This is the author’s version of a work that was accepted for publication in Environmental pollution (Ed. Elsevier). Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in García- Gómez, Héctor, et al. “Atmospheric deposition of inorganic nitrogen in Spanish holm oak forests measured with ion-exchange resins and conventional collectors” in Environmental pollution, vol. 216 (Sep. 2016), p. 653-661. DOI 10.1016/j.envpol.2016.06.027 1 Atmospheric deposition of inorganic nitrogen in Spanish holm oak 2 forests measured with ion-exchange resins and conventional 3 collectors 4 Héctor García-Gomez1,6, Sheila Izquieta-Rojano2, Laura Aguillaume3, Ignacio González- 5 Fernández1, Fernando Valiño1, David Elustondo2, Jesús M. Santamaría2, Anna Àvila4, 6 Mark E. Fenn 5, Rocío Alonso1 7 1 Ecotoxicology of Air Pollution, CIEMAT, Av. Complutense 40, Ed.70, 28040 Madrid, Spain. 8 2 LICA, Universidad de Navarra, C. Irunlarrea 1, 31009 Pamplona, Spain. 9 3 CREAF, Campus de Bellaterra (UAB), Edifici C, 08193Cerdanyola del Vallès, Spain. 10 4 Universitat Autònoma de Barcelona (UAB), Campus de Bellaterra, 08193 Cerdanyola del Vallès, Spain. 11 5 USDA Forest Service, Riverside, CA, 92507, USA 12 6 Technical University of Madrid, School of Agriculture, Avd. Complutense s/n, 28040 Madrid, Spain. 13 1. Introduction 14 A continuous increase of nitrophilous species of plants, lichens and mosses in the 15 Iberian Peninsula (Ariño et al., 2011); an increase in the N content of bryophytes 16 during the last half of the 20th century (Peñuelas and Filella, 2001); an increase in 17 nitrate concentration in headwater streams (Àvila and Rodà, 2012; Camarero and 18 Aniz, 2010); and nutritional imbalances in forests of Abies pinsapo and Pinus 19 silvestris attributed to atmospheric N inputs (Blanes et al., 2013; Sardans et al., 2015). 20 Furthermore, interactions of N enrichment with ozone, the most important air pollutant 21 in the Mediterranean region, have been described for plant species in Spain (Calvete- 22 Sogo et al., 2014; Sanz et al., 2015). 23 A recent model-based assessment of N deposition threats to habitats within the 24 Spanish Natura 2000 network showed that the most threatened habitat-types were in 25 mountainous and alpine areas ( natural grasslands and heathlands in the Pyrenees and 26 Cantabrian Ranges, Pinus uncinata and Abies pinsapo forests), and in mountain 27 forests and scrublands close to high emission sources such as the big cities of Madrid 28 and Barcelona, and sclerophyllous forests of Quercus ilex in NE Spain (García-Gómez 29 et al., 2014). These high-altitude and orographically-complex areas are difficult to 1 30 access for monitoring purposes. Besides, current chemical transport models do not 31 adequately simulate small-scale variations in deposition regimes in these areas (Boutin 32 et al., 2015; García-Gómez et al., 2014; Simpson et al., 2006). Further monitoring 33 efforts in such areas would be useful for ground validation of transport models and for 34 ecosystem conservation assessments.. 35 Although the use of automatic wet-only samplers is widely recommended for 36 monitoring atmospheric wet deposition, the expansion of monitoring networks has 37 been urged by using less expensive methods, easy to operate, and without requiring 38 frequent visits to the field (Erisman et al., 2005; Clow et al., 2015). Throughfall or 39 precipitation collectors using ion exchange resins (IER) combine all these conditions. 40 They usually consist of a collector funnel connected to an IER-containing column 41 instead of a collection bottle. Inside the tube, the precipitation flows through the resin 42 during rain or snowmelt events and the ionic solutes are adsorbed to the resin. 43 Collectors using IER have been employed to monitor atmospheric N input to forest 44 ecosystems since the early 2000s (e.g. Fenn et al., 2002; Fenn et al., 2015). This 45 technique appears highly suitable for the Mediterranean region where complex 46 orography complicates field sampling. To our knowledge, there are not published 47 studies applying this technique in Mediterranean forests of Europe. In this study, three 48 monitoring sites in three Spanish forests of Q. ilex were equipped with conventional 49 and IER collectors to monitor bulk and throughfall deposition of inorganic N (nitrate 50 and ammonium) during two years. In addition, a review of the published works using 51 IER for collecting N deposition was performed and compared with our results. This 52 study was part of the EDEN project (Effects of nitrogen deposition in Mediterranean 53 evergreen holm oak forests), for which one of the objectives was to evaluate different 54 measurement methodologies for determining atmospheric N deposition. 55 56 2. Methodology 57 2.1. Study sites 58 Three holm-oak (Q. ilex) forests in Spain, influenced by different pollution sources, 59 soil and climatic conditions were selected for this study (Fig. 1). The Can Balasc (CB) 60 site is in a forest located within a natural protected area 4 km from Barcelona. This site 2 61 is characterized by acidic soils and a Mediterranean sub-humid climate. The Carrascal 62 site (CA) is located in an agricultural area close to Pamplona (15 km), with calcareous 63 soil and a Mediterranean humid climate. This site is the most agriculturally-influenced 64 among the three forests. The Tres Cantos site (TC) is a forest located in a natural 65 protected area 9 km from Madrid, growing on acidic sandy soil and with a 66 Mediterranean semi-arid climate. The vegetation at TC was historically managed as a 67 traditional dehesa (a savannah-like agrosilvopastoral system) of Q. ilex, but the low 68 management intensity during the last decades has allowed vegetation to grow as a 69 moderately open forest. Meteorological variables were monitored in the CB and TC 70 sites, and data from the closest meteorological station were collected for the CA site. 71 Further information on the monitoring sites can be found in García-Gómez et al. 72 (2015). 73 2.2. Sampling and analytical methodologies 74 Atmospheric deposition of ammonium (NH4+) and nitrate (NO3-) was monitored for 75 two years (spring 2011 – winter 2013) in open-field plots (bulk deposition) and below 76 the forest canopy (throughfall deposition) by means of IER collectors (IECs). These 77 collectors were manufactured following Fenn and Poth (2004), and consisted of a 78 funnel (10 cm radius; NILU - Norwegian Institute for Air Research, Kjeller, Norway) 79 attached to a PVC tube with an inner diameter of 15 mm filled with a mixed-bed IER 80 (Amberlite™ IRN150) and a valve at the bottom for drainage (Fig. A1). Generic 81 residue-free synthetic filter wool was inserted in both ends of the resin tube. Replicate 82 IECs were placed in the field with the collection surface at 1.5 m height during periods 83 of 3 to 5 months. For bulk deposition measurements, 4 collectors were deployed in TC 84 and CA, and 2 collectors in CB, in open-field locations; for throughfall deposition 85 measurements, 12 collectors were deployed in TC and CA, and 8 collectors in CB, 86 below the forest canopy. The entire collector devices were replaced in the field at the 87 end of each sampling period, and visually inspected for signs of bird fouling or other 88 potential contaminations of the sample. Once in the laboratory, the columns were pre- 89 rinsed with 100 ml of deionized water and the ions extracted by percolating 200 ml 90 aliquotes of 2M solution of KCl two times, representing two consecutive extractions. 91 Extracts were measured for pH as an additional quality control of this methodology 92 since liquid samples can lose NH4+ via NH3 volatilization as the pH increases (Vlek 3 93 and Stumpe, 1978). Background levels of NO3- and NH4+ in resin columns unexposed 94 to atmospheric deposition and stored at room temperature were also determined and 95 used as blanks. Finally, a set of IECs were spiked with known quantities of NO3- and 96 NH4+ in laboratory conditions and were extracted and analysed following the same 97 procedure to determine the adsorption and recovery efficiencies of the IECs for this 98 experiment (Table 1). 99 Available published studies using similar IECs (or at least applying similar laboratory 100 tests to them) were collected, and their methodological data (when available) were 101 recorded or estimated from tables or figures. Results from this methodological review 102 are presented in Table A-1. 103 Nitrogen deposition was additionally monitored by means of conventional collectors 104 using bottles instead of resin tubes (conventional bottle collectors; CBCs). These 105 CBCs were collocated with the IECs, applying the same replication, funnel type and 106 height. The CBC samples were collected on a weekly-basis by replacing the collection 107 bottle. During periods without rain, the CBC funnels were rinsed with 100 ml of 108 deionised water every two weeks before replacing the bottle, in order to collect and 109 measure the dry deposition into the funnel. The sampling procedure, storage, analysis 110 by ion chromatography (Dionex, Sunnyvale, USA) and quality control of the 111 analytical results of the CBC samples were performed following the recommendations 112 of the ICP Forests Manual (Clarke et al. 2010), and it is further described in Izquieta- 113 Rojano et al. (2016). 114 For both measurement methodologies, a filter mesh was fixed on top of the funnel 115 walls to prevent the accumulation of litterfall in the funnels and a bug-sieve (NILU, 116 Kjeller, Norway)was placed at the bottom the funnel to protect the resins from further 117 small particle contamination. The upper edges of the funnels were equipped with an 118 external metal ring to prevent birds from perching on the tube perimeter (Fig. A1). 119 2.3. Calculations and statistical analysis 120 Nitrogen deposition in each IEC was calculated multiplying the concentration of 121 nitrate-N (NO3-N) and ammonium-N (NH4-N) by the volume of the extracting 122 solution and adding the results from the two extractions. Then the background levels 123 measured in the unexposed blank IER tubes were subtracted, and the result divided by 4 124 the collection surface. Finally, these results were corrected by the recovery efficiency 125 factor. For the CBCs, the deposition was calculated for each collection bottle by 126 multiplying the sample concentration by the volume collected (or the rinsing volume 127 for the rainless periods). Deposition values from the replicated samples were then 128 averaged for each period and plot for both methodologies. Deposition of total 129 inorganic nitrogen (TIN) was calculated by adding NO3-N and NH4-N deposition 130 values. Annual deposition values were calculated as the sum of the sampling period 131 values during the year. When the sum of periods did not exactly match with the 132 duration of a natural year, values were weighted by their added-up sampling time. 133 Deposition values (means and standard deviations) of the CBC periods were combined 134 to match the IEC periods for comparison purposes, and also grouped and combined on 135 a monthly basis to summarize and describe the intra-annual variability of the N 136 deposition. Net throughfall deposition was calculated subtracting bulk deposition 137 values from throughfall deposition values for every period and methodology. 138 To compare the two methods, the Pearson correlation coefficient (r) was calculated for 139 NO3-N, NH4-N and TIN deposition data from both methods. Correlation between 140 variables was tested using Statistica version 12 (StatSoft, Inc. Tule, OK, USA) with 141 the Pearson correlation coefficient, and with Spearman rank order correlation when the 142 data were not normal. Alpha level was set at 0.05. Additionally, two metrics 143 commonly used in model evaluation (García-Gómez et al., 2014; Boylan and Russell, 144 2006), the mean fractional bias (MFB) and root mean square error (RMSE), were 145 calculated following eq.1 and eq. 2, respectively. 1 𝐼𝐸𝐶𝑖 −𝐶𝐵𝐶𝑖 146 eq. 1 𝑀𝐹𝐵 = ∑( 𝐼𝐸𝐶𝑖 +𝐶𝐵𝐶𝑖 ) 𝑁 2 1 1 2 147 eq. 2 𝑅𝑀𝑆𝐸 = [ ∑(𝐼𝐸𝐶𝑖 − 𝐶𝐵𝐶𝑖 )2] 𝑁 148 ,where N is the number of data pairs from the IEC and CBC methods, and the index i 149 is over the time series, and including all the monitoring sites for each method. 150 Variability of deposition values among collectors (referred herein as precision of the 151 method) was calculated as the coefficient of variation (CV = standard deviation / 152 mean) for every plot and period with two or more sampling data available. 5 153 3. Results and Discussion 154 3.1. Laboratory testing of ion-exchange resin collectors 155 The values of nitrate obtained in blank unexposed IECs were 0.001 - 0.006 mg NO3-N 156 per gram of resin. These values were similar to previously reported ones (Table 1). In 157 the case of ammonium, mean values of 0.048; 0.060 and 0.074 mg NH4-N per gram of 158 resin were obtained. This blank correction was higher than those shown in Table 1, 159 which were 0.001 mg NH4-N g-1, except for Boutin et al. (2015) who nonetheless 160 reported a blank value aat least 4 times lower.. Resins made of quaternary ammonium 161 compounds, like the one use in this study, can release NH4+ (Hansen 2012; Langlois et 162 al., 2003), which could explain the relatively high NH4+ found in the blanks. Although 163 having such high blank values is not expected to cause accuracy problems, it could 164 contribute to a decrease in the precision of the measurement of low deposition values. 165 The adsorption efficiency of the IER tubes was close to 100% for both ions, the same 166 as in most of the reviewed experiments from the literature (Table 1). The recovery 167 efficiency for NO3- was comparable to that reported in other studies, but the recovery 168 efficiency was higher for NH4+. In this case, more NH4+ was recovered from the 169 spiked resins than the quantity previously added, giving a recovery factor of 112%. 170 Recovery factors above 100 % have been previously described using the same IER 171 (Fenn et al., 2002; Table 1). These results highlight the importance of lab tests to 172 explore the performance of the resin used and to provide with suitable correction 173 factors. 174 3.2. Comparison of methods 175 The comparison metrics and graphics were performed only for bulk deposition values, 176 since throughfall deposition is subjected to a higher variability due to canopy 177 interactions with atmospheric N, canopy heterogeneity, and biochemical 178 transformations provoked by litterfall or algae in the collectors (Bleeker et al., 2003; 179 Fenn and Poth, 2004). 180 In our study, a considerable number of the IEC extracts from CA exhibited a high pH 181 (mean of 4.9, maximum of 13.1), while for the other two sites a mean pH of 3.6 and a 182 maximum value of 6.3 were found. Significant negative correlations were found 183 between the pH of the extracts and NH4+ concentrations within some periods at CA 6 184 (data not shown) and a remarkable reduction in NH4+ concentration occurred generally 185 at pH values higher than 5.5. Because of this, ammonium values from extracts with a 186 pH value higher than 5 (34% of 246 extracted samples) were removed from the dataset 187 before deposition calculations, which reduced the replication of measurements (Table 188 2). The precipitation in CA was much more alkaline (with a mean pH of 7.3 and a 189 mean alkalinity of 1411 µeq l-1) than in CB or TC (with mean pH of 6.3 and 6.0; and 190 mean alkalinity of 67.6 and 27.5 µeq l-1, respectively) during the study period 191 (Aguillaume, 2015). It may therefore be that carbonates and bicarbonates present in 192 the rain at this site increased the pH of the extracts provoking the loss of NH4+ via 193 volatilization of NH3 before the analysis (Cape et al., 2012; Izquieta-Rojano et al., 194 2016). 195 Variability among IEC samplers for bulk deposition was lower in TC (mean CV of 196 12% and 11% for NO3-N and NH4-N, respectively) than in the other two sites: mean 197 CV of 22% and 35% for NO3-N and 79% and 22% for NH4-N in CB and CA, 198 respectively. These high values are similar to the highest reported ones in Table 1 and 199 they may be caused by a low replication in these sites. In CB, only two IECs were 200 installed in the open, while in CA the above-mentioned reduction in the dataset 201 provoked a drastic reduction in the actual replication. In the present study, the intra- 202 plot variability of the IEC methodology for bulk deposition measurements was in a 203 similar range to the set of previous experiments shown in Table 1. The variability 204 found with IECs was similar to that found using CBCs, with the exception of NH4-N 205 in CB (79%), which was largely higher (29% using CBCs). It might indicate that a 206 higher replication than two funnels is needed in this type of forests when IER are used 207 to measure NH4-N deposition. Moreover, the higher replication allows detecting and 208 removing questionable data, as was done for the CA dataset. 209 Comparison of IEC and CBC measurements of N bulk deposition showed a good 210 agreement for NO3-N and acceptable for NH4-N and TIN (Fig. 2). Nitrate deposition 211 estimations showed lower error (RMSE = 0.15 kg N ha-1) and a better fit to the 1:1 line 212 than NH4-N measurements. Ammonium measurements with IECs tended to 213 overestimate the lowest range and underestimate the highest ones, in relation with the 214 CBC values. The bias for NH4-N deposition was low and positive (MFB = 6% for the 215 entire dataset) and the error was about 0.56 kg N ha-1. In consequence, TIN 7 216 measurements showed an acceptable accuracy between methods (Fig. 2) but a lower 217 precision (RMSE = 0.63). This comparison was performed using the CBC results as 218 the reference for assessing the IEC results, but it must be taken into account that the 219 former methodology is not free of uncertainty (Bleeker et al. 2003; Dämmgen et al., 220 2005). Moreover, there are also differences in the quality control of the data, since a 221 thorough validation of the analytical results was not possible for IEC data (because 222 only NO3- and NH4+ were analyzed in these samples). 223 Fig. 2 shows a very good performance of IEC method for NO3-N measurements, with 224 only one period slightly out of the ±50% lines. In the case of NH4-N deposition, the 225 values with the highest deviation were from two sampling periods during the second 226 year in CA, with exceptionally low values in the IECs (2.12 and 1.54 kg NH4-N ha-1) 227 compared to the CBC values (3.15 and 3.29 kg NH4-N ha-1). These two periods 228 showed the highest mean hourly precipitation of the entire study (0.11 and 0.17 mm h- 1 229 ) and an elevated maxima hourly precipitation (9.9 and 18.2 mm h-1). The lower 230 deposition values could be therefore related to collection problems during those heavy 231 rains To protect the resins IECs have additional filters and stoppers that may impede 232 the rainwater flux during very intense storms and rainwater may temporarily 233 accumulate in the funnel, exposing a relatively large surface of sample to the 234 atmosphere. This could provoke a loss of NH4+ via volatilization of NH3, particularly 235 from rain samples with high alkalinity and pH like those found in CA. Hourly 236 averaged rainfall showed a slight positive correlation with the differences found for 237 NH4+ concentration at CA (r = -0.48; p = 0.059), which totally disappeared once the 238 two above-mentioned periods from CA were removed, suggesting that only during 239 these two periods the measurements were affected by heavy rains. When excluding 240 these two sampling periods at CA, the deposition of NH4-N estimated with the IEC 241 method was generally higher (59%) than with the CBC method. 242 Previous studies have reported that NH4+ in bulk deposition tends to be greater in IECs 243 than in CBCs (Clow, et al, 2015; Fenn and Poth, 2004; Hansen, 2012; Langlois et. al., 244 2003). There are three processes that could account for this discrepancy: 1) release of 245 NH4+ from the amine groups of the IER, 2) nitrification in the CBC samped solution 246 and 3) volatilization losses of NH3 in the CBC liquid samples (Fenn and Poth, 2004). 247 The high temperatures that are common in the Mediterranean climate could favor any 8 248 of the three potential processes (Fenn and Poth, 2004) and, in fact, there was a positive 249 relationship between the mean temperature of the sampling periods and positive 250 discrepancies of IEC over CBC for NH4-N deposition (n = 11; r = 0.66). 251 Field blanks are recommended to correct for ammonium release from the IER (Fenn 252 and Poth, 2004). However, we used laboratory blanks (i.e. IER tubes not exposed to 253 field condition), which may release less amount of NH4+ than field blanks because 254 they are not submitted to the stronger temperature oscillations experienced in the field 255 that may enhance NH4 release from resins. This could account for part of the 256 overestimation of NH4+ deposition of IECs at low concentrations.. After the study 257 period, two sealed IECs (field control) were deployed in TC during 6 months, one 258 under the canopy and the other in the open. The blank NH4+ value from the field 259 control in the open-field plot was the only one showing a clear difference with the 260 laboratory control for the same period, being 29% higher which corresponds to an 261 overestimation of 0.23 kg NH4-N ha-1 for bulk deposition for that period and site. 262 Regarding the other two processes, nitrification is not expected to occur in the open- 263 field CBCs, since the bottles are protected from solar radiation by PVC tubes , but this 264 possibility cannot be totally discarded. Secondly, the volatilization of NH3 from liquid 265 samples in open field is highly possible in this warm climate, although the narrow 266 passage through the bug-sieve in the funnel neck is expected to meaningfully reduce 267 the rate of ammonia volatilization, since this process is known to be severely restricted 268 by limiting the movement of air above the water (Vlek and Stumpe, 1978). 269 3.3. Annual nitrogen deposition to holm oak forests 270 Mean annual N deposition estimated with IECs and CBCs is presented in Fig. 3. TIN 271 bulk deposition ranged from 2.42–3.85 and 3.09–5.41 kg N ha-1 among the sites 272 according to CBC and IEC methodologies respectively. TIN in throughfall ranged 273 from 2.33–8.20 and 4.52–8.91 kg N ha-1 among the sites with CBC and IEC 274 respectively. The highest N bulk deposition occurred in CA, with an annual average of 275 2.69 kg NO3-N ha-1 and 4.15 kg NH4-N ha-1 according to the CBC method and 2.60 kg 276 NO3-N ha-1 and 2.81 kg NH4-N ha-1according to IECs. Throughfall deposition, on the 277 other hand, was the highest in CB, with an annual average of 5.83 kg NO3-N ha-1 and 9 278 2.36 kg NH4-N ha-1 according to CBC method and 5.50 kg NO3-N ha-1 and 3.41 kg 279 NH4-N ha-1, according to IECs. 280 The IEC method estimated on average a bulk deposition 0.20 kg NO3-N ha-1 y-1 and 281 0.17 kg NH4-N ha-1 y-1 lower than the CBC method, representing a MFB of -9% and - 282 5%, respectively. The comparison of annual values for throughfall showed a marked 283 difference for NH4+ in TC, where the IEC method estimated 2.25 kg NH4-N ha-1 y-1 284 more than the CBC method. This discrepancy could be related to the high uncertainty 285 of the CBC method when concentrations are close to or below detection limits 286 (Köhler, et al., 2012), as was here the case with 30% of NH4+ throughfall samples 287 below the detection limit. Besides, the open structure of this dehesa forest enhances 288 the heating of the collectors through direct radiation and, therefore, biological 289 processes in the liquid samples (nitrification) or in the accumulated debris at the 290 bottom of the IEC funnels (decomposition). Both possibilities are equally possible, 291 although collectors are shaded by PVC tubes and no correlation was found between 292 mean temperature and the discrepancies between methods for throughfall deposition at 293 this site. Despite bird dropping cannot be totally disregarded, the largest discrepancy 294 was found in spring of 2011, the period in which more inflorescences were collected 295 in litterfall samplers (55.2 g m-2). Inflorescences of Q. ilex are the smallest and most 296 easily decomposable part of the litterfall with a high N content (close to 1.5 %; Bellot 297 et al., 1999). Thus, inflorescence can mean a high return of N to the soil in the years 298 with high production of flowers (Escarré et al., 1999), such as 2011. It is therefore 299 recommended to clean or change the funnels when this situation occurs. On the other 300 hand, some of the CBC samples at TC which stood uncollected in field for 2-3 days 301 after summer rains showed algal growth in the below-canopy plot but not in the open- 302 field one. The use of biocide in the collection bottles (faltan 2 refs) is therefore 303 recommended under Mediterranean conditions, particularly under semi-arid regimes 304 like at TC site and in throughfall deposition collectors. Besides, further quality control 305 can be included by analyzing phosphates in the IEC samples to identify bird-fouling 306 contamination (Fenn et al., 2015). All these results show that, IECs seem a suitable 307 method for monitoring annual deposition in this type of forest with lower cost and 308 effort than CBC methods, once precautions mentioned above are considered. 10 309 The difference between throughfall and bulk deposition (net throughfall) reflects the 310 interaction between the atmosphere and the canopy including both the wash-off of dry 311 deposited N and the exchange with the leaf surfaces, which in the case of N iscan 312 have an important adsorption component (Lindberg et al., 1986). Negative values of 313 net throughfall indicate that at least part of the wet deposited N is being effectively 314 retained in the canopy, while positive values indicate that at least part of the dry 315 deposited N is reaching the forest floor. Annual net throughfall values were positive 316 for NO3-N with the two measurement methods at all sites (Fig. 3). This result indicates 317 a net flux of oxidized N to the forest soil at all sites coming from dry deposition, even 318 though some canopy nitrate retention could be occurring. This input of dry deposited 319 NO3-N is very high at CB, the most urban-influenced site wich experiences the highest 320 concentrations of gaseous oxidized-N compounds (García-Gómez et al., 2015). On the 321 other hand, net throughfall values for NH4-N were only clearly positive in CB and 322 negative in CA, while in TC depended on the method used and could be considered 323 close to zero. The forest canopy at CB seems to be retaining less NH4+ than CA, even 324 though dry deposition of reduced N is expected to be lower at this site than in CA 325 which is a highly agriculture-influenced site with higher NH3 concentrations (García- 326 Gómez et al., 2015). Canopy uptake of NH4+ has been described to be larger at lower 327 nitrogen foliar content (de Vries et al., 2001), which is coincident with the 328 stoichiometry of these sites, since CA showed the lowest leaf N content (1.44%) and 329 CB the highest one (1.61%). On the other hand, the apparent retention of N by the 330 canopy of Q. ilex forest needs further investigation, particularly for NH4+. The fact that 331 the forest canopy at CB seems to retain less deposited N than the other two sites, 332 together with the observation of lower retention of atmospheric gaseous N (García- 333 Gómez et al., 2015) could indicate that the Q. ilex trees in this forest cannot retain as 334 much N as the other sites. Interestingly, Q. ilex forests close to this experimental site 335 and with similar deposition load have experienced rises of NO3- concentration in 336 headwater streams considered as a sign of the onset of eutrophication (Àvila and Rodà, 337 2012). 338 Seasonality 339 In Mediterranean areas, the seasonal variability of N deposition is expected to be high 340 because of the particularities of the Mediterranean precipitation regime, in which 11 341 rainfall is not equally distributed over the year. IEC data showed seasonal variations 342 (Table 2), such as higher deposition in spring and autumn (the wettest periods), and a 343 lower deposition during the rest of the year. Data from CBCs can be better used for 344 studying intra-annual variability in detail because deposition can be presented 345 monthly; Fig. 4 and A2). Negative values for net throughfall were most commonly 346 observed with NH4+ (Fig. 4), indicating that canopy retention occurred more often in 347 the reduced form, as usually found in the literature. Relatively high positive values of 348 net throughfall can be noticed in TC during September and October in 2011 and 2012, 349 during the first rains after the summer droughts occurred, and also in March 2012, 350 after a particularly dry winter (Fig. 4). These values of positive net throughfall imply 351 relatively large and ephemeral pulses of N into the soil after the first rainfall events, as 352 those previously described in other semi-arid Mediterranean ecosystems (e.g. Meixner 353 and Fenn, 2004; Vourlitis et al., 2009). These pulses can be noticed also in CB, but 354 they rarely occurred in the CA site (Fig. A2). These ephemeral inputs, among others 355 effects, can trigger pulses of NO emissions from soil (Homyak and Sickman; 2004) 356 which can affect the atmospheric photochemistry of the forest, involved in ozone (O3) 357 and nitric acid formation (HNO3). Besides, they can provoke a flushing of inorganic N 358 to groundwater if the pulse occurs when plants and soil communities are not able to 359 use this dissolved N (like at the end of the summer, when they are withstanding 360 drought stress). This effect, known as the asynchrony hypothesis (Meixner and Fenn, 361 2004), is corroborated in TC, where high concentrations of NO3- in the soil water (up 362 to 2.01 and 0.35 mg NO3- l-1 at 20 and 40 cm depth, respectively) have been found 363 after pulses of N during late-summer and early-autumn, but not during early spring, 364 when understory annual pastures were emerged and growing and soil communities 365 were active. 366 3.4. Summary of methodological considerations 367 Some methodological recommendations on the IEC method arising from the present 368 study (some of them have been already reported in previous works) must be taken into 369 consideration. Preliminary laboratory tests on adsorption and recovery efficiency need 370 to be done in order to know the performance of the resin. The IEC method poses a 371 potential overestimation of NH4+ deposition due to the release of ammonium from the 12 372 amine groups of the IER. To avoid that, field blanks consisting of sealed IECs 373 deployed in the field are recommended over unexposed laboratory blanks in 374 Mediterranean conditions. An overestimation of NH4+ deposition could also occur 375 when a large amount of leachable debris is accumulated during the long sampling 376 periods of IECs. In Q. ilex forests it could particularly occur during the flowering 377 period and/or in open forests, like dehesas, because the structure of the canopy allows 378 high radiation exposure and heating of the collectors. Periodic cleaning of sampling 379 devices is advised during those sensitive periods. On the other hand, NH4+ deposition 380 could be underestimated when rain flow across the different parts of the IECs during 381 heavy rains is made difficult (e.g. by small nominative diameter of any part of the 382 collector, or very dense filters or sieves). A modification in the design could be studied 383 in those areas withstanding intermittent heavy rains. It is recommended to deploy at 384 least three replicate IECs at every bulk deposition sampling plot, so questionable 385 samples can be removed if necessary. There would not be a recommended replication 386 for throughfall measurements, since it may vary depending on the vegetation cover 387 type, canopy characteristics and its distribution throughout the stand; however, based 388 on the intra-site variability found in this study, at least ten replicates are recommended 389 in this kind of forest stands. Finally, we also recommend to measure the pH of the 390 sample IEC extracts, and to acidify the extract if the pH is too high. Regarding the 391 CBC method, it has proven useful for studying the temporal variability of N 392 deposition, and the use of biocides is recommended in these climatic conditions and 393 especially for throughfall measurements. An additional quality control method for IEC 394 by analysing phosphates in the extracts is also recommended. 395 4. Conclusions 396 The results of the present study show that collection methods for N deposition based 397 on ion-exchange resins can be recommended for long-term studies in the 398 Mediterranean region, since its measuring good performance was found for NO3- 399 deposition and an acceptable for NH4+ and TIN, in comparison with conventional 400 methods. This methodology is particularly recommended in remote areas and when the 401 nitrogen concentration in rain is low. However, throughfall measurements of 402 ammonium in low density forests (like dehesas) or any measurement with alkaline rain 13 403 should be considered with caution. Besides, possible contamination from the 404 accumulation of debris in the funnels must be contemplated when planning the 405 duration of the different sampling periods along the year. Absorption and recovery 406 efficiencies from the IER need to be tested in the lab to derive possible correction 407 factors needed. 408 Mean annual bulk deposition of TIN in Holm oak forests ranged from 2.42–3.85 and 409 3.09–5.41 kg N ha-1 according to CBC and IEC methodologies, respectively. 410 Analogous throughfall deposition fluxes of TIN ranged from 2.33–8.20 and 4.52–8.91 411 kg N ha-1. On average, bulk deposition estimated by the IEC method was 0.20 kg NO3- 412 N ha-1 y-1 and 0.17 kg NH4-N ha-1 y-1 lower than the CBC method. Intra-annual 413 variability studied with the CBC data showed significant input pulses of N into the 414 forest soil after dry periods. These pulses presumably originated from the washing of 415 dry deposition accumulated in the canopy and were particularly noticeable in the forest 416 site with a semiarid climate. The implication of these nutrient pulses for ecosystem 417 functioning, atmospheric chemistry and N leaching should be further investigated. 418 Acknowledgements 419 This research was funded by the Spanish Ministry of Science and Innovation (CGL2009- 420 13188-C03-02), by the Autonomous Government of Madrid AGRISOST-CM (P2013/ABI- 421 2717) and by the European Projects ECLAIRE (FP7-ENV-2011/282910) and Life RESPIRA 422 (LIFE13 ENV/ES/000417). This study was also supported by the Ministry of Agriculture, 423 Food and Environment (Resolución 15398, BOE no. 230). 424 425 426 References 427 Aguillaume, L., 2015. Nitrogen deposition at Mediterranean holm oak forests: loads 428 and indicators. PhD. Thesis. CREAF - Universitat Autònoma de Barcelona (UAB), 429 Barcelona, Spain. 430 Alonso, R., Elvira, S., González-Fernández, I., Calvete, H., García-Gómez, H., 431 Bermejo, V., 2014. Drought stress does not protect Quercus ilex L. from ozone 432 effects: Results from a comparative study of two subspecies differing in ozone 433 sensitivity. Plant Biol. 16, 375–384. doi:10.1111/plb.12073 434 Ariño, A.H., Gimeno, B.S., Pérez de Zabalza, A., Ibáñez, R., Ederra, A., Santamaría, 435 J.M., 2011. Influence of nitrogen deposition on plant biodiversity at Natura 2000 436 sites in Spain. In W. K. Hicks et al. (Eds.), Nitrogen deposition and Natura 2000. 437 Science & practice in determining environmental impacts. 438 COST729/Nine/ESF/CCW/JNCC/SEI Workshop Proceedings COST (pp. 140– 439 146). Brussels: COST office. 14 440 Àvila, A., Rodà, F., 2012. Changes in atmospheric deposition and streamwater 441 chemistry over 25 years in undisturbed catchments in a Mediterranean mountain 442 environment. Sci. Total Environ. 434, 18–27. doi:10.1016/j.scitotenv.2011.11.062 Formatat: Espanyol (Espanya) 443 Bellot, J., Sánchez, J.R., Lledó, M.J., Martínez, P., Escarré, a., 1992. Litterfall as a 444 measure of primary production in Mediterranean holm-oak forest. Vegetatio 99- 445 100, 69–76. doi:10.1007/BF00118211 446 Blanes, M.C., Viñegla, B., Merino, J., Carreira, J. a., 2013. Nutritional status of Abies 447 pinsapo forests along a nitrogen deposition gradient: Do C/N/P stoichiometric shifts 448 modify photosynthetic nutrient use efficiency? Oecologia 171, 797–808. 449 doi:10.1007/s00442-012-2454-1 450 Bleeker, A., Draaijers, G., van der Veen, D., Erisman, J.W., Möls, H., Fonteijn, P., 451 Geusebroek, M., 2003. Field intercomparison of throughfall measurements 452 performed within the framework of the Pan European intensive monitoring program 453 of EU/ICP Forest. Environ. Pollut. 125, 123–138. doi:10.1016/S0269- 454 7491(03)00142-8 455 Boutin, M., Lamaze, T., Couvidat, F., Pornon, A., 2015. Subalpine Pyrenees received 456 higher nitrogen deposition than predicted by EMEP and CHIMERE chemistry- 457 transport models. Sci. Rep. 5, 12942. doi:10.1038/srep12942 458 Boylan, J.W., Russell, A.G., 2006. PM and light extinction model performance 459 metrics, goals, and criteria for three-dimensional air quality models. Atmos. 460 Environ. 40, 4946–4959. 461 Brumbaugh, W., Earhart, S., Linder, G., Neitlich, P., 2012. Determination of annual 462 loads of inorganic atmospheric contaminants in northern Alaska by measurement in 463 snowpack and passive ion collectors, in: 33rd SETAC Conference. Long Beach. Formatat: Espanyol (Espanya) 464 Calvete-Sogo, H., Elvira, S., Sanz, J., González-Fernández, I., García-Gómez, H., 465 Sánchez-Martín, L., Alonso, R., Bermejo-Bermejo, V., 2014. Current ozone levels 466 threaten gross primary production and yield of Mediterranean annual pastures and 467 nitrogen modulates the response. Atmos. Environ. 95, 197–206. 468 doi:10.1016/j.atmosenv.2014.05.073 469 Camarero, L., Aniz, M., 2010. El sistema de seguimiento de las aguas en el nodo 470 LTER-Aigüestortes : tendencias e indicadores de los impactos de la deposición 471 atmosférica de N y S. Ecosistemas 19, 24–41. 472 Cape, J.N., Sheppard, L.J., Crossley, a., van Dijk, N., Tang, Y.S., 2010. Experimental Formatat: anglès (EUA) 473 field estimation of organic nitrogen formation in tree canopies. Environ. Pollut. 474 158, 2926–2933. doi:10.1016/j.envpol.2010.06.002 475 Cape, J.N., Tang, Y.S., González-Beníez, J.M., Mitošinková, M., Makkonen, 476 U., Jocher, M., Stolk, a., 2012. Organic nitrogen in precipitation across Europe. 477 Biogeosciences 9, 4401–4409. doi:10.5194/bg-9-4401-2012 478 Ceron, R., Ceron, J., Aguilar, C., Montalvo, C., Rustrian, E., Houbron, E., Ramirez, 479 E., Muriel, M., Cordova, A., Carballo, C., Garcia, A., Chi, H., Hernandez, M., 480 2015. Spatial and Temporal Troughfall Deposition Patterns on two regions at the 481 Center and at the Southeast of Mexico. WSEAS Trans. Environ. Dev. 11, 57–72. 15 482 Clarke , N., Zlindra, D., Ulrich, E., Mosello, R., Derome, J., Derome, K., König, N., 483 Lövblad, G., Draaijers, G.P.J., Hansen, K., Thimonier, A., Waldner, P., 2010. 484 Sampling and Analysis of Deposition. 66 pp. Part XIV. In: Manual on methods and 485 criteria for harmonized sampling, assessment, monitoring and analysis of the effects 486 of air pollution on forests. UNECE, ICP Forests, Hamburg. 487 Clow, D.W., Roop, H. a., Nanus, L., Fenn, M.E., Sexstone, G. a., 2015. Spatial 488 patterns of atmospheric deposition of nitrogen and sulfur using ion-exchange resin 489 collectors in Rocky Mountain National Park, USA. Atmos. Environ. 101, 149–157. 490 doi:10.1016/j.atmosenv.2014.11.027 491 De Vries, W., Reinds, G.J., van der Salm, C., Draaijers, G.PJ., Bleeker, A., Erisman, 492 J.W., Auée, J., Gundersen, P., Kristensen, H.L., van Dobben, H., de Zwart, D., 493 Derome, J., Voogd, J.C.H., Vel, E.M., 2001. Intensive Monitoring of Forest 494 Ecosystems in Europe, 2001 Technical Report. EC, UN/ECE, Brussels, Geneva, 495 177 pp. 496 Dämmgen, U., Erisman, J.W., Cape, J.N., Grünhage, L., Fowler, D., 2005. Practical 497 considerations for addressing uncertainties in monitoring bulk deposition. Environ. 498 Pollut. 134, 535–48. doi:10.1016/j.envpol.2004.08.013 499 Erisman, J.W., Hensen, A., Mosquera, J., Sutton, M., Fowler, D., 2005. Deposition 500 monitoring networks: What monitoring is required to give reasonable estimates of 501 ammonia/ammonium? Environ. Pollut. 135, 419–431. 502 doi:10.1016/j.envpol.2004.11.015 503 Escarre, A., Roda, F., Terradas, J., Mayor, X., 1999. Nutrient Distribution and 504 Cycling. In: Rodà, F., et al. (Eds.) Ecology of Mediterranean evergreen oak forests. 505 Springer, New York, pp 253–268. 506 Fang, Y., Yoh, M., Koba, K., Zhu, W., Takebayashi, Y., Xiao, Y., Lei, C., Mo, J., 507 Zhang, W., Lu, X., 2011. Nitrogen deposition and forest nitrogen cycling along an 508 urban-rural transect in southern China. Glob. Chang. Biol. 17, 872–885. 509 doi:10.1111/j.1365-2486.2010.02283.x 510 Fenn, M.E., Poth, M.A, Arbaugh, M.J., 2002. A throughfall collection method using 511 mixed bed ion exchange resin columns. ScientificWorldJournal. 2, 122–130. 512 doi:10.1100/tsw.2002.84 513 Fenn, M.E., Bytnerowicz, A., Schilling, S.L., Ross, C.S., 2015. Atmospheric 514 deposition of nitrogen, sulfur and base cations in jack pine stands in the Athabasca 515 Oil Sands Region, Alberta, Canada. Environ. Pollut. 196, 497–510. 516 Fenn, M.E., Poth, M.A., 2004. Monitoring Nitrogen Deposition in Throughfall Using 517 Ion Exchange Resin Columns: A Field Test in the San Bernardino Mountains. J. 518 Environ. Qual. 33, 2007–2014. 519 García-Gómez, H., Garrido, J.L., Vivanco, M.G., Lassaletta, L., Rábago, I., Àvila, A, 520 Tsyro, S., Sánchez, G., González Ortiz, A, González-Fernández, I., Alonso, R., 521 2014. Nitrogen deposition in Spain: Modeled patterns and threatened habitats 522 within the Natura 2000 network. Sci. Total Environ. 485-486, 450–60. 523 doi:10.1016/j.scitotenv.2014.03.112 16 524 García-Gómez, H., Aguillaume, L., Izquieta-Rojano, S., Valiño, F., Àvila, A., 525 Elustondo, D., Santamaría, J.M., Alastuey, A., Calvete-Sogo, H., González- 526 Fernández, I., Alonso, R., 2015. Atmospheric pollutants in peri-urban forests of 527 Quercus ilex: evidence of pollution abatement and threats for vegetation. Environ. 528 Sci. Pollut. R., 1-14. Published online: 1 December 2015. doi:10.1007/s11356-015- 529 5862-z 530 Hansen, J., 2012. Assessment of Atmospheric Nitrogen Deposition : Possible Effects 531 on Alpine Ecosystems Above 9000 Feet In Grand Teton National Park. Thesis. 532 Homyak, P.M., Sickman, J.O., 2014. Influence of soil moisture on the seasonality of 533 nitric oxide emissions from chaparral soils, Sierra Nevada, California, USA. J. Arid 534 Environ. 103, 46–52. doi:10.1016/j.jaridenv.2013.12.008 535 James, J.J., Richards, J.H., 2005. Plant N capture from pulses: Effects of pulse size, 536 growth rate, and other soil resources. Oecologia 145, 113–122. 537 doi:10.1007/s00442-005-0109-1 538 Izquieta-Rojano, S., García-Gomez, H., Aguillaume, L., Santamaría, J.M., Tang, Y.S., 539 Santamaría, C., Valiño, F., Lasheras, E., Alonso, R., Àvila, A., Cape, J.N., 540 Elustondo, D., 2016. Throughfall and bulk deposition of dissolved organic nitrogen 541 to holm oak forests in the Iberian Peninsula: Flux estimation and identification of 542 potential sources. Environ. Pollut. 210, 104–112. doi:10.1016/j.envpol.2015.12.002 543 Köhler, S., Jungkunst, H.F., Gutzler, C., Herrera, R., Gerold, G., 2012. Atmospheric 544 ionic deposition in tropical sites of central Sulawesi determined by ion exchange 545 resin collectors and bulk water collector. Water. Air. Soil Pollut. 223, 4485–4494. 546 doi:10.1007/s11270-012-1211-8 547 Krupa, S. V, 2002. Sampling and physico-chemical analysis of precipitation: a review. 548 Environ. Pollut. 120, 565–94. 549 Langlois, J.L., Johnson, D.W., Mehuys, G.R., 2003. Adsorption and Recovery of 550 Dissolved Organic Phosphorus and Nitrogen by Mixed-Bed Ion-Exchange Resin. 551 Soil Sci. Soc. Am. J. 67, 889–894. 552 Lindberg, S.E., Lovett G.M., Richter D.D., Johnson D.W., 1986. Atmospheric 553 deposition and canopy interactions of major ions in a forest. Science, 231, 141-145. 554 Meixner, T., Fenn, M., 2004. Biogeochemical budgets in a mediterranean catchment 555 with high rates of atmospheric N deposition - Importance of scale and temporal 556 asynchrony. Biogeochemistry 70, 331–356. doi:10.1007/s10533-003-4106-3 557 Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A., Kent, J., 2000. 558 Biodiversity hotspots for conservation priorities. Nature 403, 853–8. 559 doi:10.1038/35002501 560 Peñuelas, J., Filella, I., 2001. Herbaria century record of increasing eutrophication in 561 Spanish terrestrial ecosystems. Glob. Chang. Biol. 7, 427–433. doi:10.1046/j.1365- 562 2486.2001.00421.x 563 Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin, F.S., Lambin, E., Lenton, 564 T.M., Scheffer, M., Folke, C., Schellnhuber, H.J., Nykvist, B., de Wit, C. a., 565 Hughes, T., van der Leeuw, S., Rodhe, H., Sörlin, S., Snyder, P.K., Costanza, R., 17 566 Svedin, U., Falkenmark, M., Karlberg, L., Corell, R.W., Fabry, V.J., Hansen, J., 567 Walker, B., Liverman, D., Richardson, K., Crutzen, P., Foley, J., 2009. Planetary 568 boundaries: Exploring the safe operating space for humanity. Ecol. Soc. 14, 472– 569 475. doi:10.1038/461472a 570 Rodà, F., Àvila, A., Rodrigo, A., 2002. Nitrogen deposition in Mediterranean forests. 571 Environ Pollut 118, 205–13. 572 Root, H.T., Geiser, L.H., Fenn, M.E., Jovan, S., Hutten, M. a., Ahuja, S., Dillman, K., 573 Schirokauer, D., Berryman, S., McMurray, J. a., 2013. A simple tool for estimating 574 throughfall nitrogen deposition in forests of western North America using lichens. 575 For. Ecol. Manage. 306, 1–8. doi:10.1016/j.foreco.2013.06.028 576 Sanz, J., Calvete-Sogo, H., González-Fernández, I., Lin, J., García-Gómez, H., 577 Muntifering, R., Alonso, R., Bermejo-Bermejo, V., 2015. Foliar senescence is the 578 most sensitive response to ozone in Bromus hordeaceus and is modulated by 579 nitrogen input. Grass Forage Sci. 70, 71–84. doi:10.1111/gfs.12090 580 Sardans, J., Alonso, R., Janssens, I.A., Carnicer, J., Vereseglou, S., Rillig, M.C., 581 Fernández-Martínez, M., Sanders, T.G.M., Peñuelas, J., 2015. Foliar and soil 582 concentrations and stoichiometry of nitrogen and phosphorous across European 583 Pinus sylvestris forests: relationships with climate, N deposition and tree growth. 584 Funct. Ecol. n/a–n/a. doi:10.1111/1365-2435.12541 585 Sheibley, R.W., Enache, M., Swarzenski, P.W., Moran, P.W., and Foreman, J.R. 2014. 586 Nitrogen deposition effects on diatom communities in lakes from three national 587 parks in Washington state. Water, Air and Soil Pollution 225: 1857. 588 doi:10.1007/s11270-013-1857-x 589 Simkin, S.M., Lewis, D.N., C.Weathers, K., Lovett, G., Schwarz, K., 2004. 590 Determination of sulfate, nitrate, and chloride in throughfall using ion-exchange 591 resins. Water, Air, Soil Pollut. 253, 343–354. 592 Simpson, D., Fagerli, H., Hellsten, S., Knulst, J.C., Westling, O., 2006. Comparison of 593 modelled and monitored deposition fluxes of sulphur and nitrogen to ICP-forest 594 sites in Europe. Biogeosciences 3, 337–355. 595 Tulloss, E.M., Cadenasso, M.L., 2015. Nitrogen deposition across scales : hotspots and 596 gradients in a California savanna landscape. Ecosphere 6, 167. 597 van Dam, D., Heil, G.W., Heijne, B., Bobbink, R., 1991. Throughfall below grassland 598 canopies: A comparison of conventional and ion exchange methods. Environ. 599 Pollut. 73, 85–99. doi:10.1016/0269-7491(91)90016-P 600 Vlek, P. L. G., Stumpe, J. M., 1978. Effects of solution chemistry and environmental 601 conditions on ammonia volatilization losses from aqueous systems. Soil Sci. Soc. 602 Am. J. 42, 416-421. 603 Vourlitis, G.L., Pasquini, S.C., Mustard, R., 2009. Effects of Dry-Season N Input on 604 the Productivity and N Storage of Mediterranean-Type Shrublands. Ecosystems 12, 605 473–488. doi:10.1007/s10021-009-9236-6 606 Wieder, R.K., Vitt, D.H., Burke-Scoll, M., Scott, K.D., House, M., Vile, M. a., 2010. 607 Nitrogen and sulphur deposition and the growth of Sphagnum fuscum in bogs of 18 608 the Athabasca Oil Sands Region, Alberta. J. Limnol. 69, 161–170. 609 doi:10.3274/JL10-69-S1-16 610 Yamashita, N., Sase, H., Kobayashi, R., Leong, K.-P., Hanapi, J.M., Uchiyama, S., 611 Urban, S., Toh, Y.Y., Muhamad, M., Gidiman, J., Chappell, N. a., 2014. 612 Atmospheric deposition versus rock weathering in the control of streamwater 613 chemistry in a tropical rain-forest catchment in Malaysian Borneo. J. Trop. Ecol. 614 30, 481–492. doi:10.1017/S0266467414000303 615 19 616 Table 1. Compilaton of data from different studies reporting field blanks, field sampling 617 details and laboratory tests of performance for ion exchange resins collectors. NO3- NH4+ Ion exchange Number and Blank Intra-site Blank Intra-site Study Adsorption Recovery Adsorption Recovery resin type of blanks correction variability correction variability effiency efficiency effiency efficiency (mg N g-1) b (CV) (mg N g-1)b (CV) Mixed-bed 3 in total; field 0.014 ± Boutin et al., 2015 IONAC® NM- blanks Non detected 12 % a 19 % a 60 0.006 Near Near Brumbaugh et al., 2012 2-stage columns 9; field blanks detection 8–10 % 106 ± 7 % detection 5–10 % 101 ± 3 % limits limits Mixed bed Cerón et al., 2015 Amberlite™ 3; field blanks 98.6 % 98.6 % IRN150 Mixed bed 1 per site; field Clow et al., 2015 Amberlite™ < 0.003 16 % a 100 % < 0.003 21 % a 88 % blank IRN150 Mixed bed 2 per site; field Fang et al., 2011 201x7[717] & blank 0.003–0.028 9–34 % 90–99 % 90 % < 0.001 5–23 % 94–100 % 97 % 001x7[732] Mixed bed Fenn et al., 2002 Amberlite™ 5; lab. blanks < 0.001 104.4 % 0.001 104.5 % IRN150 Mixed bed Fenn and Poth, Amberlite™ Lab. blanks < 0.001 1–16 % 0.001 13–27 % 2004 IRN150 mixed bed 1 per plot; field Hansen, 2012 Rexin® blanks < 0.001 < 0.001 Mixed bed Köhler et al., 2012 Amberlite™ 1; lab. blank 0.001 95% MB 20 Mixed bed 1 field blank Approx. Approx. Sheibley et al., 2014 Amberlite™ (per site) and 3 0.001–0.003 9–36 % 90 - 91 % ≤ 0.001 7–89 % 74–96 % IRN150 lab. blanks 100 % 100 % Anion-exchange DowexTM 93.9 - Simkin et al., 2004 Monosphere 3; lab. blanks < DL 100 % 100.4 % 550-A Mixed bed Tuloss and Amberlite™ 90 - 95 % 90–95 % Cadenasso, 2015 IRN150 Mixed bed van Dam et al., DowexTM 1-X8 18 % 8% 1991 and 50W-X8 Mixed bed 3–5 per period; Wieder et al., 2010 Amberlite™ lab. blank IRN150 Mixed bed Yamashita et al., 1 per period; < 10–50 Amberlite™ < 8–25 % 2014 field blank % MB-1 Mixed bed of 90.3 - 95.5 Zhan et al., 2015 #717 anion and 2 field blanks > 99 % > 99 % 90.9–100 % #732 cation % Mixed bed Present work (CB) Amberlite™ 1–3; lab. blanks < 0.001–0.002 2–66 % 0.024–0.061 0–141 % IRN150 Mixed bed Present work (CA) Amberlite™ 2; lab. blanks < 0.001–0.006 5–67 % 0.039–0.071 22 % IRN150 Mixed bed Present work (TC) Amberlite™ 1; lab. blanks 0.001–0.004 2–20 % 100 % 99 % 0.054–0.103 4–27 % 100 % 112 % IRN150 a 618 : not CV b 619 : milligram of N released per gram of resin 20 620 621 Table 2. Bulk and throughfall deposition (kg N ha-1) using ion-exchange resin collectors. Bulk Throughfall Start date End date Season1 n2 NO3- NH4 + n NO3- NH4+ CB 05/04/11 28/06/11 SPR 2 0.54 0.33 8 1.66 2.01 05/07/11 27/09/11 SUM 2 0.42 0.86 8 1.22 0.94 04/10/11 10/01/12 AUT 2 0.43 1.03 8 1.35 0.81 17/01/12 18/04/12 WIN-SPR 2 0.39 0.01 7 1.40 0.85 14/05/12 03/07/12 SPR 2 0.49 0.43 8 1.92 0.59 11/07/12 01/10/12 SUM-AUT 2 0.29 0.01 7 1.01 0.30 16/10/12 25/02/13 AUT-WIN 2 0.30 0.61 7 0.86 0.65 CA 31/05/11 31/08/11 SUM 1 0.67 0.50 8 0.50 0.15 31/08/11 13/12/11 AUT 1 0.35 0.26 1 0.80 0.27 13/12/11 13/03/12 WIN 3 0.58 0.37 7 0.67 0.12 13/03/12 19/06/12 SPR 1 1.07 2.12 1 1.37 0.76 19/06/12 25/09/12 SUM 2 0.58 0.39 11 0.92 0.49 25/09/12 06/03/13 AUT-WIN 1 1.41 1.54 2 1.29 0.49 TC 23/03/11 21/06/11 SPR 3 0.73 1.03 11 0.54 1.55 28/06/11 27/09/11 SUM 4 0.16 0.16 12 0.80 0.28 11/10/11 21/02/12 AUT-WIN 4 0.36 0.49 12 0.85 0.71 06/03/12 21/06/12 WIN-SPR 4 0.37 0.76 12 0.64 0.77 26/06/12 23/10/12 SUM-AUT 4 0.32 0.42 11 1.69 1.12 30/10/12 25/03/13 AUT-WIN 4 0.38 0.57 12 0.22 0.65 622 1 623 : SPR: spring; SUM: summer; AUT: autumn; WIN: winter 2 624 : number of collectors used to calculate the mean deposition value 625 21 626 FIGURES 627 628 Fig. 1. Location of the study sites: CB, Can Balasc (Barcelona); CA, Carrascal (Pamplona); 629 TC, Tres Cantos (Madrid). Distribution of Q. ilex habitats in Spain is added. 630 22 631 632 Fig. 2. Comparison of bulk deposition values of nitrate, ammonium and their sum (total 633 inorganic nitrogen; TIN) measured with conventional bulk water collectors (CBC) and ion- 634 exchange resin collectors (IEC). Data from all sites and sampling periods included. 635 Comparison metrics are added in the graphic: mean fractional bias (MFB), root mean square 636 error (RMSE) and Pearson correlation coefficient (r). Dashed lines represent the line 1:1 637 (perfect fit) and lines 1:1.5 and 1:0.5 (±50%). Error bars correspond to standard deviation of 638 the period mean. 639 23 640 Fig. 3. Mean annual bulk, throughfall and net throughfall (throughfall minus bulk) deposition 641 of nitrogen (N) compounds for the 2-year sampling period (2011-2013). Measurements of N 642 deposition are represented for the two methodologies: conventional bottle collector (CBC) and 643 ion-exchange resin collector (IEC). Error bars represents the standard error of the mean (n = 2 644 for CBC and IEC; n = 3 for EMEP). 24 645 Fig. 4. Net throughfall deposition of N and precipitation on a monthly basis at the Tres Cantos 646 (TC) site. *: bars not shown because of the presence of missing data for this month. 647 25