W.J. Roem, Nature Conservation and Plant Ecology Group, Department of Environmental Sciences, Wageningen University, Bornsesteeg 69, 6708 PD Wageningen, the Netherlands (fax +31 317484845; e-mail Wilma.Roem@wur.nl).
1The atmospheric deposition of sulphur and nitrogen compounds in the Netherlands has been responsible for decreasing plant species diversity in heathland. To unravel the relative importance of nitrogen compounds on soil acidification and eutrophication, and hence on the vegetation, we carried out a factorial addition experiment and a germination experiment in heathland on nutrient-poor sandy soil.
2We changed nutrient availability and acidity independently in eight different treatments that, respectively, added nutrients or carbon in various combinations (N, P, glucose) or added acidifying or neutralizing compounds. One treatment also involved adding Al. Additions occurred five times per year during 5 years, in an area from which sods had been removed before the experiment began. The same design was used for the germination experiment, but the treatments were applied for 2 years.
3Our results showed that acidification was the most important factor in reducing species diversity. In addition, the germination of several heathland species was significantly reduced in plots with a pH below 5, and germination was very poor in plots where Al had been added.
4The number of plant species declined particularly with increasing Al in the upper soil horizons. We conclude that this relationship is responsible for the influence of acidification on plant species richness in heathland.
5The influence of nutrient availability on species composition in heathland was subsidiary to acidity, but nutrient availability influenced species composition in an independent way. The growth of the three dominant species (Molinia caerulea, Calluna vulgaris and Erica tetralix) was limited by different nutrients. Erica tetralix was limited by N, Calluna vulgaris by P and Molinia caerulea by both N and P. We argue that increased N availability will change the relative availability of N and P, which can decrease species diversity.
6Together these results show how factorial experiments can elucidate the complex ecological effects arising from sulphur and nitrogen deposition, revealing different mechanisms that change species richness and community composition.
There is still heavy atmospheric deposition of sulphuric and nitrogenous compounds in the Netherlands (0·05 Mol SOx m−2 year−1 and 4·7 g N m−2 year−1 in the experimental area in 1998; Bobbink & Heil 1993; RIVM 1999). Deposition, in turn, is often assumed to be responsible for the decline in species diversity in heathlands in recent decades (Van Dam et al. 1986; Berendse & Aerts 1987; Roelofs et al. 1996). The species composition of these communities has probably been altered by changes in the chemical status of the soil due to increasing N availability, soil acidification and mobilization of toxic metal ions (Bobbink, Hornung & Roelofs 1998). In order to maintain the last remaining fragments of species-rich heathland, it is essential that we understand the mechanisms responsible. A study in which effects of acidification and nutrient addition on plant species diversity in heathlands could be distinguished should show whether ammonium eutrophication or ammonium acidification caused the decline in species richness.
Acidification of the soil has been shown to diminish the number of species in heathlands (Houdijk et al. 1993; Roem & Berendse 2000). When the pH falls below 5 particularly endangered species will disappear. Several mechanisms whereby pH affects species performance in heath have been investigated. Van Dobben (1991) showed that the dominant species (Molinia caerulea and Deschampsia flexuosa) had a lower pH optimum than the endangered species. De Graaf et al. (1998a) observed that the endangered species Arnica montana and Cirsium dissectum performed better on N sources as nitrate than as ammonium (pH of 4 in solution). At increased ammonium concentrations these species showed symptoms of ammonium toxicity, but the frequently dominant heathland species Calluna vulgaris performed very well when ammonium was its N source.
With increasing soil acidity, the Al concentration in the soil increases. In another hydroculture experiment (De Graaf et al. 1997), Arnica montana and Cirsium dissectum showed Al toxicity, indicated by poor root development, yellowish leaves and reduced contents of Mg and P in the plants. These symptoms were less severe when Ca was added. The growth of Calluna vulgaris was not affected by Al or Ca concentrations.
All the studies described above were carried out in the glasshouse. To date, no long-term field study has been carried out on the mechanisms of acidification in heathlands and their impact on species diversity, which would allow the effects of pH and Al ion toxicity on vegetation performance and species richness to be distinguished.
In some ecosystems plant production does not respond to N addition, which suggests that communities can be P or K limited (Aerts, Wallen & Malmer 1992; Morecroft, Sellers & Lee 1994). Nevertheless, the addition of N alters species composition. It seems that the indirect effects of N addition or a shift in the ratio between the supply of N and other nutrients may also have important impacts on the plant species composition.
Prolonged high N deposition in the Netherlands, while P deposition was negligible, will have changed relative P availability in heathlands. A study in which the effects of increased N and P availability on vegetation composition in heathland are compared should therefore investigate whether the shift from N to P limitation is a serious threat to the diversity of heathland vegetation.
Changes in the chemical status of the soil influence species composition by altering the conditions for growth for adult plants, but also by altering conditions for germination and for the establishment of seedlings. Whether a new species establishes in an area depends on whether the soil conditions allow its seeds to germinate and its seedlings to establish. Tilman (1997) and Foster & Gross (1998) found a negative correlation between germination and N availability in grasslands. To be able to maintain and restore species-rich heathland it is important to know whether changes in soil acidity and nutrient availability influence the germination and seedling establishment of rare plant species (Berendse 1999). We therefore designed two experiments to study the effects of N deposition on the development of species diversity in the field. Here we describe the results of a nutrient addition experiment in heathland vegetation lasting 5 years, and the results of a germination experiment with nutrient addition lasting 2 years. To unravel the acidifying and eutrophying effects of N deposition these experiments involved different levels of soil acidity and different levels of nutrient supply. Our aim was to find out (i) whether species richness is influenced more by soil pH or the Al3+ concentration; (ii) whether germination of rare species is influenced by soil pH or the Al3+ concentration; and (iii) whether species richness is influenced by the ammonium concentration in the soil. We increased the availability of N and P independently and jointly, and compared the effects on vegetation composition with the aim of also finding out (iv) whether changes in nutrient limitation alter species richness and (v) whether they affect the germination of rare species. With these experiments we wanted to elucidate which mechanisms are reducing plant species diversity and therefore limiting the restoration of heathlands in Europe.
Materials and methods
The restoration project Het verbrande bos, near Staverden (52°17′N, 5°44′E) in the centre of the Netherlands, is part of an estate managed by the landscape conservation organization Gelders Landschap. The soil consists of aeolian sandy deposits overlying Pleistocene loam. The area was covered with pine forest before the restoration. In order to restore the former wet heathland, the trees and the topsoil were removed (sod removal), and the hydrology was changed so that seepage water was retained. In the first stage of the restoration project, sod removal took place in 1989. Since then wet heath vegetation has developed successfully, and several red list species have re-established. We started our nutrient addition experiment in 1995 and our germination experiment in 1997 in part of the area where sods had been removed in 1994. In 1995 the vegetation had just started to develop; some small plants of Calluna vulgaris, Erica tetralix and Molinia caerulea had appeared on the bare sandy soil. By 1999 the vegetation was developing into an Ericetum tetralicis in the wetter parts and into a Genisto anglicae–Callunetum typicum in the drier parts.
Experiment 1, a 5-year experiment on nutrient addition and acidification
Forty plots of 1 m2, separated by 0·5 m, were set out in a randomized design with five blocks of eight plots. Each plot was surrounded by a narrow border of plastic pushed 5 cm into the soil. The experimental site was fenced to prevent grazing. Eight different dosing treatments were then applied over 5 years, by watering the plots with the yearly dose of nutrients, acid or neutralizing agents added in five portions during the growing season (between March and October) as follows: (i) control: 50 litres tap water m−2 year−1; (ii) glucose: 250 g glucose m−2 year−1 in 50 litres tap water; (iii) N: 10 g N m−2 year−1 in 50 litres (NH4NO3); (iv) P: 5 g P m−2 year−1 in 50 litres (NaH2PO4); (v) N + P: 10 g N + 5 g P m−2 year−1 in 50 litres; (vi) Ca: 600 g CaCO3 m−2 year−1 in 50 litres tap water; (vii) acid, H2SO4, double the average atmospheric deposition: 0·48 Mol H+ m−2 year−1, 48 litres 0·005 m H2SO4 m−2 year−1 + 2 litres tap water; (viii) Al: 21·8 g Al m−2 year−1 in 50 litres tap water (AlCl3).
The glucose treatment was applied to stimulate microbial N and P immobilization and to reduce N availability for plants in the short term (Keeling et al. 1996). The N, P and N + P treatments were applied to increase nutrient availability. These four treatments together were used as the nutrient availability treatments.
The H2SO4 treatment was performed to double the atmospheric acid input in the area, which was 0·48 Mol H+ m−2 year−1 (Bobbink & Heil 1993). It was necessary to dilute this amount of acid with 50 litres tap water to achieve a pH acceptable to the vegetation (pH > 3). For this reason all plots were treated with 50 mm tap water year−1 (the mean annual precipitation in this area is 800 mm). The AlCl3 treatment was performed to increase Al3+ concentrations by the same amount as the acid treatment was expected to do. The CaCO3 treatment was intended to decrease soil acidity. These three treatments together were used as the acidification treatments. The applications started in August 1995 and ended in October 1999.
Experiment 2, a 2-year experiment on germination
In 1997, 40 plots of 0·5 × 0·25 m were set out in a randomized design of five blocks of eight plots in the same way as for the 5-year experiment. Each plot was divided in half (0·25 × 0·25 m) with a narrow strip of plastic. The same eight different treatments as in experiment 1 were applied, but for 2 years. The yearly dose was added in five portions during the growing season (between March and September). The first application was in April 1997 and the last in September 1998. These plots were located near the other plots inside the fence on a still almost bare soil in 1997.
The seeds of the following heathland species used in experiment 2 were collected from the nearby nature reserve De Leemputten in August and September 1996: Rhynchospora alba, Rhynchospora fusca, Gentiana pneumonanthe, Parnassia palustris, Succisa pratensis, Narthecium ossifragum, Epipactis palustris and Euphrasia stricta. Linum catharticum seeds from a commercial source were used. All the seeds were stored at 3 °C during the winter. In March 1997 the seeds were moistened and stored damp at 3 °C for 4 weeks prior to sowing. Equal amounts of seeds of the nine species were sown in one half of each of the 40 plots on 10 April. Germination started after 2 weeks. Rhynchospora alba, Gentiana pneumonanthe, Parnassia palustris, Succisa pratensis, Epipactis palustris and Euphrasia stricta were observed to germinate. Numbers of seedlings were counted every 2 weeks during two growing seasons (March–September). To avoid double counting, seedlings were marked with pins colour-coded per species, as soon as they had been counted.
In the 5-year experiment all plant species, including bryophytes, were recorded once a year, in September. At the same time as the species were being recorded, each plot was photographed from above (height 2 m). The resulting colour slides were used for estimating cover per species. Cover was calculated with a point quadrat method. The slides were projected on a sheet with 12 × 12 lines with 144 intersections. The species located at each intersection were counted by two people independently, differences between the two scores never exceeding three out of 144. The cover per species was estimated by the number of intersections on which the species was recorded. Mean cover values were calculated per species per treatment per year. For the three dominant species cover increase during the whole experiment was calculated.
On 13 October 1999 the total vegetation of each plot was clipped near the soil surface. The total dry weight was measured and the N, P and K concentrations in the above-ground plant material were measured in the material harvested from the inner 0·25 m2 per plot.
In March 1995 four pooled samples (three soil cores of 1·5 cm diameter) were collected in the experimental site for analysis of the starting conditions for experiment 1. In October 1999 five soil samples (3 cm diameter) from the upper 5 cm of the soil were collected in each of the plots that had been treated for 5 years and were pooled for analysis. In March 1997 three pooled samples (three soil cores, 1·5 cm diameter) from the upper 10 cm of the soil were collected in the experimental site for analyses of the starting conditions for experiment 2. In November 1998 five soil samples (1·5 cm diameter) from the upper 10 cm of the soil were collected in each half of the plots treated for 2 years and pooled for analysis.
The pH of fresh soil samples was measured after shaking 20 g soil with 50 ml demineralized water. The ammonium and nitrate concentrations were measured using a continuous flow analyser after extraction of 20 g fresh soil with 50 ml 0·1 mm KCl.
After drying at room temperature for 48 h, the cation exchange capacity (CEC) and base saturation were determined in the samples by extraction with unbuffered 0·01 m BaCl2 solution. The Ca, K, Mg, Fe and Al contents were measured in the extract on an atomic absorption spectrophotometer (AAS) (Houba et al. 1989). The organic matter content was determined after drying subsamples at 105 °C. Available P was measured after extracting 1·2 ml dried soil with 3 ml H2O.
Treatment effects on some soil variables were analysed using analysis of variance (one-way anova). Although the experiments were set up in a randomized block design using eight treatments repeated in five blocks, block appeared to have no effect and was therefore not used in the analyses. Vegetation and seedling measurements were repeated during the years of the experimental period. We tested treatment effects over the years using one-way repeated-measures anova. As interactions between year and treatments were significant, these were followed by one-way anovas for each year.
To distinguish between the effects of the nutrient treatments and the acidity treatments, two a priori contrasts were carried out within the one-way anova, for each year. The nutrient availability treatments (glucose, N, P and N + P addition) were compared with the remaining treatments, and the acidity treatments (H2SO4, Al and Ca addition) were also compared with the remaining treatments. The set of these comparisons is orthogonal. Tukey's HSD test was used to test post-hoc for differences among means.
All data were tested with Levene's test for homogeneity of variance and with Kolmogorov–Smirnov for normality. Transformation was not necessary for most data, only seedling data, N:P ratio and cover increase data were log-transformed. However, not all soil variables satisfied normality due to treatment effects, even after transformation, and the non-parametric Kendall's rank correlation test was used to analyse the relationship between the number of vascular plant species and the soil variables measured. SPSS 7.5 package (1996; SPSS, Chicago, USA) was used for all analyses.
The total list of plant species in the plots consisted of 16 vascular plant species including two ericoids, six monocots, three tree species (seedlings), five herbs and nine bryophyte species. During the experiment three rare species appeared in the plots: Drosera rotundifolia, Drosera intermedia and Lycopodium inundatum. All of these have declined sharply in the Netherlands in recent decades.
overall treatment effects
The eight treatments that were applied in these two experiments significantly influenced the number of species and the germination of four rare heathland species (Fig. 1 and Tables 2 and 3). All the treatments also had significant effects on biomass production, increase in cover of the dominant species and on nutrient ratios measured in the above-ground vegetation (Table 2).
Table 2. Experiment 1. Results of anova with repeated measures and anova for treatments and a priori contrasts for nutrient availability (nutrients) and acidity (pH) (see the Material and methods for further explanation). F-values with significance: *P < 0·05, **P < 0·01, ***P < 0·001; Trt is treatment; year × trt is the interaction between years and treatments. Degrees of freedom (d.f.) for treatments, 7; remaining error, 32; d.f. for nutrients, 3; for pH, 2; remaining error, 34
Table 3. Experiment 2. Results of anova with repeated measures and anova for treatments and a priori contrasts for nutrient availability (nutrients) and acidity (pH) (see the Material and methods for further explanation). F-values with significance: *P < 0·05, **P < 0·01, ***P < 0·001; Trt is treatment; year × trt is the interaction between years and treatments. Degrees of freedom (d.f.) for treatment, 7; remaining error, 32; d.f. for nutrients, 3; for pH, 2; remaining error, 34
The pH of the soil at the beginning of the experiments was 4·3 over the whole experimental site. At the end of the experiments the pH of the soil in the Ca treatment was higher than in the other plots, whereas the pH in the acid and the Al treatments after 5 years of addition was lower (Table 1). The Ca treatment also increased the Ca concentration in the soil and lowered the Al ion concentration. The Al treatment increased the Al ion concentration (Table 1).
Table 1. Experiment 1. Measured soil and vegetation parameters, averaged per treatment with standard errors between parentheses. Significant differences among means are indicated by different letters. Ca and Al in mmol kg−1, NH4 in mg kg−1 0. N:P ratio measured in above-ground vegetation. Degrees of freedom for treatment effect, 7; for intercept, 1; and for remaining error, 32 for all parameters
N + P
Experiment 1, vegetation
Species richness was clearly influenced by the soil pH (Figs 2 and 5). At the start of the experiment the species were distributed equally over the plots. Over the years the treatment effects on species number became significant (Table 2). The Ca treatment increased species richness. During the first years of the experiment the increase was fast, because tree seedlings appeared (although only a small proportion established; results not shown). After 1997 the number of species decreased slightly (Fig. 1). In 1999 the number of species, although still higher, was not significantly different from the control. The number of bryophyte species was increased by Ca treatment (Fig. 1). The Al treatment significantly decreased the number of species during the experiment (Fig. 2). No endangered species appeared and the number of bryophyte species was also low.
The above-ground biomass of the vegetation was not significantly influenced by the acidity treatments (Table 2 and Fig. 2), nor was the increase in cover of the three dominant species (Fig. 3).
Experiment 2, seedlings
The acidity treatments had a significant influence on the germination of all six species that germinated (Table 3). The Ca treatment significantly favoured the germination of Succisa pratensis, Parnassia palustris and Euphrasia stricta. The Al treatment was detrimental to the germination of all six species, but only for Rhynchospora alba was germination in the Al-treated plots significantly lower than in the control plots. In the unsown parts of the treated plots hardly any germination was observed. Only some seedlings of Succisa pratensis and Rhynchospora alba were observed in unsown halves of plots.
With the nutrient treatments (glucose, N, P, N + P) we influenced nutrient availability significantly. We increased the N availability with the N treatments (Table 1) and increased the P availability with the P treatments, although we did not significantly reduce the N or P availability with the glucose treatment.
Experiment 1, vegetation
Compared with the control treatment the biomass production increased significantly in the N + P treatment (Fig. 2). The nutrient treatments significantly changed the N:P and N:K ratios measured in the total above-ground vegetation (Table 2).
Species richness was not significantly influenced by the nutrient availability treatments (Fig. 2 and Table 2). The P treatment clearly increased species richness during the first years (Fig. 1) but this effect diminished in 1999. Phosphorus addition increased the number of bryophyte species (Fig. 1). The N + P treatment showed the same pattern for species richness as the control, but the number of species decreased in 1999 (Fig. 1). In these fertilized plots, vegetation biomass increased during the experiment and light limitation probably occurred in 1999 (Fig. 2). This treatment, in which P was also added, slightly increased the number of bryophyte species (Fig. 1).
The increase in cover of the three dominant species was significantly influenced by nutrient treatments. Because the interaction between species and treatments was significant, we analysed the species separately (Table 2). At the beginning of the experiment the cover of the three species did not differ significantly among treatments. This changed over the years; in 1999 treatment effects were clearly significant in all three species. The nutrient availability was significantly correlated with the responses of two of the three species (Table 2).
The cover of Calluna vulgaris responded to the P application (Fig. 3) by increasing steadily over the years. The N + P treatment increased this cover even more than the P treatment. The N treatment did not influence the cover of Calluna vulgaris.
The cover of Erica tetralix was increased by the N treatment (Fig. 3). The P treatment decreased the cover of Erica tetralix and the N + P treatment did not increase the cover of Erica tetralix, probably due to competition for light, this species being smaller than the other two. The cover of Molinia caerulea was increased significantly by the N + P treatment (Fig. 3) but only slightly by the N and P treatments. The glucose treatment did not decrease the cover of all three species compared with the control treatment.
Experiment 2, seedlings
The nutrient availability treatments were not observed to influence the germination of the heathland species in this experiment (Table 3 and Fig. 4).
relations between species richness and measured variables
After analysing the overall treatment effects and the a priori contrasts we analysed the effects of several variables on species richness independently. We entered soil Al concentration and soil pH as a covariable in the anova (Table 4). Both parameters explained a large part of the treatment effects on species richness. Within the nutrient treatments they explained a significant part of the remaining covariance.
Table 4. Experiment 1. Results of anova of treatment effects and a priori contrasts with number of vascular plant species dependent and different parameters included as covariables in 1999. F-values with significance: *P < 0·05, **P < 0·01, ***P < 0·001. Degrees of freedom (d.f.) for treatments, 7; for covariable, 1; and for remaining error, 31; d.f. for nutrient treatments, 3; and remaining error, 35; d.f. for pH treatments, 2; remaining error, 36
Dependent variable: number of vascular plant species
The nutrient ratios measured in the above-ground vegetation were significantly affected by the nutrient treatments (Table 2). To analyse the relationship between species richness and N:P ratio, we added N:P ratio as a covariable (Table 4) in the anova. The treatments still had a significant influence on the number of species, but the N:P ratio also explained a significant part of the covariance. In the a priori contrasts the N:P ratio added a significant influence on the number of species (Table 4). Although the highest numbers of species were found in vegetation with an N:P ratio between 14 and 16, there was no significant correlation between species richness and N:P ratio (Fig. 5). Neither was there a significant relationship between species richness and above-ground biomass (Fig. 2).
Linear regression of plant species richness and soil parameters showed several significant correlations (Fig. 5) for the acidity-related soil parameters as well as for NH4. Of all the parameters tested, the Al concentration in the upper soil was the best predictor of species richness.
In this study we investigated the establishment of species-rich heath vegetation after sod removal. By choosing this starting condition, we obtained a condition with relatively low contamination of deposited N. The advantage of this experimental set up was the possibility of following the establishment of the vegetation during the years of the experiments, but the disadvantage was that only in the last year of the experiments had the vegetation developed a closed canopy in most plots.
The influence of the treatments modifying soil acidity on species richness was very clear over the years of the experiments (Figs 1 and 2): the soil acidity was highly correlated with species richness (Fig. 5). In agreement with earlier studies (Houdijk et al. 1993; Roem & Berendse 2000), we found that on these nutrient-poor weakly buffered soils the greatest threat to species richness is acidification. Increasing the pH of the soil by treatment with CaCO3 increased both the number of plant species and the number of bryophyte species, which confirms the findings of De Graaf et al. (1998b). During the first years of the experiment numbers of species were especially high, which implies that the increased pH favoured the germination of plant species (Fig. 1). Not all these species subsequently established in the plots, probably due to low nutrient availability. At the end of the experiment, species numbers were still higher than the control.
The three endangered species present in experiment 1 were all species of acid soils (Drosera rotundifolia, Drosera intermedia and Lycopodium inundatum). It seems likely that no seeds of other endangered species were present, even though a population with flowering plants of rare species occurred only 200 m from the plots. The fact that hardly any germination of endangered species was observed in the unsown halves of experiment 2 plots endorses this claim.
In the design of these experiments we calculated amounts of Al3+ ions to add for the Al treatment to be comparable with the amount of Al ions available in the acid treatment. Although we succeeded in reaching the same pH in both treatments, the Al concentrations in the Al treatment were significantly higher (Table 1). Species richness responded differently to the two acidifying treatments (Fig. 1). In the H2SO4 treatment, species numbers were only slightly decreased compared with the control. In contrast, the Al treatment had a disastrous impact on the vegetation: species numbers decreased significantly. Moreover, total plant cover was low (results not shown). Endangered heathland species have been shown to be susceptible to Al toxicity (De Graaf et al. 1997) in the glasshouse, whereas Calluna vulgaris was not harmed by Al. In this field study we found that almost all the species were clearly susceptible to increased Al ion concentrations in the soil. The only species to appear in the Al plots were the three dominant species Calluna vulgaris, Erica tetralix and Molinia caerulea. The Al concentration in the soil explained a large part of the treatment effects of the addition experiment (Table 4) and showed the highest negative correlation with species richness (Fig. 5). The availability of Ca counterbalanced the Al toxicity for the vegetation and increased the pH, reducing the Al ion concentrations. In the germination experiment, the Al treatment reduced the germination of all six species investigated to almost zero (Fig. 4); the acid treatment was less detrimental. We therefore conclude that elevated Al ion concentrations caused by acidification of the soil is the most important threat to species richness in heathlands.
In agreement with earlier studies (De Graaf 2000) we found that in five out of the six species investigated, the germination was significantly favoured by the Ca treatment and thus by the increased pH of the soil. Only Rhynchospora alba, a species of acidic soils, did not germinate better in the Ca-treated plots (Fig. 4). Rhynchospora alba showed the highest numbers of seedlings in the control treatment while the Al treatment significantly decreased the germination in 1998. In the germination experiment the pH of the soil was significantly increased in the Ca-treated plots only (results not shown); after 2 years of addition the soil pH of the acid and Al-treated plots had not yet significantly decreased. The opposite was the case for the Al ion concentration in the soil: it was significantly lowered in the Ca-treated plots only. In spite of this, by comparison with the other treatments, the acidification treatments together significantly influenced the germination of all six species (Table 2). We therefore conclude that increasing the soil pH above pH 5 is the most important germination condition for rare heathland species, but that increased Al ion concentrations in the soil seem to be detrimental to the germination of all species.
The increased ammonium concentration in the soil in the Al treatment (Table 1) cannot explain the decreased numbers of species. In the N treatment, ammonium concentrations were even higher but there were no detrimental effects on numbers and cover of plant species. De Graaf et al. (1998a) reports that Calluna vulgaris performed well on ammonium nutrition. In our study Calluna vulgaris in the Al treatment did not perform well, although differences with the control treatment were not yet significant after 5 years. The growth of Calluna vulgaris was not negatively influenced by the N treatment, although ammonium concentrations in these plots were significantly higher than in the Al plots. The ammonium concentrations in the soil could not explain the numbers of species in this experiment. The correlation between ammonium concentration and species richness (Fig. 5) was attributable solely to the increased ammonium concentration in the Al treatment (Table 1). There was no correlation between ammonium concentration and soil pH. The increased ammonium concentration in the Al-treated plots could have been caused by the strongly reduced N uptake in the Al plots, so that the mineralized N accumulated as in the soil (Roelofs et al. 1996). We conclude that the acidification accompanying ammonium deposition has the potential to reduce plant species richness because the accompanying acidifying effect results in increased concentrations of Al3+ ions in the soil.
The nutrient treatments together had no significant influence on species richness in this experiment (Fig. 2). Only P addition favoured bryophyte species (Fig. 1): Dicranella heteromalla in particular increased its cover in the P treatments in the first few years of the experiment, which is in agreement with the findings of Tyler, Tyler & Tyler (1995). By contrast, nutrient availability was significantly correlated with the abundance of the dominant species (Table 2) and to above-ground biomass production. In agreement with earlier work (Roem & Berendse 2000), the influence of nutrient availability on species composition in heathland was subsidiary to acidity but nutrient availability had its own influence on the species composition.
The cover of the three dominant species responded differently to the nutrient addition treatments (Fig. 3). The growth of Calluna vulgaris was limited by P and not by N, the growth of Erica tetralix was limited by N and definitely not by P, and the growth of Molinia caerulea was limited by N and P together. The N + P treatment increased the biomass of Calluna vulgaris and Molinia caerulea so that competition for light became a disadvantage for Erica tetralix, the shortest species of the three. Aerts (1990) investigated the same three species and concluded that Molinia caerulea crowded out the other two species when nutrient availability increased. In our experiment Erica tetralix was almost outcompeted after 5 years in the N + P treatment but the other two species were both growing well.
Under these nutrient-poor conditions, the growth of the three dominant species, growing together in the same plots, was limited by different nutrients. The coexistence of these three species in the same vegetation appears to have been favoured by a relatively low availability of both N and P. Disturbance of this resource balance might bring about the competitive exclusion of one or two of the species. Our findings suggest that species diversity in heathland is favoured by low nutrient availability of both nutrients, and changes in the availability of one nutrient might disturb the balance.
The results we report are in agreement with the resource balance hypothesis (Braakhekke & Hooftman 1999) that predicts that species will coexist in a competitive equilibrium in a homogeneous environment when actual resource supply ratios are balanced. This conclusion is also in agreement with the resource transport and competition theory of Huston & DeAngelis (1994), which predicts that shifts from limitation by one type of resource to limitation by another type can dramatically alter the nature and intensity of competitive interactions.
The high values for the N:P ratios measured in the above-ground vegetation in the control plots (Table 1) indicate that P and probably also K are the nutrients limiting the vegetation growth in this area (Koerselman & Meuleman 1996). This is in agreement with earlier findings (Roem & Berendse 2000). Our results indicate that N still limits growth of at least one of the dominant species, Erica tetralix, in spite of the high N deposition in this region. Only in the N + P treatment did we find a significant increase in biomass production. Changes in nutrient availability induced changes in the nutrient ratios measured in the vegetation. We found that the N:P ratio in the above-ground vegetation influenced the number of species (Table 4), which is in agreement with the theory of the resource balance hypothesis. We therefore conclude that increased N availability due to high N deposition threatens species composition by disturbing species’ coexistence at balanced supplies of resources.
In this study we distinguished the acidifying and eutrophying components of the effects of N deposition on heathland vegetation. Acidification was the most important factor reducing species diversity, due to increased Al ion concentrations in the upper soil layer and resulting toxicity for many species. The increased N availability altered the equilibrium between the dominant species and decreased species diversity.
The current restoration management of the area we studied is succeeding in restoring a relatively species-rich heath vegetation, probably due to the hydrological measures that have increased the soil pH in parts of the area, combined with the removal of accumulated N in the soil with the sod removal. We showed in this study that acidification is the most serious threat to European heathlands but that accumulation of N can decrease species richness. The most effective restoration measures in the short term will be those combating acidification but, without measures to diminish or to remove deposited N, species richness will decrease again in the longer term.
The authors are grateful to Gelders Landschap for granting permission to carry out the experiment in their nature reserve, and to J. Van Walsem for technical assistance. Critical comments of Dr H. de Kroon, Dr S. Güsewell and Dr M. Heijmans helped to improve earlier versions of the manuscript. J. Burrough advised on the English.