Introduction

Penguin rookeries and other colonies of marine animals are major sources of nitrogen (N) and phosphorus (P) input to Antarctic coastal ice-free terrain (Tatur 2002; Riddick et al. 2012). These create hotspots of eutrophication where aerosol and volatilized ammonia (NH3) generated from marine-derived excreta can enrich surrounding areas through atmospheric dispersion and deposition processes in addition to localized surface run-off. Numerous reports describe how lichen and plant communities are modified by the abundance of excreta generated in penguin rookeries (Beyer et al. 2000; Leishman and Wild 2001; Tatur 2002; Smykla et al. 2007). Active nesting areas are usually devoid of plants and lichens due to a combination of heavy physical disturbance by trampling and the toxic effects of fresh faecal material. However, at the perimeters of rookeries distinct communities frequently develop of algae, cyanobacteria, lichens (e.g. members of the Teloschistales) and mosses and, in the maritime Antarctic, vascular plants (Smith 1972, 1988; Smykla et al. 2007), that are tolerant of eutrophicated conditions. This visually evident effect of eutrophication is largely restricted to a zone 100–500 m beyond rookery perimeters. However, volatilized NH3 can be dispersed to much greater distances.

Wodehouse and Parker (1981) demonstrated an NH3 concentration ([NH3]) gradient in air along a 200 m transect downwind of an Adélie penguin rookery occupied by c. 6 × 103 breeding pairs of birds on Anvers Island on the Antarctic Peninsula. Similarly, several authors document higher [NH4 +] values in snow and air sampled at sites near to penguin rookeries than in comparable samples from localities remote from rookeries (Greenfield 1992; Legrand et al. 1998; Park et al. 2007; Nędzarek and Rakusa-Suszczewski 2007) and Crittenden (1998), working in the Windmill Islands, recorded an increase in [NH4 +] in sequential samples of meltwater from freshly fallen snow during a c. 6 day period of snow melt and suggested that this might have been a result of dry deposition during the melt period of NH3 from rookeries at a distance of ≥2 km. Lindeboom (1984) studied N flow through penguin rookeries on subantarctic Marion Island and reported that the odour of NH3 was apparent at distances of up to 10 km from the source. Moreover, Erskine et al. (1998) consider that much of the N capital on subantarctic Macquarie Island could have originated from NH3 volatilized from royal penguin rookeries over periods of thousands of years. Dispersion of NH3 from rookeries can, therefore, potentially have an influence over large areas surrounding the emission source. However, detailed estimates of NH3 dispersion from a penguin rookery have not been previously published. While there is a dearth of information on N deposition rates in Antarctica, it is probable that at sites remote from animal colonies such rates [and also those of biological N2-fixation e.g. Wynn-Williams (1990), Whitton and Potts (2000)] are very low. For example, [NO3 −] and [NH4 +] values in falling snow (0.3–16.2 µmol l−1, Maupetit and Delmas 1992) or freshly deposited snow (0.5–7.6 µmol l−1, Crittenden 1998; Mulvaney et al. 1998) in the Antarctic are comparable with, or lower than, values in remote Arctic locations (e.g. Talbot et al. 1992; Walker et al. 2003).

Lichens are the major components of Antarctic vegetation both in terms of biomass and diversity (Longton 1988; Øvstedal and Lewis Smith 2001). Inorganic ions and some simple organic compounds such as amino acids are absorbed efficiently over the thallus surface from precipitation, dry deposits and overland flow. Lichens produce surface bound enzymes capable of hydrolysing peptides and organic phosphates thus releasing accessible N and P from organic deposits (Hogan 2009; Hogan et al. 2010a; Higgins and Crittenden unpublished data). Lichens typically occur in habitats in which available forms of N and P are scarce and many species are highly sensitive to N enrichment. In European settings this sensitivity is reflected in, for example, strong positive correlations between N deposition rates and thallus N concentration (Bruteig 1993; Hyvärinen and Crittenden 1998; Hogan et al. 2010a). In the terricolous heathland lichen Cladonia portentosa, N enrichment increases thallus N:P mass ratio, rate of phosphomonoesterase (PME) activity (Hogan et al. 2010a, b) and PO4 3− uptake efficiency (Hogan 2009). These shifts in enzyme and uptake capacities have been interpreted as physiological adjustments driving to maintain cellular N:P stoichiometry under conditions of relative N excess. Hence, while heavy eutrophication in the immediate vicinity of penguin rookeries might induce striking changes in the species composition of lichen communities, more subtle changes in lichen physiology might be used to indicate deposition of smaller but physiologically significant quantities of penguin-derived NH3 at more remote locations.

In addition, 15N natural abundance in penguin-derived NH3 might be sufficiently elevated to provide an isotopic signal that is useful in mapping NH3 capture by lichens in the surrounding terrestrial environment. This suggestion is prompted by (i) the elevated 15N natural abundance documented in penguin biomass and excreta compared to atmospheric N2 (Mizutani and Wada 1988), a condition that arises from the high trophic position of penguins and the process of trophic enrichment of 15N whereby 15N natural abundance in marine animals increases with increasing trophic status (Rau et al. 1992; Kelly 2000; Michener and Kaufman 2007), and (ii) the very high 15N natural abundance in ornithogenic rookery soils (e.g. Mizutani et al. 1985a, 1986) which is generated by discrimination against 15N–NH3 during volatilization (Högberg 1997; Frank et al. 2004) but which is likely to yield emissions of 15N-enriched NH3 despite this isotopic fractionation. The use of 15N natural abundance values in plants to trace N sources has frequently proved problematic due to spatial and chemical heterogeneity of soil N pools and the diversity of N capture strategies and N economies in plants (e.g. variation in rooting depth, mycorrhizal type and infection intensity, internal N recycling efficiency) (Handley and Scrimgeour 1997; Högberg 1997; Hobbie and Högberg 2012). By contrast, environmental circumstances relating to N capture in lichens are potentially far less complex: atmospheric deposition is spatially and chemically more homogeneous and N uptake strategies in lichens are probably considerably less variable than in plants. Lichens in remote background locations are often strongly 15N-depleted (Tozer et al. 2005; Huiskes et al. 2006; Fogel et al. 2008) rendering a local positive 15N signal easier to detect. Hence seeking a relationship between lichen chemistry and a point source of NH3 emissions is likely to be associated with fewer confounding factors than in plant/soil systems.

Here we report the results of an interdisciplinary study in which NH3 dispersion is quantified in the vicinity of a major Adélie penguin rookery at Cape Hallett in continental Antarctica and lichen physiological responses are used to determine the area of impact of NH3 deposition on the terrestrial biota. The results are timely given a growing body of evidence that Antarctic cryptogamic and soil microbial communities are frequently N limited (e.g. Wasley et al. 2006).

Materials and methods

Study site

Cape Hallett (72°19′S, 170°16′E) is situated at the southern end of Moubray Bay, northern Victoria Land, in the western Ross Sea at the northern tip of the Hallett Peninsula (Fig. 1a and Online Resource 1). An Adèlie penguin (Pygoscelis adeliae) rookery, estimated in 1998/99 to be occupied by 39,000 breeding pairs (Gordon 2003), is located on Seabee Hook, a recurved spit projecting c. 1,200 m west from the high rock ridge forming the Cape. Seabee Hook is c. 41 ha in total area, generally <5 m above sea level and covered by ornithogenic soils. A description of the vegetation at Cape Hallett is provided by Rudolph (1963) and Brabyn et al. (2006). Ridges, bluffs and gravel beaches around the perimeter of Edisto Inlet support scattered, locally large lichen populations including those of Buellia frigida, Umbilicaria decussata, and Usnea sphacelata. These species do not occur within 0.5 km of the penguin rookery; here putative N-tolerant species occur. These include Xanthomendoza borealis and species of Caloplaca, Candelariella, Physcia and Xanthoria, lichens that are typical of penguin rookeries and other eutrophicated sites elsewhere in Antarctica (Gremmen et al. 1994; Smith 1988; Øvstedal and Lewis Smith 2001). Several species of bryophyte also occur on gravel terraces and slopes within 0.5 km of the rookery.

Fig. 1
figure 1

Geographical detals of study area. a Location of sampling sites in the vicinity of Cape Hallett. Site numbering follows that in Table 1 which gives details of sampling undertaken at each location; the wind rose reports the relative frequency of wind directions at 10° intervals during January–May 2006. Insets show the location of Cape Hallett on the Antarctic continent and location of sampling sites on and adjacent to Seabee Hook (scale bar in inset 250 m); the asterisk marks the position of the automatic weather station. Dashed lines and whiskers indicate location of glaciers. b Simulated dispersion of ammonia from the Cape Hallett penguin rookery to concentrations ≥0.1 µg m−3 using the ADMS model. Light and dark blue areas indicate dispersion fields based on the high and low emission estimates, respectively (see “Materials and methods” section), and average wind frequencies during the entire measurement period (described by wind rose shown). (Color figure online)

Full size image

Ammonia sampling

Atmospheric ammonia concentration was measured at 11 sites in the vicinity of Cape Hallett: at the centre of Seabee Hook, at four locations on its perimeter (N, S, W and E), and at six locations on the perimeter of Edisto Inlet up to a distance of 15.3 km from the rookery (Fig. 1a; Table 1). Mean [NH3] values were determined at each site using Adapted Low-cost Passive High Absorption (ALPHA) samplers (Tang et al. 2001) mounted on wooden poles at 1.5 m above the ground. Samplers were exposed in triplicate at each site for three periods: 26 December 2005–10 January 2006, 11–17 and 17–23 January 2006 and at three sites [3, 5 and 6 (Table 1)] between 8 December 2004 and 24 January 2005. Measurements were made at all sites simultaneously. After exposure, the samplers were stored at between 0 and 5 °C until return to the UK for analysis following Tang et al. (2001) to yield mean [NH3] in air during each sampling period.

Table 1 Location of sampling sites at Cape Hallett and around Edisto Inlet, the type of sample collected and distance from rookery centre
Full size table

Bulk NH3 collection for isotopic analysis was undertaken at the centre of the rookery by drawing air sequentially through a Teflon pre-filter and a Whatman No 41 filter paper impregnated with citric acid. Air was drawn through the filters using a Capex 1D 12 V DC diaphragm pump (Charles Austin Pumps Ltd, Byfleet, UK) with a capacity of 15 l min−1.

Dispersion modelling

Mean hourly wind speed, wind direction (and standard deviation), air temperature, relative humidity and solar radiation data were recorded by an Automatic Weather Station (Fig. 1a). These data were used to predict mean [NH3] values resulting from the rookery emissions using three different dispersion models: ADMS (v4.1) (Carruthers et al. 1994; Hill 1998; Dragosits et al. 2002; CERC 2009), LADD (Dragosits et al. 2002) and the Lagrangian stochastic model of Flesch et al. (2004), which is implemented in the WindTrax software (V.2.0.8.3). The rookery was modelled as a ground–level area source and an arbitrary emission rate was assigned (10 tonnes NH3 ha−1 year−1). The simulation assumed a flat terrain; while this is unrealistic for dispersion across land where gradients of up to 75 % are present, it is a realistic assumption for short-range dispersion across the rookery and the sea. The complex terrain option of ADMS 4 was also tested in an attempt to better simulate the dispersion over elevated terrain. However, when using this option, the model makes some assumptions regarding atmospheric turbulence (such as setting limits on the atmospheric stability) that are not appropriate for the flatter regions of the domain, where most of the atmospheric measurements were located. For this reason, it was decided not to use the complex terrain option for the final simulations. Full details of the model parameterisations are given by Theobald et al. (2013). The predicted concentrations at the seven measuring stations closest to the centre of the colony were fitted to the measured values by applying a factor to reduce the geometric variance between predicted and measured concentrations (Chang and Hanna 2004). Since the concentration predictions are proportional to the emission rate used in the model, this factor can be used to estimate the emissions during each period. The factors obtained for the three dispersion models for the 2 measurement periods with the most reliable meteorological data (11–17 and 17–23 January 2006) ranged from 0.03 to 0.08, giving emission estimates between 0.3 and 0.8 tonnes NH3 ha−1 year−1 (Theobald et al. 2013). For a rookery occupation of 39,000 breeding pairs distributed over an area of 33.2 ha, these emission estimates are equivalent to an emission of between 0.8 and 1.8 g NH3–N per breeding pair per day.

Collection of ecological materials

Lichens were collected from 12 locations around the perimeter of Edisto Inlet up to a distance of 26.8 km from the rookery (Table 1; Fig. 1a): U. sphacelata from 8 locations, U. decussata from 5 locations and X. borealis from 4 locations. Lichens were cut from rocks with a scalpel. The terminal 10–20 mm of thallus branches of U. sphacelata were selected for analysis whereas in the cases of U. decussata and X. borealis whole thalli were analyzed. After collection, lichens were stored air-dry in polyethylene vials at ambient temperature while in the field and at 5 °C on return to the UK. Samples of the surface 5 mm of rookery soil were collected from guano-rich areas between nests. At each location, replicate samples of lichen or soil were taken from spots ≥10 m apart. Samples of Adélie penguin chick leg muscle were dissected from fresh kills by south polar skuas and freeze dried. Powder free latex gloves were worn at all times when handling these materials both in the field and in the laboratory to minimize contamination.

Determination of 15N natural abundance and total N and P concentrations

Lichen samples for total P analysis were oven-dried at 80 °C, weighed, digested using the sulphuric acid-hydrogen peroxide procedure (Allen 1989) and then phosphate assayed colorimetrically by the malachite green variant of the methylene blue method (van Veldhoven and Mannaerts 1987). Lichen samples and other materials (penguin muscle, soil and membrane filters) for isotopic analysis were oven-dried at 40 °C, reduced to powder in a ball mill, re-dried at 40 °C and weighed into ultra-clean tin capsules (Elemental Microanalysis Ltd., Oakhampton, UK) using a Cahn C-31 microbalance (Scientific and Medical Products Ltd., Cheadle, UK). 15N natural abundance was determined by continuous-flow elemental analyser-isotope ratio mass spectrometry (ANCA-SL coupled to 20-20 IRMS, Europa Scientific, Crewe, UK). The analytical protocol was adapted from that normally used for plant material (c. 2 % N) to cope with the much lower N content of some of the lichen materials analyzed (c. 0.4 % N). Samples of 20 mg were used and to ensure complete combustion the oxygen pulse was increased by 50 %. Batches of no more than 50 samples were analysed and the CO2 trap changed after each batch. The combustion ash was also removed more frequently than normal. Working standards containing 100 µg N, freeze dried from a leucine-citric acid mixture (~2 %N), were calibrated for δ15N against IAEA-N1 and IAEA-N2 (IAEA, Vienna). Precision for the working standards during a run was monitored and was routinely 0.3–0.4 ‰. Pairs of working standards were run with every ten samples. The in-line elemental analyzer also produced measurements of total N in the samples analyzed. Values of the ratio 15N/14N (R) are expressed in δ notation in per mil. units (‰) according to the following equation: δ15N = [(Rsample/Rstandard) − 1] × 1,000, where Rstandard is that of the atmosphere. The proportion of lichen thallus N that originated from penguin-derived NH3 (\(F_{{{\text{NH}}_{ 3} }}\)) was estimated using a two source mixing model (e.g. Fry 2006): \(F_{{{\text{NH}}_{ 3} }} = (\updelta_{\text{L1}} - \updelta_{\text{L2}} )/(\updelta_{{{\text{NH}}_{ 3} }} - \updelta_{\text{L2}} )\) where δL1, δL2 and \(\updelta_{{{\text{NH}}_{ 3} }}\) are, respectively, the δ15N values in lichen exposed to penguin-derived NH3, lichen from a background site remote from penguin influence, and penguin derived volatilized NH3.

Phosphomonoesterase assay

Phosphomonoesterase (PME) activity was assayed in the field using the p-nitrophenyl phosphate (pNPP) colorimetric method [Bessey et al. (1946) as modified by Hogan et al. (2010a)]. Air dry lichen samples (5–35 mg depending on species) were added to 2.9 ml assay medium comprising citric acid-trisodium citrate buffer made up in melted snow and the assays initiated by the addition of 0.1 ml analogous substrate solution. Samples were then placed in a water-bath at c. 5 °C (measured range 4.8–5.7 °C) for 30 min in the dark after which the reaction was terminated by transferring 2.5 ml assay medium into 0.25 ml terminator solution (1.1 M NaOH, 27.5 mM EDTA, 0.55 M K2HPO4) and the absorbance measured at 405 nm using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, USA). Post assay, thalli were blotted dry, oven dried for 24 h at 80 °C and weighed. Enzyme activity was expressed as μmol substrate hydrolysed g−1 dry mass h−1 using p-nitrophenol to calibrate the assay. No-lichen and no-substrate controls were included. Among the different experiments performed the pH of the bathing solution was varied between 2.5 and 6.9 and the substrate concentration between 0.25 and 8.0 mM.

Adding air dry lichen samples directly to the assay medium will produce a small over-estimation of activity due to water entering the non-free space (symplast) of the thallus. Using Beckett’s (1995) estimates of the free-space volume in lichens from xeric habitats (a mean of 32 % of thallus volume), we predict that this over-estimation was probably in the range of 0.2–1.0, 0.2–1.8 and 0.4–1.2 % in U. sphacelata, U. decussata and X. borealis, respectively, being dependent on the mass of lichen sample being assayed. It is widely assumed that the pNPP assay detects cell wall-bound PME activity because the hydrolysis product, p-nitrophenol, accumulates in the assay medium at a rate that is constant and pH dependent (Bartlett and Lewis 1973).

Data analysis

Data for atmospheric variables, lichen chemistry, PME activity and distance from the rookery were subjected to either correlation analysis, linear regression or one-way ANOVA conducted in Sigma Plot 11 (Systat Software Inc.). All data were checked for normality and homogeneity of variances and where test assumptions were not met, either log-transformation or non-parametric correlation analysis were applied.

[NH3] predicted by the NH3 dispersion model was tested as a quantitative predictor and lichen species as a categorical predictor of \(F_{{{\text{NH}}_{ 3} }}\) in a general linear model [GLM in R (R Core Team 2014)] using sequential deletion. The residuals were normally distributed (Shapiro–Wilk W = 0.95, P = 0.44) and so significance was assessed using F tests.

Results

Raw data have been deposited with the Polar Data Centre, British Antarctic Survey, Cambridge, UK. doi: http://doi.org/v27.

Ammonia concentrations and δ15N values

Mean [NH3] values of up to 75 µg m−3 were recorded at the rookery centre while values in the range 6–81 µg m−3 were recorded on its perimeter (sites 2–5) where they were weakly but significantly (P ≤ 0.05) correlated with wind direction (data not shown). Concentrations declined exponentially with distance from the rookery; the lowest concentration recorded was 0.05 µg m−3 at 15.3 km and the lowest mean concentration that is significantly greater than this value is 0.12 µg m−3 estimated to have occurred at 8.9 km (Fig. 2a). Note that the maximum mean [NH3] values recorded were insufficient to saturate the passive samplers during a 2 week exposure period (Tang et al. 2004); [NH3] values >200 µg m−3 would be necessary for saturation to be approached (Tang, personal communication). Ammonia captured on filter paper had a mean δ15N value of +6.94 ± 2.26 ‰ (mean ± SE, n = 4). This can be compared with mean values for guano deposits and penguin chick muscle of +20.32 ± 1.33 (n = 10) and +8.29 ± 0.08 ‰ (n = 10), respectively.

Fig. 2
figure 2

Relationships between distance from the rookery centre and a NH3 concentration in air, b lichen δ 15N values, c total thallus N (open symbols) and P (closed symbols) concentrations, and d lichen PME activity. Lichen species: U. sphacelata (black circle, white circle), U. decussata (white up-pointing triangle, black up-pointing triangle) and X. borealis (black square, white square). Points are mean values ± 1 SE: a n = 3; b–d n = 1–19 (note that in a SE bars do not exceed the diameter of the points). Where appropriate, regression lines are shown (solid line) together with 95 % confidence intervals (dashed line): a, r 2 = 0.87, P = < 0.001; b U. sphacelata, r 2 = 0.76, P = 0.005 [note that the regression of δ 15N values for all species combined on log10 distance is also significant (r 2 = 0.57, P = < 0.001)]; c, no regression line shown but the log–log relationships for all species combined is significant (N, r 2 = 0.60, P < 0.001; P, r 2 = 0.49, P = 0.002); d, U. sphacelata, r 2 = 0.996, P = 0.042. 95 % confidence intervals are used to estimate the lowest elevated value of the dependant variable (dotted lines) compared to the best estimate of background value, and the distance at which this is predicted to occur

Full size image

Lichen and soil chemistry

15N natural abundance in lichens was significantly positively related to proximity to the rookery (Fig. 2b). At Football Saddle, 27 km southwest of Seabee Hook, the mean δ15N values for U. sphacelata and U. decussata were −15.5 ± 0.7 (n = 4) and −17.5 ± 0.3 ‰ (n = 19), respectively, increasing to −2.1 ± 0.5 (n = 9) and −7.8 ± 0.6 (n = 11), respectively, within 1-2 km of the rookery centre. By contrast, δ15N values for X. borealis ranged between −0.9 ± 1.2 (n = 10) to +9.0 ± 0.9 ‰ (n = 9). Thallus δ15N signature was strongly related to [NH3] when data for lichens from the three sites at which both [NH3] in air and δ15N in lichens were measured were combined (r 2 = 0.93, n = 4, P = 0.008). Assuming that δ15N values for U. sphacelata and U. decussata at 27 km represent the background values for populations uninfluenced by bird-derived NH3 or excreta and that the effects of other processes that might modify δ15N signatures are assumed to be negligible (see discussion below), then the two-source mixing model suggests that up to 60 % of thallus N in U. sphacelata, and up to 40 % in U. decussata could have been derived from NH3 at sites closer to the rookery. Applying a mean background value of −16.5 ‰ to X. borealis suggests that 66–100 % of thallus N could have originated from penguin-derived NH3. Note that the mixing model assumes that the quantity of thallus N originating from penguin-derived NH3 is 0 and 100 % when δ15N values in, e.g. U. sphacelata, are −15.5 and +6.9 ‰, respectively. In the general linear model to test the relationship between \(F_{{{\text{NH}}_{ 3} }}\) and [NH3], both modelled [NH3] (see below) and species identity were significant predictors (Table 2, Fig. 3). Nearly all the differences among species obviously arose because the mean for Xanthomendoza borealis was higher than the other two taxa. Modelled [NH3] and lichen species identity accounted for 72 % of variation in the \(F_{{{\text{NH}}_{ 3} }}\) values.

Table 2 General linear model statistics for predictor variables in the model relating \(F_{{{\text{NH}}_{ 3} }}\) in U. sphacelata, U. decussata and X. borealis to [NH3] (as predicted by the dispersion model) and lichen species identity
Full size table
Fig. 3
figure 3

Relationship between measured \(F_{{{\text{NH}}_{ 3} }}\) in the three lichen species studied and values modelled from [NH3] as a quantitative predictor and species identity as a categorical predictor (Table 2). Plotted y values are means ± 1 SE (n = 1–19). Lichen species: U. sphacelata (white circle), U. decussata (white up-pointing triangle) and X. borealis (white square) (R 2 = 0.72). Data are for 11–17 January 2006. Note that data from Site 8 for X. borealis were significant outliers and were omitted from the analysis

Full size image

Thallus N and P concentrations were also significantly related to proximity to the rookery when data for the three lichen species were combined (Fig. 2c). However, this relationship is strongly influenced by high concentrations in X. borealis (N concentration c. 3–7 times higher, and P concentration c. 5 times, than values in U. sphacelata and U. decussata) and there were no clear relationships between thallus N and P concentrations and distance from the rookery in any of the three species examined separately. In all species there was marked positive correlation between δ15N values and P and N concentrations (Table 3; Fig. 4a), and a trend for negative correlation between δ15N values and N:P mass ratio (Table 3).

Table 3 Bivariate Pearson correlation coefficients (r) between δ15N values, concentrations of thallus N and P ([N], [P]) and N:P mass ratio in U. sphacelata (Us), U. decussata (Ud) and X. borealis (Xb)
Full size table
Fig. 4
figure 4

Relationships in U. sphacelata (white circle), U. decussata (black up-pointing triangle) and X. borealis (white square) between a δ 15N value and total P concentration (see Table 3 for statistics), and b PME activity and thallus N:P mass ratio [mean values plotted ± 1SE (n = 10–11); note that the probability of obtaining a progressive increase in PME activity with increasing N:P ratio in all cases by chance is 1 in 24]

Full size image

The guano rich surface soil sample had a 15N enrichment of +20.3 ± 1.3 ‰, and a total N concentration of 3.4 ± 0.5 %.

Phosphomonoesterase activity

Phosphomonoesterase activity was readily detected in each of the three lichen species assayed. There was a significant effect of assay pH on activity in X. borealis with an apparent optimum at pH 6.6 but not in either U. sphacelata or U. decussata although there was a trend for higher rates at lower pH values (Online Resource 2). Accordingly, subsequent assays were performed at pH 6.6, 2.5 and 3.0 for X. borealis, U. sphacelata and U. decussata, respectively. PME activity was highest in U. sphacelata and responded strongly to substrate concentration up to 8 mM; material collected at 10 km from the rookery had K m and V max values, calculated from a Hanes-Woolf plot, of 2.03 mM and 110 µmol substrate hydrolyzed g−1 dry mass h−1. A saturating substrate concentration of 8 mM pNPP was used in subsequent assays.

In U. sphacelata there was a strong positive relationship between proximity to the penguin rookery and PME activity (Fig. 2d); enzyme activity at 2 km from the NH3 source was 5 times higher than that in samples at 15 km. Similarly, activity in U. decussata at 2.4 km was 3.4 times higher than that in samples from 10 km. By contrast, activity was low in X. borealis at both sites which were <1 km from the rookery centre. PME activity was also positively related to thallus N:P mass ratio (Fig. 4b).

Ammonia dispersion model

The dispersion model was used to predict the area surrounding the emission source in which [NH3] values were ≥0.1 µg m−3 (Fig. 1b). This concentration approximated to that associated with the estimated minimum significant elevation of δ15N and PME activity in U. sphacelata, changes which are used here as evidence of impact. This impacted area was 35 and 270 km2 for the low and high emission estimates, respectively (see “Dispersion modelling” section above), during the entire measurement period, providing estimates of the dimension of the rookery’s potential impact zone. The simulations used the lower range of the dry deposition velocity from Theobald et al. (2013) since this gave the best spatial correlation between the modelled and measured [NH3]. This suggests that the dry deposition rate to sea ice is lower than might be expected. At Cape Hallett, the influence of NH3 dispersion on the terrestrial environment is substantially reduced because, due to the prominent south-westerly winds, much of the NH3 plume falls over the sea (ca 90 % based on the frequency of winds from the 40°–350° sector during the [NH3] measurement period). It should be noted that although the dispersion models are more uncertain when applied to such locations with complex topography, the model simulations should still provide an estimate of the order of magnitude of the impacted area.

Discussion

Numerous reports have described the impacts of penguin rookeries on Antarctic vegetation (e.g. Beyer and Bölter 2002; Smykla et al. 2007). However, this is the first study to determine the spatial extent of NH3 dispersion from a rookery and its effects on terrestrial biota. At Cape Hallett, lichen species such as X. borealis and Xanthoria elegans, and the moss Bryum subrotundifolium occur within the immediate vicinity of the rookery but are infrequent or absent beyond 1 km from the rookery perimeter. These apparently nitrophilous species demarcate a zone of intense eutrophication. However, the present study shows that the influence of the rookery extends far beyond this zone of marked community change. Our data suggest that NH3 deposition has modifying effects on lichen chemistry and physiology to a distance up to 10 km inland (upwind according to the predominant wind direction, south south-westwards). Measurements of 15N natural abundance indicate that a major proportion of thallus N could originate from penguin-derived NH3, the proportion potentially remaining as high as 20 % in U. sphacelata at c. 13 km, and \(F_{{{\text{NH}}_{ 3} }}\) estimates were strongly related to modelled [NH3] (Table 2 and Fig. 3). Similarly, PME activity in U. sphacelata was significantly elevated to a distance of c. 10 km inland (Fig. 2d). These measurable changes in δ15N and PME activity were detectable down to an [NH3] value of c. 0.1 μg m−3 and, hence, we selected this concentration to define the boundary of the modelled NH3 dispersion envelope (Fig. 1b). This model implies that [NH3] values ≥0.1 μg NH3 m−3 occur over an area of up to 300 km2 surrounding the emission source. A level of 0.1 μg NH3 m−3 might seem low. For example, the proposed Critical Level of NH3 for lichen and bryophyte rich vegetation in Europe is 1 μg m−3 (Cape et al. 2009). However, the high average wind speeds (e.g. Bargagli 2005) and attendant turbulence over rough terrain in Antarctica will result in comparatively high NH3 deposition rates to terrestrial surfaces even at low atmospheric concentrations while reference conditions in Antarctica represent an extremely oligotrophic situation allowing comparison to lower [NH3] levels. The value of 0.1 μg NH3 m−3, which we consider to be ecologically significant at Cape Hallett, can be compared with 0.005–0.03 μg NH3 m−3 recorded at Antarctic sites remote from animal colonies (Gras 1983) and a mean value of 0.06 NH3 m−3 recorded over the Southern Ocean (Ayers and Gras 1980). It should be noted that ammonia emissions from Adèlie penguin rookeries are likely to be strongly seasonal since they are mostly occupied in the summer months [mid-October to mid-February (Gordon 2003)] and NH3 volatilization rates in frozen soil are extremely low (Zhu et al. 2011). On the other hand, there is little inter-seasonal or inter-annual variation in the frequencies of wind directions over Cape Hallett (Gordon 2003) so that there is likely to be little temporal change in the NH3 dispersion pattern. The average tropospheric lifetime for NH3 is c. 0.9 days but in the pure air of Antarctica with low concentrations of potentially reactant gases (e.g. sulphur dioxide) and aerosols (Adams et al. 1999) the lifetime could be longer.

The δ15N values presented here for penguin tissues, guano-rich soil, atmospheric NH3 and lichens compare favourably with those documented in the literature for broadly comparable samples (Table 4). There is a dearth of published data on 15N natural abundance in atmospheric NH3. The three positive δ15N values reported in Table 4 (+2.4 to +6.9 ‰) were for air samples drawn through either H2SO4 or acid impregnated filters, whereas the negative δ15N value (−10.0 ‰) recorded by Erskine et al. (1998) was for NH3 collected passively relying on diffusion into inverted jars which could have greatly underestimated [15N–NH3] due to isotopic fractionation during diffusive transfer (Högberg 1997; Tozer et al. 2005). The positive δ15N values for NH3 at Cape Hallett likely reflect very high 15N enrichment in the guano/soil matrix resulting from the cumulative effects of discrimination against the heavier isotope during NH3 volatilization over periods of years and perhaps decades, δ15N values in the range +33 to +50 having been reported for NH4 +, NH3 and NO3 − in bulk samples of Adelie rookery soils (Mizutani et al. 1985a, 1986; Mizutani and Wada 1988; Zhu et al. 2009). Emission of 15N-depleted N2O resulting from denitrification, although at rates that are probably at least 2 orders of magnitude lower than those for NH3 release, will contribute to this process further (Zhu et al. 2008a, b). Isotopic fractionation for NH3 volatilization is 30–40 ‰ (Mizutani et al. 1985b; Högberg 1997; Frank et al. 2004). Applying this degree of fractionation to soil N with the above levels of 15N enrichment, assuming that these are representative of NH3 in the soil solution at the soil/air interface, yields δ15N values for volatilized NH3 of −7 to +20 encompassing our four measurements in the range +1.2 to +10.9. An alternative approach to understanding this question is as follows. If all N in a unit of excreta were volatilized as NH3 (plus some N2O) then the volume weighted average δ15N value of the emissions would be the same as that of the faeces (in this case c. +8 ‰), fractionation merely creating transient changes in δ15N values. Since the rookery is >1,000 years old (Gordon Gordon 2003) and has therefore received a long series of annual inputs of excreta ranging from fresh faeces in the current year to those sufficiently old for the N capital to be close to exhausted, it is probable that the volatized NH3 now has a δ15N signal close to the average value. The low δ15N values for lichens at remote sites imply that atmospheric supplies of non-biogenic N taken up by lichens are generally highly 15N depleted. This view is consistent with the low δ15N values for NO3 − in Antarctic Dry Valley soils [−9 to −26 ‰, Wada et al. (1981)], which Michalski et al. (2005) consider result from 15N-depleted NO3 − in precipitation, for NO3 − in Antarctic atmospheric aerosol [−50 to +5, (Wagenbach et al. 1998)] and for NO3 − in fresh snow on Svalbard [−18 ‰, Heaton et al. (2004)]. The particularly high 15N enrichment in X. borealis (δ15N = −0.9 to +9.0 ‰) reflects its preference for eutrophicated habitats close to the rookery perimeter; note that a mean value >+6.9 for X. borealis at Site 7 suggests that at such sites deposition of particulate matter, perhaps as aerosol generated by penguin physical activity in a windy environment, is influencing lichen 15N signatures in addition to NH3. δ15N values for X. borealis can be compared with those for Caloplaca eudoxa (δ15N = +3.6 to +20.6) growing on the Namib Desert coast in immediate proximity to a major seal colony (Crittenden et al. unpublished data). Accordingly, our δ15N measurements at Cape Hallett, together with corroborative data from the literature, are consistent with an N transfer pathway penguins → soil → atmosphere → lichens.

Table 4 15N natural abundance values reported in the present work and documented in the literature for Adélie penguin tissues, Adélie guano and rookery soils, volatilized ammonia gas and lichens
Full size table

A key assumption in the calculation of \(F_{{{\text{NH}}_{ 3} }}\) is that δ15N values in lichen thalli mirror closely those in atmospheric N sources accessed by the thallus. However, several processes could modify this relationship. First, isotopic fractionation during diffusion of NH3 through air to the surfaces of both lichens and snow cover will result in discrimination against 15N entering the pool of NH3/NH4 + available for uptake at the thallus surface (Tozer et al. 2005). However, the turbulent nature of air flow in Antarctica will reduce diffusion path lengths over these surfaces and tend to minimise this effect. Second, fractionation in favour of 14N can occur during uptake in both fungi and plants (Högberg 1997, Hobbie and Högberg 2012). This effect is probably insignificant under N-limited conditions; given the high efficiency of inorganic N capture by lichens (Crittenden 1998), it is probable that most NH3/NH4 + molecules in the thallus free space will be taken up. However, it could become significant at eutrophicated sites if N is available in excess of uptake. Third, in mat-forming lichens that produce copious dead basal thallus or “litter” (e.g. species of Cladonia subgenus Cladina) there are strong gradients in δ15N values from the growing apices downwards towards the thallus bases; it is thought that these gradients are related to thallus development, age and internal N recycling (Ellis et al. 2003, 2005). The little available evidence suggests that such gradients are less marked in lichens with other growth forms (Ellis et al. 2003). Fourth, lichens lose traces of N in reproductive propagules and organic solutes leached during precipitation (Crittenden 1983, 1998), and further losses could occur due to “pruning” resulting from wind damage, but how these losses influence thallus δ15N values is unknown. The first two processes will lead to underestimations of \(F_{{{\text{NH}}_{ 3} }}\), the third would contribute to intra and inter species variation in \(F_{{{\text{NH}}_{ 3} }}\), but it is unlikely that any would invalidate the relationship between lichen δ15N values and distance from the rookery and that between \(F_{{{\text{NH}}_{ 3} }}\) and modelled [NH3].

The inferred positive correlation between [NH3] and PME activity in U. sphacelata and U. decussata (comparing Fig. 2a, d) is corroborated by data for C. portentosa in mainland Britain (Hogan et al. 2010a). PME capacity in this heathland lichen was higher at the most N-polluted heathlands (annual wet inorganic N deposition of 33 kg ha−1) by a factor of 2.3 compared with rates at background sites (4.1 kg N ha−1 year−1). In reciprocal transplants, activity increased or decreased to that of native lichen in the new environment within 6–12 months Hogan (2009) and, furthermore, the relationship between N deposition and PME activity was reproduced in a field manipulation experiment involving N- and P-enrichment treatments, thus demonstrating causality Hogan et al. (2010b). These findings for C. portentosa provide strong evidence to suggest that higher PME activity in U. sphacelata and U. decussata near to the penguin rookery is caused by greater N availability. Up-regulation of PME activity in response to N enrichment has been observed in a wide range of plant-soil systems [see Hogan et al. (2010a) for a discussion], a frequent explanation for which is that fertilization with N releases plants and microorganisms from N-limitation but in turn induces P-limitation. In C. portentosa, this link between PME capacity and the relative availability of N and P was also evident in striking positive relationships between PME capacity and thallus N:P mass ratio. Evidence for such a relationship between N:P mass ratio and PME capacity was also found in lichens at Cape Hallett (Fig. 4b).

Given the trend in U. sphacelata and U. decussata for increasing PME activity with greater proximity to the rookery, why was activity in X. borealis so low (Figs. 2d, 3b)? X. borealis is an ornithocoprophilous lichen growing near the rookery perimeter and exposed to aerosol and NH3 deposition rates that are probably toxic to U. sphacelata and U. decussata. Its total thallus N and P concentrations are up to 5 times higher than those in U. sphacelata and U. decussata, up to 100 % of its N capital could have been penguin-derived and it is probably N and P saturated. We suggest that moderated nutrient uptake rates and PME activities are likely to be advantageous traits in lichens growing in such a heavily eutrophicated habitat. Interestingly, similar differences in PME capacity have been found between eutrophication tolerant and intolerant epiphytic lichens in the British Isles (Lewis 2012).

Not all relationships between the measured variables and distance from the rookery were unambiguous. Despite evidence that a significant fraction of thallus N might be derived from volatilized NH3, total thallus N concentration was only weakly related to proximity to the rookery. An obvious explanation is that increased N uptake promotes lichen growth rate resulting in growth dilution of thallus N. An alternative explanation is that there might be significant additional local sources of bird-derived nutrient enrichment e.g. from south polar skuas, at some sites distant from the rookery. The strong relationship between δ15N and [P] values is consistent with this suggestion. However, bird colonies were not evident at lichen collection sites >1 km from the rookery and the link between δ15N and [P] could equally be due to long distance dispersal of penguin-derived P in the form of P-laden aerosol and/or phosphine (Zhu et al. 2006), and/or up-regulation of PME activity in N-enriched lichens. South polar skuas nest in the vicinity of the penguin rookery and snow petrels have been occasionally sighted; note that Cocks et al. (1998) showed that δ15N values in lichens are markedly elevated within 10 m of major snow petrel colonies. However, according to Gordon (2003) populations of both species in the study area are small (c. 100 skuas and one small colony of snow petrels on the West coast of Edisto Inlet) while sightings of other species such as giant petrel and Wilson’s storm petrel are infrequent. The effect of these birds, if any, is likely to have been a contributory factor to variability of lichen data, due to highly dispersed in-flight defecation, rather than a major confounding factor.

A high capacity for surface PME activity in an Antarctic lichen assayed at c. 5 °C is perhaps surprising. It is widely assumed that extracellular phosphatases promote the release of inorganic phosphate from organic P in the environment (usually soil) and, hence P uptake. The sources of organic P available to U. sphacelata and U. decussata in the unpolluted aerial environment might not be immediately obvious. However, Yoshioka et al. (2009) analyzed rainfall in Matsue, Japan, and found that the volume weighted concentration ratio of total P to soluble reactive P was on average 5:1 raising the possibility of a significant organic-P component. Hence, extracellular PME might increase the efficiency of P capture from trace organic solutes in precipitation. It is also possible that high surface enzyme activities in lichens are adaptations for exploiting infrequent, spatially unpredictable but highly eutrophicating events associated with deposition of animal excreta in otherwise oligotrophic areas. At Cape Hallett this would be excreta from bird species such as skuas and snow petrels flying at some distance from their nesting and roosting sites (Tomassen et al. 2005; Bokhorst et al. 2007).

There is uncertainty concerning the extent to which nutrients might limit production among Antarctic terrestrial microbial and cryptogamic communities. Earlier opinion, and one still reflected in some more recent literature, is that nutrients are generally not limiting (e.g. Longton 1988; Beyer et al. 2000; Robinson et al. 2003; Wasley et al. 2006; Block et al. 2009). However, there is a growing body of evidence deriving from nutrient amendment experiments suggesting that Antarctic soil bacterial and fungal communities are frequently N-limited (Davey and Rothery 1992; Dennis et al. 2013 and references therein]. Wasley et al. (2006) reported that addition of nutrients to lichen and bryophyte communities in the Windmill Islands, continental Antarctica, resulted in increased photosynthetic electron transport rate and total chlorophyll and thallus N concentrations in crustose lichens, Usnea spp and bryophytes and this was interpreted as indicating nutrient limitation in these communities [cf. Crittenden et al. (1994)]. The present work at Cape Hallett provides additional physiological evidence that Antarctic lichens respond strongly to the relative availabilities of N and P. Current warming trends in the Antarctic could result in a greater availability of both colonisable surfaces and water in seasonal ice-free terrain (Convey 2011) and there is speculation that expansion of existing lichen populations and possibly new local introductions might occur in the future (Frenot et al. 2005; Green et al. 2011; Chown et al. 2012). At the same time it is estimated that warming will lead to significantly increased NH3 emission from penguin and other seabird colonies (Sutton et al. 2013). Combining these two effects, it seems possible that NH3 dispersion fields centred on penguin rookeries and other large animal colonies where potential N-limitation is ameliorated (cf. Vidal et al. 2003; Casanovas et al. 2013, Haussmann et al. 2013) might be amongst those sites in which such changes are most likely to occur. Further afield there are many locations in the Arctic at which NH3 emitted from large seabird colonies is likely to influence otherwise N-limited terrestrial ecosystems (Wainright et al. 1998; Riddick et al. 2012; Zwolicki et al. 2013). However, Riddick et al. (2012) estimate that between 61 and 84 % of global NH3 emissions from seabirds occurs in Antarctica and islands in the Southern Ocean, most being attributable to penguins; these authors consider principal penguin rookeries to be the largest biogenic point sources of NH3 globally. Accordingly, NH3 impacts in the vicinity of penguin rookeries in particular merit further study.