1. Atmospheric deposition of nitrogen (N) is a global problem resulting in negative consequences for biodiversity due to direct toxicity, increases in invasive species, increased susceptibility to environmental stresses and soil-mediated effects of acidification and eutrophication.
2. Reductions in plant species richness related to N deposition have been observed in a number of habitats including calcifuge (acid) grasslands but the mechanisms of this decline have not been fully investigated. We test the hypotheses that along a large-scale gradient of N deposition there is (i) an increase in species tolerant of low pH conditions as a result of soil acidification and (ii) an increase in competitive and nitrophilic species as a result of soil eutrophication. As competitive species can occur in low pH habitats, both of these hypotheses could be true.
3. Using plant characteristics, we examined changes in vegetation species composition along the gradient of N deposition in the UK. Mean C–S–R signatures were used to identify the competitive response of plant communities together with Ellenberg N (nitrogen) scores to identify increases in nitrophilic species. Ellenberg R (reaction, pH) scores were used to identify change in response to soil pH together with an index of soil acidity preference developed using regional survey data.
4. Mean C–S–R signatures showed no significant correlation with N deposition, nor did mean Ellenberg N scores. Ellenberg R scores and the index of soil acidity preference showed significant relationships with N deposition indicating an increased dominance of acid-tolerant species.
5. The results suggest that soil acidification as opposed to eutrophication and consequent competition between species is contributing to shifts in species composition and diversity linked to N deposition in calcifuge grasslands. Soil acidification may be leading to reduced nutrient availability preventing the effects of N addition from being apparent.
The creation of reactive nitrogen (N) has increased globally by 120% since 1970 and continues to increase each year (Galloway et al. 2008). The global N cycle has now reached the point where more N is fixed annually by human-driven than by natural processes (Vitousek 1994). There are important consequences of this increase for deposition of reactive nitrogen. Deposition exceeds 10 kg N ha−1 year−1 in many areas of the world, 20 times more than the estimated deposition of 0·5 kg N ha−1 year−1 where there is no human influence (Galloway et al. 2008). About 10 kg N ha−1 year−1 is exceeded in large parts of Europe, with only remote regions, such as northern Sweden and Norway or northern Scotland, below this level.
The consequences of high N deposition for biodiversity are increasingly well-known. They include direct toxicity (e.g. van den Berg et al. 2008), increased susceptibility to pests and disease (e.g. Brunsting & Heil 1985), increased susceptibility to environmental stresses (e.g. Caporn, Ashenden & Lee 2000) and soil-mediated effects of acidification and eutrophication. Amongst these, acidification and eutrophication are believed to be the most important and may drive other changes listed above (Bobbink, Hornung & Roelofs 1998).
Eutrophication can lead to an increased vigour of potentially robust, nutrient-responsive species (competitors, sensuGrime 1974). These species are better able to take advantage of the additional nitrogen and out-compete the smaller, stress-tolerant species for nitrogen and other resources, including light and water, leading to an overall reduction in species richness as a result of competitive exclusion and reduced recruitment (Stevens et al. 2004a). The most dramatic species losses are likely to occur in communities initially at low–intermediate levels of nutrient availability (Al-Mufti et al. 1977) and will not be as strong where systems are N-saturated. Competitive exclusion and reduced recruitment has been supported by reported increases in sward productivity (e.g. Tilman 1993; Wilson, Wells & Sparks 1995) and increases in species typical of more fertile conditions (e.g. Kirkham & Kent 1997) following experimental fertilization.
Atmospheric N deposition can also result in soil acidification both directly as a result of acid deposition (nitric acid) and indirectly through processes and reaction in soil and water. The other potential cause of species loss is a reduced species pool as a result of increased soil acidity and consequent mobilization of metals, loss of base cations and changes in the balance between nitrogenous compounds (Jefferies & Maron 1997; Crawley et al. 2005). Few species are commonly found on soil with pH below 4·5 (Grime & Hodgson 1969; Grime, Hodgson & Hunt 2007), such that if soils are acidified below this pH, the potential species pool is considerably reduced. This pH threshold coincides with the aluminium (Al) buffering range where Al ions become increasingly available in the soil (Magistad 1925). A number of heavy metals behave in a similar manner to Al, becoming more biologically available in the soil as soil becomes more acid. Once mobilized into solution, Al can be directly toxic to plants, inhibiting root extension and resulting in abnormalities in lateral roots (Andersson 1988). Additionally, at low pH, nitrate uptake is reduced by free Al3+ and this increased Al can have a toxic effect on mycorrhizal symbionts (e.g. Lazof et al. 1994). Other macro- and micronutrients including calcium and magnesium (base cations) are commonly found at higher levels in higher pH soils and are among the first ions to leach from the soil as pH declines. A key role for reductions in soil pH in the mechanisms of declining species richness is suggested by the association between acid deposition, reduced soil pH (e.g. Skiba et al. 1989; Blake et al. 1999; Stevens et al. 2006; Horswill et al. 2008) and increased availability of metals (e.g. Roem, Klees & Berendse 2002; Stevens, Dise & Gowing 2009).
The extent to which each of these mechanisms is responsible for reductions in species richness related to N deposition is not known, but is essential to our understanding of N-deposition impacts. Here, we use data from a national survey of acid grasslands across the UK to test two hypotheses along a gradient of atmospheric N deposition, where declines in species richness have been observed (Stevens et al. 2004a,b). The hypotheses are that there is an increase in species tolerant of low pH conditions as a result of soil acidification and an increase in competitive and nitrophilic species as a result of soil eutrophication. There are a number of potentially competitive species found growing in these habitats so either or both of these mechanisms could be responsible for species changes. Because pH and enrichment of soil N are both potentially related to N deposition and are highly correlated with each other within calcifuge grasslands (Stevens et al. 2004a,b), we use plant community characteristics to determine whether the plant community is responding to changes in pH, nutrient enrichment or both. The data used in this study come from 68 grasslands with species richness ranging from 6 to 27 species per 2 × 2 m quadrat. The results of earlier investigations showed a significant decline in species richness related to increasing N deposition along the gradient of deposition found in the UK (Stevens et al. 2004a,b). Changes in species composition, reduced pH, increased ammonia concentration (Stevens et al. 2006) and changed soil metal concentrations (Stevens, Dise & Gowing 2009) were also observed with increasing N deposition.
Materials and methods
Calcifuge (acid) grasslands belonging to the community U4 Festuca ovina–Agrostis capillaris–Galium saxatile grassland (Rodwell 1992) (closely allied to the Violion caninae association described in Schwickerath 1944) were surveyed along the gradient of N deposition found in Great Britain. Sixty-eight sites were selected from Natural England, Countryside Council for Wales and Scottish Natural Heritage databases using stratified random sampling to cover the range of N deposition in the UK. Grasslands surveyed consisted primarily of protected areas including Sites of Special Scientific Interest (SSSI), National Nature Reserves (NNR) and National Parks. Selected sites were in both upland and lowland areas and none was artificially fertilized.
Deposition of inorganic N (oxidized and reduced) and acid deposition (N deposition plus sulphur deposition) was modelled by the Centre for Ecology and Hydrology (CEH) in Edinburgh, using the CEH National Atmospheric Deposition Model (Smith et al. 2000). This provides values for both wet and dry deposition and oxidized and reduced deposition at a 5-km resolution. Total inorganic N deposition at the surveyed sites ranged from 6·2 to 36·3 kg N ha−1 year−1.
All of the sites were surveyed during the summers of 2002 and 2003, beginning in the south of England and moving north. Full details of the field sites are given in Stevens (2004). A full description of each site was made, including location (latitude/longitude), altitude, aspect, slope, grazing intensity (on a scale of one to three estimated by eye based on vegetation height and amount of animal faeces) and the presence of any enclosures. Vegetation at each site was surveyed using five 2 × 2 m quadrats randomly placed within a block 1 ha in extent, excluding areas of different vegetation communities, footpaths or tracks and areas of intense animal activity such as adjacent to feeding or water troughs. Sites with large point sources of ammonia in the vicinity were avoided. Site characteristics are given in Table 1. All higher plants and bryophytes were identified to species level and their percentage cover, estimated by eye, was recorded. These data were then compiled to give a species list for each site.
Table 1. Site characteristics of the 68 sites surveyed
Total N deposition (kg N ha−1 year−1)
Total S deposition (kg N ha−1 year−1)
C : N (by mass)
Mean species richness (per 2 × 2 m quadrat)
Species lists from each site were used to calculate mean cover-weighted and unweighted Ellenberg R (reaction – soil pH) and N (soil nutrient) scores (Ellenberg et al. 1991) using scores recalculated for plants in Great Britain (Hill et al. 1999, 2007) and mean C–S–R signature (Hunt et al. 2004; Grime et al. 2008) for the community. Ellenberg scores are indicator values based on Ellenberg’s expert opinion based on his own experiments and research to give a plants’ realized ecological niche (Hill et al. 1999). C–S–R scores provide a classification of the external factors that affect vegetation grouped into two classes: stress (i.e. phenomena that restrict production) and disturbance (i.e. phenomena that destroy biomass). C–S–R scores are derived from vegetation surveys, knowledge of plant functional traits and field and mesocosm experiments conducted by the Unit of Comparative Plant Ecology at Sheffield University (Grime et al. 2008). An index of soil acidity preference was also calculated using data for over 600 sites from Grime & Lloyd (1973). Although providing similar information to the Ellenberg score, this provides a continuous scale and is tailored specifically to the threshold pH of 5. The index of soil acidity preference was calculated as the proportion of sites on soils with pH of 5 or below each species occurred on compared to the proportion of sites with soils of a pH above 5 using the following formula:
where a is the number of sites with soil of pH 5 or below on which the species occurs, and b the Number of sites with soils of pH above 5 on which the species occurs
This is the probability of a species occurring on a soil with a pH <5 relative to the probability of a species occurring on any of the sampled grasslands. A score of 0 indicates a strong preference for soil with a pH above 5, a score of 1 indicates a strong preference for soil with a pH of 5 or below, and a score of 0·5 indicates no preference. pH 5 was selected as the best value for use in this index because it is close to the point of aluminium mobilization and the division represented the largest change in species preferences within this range. For example, Calluna vulgaris, a strongly calcifuge species, has a score of 0·98 whereas Bellis perennis, which is not found on the most acid soils, has a score of zero. Values for the acid preference of species included in this survey and some further examples can be found in the Supporting Information. Where there were no data on the distribution of a species, this species was not included in the calculation.
For all scores, means were calculated without cover-weighting. Data were analysed using Pearson correlation coefficients, simple and forward stepwise multiple regression analysis in spss v17.0 (SPSS Inc., Chicago, IL, USA) with nitrogen deposition as the independent variable, and mean Ellenberg, CSR and soil acidity preference index scores as dependant variables. In the stepwise multiple regression latitude, longitude, altitude, topsoil pH, mean monthly rainfall, mean annual temperature, slope and S deposition were used as independent variables with the N deposition included as an independent variable in all models, and mean Ellenberg, CSR and soil acidity preference index scores were the dependant variables.
For all of the scores used, the best-defined relationships were with unweighted scores as opposed to the cover-weighted scores. This may be because the dominance of a few grass species in this community meant that differences in scores between sites were reduced. Forbs of small stature and cover, common in this community, may also have been insufficiently represented.
Correlations between the indices examined in this study showed that Ellenberg N and R scores for sites were closely correlated (r = 0·83, P < 0·001) but there were no other significant correlations between indices (Table 2). Because C, S and R scores are interrelated, only C scores were considered.
Table 2. Correlation coefficients and significance values for comparison between indices
Ellenberg R and Ellenberg N
Ellenberg R and Acid Index
Ellenberg R and C score
Ellenberg N and Acid Index
Ellenberg N and C score
Acid Index and C score
Despite compositional shifts in the vegetation (reported in Stevens et al. 2004b,a), very little evidence were obtained of functional shifts in the vegetation. Calculating mean C–S–R signatures for each site shows no significant change in the S score (stress tolerance) with increasing N deposition (P = 0·11) nor any significant increase in the C score (competitors) (P = 0·81). There was also no significant change in the R score (P = 0·06; Fig. 1). Results for acid deposition were very similar showing no change in C or S score but with a weak but significant negative relationship between ruderals and acid deposition (r2 = 0·07, P = 0·01). C–S–R scores range from 0 to 1 so the changes in competitor scores are small (range 0·12–0·36) indicating that none of the communities is dominated by competitive species. There is a larger range of S scores (0·36 to 0·79) but the range of R scores is also low (range 0·07–0·34). Adding other measured variables (latitude, longitude, altitude, topsoil pH, mean monthly rainfall, mean annual temperature, slope and S deposition) into the regression equations improved the explanatory power of the relationships. The relationship between the C score and N deposition was improved by the addition of soil pH, altitude and latitude resulting in a model that accounted for 11% of the variation in the C score (P < 0·05). The relationship between S score and N deposition was improved by the addition of S deposition, latitude, longitude, altitude and soil pH to the model resulting in 40% of the variation in S score being accounted for (P < 0·001). The addition of the same variables also increased the explanatory power of the model for the R score resulting in 44% of the variation in R score being accounted for (P < 0·001). A summary of these models is provided in the Supporting Information.
Mean Ellenberg N scores revealed a similar result to C scores. There was no significant correlation between Ellenberg N score and N deposition (r2 = 0·03, P = 0·14; Fig. 2a). Ellenberg N scores ranged from 2·3 to 4·7 representing a shift from indicators of infertile sites towards indicators of intermediate fertility. Mean Ellenberg R scores showed a weak but significant negative correlation with N deposition (r2 = 0·06; P = 0·02), although very little of the variation is explained (Fig. 2b). There were no significant relationships between Ellenberg scores and acid deposition (R: P = 0·35; N: P = 0·12). The addition of other variables measured in this study did not improve the relationship between N deposition and Ellenberg scores. Ellenberg R scores ranged from 3·3 to 5·3 representing a shift from acidity indicators found mainly on acid soils to indicators of moderately acid soils where N deposition was lower. The magnitudes of changes using Ellenberg scores were very small. This is possibly because Ellenberg scores provide a discontinuous scale. As changes may be only subtle as a result of N deposition and the pH range of the grasslands in this investigation is small, the scale may not be sensitive enough to detect changes that are occurring. Despite this limitation, Ellenberg N scores have been used to identify changes associated with N deposition in a number of other habitats [e.g. Bennie et al. 2006 (calcareous grasslands); Smart et al. 2005 (infertile grasslands and moorland); Pitcairn et al. 2002 (woodland)]. Ellenberg scores are system-dependent to some extent (Wamelink et al. 2002), but as this study was conducted in a single habitat and region, this problem is reduced.
The correlations using the soil acidity index yielded similar results as those using the Ellenberg R scores, showing a significant correlation (r2 = 0·23, P < 0·001) between index score and N deposition with the species present at high N deposition showing a stronger preference for acid habitats (Fig. 3). The relationship between index score and acid deposition is very similar to that with N deposition (r2 = 0·28, P < 0·001). The addition of altitude, latitude and pH to the regression model resulted in a model that explained 45% of the variation on the index of acidity preference score and N deposition (P < 0·001; see Tables S1 and S2).
Soil pH shows a clear relationship with N deposition along the gradient used in this study (Stevens et al. 2004a,b). This seems to be reflected in the community composition, which shows an increase in the proportion of species with a preference for acid conditions. This is demonstrated by the Ellenberg R score and, to a greater extent, by the index of soil acidity preference scores, although this relationship is still not strong. The slight improvement in the relationship by looking at total acid deposition instead of N deposition against acid index score shows that acidification from sulphur deposition and its historic impact may still be apparent. Given strong evidence for mobilization of Al and heavy metals (Stevens, Dise & Gowing 2009) along this gradient and in experimental studies in similar communities (Blake et al. 1999; Horswill et al. 2008), it is likely that the concentration of metal ions could play an important role in reducing species richness.
Indeed, Grime & Hodgson (1969), working with many of the most abundant of the species encountered in this investigation, showed a clear relationship between species occurrence on acidic soils and resistance of the seedling root to aluminium toxicity. Using an experimental approach, Roem, Klees & Berendse (2002) also conclude that aluminium toxicity is the main driver of species reductions in heathland and acid grasslands in the Netherlands. We conclude that such impacts of increasing acidity are limiting the pool of species able to survive.
One reason why the data do not appear to support the hypothesis that the loss of species with increasing deposition is related to eutrophication effects may be the interaction between soil pH and nutrient availability in the soil. Availability of a number of nutrients is reduced by declining pH. Maximum availability of phosphorus in the soil occurs between pH 5·5 and 7·5, so as pH is reduced below 5·5 as a result of acidification, there is less phosphorus available. Base cations (including the macro- and micronutrients calcium, magnesium and potassium) are readily leached from acidified soils, and concentrations along this deposition gradient are significantly related to pH (Stevens, Dise & Gowing 2009). Long-term investigation of the unfertilized, unlimed plots at the Park Grass experiment at Rothamsted, UK, has shown reductions in concentrations of exchangeable calcium and reductions in cation exchange capacity and base saturation over the 120 years of atmospheric deposition (Blake et al. 1999).
Nitrification is inhibited at low soil pH because the Nitrosomas bacteria, responsible for nitrification, have optimum pH requirements of 7–8. This has been demonstrated in several habitats including grasslands and heathlands (e.g. Roelofs et al. 1985; Dorland et al. 2004). If ammonium accumulates in the soil due to low nitrification, this could also reduce denitrification activity (Sanchez-Martin et al. 2008). At low pH, nitrate uptake is reduced by free Al3+, which can also have a negative impact on mycorrhizal symbionts (e.g. Lazof et al. 1994). As a result of the changes in processing the predominant form of N available in the soil, some plants have become adapted to using one form or another according to their preferred growing conditions. Plants adapted to acidic soils with low nitrification rates (more N available as ammonium rather than nitrate) use ammonium as their preferred N source. They can tolerate high ammonium concentrations without toxic effects and are less efficient at using nitrate (Britto & Kronzucker 2002). The consequence of all these changes in nutrient availability with pH is that N, and other important nutrients, may be less available to plants growing in acidified soils even though N is added. Even if there is additional N available to plants, other nutrients could become limiting and responses to nutrient enrichment may not become manifest. As N deposition increases and the soil becomes increasingly acidic, those species less tolerant of acid conditions are no longer found and the community becomes increasingly dominated by acid specialists.
Ellenberg R and N scores for sites were closely correlated with each other but Ellenberg R scores were not correlated with the index of soil acidity preference as might have been expected. This is likely to be because the scores are derived in a very different manner and the index of soil acidity preference scores is based on data from grasslands only, whereas the Ellenberg scores represent a tolerance across the range of habitats a species occurs in.
The lack of an increase in the Ellenberg N score suggests that any enrichment of the soil by atmospherically deposited N is not available to the vegetation and/or is not of sufficient magnitude to bring about significant changes in the plant community composition. Increased KCl-extractable ammonium in the soil samples noted by Stevens et al. (2006) may be a result of growth, and therefore uptake being limited by a lack of other resources. The range of Ellenberg N scores reported in this study is small, and the values remain at the lower end of the scale. This is because the Ellenberg scores encompass the whole range of situations plants grow in including the most fertile. In this study, we were only concerned with semi-natural grasslands where nutrient inputs are low. These levels of input are considerably lower than the rates of fertilizer application to agricultural grasslands. Experimental evidence in some communities has shown that over time, sites with high atmospheric deposition might be expected to show a slow increase in Ellenberg N score (e.g. Smart et al. 2005).
The lack of a significant change in the C score with increasing N deposition further implies that competition for nitrogen is not the prime mechanism controlling community composition. Under our eutrophication hypothesis, we might expect to see reductions in stress-tolerant species with increasing N deposition, but this is not the case in this grassland community. Although large amounts of N are deposited in some of these grasslands, Phoenix et al. (2003) demonstrated in the same community type that even at high N inputs, large amounts of N are immobilized in the soil, potentially in forms that are inaccessible to plants. Experimental N additions have shown mixed responses in terms of vegetation productivity, but Horswill et al. (2008) cite declines in productivity at high levels of N addition as evidence that competition is not responsible for declines in species richness. Although we know that soil KCl-extractable ammonium increases with increasing N deposition, neither KCl-extractable nitrate or total nitrogen increased along the N deposition gradient (Stevens et al. 2006).
The results from this study contrast with those found in calcareous grasslands. For example, Bennie et al. (2006) report increases in the Ellenberg N score of calcareous grasslands between 1953 and 2003, which they relate to N deposition. The change in Ellenberg N suggests that in calcareous grasslands, where soils are well buffered against changes in pH, eutrophication may be a more significant driver of change than in acid soils. In calcareous soils, pH will not be reduced to a sufficient level to cause the changes in soil chemistry reported above. This has important implications for semi-natural communities sensitive to N deposition suggesting that, in communities with base rich soils, eutrophication may be the most important driver of species change whereas, in communities with acid soils, acidification may be more important. In very resource poor communities, this relationship may not hold as other nutrients or limiting resources may be more important. Further work is needed to investigate the relative importance of acidification and eutrophication in driving species composition changes in response to N deposition across a range of habitats. Management options to mitigate the effects of N deposition will differ depending on whether the change is driven by acidification or eutrophication, and indicators used to detect damage from N deposition must also be different.
Stevens et al. (2004a,b) reported that relationships between species richness and total inorganic N deposition were slightly stronger than those with soil pH, indicating that acidification alone does not explain the full effect of N deposition. Although the evidence from this study supports the hypothesis that soil acidification is the dominant process responsible for a decline in species richness with increasing N deposition, it is likely that there is an element of eutrophication and other soil chemical changes are involved too, the effects of which may be partly missed by shifts in the limiting nutrient.
The authors would like to thank the European Science Foundation for funding C.J.S. to visit Sheffield University to do this research. C.J.S. is grateful to Nancy Dise and Owen Mountford for their guidance in developing the original gradient study from which data for this investigation were derived. We are grateful to an associate editor and three anonymous referees for helpful comments on earlier versions of this manuscript.