THE ECOLOGICAL MEANING AND PREDICTIVE ACCURACY OF nDEV SPECIES’ VALUES
Three observations make us believe that the Ndev values have an ecological meaning:
The Ndev values of the most common species were positively correlated between neighbouring regions.
They were significantly positively correlated with species changes in frequency observed in long-term studies in Skåne and to some extent in those in Central Europe.
3 They showed a good agreement with the responses of species found in fertilization experiments.
The third result in particular indicates that Ndev is indeed related to the species’ abilities to respond to enhanced nitrogen levels, although the same conclusion may be drawn from (2), since nitrogen deposition and enrichment of ecosystems (Bobbink et al. 1998), as well as soil acidification and forest management (Falkengren-Grerup et al. 1998; Diekmann et al. 1999), have occurred over recent decades. Some species with positive Ndev scores appear to increase in all the regions, including several ferns (Athyrium filix-femina, Dryopteris spp.) and species such as Epilobium angustifolium and Rubus idaeus, while others with negative scores, such as Maianthemum bifolium and Vaccinium spp., show consistent decreases.
Ndev values are constructed using the mean field values of percentage minNO3− and pH at which species performance is optimum. Failure to consider the amplitudes around this optimum may explain why the Ndev values fail in some cases to explain responses observed in the field. In general, although correlations of Ndev with the observed species were usually significant, r2 values were rather low. Galium odoratum, for example, attained a high positive Ndev score in Skåne, although its abundance at least in permanent forest plots (Table 2) has decreased. This species obviously prefers relatively nitrate-rich sites, but is unable to tolerate the high hydrogen and aluminium ion concentrations on acid soils (Falkengren-Grerup & Tyler 1992). It may therefore have a skewed response to pH that is not taken into consideration by the current formulation of Ndev. Moreover, the accuracy of Ndev in general is dependent on the homogeneity of the data set used (improved by a high number of sample plots, with the species under consideration present in most –Galium odoratum was present in only 5% of the sites in Skåne, the required minimum – and, above all, a normal or even distribution of values of the measured variables, i.e. pH and percentage minNO3−). The responses shown were derived from both fertilization experiments and permanent plot studies from different ecosystems and over different time spans, and cover a strong gradient in nitrogen deposition and accumulation in an area where successional changes in deciduous forests are often slow. The sites may also have been affected by forestry and other management practices, including clear-cutting, thinning, coppicing and grazing, etc. The opening of the forest in particular may have considerable effects on species richness and composition of the herbaceous flora that may be difficult to separate from those of eutrophication. Environmental changes of various kinds may also have different effects on plant species, depending on the vegetation type in question and on the presence of species with high competitive abilities.
Theoretically, a soil with 100% nitrification but a very low rate could have a higher percentage minNO3− value than a soil with a higher production of nitrate and any amount of ammonium, despite its lower Ndev value. However, this is more a theoretical than a practical problem with the use of Ndev as the correlation between minNO3− and percentage minNO3− is positive and highly significant for the studied sites (r2 = 0.512, P < 0.001, n = 648). We therefore believe that percentage minNO3− can be used to represent the complexity of nitrogen availability in a soil: it takes into account the form of inorganic nitrogen produced in the soil and also, indirectly, through the correlation with the nitrification rate, the amount of both ammonium (low percentage minNO3−) and nitrate.
SPECIES’ RESPONSES AND LIFE HISTORY TRAITS
Various mechanisms responsible for vegetation changes due to increased nitrogen deposition and availability have been recognized, namely changes in soil chemistry per se, an increased density of the tree canopy (Kellner & Mårshagen 1991), altered patterns of herbivory (Ellenberg 1988) and changes in the mycorrhizal structure of the soil (Arnebrant 1994). The changes in competitive relationships between species, however, appear to be of primary importance (Kellner & Redbo-Torstensson 1995; Bobbink et al. 1998), and these are likely to be affected by life history traits. Using functional traits or groups of species sharing common attributes has the advantage of arriving at more general conclusions than a single-species approach.
Some attributes were, at least partly, correlated with observed responses, although less well than the ecological measure (Ndev) based on soil chemistry. Neither the analysis of taxonomy nor life form gave any clear results. This was not unexpected as both are large and heterogeneous groups. The physiological amplitude of single taxonomic groups such as monocots is very broad, and the same is true for most Raunkiaer life forms (such as the hemicryptophytes) defined on the basis of the position of regenerating buds over the inactive season. Positive and negative changes in abundance were observed in both broad- and narrow-leaved grasses, in contrast to studies showing that narrow-leaved grasses were favoured by nitrogen deposition (Rosén et al. 1992), but consistent with observations that potential growth rate varies with factors other than leaf width (Falkengren-Grerup 1998).
Plant height was the best predictor among the life history traits tested and was also significantly positively correlated with Ndev (for Skåne, for example, r2 = 0.109, P = 0.007, n = 66). This means that tall-growing species are generally favoured by a high nitrogen deposition and percentage minNO3−: pH ratio at any given site, at the expense of smaller species; such species are able to overgrow and thereby outshade and outcompete smaller species when conditions are favourable. Height was positively related to RGRmax (r2 = 0.168, P = 0.024, n = 30), an observation also made by Grime & Hunt (1975) in an experimental study of grassland species. Leaf nitrogen concentration showed higher correlations with the observed species trends than the other two physiological traits. In their screening of life history attributes Grime et al. (1997) found a close correlation between foliar nutrient concentrations and the capacity for rapid growth under productive conditions (see also Meerts 1997). The primary strategies of the C-S-R model developed by Grime (1979), on the other hand, did not show any agreement with the observed species trends and do not appear to be good predictors of the species’ responses to nitrogen deposition. The large majority of species in deciduous forests are competitors or stress-tolerant and/or ruderal competitors, and the mean responses of species that were observed did not differ between strategy types.
Plant anatomy was linked to the observed species trends and was also intercorrelated to N-conc (r2 = 0.168, P = 0.009, n = 40; cf. Meerts 1997). Many species with a more or less helomorph anatomy (Aegopodium podagraria, Anthriscus sylvestris, Mercurialis perennis, Urtica dioica) also have a high foliar nitrogen concentration and appear to be favoured by high nitrogen deposition. This coincides with the observation that plants cultivated under nitrogen shortage develop a more xeromorph leaf anatomy than plants cultivated on fertile soil (Müller-Stoll 1947; cited in Ellenberg 1996). In other words, increased nitrogen availability may favour an increasingly hydro-helomorph architecture. A second observation made in grasslands is that fertilization of dry grasslands causes the vegetation to develop a more mesic character (i.e. nitrogen in a way ‘replaces’ water), because plants on nitrogen-rich soil have a better water-use efficiency (Ellenberg 1996).
We did not expect any correlation between phenological development and species trends, because the scores reflect the sequence of flowering rather than the growth of roots, shoots and leaves. However, there seems to be a positive correlation between the seasonal development of different parts of the plant, and there was a trend (albeit with low r2) for late-developing species to increase in abundance. One reason for this may be that phenological development and plant height are highly positively correlated (r2 = 0.368, P < 0.001, n = 130).
The results of our study allow us to define a so–called attribute syndrome, i.e. a set of co-occurring attributes (Lavorel et al. 1999). We predict that species that are found in sites with a high ratio of percentage minNO3− to pH or have tall stature, hydro- to helomorph anatomy, high leaf nitrogen concentration and late phenological development, will increase in abundance in response to enhanced nitrogen levels. Athyrium filix-femina, Epilobium angustifolium and Urtica dioica are examples where these attributes co-occur, but Calamagrostis arundinacea and Pteridium aquilinum, two tall and late-developing species, have negative Ndev values and, partly, negative trends, and Moehringia trinervia, a small, rather early developing species with positive Ndev values, appears to increase in abundance. Other, less well-known attributes of forest vascular plants therefore appear to affect the species’ responses to nitrogen enrichment, such as morphological characteristics of the plant or of the plant-mycorrhizal root system, the nutrient (other than nitrogen) content of the plant, etc. We need to gain a better knowledge of the basic life history of species in order to be able to make more accurate predictions of their responses to atmospheric deposition. Another interesting task would be to study whether relevant life history traits differ in importance between deciduous forest sites with different soil acidity and nutrient regimes, or between deciduous forests and other habitat types suffering from airborne nitrogen deposition, such as grasslands, heaths and wetlands.