Year-round observations of water-soluble ionic species and trace metals in Sapporo aerosols : implication for significant contributions from t rrestrial biological sources in Northeast Asia

Introduction Conclusions References

and ice nuclei (IN) and thus altering their hygroscopic properties on local, regional and global scales (Ramanathan et al., 2001;P öschl et al., 2010).They also have an adverse effect on the human health (Nel, 2005).Such impacts of aerosols are largely depend on aerosol loading and chemical composition (Baustian et al., 2012;D éandreis et al., 2012;Dentener et al., 1996;Nel, 2005;Schaumann et al., 2004;Sciare et al., 2005;Zhang et al., 2012) that are spatially and seasonally highly variable in the lower atmosphere.Aerosol concentrations vary in the range of 1-100 µg m −3 and their dominant chemical components are characterized by sulfate, nitrate, ammonium, sea salt, minerals, organics and elemental carbon, each of which typically contribute about 10-30 % of the overall mass load (P öschl, 2005).
Aerosol SO 2− 4 could affect the radiation budget directly by reflecting solar radiation and indirectly by altering the physical properties of aerosols and the CCN activity (D éandreis et al., 2012).Inorganic salts are more water-soluble and their solubility is much higher than organics, enhancing CCN activity of the particles (Zhang et al., 2012).On the other hand, mineral dust adds complexity, because it can play a significant role in radiative forcing (Dentener et al., 1996) as well as in the uptake of acidic gases (Goodman et al., 2000).Trace elements are typically present in elevated concentrations, especially in urban aerosols, and can play an important role in the risk of human health due to their high bioreactivity (Moreno et al., 2011;Schaumann et al., 2004).
High aerosol loadings are commonly observed in East Asia due to anthropogenic activities (Carmichael et al., 2009).Radiative forcing and climatic effects over East Asia have been shown to be large (Nakajima et al., 2007) with significant seasonal variations due to the enhanced concentrations of dust in spring (Li-Xia and Zhi-Wei, 2011) and/or SO 2− 4 in summer (Su et al., 2012).East Asian aerosols, including dusts from arid regions in Mongolia and North China, could be further transported across the Pacific Ocean (Alfaro et al., 2003;Kawamura et al., 2004;Uno et al., 2011) with an impact on the regional to global climate.However, the aerosol composition, sources and seasonal variations, which are crucial in reducing the uncertainty in modeling of the aerosol impacts, are still far from being fully understood, despite the extensive studies over the past decade (e.g.Huebert et al., 2003;Nakajima et al., 2007).For example, Huang et al. (2006) reported an underestimation of the annual averaged model-simulated aerosol optical depth (AOD) by ∼ 45 %.They interpreted that the obtained negative intercept is due to the inclusion of only the anthropogenic aerosols (SO 2− 4 , black carbon and organic carbon), suggesting that the contributions from natural sources might also be significant in this region.
It is generally believed that anthropogenic activities (largely coal combustion) and volcanic emissions are two major sources of particulate non-sea-salt (nss)-SO 2− 4 (Andres and Kasgnoc, 1998;Moore et al., 2006;Smith et al., 2011).Biological production of dimethylsulfide (DMS) is also considered as a significant source of nss-SO 2− 4 at least in marine regions (Moore et al., 2006;Oduro et al., 2012).Unlike nss-SO 2− 4 , methanesulfonate (MSA − ) can only be produced by the oxidation of DMS that originates from the microbial activity in the ocean.Abundant levels of MSA − and nss-SO 2− 4 were reported in the ocean/coastal atmosphere (Bates et al., 1992;Gaston et al., 2010;Gondwe et al., 2004;Legrand and Pasteur, 1998;Yang et al., 2009;Kouvarakis et al., 2002;Mihalopoulos et al., 1997).Recently, Miyazaki et al. (2012) reported significant concentrations of MSA − in forest aerosols and interpreted a pathway for the emissions of DMS from forest floor.
Other ionic species such as K + , Na + , Mg 2+ , NH + 4 and Cl − are derived from biomass burning, oceanic and/or biogenic emissions (Andreae, 1983;Graham et al., 2003;P öhlker et al., 2012).Graham et al. (2003) reported significant amounts of K, Na and Mg in the natural Amazonian aerosol and attributed to biogenic matter.Recently, P öhlker et al. ( 2012) found K-rich particles along with substantial amounts of NH + 4 , Cl − and SO 2− 4 in the Amazon region and attributed to biogenic emissions.On the other hand, metalliferous particles are directly emitted from natural sources such as volcanic eruptions, dust storms, rock weathering, forest fires and biogenic emissions at significant levels (Nriagu, 1989).Based on chemical (organic and trace metal) composition of individual aerosol particles, Van Malderen et al. (1996) reported that the main part of the particles above Lake Baikal, Siberia, could be related to natural sources.These studies suggest that the significant contributions of water-soluble ionic species and trace metals from the vegetation/biomass burning.However, less attention has been paid on terrestrial biogenic contribution of inorganic aerosols and such studies are limited to the forest regions (Graham et al., 2003;Maenhaut et al., 2002;Pauliquevis et al., 2012).
Sapporo is located in the west of Hokkaido Island, northern Japan (43.07 • N, 141.36 • E) and is an ideal site for the study of air masses delivered from Siberia, North China, and surrounding oceans (Aggarwal and Kawamura, 2008;Yamamoto et al., 2011).Previous studies of Sapporo aerosols for chemical and stable carbon (δ 13 C) isotopic compositions demonstrated that the spring and/or summer aerosols are largely influenced by the outflows from East Asia and Siberia and are characterized by anthropogenic and biogenic (including biomass burning) contributions with the atmospheric processing during long-range transport (Agarwal et al., 2010;Aggarwal andKawamura, 2008, 2009;Pavuluri et al., 2012).Recently, Yamamoto et al. (2011) detected terrestrial biomarkers in the Sapporo snow samples, which were interpreted by the long-range transport from Siberia, Russian Far East and Northeast China based on stable carbon (δ 13 C) and hydrogen (δD) isotope ratios of biomarkers together with backward trajectory analyses.However, the seasonality of biotic contributions is not clearly understood due to the lack of year-round observations.To our knowledge, no study has been conducted for trace metal compositions in the aerosols from Northeast Asia.
To better understand the characteristics and sources of atmospheric aerosols and their seasonal variations over Northeast Asia, we conducted one-year observation of total suspended particulate matter (TSP) collected from Sapporo, northern Japan for inorganic ions and trace metals.In this paper, we present a significant contribution of water-soluble ionic species and trace metals from terrestrial biological sources in Northeast Asia based on their seasonal variations and factor analysis.We also discuss the relations of selected trace metals (e.g.Ni, Cu, As) to specific organic tracers for pollen (sucrose) and fungal spores (mannitol).

Aerosol sampling
Aerosol (TSP) sampling (n = 21) was performed from 2 September 2009 to 5 October 2010 on the rooftop (ca.20 m above the ground level (a.g.l.)) of the Institute of Low Temperature Science (ILTS) building, Hokkaido University, Sapporo, northern Japan using a pre-combusted (450 • C, 4 h) quartz fiber filter and high-volume air sampler (∼ 65 m 3 h −1 ).The geographical details of the sampling site has been described elsewhere (Pavuluri et al., 2013).Briefly, the population of Sapporo, the capital of Hokkaido (the northernmost island of Japan), was 1.9 million in 2010 and the major industries include retail and tourism as well as manufacturing of foods, beverages, paper and pulp and machinery, from which the input of pollutants into the atmosphere is expected to be small.The climate in this region is strongly affected by the East Asian monsoon that controls the air mass source regions and compositions of atmospheric aerosols at the receptor site.Each sample was collected for ca. 2 consecutive weeks.Filter samples were placed in a pre-combusted glass jar with a Teflon-lined screw cap separately and stored in dark room at −20 • C prior to analysis.

Water-soluble ionic species
Water-soluble ionic species were measured by ion chromatograph (761 Compact IC, Metrohm, Switzerland).An aliquot of filter (1.2 cm in diameter disc) was extracted with 10 mL Milli Q water under ultrasonication for 20 min and filtered using a syringe filter (GL Sciences Chromatodisc Type A, 0.45 µm) and then injected into IC.For anion ) solution with a flow rate of 1.0 mL min −1 .A calibration curve was evaluated by the analyses of a set of authentic standards along with a sequence of filter samples.The analytical errors in duplicate analysis of filter samples were within 14 % for MSA − and 4 % for the rest of the ionic species.The concentrations reported here are corrected for four field blanks that were collected every season.

Trace metals
Trace metals were measured using an inductively coupled plasma mass spectrometry (ICP-MS, Thermo Electron X Series) after the acid microwave digestion of samples as reported in Theodosi et al. (2010).Briefly, a filter disc (1.0 cm in diameter) was placed in Teflon vessel (DAP-60 K, 60 mL/40 bar), to which concentrated nitric acid was added and then digested using a microwave digestion system (Berghof MWS-2).Indium (CPI International, S4400-1000241) was used as internal standard, and a calibration curve was evaluated using standard certified solutions (CPI International).
Recoveries obtained with the use of certified reference materials ranged from 90.0 to 104.1 %.The concentrations reported here are not corrected for field blanks because none of the trace metals measured were found to be significant in the field blanks.

Organic tracers
Details of the procedure for the determination of organic tracer compounds are described elsewhere (Fu et al., 2010).Briefly, the tracer compounds were extracted with dichloromethane/methanol (2 : 1; v/v) under ultrasonication and derivatized with 50 µL of N,O-bis-(trimethylsilyl)triflouroacetamide (BSTFA) with 1 % trimethylsilyl chloride and 10 µL of pyridine.The tracer compounds were then measured using a capillary  , respectively, whereas the precipitation occurred occasionally with the highest amount (range 5.5-153 mm) in summer (June to August) during rainy season with the highest frequency (range 14.3-73.3%) in winter (December-February) during snowfall season.September to November and March-May are classified as autumn and spring, respectively.The ground surfaces in Sapporo are covered with snow from late December to early April.

Air mass trajectories
Ten-day backward air mass trajectories arriving in Sapporo at 500 m a.g.l. were computed for every 48 h during each sample period using HYSPLIT model (Draxler and Rolph, 2012).Their three dimensional plots are presented in Pavuluri et al. (2013).Briefly, the air masses that arrived in Sapporo during autumn, winter and spring mostly originated from Siberia passing over Northeast Asia and the Japan Sea whereas in summer they mostly originated from the East China Sea and/or the western North Pacific passing over coastal region and/or the Japanese main island.The air parcels were mostly travelled at lower than 2000 m a.g.l. in all seasons.) and trace metals (Al, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Cd and Pb) in TSP samples (n = 21) measured in this study.TSP masses that were gravimetrically measured ranged from 13.5 µg m −3 to 73.8 µg m −3 with an average of 30.0 ± 12.7 µg m −3 .
Similarly, the third and fourth dominant species NH +

and Ca
2+ indicate a significant contribution from biogenic and/or biomass burning emissions and soil dust, respectively.High concentrations of MSA − (Table 1), which are comparable to those (range ∼ 10-100 ng m −3 ) reported in marine (Ayers and Gras, 1991;Mukai et al., 1995;Yang et al., 2009) and forest (Miyazaki et al., 2012)  biological (phytoplankton, higher plants and/or soil organisms) contributions during the campaign.It is well known that non-sea-salt contents of ionic species are estimated using the Na + as reference for sea-salt, however, we did not estimate them in this study to avoid potential underestimation because Na + could be contributed from other sources than the ocean.

Trace metals
Ca was found as the most abundant metal ranging from 196 ng m −3 to 2920 ng m −3 (Table 1).Its relative abundance in all the metals detected is 45 ± 5.2 % followed by Fe (27 ± 4.5 %).Al (21 ± 3.1 %) was found as the third most abundant followed by Zn 1).The total mass of Al, Ca, Fe, Ti and Mn that mainly originate from the crust accounted for 95±2 % of all the trace metals determined in this study.On average, they accounted for 7.45 ± 2.09 % (range 2.29-10.6%) of TSP.The amount of dust estimated from Al, whose mass ratio relative to the Asian dust is 7 % (Zhang et al., 2003), accounted on average for 24 ± 8 % (range 7-36 %) of TSP, suggesting that the soil dust is one of the major sources of Sapporo aerosols.On the other hand, Ca concentrations obtained from ICP-MS measurements were always higher by a factor of 1.3-3.2than water-soluble Ca 2+ obtained from IC measurements (Table 1).The insoluble fraction accounted for 26-69 % (average 51 ± 11 %) of total Ca, which may be attributed to the existence of insoluble minerals of Ca such as calcium oxide and feldspar.Similarly, the fourth most abundant Zn and high levels of Cu and Ni suggest that biogenic sources might be important.However, Pb was found to be almost equally abundant together with Ni, indicating that the contributions from anthropogenic activities might also be significant.

Seasonal variations of ionic species and trace metals
Seasonal variations of ionic species and trace metals are shown in Fig. 2 to mid spring.They remained high until mid summer followed by a gradual decrease to winter, although the trend is not clear in autumn (Fig. 2a).This pattern is similar to that of volatile organic compounds (VOC) emission from vegetation; that is, their emissions are negligible in winter but increased toward the growing season (mid spring to mid summer) (Guenther, 1997).The ambient temperature, which is a key factor to control the BVOC emission, showed a distinct seasonal trend in Sapporo with winter minima and summer maxima (Fig. 1).A similar pattern should prevail in Northeast Asia.Interestingly, seasonal variation of SO 2− 4 is characterized by an early summer maximum, being different from that of NO − 3 (Fig. 2a), whose concentrations are higher in autumn to spring but lower in summer.The air masses arriving over Sapporo originated from oceanic regions in summer (Pavuluri et al., 2013).These comparisons suggest that biogenic emissions might be a major source of SO 2− 4 in Sapporo, although we could not preclude anthropogenic activities.Concentrations of Cl − are higher in autumn to spring than in summer whereas Na + and Mg 2+ peaked in autumn and early spring and minimized in winter as well as in summer (Fig. 2a).The seasonal variations of ions may be significantly influenced by biogenic sources and/or soil dust, in addition to marine contribution.
Concentrations of Ca 2+ stayed low in autumn and summer and slightly decreased in winter, however peaked in early spring (Fig. 2a).Its concentrations may be associated with wet scavenging by snow in winter when snow precipitation frequency occurred (Fig. 1).Its peak in early spring may be due to the soil dust injection to the atmosphere from the arid regions of Mongolia and Northeast China (Wang et al., 2004;Zhang et al., 2003)   whereas K + showed a decrease toward late winter followed by an increase in early spring (Fig. 2a).The higher levels of these species during mid autumn to mid winter might be due to significant contributions from forest fires and/or biomass burning on local and/or regional scale whereas the increase of K + from late winter to early spring might be associated with the long-range transport of soil dust.
Crustal metals such as Al, Ca, Fe, Ti and Mn did not show any significant seasonal variations, although their concentrations are relatively low in winter (Fig. 2b) probably due to wet scavenging (Fig. 1).They showed a very sharp increase in early spring (1-15 April) followed by a gradual decrease to late autumn (Fig. 2b).Being similar to Ca 2+ , these trends indicate a significant contribution of soil dust particles that were probably transported from the arid regions in Mongolia and Northeast China.Ni showed a gradual decrease from autumn to mid winter and a gradual increase to early summer followed by a decrease in mid summer (Fig. 2b).This seasonal pattern is comparable to that of VOC emission from vegetation (Guenther, 1997).The seasonal trends of both Cu and As were also quite similar to that of Ni in autumn to mid spring and the rest of the period (Fig. 2b).In fact, Cu has often been considered to be of biological origin and found abundantly present in forest aerosols (Graham et al., 2003).These results suggest that Ni, Cu and even As may mainly be derived from biogenic sources (see Sect. 3.5.4for further discussion), although they could partly be derived from anthropogenic sources.
Zn has also been considered as a biogenic metal derived from biomass burning (Graham et al., 2003;Pauliquevis et al., 2012).Its concentrations are rather constant in winter to late spring, although a high value was observed in early summer (Fig. 2b).Zn may be derived from biogenic sources at least in spring.V and Cr did not vary significantly in autumn but showed a gradual decrease to winter followed by a sharp increase in early spring.Thereafter V stayed relatively high until late summer whereas Cr decreased gradually to early summer (Fig. 2b).Cr and V may have been significantly contributed from soil dust in early spring and anthropogenic sources in the rest of the period.In contrast, Cd and Pb did not show any clear trend throughout the campaign (Fig. 2b), suggesting that they are mainly derived from anthropogenic activities.

Seasonality in ionic balance and correlations
The equivalent ratios of total cations (Na + , NH + 4 , K + , Ca 2+ and Mg 2+ ) to anions (Cl − , NO − 3 and SO 2− 4 ) ranged from 0.89 to 1.16 with an average of 1.00 ± 0.07 (n = 21) and seasonal averages of 1.03±0.09(autumn, n = 8), 0.95±0.02(winter, n = 3), 0.98±0.05(spring, n = 5) and 1.01 ± 0.07 (summer, n = 5).Although the overall slope of linear regression line for total cations and anions (Fig. 3) is 0.83 ± 0.03, the slopes for individual seasons are slightly different; 0.67 ± 0.07 in autumn, 0.86 ± 0.08 in winter, 0.86 ± 0.04 in spring and 0.85 ± 0.06 in summer.This result may be caused by a lack of measurements of H + , HCO − 3 , CO 2− 3 , PO 3− 4 and organic acid anions.As discussed in previous section, the seasonal trends of most ionic species showed a distinct seasonal pattern (Fig. 2a), reflecting their source type/strength in each season.Therefore, it is reasonable to consider that the data sets should be grouped seasonally for the analyses of ionic species and their characteristics although the numbers of winter samples are small.
Average NH + 4 /SO 2− 4 equivalent ratios were 0.36 ± 0.3 (autumn), 1.16 ± 0.09 (winter), 0.47 ± 0.19 (spring) and 0.47± 0.14 (summer).These results suggest that the atmospheric NH 3 was not abundant enough to neutralize H 2 SO 4 , except for winter.In addition, NH + 4 showed a strong correlation with SO 2− 4 in summer while the correlations are weak in other seasons (Fig. 4a).Similarly, NH + 4 showed a good correlation with NO − 3 in autumn and winter (Fig. 4e) and a strong correlation with Cl − in winter (Fig. 4j) although the correlations are insignificant in other seasons (Fig. 4e, j).Hence, it is likely that SO in spring (Fig. 4b) may suggest that they were significantly contributed from biogenic emissions (see Sect. 3.5.4for further discussion).
The equivalent ratios of Cl − /Na + ranged from 0.14 to 1.17 (average 0.72 ± 0.39) in autumn, 1.05-1.54(1.29 ± 0.25) in winter, 0.14-1.29 (0.76 ± 0.45) in spring and 0.01-0.09(0.04 ± 0.03) in summer.Cl − showed a good correlation with Na + in winter and spring whereas it was moderate in autumn (Fig. 4i).In summer, there was no relation due to negligible amounts of Cl − , despite that the summer air masses arriving to Sapporo originated from oceanic regions (Pavuluri et al., 2013).The negligible amounts of Cl − in summer can be explained by a significant depletion of Cl − through HCl formation followed by its volatilization from the aerosols during the long-range transport.On the other hand, Cl − showed a good correlation with NH + 4 and K + in winter.Although the correlation is positive in autumn and spring, the plotted data points are somehow scattered (Fig. 4j, k).These results imply that Cl − should have been significantly contributed from biomass burning at least in winter that is possibly present in the form of KCl in addition to NH 4 Cl and NaCl.

Factor analysis and source identification
Factor analysis is widely used to identify and apportion the sources of particulate pollutants at receptor sites (Moreno et al., 2011;Pauliquevis et al., 2012).To better characterize the different sources of inorganic ions and trace metals in Sapporo aerosols, varimax rotated factor analysis was conducted here for all the data set using IBM SPSS statistics version 20 for Mac.Five factors were obtained with Eigen values greater than 1 that explain 88 % of the total variance of the data set (Table 2), representing four source categories: (i) soil dust emission, (ii) marine and biological source, (iii) anthropogenic source, and (iv) terrestrial biogenic emission.

Soil dust contributions
The first factor, which accounted for 32.5 % of the total variance, is strongly loaded with Ca 2+ , Al, Ca, Ti, Mn, and Fe, which are typical crustal components, as well as Cr.Al and Ca/Al and Fe/Al ratios have been widely used as tracers for suspended soil and long-range transport of mineral aerosols (Alfaro et al., 2003).On the other hand, NO − 3 , SO 2− 4 , Mg 2+ , V, Ni and As showed moderate loadings (≥ 0.4) whereas MSA − , Na + , K + and Pb showed minor (≥ 0.2) loadings with factor 1 (Table 2).It is likely that V and Pb are mostly originated from anthropogenic activities whereas biogenic emissions and biomass burning are associated with MSA − , K + and Ni, which should be adsorbed on soil dust particles (Moreno et al., 2011;Pauliquevis et al., 2012).NO As are likely derived from mixed sources of anthropogenic and biogenic emissions (Graham et al., 2003;Nriagu, 1989;Yamasoe et al., 2000).Further, the association of Na + and Mg 2+ with factor 1 can mostly be linked to their minor contributions from soil dust (Guo et al., 2007).Hence, this factor can mainly be attributed to soil dust but the mixed sources of anthropogenic and biogenic emissions may also be associated.

Marine and biological sources
The second factor is strongly loaded with Cl − , Na + and Mg 2+ that accounted for 19 % of total variance (Table 2).These ions are originated mostly from ocean because Na + is emitted from ocean surface; thus factor 2 should be associated with sea-salt emission.
On the contrary, NO − 3 that can be produced from fossil fuel combustion and/or biomass burning showed a moderate loading (0.37) with factor 2. HNO 3 may react with alkaline particles such as NaCl, which are present in coarse mode.Seasonal variation of Cl − is similar to that of NO − 3 (Fig. 2a), suggesting that Cl − should have significantly been derived from biomass burning in addition to marine emissions.As noted earlier, soil dust might also be a minor source of Na + and Mg 2+ .Na suggest that the soil dust importantly contribute to Na + and Mg 2+ at least in winter and spring.However, Na + and Mg 2+ showed negative relations with Al in autumn (r = 0.37 and 0.43, respectively) and summer (0.60 and 0.19, respectively).
Interestingly, Na + is most abundant in autumn (Fig. 5), although the peak values in spring are comparable to those in autumn (Fig. 2a).The average and median values of Mg 2+ in autumn and spring are the same (0.19 µg m −3 ), despite the enhanced soil dust contribution in early spring.The abundant Na + and Mg 2+ in autumn suggest a significant contribution from season specific source(s), probably biogenic emissions such as fungi.Fungi can act as biological weathering agents through biomechanical and biochemical processes and play an important role in biogeochemical cycling of nutrients and metals (Burford et al., 2003).Fungal activities on soil surface and plants emit the spores and hyphae, which are enriched with several elements including Na and Mg at significant levels (Burford et al., 2003;Hammer et al., 2011).Their activities on fallen leaves may be increased during defoliation season.In fact, the concentrations of mannitol that is mainly produced by various fungi were found highest in the autumn aerosols (20.3 ± 13.3 ng m −3 ): winter (0.42 ± 0.37 ng m −3 ), spring (2.27 ± 1.13 ng m −3 ) and summer (7.07 ± 3.38 ng m −3 ) (Pavuluri et al., 2013).In summer, the air masses that arrived in Sapporo mostly originated from oceanic region and passed over the coastal regions of Japanese Islands.Hence, it is likely that the contributions from marine emissions could be significant in summer.However, the concentrations of both Na + and Mg 2+ were found to be lowest in summer (Fig. 5).This result implies that the biological contribution of these elements is more significant in autumn, winter and spring than summer.However, we cannot preclude some contribution of marine aerosol due to mixing of oceanic air masses with terrestrial air masses along the transport pathway.

Anthropogenic sources
The fourth factor is found to associate with Cd and Pb, which are most likely derived from the anthropogenic activities.Factor 4 accounted for 13 % of the total variance (Table 2).NO − 3 that can originate from both fossil fuel combustion and biomass burning also showed a significant loading (0.58) with this factor.Hence, this factor can be interpreted as anthropogenic emissions.However, NO − 3 is well correlated with Cl − as noted earlier, suggesting that these anions may have common origin; biomass burning.In addition, Cr, Zn and As as well as K + are all associated with factor 4. As discussed earlier, these elements might be contributed from biogenic/biomass burning emissions, although we could not preclude the input from industrial activities.Nriagu (1989) reported that the biogenic contributions can account for 30-50 % of the total As and several other trace metals.Therefore, factor 4 should be associated with a mixing of biogenic and/or biomass burning emissions as well, although the anthropogenic emissions are major.

Terrestrial biogenic emissions
Factor 3 is loaded with Cu, Ni and As whereas factor 5 is with K + , Zn and SO 2− 4 (Table 2).In addition, V showed a moderate loading with both factors whereas NH showed a considerable loading with factor 5 (Table 2).It is of interest to note that MSA − , which is likely produced by the oxidation of DMS originated from phytoplankton, higher plants and soil as well as biomass burning, showed a minor loading (0.24) with factor 5. Ni, Cu, As, Zn and SO 2− 4 can originate from natural and/or anthropogenic sources (Graham et al., 2003;Moreno et al., 2011;Pauliquevis et al., 2012), whereas K + originates only from biomass burning and/or biogenic emissions (Andreae, 1983;Graham et al., 2003).Generally, V is considered as a tracer of oil and coal combustion, however, it has been detected in forest aerosols from the Amazonian atmosphere, which is quite pristine (Graham et al., 2003;Pauliquevis et al., 2012).NH mainly from biological sources and/or biomass burning (Graham et al., 2003;P öhlker et al., 2012).
Ni, Cu and As showed a seasonal trends (Fig. 2a) similar to that of VOC emission from vegetation (Guenther, 1997) as noted earlier.Their seasonal trends are comparable to those of MSA − (Fig. 2a) and similar to that of mannitol in Sapporo aerosols (Pavuluri et al., 2013), which is considered as a tracer for bioaerosol (Gilardoni et al., 2011).As shown in Fig. 6, Ni, Cu and As positively correlate with mannitol in Sapporo aerosols (r 2 = 0.33, 0.47 and 0.28, and p = 0.006, 0.001 and 0.013, respectively).In contrast, the tracers for combustion sources; (i) hopanes (C 27 -C 32 hopanoid hydrocarbons), specific biomarkers of petroleum and coal (Schauer et al., 2002 and references therein), (ii) levoglucosan, an excellent marker for biomass burning (Simoneit, 2002), and (iii) elemental carbon (EC) that is derived from combustion sources, did not show any significant seasonal trend in Sapporo aerosols (Pavuluri et al., 2013).It is also of interest to note that Ni, Cu and As showed negligible correlations with EC (r 2 = 0.03, 0.08 and 0.18, respectively) and even negative correlations with hopanes (r 2 = 0.22, 0.01 and 0.001, respectively) and levoglucosan (r 2 = 0.25, 0.03 and 0.03, respectively).Zn also showed a gradual increase from winter minimum to late spring maximum although its level was relatively stable in summer and autumn, except for a sporadic peak in early summer (Fig. 2b).We found a positive correlation (r 2 = 0.60) between Zn and sucrose whereas weak or negligible correlations with hopanes (r 2 = 0.24), levoglucosan (r 2 = (−)0.001)and EC (r 2 = 0.12) in Sapporo spring aerosols (Pavuluri et al., 2013).Because sucrose is a tracer for pollens derived from terrestrial plants (Fu et al., 2012), this result suggests that Zn is emitted from the terrestrial biota to the atmosphere in spring although no relation was found in other seasons.Sucrose was abundantly detected in spring aerosols from Sapporo (Pavuluri et al., 2013).It is noteworthy that the association of Ni and Cu and Zn with different factors (Table 2) is similar to that reported in forest aerosols (Graham et al., 2003), in which both Cu and Zn are attributed for biological matter.These results and comparisons imply that Ni, Cu, As and Zn and On the other hand, K + in winter showed a strong correlation (r 2 = 0.97) with levoglucosan but the correlation was weak (r 2 = 0.22) in autumn and no relation was found in other two seasons.The average concentration of K + (Table 1) is comparable to those in forest aerosols from Balbina, Central Amazonia (up to 140 ± 88 ng m −3 in fine mode during dry season), where local biomass burning is insignificant and long-range transported biomass burning aerosol is not severe (Pauliquevis et al., 2012).Orsini et al. (1982) reported a strong correlation between S and K in fine particles from a remote site in the Amazon forest, which they interpreted by gas-to-particle conversion of gases emitted from higher plants.P öhlker et al. ( 2012) detected K-rich particles in the Amazon forest and attributed to biogenic primary emissions particularly of fungal spores.In addition, SO 2− 4 showed a linear relation with MSA − (Fig. 7), which has been widely used as a tracer to estimate the biogenic nss-SO 2− 4 (Bates et al., 1992;Kouvarakis et al., 2002;Mihalopoulos et al., 1997;Yang et al., 2009).Although the correlation varies from season to season (r 2 = 0.35 in autumn, 0.05 in winter, 0.25 in spring and 0.12 in summer), these results suggest that the SO 2− 4 is significantly derived from biological sources throughout the year.enhanced emission of DMS from phytoplankton bloom followed by the subsequent oxidation (Gaston et al., 2010;Legrand and Pasteur, 1998;Mihalopoulos et al., 1997).On the other hand, higher plants and soils as well as biomass burning could emit gaseous sulfur species including DMS (Geng and Mu, 2006;Meinardi et al., 2003;Yi et al., 2010) and may contribute to the atmospheric budget of MSA − and SO 2− 4 .VOC emission from vegetation is higher in growing season and lower in winter (Guenther, 1997).Recently, Miyazaki et al. (2012) reported significantly high levels of MSA − , with a maximum during May and July, in the forest aerosols collected at two levels (∼ 2 m and ∼ 15 m) of the experimental forest in Sapporo, where the canopy height is ∼ 20 m.
MSA − was found to inversely correlate with Na + during the campaign (n = 21).
A strong anti-correlation (r 2 = 0.42) was found in summer when the air masses arriving in Sapporo originated mostly from oceanic regions (Pavuluri et al., 2013).Interestingly, a negative correlation was found between MSA − and levoglucosan during the whole campaign.A strong anti-correlation was found in autumn (r 2 = 0.32) and winter (r 2 = 0.34) when levoglucosan became more abundant (Pavuluri et al., 2013).Further, the concentration range of MSA − (Table 1) is comparable to that (9-95 ng m −3 ) in the forest aerosols in Sapporo (Miyazaki et al., 2012).As noted earlier, MSA − showed a gradual increase from winter to growing season followed by a gradual decrease to autumn (Fig. 2a).In autumn, the microbial activities on surface soil and fallen leaves may be significant, while MSA − did not show any seasonal trend (Fig. 2a).Hence, it is more likely that MSA − in Sapporo aerosols should have been produced from DMS that mainly originated from higher plants and soil organisms rather than phytoplankton in the ocean and biomass burning in the continent.
Sulfate also showed a negative correlation with Na + , being similar to the case of MSA  with Na + .These results together with air mass trajectories imply that the long-range atmospheric transport of soil dust, terrestrial biogenic emissions including microbial activities, and biomass burning are major sources of Sapporo aerosols although anthropogenic emissions and sea-salt emissions are minor sources of atmospheric aerosols in Northeast Asia.In addition, the seasonality of the source strength is significant, except for anthropogenic emissions.This study suggests the importance of terrestrial biological contributions to inorganic aerosols and their climatic impacts on regional to global scale.

1
Introduction Atmospheric aerosols have an impact on the Earth's climate system directly by absorbing and reflecting solar radiation and indirectly by acting as cloud condensation (CCN) Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

6591
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

6593
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |
aerosols, further support the significant 6597 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | . Seasonal variations of MSA − and SO 2− 4 are both characterized by a gradual increase from winter Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

+
4 then showed a gradual decrease toward early spring 6599 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | + showed a strong (r = 0.97 in winter) and good (r = 0.60 in spring) correlation with Al, a tracer for crustal origin.Mg 2+ and Al also showed a correlation in winter (r = 0.97) and spring (0.83).These results 6603 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | even V should have been derived from different types of biological sources rather than anthropogenic activities.
As shown in Fig.8, both MSA − and SO 2− 4 were abundant in summer followed by spring.Their concentrations were lowest in winter.Average concentrations of MSA − were 27.4 ± 8.3, 21.9 ± 5.5, 49.2 ± 20.2 and 64.4 ± 14.5 ng m −3 in autumn, winter, spring and summer, respectively whereas those of SO 2− 4 were 2.96 ± 0.77, 2.65 ± 0.09, 3.81 ± 0.50 and 4.45 ± 1.26 µg m −3 .Abundant MSA − and biogenic SO 2− 4 were reported in the marine and coastal atmospheres during summertime, which were attributed to the 6607 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

−.
This result suggests that SO 2− 4 should have same or similar origin with MSA − : terrestrial biogenic emissions.No correlation (r 2 = 0.03) was found between SO 2− 4 and levoglucosan although anti-correlations were found between MSA − and levoglucosan (r 2 = 0.32) in autumn.The negative correlation between SO 2= 0.42) in summer.The Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | while crustal elements including Na + , Ca 2+
Table 1 presents statistical summaries of water-soluble ions (MSA − , Cl − , NO in the forms of H 2 SO 4 and/or NH 4 HSO 4 in addition to (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl.On the other hand, both SO 2− 4 and NO − 3 showed a positive relation with metal species: K + , Ca 2+ and Mg 2+ , except for autumn (they negatively correlate with Mg + 4 are present 2+ ), although the correlations are rather poor and Mg 2+ peaked in early spring.On the other hand, Cl − and NO Zn showed a gradual increase from winter to late spring.Cd and Pd did not show any clear seasonal trend, staying relatively constant throughout the year.We found that Ni, Cu and As positively correlate with mannitol (a tracer for fungal spores), while Zn positively correlates with sucrose (tracer of pollen) in spring.Similarly, K+ showed linear relation with levoglucosan (a tracer of biomass burning) in au-

Table 1 .
Statistical summary for the concentrations (ng m −3 ) of water-soluble ionic species and trace metals in atmospheric aerosol (TSP) samples (n = 21) collected from Sapporo, northern Japan during 2 September 2009 to 5 October 2010.

Table 2 .
Varimax rotated factor loadings for water-soluble ionic species and trace metals in the Sapporo TSP samples (n = 21) during the campaign.