• The adaptive responses to atmospheric nitrogen deposition for different European accessions of Arabidopsis lyrata petraea were analysed using populations along a strong atmospheric N-deposition gradient.
• Plants were exposed to three N-deposition rates, reflecting the rates at the different locations, in a full factorial design.
• Differences between accessions in the response to N were found for important phenological and physiological response variables. For example, plants from low-deposition areas had higher nitrogen-use efficiencies (NUE) and C : N ratios than plants from areas high in N deposition when grown at low N-deposition rates. The NUE decreased in all accessions at higher experimental deposition rates. However, plants from high-deposition areas showed a limited capacity to increase their NUE at lower experimental deposition rates. Plants from low-deposition areas had faster growth rates, higher leaf turnover rates and shorter times to flowering, and showed a greater increase in growth rate in response to N deposition than those from high-deposition areas.
• Indications for adaptation to N deposition were found, and results suggest that adaptation of plants from areas high in N deposition to increased N deposition has resulted in the loss of plasticity.
Atmospheric pollutants are a widespread phenomenon, most having increased globally as a direct consequence of human impact. Anthropogenic atmospheric nitrogen deposition is recognized as a major threat to seminatural ecosystems, and an increasing threat to global hotspots of biodiversity (Phoenix et al., 2006). Many studies have linked increased atmospheric N deposition to a decrease in plant diversity (Tilman, 1987, 1993; Gough et al., 2000; Smart et al., 2003, 2005; Stevens et al., 2004). The mechanisms leading to these effects include increased susceptibility to environmental stressors such as drought and frost (Carroll et al., 1999), increased susceptibility to pests and diseases (Brunsting & Heil, 1985), and the domination of faster-growing, nitrophilous species, leading to the competitive exclusion of plants adapted to low N availabilities (Tamm, 1991; Falkengren-Grerup 1998). Taken together, these studies provide strong evidence that plant communities of low nutrient status change as a result of elevated N deposition.
Atmospheric N deposition in Europe has increased from an estimated background rate of 1–5 kg ha−1 yr−1 in the early 1900s to as high as 30–60 kg ha−1 yr−1 in the 1980s and early 1990s in the worst affected regions (Bobbink & Lamers, 2002; Fowler et al., 2004), mainly caused by anthropogenic activities. Generally, reduced forms of N are deposited heavily near their agricultural sources and in areas with high wet deposition, whereas oxidized forms of N are emitted from industrial sources and transported over wider areas. In Western Europe, the highest deposition of N can be found in heavily populated areas with intensive farming, such as the Netherlands, Denmark and some southern and central parts of the UK, whereas the lowest N deposition is found in less populated areas such as Scotland and Scandinavia. Although, in most European countries, emissions of oxidized N declined after the 1990s as a result of more stringent legislation, current deposition rates are still an order of magnitude higher than in preindustrial times (Fowler et al., 2004; Galloway et al., 2004).
Nitrogen availability has been identified as an important factor influencing the distribution of plant species, exerting a strong selective pressure on natural populations. Provided there is enough genetic variation and the timescale is long enough for these processes to take place, plants may evolve an adaptive response to N availability. Recent reports indicate differential fitness in response to N enrichment, both between (Falkengren-Grerup, 1998; Van den Berg et al., 2005) and within (Vergeer et al., 2003) species, which may result in evolutionary shifts in populations. However, to date there is surprisingly little evidence of evolved responses to N deposition.
There are two ways to study evolutionary changes of plants to long-term exposure to air pollutants. In the first approach, populations either are followed in time as they experience increased air pollutant levels, or are introduced into an already polluted area and their evolutionary change followed over time (Wilson & Bell, 1985). A second approach is to compare populations with different histories of exposure, including locations where air pollutants have been elevated over evolutionary timescales (and thus where selection for adaptation has been strong) with those where air pollutant concentrations have remained low (Fordham et al., 1997). This can be studied on a relatively fine scale (within regions) as well as on a coarse scale (between regions or countries). On a fine scale, effects of gene flow may limit adaptation, whereas on a coarse scale, interference of secondary factors such as climate will become more important. The few studies that address the possibility of N-induced evolution of adaptive genotypes were performed at a relatively fine (subnational) scale on plants originating from sites with different air pollution characteristics (Dueck et al., 1988; Taylor & Bell, 1988). In these studies, which mainly focused on plant fitness, adaptive responses to increased atmospheric N deposition were not found. Studies on a coarser scale, comparing specific plant characteristics or physiological traits in plants from regions with long-term differences in atmospheric N pollution, may provide a better test of possible evolutionary adaptations, but have not been undertaken to date.
In this study, we focus on a single species with a distribution range spanning a large gradient in atmospheric N deposition, to test for regional differences in adaptations to N usage. Our focal species, Arabidopsis lyrata petraea, is a short-lived perennial that grows in seminatural rocky habitats. The fact that the species largely grows as solitary individuals in low-competition (and low-herbivory) environments without any mycorrhizal associations makes it an ideal system to study the effects of abiotic stressors such as atmospheric N deposition. We selected plants from six populations divided over three regions (Iceland, Norway and the British Isles) along an atmospheric N-deposition gradient. The gradient ranges from Ireland and north-west Wales, with relatively high atmospheric N-deposition rates of approx. 27 kg N ha−1 yr−1, to Norway and Iceland, with intermediate and low atmospheric N-deposition rates of approx. 12 and 3 kg N ha−1 yr−1, respectively (Tarrasón et al., 2007). Conveniently, this provides us with threefold differences in deposition rate between each region in succession. As nutrient availability in the soil in subarctic regions is lower than in temperate regions (Shaver & Chapin, 1980), soil conditions were analysed in each site. The plants were grown in a climate chamber under controlled laboratory conditions with simulated atmospheric N deposition corresponding to the total N loads for each of the three regions (27, 9 and 3 kg N ha−1 yr−1, respectively). The effect of N-deposition rates on both life-history traits and plant performance, and plant physiological traits such as C : N ratio and nitrogen-use efficiency (NUE), were analysed.
We hypothesized that plants originating from the medium to low N-deposition areas (Norway and Iceland) would use mineral N more efficiently than plants from high N-deposition areas (British Isles). We also hypothesized that plants from British Isles populations would be at a selective disadvantage at low N-deposition rates and would perform less well than plants from Norway and Iceland. As we expect the plants from high N-deposition areas to be adapted to their local conditions, we hypothesized that these plants would perform better than plants from low N-deposition areas when grown at high N-deposition rates.
Materials and Methods
Study species and site selection
Arabidopsis lyrata spp. petraea (L.) is a self-incompatible, insect-pollinated, short-lived perennial species (Schierup, 1998). In northwest Europe, the species occurs in Iceland, western Norway, Sweden (along a restricted part of the Bothnian coast), the Faroe and Shetland Islands, Scotland, Wales and Ireland (Fig. 1) (Jalas & Suominen, 1994). It is also known from a small area north of Lake Onega in Russian Karelia (Hultén & Fries, 1986), and from scattered refugial populations in central Europe.
In the summer of 2005, seeds of A. l. petraea were collected from six populations divided over three regions (Iceland, Norway and the British Isles). The locations were selected based on the prevailing atmospheric N deposition according to measured and modelled data (EMEP unified model rev. 2.8, http://projects.dnmi.no/~emep/index_model.html) derived from the current EMEP Status Report and country-specific notes (Tarrasón et al., 2007). Modelled N-deposition data were on a 50 × 50-km grid, and values of selected grid cells did not differ markedly between neighbouring grid cells, reducing the effects of scale in the selection (see also Simpson et al., 2006a, 2006b). Within each region, high- and low-altitude populations were selected (Table 1) to test for climatic differences. Within each population, seeds were sampled from 20 randomly selected plants.
Table 1. Environmental conditions at the collection sites
Latitude and longitude
Altitude (m asl)
Total atmospheric nitrogen deposition (kg ha−1 yr−1)
From each site, two soil samples, each consisting of 10 subsamples, were randomly taken with an auger (depth 10 cm, diameter 3.0 cm) after removal of stones, vegetation and litter. Samples from each site were mixed and stored in polyethylene bags in the dark at 4°C until further analysis. In the lab, all remaining roots were removed from the soils before extraction. Samples for nitrate () and ammonium () analysis were extracted using 35 g FW soil in 100 ml 0.5 KCl extraction buffer after shaking for 1 h at 100 rpm. For pH analysis, a 35-g FW sample was shaken with 100 ml deionized water for 1 h at 100 rpm, then pH was measured in the soil suspension. Soil moisture content was measured after drying 15 g fresh soil at 105°C for 24 h to calculate the and content; organic content, determined as loss on ignition, was measured after drying 5 g dry soil at 500°C for 5 h. and concentrations were determined colorimetrically using a Bran and Luebbe (Norderstedt, Germany) Auto Analyser III.
Site specific data
Temperature Local temperature was measured at each site every 2 h for 1 yr, using six Thermochron iButtons (Maxim/Dallas, Dallas, TX, USA). The numbers of degree days above 20°C and below 0°C were calculated for each site based on a full year of iButton records. The growing season length was defined as beginning when the maximum temperature on five consecutive days exceeds 5°C and ending when the maximum temperature on five consecutive days is below that threshold.
Precipitation Precipitation data for the sites were obtained by using data from the nearest available monitoring station from the local and national meteorological institutes, apart from the Welsh population, for which data from the Yr Wyddfa (Snowdon) Environmental Change Network (ECN) monitoring site were used (http://www.ecn.ac.uk). At each weather station, rainfall was measured hourly or daily using comparable methods that meet the criteria of the World Meteorological Organization. All monitoring stations lay within 25 km of their corresponding populations.
Experimental set up
Seeds from all populations (20 families each) were sterilized by washing them for 10 min in 10% sodium hypochlorite solution. Sterilized seeds were germinated in trays containing sterilized sharp sand. Deionized water was used to water the germinating seeds. After 14 d, the young seedlings were transplanted individually into sterilized, small plastic pots filled with sterilized sharp sand (pH = 5.55 with nutrient levels close to the detection limit). After planting, the plants were allowed to acclimatize for 2 wk before treatments started, during which they were watered with deionized water.
Plants were divided between three N-treatment groups so that all groups had representatives of all families from all populations. This resulted in a total of 360 plants divided over three N treatments, with 20 plants in each experimental unit (3 N treatments × 3 regions × 2 populations (low and high altitude) × 20 plants (one plant per family)). The plants were then subjected to different N treatments by means of artificial rain application made from deionized water. Nitrogen was applied at rates of 3, 9 and 27 kg ha−1 yr−1, closely reflecting the atmospheric N-deposition rates at the three field sites.
The composition of the artificial rainwater was based on the rainwater composition of the midlands of Ireland (Clara bog, County Offaly; 53°19.504′ N, 7°36.544′ W) measured in 2000 and 2001 (Tomassen et al., 2004) and contained (in µmol l−1) 66.0 Na+, 107.5 Cl−, 4.0 , 11.4 K+, 7.5 Mg2+, 11.5 Ca2+, 0.02 Zn. A higher concentration of phosphate was added (20 µmol l−1 , approx. 10 times higher than the total P concentration measured in the rainwater) to prevent phosphate-limitation effects. The 3-kg N treatment contained 13.4 and 13.4 ; the 9-kg N treatment contained 40.0 and 40.0 ; the 27-kg N treatment contained 120.1 and 120.1 , respectively (µmol l−1). The pH of the different artificial rain solutions was set at 5.5 before watering using 1 mmol−1 HCl and/or 1 mmol−l NaOH. Plants were watered three times a week, corresponding to an annual rainfall of 800 mm, which in previous experiments proved to keep the substrate continuously moist and close to natural conditions without drying or waterlogging the plants. We therefore felt that a rainfall treatment of 800 mm was the best treatment for our experiment.
Plants were fully randomized and kept in a growth cabinet. Every 2 wk, all plants were reorganized according to a random design. Plants were grown under a day : night regime of 16 : 8 h, temperature 20 : 15°C (day : night), relative humidity 55%, photon fluence rate 150 µmol m−2 s−1 and ambient CO2 concentrations.
Measurements: specific plant characteristics
Every 4 wk, the number of leaves was counted and the length of the longest leaf was measured. Senescent leaves were distinguished from ‘fresh’ (nonsenescent) leaves and were counted separately. Flowering time (number of days from germination to flowering) and length of inflorescence were measured when there was a complete expansion of the petals on the first flower. Plants were harvested at the time of flowering, ranging from 145 to 269 d after germination for the first- and latest-flowering plant, respectively. After washing, inflorescences were separated from leaves, and senescent leaves were distinguished from fresh leaves. The fresh leaves and inflorescences were dried for 48 h at 70°C, after which the dry weight was determined. A leaf-correction factor was calculated for each plant by dividing the dry weight of fresh leaves by the number of fresh leaves multiplied by the length of the longest leaf. This enabled us to estimate the total produced aboveground biomass at earlier time points, which included fresh and senescent leaves. The total aboveground biomass was then estimated for every 4 wk by multiplying the total number of leaves produced (fresh and senescent) by this correction factor. We then calculated the growth rate of total aboveground biomass as the slope of the linear regression equation for each individual plant between estimated total aboveground biomass and the time in days after germination. Leaf lifespan was expressed as the number of days from when a leaf first appeared until it became senescent. Leaf lifespan was calculated using the assumptions that the first senescent leaf was the first leaf produced, and no leaves fell off without being noticed. The first two senescent leaves, which were likely to be the cotyledons, were omitted from the analyses.
For C : N analysis of the shoots, dried plant material (fresh leaves only) was ground in a ball mill and analysed using a C : N analyser (VarioMacro, Elementar, Hanau, Germany). Aboveground NUE was estimated by dividing the total aboveground biomass (g DW) by the amount of N (g) in fresh leaves (based on Chapin, 1980; Vitousek, 1982).
We used the following parameters to analyse the effect of atmospheric N deposition on plant performance: flowering time; total aboveground biomass at time of flowering; and average growth rate of total aboveground biomass. The C : N ratio, percentage C, and percentage N in the biomass of fresh leaves, in combination with NUE and leaf lifespan, were used to analyse the N efficiency of plants.
The proportion of plants that survived was analysed using a generalized linear model with logit-function. Nitrogen-deposition treatment, population and altitude were treated as fixed factors. As no significant effects of altitude, or interactions between altitude and the other factors, were observed, different altitudes were grouped. Nitrogen-deposition treatment and origin (region) were treated as fixed factors in subsequent analyses. The same method was used to analyse the proportion of plants that flowered; in this analysis, only plants that survived to this stage were included. The following analyses were performed on flowering plants only: the explanatory variables biomass of the fresh leaves, inflorescences and the total aboveground biomass, the inflorescence length, total number of leaves (fresh and senescent), proportion of leaf senescence, C and N content of fresh leaves and C : N ratio, which were all analysed using generalized linear models with N deposition and origin (region) as fixed factors. Again, we grouped data at different altitudes, as initial tests indicated no differences. Total aboveground biomass, biomass of inflorescences and inflorescence length were arcsine square-root transformed in order the meet the criterion of a normal distribution. Parametric survival models for censored data with a lognormal distribution were used to analyse the effect of N deposition and origin (region) on flowering time. Data at different altitudes were grouped as initial tests indicated no differences. Nitrogen deposition and origin were treated as fixed covariates. The average growth rate of each individual was expressed as the slope of the line between total aboveground biomass and time. The effect of N deposition and origin (region) on average growth rate was analysed subsequently using generalized linear models in which only the slopes of those lines with R2 > 0.8 were included. This resulted in the deletion of 18 out of 207 data points, which were evenly distributed among treatments. Different altitudes were grouped, as no differences were observed in initial tests.
To analyse the NUE data, we initially fitted a linear regression model with N-deposition treatment, population and altitude as fixed factors, including all interaction terms. Model validation gave no indication of nonlinearity, but revealed heteroscedasticity. In this case, we used linear regression with the generalized least squares extension (Pinheiro & Bates, 2000; West et al., 2006), which uses variance−covariate terms to allow for unequal variance (Pinheiro & Bates, 2000), thus benefiting from retaining the original variance structure in the data. To find the minimal adequate model, we adopted the approach outlined by Verbeke & Molenberghs (2000); Diggle et al. (2002). First the optimal structure of random components was determined using restricted maximum likelihood estimation. Then maximum likelihood estimation was used to determine the optimal fixed-effects structure. The optimal random structure was determined with the use of Akaike information criteria (Sakamoto et al., 1986; Burham & Anderson, 2002) in conjunction with likelihood ratio tests and plots of residuals versus fitted values. The optimal fixed structure was established by applying a backward selection using the likelihood ratio test. The numerical output of the optimal model was obtained using restricted maximum likelihood estimation (West et al., 2006). The importance of each explanatory factor in the minimum adequate model was assessed by comparing this model with a reduced model (with all the terms involving the factor of interest removed), using the likelihood ratio test. All analyses were performed using the nlme package (ver. 3.1, Pinheiro et al., 2006) in the r (ver. 2.3.0) statistical and programming environment (R Development Core Team, 2005).
The leaf lifespan data were analysed using a linear regression as performed for NUE. Model validation here gave no indication of violation of heterogeneity of variance, nonnormality, nonlinearity or strongly influential points. We therefore proceeded with a backward selection using the F-ratio test.
Table 1 shows that regions with low atmospheric N deposition have low plant-available N in the soil (estimated as the summed and concentrations), and that regions with high atmospheric N deposition have high plant-available N; and that, in general, higher-latitude sites have lower plant-available N. Soil pH did not vary between the sites. Organic content was very low in all sites, and differed only between the low- and high-altitude site of the British Isles. Base cations (K, Na, Mg) and phosphate (total P) did not differ significantly between the different sites (data not shown). Average precipitation across all sites was 1684 mm yr−1, and most sites were close to this average figure. The only exceptions were the high-altitude British Isles site, with higher precipitation, and the Icelandic low-altitude site, with lower precipitation. The shortest growing season was found in Icelandic and Norwegian high-altitude sites, whereas the longest growing season was found in the British Isles and Norwegian low-altitude site. By contrast, the longest days during this growing season were found in Iceland and Norway (> 16 h).
Plant responses to different nitrogen treatments
Table 2 summarizes the results for each region and experimental N-deposition rate, together with the statistical significance of the origin, N deposition and their interaction. No significant differences between plants from low- and high-altitude populations were found within the same region in terms of biomass production, growth rate, flowering time, flowering percentage, percentage of leaf senescence, N or C content in the leaves, or C : N ratio. It was therefore decided to group low- and high-altitude populations within each region in subsequent analyses. Significant altitude effects were found for the variables NUE and leaf lifespan. Low- and high-altitude sites were therefore not pooled for these variables. However, as no consistent effect of altitude was found between regions, we interpret these as a population effect, rather than an altitude effect.
Table 2. Arabidopsis lyrata petraea plant characteristics of plants from different regions grown at different atmospheric nitrogen-deposition rates
3 kg ha−1 yr−1
9 kg ha−1 yr−1
27 kg ha−1 yr−1
Origin × Ndep
Asterisks represent the importance of the explanatory factor assessed by a comparison of the minimum adequate model with a reduced model using the likelihood ratio test.
Averages and levels of significance: ns, not significant; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.
Plants from the British Isles had a significantly lower survival percentage than Norwegian and Icelandic plants, but N deposition did not affect plant survival. Norwegian and Icelandic plants had a higher aboveground growth rate than plants from the British Isles (Fig. 2). All populations grew more quickly at higher N deposition, but a significant interaction effect (df = 4, F = 2.983, P = 0.021) showed that this response was much stronger in Norwegian and Icelandic plants. Although plants from the British Isles had the lowest growth rate, they had a higher aboveground biomass at time of flowering than Norwegian and Icelandic plants (Table 2). This is explained by differences in flowering time, with Norwegian and Icelandic plants having a much shorter time to flowering than plants from the British Isles (Table 2). All plants had a higher biomass at time of flowering when grown at the high rate of N deposition. Norwegian and Icelandic plants flowered significantly earlier at higher N deposition, whereas the flowering time of the British Isles plants did not change (Table 2; N deposition × origin: df = 4, F = 5.7673, P < 0.001). Flowering percentages were not significantly affected by treatments, nor were there any differences between the different origins. Also, no N effects were observed on the biomass of the inflorescences. The inflorescences of the Icelandic plants were smaller than those of Norwegian and British Isles plants, irrespective of N treatment.
Plants from Norwegian and Icelandic populations had a significantly higher proportion of senescent leaves than the plants from the British Isles at all N treatments (Fig. 3). However, the proportion of leaf senescence in plants from the British Isles decreased at the high N-deposition rate, whereas it increased in Icelandic plants (N deposition × origin: df = 4, F = 5.0814, P < 0.001). In Norwegian and Icelandic plants, total leaf number produced did not change with higher N-deposition rates, but plants from the British Isles produced a greater total number of leaves at higher N-deposition rates (N deposition × origin: df = 4, F = 20.45, P < 0.001). Leaves from Norwegian and Icelandic plants had a significantly shorter lifespan than leaves from plants from the British Isles (Table 2). Higher N deposition increased the leaf lifespan of plants from the British Isles, whereas it decreased the lifespan of the Norwegian leaves.
The C : N ratios in the fresh leaves of all populations decreased significantly with increasing N-deposition rates (Fig. 4). Icelandic plants showed the strongest response with a decrease of approx. 50% with increasing N-deposition rates, whereas the response of plants from the British Isles was < 25% (Table 2). At 3 kg N ha−1 yr−1, C : N ratios were highest in the Icelandic plants. However, at the medium and high N-deposition rate there was no difference in C : N ratio between the populations (N deposition × origin: df = 4, F = 7.898, P < 0.001). The decrease in C : N ratio with increasing N-deposition rates was mainly driven by an increase in the N content of leaves in the plants from the British Isles (Table 2), whereas this shift in Norwegian and Icelandic plants was driven by a significant increase in N as well as a significant decrease in C content (Table 2).
The NUE decreased with increasing N-deposition rate of all plants, but the decrease in NUE at elevated N-deposition rates was far more pronounced in the Norwegian and Icelandic plants than the plants from the British Isles (Fig. 5). Plants from the British Isles always used their N less efficiently than did the Norwegian and Iceland plants, even at low N-deposition rates (Fig. 5).
Plants from all populations responded positively to N addition in terms of a higher growth rate and the production of more biomass, which is consistent with numerous N-fertilization experiments (Jones et al., 2002; Van den Berg et al., 2005). In this study, however, we also found for the first time clear differences between plants from different origins, with significant N deposition × origin interactions for the growth rate of the total aboveground biomass, proportion of senescent leaves, time to flower, leaf lifespan, NUE, and percentage of N and C : N ratio in the fresh leaves.
Differences in plant performance
The large differences in flowering time between plants from the British Isles and plants from Norway and Iceland could be an adaptation to the different local climates rather than the different local N-deposition rates. There are some indications that local conditions favour different flowering phenology in different regions. Flowering time was observed to be highly plastic in a field-based reciprocal transplant common-garden experiment performed by the authors (data not shown), with accessions from all four areas (and several others) planted out in sites including Wales, Sweden and Iceland. All plants had lower flowering percentages and longer times to flowering when planted in the lower-latitude field sites, and higher flowering percentages and shorter flowering times in the higher-latitude field sites. Nonetheless, in each site (as in the controlled-environment trials reported here) the Norwegian and Icelandic accessions flowered earlier than those from the British Isles. There thus appear to be both genetic adaptations and plastic responses to regional differences in conditions, which lead to the differences in reproductive timing. However, the differences observed in this experiment in the responsiveness of flowering to N availability suggest that local climate is not the only factor influencing reproductive timing. The results reported here show that increased N availability significantly decreased time to flowering in Norwegian and Icelandic plants, but no significant effect was found on those from the British Isles. These results are in partial agreement with another study in which the response in flowering time on N addition was analysed for the related Arabidopsis thaliana (Pigliucci & Schlichting, 1998). In this study, a significant N effect on flowering time in one out of four accessions was observed, indicating differences between plants from different accessions.
Total aboveground biomass was analysed at time of flowering to ensure that all plants were measured in the same life stage. However, because of the differences in flowering time between plants of the different regions, the number of days over which biomass was formed differed substantially. Plants from the British Isles had a longer time to flowering, and therefore used more time to produce their biomass. Hence the calculated aboveground growth rate until flowering provides a better measure to compare the growth response as it corrects for differences in flowering time.
Of the three plant performance parameters, the average aboveground growth rate of Icelandic and Norwegian plants increased significantly more when grown at higher N levels than plants from the British Isles. In addition, the time to flower in these plants decreased significantly at high N-deposition rates, suggesting that they initiate flowering based on a size threshold, which is reached more quickly at faster growth rates. This explains why there is no significant difference between populations in effects of N deposition on aboveground biomass, despite the average aboveground growth rate showing this differential response.
Differences in nitrogen-use efficiency, leaf lifespan and leaf senescence
All plants in our study were found to use N less efficiently at higher N-deposition rates. This is in line with other studies showing decreased NUE under conditions of higher N availability (Vitousek, 1982; Aerts & De Caluwe, 1994; Bowman, 1994). Although these studies compared between-species differences, and not within-species as here, their conclusions are applicable to our study system. We found that plants from Norway and Iceland (regions with low N deposition) use N more efficiently than plants from the British Isles (where N deposition is substantially higher), correlating well with the differences in regional N-deposition rates. Thus our observations confirm the expectation that plants from low-N habitats decrease their NUE in response to increased N availability, suggesting evolutionary adaptations to regional deposition rates – a result not reported previously. Moreover, our results suggest that A. l. petraea plants from high-N areas of the British Isles have largely lost the ability to use N highly efficiently when grown at low rates of N availability. It is not unlikely that similar responses have evolved in other plant species facing higher N availabilities.
Plants may adapt to nutrient-poor habitats, either by maximizing the assimilation of N or by reducing N losses. Berendse & Aerts (1987) showed that in nutrient-poor environments, selection is for traits that minimize nutrient losses rather than conferring a high NUE per se. Nutrient losses can be strongly reduced by increased leaf longevity (Chapin, 1980; Berendse & Aerts, 1987). Negative correlations between leaf lifespan and the nutrient status of the habitat have been found in several studies (Aerts, 1990; Reich et al., 1992). Next to nutrient availability, leaf lifespan has often been recognized to be related to photosynthesis (Chabot & Hicks, 1982; Reich et al., 1992), with younger leaves having a higher net photosynthesis rate than older leaves (Westoby et al., 2002; Wright et al., 2002). Plants originating from areas with high N availability (British Isles) had longer leaf lifespans than plants from areas of low N deposition. However, the responses to the experimental N deposition were inconsistent, while the Icelandic and British populations showed a small increase in leaf lifespan at the highest rate of N deposition, the Norwegian population had a reduced leaf lifespan in this treatment. This is consistent with the findings reviewed by Aerts & Chapin (2000), who note that various effects of increased nutrient availability on leaf lifespan have been found, suggesting that leaf lifespan is not unambiguously controlled by nutrient availability.
Nutrient resorption from senescing leaves is another trait suggested to minimize nutrient losses. Resorption of nutrients from senescing leaves is an efficient adaptation to nutrient-poor conditions, because it enables plants to reuse these nutrients and thereby leads to a higher nutrient retention (Chapin, 1980; Aerts, 1990). Nutrients that are resorbed during senescence are directly available for further growth, whereas nutrients that are not resorbed will eventually be circulated via litter fall. The significantly higher proportion of leaf senescence in Icelandic and Norwegian plants, as compared with plants from the British Isles, can therefore be a response to the low-nutrient conditions in these countries. Taylor & Bell (1988) compared the responses to lower concentrations of NO2 of a population occurring at high NOx-polluted sites (close to a nitrogen fertilizer factory) grown in soils of 100 or 10 mg N kg−1. Exposure to elevated concentrations of NO2 caused a significant increase in growth in both this population and one from a site with low NOx levels. However, there was a significant difference between populations in terms of leaf senescence and root : shoot ratio; the population from the site high in NOx had higher leaf senescence at low than at high soil N concentrations. Taylor & Bell (1988) attributed this effect to a poor performance of the population from the high-NOx site when grown at low N availability. Their results are in line with our findings, which also show a higher proportion of leaf senescence for the plants from the British Isles when grown at low N-deposition rates.
Differences in C : N ratio
Decreases in C : N ratios with increasing experimental N deposition have been reported in a number of N-addition experiments (Leith et al., 1999; Carroll et al., 2003), in line with our study. By contrast, a recent survey of plant tissue along an N-deposition gradient in the UK revealed no significant N-deposition effects on tissue N (Stevens et al., 2006), and it was suggested that the responses in plant tissue N are very plastic. Our study revealed plastic responses for plant tissue N. In addition, a significant reduction in plant tissue C was found in Norwegian and Icelandic plants with increasing N deposition, whereas plant tissue C remained stable in plants from the British Isles. This negative correlation in the Norwegian and Icelandic plants may be attributed to a higher C demand as a result of increased assimilation of the different N forms and respiration in combination with low levels of C assimilation caused by senescence (Noodén & Guiamet, 1989). As a result of the negative response of tissue C to N addition, a significant N deposition × origin effect was found for C : N ratio. Icelandic and Norwegian plants reduced their C : N ratio to that of plants from the British Isles when grown at high N deposition, but plants from the British Isles did not increase the C : N ratio to the same degree as Norwegian and Icelandic plants when grown at low N deposition. These results thus show a highly plastic response of C : N ratio in the Norwegian and Icelandic plants in contrast to plants from the British Isles.
Possible interference with climate
One difficulty with interpreting the regional differences is that our focal regions differ in many ways, not just in atmospheric N deposition. We cannot exclude the possibility that the differences reported here have arisen in response to some other aspect of these regional environmental differences. The most striking difference between our focal regions is in climate, with dramatic differences in temperature (degree days < 0°C in particular), and thus in growth-season length. We therefore selected populations from low- and high-altitude areas within each region, to allow a within-region comparison of the effects of temperature and growth-season length. These within-region differences were larger than the between-region differences (except for the difference in growth season between Iceland and the British Isles), but no major within-region differences in plant response were observed for most traits, apart from NUE and leaf lifespan. Although the within-region differences measured for NUE and leaf lifespan were significant, the differences between low- and high-elevation sites were not in the same direction and did not show such strong responses compared with the between-region effects. The same applies to the different N treatments. This suggests that within-region effects such as temperature and growing season length had a minor effect on evolution of the traits measured on a regional scale in this experiment. However, we cannot rule out the possibility that climatic effects have interfered at coarser scales, in particular for climatic effects that differ between regions.
Evolutionary response to nitrogen deposition?
The results of this study clearly demonstrate that plants from regions with different N deposition have the potential to respond differently to N, and this is likely to be an adaptive response to the different selection pressures they face. It is important to note that, while the Icelandic and Norwegian plants are likely to have experienced relatively small changes in N deposition over the past 50 yr, plants from the British Isles will have experienced substantial increases in the current rates of deposition, and even if N-deposition inputs are reduced because of legislation, the ecological effects will persist for many years (Power et al., 2006). Our results clearly show that adaptation of these populations to the local increase in N deposition has resulted in the loss of plastic responses to reductions in N availability.
These results were found at N-deposition rates of 27 kg N ha−1 yr−1 at the most, in a relatively short period. 27 kg N ha−1yr−1 is relatively low in terms of application rates used in N-fertilization experiments (Bergholm et al., 2007; Pilkington et al., 2007), and is comparable with current and predicted future N-deposition rates in many ecosystems (Galloway et al., 2004; Phoenix et al., 2006). Our results demonstrate the potential for atmospheric N deposition to be a selective force for adaptation, and demonstrate that an adaptive response to an environmental stress can evolve over a relatively short timescale of a few decades. Similar conclusions have been drawn by Wu et al. (1975), who found that adaptation of the perennial grass Agrostis capillaris to elevated SO2 evolved in a period of 22–34 yr; by Wilson & Bell (1985), who provided strong evidence for evolutionary adaptation to SO2 by the grass Lolium perenne at highly polluted sites in the UK in a period as short as 4–5 yr; and by several other studies of evolutionary responses to climate change (Davis et al., 2005; Jump & Peñuelas, 2005; Thomas, 2005). The rate at which adaptation can evolve partly determines the rate at which populations can colonize new habitats (García-Ramos & Rodriguez, 2002), and the probability that populations will go extinct in their present ranges in the face of environmental change (Bürger & Lynch, 1995). The rate of adaptation is therefore of major interest, as it is critical for predicting plant response to global change (Davis et al., 2005).
Implications for biodiversity and nature management
Current nature management and policy-making with respect to atmospheric N deposition is based on critical loads, empirically derived maximum tolerable loads for a given ecosystem (Bobbink et al., 2003). These critical loads do not take into account possible adaptive responses of plants or ecosystems to elevated N-deposition rates. If species adapt to the local deposition rate, critical loads cannot be extrapolated to other areas with different rates or histories of N deposition. Differential capacity between species to adapt to increased N deposition may be an important long-term driver of changes in species composition that has not been considered previously. In Europe, N deposition is now decreasing, and our results suggest that the reverse adaptation to lower rates of N availability may also be a significant factor in response to this change. Warming of high-latitude regions is likely to lead to greater nutrient cycling (Chapin et al., 1995; Kirschbaum, 1995; Rustad et al., 2000), hence the capacity and speed of adaptation of populations to increased N availability is also relevant in this context. The presence of adaptive genotypes, and the rate of, and capacity for, adaptation to changing N deposition may change perspectives on several aspects of research, predictive modelling and policy development in global environmental change.
This study was funded largely by the Natural Environment Research Council under the Post-Genomics and Proteomics Programme (NE/C507837/1). In addition, the work was partially funded by the Frye Stipendium from the Radboud University Nijmegen, the Netherlands. The authors would like to thank the UK Environmental Change Network (ECN) and the meteorological weather stations for providing data, and English Nature, the Countryside Council for Wales, Scottish National Heritage, and the landowners who permitted access to their land.