to 2020 on the global scale. The results gave a geometric mean bulk P deposition value of 0.32 kg ha ? 1 yr ? 1 , or a global P budget of 4.4 Tg yr ? 1 . Compared with the period 1959—2000,the authors found an elevated P deposition in Europe and Asia during 2001—2020, likely due to the increasing agricultural emissions and fossil fuel combustion—related sources in addition to dust emissions. The findings highlight the need to quantify the impacts of elevated P deposition from anthropogenic emissions on long-term ecosystem development in the context of carbon neutrality and clean-air actions.

1. Introduction

The supply of phosphorus (P), a principal macronutrient, to ecosystems is a limiting factor for plant productivity, and hence governs nitrogen fixation and carbon storage in terrestrial and aquatic ecosystems( Vitousek et al., 2010 ). A recent assessment indicated that 43% of the natural terrestrial land area is significantly limited by P, whereas 18%is relatively nitrogen limited, changing the conventional view that nitrogen limitation is more important than P limitation ( Du et al., 2020 ).This was also evidenced by a global meta-analysis by Hou et al. (2020) ,which indicated that 46.2% of 652 P-addition field experiments demonstrated a significant P limitation on plant production. Globally, P addition increases plant production by 34.9% in natural terrestrial ecosystems, which is 7.0%—15.9% higher than previously estimated( Hou et al. 2020 ). These studies highlight a more pervasive P limitation on plant production than conventionally believed.

In ecosystems, the natural sources of P are believed to be mineral weathering and erosion processes ( Gardner, 1990 ). With the application of P fertilizer on farmlands and the combustion of fuels in developed regions ( Wang et al., 2015 ), however, substantial anthropogenic P has been emitted to the atmosphere and transported to surrounding areas( Du et al., 2016 ). Atmospheric P can be removed by wet and dry deposition, which serves as a significant and even dominant source in some ecosystems ( Tipping et al., 2014 ). A recent source investigation using phosphate oxygen isotopes found that coal burning contributed 24%—39% to wet P deposition in Daya Bay ( Wu et al., 2021 ). Aside from dust sources ( Gross et al., 2016 ), biomass burning in Africa was also identified as a substantial source of soluble P in receiving ecosystems, e.g.,the Amazon, tropical Atlantic Ocean and Southern Ocean ( Barkley et al.,2019 ).

Since a sustained P input partially determines whether plant productivity is ultimately limited by P or nitrogen in ecosystems ( Cleveland et al., 2013 ; Du et al., 2020 ), quantifying atmospheric P deposition to land and water is important for understanding the dynamics of various ecosystems ( Tipping et al., 2014 ). Based on observational experiments at 49 sites in marine and remote continental areas in the 1970s, the annual transfer of P flux to and from the atmosphere was estimated to be 4.56 Tg yron a global scale ( Graham and Duce, 1979 ). Of this flux,3.21 Tg was deposited to land, which is equivalent to a deposition of 0.22 kg hayr. In a recent meta-analysis of 253 sites covering the period 1954—2012, Tipping et al. (2014) estimated a global P budget of 3.7 Tg yr, with a P deposition to land of 0.27 kg hayr.

With a chemical transport model, however,Mahowald et al. (2008) simulated a global atmospheric P deposition of only 1.39 Tg yr. This value is less than 50% of the global deposition pool interpolated from measurements, highlighting a poor agreement between modeled and observed atmospheric P deposition.While the gap can be filled by an increase in combustion-related emissions in the model ( Wang et al., 2015 ), uncertainties still exist in the previous observational compilations. As Tipping et al. (2014) recognized, although their data covered most land surfaces of the globe,they were biased towards sites in North America and Europe (

n

= 209,or 83% of the measurement locations). For Asia, however, their database only compiled eight locations, which was a large omission of observations, especially in China.

In view of this, we carried out a more comprehensive collation of published data on atmospheric P deposition, with a focus on observations in China. Based on investigations of the forms (inorganic vs organic) and pathways (wet vs dry) of atmospheric P deposition, we extended the global database for total P deposition to cover the period 1954—2020. With this new compilation, we investigated the long-term trend and global pattern of atmospheric P deposition. The dataset can be used to validate models related to atmospheric P deposition, thus providing a baseline for evaluations of the nutrient status of ecosystems.

2. Methods

2.1. Data collection

Based on the recent database compiled by Tipping et al. (2014) that covered the period 1954—2012 for 253 sites, we further scoured the peerreviewed literature for studies that observed atmospheric wet and/or dry deposition of in/organic P, especially since 2013, and ultimately added data for an additional 143 sites. In particular, we included data compilations for the period 1981—2020 in China (

n

= 69). The new compilation improves upon the previous one in that there are more sites (Fig. S1), with a total of 396 locations, 23.4% of which covers Asia —larger than the 3.2% in Tipping et al. (2014) . We provide our data compilation with information on the reported atmospheric P deposition (mass/area/time; Fig. S2), geographic locations ( Fig. 1 ), sampling methods (Fig. S3), and analytical methods (Fig. S4).

2.2. Data summary

The majority of the published data on atmospheric P deposition were on bulk deposition (86.1%; Fig. S3), which is the focus of this synthesis.There were considerable methodological differences among the studies in terms of how P was measured. Methods to quantify bulk TP (on an unfiltered sample) or FTP (on a filtered sample) were generally based on the ammonium molybdate colorimetric method, which accounted for 70.4% of the publications (Fig. S4).

Studies also varied in terms of whether wet, dry, or both were quantified, and the inorganic and organic species in depositions over which data were reported. To gain a holistic picture of the total deposition (wet and dry deposition of inorganic and organic P), the extrapolation of the reported results (see below) can only be regarded as an approximation of bulk deposition.

2.3. Data processing

Based on the studies that investigated dry and wet deposition of P concurrently at a single site, we obtained the percentage of P deposition as dry deposition ranging from 15.0% to 85.3%, with a mean of 54.5%(

n

= 16) (Fig. S5). To estimate the total deposition, we assumed that dry deposition was equal to that of wet deposition when the wet or dry fraction was absent. In addition, we summarize the P deposition that was conducted separately as wet (

n

= 34, Table S1) or dry deposition(

n

= 5, Table S2) for reference.Furthermore, not all observations determined P forms concurrently.Based on the available dataset (Table S3), we obtained the organic and inorganic P deposition as 0.39 and 0.29 kg hayr, respectively. The proportion of organic P in the total P varied from 18.8% to 83.3%, with an average of 52.5% (

n

= 28). Thus, the organic to inorganic P ratio was very close to 1:1, which can be roughly used to estimate the total P deposition.

3. Results and discussion

3.1. Global P deposition

The full dataset of annual observed total deposition of atmospheric P is plotted in Fig. S2. Taking all the data together, the global P deposition covering the years 1954—2020 was 0.66 ± 1.32 kg hayr(

n

= 396).This overall average value decreased to 0.58 ± 0.72 kg hayr( Table 1 ) if the calculation omitted two outliers from South America(14.7 kg hayr) and Asia (17.4 kg hayr). Since a simple average is probably not the best value to represent a summary statistic of global P deposition, the geometric mean and median values were also calculated, as 0.32 and 0.32 kg hayr(

n

= 394), to avoid bias towards larger values ( Table 1 ). Except for the median of 0.63 kg hayr(

n

= 20) reported by Newman (1995) , our results are comparable to the median value of 0.33 kg hayr(

n

= 33) collated by Gibson et al. (2010) , but slightly higher than that of 0.25 kg hayr(

n

= 45) reported by Tsukuda et al. (2006) , 0.28 kg hayr(

n

= 86)by Mahowald et al. (2008) , 0.22 kg hayr(

n

= 49) by Graham and Duce (1979) , and 0.26 kg hayr(

n

= 253) by Tipping et al. (2014) .In their landmark global meta-analysis study,Tipping et al. (2014) estimated a geometric mean of P deposition as 0.28 kg ha1 yr1 (

n

= 253), which is slightly lower than our estimate. This discrepancy could be well understood when the geometric mean values were grouped by continental region in each study.As shown in Table 1 , our geometric mean values of atmospheric P deposition in Africa, Europe, America, and Oceania were close to the results reported by Tipping et al. (2014) . In Asia, however, our geometric mean value was twice that reported by Tipping et al. (2014) ,resulting in a higher global geometric mean value in this study.

Table 1 Summary of the total atmospheric P deposition (kg ha ? 1 yr ? 1 ).

We further examined the long-term trends of global P deposition in our new dataset ( Fig. 2 ). The global P deposition varied insignificantly during the period 1959—2000, with an overall mean of 0.45 ± 0.42 kg ha1 yr(

n

= 133). This value reached up to 0.65 ± 0.83 kg hayr(

n

= 261) during the period 2001—2020. Upon close inspection of Fig. 2 , these elevated values were mostly identified in Europe and Asia in the most recent two decades. This trend was not reported by Tipping et al. (2014) , due to their data coverage being limited to 2012.

Fig. 1. Locations with measurements of atmospheric P deposition compiled in this study (black dots), with the global mapping of nitrogen and P limitation (colors)indicated by Du et al. (2020) .

Fig. 2. Global atmospheric P deposition from 1959 to 2020 ( n = 394, with two outliers removed).

Fig. 3. Global distribution of total atmospheric P deposition from 1959 to 2020 ( n = 394, with two outliers removed).

3.2. Sources of P deposition in Asia and Europe

The elevated deposition of atmospheric P in Europe may be explained by the increased emissions from agricultural soils and dust. In fact, an increasing trend for the dry deposition of P was well recorded in the Mediterranean region due to the prevalence of Saharan dust( de Fommervault et al., 2015 ). For example, the atmospheric dry deposition of P in the Cap Ferrat, eastern Mediterranean, was 0.017—0.079 kg hayrin 1998 ( Migon et al., 2001 ), but it reached 0.20 kg hayrin 2011 ( de Fommervault et al., 2015 ). In Finokalia, western Mediterranean, the dry P deposition also increased eightfold from 0.04 kg hayrin 2001 ( Markaki et al., 2010 ) to 0.32 kg hayrin 2013 ( Violaki et al., 2018 ). In addition, agricultural emissions from livestock farming may also contribute to high P deposition in an area of Germany (

n

= 11), as noted by Tipping et al. (2014) .

Although the frequency of sand dust weather decreased in Asia in recent years, dust emissions may also account for the high P deposition.For example, atmospheric P deposition in Jiaozhou Bay was strongly affected by air masses passing Northwest China ( Xing et al., 2018 ). In addition, P originating from animal husbandry can supply substantial P input to Lake Qinghai, with a wet P deposition of 1.86 kg hayr( Zhang et al., 2019 ). Besides, industrial emissions were found to be an important source of atmospheric P deposition ( Chiwa, 2020 ). In Ashiu experimental forest, 59%—89% of total soluble P came from the burning of fossil fuels ( Tsukuda et al., 2005 ). In Daya Bay, coal-burning also contributed to 39% of wet P deposition ( Wu et al., 2021 ). All this evidence suggests that anthropogenic emissions have contributed to atmospheric P deposition in Asia.

Globally, the mineral dust input to the atmosphere has been estimated to be 0.93 Tg P yr, or 27% of total P emissions ( Wang et al.,2015 ). More specifically, the dust emissions were lower than the combustion-related emissions of 1.8 Tg P yr, which represented over 50% of global atmospheric sources of P ( Wang et al., 2015 ). This contribution is far higher than the previously reported value of 5%( Mahowald et al., 2008 ). Therefore, in the future, more attention should be paid to P emissions from human-related sources, in addition to natural origins.

3.3. Global P budget and implications for ecosystems

Global P deposition amounts based on the geometric mean value of 0.32 kg hayr( Table 1 ) and a land area of 117 ×109 km(omitting high latitudes) yielded a total P deposition of 4.4 Tg yrto land. This is slightly higher than the earlier estimate of 3.5 Tg yr( Tipping et al., 2014 ), likely because of the omission of the elevated deposition amounts in Asia. When compared with simulated values, our observational estimate is much higher: for example, compared with the simulated value of 1.4 Tg yrreported by Mahowald et al. (2008) .While Mahowald et al. (2008) argued that coarse particles were not considered in their global modeling, Wang et al. (2015) suggested that combustion-related emissions from fossil fuels and biofuels are larger than previously thought. After accounting for P emissions to the atmosphere from all combustion sources, Wang et al. (2015) estimated total global emissions of atmospheric P as 3.5 Tg yr1 , of which about 77%(2.7 Tg yr) was deposited over land. One reason that our estimate is greater than this value of 2.7 Tg yris that an elevated P deposition occurred in the most recent two decades, which was not covered in their modeling period. On the other hand, emissions from agricultural land that has received large quantities of P fertilizer was also not considered in previous simulations and should be included in future global and regional modeling work.

Although there are still uncertainties in obtaining global P budgets,it is clear that the perturbation of the global P cycle by anthropogenic emissions is much larger than traditionally believed. As a consequence,currently, more ecosystems than previously thought have been fertilized by increased P deposition. To elucidate the possible effects of this cascading trend, we further mapped atmospheric P deposition for the period 1954—2020 ( Fig. 3 ) and found that it was higher in Africa, America, Asia, and Europe, where natural terrestrial land areas are relatively P limited ( Fig. 1 ) ( Cleveland et al., 2013 ; Du et al., 2020 ). Our results highlight that elevated atmospheric P deposition in Asia and Europe may become an important source for P inputs, affecting plant productivity in ecosystems. Therefore, the future effect of atmospheric P deposition on terrestrial carbon sinks may be even larger than previously thought.While control policies aimed at reducing the emissions of aerosols will improve air quality, the trade-offeffects of atmospheric P deposition should also be accounted for owing to their fertilizing impacts on the terrestrial and oceanic sinks of carbon dioxide.

4. Conclusion

In this study we compiled data on atmospheric wet and dry deposition of in/organic P on a global scale, including locations at 396 sites and covering the period from 1959 to 2020. The whole dataset gave a geometric mean total P deposition value of 0.32 kg hayr, or a global P budget of 4.4 Tg yr.

Based on this new compilation, elevated P deposition was found in Europe and Asia during 2001—2020. The high deposition of P in these regions may be due to the increasing agricultural emissions from livestock farming or fertilized soils and fossil fuel combustion—related emissions,in addition to dust emissions.

To reduce the gaps between observations and simulations, future observations should focus on separate measurements of wet and dry deposition of in/organic P. Detailed emissions inventories of P are also needed, to refine atmospheric chemistry models and thus allow the impacts of P deposition on carbon sinks in ecosystems to be well evaluated.

Funding

This work was supported by the National Key Research and Development Project [grant numbers 2016YFD0800302 and 2017YFC0210103].

Declaration of Competing Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.aosl.2021.100051 .