II Variability of N : P ratios in response to nutrient supply
III Critical N : P ratios as indicators of nutrient limitation
IV Interspecific variation in N : P ratios
V Vegetation properties in relation to N : P ratios
VI Implications of N : P ratios for human impacts on ecosystems
Nitrogen (N) and phosphorus (P) availability limit plant growth in most terrestrial ecosystems. This review examines how variation in the relative availability of N and P, as reflected by N : P ratios of plant biomass, influences vegetation composition and functioning. Plastic responses of plants to N and P supply cause up to 50-fold variation in biomass N : P ratios, associated with differences in root allocation, nutrient uptake, biomass turnover and reproductive output. Optimal N : P ratios – those of plants whose growth is equally limited by N and P – depend on species, growth rate, plant age and plant parts. At vegetation level, N : P ratios <10 and >20 often (not always) correspond to N- and P-limited biomass production, as shown by short-term fertilization experiments; however long-term effects of fertilization or effects on individual species can be different. N : P ratios are on average higher in graminoids than in forbs, and in stress-tolerant species compared with ruderals; they correlate negatively with the maximal relative growth rates of species and with their N-indicator values. At vegetation level, N : P ratios often correlate negatively with biomass production; high N : P ratios promote graminoids and stress tolerators relative to other species, whereas relationships with species richness are not consistent. N : P ratios are influenced by global change, increased atmospheric N deposition, and conservation managment.
This review aims to summarize our current knowledge on relationships between plant N : P ratios and the composition and functioning of natural and seminatural vegetation, mainly from temperate and boreal regions. Four main questions are addressed: (1) how variable are plant N : P ratios, and how does this variation depend on N and P concentrations; (2) what differences in plant and vegetation properties are associated with variation in N : P ratios; (3) what mechanisms cause these differences; and (4) do N : P ratios indicate whether biomass production is N- or P-limited?
Ecological research in temperate and boreal regions has tended to stress the role of N in plant growth and ecosystem processes (e.g. Roy & Garnier, 1994). Mechanisms determining plant responses to P deficiency have mainly been investigated in agricultural research; this literature is referred to here when studies with wild plants are insufficient. Several important topics in relation to plant nutrition are not addressed in this review as they have been dealt with extensively elsewhere. These include soil properties and processes that determine nutrient availability to plants, as well as root properties and mechanisms of nutrient acquisition (Marschner, 1995; Raghotama, 1999; Frossard et al., 2000; Whitehead, 2000). All these factors are implicit in the definition of ‘nutrient availability’ in the following, as they jointly determine nutrient uptake by plants. Finally, as well as describing relationships between N : P ratios and ecological patterns or processes, this review seeks to explain these relationships by examining how regulation mechanisms depend on N and P concentrations in plants.
The first part of this review deals with plastic responses of plants to variation in nutrient supply (Sections II–III). Interspecific differences and the species composition of the vegetation are then considered (Sections IV–V). The last part discusses implications for human impacts on ecosystems (Section VI) and the use of N : P ratios in ecological research (Section VII).
II. Variability of N : P ratios in response to nutrient supply
1. Patterns of variation
Nitrogen and phosphorus concentrations in plant biomass, [N] and [P], are determined by the balance of N and P uptake, C assimilation, and the losses of C, N and P through turnover, leaching, exudation, herbivores and parasites (Chapin & Shaver, 1989; Aerts & Chapin, 2000; Eckstein & Karlsson, 2001). Biomass N : P ratios (quotient of [N] and [P]) are not directly influenced by the plant's C economy but reflect the balance between uptake and losses of N and P. N : P ratios in individual plant parts may depend additionally on internal nutrient translocation. The N : P ratios can be expressed either as mass ratios (g N/g P) or as atomic ratios (mol N/mol P), which differ by a factor of 2.21. The use of molar ratios is most common in physiological research and in aquatic ecology because they reflect the actual stoichiometric relationships (Sterner & Elser, 2002), but most literature in terrestrial ecology reports mass ratios; to facilitate the comparison with this literature, mass N : P ratios will also be used here.
The average N : P ratio of terrestrial plant species at their natural field sites is 12–13 (Elser et al., 2000a; Güsewell & Koerselman, 2002; Knecht & Göransson, 2004), similar to the average N : P ratio of aquatic plants and algae (Geider & La Roche, 2002). However, N : P ratios can vary widely, and individual measurements range from approximately 1–100 (Schmidt et al., 1997; Phoenix et al., 2003; Tomassen et al., 2004). N : P ratios vary among and within species. For wetland plants grown at their natural field sites, Güsewell & Koerselman (2002) found that intraspecific variation was more important than interspecific variation: N : P ratios varied on average by 34% (= SD/mean) among sites for a given species, but only by 25% among species that co-occur at a site. This contrasts with [N], which varied more among species (26%) than among sites (22%); for [P] both coefficients of variation were equal (47%).
The N : P ratios of individual species may vary fifty-fold in response to natural or experimental variation in N and P supply, reflecting variation in [N] or [P] or both together. Relationships between [N], [P] and N : P ratios depend on the set of nutrient conditions to which plants are exposed, and on the plasticity of a species in [N] and [P]. When a species is sampled at different field sites, [N] and [P] typically correlate positively with each other (Fig. 1a) because the same sites tend to be N- and P-rich, and because nonnutritional factors (e.g. shade, drought) have similar impacts on [N] and [P]. When plants are grown experimentally with N and P supplied in differing proportions, [N] and [P] tend to correlate negatively (Fig. 1b). Variation in N : P ratios is primarily determined by variation in [P] with graminoid species because their [N] is relatively stable (Fig. 1a,b); variation in [N] is often more important in determining the N : P ratios of woody plants, bryophytes or lichens (Fig. 1c).
Although biomass N : P ratios reflect the relative availability of N and P the correspondence is not exact because of homeostatic regulation by plants: 10-fold variation in N : P supply ratios causes only two- to three-fold variation in biomass N : P ratios (Güsewell & Koerselman, 2002). In Carex curta, shoot biomass N : P ratios increased from 2.5–4.5 to 23–45 as the N : P ratio of nutrient solutions increased from 0.6 to 405 (Fig. 2). The relationship between plant N : P ratios and N : P supply ratios can be summarized by the regulatory coefficient H, defined as the inverse slope of a regression line fitted to log-transformed variables (Sterner & Elser, 2002). Various growth experiments with herbaceous plants grown at N : P supply ratios between 1 and 100 revealed regulatory coefficients between 1.7 and 4.6 (recalculated from Shaver & Melillo, 1984; Ryser & Lambers, 1995; Güsewell & Bollens, 2003; Güsewell, 2004). Regulation of N : P ratios in plants is therefore stronger than in algae or fungi (H = 1–2, Rhee, 1978; Sterner & Elser, 2002) but weaker than in animals or bacteria (H = 3, Sterner & Elser, 2002; Jaenike & Markow, 2003; Makino et al., 2003). The degree of regulation depends on overall nutrient supply and on light intensity (Fig. 2).
Biomass N : P ratios are regulated primarily through adjustment of N and P uptake rates induced by whole-plant signalling mechanisms that are sensitive to the composition of the phloem (Imsande & Touraine, 1994; Raghotama, 1999; Forde, 2002). This regulation is both positive and negative: N-deficient plants increase the rate of N uptake and reduce the rate of P uptake while P-deficient plants do the opposite (Aerts & Chapin, 2000). Nitrate uptake is downregulated by feedback inhibition when nitrate accumulates in roots instead of being exported to shoots (Gniazdowska et al., 1999), or when amino acids not used for growth cycle back from shoots to roots (Imsande & Touraine, 1994). Ammonium uptake is downregulated less effectively, so that the ratio of ammonium to nitrate uptake increases in P-deficient plants (Schørring, 1986). Ammonium nutrition causes higher [N] in plants than nitrate nutrition with the same N and P supply (Andrews et al., 1999). Phosphorus uptake is down-regulated in response to high phosphate concentration in the phloem (Marschner et al., 1996; Schachtman et al., 1998). As an alternative to reduced uptake, storage of P as polyphosphate in stems, roots and seeds avoids toxic effects and can support growth at other times (Handreck, 1997). These storage tissues can have very high [P] (and low N : P ratios) when plants are grown at high P supply (Olsen & Bell, 1990).
Mechanisms regulating biomass N : P ratios have been studied mainly in laboratory experiments, but they can also be observed in field-grown plants. In Hawaiian rainforest the P uptake capacity of excised roots was reduced by P fertilization, whereas the N-uptake capacity was reduced by N fertilization at a N-limited site and increased by P fertilization at a P-limited site, reflecting changes in plant N and P requirements (Treseder & Vitousek, 2001). Phosphatase secretion and the P-uptake capacity of roots have been used to assess the P status of plants in the field (Harrison et al., 1991; Johnson et al., 1999; Phoenix et al., 2003).
Nutrient losses through root exudation and senescence may contribute further to the homeostatic regulation of biomass N : P ratios. Phosphorus deficiency may cause a substantial release of N from roots as phosphatases or amino acids (Ratnayke et al., 1978; Jones et al., 1994; Treseder & Vitousek, 2001). Nutrient losses with leaf litter depend on the efficiency of nutrient resorption during leaf senescence. Typically 30–90% of a leaf's N and P pool is resorbed (Aerts & Chapin, 2000). Nutrient resorption efficiency is often similar for N and P (Aerts, 1996), but plant species adapted to P-limited sites tend to resorb P more efficiently than N (Walbridge, 1991; Wright & Westoby, 2003). Plastic adjustments of resorption efficiency sometimes occur, but not consistently (Aerts, 1996). In plants with [P] far above a species’ normal range, P resorption may be reduced and P may accumulate even in the senescent leaves, while N resorption remains unchanged (Chapin & Moilanen, 1991; Uliassi & Ruess, 2002). Downregulation of N resorption has been reported only occasionally (Van Heerwaarden et al., 2003). For both elements these changes in resorption efficiency are not consistently related to biomass N : P ratios (S. Güsewell, unpublished). Nutrient losses with root litter are likely to be important as there seems to be little or no resorption during root senescence, but this has rarely been quantified (Aerts et al., 1992; Gordon & Jackson, 2000).
3. Relationships between N : P ratios and plant properties
Plants with high and low N : P ratios differ in various traits even if they grow equally fast (Marschner et al., 1996; Güsewell, 2004). Evidence from experiments with N- and P-deficient plants suggests that contrasting morphological responses to N and P supply could be mediated by cytokinins. High cytokinin levels in shoots stimulate shoot growth and delay leaf senescence (Smart, 1994; Gan & Amasino, 1997). The production of cytokinins and their transport from roots to shoots decrease immediately after a reduction in N supply; this represses growth before [N] becomes directly limiting (Lambers et al., 1998; Forde, 2002). Conversely, increased N concentration in the rooting medium stimulates cytokinin production even in plants that are already taking up N at maximal rate (De Groot et al., 2003). The effect of P supply on cytokinin levels is less rapid and less pronounced, and cytokinin production is less influenced by the P concentration in the rooting medium when P is not limiting (De Groot et al., 2003).
Leaf senescence is often accelerated by nutrient deficiency as nutrients are translocated from old to young leaves to support their growth (Schachtman et al., 1998). In perennial plants, P deficiency (high N : P ratio) accelerates leaf senescence more than N deficiency (low N : P ratio). For example, the life span of leaves of the evergreen rainforest tree Metrosideros polymorpha was shorter at a P-limited old site than at a N-limited young site, and it was also reduced by N fertilization (Cordell et al., 2001). In growth experiments with wetland forbs and graminoids, leaves of P-deficient plants senesced more quickly than leaves of N-deficient plants with similar biomass (Bollens, 2000; S. Güsewell, unpublished). The faster leaf senescence of P-deficient than of N-deficient plants might be caused by the fact that growth is not reduced immediately by P deficiency (Limpens et al., 2003a), so that growing leaves still act as sinks and cause the mobilization of P from older leaves (Usuda, 1995). Depletion of inorganic P (Pi) may cause the accumulation of carbohydrates, the inhibition of photosynthesis and a reduced N content, accelerating leaf senescence (Christensen et al., 1981; Halsted & Lynch, 1996; Chen & Lenz, 1997). Conversely, P deficiency did not accelerate leaf senescence of soybean plants that maintained [Pi] by reducing growth, so that the carbon metabolism was not disturbed (Crafts-Brandner, 1992).
Root turnover has been found to increase after N fertilization in several studies (Pregitzer et al., 1995; Fransen & de Kroon, 2001). In a P-poor fen, N fertilization increased the proportion of dead roots whereas P fertilization increased the proportion of living roots (El-Kahloun et al., 2003). However, Ostertag (2001) reported opposite effects for Hawaiian forest: root turnover was slightly faster at the N-limited site than at the P-limited site and was increased by P fertilization. The available data are not sufficient for generalizations about the effects of N vs P on root turnover.
Reproduction may require a considerable fraction of a plant's N and P: the reproductive allocation of N and P typically exceeds that of C and may reach 50–60% (Fenner, 1986). Seeds generally have higher N and P concentrations and lower N : P ratios than vegetative structures; seed N : P ratios reported for wild herbaceous plants range from 1.5 to 15 (Ascencio & Lazo, 1979; Fenner, 1986; Lewis & Koide, 1990; Miao et al., 1998; Köhler et al., 2001). Accordingly, flowering plants (or shoots) may have lower N : P ratios than nonflowering plants of the same species. For example, the N : P ratio of Milium effusum in subarctic forest was 7.5 for flowering shoots but 13.1 for nonflowering ones (Eckstein & Karlsson, 1997).
The low N : P ratio of seeds relative to that of shoots suggests that P limitation should affect reproductive output more than growth. Some experiments support this expectation: In Succisa pratensis the combination of high N and low P supply increased shoot biomass and seed number per plant but reduced seed size, germination rates and overall reproductive success (Vergeer et al., 2003). Often, however, the effects of nutrient supply on reproductive output correlate closely with those on plant biomass, suggesting no specific role for N or P (Allison, 2002). In ruderal Juncus species low P availability reduced plant growth and flowering, and in the field this effect was related to increased N availability (Brouwer et al., 2001). The effects of nutrient supply on nutrient concentrations of seeds are variable: In Senecio vulgaris nutrient-deficient plants increased their reproductive allocation of N and P relative to C and maintained constant nutrient concentrations in seeds (Fenner, 1986), whereas P-deficient plants of Sinapis alba and Abutilon theophrasti produced seeds with lower P content (Lewis & Koide, 1990).
III. Critical N : P ratios as indicators of nutrient limitation
1. Definition and assessment of nutrient limitation
Nutrient concentrations in plant tissues were first used as a way of assessing nutrient limitation in agriculture and forestry. Here plants are regarded as nutrient-deficient or nutrient-limited if their production (relative growth rate, RGR; annual herbage production; tree diameter increment; grain yield, etc.) is lower by a certain percentage (e.g. 10% or 20%) than the maximum reachable at high nutrient supply (Wells et al., 1986; Föhse et al., 1988). Limiting elements are those of which supply must be increased to obtain maximal production (Montañés et al., 1993; Sinclair et al., 1997).
In ecology it is widely accepted that ‘nutrient limitation to a particular process is demonstrable when a substantial addition of a particular nutrient increases the rate or changes the endpoint of that process’ (Vitousek & Howarth, 1991). Nutrient limitation is assessed in ‘well-controlled, well-replicated nutrient addition experiments’; the process is regarded as nutrient-limited the difference between fertilised and unfertilized samples is statistically significant (Vitousek & Howarth, 1991).
The agricultural and ecological concepts of nutrient limitation differ in the reference against which limitation is assessed. In the first case the reference is the maximal reachable yield, and the degree of nutrient limitation (= deficiency) can be quantified as the percentage difference between maximal and actual production. In the second case the reference is the current, unfertilised system, and the degree of nutrient limitation (= potential responsiveness) is quantified as the percentage increase of the process rate or endpoint after fertilization.
The type of nutrient limitation is assessed by comparing the degree of limitation for several elements (with the unfertilised situation as reference). The ‘Liebig law of the minimum’ proposes that growing plants require essential elements in certain proportions; if actual proportions differ from ideal ones, growth is limited by the element that is relatively scarcest (Sinclair & Park, 1993). For N and P, the ‘law of the minimum’ suggests that as long as [N] and [P] are suboptimal, there is a ‘critical N : P ratio’ below which growth is limited only by N and above which growth is limited only by P. When the biomass N : P ratio is equal to the critical value, plant growth is colimited by N and P, that is, both N and P must be added to obtain a significant increase. The ‘law of the minimum’ is an idealization (Sinclair & Park, 1993). In fact both N and P can stimulate growth or other processes because N supply often influences how efficiently P is acquired and used, and vice versa (Treseder & Vitousek, 2001; Güsewell et al., 2003a; Güsewell, 2004).
2. Critical N : P ratios for growth of young plants
Critical (also called ideal or optimal) N : P ratios describe the proportion of N and P in plants whose growth is limited to the same degree by N and by P. Critical N : P ratios have sometimes been derived from relationships between [N] and [P] in many plant samples, but the N : P ratios thus obtained may reflect nutrient conditions at the sampling sites rather than plant requirements (Sinclair et al., 1997). Experiments to determine critical N : P ratios can be designed in various ways, of which three are shown in Fig. 3:
1In two separate sets of treatments (N- and P-limited growth) the supply of either N or P is varied while the other element is freely available (optimal supply). Plant growth rates and biomass nutrient concentrations are determined for each treatment. Response curves are fitted to the relationships between [N] and growth rate for the treatments with N-limited growth, and to relationships between [P] and growth rate for the treatments with P-limited growth (Ingestad, 1979; Ågren, 1988). The critical N : P ratio is a function of growth rate, calculated as the ratio between [N] in N-limited plants and [P] in P-limited plants with corresponding growth rate (Fig. 3a).
2The supply of both N and P is varied in a factorial design. A response surface is fitted to the relationship between growth and the concentrations of both elements (Sinclair et al., 1997). Critical N : P ratios are those of plants whose growth would increase similarly if the supply of either N or P is raised to a nonlimiting level (Sinclair et al., 1996) (Fig. 3b).
3N : P supply ratios are manipulated through inverse variation in N supply and P supply. Raising N supply while reducing P supply increases growth as long as plants are N-limited but reduces growth when plants are P-limited (Güsewell, 2004). Critical N : P ratios are therefore those of plants with maximal growth across the N : P ratio treatments. This can be repeated with different absolute levels of N and P supply to yield critical N : P ratios for different growth rates (Fig. 3c).
The first approach was formalised by Ågren (1988) for young plants growing exponentially under steady-state conditions (Ingestad & Lund, 1979) . RGR was modelled as a linear function of the internal concentration of the growth-limiting nutrient (cN, cP) with fixed slope (nutrient productivity, PN or PP) and a minimal N or P concentration required for any growth to occur, cN,min or cP,min. The condition RGR= PN(cN − cN,min) = PP(cP − cP,min) then yields a non-linear relationship between critical N : P ratio and RGR:
The model was revised recently to include a quadratic relationship between RGR and the C : P ratio (Burns et al., 1997; Ågren, 2004). For Betula pendula these calculations yielded critical N : P ratios between 8 and 12, with a maximum at intermediate RGR (Ågren, 2004). Critical N : P ratios for tomato, Lycopersicon esculentum, increased from 4–5 at low RGR to 8–9 at near-maximal RGR (De Groot et al., 2003). Using the second approach (response surface modelling) for the yield of field-grown Trifolium repens, Sinclair et al. (1997) found that critical N : P ratios decreased from 18 to 13 with increasing yield (faster growth). Thus critical N : P ratios determined in short-term growth experiments differ among species and depend in a species-specific way on the growth rate of plants. Relationships between RGR and N : P ratios have also been described for algae (Elrifi & Turpin, 1985) and bacteria (Makino et al., 2003).
Interspecific differences in critical N : P of young plants ratios reflect differences in plant traits and physiological mechanisms that determine how efficiently N and P are used for growth. Nitrogen requirements are largely determined by photosynthetic N use efficiency (Garnier et al., 1995), which varies widely among species because of differences in leaf morphology, internal distribution of N and production of N compounds not involved in photosynthesis (Lambers & Poorter, 1992). Variation in P requirements can be related to, for example, rates of internal phosphate recycling (Nanamori et al., 2004); the ability to export and transform sugars from chloroplasts under P deficiency (Nanamori et al., 2004); or the availability of alternative P-saving metabolic pathways for glycolysis and mitochondrial respiration (Vance et al., 2003).
3. Critical N : P ratios of older plants
Critical N : P ratios of established plants may differ from those of young plants in growth experiments because of the different age and function of tissues. Young plants consist largely of immature leaves that assimilate and grow simultaneously, so that the demands of N and P are given by the stoichiometry of basic biochemical processes (photosynthesis, respiration protein synthesis, DNA duplication and DNA transcription). When plants are older, growth is restricted to active meristems (e.g. young leaves, shoot tips or inflorescences). Mature leaves are still photosynthetically active but no longer grow, which greatly reduces the P requirements for RNA. Nucleic-acid P can therefore be mobilized and translocated to young leaves, leading to higher plant-level N : P ratios (Usuda, 1995). In Eucalyptus cloeziana seedlings grown at low P supply, the N : P ratio of developing leaf blades was c. 15 regardless of N supply, whereas N : P ratios of mature leaves ranged from 11 at low N supply to 48 at high supply, and those of old leaves from 16 to 55 (Olsen & Bell, 1990).
During leaf senescence, the N and P concentration is reduced further (Batten & Wardlaw, 1987). If relatively more P than N is resorbed at this stage, and if senesced leaves are not shed immediately, the plant-level N : P ratio will increase again. This increase is most pronounced in graminoids at P-poor sites: the evergreen sedge Cladium mariscus had N : P ratios of 6–8 in its young leaves but 15–25 in shoot bases, roots and rhizomes, and 25–40 in senesced leaves (Pfadenhauer & Eska, 1986); young plantlets developing vegetatively on the inflorescences had a N : P ratio of 12 (Miao et al., 1998). The deciduous sedge Rynchospora alba had N : P ratios of 7–9 in the growing inflorescences but 16–25 in the rest of the shoot (Ohlson & Malmer, 1990). In the deciduous grass Molinia caerulea N : P ratios of young shoots (April) were 9.2–9.7, whereas those of old shoots (September) were 21–36 (El-Kahloun et al., 2000). Thus, despite low [P] and high N : P ratios at the whole-shoot level, those in actively growing parts were close to the optimal values found in short-term growth experiments.
The direct reallocation of nutrients from old to young leaves plays an essential role in the nutrition of graminoids adapted to nutrient-poor sites (Jonasson & Chapin, 1985; Bausenwein et al., 2001). In woody plants and forbs the importance of this process varies widely as species differ in their ability to maintain essential metabolic processes at low internal nutrient concentrations (Halsted & Lynch, 1996; Vance et al., 2003). Some species show the same patterns in nutrient concentrations and N : P ratios as graminoids (Olsen & Bell, 1990), whereas others have high nutrient concentrations and low N : P ratios (as in young leaves) even at nutrient-poor sites (Jonasson, 1989; Niva et al., 2003). Differences in the internal reallocation of nutrients are essential in explaining why relationships between nutrient limitation and N : P ratios of mature plants differ among species.
Further changes in growth responses of perennial plants to N : P supply ratios can be caused by differences in nutrient conservation from one growing season to the next (Güsewell et al., 2003a). When Carex curta was grown for 6 or 18 months at widely varying N : P supply ratios, shoot biomass peaked at N : P ratios of 15–26 after 6 months, whereas it was almost the same across a broad range of N : P ratios (0.6–26) after 18 months, (two growing seasons; Fig. 4). According to nutrient budgets, plants with low N : P ratios retained nutrients better during the winter, which caused their shoot biomass to be greater in the second season than in the first (S. Güsewell, unpublished data).
4. Nutrient limitation of plant communities
N : P ratios of vegetation (pooled above-ground biomass of all species) have been compared with the results of field fertilization experiments by several authors to identify critical N : P ratios discriminating between N and P limitation at vegetation level (Fig. 5). All authors agreed that low N : P ratios indicate N limitation whereas there is no consistent interpretation of intermediate and high N : P ratios. Some authors suggest that high N : P ratios indicate limitation by P alone, but others report N limitation or colimitation up to high N : P ratios (Downing & McCauley, 1992; Olde Venterink et al., 2003). Clearly, N : P ratios cannot be the only criterion for the assessment of nutrient limitation as biomass production might also be limited by elements other than N or P (van Duren & Pegtel, 2000) or by light or climatic factors (Spink et al., 1998). Wassen et al. (1995) proposed that [N] < 13–14 mg g−1 and [P] < 0.7 mg g−1 are additional conditions for N and P, respectively, to be limiting. Güsewell & Koerselman (2002) found no maximal [N] for N limitation, but [P] < 1.0 mg g−1 appeared to be necessary for limitation by P alone (as in Lockaby & Conner, 1999). Finally, Olde Venterink et al. (2003) set N : K (potassium) < 2.1 and K : P > 3.4 as conditions to exclude K limitation. Overall it appears that biomass production is most likely to be enhanced by N fertilization in vegetation with N : P ratio < 10 and by P fertilization in vegetation with N : P ratio > 20, whereas within this range, the effects of fertilization are not unequivocally related to N : P ratios.
Although fertilization experiments are often regarded as the best way to establish the type of nutrient limitation (Vitousek & Howarth, 1991; van Duren & Pegtel, 2000), they do not always yield unambiguous results regarding nutrient limitation because their outcome may depend on experimental methods such as the time of year, frequency and intensity of fertilization or the variable(s) measured to assess vegetation responses (Nams et al., 1993; Boeye et al., 1999) The outcome of fertilization experiments can also be time-dependent, for various of reasons:
1Fertilization may enhance the biomass production only after several years, particularly if current-year growth depends to a large extent on below-ground reserves or on buds formed during the preceding growing season (Lipson et al., 1996; De Kroon & Bobbink, 1997), or in evergreen woody vegetation where current-year growth represents only a small proportion of the biomass (Aerts, 1989; Shaver & Chapin, 1991).
2The species composition of the vegetation may change after fertilization. First-year effects of fertilization are often determined by the responses of species that dominate the ori-ginal plant community, but in following years, subordinate or even new species with different nutrient requirements may reach dominance in plots fertilised with the ‘nonlimiting’ nutrient and produce more biomass than species in control plots (Stöcklin et al., 1998). For example, in a Dutch fen, where plots received N or P fertilizer five times in the course of three months, biomass production during this period was increased by N, mainly due to the response of the sedge Carex panicea (Van der Hoek et al., 2004). One year later, however, the fast-growing grass Holcus lanatus had become more dominant in the previously P-fertilised plots, in which biomass production was now increased relative to controls (Van der Hoek et al., 2004). Shifts in species composition may cause the effects of fertilization to last much longer than increased nutrient availability in soil. Long-term effects of N fertilization on heath vegetation were caused by a change in communities of bryophytes and fungi (Gordon et al., 2001). Strengbom et al. (2001) detected such long-term effects as much as 47 yr after N fertilization in a boreal forest. In an alpine grassland the effects of fertilization also remained visible 50 yr after the last fertilization (Hegg et al., 1992; O. Hegg, pers. comm.).
3Soil processes may be modified by the addition of one element, resulting in increased availability of the other element. Phosphorus fertilization often promotes both the growth of legumes and nitrogenase activity in their nodules (Niklaus et al., 1998; Gentili & Huss-Danell, 2003). As a result, N availability to non-fixers may rise after P fertilization and stimulate their growth. Thus, long-term P fertilization has effects similar to short-term N fertilization, so that ‘what appears as N limitation (to nonfixers) could actually be P limitation in disguise’ (Vitousek & Howarth, 1991; Vitousek & Field, 1999). Long-term effects of P fertilization have also occurred in vegetation where symbiotic N2 fixation did not constitute an important source of N (Egloff, 1983; Seastedt & Vaccaro, 2001). Nitrogen fertilization can stimulate the mobilisation of P through an increase in phosphatase production and/or activity (Johnson et al., 1999; Treseder & Vitousek, 2001).
The possibility that fertilizer effects on biomass production reflect changes in community structure and functioning or in soil processes, rather than the direct growth responses of plants to nutrient availability, has caused some authors to question the application of the concept of nutrient limitation to plant communities (Chapin et al., 1986b; Körner, 2001). Responses to fertilization remain a convenient operational definition of nutrient limitation as long as the ecological context and time scale are clearly stated (Vitousek & Howarth, 1991). However, these effects do not necessarily reflect the physiological status of plants (‘nutrient deficiency’) or nutrient availability in soil before fertilization (Güsewell et al., 2002).
5. Effects of fertilization on plant species within communities
Relationships between biomass N : P ratios and responses to fertilization are even more complex for individual plants or plant species within established communities. There are many reports of fertilization experiments where species responses to N or P addition did not correspond to the type of nutrient deficiency suggested by biomass N : P ratios (Mamolos & Veresoglou, 2000; Tomassen et al., 2004). Species interactions are important factors in causing such responses. It is well known that species grown in monoculture and in mixture can respond differently to nutrient addition (Austin et al., 1985; Heil & Bruggink, 1987; DiTommaso & Aarssen, 1991; Wetzel & van der Valk, 1998). In mixed vegetation, addition of the growth-limiting nutrient typically stimulates only some of the species (Güsewell et al., 2003b); these may be the dominant species, or the faster-growing ones, or those whose spatial or temporal pattern of nutrient uptake corresponds best to the timing of fertilization (Mamolos & Veresoglou, 2000). Other species might also have been enhanced by the nutrient addition had they grown alone, but increased competition from the most responsive species negates the effect (Güsewell et al., 2003b). The opposite effect is also possible: a P-deficient species may be promoted by N fertilization if this enables it to increase its P uptake at the expense of its neighbours.
Contrasting effects of N and P fertilization on plant species within communities were predicted by the resource-ratio model of competition (Tilman, 1982). According to this model, species with particularly low [N] or [P] can become dominant in the long term under N- or P-limited conditions because of their low requirements for these elements. Furthermore, species with inherently low N : P ratios are predicted to dominate in N-limited vegetation, and species with inherently high N : P ratios, in P-limited vegetation. Phosphorus fertilization should therefore promote species with low N : P ratios, and N fertilization species with high N : P ratios (Tilman, 1997). Thus, the long-term responses predicted by this model are opposite to the short-term responses predicted by the law of the minimum.
The resource-ratio model was originally developed for planktonic communities (Tilman, 1981), and its predictions were supported by various studies comparing the biomass N : P ratios of phytoplankton species, the N : P ratios of waters where they dominate, their responses to water fertilization, and competition experiments at various N : P supply ratios (Rhee & Gotham, 1980; Smith, 1982; Stockner & Shortreed, 1988; Fujimoto et al., 1997). In Mediterranean grasslands, Mamolos et al. (1995) and A.P. Mamolos & D.S. Veresoglou (unpublished) found that N fertilization promoted mainly the species with highest [N] and highest N : P ratio (except for legumes), whereas P fertilization mainly promoted the species with the highest [P] and lowest N : P ratio. For 15 herbaceous wetland species, the N : P ratios of plants grown in pots at low nutrient supply (Güsewell et al., 2003a) correlated positively with the mean and maximal N : P ratios of the same species in the field but not with their minimal N : P ratios (Güsewell & Koerselman, 2002; Fig. 6). Thus species with inherently low N : P ratios occurred only at N-limited wetland sites, whereas species with inherently high N : P ratios could occur at N- or P-limited sites.
IV. Interspecific variation in N : P ratios
1. Patterns of variation
Interspecific variation in biomass nutrient concentrations and N : P ratios has been studied either by growing plants under the same conditions or by comparing the means of plant populations sampled at several sites per species. When many species are compared, N and P concentrations always correlate positively with each other (Fig. 7). N : P ratios and [N] are largely unrelated, while N : P ratios and [P] correlate negatively (Fig. 7). Interspecific variation in N : P ratios is therefore primarily determined by variation in [P]. These relationships are nearly identical for plants grown under standardized conditions in the glasshouse (Fig. 7a) and plants sampled in the field (Fig. 7b). Similar correlations have also been reported by Garten (1976), Bedford et al. (1999), Güsewell & Koerselman (2002) and others. Niinmets & Kull (2003) found no correlation between [N] and [P] among species in a wooded meadow and a bog, but this was probably because concentrations were closely similar in all species.
2. Relationship with growth rate
Across all groups of primary producers (terrestrial and aquatic, including phytoplankton), N : P ratios were found to correlate positively with body size and with the thickness of assimilating tissues, and negatively with maximal relative growth rate, RGRmax (Nielsen et al., 1996; Sterner & Elser, 2002). The same type of relationship was also found for microbes and several animal taxa (Sterner & Elser, 2002; Elser et al., 2003; Woods et al., 2004). As a possible explanation, Elser et al. (2000b) proposed that RGRmax is determined by the rate of protein synthesis in ribosomes; this would cause RGRmax to correlate with the amount of ribosomal RNA which is, in turn, related to the number of ribosomal DNA copies and the length of the intergenic spacer regions. For plants this hypothesis was supported by the finding that fast-growing strains of some crop plants had longer intergenic spacer regions than slower-growing strains (Elser et al., 2000b). Because of the high P content of nucleic acids (8.7%), high [P] appears to be an inevitable consequence of fast growth as long as growth rate is determined by the rate of protein synthesis and as long as nucleic acids contain a substantial fraction of cell P. In leaves of vascular plants, nucleic acids may contain 30% of total P but this fraction decreases considerably when P supply is not limiting (Chapin et al., 1986a).
The relationship found by Nielsen et al. (1996) across all groups of primary producers also holds within these groups: in several independent data sets comparing terrestrial vascular plant species, RGRmax correlated positively with [N] and [P], and negatively with foliar N : P ratios (Table 1a). Foliar N : P ratios also correlated negatively with Ellenberg's indicator values for nutrients, meaning that species from nutrient-rich sites have, on average, lower N : P ratios but higher [N] and [P] than species from nutrient-poor sites (Table 1b). Previous research on determinants of interspecific variation in RGR and associated trade-offs has focused on variation in specific leaf area (SLA), leaf life span and N concentration (Cornelissen et al., 1997; Poorter & van der Werf, 1998; Reich et al., 1998), but the data in Table 1 suggest that [P] is equally important.
Table 1. Correlations between N or P concentrations or N : P ratios in the biomass of plant species and (a) maximxal relative growth rate (Grime et al. (1988)) or (b) nutrient indicator values (Ellenberg et al. (1991))
A further difference in the influence of [N] and [P] on RGRmax becomes apparent if the effect of leaf structure is taken into account. In the data set of Nielsen et al. (1996), both [N] and [P] correlated with leaf thickness, but joint effects on RGR differed: The correlation of RGRmax with [N] was almost entirely accounted for by variation in leaf thickness, whereas the correlation of RGRmax with [P] was partially independent of leaf thickness (Nielsen et al., 1996). The same appeared in a reanalysis of growth parameters in tree seedlings (Cornelissen et al., 1997; Castro-Diéz et al., 2000): after accounting for the correlation between SLA and RGRmax (r = 0.66), the effect of [N] on RGRmax was no longer significant (P = 0.13) but the effect of [P] was still significant (P = 0.038).
Growth forms A compilation of data sets from several field surveys or reviews shows that [N], [P] and N : P ratios do not differ consistently between woody and herbaceous plant species (Table 2). Evergreen woody species generally have lower [N] and [P] than deciduous ones, but N : P ratios do not differ. Graminoids generally have lower [N] and [P] and higher N : P ratios than forbs; their greater ability to maintain growth at low internal P concentration has also been demonstrated in controlled experiments (Caradus, 1980; Halsted & Lynch, 1996; Falkengren-Grerup, 1998). Halsted & Lynch (1996) attributed this difference to the localization of leaf growth in basal meristems, reducing P requirements for nucleic acids in the remaining part of the leaf. The difference in N : P ratios between graminoids and forbs contrasts with the suggestion of Knecht & Göransson (2004) that ‘terrestrial plants require nutrients in similar proportions’– these authors lumped together graminoids and forbs in one group.
Table 2. Differences in N and P concentrations and N : P mass ratios among plant growth forms in various vegetation types of the world
Nutrient acquisition mechanisms Species with particular mechanisms for the acquisition of either N or P tend to dominate on soils where this element is otherwise poorly available. Thus species with symbiotic N2 fixation are generally favoured by high availability of P and low availability of N (Smith, 1992; Vitousek & Field, 1999); species that take up amino acids typically occur under conditions that inhibit N mineralization (Northup et al., 1995; Northup et al., 1998; Lipson & Näsholm, 2001); and species with effective P solubilization dominate on soils with strong P sorption (Ström, 1997; Lamont, 2003). However, these mechanisms are not necessarily related to biomass N : P ratios: N2-fixers often have higher N : P ratios than co-occurring nonfixers (Güsewell et al., 2003b), and species able to solubilize P may have relatively low N : P ratios even on soils with low P availability. Mycorrhizal associations and the secretion of root phosphatases are found in the large majority of plant species and therefore do not constitute specific adaptations to N- or P-poor soils, but their expression and the costs and benefits for plants depends on N and P availability (Wetzel & van der Valk, 1996; Nielsen et al., 1998; Van der Heijden & Kuyper, 2001; Vance et al., 2003).
Nutrient conservation mechanisms The reduction of nutrient losses from plants reduces the need to acquire new nutrients and is an important adaptation to growth at nutrient-poor sites (Aerts, 1999; Aerts & Chapin, 2000). Long life span and low nutrient concentrations in litter appear to be the main adaptations under both N and P-limited conditions (Escudero et al., 1992; Wright & Westoby, 2003). In contrast, efficient resorption of nutrients during leaf senescence appears to be an adaptation more to P-limited conditions than to N-limited conditions (Aerts & Chapin, 2000).
Photosynthetic pathways Because of their greater photosynthetic N-use efficiency, C4 plants can grow optimally with lower [N] than C3 plants. They are therefore advantaged on N-poor, dry soils. Some studies have suggested that C4 plants might also perform better than C3 plants under P limitation by avoiding detrimental effects of P deficiency such as starch accumulation, drought sensitivity and increased photorespiration, which have been observed in P-deficient C3 plants (Herold et al., 1976; Halsted & Lynch, 1996). The advantages provided by either photosynthetic pathway are apparently not specifically related to N or P limitation.
C-S-R strategies When species are classified according to Grime's (2001) primary strategies (C = competitive, S = stress-tolerant, R = ruderal) three main differences are apparent (Table 3): species with S, CS or RS strategies have lower [N] than other species; species with S and (even more) CS strategies have low [P] and high N : P ratios; species with R or CRS strategies have low N : P ratios. The significance of low nutrient concentrations for the stress-tolerant growth strategy has been discussed extensively elsewhere (Grime et al., 1997; Aerts, 1999; Aerts & Chapin, 2000). The high N : P ratios of S and CS species partly reflect slow growth, and partly (in CS species) their efficient internal cycling of P. Ruderal species are often annual species or short-lived perennials with high reproductive allocation, leading to greater P requirements (cf. Thompson et al., 1997).
Table 3. Differences in N and P concentrations and N : P mass ratios among plant strategies in the established phase (Grime (2001))
V. Vegetation properties in relation to N : P ratios
1. Patterns of variation
Vegetation-level variation in [N], [P] and N : P ratios is caused by the combination of differences in species composition and in nutrient availability among plant communities. It is therefore not surprising that relationships between [N], [P] and N : P ratios at vegetation level are intermediate between those found in intraspecific (Fig. 1) and interspecific (Fig. 7) comparisons: [N] and [P] generally correlate positively, but the relationship can be weak. N : P ratios are determined by [P] more than [N], that is, vegetation with low N : P ratio is primarily P-rich (Koerselman & Meuleman, 1996; Lockaby & Conner, 1999). These relationships also hold across ecosystem types at a worldwide scale (Cebrián et al., 1998).
Chemical (mineral soil) and biological (microbial biomass, organic matter) or combined
Effect of fire
N volatilisation, followed by increased N mineralisation and N fixation
Mineralisation and increased availabillity; some losses as dust
Predominant form in soil
Mineral and organic
Predominant form in organic matter
Carbon-bonded (release requires oxidative breakdown of the carbon structure)
Ester-bonded (easily released through hydrolysis by phosphatase)
Temporal patterns in vegetation N : P ratios have been analysed in relation to concerns about the effects of increased atmospheric N deposition on vegetation diversity and functioning. In many areas, comparisons of samples from different times showed an increase in N : P ratios during the 20th century. In some cases this increase was caused mainly by increasing N concentration (Pitcairn & Fowler, 1995), in other cases by decreasing P concentration (Verhoeven & Schmitz, 1991). Some of these changes may also be caused by natural succession (Koerselman & Verhoeven, 1995).
2. Relationships with biomass production
Relationships between primary production and either nutrient concentrations or N : P ratios (for a given ecosystem type and geographical range) are generally weak. Unimodal relationships with maximal biomass production at intermediate (‘optimal’) N : P ratios have been observed only occasionally. For wetland forests Lockaby & Conner (1999) found the greatest amount of litterfall (an estimate of annual net primary production) in forests with litter N : P ratios of 12. More often there is either no relationship, or a decline in biomass production with increasing N : P ratio (spatially or temporally). In herbaceous wetlands, high biomass production (>800 g m−2) was only found in vegetation with a N : P ratio <15 (Olde Venterink et al., 2001a; Olde Venterink et al., 2003). The biomass production of road verges correlated positively with [N] and [P] but negatively (though weakly) with tissue N : P ratios (Schaffers, 2002). In Czech forests, increasing N : P and N:K ratios associated with increasing N deposition between the middle and end of the 20th century went along with a reduction in biomass production (Hofmeister et al., 2002).
3. Relationships with species composition
Based on the differences in N : P ratios among functional groups of plants, the main relationships to be expected between N : P ratios and the species composition of vegetation are (1) more graminoids and fewer forbs in vegetation with a high N : P ratio and (2) more stress-tolerant species and fewer ruderals in vegetation with high N : P ratio.
The expected patterns have been found in vegetation surveys along natural gradients in N vs P availability (Theodose & Roths, 1999); in shifts of species composition caused by increased atmospheric N deposition; and in fertilization experiments. One of the most consistently reported effects of increased atmospheric N deposition in natural or seminatural temperate vegetation is the increased dominance of graminoid species with low P and K requirements, most of which are classified as stress-tolerant competitors (De Kroon & Bobbink, 1997; Bobbink et al., 1998). These species are also promoted by N fertilization in field or laboratory experiments (Bobbink et al., 1988; Carroll et al., 2003; Tomassen et al., 2003; Tomassen et al., 2004) or by other measures (e.g. drainage) that enhance the availability of N relative to P (Beltman et al., 1996). Forbs are promoted mostly by P or N + P fertilization, and bryophytes by P fertilization.
Species that have become more abundant because of N enrichment on P-poor soils are mainly species of rather low-producing vegetation, and not typical ‘nitrophilous’ species. In southern Swedish forests changes in forest vegetation due to high N deposition were unrelated to Ellenberg's nutrient indicator values (Falkengren-Grerup, 1998) and (in this case) to taxonomy, life form, RGR and habitat type (Diekmann & Falkengren-Grerup, 2002).
Relationships between the N : P ratios of vegetation and plant strategies were investigated for wetland vegetation in north-west Europe by Willby et al. (2001). The proportion of stress-tolerant species correlated negatively with the N, P and K concentrations of vegetation biomass, and positively with N : P ratios. Correlations with the proportion of ruderal species were opposite. According to indications provided by N : P : K ratios, half the plots dominated by stress-tolerant species were limited by P or colimited by P and N or K, whereas none of the plots dominated by ruderals and <20% of the plots dominated by competitors appeared P-limited (Willby et al., 2001).
4. Relationships with species richness
Relationships between vegetation N : P ratios and species richness are of particular interest for the assessment of anthropogenic impacts on natural or seminatural vegetation (Verhoeven et al., 1996a; 1996b; Bobbink et al., 1998; Bedford et al., 1999; Olde Venterink et al., 2001b). Of the numerous factors and mechanisms controlling the species diversity of plant communities, several are related to nutrient availability and might therefore cause differences in diversity between N-limited and P-limited vegetation.
Species coexistence facilitated by colimitation If nutrient supply to the vegetation is on the whole balanced so that biomass production is colimited by N and P, some species might be limited by N and others by P. This could be caused by interspecific variation in N and P uptake (Koerselman & Meuleman, 1996), or by differing N and P requirements of the species (Mamolos et al., 1995). According to the ‘resource balance hypothesis’, differential nutrient limitation would reduce interspecific competition and allow more species to coexist (Braakhekke & Hooftmann, 1999). In various types of herbaceous vegetation, maximal species density was indeed found at intermediate N : P ratios (6–20), but the effects of N : P ratios on species density were always weak and partially confounded with those of biomass production and/or soil pH; only communities with a N : P ratio > 25 were markedly less species-rich (Braakhekke & Hooftmann, 1999; Roem & Berendse, 2000; Van der Welle et al., 2003). Morevoer, fertilization with N or P to create a more balanced nutrient supply did not increase species richness in two wet grasslands (Aerts et al., 2003). Thus available data provide only weak support for the resource balance hypothesis.
Species coexistence facilitated by P limitation Interspecific competition might be weaker in P-limited than in N-limited vegetation for several reasons: (1) The numerous chemical forms of P in soil might reduce or even allow facilitation if individual species exploit different P fractions according to their specific P-acquisition mechanisms (Kamh et al., 1999). (2) Competition for a resource with low mobility in soil might be weaker than for a highly mobile resource (Huston & De Angelis, 1994). It could therefore be that more species are able to coexist in P-limited vegetation. In agricultural grasslands there was indeed a clear relationship between species richness and P availability in soil: only grasslands with less than 0.05 mg g−1 extractable P contained more than 20 species per 100 m2 (Janssens et al., 1998). For Dutch wetlands Verhoeven et al. (1996a) proposed that species-rich communities require the inflow of base-rich groundwater to maintain a low P availability, and Olde Venterink et al. (2003) showed that threatened wetland species occur mainly in P-limited vegetation. In Swedish rock vegetation species richness was reduced when P availability was high (Tyler, 1996). In contrast, species richness in mountain pastures was not clearly related to P availability (Müller et al., 2003; Jewell et al., 2002). In short-term fertilization experiments, addition of P often had little effect on species richness; positive and negative effects both occurred (Gough et al., 2000; Pauli et al., 2002; Roem et al., 2002). The role of low P availability for species richness is therefore not unequivocal.
Species exclusion caused by P limitation Nitrogen fertilization reduced the small-scale species richness (species density) in herbaceous vegetation throughout the Northern hemisphere (Gough et al., 2000) and negative effects on the diversity of the forest understory have also been reported regularly (Strengbom et al., 2001), as well as reduced species diversity as a result of increased atmospheric N deposition (Bobbink & Lamers, 2002; Falkengren-Grerup & Diekmann, 2003). The reduction in species diversity was partly associated with an increase in biomass production (Gough et al., 2000) but also occurred when biomass did not change (Hofmeister et al., 2002). Reproduction and survival of several species proved to be impaired by high N deposition (Brouwer et al., 2001; Gotelli & Ellison, 2002; Vergeer et al., 2003), presumably as a result of P deficiency or another nutritional imbalance. However, increased N deposition can also cause direct toxic effects, so that the relative roles of N toxicity vs P limitation for reduced fitness still have to be established.
Species exclusion caused by graminoids Increased dominance of graminoids might often explain the reduction in species richness caused by shifts in N : P ratios. Species richness correlated negatively with the biomass of grasses and positively with that of forbs in several herbaceous communities (Willems & van Nieuwstadt, 1996; Theodose & Roths, 1999). In fertilization experiments, negative effects of N fertilizer on species richness were often associated with an increased dominance of one (or few) species of clonal graminoids, as seen in Brachypodium pinnatum calcareous grasslands (Bobbink, 1991); Carex scopulorum in alpine tundra (Theodose & Bowman, 1997); Agrostis capillaris in a flood meadow (Joyce, 2001); or Molinia caerulea in a bog (Tomassen et al., 2004). In contrast, species richness was unaffected by N fertilization in an old-field community entirely dominated by forbs (Huberty et al., 1998) and reduced by PK fertilization when this increased the dominance of the grass Festuca rubra in a Dutch calcareous grassland (Willems & van Nieuwstadt, 1996). According to these results shifts in the dominance of clonal graminoids, rather than shifts in the type of nutrient limitation, determine changes in species diversity.
Diversity reduced by shift from historical to new type of nutrient limitation Nitrogen limitation has predominated historically in the majority of ecosystems of the Northern hemisphere, given their young and mostly P-rich soils (Vitousek & Howarth, 1991). It therefore seems that N-limited sites have been more abundant and ecologically more diverse than P-limited sites, offering more opportunities for speciation to relatively fast-growing, P-demanding species. A greater regional species pool could therefore evolve for N-limited than for P-limited sites (Willems & van Nieuwstadt, 1996). For European wetlands, published data on N : P ratios in plant species at multiple sites indicate that few species occur naturally with high N : P ratios (Güsewell & Koerselman, 2002; Güsewell et al., 2005).
World regions where P limitation has prevailed for a long period have more species adapted to P-limited conditions. The great species diversity of the Proteaceae (>1600 species), particularly in Mediterranean regions of Australia and South Africa (>1000 species), is thought to be the result of in situ speciation on generally P-deficient soils (Cowling & Lamont, 1998). Almost all Proteaceae possess cluster roots improving their P uptake (Lamont, 2003); many species have efficient P conservation mechanisms (Specht & Groves, 1966), and many are presently threatened by P enrichment (Handreck, 1997). In these regions, P availability in soil was negatively related to the number of native species (Handreck, 1997; Morgan, 1998), and positively with the spread of alien species (e.g. the grass H. lanatus).
In conclusion, relationships between N : P ratios and species richness observed in individual studies can be caused by differences in species availability, recruitment and competition. The relative importance of these factors in controlling species diversity has been much debated in recent years (Grace, 1999; Herben, 2000; Zobel et al., 2000; Grace, 2001; Zobel, 2001), showing that each of them can be important, and that they interact (Foster, 2001; Lord & Lee, 2001). Any shift in N : P ratio may therefore affect species richness, but no general prediction appears to be possible regarding the direction of the effect.
VI. Implications of N : P ratios for human impacts on ecosystems
The evidence reviewed here indicates that changes in N and P availability to terrestrial vegetation cause shifts in plant N : P ratios because of plastic adjustments of the plant's nutrient concentrations (Section II) and often also caused by shifts in plant species composition (Sections IV–V). Changes in plant N : P ratio affect the functioning of terrestrial vegetation directly (through the physiological responses of plants) but also indirectly, through responses of herbivores and decomposers, which can feed back on nutrient availability to plants (Elser & Urabe, 1999; Güsewell et al., 2004), and through changes in plant–microbe interactions (Smith, 2002). All these effects are expected to interact with other effects of global environmental change and with effects of ecosystem management.
Shifts in N : P ratios can help explain some of the changes in ecosystem functioning caused by increased N deposition (Bobbink et al., 1998; Lee & Caporn, 1998; Aerts & Bobbink, 1999; Bobbink & Lamers, 2002): litter accumulation, sensitivity to drought, fungal infection, impaired reproduction and dominance of graminoids might be caused by increased N : P ratios rather than to increased availability of N per se. It has often been suggested that a low P availability in soil is important to limit the impact of increased atmospheric N deposition, for example to avoid the spread of faster-growing plant species (Verhoeven et al., 1996a; Lammerts & Grootjans, 1997; Limpens et al., 2003a; 2003b). However, detrimental effects of high N deposition that are related to high N : P ratios would rather be strengthened by low P availability (Brouwer et al., 2001). Because studies reporting detrimental effects of high N deposition on terrestrial vegetation have often not included measurements of P availability or P concentrations in plants, the role of P limitation in these effects is not yet sufficiently known (Tomassen et al., 2003; 2004). This information would however, be important for management decisions (Gordon et al., 2001).
If effects of increased N deposition are partly caused by P limitation, they might be strengthened by elevated CO2 concentrations. Elevated CO2 generally increases the rate of C assimilation and thereby reduces the concentration of the growth-limiting element in plant biomass (Poorter & Pérez-Soba, 2001; Poorter & Navas, 2003). On average the biomass N concentration is reduced by 15% at elevated CO2 with individual changes ranging from –60% to +50% (Cotrufo et al., 1998). The effects of elevated CO2 on plant P concentrations have been much less investigated and have proved inconsistent (Niklaus et al., 1998; Gifford et al., 2000), presumably because many CO2-enrichment experiments were carried out with high P supply. For a calcareous grassland on P-poor soil, Stöcklin et al. (1998) and Stöcklin & Kroner (1999) found strong interactions between the effects of P supply and those of elevated CO2. In strawberry, elevated CO2 promoted starch accumulation in leaves of P-deficient plants and thus strengthened effects of P deficiency such as accelerated senescence (Chen & Lenz, 1997).
Climate warming might increase the availability of N relative to P by accelerating decomposition and N mineralization (Vitousek & Howarth, 1991). A meta-analysis of 32 warming experiments revealed an average increase in N mineralization by 46% (Rustad et al., 2001), whereas little is known about the effects of warming on soil P availability. Climate warming may modify the nutrient status of plants indirectly, for example by interfering with leaf senescence and thus, reducing nutrient resorption (Robinson et al., 1998; Norby et al., 2000; Aldous, 2002). Effects are likely to be similar for N and for P, given the close correlation between N and P resorption efficiency (Aerts, 1996). However, the implications may be larger for a plant's P budget than for its N budget because of the normally greater P resorption efficiency (Aerts & Chapin, 2000). The existence of an overall negative relationship between temperature and nutrient concentrations in organisms might lead to unexpected stoichiometric effects of climate warming (Woods et al., 2003).
Research on the effects of global environmental change has been carried out mainly under the paradigm of N limitation (Norby et al., 2001; Rustad et al., 2001). At the global scale, strong correlations between ecological process rates and both N and P concentrations in plant material appear to justify this approach (Cebrián et al., 1998; Wright et al., 2004). At the local scale, however, there is large variation in N : P ratios and this variation may influence the effects on terrestrial vegetation of atmospheric N deposition, elevated CO2, climate warming and management. Our knowledge of these interactions is still limited (Gifford et al., 2000). Predictions of changes and mitigation measures might be improved by the inclusion of N : P ratios in community and ecosystem models (Elser & Urabe, 1999).
The balance between N and P supply, as reflected in N : P ratios of plant biomass, influences the functioning of terrestrial vegetation at all levels: the growth and reproduction of individual plants, plant species interactions, community composition and botanical diversity all depend in a distinct way on the availability of N and P to the vegetation, and the resulting differences in plant chemistry may determine the activity of herbivores, parasites, pathogens and decomposers.
N : P ratios in terrestrial vegetation have so far been used mainly to assess whether N or P is more limiting for biomass production; research efforts have focused on the reliability of this indication have and what the critical N : P ratio should be (Wassen et al., 1995; Koerselman & Meuleman, 1996; van Duren & Pegtel, 2000; Güsewell et al., 2003b; Tessier & Raynal, 2003). The lack of a definite answer reflects the ambiguity of the concept of nutrient limitation: ‘limitation’ is defined by differences between process rates and therefore depends on the process, the type of comparison and the time scale over which it is assessed. Plant N : P ratios are a useful variable to consider in ecological research precisely because they reflect the gradual and dynamic character of nutrient limitation better than fixed categories (N-limited vs P-limited) and their unique assessment through fertilization.
The main applications of N : P ratios in future ecological research will probably be the investigation of changes in ecosystem functioning caused by anthropogenic changes in the balance between N and P availability, both directly (e.g. N deposition) and indirectly (e.g. climate warming). Shifts in N : P ratios have proved to be related to changes in plant performance and in the species composition of the vegetation. As shown by similar studies in aquatic ecology (Sterner & Elser, 2002), N : P ratios may be particularly useful to investigate the effects of these changes across trophic levels, and the implications for element cycles.
Comments from P.J. Edwards, E. Frossard, H. Olde Venterink, Ch. Küffer, D. Veresoglou, R. Häsler, W. Landolt, S. Bassin, and three referees helped to improve the manuscript. I would like to thank W. Koerselman and J.T.A. Verhoeven for several years of stimulating research collaboration on N : P ratios (funded by TMR grant ENV4-CT97-5075 from the EU) and A. van der Werf for suggesting that write this review.