Measurement report: Short-term variation of ammonia concentration in an urban area: contributions of mist evaporation and emissions from a forest canopy with bird droppings

Short-term variations of NH3 concentrations in the urban atmosphere are affected by local meteorological conditions and variations of natural and anthropogenic sources. To investigate potential sources and processes of NH3 variation in an urban area, hourly NH3 and NH4 concentrations were measured from 10 November 2017 through October 2019 in Nagoya, a megacity located in central Japan. Monthly averages of NH3 concentrations were high in summer and low in winter. Daily minimum NH3 concentrations were almost linearly correlated with daily minimum air temperature. In contrast, daily maximum NH3 concentrations revealed an exponential increase with temperature, suggesting that different processes with air temperature acted during the nighttime and daytime. Short-term increases of NH3 concentrations of two types were examined closely. The first 15 is a rare but large increase (11 ppb for 2 hr) after mist evaporation during daytime. It is noteworthy that an event of this magnitude was identified only once during two years of observations at Nagoya even though evaporation of mist or fog droplets is expected to be frequent after rain. The second short-term increase was a large morning peak in summer. After selected days were fulfilled with non-wet and weak wind conditions, the amplitude of diurnal variation of NH3 concentration (daily maximum minus minimum) was analyzed: the amplitude was small 20 (ca. 2 ppb) in winter but it increased from early summer along with new leaf growth. It peaked in summer (up to ca. 20 ppb) during intense addition of droppings from hundreds of crows on trees in the campus assembled before roosting. The high daily maximum NH3 concentration was characterized by a rapid increase occurring 2–4 hr after local sunrise. Daily and seasonal findings related to the morning peak implied that stomatal emission at the site was responsible for the increase. The yearly difference between daily amplitudes during the two summers 25 was explained by the difference in the input amounts of reactive nitrogen derived from bird droppings and some rain, suggesting that the canopy of a small forest affected by the bird droppings might act as a temporary but strong source of NH3. https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c © Author(s) 2020. CC BY 4.0 License.


Introduction 30
Ammonia (NH3) plays an important role in various atmospheric chemical processes (Behera et al., 2013). For example, NH3 is the major precursor of fine aerosol particles of constituents such as ammonium sulfate and ammonium nitrate (Seinfeld and Pandis, 2016). In addition, aerosol particle acidity is modified by neutralization with NH3 (e.g. Murphy et al., 2017;Song and Osada, 2020). Aerosol particles affect human health and climate; therefore, reduction and control of aerosol concentration are desired for many situations (Dockery et al., 1993;35 IPCC, 2013). As an important gaseous precursor of aerosol particles, NH3 sources and factors affecting concentrations have been studied for decades. Various natural and anthropogenic sources of NH3 are known (Sutton et al., 2008;Behera et al., 2013). Although agricultural NH3 sources (domestic animals, fertilizer loss, etc.) are dominant emitters on a global scale, non-agricultural sources (motor vehicles, industries, garbage, sewage, humans, wild animals, etc.) are also major contributors, especially in urban areas (e.g., Perrino et al., 2002;40 recent years (Ueta et al., 2003;Vuorisalo et al., 2003). They form large roosts in scattered forests in urban areas and drop excreta from trees and wires to the ground during pre-roosting assembly and when resting in roosts.
However, the potential of NH3 emissions related to bird droppings in urban green areas has not been studied. 65 Analysis of hourly concentrations in the atmosphere is useful to explain the sources and processes of ambient NH3. For example, a temporal correlation between vehicular exhaust species such as NOx, CO, and elemental carbon in urban area has been inferred for vehicular emissions of NH3 (e.g. Perrino et al., 2002;Nowak et al., 2006;Osada et al., 2019). Moreover, temporal analyses have been made of NH3 concentrations at grasslands, which have allowed elucidation of the link between the morning peaks and dew formed on plant surfaces during 70 the previous night (Wentworth et al., 2014;. Hourly NH3 measurements are also a key technique to ascertain the bidirectional exchange of NH3 through the canopy layer (e.g. Wyers and Erisman, 1998;Nemiz et al., 2004;Kruit et al., 2007;Hansen et al., 2013) because NH3 transfer is governed by rapidly changing meteorological (sunlight availability, temperature, relative humidity, etc.) and plant physiological (stoma opening and closing, etc.) parameters (Schjoerring et al., 1998;. In fact, NH3 exchange between plants and ambient air occur 75 mainly through stoma when they open during daytime for photosynthesis (Farquhar et al., 1980). Therefore, the degree and direction of the NH3 exchange are expected to vary diurnally, highlighting the importance of hourly measurements of related parameters.
To investigate potential sources and processes controlling variation of NH3 concentration, hourly data of NH3 and NH4 + were measured from November 2017 to October 2019 in Nagoya, central Japan. The data were analyzed 80 by particularly addressing various time scales and the amplitude of diurnal variation in relation to potential reactive nitrogen sources and plant physiology near the site. These data are expected to elucidate ambient NH3 levels in urban areas with scattered forest affected by large amounts of bird droppings.

Observation 85
Atmospheric observation was conducted at Nagoya University in Nagoya city located in the central area of Honshu Island of Japan (Fig. 1a). Industrial area with busy port is located at about 10 km southwest from the campus of Nagoya University (Fig. 1b). The Nagoya city population is about 2.3 million. Despite the large city population engaged in numerous industrial activities, the air pollution level is not so high. Recent levels of the annual mean PM2.5 concentration are approximately 12 µg/m 3 (Nagoya City, 2019: 90 http://www.city.nagoya.jp/kankyo/page/0000117927.html). The nearest agricultural activities (farming land) are done about 4 km southeast from the campus. Garbage collection in the city requires 1) that burnable waste including food waste and other materials be packed into predesignated plastic bags and 2) that the garbage bags must be put out in a specified collection place by 8:00 a.m. on the regular (twice per week) collection day, https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License. preventing unnecessary NH3 emissions during garbage collection. However, the garbage bags might be pecked 95 by crows when deterrents to bird pecking are insufficient, providing the possibility of food supply for adaptation of omnivorous animals, such as crows, in urban areas . The observation site is located within the campus. Therefore, effects of residential garbage are expected to be small. The annual mean air temperature in Nagoya is 15.8°C; the annual mean rain amount is about 1540 mm (Japan Meteorological Agency: http://www.jma.go.jp/jma/index.html). Seasons in Nagoya have warm-humid summers with southern winds from 100 the Pacific Ocean and cold-dry winters with winds from the northwest, originating from continental Eurasia.
Measurements of NHx (NH3 and NH4 + in fine particles) were taken at Nagoya University (35. 16°N, 136.97°E), located in an eastern residential area of Nagoya city. Meteorological data (air temperature, relative humidity, etc.) were obtained from the Nagoya Local Meteorological Observatory, ca. 2 km north from Nagoya University (data available from https://www.jma.go.jp/jma/index.html). NOx concentrations were observed at the 105 Nagoya national air pollution monitoring site located ca. 2 km north from Nagoya University (data available from http://soramame.taiki.go.jp/). The equipment used for NHx measurements was set in a room located on the seventh floor of the Environmental Studies Hall on the main campus of Nagoya University. The northeastern side of the building faced upslope with a small forest mixed with deciduous and evergreen trees (Fig. 1c). Scattered trees and buildings 110 are located on the other side of the hall. Hourly measurements of NHx were conducted using a semi-continuous microflow analytical system (MF-NH3A; Kimoto Electric Co. Ltd.; Osada et al., 2011). Two identical sampling lines were used to differentiate total ammonium (NH3 and NH4 + ) and NH4 + (p) alone after removal by a H3PO4 coated denuder. After passing an impactor (cut-off diameter of about 2 μm) and an inner frosted glass tube (one coated; the other uncoated; both are 3 mm inner diameter and 50 cm long), pure water droplets were added to the 115 sample air at 100 μl min −1 . The equilibrated sample water was analyzed respectively using a microflow fluorescence analyzer to quantify NH4 + in the lines of NH4 + and total ammonium. The NH3 concentration was calculated based on their difference. The sample air flow rate of the NHx system was 1 L min −1 . The temporal resolution was ca. 30 min for one pair of NH4 + and total ammonium measurements. The detection limit of NH3 concentration was about 0.1 ppbv (Osada et al., 2011) under stable atmospheric NH3 and NH4 + concentrations. 120 Equivalence of two sample lines and the span of the calibration slope was checked monthly using NH3 standard gas at about 4 ppb diluted from 100 ppm (Taiyo Nippon Sanso Corp.). The NHx system was calibrated monthly using a standard NH4 + solution prepared from a certified 1000 ppm solution (Fujifilm Wako Pure Chemical Corp.). https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License.

Diurnal variation during summer and winter
Results of measurements taken in summer (July, 2018; Fig. 2) and winter (December, 2018;Fig. 3) are presented to explain the relation between NH3 concentrations and other parameters. In summer (Fig. 2), the Pacific highpressure system dominates the Japan archipelago, engendering continuous good weather with land−sea breeze cycles: south−southwest winds during afternoon and north−northeast winds after midnight to early morning. 130 When good weather continued, for example of 14-24 July, regular diurnal variations were visible in air temperature and wind speed: high in the afternoon and low in the midnight to early morning. For this period, the diurnal variation of NH3 concentration was extremely wide. In other words, the difference between maximum and minimum concentration ranges from nearly 10 to more than 20 ppb. In contrast, the NH3 concentration dropped to a few ppb and remained constantly low for the duration of the rain with small diurnal variation as that 135 found in 4-7 July when Baiu front (East Asian rainy front) was active near the site. A similar tendency of low NH3 concentration during rainy days was also found during observations, as reported at other places (Roelle and Aneja, 2002;Ellis et al., 2011). Increased contents of soil pore water dilute NH4 + in liquid phase and inhibit evaporation as NH3. Furthermore, wet surfaces of cuticular of leaf absorb ambient NH3 under high relative humidity during rain. Moreover, NH3 concentrations were low during the day of higher wind, such as 23 July. 140 The NOx concentrations in July were mostly below 0.02 ppm with no clear correlation with NH3 variation. Under low winds during winter, a surface inversion layer often developed in the lower atmosphere, preventing vertical diffusion of locally emitted pollutants (e.g., Kukkonen et al., 2005. It is particularly interesting that temporal variation of daily minimum NH3 concentration in winter is roughly following the day to 150 day variation of the daily minimum temperature. For example, a higher minimum NH3 concentration of about 3 ppb during 2−4 December decreased to ca. 1 ppb around 10 December, which corresponds to a decreasing trend of high to low air temperatures for these days. In contrast to the modest variation in July, NOx concentrations in December showed large variation: it was displayed frequently as more than 0.05 ppm during calm winds. In Fig. 3, the concentration peaks on 3-4, 11, 155 19-23, and 25-26 December were associated with low winds. High NOx correlates well with high NO concentrations (not shown), suggesting strong contributions of emissions from internal combustion. In addition https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License.
to the NOx variation, the NH3 concentration increased also under low winds, as described earlier. The similarity of NH3 temporal variation with NOx suggests that emissions from motor vehicles partly contribute to ambient NH3 concentration in winter, as reported in Tokyo, Japan . 160 Figure 4 depicts average diurnal variations of NH3 and NH4 + concentrations, air temperature (Ta), wind speed (WS) and wind direction for July (left column) and December (right column) in 2018. For July, large NH3 variation was visible as sharp peaks at around 8 o'clock in the morning. In contrast, broad and modest maximum NH3 concentrations were observed slightly before noon in December. The start timing of the NH3 increase was in general accord with the hours of rising temperature and dropping relative humidity. About 2-3 h delay of the 165 NH3 peak time between summer and winter might be related to the difference of local sunrise: about 4:50 for July and 6:50 for December. The diurnal variation of NH3 concentration in December showed no large morning peak at rush hour around 7−9 o'clock. Therefore, contributions of rush hour emissions are apparently limited for only those days under calm wind conditions in Nagoya.
Regarding the relations between NH3 concentration and temperature, dissociation of particulate NH4NO3 in 170 the atmosphere is known to be strongly related to temperature: partitioning toward gas phase is favored under higher air temperatures (Mozurkewich, 1993). However, NH4 + concentrations in both seasons were not simply a mirror of diurnal temperature variations. Although a small decline of NH4 + concentration around noon time in December might result from the dissociation of NH4NO3, it is difficult to discern because of the large variation.
Another interesting point is the relation between the average air temperature and the lower tenth percentiles for 175 NH3 concentrations, which both show a maximum at around noon for July and December. This finding is discussed later in greater detail. Figure 5 presents an example of a sudden increase from 2 ppb to 13 ppb of NH3 concentration during 2 hr 180 associated with drying mist after rain on 15 November, 2017. This magnitude of NH3 increase after rain was rarely observed during the study period of two years. Mist is defined as reduced horizontal visibility between 1 and 10 km by suspending water droplets in the atmosphere. In Nagoya, mist is often observed before or after rain.

Peak after mist evaporation
In this case, rain ceased in the early evening of November 14, but the mist continued until 10 am of November 15. Associated with relative humidity dropped sharply from ca. 90% at 10 AM to ca. 40% at noon, mist 185 disappeared and NH3 concentration abruptly increased as a mirror of temporal variation of relative humidity (RH).
Slight enhancement of NH3 concentration after the rain has been described in some reports of the literature (Roelle and Aneja, 2002;Ellis et al., 2011). They discussed the hypothesis on enhancement associated with the combination of an increase in the ammoniacal nitrogen concentration in the soil and diffusion from the soil to air after the drying pore solution. However, this process requires more time after cessation of rain to decrease soil 190 https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License. moisture; then, it is too slow to raise the atmospheric NH3 concentration. In contrast, Wentworth et al. (2014Wentworth et al. ( , 2016 reported that rapid increase of NH3 in the morning was attributed to evaporation of dew containing high concentrations of NH4 + : the pH of dew was slightly acidic, but near neutral. As their results and the rate shown in Fig. 5, the sudden event of NH3 concentration increase was regarded as resulting from evaporation of the mist droplets rather than soil-microbe-related processes. A similar rapid NH3 increase up to 15 ppb during 4 hr was 195 observed in Nagoya after drying hydrometeor and wet surfaces (Osada et al., 2018). During that event, mist was also observed after rain; the rain pH was ca. 5.6 with higher contents of sea salt and Ca 2+ concentrations. Although the pH of the rain sample of 14 November in this study was unfortunately not known, the volume weighted mean pH of the rain samples from 13-20 November, 2017 was 6.00 at Nagoya City Institute of Environmental Sciences, which is higher than the annual mean (4.99) of rain pH in 2017 (Nagoya City Institute of Environmental Sciences, 200 2018). In addition, because the duration of mist after the rain was unusually long (16 hr), ambient NH3 was dissolved more into the mist droplets as a reservoir. Subsequently, a large amount of NH3 was released from the mist droplets after evaporation, engendering a spike of the NH3 concentration. Figure 6 presents results of data analysis of NH3 concentrations. Daily and monthly NH3 concentrations (Fig. 6, top panel) show clear seasonal variation: high in summer (maximum in August) and low in winter (minimum in January). The monthly minimum (1.6 ppb) in January 2018 was almost equal to that (1.7 ppb) in January 2019, although the monthly maximum (7.0 ppb) in August 2018 was higher than that (4.9 ppb) in August 2019.

Emission from tree canopy around the site: relation to bird droppings 205
Furthermore, day-to-day variation was also greater in 2018 than in 2019. To examine the relation with source 210 factors, hourly NH3 concentrations were analyzed for two subjects: daily minimum and diurnal range (maximum minus minimum) under dry (RH <70%) and weak wind (<3 m s -1 ) conditions as fulfilled for both daily mean values. Reasons for the meteorological limitations were the following. Wet surfaces on building walls, litter, soil, and leaves can act as NH4 + reservoir, which might change ambient NH3 concentration shortly after evaporation.
To avoid this effect, the daily average of relative humidity was set to below 70% for extraction as "non-wet days". 215 As Figs. 2 and 3 show, the wind speed exhibited a strong effect on local source dilution. Therefore, a day of weak wind was selected to illustrate a stronger effect of the local source.
The daily minimum NH3 concentrations are shown together with the daily minimum air temperature (Fig. 6, middle panel). As briefly described earlier for Fig. 4, day-to-day variation of daily minimum NH3 concentration in December covaries with the baseline trend in daily minimum temperature. Analogous to this, the seasonal 220 variation of daily minimum NH3 concentrations follow closely with the seasonal variation of the daily minimum temperature: high in summer with larger variation, and low in winter with less variation during the month.
Monthly averages of daily minimum NH3 concentrations were higher in summer of 2018 (ca. 4 ppb) than those https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License. of 2019 (ca. 2.8), but almost identical values (0.7−1 ppb) were obtained for the respective winters. Daily minimum values of concentration and temperature were usually observed in the early morning before sunrise (Fig. 4). 225 Therefore, ambient NH3 observed under these circumstances (early morning under dry and weak wind) is regarded as derived from very local in origin because the ambient air must stay near the site for several hours.
Because of this, NH3 concentrations are equilibrated with local surfaces that can exchange NH3 bidirectionally, possibly surfaces of plants and soils. Stomata of plants do not open before sunrise. Therefore, stomatal gas exchange is expected to be negligible. Furthermore, the plant surface is less effective as the NH4 + reservoir 230 because of RH <70%. However, pore water or moisture in soil can remain. They might act as a bidirectional exchange source of NH3. For NH3 equilibrium between soil pore water and air, known as a compensation point, temperature, pH and NH4 + concentrations in the solution are important parameters aside from the atmospheric NH3 level (Farquhar et al., 1980). We discuss this point later in greater detail.
Another is analysis of the amplitude of diurnal variation as the difference between maximum and minimum 235 of the day (denoted as daily max−min; bottom panel of Fig. 6). As anticipated from the difference between summer and winter in daily NH3 variations portrayed in Figs Apoplastic fluid in stoma of plant and pore water in soil are assumed as the major reservoirs of NH4 + . To equilibrate apoplastic fluid with ambient atmosphere, stoma must be opened, which is regulated by plant 260 physiology relating to photosynthesis. Consequently, the daily maximum NH3 concentrations were observed at about 2−4 hr later from sunrise. In other words, the initial stage of stoma opening is synchronized well with the timing of the morning increase. The daily maximum NH3 concentrations are shown versus the average air temperature of the day (Fig. 7a). The leaf temperature was not measured in this study. The ambient temperature was used as a surrogate of the leaf temperature. In Fig. 7a, two hypothetical compensation curves are also shown 265 using  of 1500 and 200. Above an air temperature of about 10-15°C, most observed data were shown between these two curves, suggesting that the of the forest canopy around the site was in the range of 200-1500.
According to compilation by Zhang et al. (2010) and by Massad et al. (2010),  in the literature was several tenths to 10 5 depending on the richness of reactive nitrogen available for the plant, types of ground, and vegetation. For stomatal emission potential of NH3, the range of 300-3000 for trees of deciduous and evergreen forest was 270 proposed by Zhang et al. (2010). A similar range of the values was also listed by Massad et al. (2010).
Furthermore, daily minimum NH3 concentrations are also shown versus the minimum air temperature of the day (Fig. 7b). As described earlier, the condition observed for daily minimum NH3 concentration is connected to the emission potential from soil around the site because stomatal emissions are negligible. Although soil temperatures were not measured in this study, the minimum air temperature was used as a surrogate for nighttime 275 soil temperatures. In Fig. 7b, two hypothetical compensation curves are shown using  of 500 and 200. Observed data were in the range of the hypothetical compensation curves only for minimum air temperature above 20°C.
Below 20°C, most observed data were over the curve for  of 500. Two possibilities are considered for these relations. One is that higher  for soil is responsible for winter because litter from deciduous trees can be decomposed by microbial activity. Also, subsequent NH4 + production raised  higher than 500. Another 280 possibility is a contribution from vehicular emissions because of frequent stagnant air pollution on elemental carbon in winter in Nagoya (Yamagami et al., 2019).
As shown separately in Fig. 7, concentrations of NH3 in summer of 2018 were higher than those of 2019 for comparison with the same temperature.  for the canopy of a site varies with various parameters such as seasonal variation of plant's stage of growth and supply of reactive nitrogen (Schjoerring et al., 1998;Massad et al., 285 2010). Senescent and mature leaves have high potential for NH3 emissions (Mattsson and Scjoerring, 2003). For deciduous trees in this study, new leaves start to grow in April and mature after June. They turn red in November.
The duration of active leaf of deciduous trees roughly accords with the season of the higher daily max-min shown in Fig. 6c. However, the values of the daily max-min in summer differed between 2018 and 2019, with no great https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License. change of trees in the campus. Therefore, deposition of reactive nitrogen at the site might be the reason for the 290 difference. Wet deposition (rain) is the major input of reactive nitrogen. According to reports of rain composition in Nagoya City (http://www.city.nagoya.jp/shisei/category/53-5-22-8-1-2-0-0-0-0.html. Personal communication from Yamagami and Nakashima, 2019), monthly average wet depositions of NO3from May to September were 2.3 mmol m -2 in 2018 and 2.0 mmol m -2 in 2019, respectively. Similarly, monthly average wet depositions of NH4 + during May-September were, respectively, 2.9 mmol m -2 in 2018 and 2.4 mmol m -2 in 2019. Wet 295 depositions of these species during warm months were slightly higher (ca. 15%) in 2018 than those in 2019.
However, the observed differences (ca. 30%) in the average daily max-min between 2018 and 2019 were almost double for the difference of wet depositions, requiring more input difference.
To fill the gap of the yearly difference, the possibility of bird droppings at the site is discussed below. From June or July to September or October, rooftops of the buildings and trees in the campus are used frequently by 300 large numbers, actually hundreds, of crows for pre-roosting assembly or flight line assembly in early evening before going to roost, presumably located near the campus. Normally, a murder of crows stays a short time (mostly less than 2 hr). They then leave to their primary roost area (Nakamura, 2004). More crows gathered in the murder in the early evening in 2018 than in 2019 based on visual impressions of the crow density on the trees and building rooftops. Some crows unusually stayed overnight on the campus in the summer of 2018, but that behavior was 305 rarely observed in 2019.
Bird droppings are rich in reactive nitrogen: nitrogen contents in dry weight of droppings are 3.5% for chickens (Nakamura and Yuyama, 2005) and 4.7% for crows (Fujita and Koike, 2007). The major reactive nitrogen of bird droppings is the uric acid, which is readily transformed into NH4 + by microbial activity in the soil. It is later incorporated and used by plants through roots. To evaluate the effects of bird droppings at the site, 310 the flux of reactive nitrogen added by bird droppings over the unit area (Fluxbd, mol m -2 day -1 ) is estimated as shown below.

315
In that equation, Freq. (number m -2 day -1 ) represents the input frequency of excreta shot per day over unit area, W (g shot -1 ) stands for the dry weight of excreta per an excreta shot, R (%) denotes the nitrogen content per dry excreta weight, and 14 is the atomic weight of nitrogen for conversion. For simplicity, the following values are used to estimate Fluxbd: Freq is once per day per square meter, W is 1 g per shot, and R is 4.7% (Fujita and Koike, 2007). Evaluating the relevance to the assumptions is difficult, but it is believed to the best guess from the 320 dropping situations observed around the building (Appendix Photograph 1). The estimated result is 3.4 mol m -2 day -1 , which is converted as ca. 100 mmol m -2 month -1 . This value is nearly 40 times higher than the NH4 + flux https://doi.org/10.5194/acp-2020-244 Preprint. Discussion started: 14 April 2020 c Author(s) 2020. CC BY 4.0 License. by rain. Fluxbd includes large uncertainty depending on the crow density and their habitat in the campus. However, it is useful for comparison with reactive nitrogen flux by rain. Even assuming Freq was one-tenth of the initial assumption above (0.1 number m -2 day -1 ), Fluxbd is still larger than the NH4 + flux by rain. In this study, the dense 325 area of bird droppings was not so large in the campus. However, the excess inputs of reactive nitrogen brought by crows to a small area might engender strong local emissions of NH3 from the soil and through the forest canopy.
Indeed, Fujita and Koike (2007) pointed out that jungle crows brought substantial amounts of nutrients to their roost of fragmented forests in an urban area. Populations of crows and the distribution of crow roosts vary with food availability and trees for sleeping and breeding. Crows have adapted well to urban areas. Therefore, they are 330 ubiquitous. Their populations are often increasing (Ueta et al., 2003;Vuorisalo et al., 2003). Through their habits of roosting in scattered small forests in urban areas, reactive nitrogen supplied by crows might be oversaturated for tree growth and emitted from the canopy. Large and continuous NH3 sources are limited for recent urban areas, except for automobile exhaust. Therefore, emissions from the tree canopy reinforced by bird droppings have become more important for neutralizing acidic urban aerosol particles. 335

Summary and Conclusions
Hourly measurements of NH3 and NH4 + were conducted from November 2017 through October 2019 in Nagoya, central Japan. Monthly average NH3 concentrations were high (7.0 ppb and 4.9 ppb, respectively, for August in 2018 and 2019) in summer and low (1.6 ppb and 1.7 ppb for January 2018 and 2019, respectively) in winter. 340 During the study period, a surge event (11 ppb during 2 hr) was observed after mist evaporation during daytime, which was very rare at Nagoya, even though evaporation of mist or fog droplets are expected to be frequent after rain. A plausible condition of the surge event was discussed in terms of composition and pH of rain. The amplitude of diurnal variation of NH3 concentration (daily maximum minus minimum) was small (ca. 2 ppb) in winter and large (ca 20 ppb) in summer. The daily max-min increased from late spring synchronized with new leaf growth 345 and peaked in summer during intense addition of droppings from hundreds of crows assembled on trees and rooftops near the site before going to their roosts. The large diurnal variation of NH3 concentration was characterized by a peak at 2-4 hr after sunrise. The timing of seasonal and daily increases of the late morning NH3 peak imply that reactive nitrogen inputs from crow droppings and rain increased NH3 emissions from the tree canopy. Preliminary estimates suggest that reactive nitrogen input by crow droppings was greater than the 350 effect of wet deposition. Therefore, as populations of crows increase in some urban areas through adaptation, the reactive nitrogen supplied by crow droppings might become an increasingly important source of NH3 emissions in urban areas.