Alkaline air: changing perspectives on nitrogen and air pollution in an ammonia-rich world

Ammonia and ammonium have received less attention than other forms of air pollution, with limited progress in controlling emissions at UK, European and global scales. By contrast, these compounds have been of significant past interest to science and society, the recollection of which can inform future strategies. Sal ammoniac (nūshādir, nao sha) is found to have been extremely valuable in long-distance trade (ca AD 600–1150) from Egypt and China, where 6–8 kg N could purchase a human life, while air pollution associated with nūshādir collection was attributed to this nitrogen form. Ammonia was one of the keys to alchemy—seen as an early experimental mesocosm to understand the world—and later became of interest as ‘alkaline air’ within the eighteenth century development of pneumatic chemistry. The same economic, chemical and environmental properties are found to make ammonia and ammonium of huge relevance today. Successful control of acidifying SO2 and NOx emissions leaves atmospheric NH3 in excess in many areas, contributing to particulate matter (PM2.5) formation, while leading to a new significance of alkaline air, with adverse impacts on natural ecosystems. Investigations of epiphytic lichens and bog ecosystems show how the alkalinity effect of NH3 may explain its having three to five times the adverse effect of ammonium and nitrate, respectively. It is concluded that future air pollution policy should no longer neglect ammonia. Progress is likely to be mobilized by emphasizing the lost economic value of global N emissions ($200 billion yr−1), as part of developing the circular economy for sustainable nitrogen management. This article is part of a discussion meeting issue ‘Air quality, past present and future’.

: Trade in nao sha (sal ammoniac) and other weighed commodities recorded in one year (c. 620) from a weighing station near Turfan. Calculations are based on records summarized by Skaff (1998, pp 81-93 Notes: a, Single transactions for medicine, muscovado sugar, turmeric, brass and copper; b, Assuming the 'scales tax' of 0.9% for silver applied to all weighed commodities; c, Based on a slave from Samarkand sold in Turfan in 639. Notes: a, Based on the mean of three transactions; b, Derived as the reciprocal from the mean rate (kg/nuqra dirham).

Supplementary Section 2: Long-term changes in nitrogen prices
Comparison of N prices through centuries is complicated by different rates of exchange between multiple currencies. To simplify the comparison, Figure S1 shows N prices normalized to the nominal prices of gold and silver. Table S3 shows the data on which these estimates are based. Figure S1 indicates that, relative to the price of silver, nitrogen compounds (mainly ammonium, ammonia and urea) are currently about 2-3 orders of magnitude cheaper than the earliest records. Relative to the price of gold, present-day nitrogen compounds are 3-4 orders of magnitude cheaper than the early records. The highest relative price for nitrogen noted here concerns purchase (c. 1500) of sal ammoniac among a wide range of alchemical ingredients used at the Scottish royal court (Read, 1938). Figure S1: Changing price of nitrogen over the centuries indicated by the ratio to contemporary silver and gold prices (For data see Tables S3 and S4). Table S3: Comparison of nitrogen price over the centuries with gold and silver pricesbased on nominal price (For basis see Table S4).  Table S4: Basis for the comparison on nitrogen prices to silver and gold prices ( Figure S1 and Table S3).

Supplementary Section 3: Ammonia in Greek alchemical literature
Over the last century, it has been most common to assume that ammonia and ammonium were not known in the Greek alchemical literature characteristic of c. 100 BC to 700 AD. Instead, it has been suggested that they were first known to Arabic and Persian alchemists after around 800 AD (e.g., Ruska, 1923;Needham et al., 1980, p 435). Stapleton (1953, p 40) was unusual in supporting earlier Greek alchemical knowledge of sal ammoniac, noting reference to it in a treatise (in later Arabic translation) attributed to the Greek alchemist Zosimus, who flourished c. 300 AD. By contrast, Hallum (2008) considered that this reference represented a later interpolation by an Arab scribe/editor. While Hallum may be correct as such, this does not distract from the important point the supposed Arab scribe was making: "It is nushādir, I believe, truly, truly" ('Tome of the Art', line 48, trans. Hallum, 2008, p 362). At minimum, the implication is that the Arab scribe considered the Greeks to have been familiar with ammonium salt. The text, as translated by M. Turāb 'Ali (version corrected by H.E. Stapleton, 1948) reads as follows, with Theosebia addressing Zosimus: She said: "Inform me about Nushadur (Sal-Ammoniac)". He replied: It is a bird that flies without a wing. It is called "The Eagle" on account of its flying into the air (when heated)… She said: "Inform me about the Generous Stone which the Sages have mentioned". He replied: It is a Noble Stone. I will not call it by its name, but if you (so) desire, I will let you know its signs, its benefits and its names, without naming it.
She said: "Do so". He replied: It is a Stone and not a Stone. It has a Spirit, a Soul and a Body. If it is liquefied, it becomes liquefied. If it is coagulated, it becomes coagulated. It is of white colour, spacious, swift, light and sultry [or 'briny']: the tongue cannot describe it. In my opinion it is Nushādur. Verily, verily, I have described it as being with the rich, the poor and all peoplein markets, on roads and in dung heaps. None of the Sages can do without it in any of their operations… You should not be deceived by the numerous descriptions of the Sages: because in the "Stone of the People" there will be (found) Four Natures, in the analysis by fire, and Three by appearance, every one of them surrounding its companion. So ponder over this and pay no attention to the unnecessary things which the Sages have inserted in their books in order to conceal it from the (common) people. And I have named it to you by its true name (TS Stapleton 104, pp 4-7, History of Science Museum, Oxford).
Other arguments for Greek alchemical knowledge of ammonia and ammonium include: a) Several other Arab/Persian authors attribute earlier use to the Greeks, including: al-Jāhiz (Wiedemann, 1907, pp 76-77) and al-Rāzī ('Book of Evidences', see Stapleton and Azo, 1910). The same is seen in an anonymous Greek text that relates organic alchemical distillation to the four 'elements', Water, Earth, Fire, Air. Although Colinet (2000) concluded that this work was originally written in Arabic, synonyms added by the Greek editor attribute the procedures as being of prior Greek origin (see Appendix A).

b) Several
Greek alchemical texts describe processing of organic matter, where 'cover names' (Decknamen) demonstrate how organic substances may also be present in other books. For example in a text 'On the same divine water', attributed to Zosimus (Mertens, 1995, pp 30-33), the first fraction of distillation (ammonium carbonate) is referred to as 'rain water' (i.e., macrocosmic allusion), with subsequent distillation fractions referred to as 'radish oil' and 'castor oil'. The philosophical perspective of this work is illustrated by the reference to the awful smell, none of which should escape, because it is "the firstprinciple of the art (huparchei hē technē)", applying a characteristic Neoplatonic term. c) Several processes seen in Islamic alchemy are recognizable in preceding Greek alchemy. Although the earlier texts are deliberately more obscure, it is often possible to identify contrasting descriptions of the same process. This is illustrated by the process of al-Rāzī (Section 2.2): the tincture of burnt oil (Fire) is extracted along with the acid by suspending the coagulated oil in ammonia / ammonium carbonate solution (Water), then repeating 5-6 times, thereby 'marrying' the Fire and Water. The same process can be found in an Arabic translation of Zosimus (Mus ̣ḥ af as ̣-s ̣uwar, Abt, 2011, pp 486-492), including reference to the suspended matter, repeated washing, marriage and colours.
The alchemists' multiple description of the same process with different 'cover names' appears to be a characteristic approach of gaming in early alchemical literature, where texts represent literary puzzles to be solved. Contrary to Hallum (2009), the Mus ̣ḥ af as ̣-s ̣uwar is here noted to meet all the criteria of an authentic (if edited) translation of Zosimus (e.g., all appropriate substances and authorities named or not named). However, in case this should be a point of contention, it is worth noting that the same alchemical marrying of Water and Fire can be seen in the Phusika kai Mustika traditionally attributed to Democritus (Martelli, 2013, pp S86-95). Here the procedure is repeated in ten short operations using apparently different ingredients (concerning which Zosimus dedicated a 10-chapter commentary; Abt, 2016). For example, an operation of Democritus to "make cinnabar white", uses "oil, or vinegar, or honey, or brine, or alum", where these five substances appear to be cover names for five additions of the ammonia solution, which extract the redness, so whitening the coagulated oil (the Cinnabar). The subsequent recipe even mentions "castor oil or radish oil".
It should be emphasized that this is not a new interpretation, but was the traditional reading of the alchemists themselves. In the margin of a famous alchemical manuscript (Venezia, Biblioteca Nazionale Marciana, Gr. Z. 299 (=584), see f. 68 v.) formerly owned by Cardinal Basilios Bessarion (likely brought with him from Constantinople, c. 1438), Bessarion (?) commented of the Cinnabar-whitening above: "Alum is aether and [Mercury (hudrarguros, literally water-silver)] and copper-without-shadow", succinctly demonstrating his understanding of these synonyms. It suggests we should "sharpen our mind" (Synesius, Commentary on the Book of Democritus, trans. Martelli, 2013, p S129) to the idea that ammonium, nūshādir, aether and alum may all be one thing, their meaning summarized lightly in the overall synthema , a form applicable to all.
While such recipes of the earlier Greek alchemical literature have been interpreted over the past century as technical procedures using the chemicals named, the perspective outlined here thus indicates that they are better understood as presenting a few covert processes in multiple ways, as playful puzzles to be decoded. This suggests that first appearances of key features in alchemical literature (e.g., new substances, processes and philosophical interpretations) may not actually be new at all. Rather, it seems more likely that such innovations mainly reflect increasing openness in the description of existing alchemical activity and of its meaning as understood by the alchemists. As a result, we may expect to find ammonia processes in many other alchemical writings, including the exuberant periphrasis of Stephanus of Alexandria (Papathanassiou, 2017) and the Greek alchemical poems, such as that attributed to Heliodorus (Goldschmidt, 1923). A further conclusion, that both ammonia and alchemy are similarly recognizable in the extant philosophical fragments of Empedocles and Heraclitus and in several romances, such as the Aethiopica of Heliodorus and the tale of Cenerentola, is likely to be contentious and must be a matter for future debate. This perspective highlights the extreme difficulty and uncertainty in tracing early knowledge of ammonia.

Supplementary Section 4: Air quality changes and Alkaline Air
Whereas 'alkaline air' was the original name given to gaseous ammonia by Priestley, the main text (Section 1) also defines alkaline air from an environmental perspective as "air where alkaline gases (primarily NH3, but in principle also including volatile amines) dominate over those that are acidic in nature (primarily SO2, NOx, HCl etc)." Based on this concept, we define an additional term the 'gaseous alkaline fraction' in air based on the relative composition of alkaline and acidic components. Values larger than 0.5 can be considered as representing 'alkaline air' from this environmental perspective.
The total of gaseous alkaline components (alkaline) and acidic components (acid) is summed to give a combined total (alk+acid), calculated as follows (expressed in nmol m -3 or ppb): For the present calculation, alkaline is simply taken as the gaseous NH3 concentration in air (nmol m -3 or ppb). As SO2 is oxidized in the atmosphere (and when deposited on surfaces) to H2O4 it is considered as dibasic. The value of alk+acid may be calculated including or excluding NOx, since it is rather insoluble in water, so that its effect (especially when deposited on plant surfaces) may be much less in practice than indicated by NOx concentration. Other components typically existing at lower concentrations, such as amines, nitrous acid and organic acids could also be included in a more comprehensive approach where data are available. The approach could similarly be extended to consider the contribution of non-neutral particulate matter, including calcareous particles (i.e., as total alkaline fraction), as well as the contribution of increasing CO2 concentrations.
Based on values of alk+acid, the gaseous alkaline fraction (alk) is calculated simply as: Application of this approach to the UK (see Figure S2), indicates that alkaline has increased from 30% in 1985 to 88% in 2017 (when excluding NO2). Including NO2 in the calculation of alkaline gives values ranging from 18% in 1997 to 37% in 2017. Based on the assumption that NO2 concentration mainly represents potential rather than effective acidity (especially when considering immediate effects on leaf surfaces), the mean UK atmosphere can be considered as having represented alkaline air (alkaline>50%) since 1997. An alternative way to express such differences would be to calculate  values based on the relevant dry deposition fluxes to ecosystems.
Values of alkaline for the globe (Figure 3) are based on modelled gaseous concentrations for the surface atmosphere (derived from the model first layer 0 -45 m) using the EMEP-WRF global model (Vieno et al., 2016) an implementation of the EMEP MSC-W model (Simpson et al., 2012) using meteorology from the WRF model (Skamarock et al., 2019). Global emissions for 2010 are taken from the ECLIPSEv6a emissions inventory (IIASA -GAINS model). HCl is not simulated in this model and is excluded from the estimation. The effect of including or excluding NOx concentrations is shown in Figure 3A and 3B.

Supplementary Section 5: Lichen responses to atmospheric ammonia Methodology for Lichen Recording at the Farm Transect
A transect was established in 2002 in a mixed woodland adjacent to a poultry farm near Earlston, southern Scotland, in order to assess the effects of atmospheric NH3 on epiphytic lichen communities. Passive diffusion tubes were set up to monitor NH3 concentrations on a transect through the woodland at 16,46,76,126,200 and 276 m from the poultry buildings in a northerly direction (Sutton et al., 2004, pp. 75-86). For other nitrogen indicators at this transect, see Pitcairn et al. (2002Pitcairn et al. ( , 2006. At each location established for NH3 monitoring, 5 trees of Pinus sylvestris or Picea sitchensis with girths >40 cm were selected for recording lichens on trunks using a ladder quadrat of 5, 10 x 10 cm square quadrats placed on 4 cardinal aspects (North, South, East, West) between 1-1.5 m height on the trunk and lichens recorded in each quadrat. Frequency of epiphytic lichens occurring on twigs of Betula pubescens was recorded using presence on 5 sampled twigs in the vicinity of NH3 sampling sites, as well as at a nearby nature reserve Gordon Moss (5.8 km) as a background site. Bark pH was recorded for each aspect on the trunks using a flat tip electrode, averaged for each tree and this was adapted to record bark pH on branches (Wolseley et al. 2006). Means were calculated for each sampling location. Crustose, fruticose and foliose lichens were recorded in all quadrats. Using indicator species identified by van Herk (2000), nitrophytes were distinguished from acidophytes. Based on the indices for acidophytes (LA) and nitrophytes (LN), the combined index LAN was calculated as LAN = LA -LN.

Methodology for Lichen Recording at the UK scale
The UK wide survey of epiphytic lichens was conducted in 2004 in the vicinity of National Ammonia Monitoring Network Sites (Tang et al., 2019a). This combined intensive recording of epiphytic lichens at a few sites with a simplified recording of macrolichens, with fieldwork undertaken by agency staff at 32 sites across the UK . Recording methods were the same as those used in the Earlston farm transect, combining bark pH with frequency of macrolichen species on trunks and twigs of trees adjacent to the NH3 monitoring sites (Wolseley et al. 2009). Further research investigating the effects of NH3, atmospheric pollutants and climatic factors on epiphytic lichens was undertaken during a PhD project (Lewis, 2012) leading to the preparation of a Nitrogen Air Quality Index (NAQI) using a Lichen Indicator score (Wolseley et al. 2013).

Supplementary Results from the Farm Transect
As shown in Figure S2, bark pH of both trunks and twigs was found to be positively correlated with NH3 concentrations (A). Other relevant indicators measured at this site recorded trunk lichen diversity according to a German protocol (VDI, 1995) and European Protocol (LDV) (Asta et al., 2000) (B), mean score according to Ellenberg nitrogen values of species present (Wirth, 1992) (C), and measured foliar nitrogen concentration of two key species, Hypogymnia physodes and Xanthoria parietina (D). While neither the VDI nor the LDV lichen diversity indices were found to correlate uniformly with the full range of NH3 concentrations, positive correlations were seen for both the Ellenberg nitrogen index and foliar N concentration. A.
D. Figure

Supplementary Results from the UK-wide lichen survey
The full dataset for the UK-wide survey is shown in Table S5, from which the main relationships were derived based on the LAN indices for trunks and twigs. Of the 32 sites, it was possible to obtain lichen survey data at 30 sites, allowing calculation of LAN values at 24 sites for trunks (11 of which were for oak) and 26 sites for twigs (14 of which were for oak).
Where it was possible to combine twig and trunk results, this provided 50 samples overall and 25 samples for oak. A summary of the main relationships between LAN values possible drivers is given in Figure S4 expressed as correlation coefficients (R 2 ).
Consistently higher correlations were found when considering only the data for lichens growing on oak bark, compared with the full dataset, which can be explained by natural difference in bark pH (reflective of different bark chemistry) between tree species. Higher correlations were found between LAN and NH3 concentrations than with N deposition.
Weak or insignificant relationships were found for LAN with mean SO2 concentrations and annual precipitation, with higher correlations with bark pH (significant in all tests shown in Figure S4). The highest correlation for a single factor relationship was found for the ratio of N deposition to S deposition, reflecting a strong negative correlation with N deposition and a weak positive correlation (only significant for twigs) of S deposition to LAN. Although harder to understand than NH3 concentrations, this suggests itself as another indicator of atmospheric acid-base relations. Indeed, the N deposition to S deposition ratio was found to be positively and closely correlated with NH3 concentrations (R 2 =0.87, n=32). Figure S4 also shows the correlations with two combined two-factor indicators, where the correlations between LAN and NH3 + 4 (bark pH) are all highly significant. The correlations for LAN and Ndep/Sdep + 2.8 (bark pH) are slightly higher than for NH3 + 4(bark pH). Again this suggests that the Ndep/Sdep indicator has additional utility as a complement to the simpler indicator of NH3 concentrations. The multipliers 4 and 2.8 used in the combined pollutionbark indices are the values that maximize the overall correlation with LAN.

Supplementary Section 6: Effects of controlled ammonia exposure on peatland vegetation Experimental Site
The experiment is carried out at Whim bog in the Scottish Borders (3° 16' W, 55° 46' N) on 3-6 m of peat. Mean temperatures of the air and soil (at 10 cm depth) were 7.9 °C and 7.6 °C respectively with mean annual rainfall (2011 to 2016) of 1141 mm (range: 734-1486 mm). The peat is very acidic, with pH 3.4 (3.27-3.91 in water). The vegetation is mainly classified as a Calluna vulgaris -Eriophorum vaginatum blanket mire community (M19 in the UK National Vegetation Classification). Prior to treatment and in controls, the replicate plots represent highly variable mosaics containing Calluna vulgaris and Sphagnum capillifolium hummocks, with hollows containing S. fallax and S. papillosum. Other common species include Erica tetralix and the mosses Hypnum jutlandicum and Pleurozium schreberi.

Experimental Treatments
Nitrogen is applied to the site using two different treatment systems, one for dry deposition of NH3 gas, and one for wet deposition of NH4 + and NO3in solution. Treatments commenced in June 2002 and continue through the year, except during winter (i.e., freezing conditions prevent operation of the wet release system). NH3 deposition is manipulated using a free-air release system, supplied from a cylinder of liquid NH3, diluted with ambient air and released from a perforated 10 m pipe, 1 m above ground. NH3 is released only when wind direction was in the south-west (from 180° to 215°) and wind speed exceeded 2.5 m s -1 (Leith et al., 2004). Ammonia concentrations are measured 0.1 m above the vegetation using ALPHA samples (Tang et al., 2001) at 8 distances downwind of the source, as well as background. Ammonia dry deposition is calculated from the concentration measurements using the method of Cape et al. (2008). This bespoke method applies a canopy resistance (Rc) based inferential modelling approach using monitored NH3 concentrations at 0.1 m above the canopy. The main uncertainty is the extent to which Rc values change as a function of NH3 concentration. Therefore a specific concentration-dependent function of Rc is used, derived from detailed chamber measurements for this habitat by Jones et al. (2007).
Wet deposition of NH4 + and NO3is increased in replicated mesocosms in a randomised block design, using a water sprayer system using collected rainwater to dilute solutions of NH4Cl or NaNO3, to cover each plot (12.8 m 2 ) uniformly. In addition to a background level of 8 kg N ha -1 y -1 , three treatment levels are applied, aiming to provide total (background + treatment) nitrogen of 16, 32 and 64 kg N ha -1 y -1 .
The dry ammonia transect is 10 m wide by 60 m long, which was designed to allow for multiple samples of individual species at each concentration along the transect. This provides a sufficiently large sampling pool to assess changes in the species within the habitat from the line source, allowing variation to be considered with distance from the NH3 source, as well as variation between permanent replicated quadrats at each sampling distance, exposed to the same NH3 levels. Use of multiple NH3 line sources of this scale would increase the sampling pool, but would not be feasible due to the environment impact at this experimental study site, and because of the multiplication of sampling effort and resources that would be required.
Conversely, use of smaller transects would not provide the real world scenario that is created by the Whim experiment.
The major changes in species along the NH3 transect demonstrate the successful design, while increasing the sampling pool further would be unlikely to improve relationships substantially. If substantial extra financing were available, rather than increasing replication, it would be considered a priority to invest in additional experiments of the type developed at Whim Bog, for other ecosystem types, to address the extent to which the observed differential responses between NH3, NH4 + and NO3are also found in other habitats.
The NH4 + and NO3wet treatments have four 12.5m 2 plots for each N level, and were also designed in size to allow for multiple samples of individual species within each plot. This provides a high level of replication for each species, across a number of replicated plots.
Effects of the treatments on peat carbon and nitrogen interactions have been reported elsewhere by Kivimäki et al. (2013) and Sheppard et al. (2013).

Species survey
Plant species composition has been surveyed in all plots over the course of the experiment, usually every two years. In each experimental plot, three permanent quadrats (40 x 40 cm) were established. These were sub-divided into 16 sub-quadrats (10 x 10 cm). At each survey, the percent cover of all species is recorded at the sub-quadrat level. For each species, the 16 sub-quadrat values are averaged to give a mean cover for each quadrat. For further details of the site, treatments and plant survey analysis, see Sheppard et al. (2014) and Levy et al. (2019).

Leaf surface pH
Surface pH of Sphagnum capillifolium and Cladonia portentosa were scored by modifying the methodology of Costa et al. (2005). Intensive measurements were done in situ at the CEH Whim experimental field site with a pH meter (Horiba LAQUAact-pH 110) having a flatbottom pH electrode from June to September 2018, supplemented by subsequent measurements in 2019. The surface of these Sphagnum and Cladonia species were moistened with deionized (DI) water and were allowed to wet. After 2-3 minutes, a drop of DI water (0.1 ml) was pipetted on the surface of Sphagnum and Cladonia and left for 10 seconds prior to pH measurement; the measurement was then repeated to ensure a stable reading. The flat tip of the pH electrode was put on the individual Sphagnum and Cladonia plants held by the tweezer from beneath, and the pH was recorded on the screen of the digital pH meter. The temperature was set at MTC (20 °C). Before the measurements, the pH meter electrode was immersed in KCl (3.33 mol L -1 ) solution and then calibrated with pH 4 and pH 7 solutions.

Physiological indicator measurements
An indication of plant photosynthetic functioning was provided by measuring fluorescence of two lichen species, transplanted into the NH4 + , NO3and NH3 plots: Evernia prunastri (representative of an acidophyte species) and Xanthoria parietina (representative of a nitrophyte species). Fluorescence measurements using a Plant Efficiency Analyzer (Handy PEA, Hansatech Ltd, Norfolk, UK) were carried out at three periods allowing up to 6 months field exposure of transplants (Munzi et al., 2014). Fully hydrated samples were dark adapted for 10 min, then illuminated for 1 s with a saturating (3000 μmol m −2 s −1 ) red-light pulse and fluorescence emission was recorded for 1 s. The parameter Fv/Fm, the maximum quantum yield of primary photochemistry, was used as a proxy of the intrinsic efficiency of photosystem II (PSII) and therefore as a vitality index for the samples, which corresponds to: where F0 and Fm are the minimum (before the light pulse, when all the reaction centres of PSII are open) and maximum chlorophyll a fluorescence (after the light pulse, when all the reaction centres are closed) emission, and Fv is the variable fluorescence, the difference between the maximum and the minimum. Optimal values for healthy lichens and mosses vary around 0.6 and 0.8.

Plant species cover
Data are examined here for five common species. These were the only species that frequently occurred with more than 5 % cover: Calluna vulgaris, Sphagnum capillifolium, Cladonia portentosa, Hypnum jutlandicum and Pleurozium schreberi. In addition, Eriophorum vaginatum was abundant with over 5% cover, as further analysed by Levy et al. (2019). This plant is not considered here, since it was found to benefit from gaseous NH3 exposure relative to the more sensitive plant species. The data reported by Levy et al. (for 2002Levy et al. (for -2016 are here extended for plant cover sampling up to and including 2019.
The results of the plant cover sampling are illustrated in Figure S5a-c for Calluna vulgaris, S. capillifolium and C. portentosa, respectively. It is immediately obvious that the effects of NH3 on percentage cover are much larger and occur more rapidly than for wet deposited NH4 + and NO3 -. Of the three species shown, C. portentosa is found to be the most sensitive and C. vulgaris the least sensitive.
In order to compare the outcomes for the five key species, the data for each treatment (by nitrogen form, N application rate and species) are here used to estimate the time taken to reduce cover by 50% compared with the start of the experiment (ET50). Linear interpolation was used, except where the data were obviously non-linear, where simple decay curves were fitted (as shown in Figure S5).
The Eradication Dose 50 (ED50) is defined as the annual nitrogen deposition multiplied by the time taken to reduce cover by 50% compared with the start of the experiment (kg N ha -1 ). In Figure 6 (main text) the calculated results for ED50 are plotted versus the annual N deposition in the experimental treatments in order to visualize the outcomes, and see to what extent the ED50 may itself depend on the N deposition rate. A summary of the ET50 and ED50 results and the calculations underpinning Figure 7 is given in Table S7.
Statistical comparison was based on reciprocal values of ED50, which helped normalize the data compared with the skew distribution of ED50 values ( Figure 6). This also allowed the reciprocal value of ED50 to be set at 0 for treatments where plant cover was stable or increased over time. Making this correction would have tended to underestimate the differences between treatments shown in Figure 7.
Since the experiment was established in 2002, background nitrogen deposition has decreased by around 20% (Levy et al., 2020) and it is possible to calculate the likely effect on the estimation of total N dose of the 18 year exposure period of the experiment. Based on total N deposition in 2018 being 20% less than in 2001, then the deposition rates for 2018 would be 6.4 (background), 14.4, 30.4 and 62.4 kg -1 y -1 . Since the background N deposition rate decreased approximately linearly through the period, then the total accumulated N doses of the 18 year period can be estimated as being slightly smaller than indicated by the nominal annual rates of 8 (background), 16, 32, 64 kg N ha -1 y -1 , by 10% (background), 5%, 3% and 1%, respectively. This indicates only a minor effect on the calculated ED50 values shown in Figure 6 and Table S7. A further change during the life of the experiment is that background precipitation is no longer as acidic as it was during 2001 ( Figure 2C).

Interactions with leaf surface pH
Wet deposition of NH4 + and NO3is also associated with potential acidity and alkalinity interactions. Hence plant uptake of deposited NH4 + (to become R-NH2 in organic compounds) releases H + , with additional acidity released if NH4 + is nitrified to NO3 -, which may then be leached accompanied by loss of base cations (Sutton et al., 1993). By contrast, reduction of NO3to NH4 + consumes H + , although this effect can be partly or fully offset if this is then converted to organic R-NH2 forms. In an acidic peatland system, with limited nitrification expected, a net acidifying effect of NH4 + may be expected (to the extent that this is taken up by plants), while NO3has potential for a weak alkaline effect. As these effects are dependent on plant uptake and also result from low aqueous N concentrations (compared with high ammoniacal concentrations on plant surfaces from exposure to gaseous NH3), then much smaller pH effects are expected compared with gaseous NH3. Comparison of the surface pH of Sphagnum capillifolium between wet deposition NH4 + and NO3treatments showed no significant differences (cf. Figure S8). By contrast, limited measurements of lichens growing on wooden walkways ('board-walks') adjacent to some plots did show pH differences (data not shown), which may be examined in a future publication.

Differences in physiological response to N forms
The photosynthetic indicator Fv/Fm showed that both the transplanted lichen species were more affected by exposure to gaseous NH3 than to wet deposited NH4 + and NO3 -( Figure S6). In the case of the nitrophyte species Xanthoria parietina, there was no apparent effect of either wet deposited NH4 + or NO3on Fv/Fm, whereas for the acidophyte species Evernia prunastri, the effect of NH4 + was intermediate between NO3and gaseous NH3.  Figure S5a) and the calculated Eradication Dose 50 (ED50). Notes: ND, not determined where cover was stable or increased over time. The reciprocals of these values were set to zero ( a ).  Figure S5a: Response of Calluna vulgaris cover to addition of gaseous NH3 and wet deposition of NH4 + and NO3at rates of 8 (control), 16, 32 and 64 kg N ha -1 y -1 (averages of replicated plots).

Figure S5b
: Response of Sphagnum capillifolium cover to addition of gaseous NH3 and wet deposition of NH4 + and NO3at rates of 8 (control), 16, 32 and 64 kg N ha -1 y -1 (averages of replicated plots).

Figure S5c
: Response of Cladonia portentosa cover to addition of gaseous NH3 and wet deposition of NH4 + and NO3at rates of 8 (control), 16, 32 and 64 kg N ha -1 y -1 (averages of replicated plots). Figure S6: Estimation of the photosynthetic indicator Fv/Fm in transplants of two lichen species, exposed for up to 6 months at Whim Bog to wet deposition of NH4 + or NO3and to dry deposition of NH3. Total deposition is shown based on transplant exposure-time, considering the timing of preceding wet and dry N release periods. Least square regressions were fitted as Fv/Fm = a ln(N input) + b. A.
B. Figure S8: Surface pH of Sphagnum capillifolium and Cladonia portentosa measured along the NH3 transect at Whim Bog, shown in relation to distance from source (A) and NH3 concentrations (B). For Sphagnum, 10 measurements were made per location on in situ plants (July 2018). For Cladonia, specimens were transplanted into the transect on 16 August 2018, with pH measurements on 6, 12 September 2018, 11 October 2018 and 24 September 2019. Values for Cladonia are means and standard deviations of the four sampling periods (based on triplicate measurements at each distance).