During summer 2004, NH3, size-resolved particulate NH4+, and associated characteristics of surface air were measured continuously on Appledore Island, off the southern Maine coast as part of the International Consortium for Atmospheric Research on Transport and Transformation (ICARTT). NH3 concentrations ranged from <0.6 to 123 nmol m−3 with maxima around local noon and minima near dawn. Particulate NH4+ ranged from 10.3 to 191 nmol m−3. The transport of emissions from intensive agricultural activities in the eastern United States was an important source of total NH3 (NH3 + NH4+) over the Gulf of Maine during summer. Under cleaner northwest flow, total NH3 concentrations were relatively low (median = 50.0 nmol m−3) and partitioned roughly equally between phases; under the more polluted midwest flow, total NH3 concentrations were substantially higher (median = 171 nmol m−3) but dominated by particulate NH4+. Because particulate NH4+ was associated primarily with the highly acidic sub-μm size fractions with low deposition velocities (median flux = 1.5 μmol m–2 day–1), dry-deposition fluxes were dominated by the gas phase (median = 6.2 μmol m−2 day−1). Consequently, phase partitioning with pollutant-derived sulfur aerosol substantially increased both the atmospheric lifetime of total NH3 against dry deposition and the relative importance of removal via wet- versus dry-deposition pathways. Total NH3 accounted for 32% of the dry-deposition flux of inorganic N to the Gulf of Maine during summer. The combined dry deposition of total NH3 and wet deposition of NH4+ via precipitation contributed 40% of the corresponding total atmospheric N flux.
 Globally, the anthropogenic production of reactive nitrogen (N) has increased by a factor of 10 over the last hundred years and is expected to increase by an additional factor of 2 over the next 50 years [Galloway et al., 2004]. Increased emissions have resulted in a greater atmospheric burden of fixed nitrogen and associated increases in atmospheric deposition to the northeastern United States (US) [Galloway et al., 1984]. Atmospheric deposition is one of the principal pathways by which nitrogen enters coastal waters along the eastern and southern US. Castro et al.  estimate that wet deposition of inorganic nitrogen and the dry deposition of total NO3 (gaseous HNO3 + particulate NO3−) from the atmosphere contributes from 7% to 60%, or about 20% on average, to the total nitrogen loads to 42 coastal bays and estuaries along the Atlantic and Gulf coasts of the eastern U.S. Because this analysis does not account for the dry deposition of gaseous NH3 and particulate NH4+, the estimated relative contributions of atmospheric inputs are considered lower limits. Surface waters in many coastal regions are nitrogen limited and, consequently, exogenous inputs of pollutant nitrogen can lead to excess phytoplankton production and associated degradation of ecosystems [Nixon, 1995; Paerl, 1985, 2002; Larsen et al., 2001; Paerl et al., 2002]. Deposition fluxes of atmospheric nitrogen are typically greatest during summer when surface coastal waters are often nitrogen limited and consequently sensitive to exogenous nitrogen inputs [Scudlark and Church, 1999; Walker et al., 2004].
 Ammonia is emitted from numerous natural and anthropogenic sources (Table 1). Natural sources include volatilization from the plant-soil system, which varies as functions of soil type, pH, and temperature. Forests account for 20% to 40% of the NH3 emissions in Maine and New Hampshire [Langford et al., 1992]. Ammonia is also emitted from animal excreta, decomposing vegetation, wildfires, and the surface ocean. Globally, natural sources of NH3 are relatively less important than anthropogenic sources [Galloway et al., 2004]. Biomass burning (both natural and anthropogenic) accounts for about 14% of the total global flux. The major anthropogenic sources of NH3 in eastern North America are agriculture (animal waste [60%] and chemical fertilizers [16%]), human breath and perspiration (7%), domestic animals (7%), sewage treatment plants and septic systems (6%), industrial point sources (2%), and mobile sources (2%) [Strader et al., 2004]. Since natural and anthropogenic sources of NH3 vary spatially over eastern North America and the western North Atlantic Ocean, relative contributions from different sources and associated atmospheric concentrations in coastal New England vary as a function of transport region [Parrish et al., 1993; Driscoll et al., 2003].
 On the basis of its thermodynamic properties (Henry's law and dissociation constants), NH3 partitions primarily with acidic sub-μm aerosol size fractions as a function of the pH and the volume of aerosol liquid water. Consequently, NH3 phase partitioning in coastal New England is also influenced by the relative amounts of highly acidic sulphate aerosol (a reaction product of upwind pollutant SO2 emissions) with which NH3 chemically interacts. Phase partitioning has potentially important implications for the atmospheric lifetime of total NH3; gaseous NH3 has a typical atmospheric lifetime against dry deposition of less than a day whereas sub-μm aerosols, with which most particulate NH4+ is associated, have lifetimes on the order of many days [Dentener and Crutzen, 1994]. Ammonia may also react with gaseous H2SO4 and H2O to form new particles [Korhonen et al., 1999; Kulmala et al., 2000].
 The sinks for total NH3 in air are wet, cloud-water, and dry deposition to the surface. Wet-deposition fluxes have been reliably quantified by numerous programs [e.g., Meyers et al., 2001]. In contrast, dry-deposition fluxes are associated with substantially greater uncertainties and relatively few estimates are available. Dry fluxes are typically modeled on the basis of measured concentrations of gaseous NH3 and size-resolved particulate NH4+, the nature of the receiving surface (ocean, forest, grass, etc.), and associated meteorological conditions (wind velocity, relative humidity, and temperature) [e.g., Hummelshøj et al., 1992; Valigura, 1995]. On average, the dry-deposition flux of atmospheric N to the mid-Atlantic coastal ocean during summer accounts for 43% of the total (wet + dry) deposition flux; NH3 corresponds to 60% of the dry N flux [Russell et al., 2003]. In contrast, Jordan and Talbot  estimated that dry deposition to the coastal Gulf of Maine contributes only 10% to 20% of the total deposition flux of atmospheric N. However, their analysis did not consider the direct deposition of NH3 vapor and, consequently, may have underestimated the relative contribution of dry deposition.
 During the summer of 2004, the chemical and physical evolution of polluted air along the New England coast was investigated from surface, aircraft, and ship platforms as part of the ICARTT campaign. On the basis of a subset of data generated under the auspices of Chemistry of Halogens at the Isles of Shoals (CHAiOS) ICARTT, this study evaluated the major processes that control the cycling of NH3 and particulate NH4+ over coastal New England and the associated deposition fluxes to the Gulf of Maine.
2.1. Site Description
 Between 6 July and 12 August 2004, air was sampled on Appledore Island, Isles of Shoals, approximately 10 km off the coast of New Hampshire and Maine, USA (42.90°N, 70.62°W, Figure 1). Sampling equipment was installed on top of a World War II-era surveillance tower 43 m above the ocean surface. Samples were processed in a laboratory about 150 m from the tower.
2.2. Measurement Techniques
 Size-resolved aerosols were sampled over discrete daytime (∼15 hours) and nighttime (∼9 hours) periods using modified (with the addition of a top “0” stage) Graseby Anderson model 235 high-volume (about 1 m3 min−1) cascade impactors configured with Lexan (frosted polycarbonate) substrates and quartz back-up (Pallflex 2500 QAT_UP) filters. Bulk aerosol was sampled in parallel on a 20.3-by-25.4-cm quartz filter at an average flow rate of 1.2 m3 min−1. Average 50% aerodynamic cut diameters for the impaction stages were 18, 9.9, 3.9, 2.2, 1.1, and 0.56 μm, which yielded average geometric mean diameters (GMDs) for the size-resolved samples of 25, 13, 6.2, 2.9, 1.5, 0.77, and 0.39 μm. Immediately after recovery, substrates and filters were halved in a class-100 clean bench, transferred to clean polyethylene bags, sealed in glass Mason jars, and frozen. After the campaign, samples were transported frozen to Mount Washington Observatory, extracted in 18 Ω deionized water (DIW) and analyzed by ion chromatography (IC) for NH4+, NO3−, SO42−, Cl−, Na+, and other major ionic constituents [Pszenny et al., 2004; Fischer et al., 2006]. Sea-salt (ss) and non-sea-salt (nss) constituents were differentiated using Na+ as a sea-salt referencing species [Keene et al., 1986].
 Gaseous NH3, HNO3, HCl, and other soluble reactive trace gases were sampled in parallel over two-hour intervals with tandem mist chambers containing DIW [e.g., Keene et al., 2004]. Sample air was drawn through a size-fractioning inlet that inertially removed super-μm aerosols; sub-μm aerosols were removed downstream with an inline Teflon filter [Keene et al., 1993]. Samples were analyzed by IC on site, usually within 12 hours of collection.
2.3. Meteorological Conditions and Surface Ocean Temperature
 Local air temperature and relative humidity (RH) were recorded at the Isles of Shoals, NH, meteorological station (IOSN3) operated by National Oceanic and Atmospheric Administration (NOAA). The station is located on White Island approximately 2.8 km S of the sampling site. Sea surface temperature was measured at two buoys (44013 and 44007) operated by NOAA's National Data Buoy Center that were moored 87 km S and 80 km NE, respectively, of the sampling site.
2.4. Regional Transport and Air Mass History
 Large-scale atmospheric transport was evaluated using both HYbrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) back t rajectories [Draxler and Rolph, 2005] and Lagrangian particle dispersion (FLEXPART) retroplumes [Stohl et al., 2005]. HYSPLIT trajectories were calculated every 6 hours (0, 6, 12, and 18 UTC) and archived by the Plymouth State Weather Center (http://pscwx.plymouth.edu/ICARTT/archive.html). FLEXPART was run in backward mode [Stohl et al., 2003; Seibert and Frank, 2004]. Every 3 hours, 40,000 particles were released from the location of the measurement site and followed backward in time for 20 days to form so-called retroplumes. The retroplume model output consists of emission sensitivities, which were plotted both as atmospheric column-integrated values to visualize the overall transport, and as footprint values. The footprint is the emission sensitivity in the lowest model output layer (0–150 m above the surface) since most of the emissions occur near the ground. Therefore the footprint indicates the potential for emission uptake for the sampled air mass. Transport to Appledore over the course of the experiment was categorizing into six regional flow sectors based on visual inspection of the HYSPLIT trajectories and the FLEXPART column-integrated and footprint emission sensitivity plots (Figure 2 [Fischer et al., 2006]).
 Because of the short sampling intervals (2 hours), each mist chamber sample was categorized by a single transport sector. The longer-duration aerosol samples were grouped by the predominant transport sector. If air was consistently transported from a single sector for greater than 75% of the sampling interval, then that sample was operationally categorized by that sector. This procedure represented a compromise between the limitation associated with the relatively long aerosol sampling intervals and the objective of interpreting transport regimes in terms of associated particulate NH4+ concentrations. This criterion allowed 44 of 60 aerosol samples to be classified into one of the six transport sectors. The longer-duration aerosol samples were associated with multiple paired mist chamber samples. To evaluate multiphase variability as a function of transport sector, the paired gas phase data were averaged over the period of each aerosol sampling interval; data for overlapping gas phase samples at the beginning and end of each aerosol sampling interval were weighted proportionately. Daily weather maps from the Department of Commerce (http://www.hpc.ncep.noaa.gov/dailywxmap/) were visually inspected to further interpret the role of air mass history and, specifically, the potential importance of scavenging and deposition via precipitation in fetch regions on the chemical characteristics of sampled air.
 Dry-deposition fluxes of size-resolved particulate NH4+ to the surface ocean were modeled on the basis of the measured NH4+ concentration and corresponding GMD for each size fraction, wind velocity, air temperate, and RH following Hummelshøj et al. . The laminar sublayer was assumed to be at 98% RH [Lewis and Schwartz, 2004] and the corresponding GMD for a given size fraction in the sublayer was based on the measured ionic composition and the hygroscopicity models mentioned above.
 Dry-deposition fluxes of gaseous NH3 to the surface ocean were modeled on the basis of the measured mixing ratios, wind speed, air temperature, and surface-seawater temperature following Valigura . This approach assumes no limit on gas solubility in the surface seawater. Although solubility limitations of this nature may be important for heavily polluted estuarine and nearshore coastal waters [e.g., Larsen et al., 2001], available evidence suggests that such limitations are not significant in offshore coastal waters along the eastern seaboard during summer [e.g., Russell et al., 2003]. Reported values of NH4+ concentrations for the surface of the Gulf of Maine during summer range from 0.01 μM to 0.07 μM [Townsend, 1998; Runge and Jones, 2003]. Assuming that these data are representative of surface waters surrounding Appledore, the corresponding calculated equilibrium gas phase concentrations based on an assumed pH of 8.0 [Butler, 1982] range from 9 to 63 pptv of NH3. Only 2% to 7% of the ambient gas phase concentrations during the ICARTT campaign were less than or equal to these equilibrium gas phase concentrations. In addition, dry-deposition fluxes of NH3 associated with the lowest 7% of ambient NH3 concentrations accounted for less than 3% of the dry-gaseous flux. On the basis of the above, we infer that solubility limitations on the NH3 dry-deposition fluxes were unimportant.
3. Results and Discussion
3.1. Summary of Measurements
 NH3 and NH4+ concentrations during this study varied over large ranges (Figure 3 and Table 2). Median concentrations of NH3 and NH4+ were similar to those measured at other locations in eastern New England but generally less than those at coastal locations in the mid-Atlantic region of the eastern U.S. (Table 3). These differences reflect the relatively closer proximity of the southern sites to intensive agricultural activities (commercial poultry and hog production, fertilizer applications, etc.) and associated NH3 emissions in the mid-Atlantic region [Russell et al., 2003].
Table 2. Summary Statistics for NH3 and NH4+ Measured During This Campaign
Maximum, nmol m−3
Minimum, nmol m−3
Average, nmol m−3
Median, nmol m−3
Number of Samples
NH4+ corresponds to the sum over all impactor stages.
Total NH3 refers to the subset of paired particulate NH4+ and gaseous NH3 data that were classified by source region (see section 2).
 When comparing data from different coastal or marine sites, it is important to recognize that a bulk sampling protocol such as that employed by the Clean Air Status and Trends Network (CASTNET) may not reliably quantify particulate phase concentrations for compounds with pH-dependent solubilities such as NH3. The pH of the bulk mix of super-μm sea-salt and sub-μm sulfur aerosols is typically greater than that of the sub-μm aerosol size fractions with which most NH4+ is associated in the atmosphere [Keene et al., 1990]. The higher pH results in supersaturation of bulk aerosol mixture relative to ambient NH3 vapor, subsequent volatilization of NH3, and associated negative artifacts in measured NH4+. Because of this effect, reported NH4+ concentrations that are based on analysis of aerosol sampled in bulk at coastal or open ocean locations should be considered lower limits.
 The time series data for NH3 indicate a fairly consistent pattern of higher concentrations during the day (median for all values = 419 pptv) and lower at night (median for all values = 225 pptv). To characterize relative diel variability, NH3 concentrations measured during each individual day (based on the midpoint of the sampling interval) were normalized to a common scale ranging from 0 to 1 by subtracting the minimum for the day and then dividing by the range for the day. The ranked values for every day were then binned into twelve 2-hour increments (e.g., 2300 to 0100; 0100 to 0300, etc.) and plotted (Figure 4). On the basis of median values, NH3 peaked around local noon and decreased to minimum values near dawn. Higher concentrations during the midday may be explained in part by the temperature dependence of NH3 emissions sources [Dentener and Crutzen, 1994]. However, the depth of the mixed layer over the upwind continent also varies over diel cycles in response to surface heating [Arya, 2001] and thereby contributes to variability in the composition of near-surface air sampled at Appledore. At night, total NH3 in the relatively shallow mixed layer is depleted via dry deposition to the surface. As the depth of the mixed layer increases in response to heating following sunrise, air from aloft that was not depleted by nocturnal deposition is entrained into the mixed layer, which contributes to rising concentrations in near-surface air during morning.
 Between 20 and 23 July, concentrations of NH3 remained consistently near or below the detection limit; virtually all total NH3 was in the form of NH4+ (Figure 3). These unusually low NH3 concentrations were associated with a major pollution episode as indicated by some of the highest concentrations of nss SO42− (Figure 3) and lowest aerosol pHs for the two smallest size fractions (GMD 0.77 μm fraction ranged from −0.68 to 0.72 and GMD 0.39 μm fraction ranged from −0.50 to 0.28) during the campaign. High aerosol acidity favors partitioning of NH3 with the particulate phase [Keene et al., 2004], which is consistent with these observations. In addition, the highest HNO3 mixing ratio (7546 pptv) [Fischer et al., 2006] and one of the highest HCl mixing ratios (3144 pptv) [Keene et al., 2007] measured during the campaign also occurred during this period. The influence of acidic pollutant aerosol on NH3 phase partitioned is addressed in more detail below.
Figure 5a depicts the distribution of size-resolved NH4+ during this campaign. The highest concentrations of both NH4+ and nss SO42− (Figure 5b) were associated with sub-μm size fractions. These size fractions also exhibited the highest aerosol acidities (Figure 5c). These relationships are consistent with those observed elsewhere (e.g., see citations in Table 3 and references therein) and reflect the pH dependence of NH3 phase partitioning (discussed in more detail below).
3.2. Chemical Composition Versus Air Mass Source Region
 On the basis of median values, the concentrations of particulate NH4+, total NH3, and nss SO42− varied substantially among source regions with highest values associated with the more polluted midwest and southwest transport regimes (Figure 6). The nonparametric Median Test was employed to assess the null hypothesis that the independent distributions for the subsets of total NH3 data corresponding to the six source regions are from populations with the same median. The null hypotheses for the midwest and the southwest urban sectors were rejected at 95% confidence. Two factors contribute to the differences in total NH3 between transport regions. First, regional emissions inventories indicate greater anthropogenic sources for NH3 (primarily agricultural activities including livestock farming and fertilizer application) in the midwest and southwest sectors relative to the other transport sectors [e.g., Bouwman et al., 1997]. Second, as discussed in more detail below, the higher concentrations of acidic S aerosol associated with air transported from the more polluted midwestern and southwestern sectors leads to the partitioning of larger fractions of total NH3 with sub-μm aerosols (Figure 7). Relative to gas phase NH3, sub-μm size fractions have lower dry-deposition velocities (Figure 8) and, thus, longer atmospheric lifetimes against deposition.
 To evaluate the importance of transport from the more polluted midwest and southwest sectors (N = 14) relative to the cleaner sectors (N = 30) as a source for total NH3 in coastal New England, the median values for the combined polluted (133 nmol m−3) and the combined clean sectors (52 nmol m−3) were weighed on the basis of the corresponding frequencies of transport from those sectors (0.32 and 0.68 respectively). These weighted contributions indicate that the relatively less frequent transport from the midwest and southwest sectors accounted for 55% of the median total NH3 concentration in near surface air during the experiment.
 During two periods (17 July and 31 July), flow from the southwest was associated with relatively low concentrations of total NH3 and nss SO42− (<40 nmol m−3 each, Figure 7a). Each of these sampling periods was preceded by widespread precipitation over the southwest source region that persisted for several days (see http://www.hpc.ncep.noaa.gov/dailywxmap). This evidence suggests that precipitation scavenged these air parcels during transport thereby accounting for the relatively lower concentrations measured at Appledore.
3.3. Phase Partitioning
 The phase partitioning of NH3 with deliquesced aerosol solutions is controlled primarily by the thermodynamic properties of the system expressed as follows:
where KH and Kb are the temperature-dependent Henry's Law (62 M atm−1) and dissociation constants (1.8 × 10−5 M), respectively, for NH3 and Kw is the ion product of water (1.0 × 10−14 M) [Chameides, 1984]. It is evident that for a given amount of total NH3 in the system, higher aqueous concentrations of particulate H+ will shift the partitioning of NH3 toward the condensed phase. Consequently, under the more polluted conditions characterized by higher concentrations of acidic S aerosol, the ratio of NH3 to NH4+ decreased (Figures 6 and 7). It also follows that for a multiphase system such as this in which aerosol acidity varies as a function of particle size, NH3 will partition preferentially with the more acidic sub-μm size fractions as is evident in our measurements (Figures 5 and 7). In addition, because virtually no NH3 partitions with super-μm sea-salt size fractions under these conditions (Figure 5), the chemical processing of sea salt and NH3 in polluted coastal air is largely decoupled. The large surface-to-volume ratios associated with the sub-μm aerosol size fractions facilitate rapid rates of equilibration (seconds to minutes) [Meng and Seinfeld, 1996] and, consequently, NH3 phase partitioning is near thermodynamic equilibrium under most conditions.
 Although concentrations of total NH3 were positively correlated with nss SO42− (Figure 7a), ratios of NH3 to NH4+ were negatively correlated (Figure 7c) because of the higher aerosol acidities (Figure 7b), greater aerosol volumes per unit volume of air (as indicated by nss SO42−), and associated influences on NH3 phase partition (relationship 1). Consequently, the concentrations of NH3 vapor were not correlated with nss SO42− (Figure 7d).
3.4. Dry Deposition
 The modeled dry-deposition velocities of gaseous NH3 to the ocean surface were substantially greater than those for the sub-μm aerosol size factions (Figure 8) with which most NH4+ was associated (Figure 5). Because the presence of high concentrations of acidic S aerosols in air transported from midwest and southwest shifted NH3 partitioning toward the longer-lived particulate phase (Figure 7), chemical interaction with acidic pollutant aerosols effectively increased the atmospheric lifetime of total NH3 against dry deposition. This shift has important consequences for NH3 cycling. Because gaseous NH3 dominates the dry-deposition flux of total NH3 (Figure 9), processes that influence NH3 phase partition (primarily interaction with pollutant nss SO42−) may be more important than the corresponding emission fluxes of NH3 in controlling the dry deposition of total NH3. For example, on the basis of median values for different transport sectors, relatively low concentrations of total NH3 associated with the northerly transport sector (56.3 nmol m−3, Figure 6) sustained the highest dry-deposition fluxes of total NH3 (10.7 μmol m−2 day–1, Figure 9). In contrast, the significantly higher (factor of 3) concentrations of total NH3 associated with the more polluted midwest transport section (Figure 6) sustained lower (30%) dry-deposition fluxes of total NH3 (Figure 9). The nonparametric Median Test was employed to assess the null hypothesis that the independent distributions for the subsets of total NH3 dry-deposition fluxes corresponding to the six source regions are from populations with the same median. The null hypotheses were accepted for all subsets at 95% confidence.
 Assuming steady state, a well mixed boundary layer over the upwind continent of 1000 m depth, and no precipitation, the median concentrations and dry-deposition fluxes of total NH3 discussed above correspond to atmospheric lifetimes against dry-deposition of ∼24 and ∼13 days, respectively, for air transported from the midwest and southwest sectors. The corresponding atmospheric lifetimes of total NH3 in air transported from the north, northwest, and south coastal regions were ∼5, ∼6, and ∼8 days, respectively. It is evident from the above that pollutant aerosol substantially increased the atmospheric lifetime of total NH3 against dry deposition. However, the actual atmospheric lifetimes of total NH3 may diverge substantially from those in the above example as a result of wet deposition, variability in the depth of the mixed layer, and transport of continental outflow above the marine boundary layer [e.g., Neuman et al., 2006].
 The relatively longer atmospheric lifetime of total NH3 against dry deposition under more polluted conditions implies that, on average, total NH3 would accumulate to higher atmospheric concentrations under these conditions and also be subject to atmospheric transport over longer distances. Consequently, the importance of removal via wet relative to dry deposition would also increase. It follows that the sporadic nature of precipitation would contribute to greater heterogeneity in NH3 deposition fields and potential ecological responses downwind of major S-emission regions.
3.5. Wet Versus Dry Deposition
 The relative average concentrations of individual inorganic N species and the corresponding dry-deposition fluxes are compared in Figures 10a and 10b. Each average flux corresponds to the integrated flux over all sampling intervals divided by the number of sampling days. Organic N was not quantified as part of this study, but available evidence suggests that it accounts for a relatively minor fraction of atmospheric nitrogen deposition in this region [Keene et al., 2002b]. Although particulate NH4+ accounted for almost half of the N associated with these inorganic species in ambient air, it contributed only 11% of the corresponding average dry-deposition flux. Dry fluxes were dominated by HNO3 (19.4 μmol m−2 day−1) followed by NO3− (10.0 μmol m−2 day−1), NH3 (9.3 μmol m−2 day−1), and particulate NH4+ (5.0 μmol m−2 day−1). The integrated average fluxes (Figure 10b) suggest that NH3 accounted for 66% of the dry-deposition flux of total NH3. In contrast, the median fluxes (Figure 9) suggest somewhat greater relative contributions of NH3 ranging from 72% to 91%. These differences reflect the relatively greater contributions from extreme values to the integrated dry flux of particulate NH4+ versus that of NH3. Similar relationships are evident in relative differences between median and average concentrations of NH4+ versus NH3 (Table 2). The average concentration for particulate NH4+ is 60% greater than its median concentration whereas the average concentration of NH3 is only 19% greater than its median. The median values provide useful statistics for characterizing and comparing the central tendencies of the populations. In terms of total N loadings, however, integrated fluxes that reflect contributions from extreme values are more relevant statistics.
 Wet-deposition fluxes of NH4+ and NO3− over the course of the experiment were quantified at Casco Bay, Maine (about 50 km N of the Appledore Island), by the National Atmospheric Deposition Program (http://nadp.sws.uiuc.edu/). Assuming that the averages for wet fluxes for NH4+ and NO3− at this site (31.5 and 55.7 μmol m−2 day−1, respectively) are representative of those at Appledore, the combined dry-deposition fluxes of NH3, HNO3 and particulate NH4+ and NO3− contributed 29% of the total (wet plus dry) deposition flux of inorganic N during the campaign (Figure 10c). Accounting for the contributions from the dry deposition of NH3 explains much of the difference between the greater percentage contribution of dry to total N deposition estimated during this investigation relative to that previously reported for the same region (10% to 20%) by Jordan and Talbot .
 In contrast to the results of this study (Figure 10c), dry deposition contributes a substantially higher percentage (43%) of the total deposition flux of atmospheric N at Lewes, Delaware, during summer [Russell et al., 2003]. The lower fraction of gaseous NH3 relative to particulate NH4+ at Appledore compared to Lewes (Table 3) contributed to these differences. The higher concentrations of pollutant S aerosol over coastal New England during summer lead to the partitioning of most total NH3 with the particulate phase, which increased its atmospheric lifetimes against dry deposition and thereby, increased the relative importance of removal via wet versus dry pathways.
 1. Because of the pH-dependence of NH3 phase partitioning, particulate NH4+ was associated almost exclusively with highly acidic sub-μm aerosol size fractions, which have long atmospheric lifetimes against deposition (many days) relative to NH3 vapor (less than one day). Because virtually no NH4+ partitions with the relatively less acidic sea-salt size fractions, the cycling of sea salt and NH3 in polluted coastal air are largely decoupled.
 2. The dry-deposition flux of total NH3 was dominated by gaseous NH3.
 3. Air transported from the cleaner northern and northwestern sectors typically contained relatively lower concentrations of total NH3 and particulate nss SO42− and H+. Consequently, total NH3 partitioned roughly equally between gaseous NH3 and particulate NH4+ and, thus, deposited relatively rapidly to the surface.
 4. Air transported from the southwest and midwest typically contained substantially higher concentrations of total NH3 and particulate nss SO42− and H+. Consequently, total NH3 partitioned primarily as particulate NH4+ and, thus, deposited relatively slowly to the surface.
 5. The shift in phase partitioning from NH3 to particulate NH4+ associated with the presence of pollutant S aerosol increased the atmospheric lifetime of total NH3, which we infer led to higher atmospheric concentrations, transport over longer distances from sources, an increase in the relative importance of removal via wet versus dry deposition pathways, greater heterogeneity in N deposition fields, and greater potential ecological influences.
 6. During this campaign, total NH3 accounted for 32% of the dry-deposition flux of inorganic N to the coastal Gulf of Maine. The combined dry deposition of total NH3 and wet deposition of NH4+ via precipitation contributed 40% of the corresponding total atmospheric N flux.
 We thank J. Moody and J. Galloway at the University of Virginia (UVA) and the anonymous reviewer for helpful comments on the manuscript. The Shoals Marine Laboratory provided outstanding logistical support, R. Talbot and the AIRMAP Program at the University of New Hampshire (UNH) shared facilities and data, Plymouth State University archived HYSPLIT back trajectories for the program, and all participants contributed to an intellectually stimulating and thoroughly enjoyable field campaign. Principal financial support was provided by the National Science Foundation through awards to UVA (ATM 0401628) and UNH (ATM 0401622). Additional support was provided by NOAA through contracts NA03OAR4600122 and NA04OAR4600154 to UNH. This paper is contribution 133 to the Shoals Marine Laboratory.