• Shifts from nitrogen (N)- to phosphorus (P)-limited growth due to high N deposition may alter the functioning of wetland vegetation. This experiment tested how N vs P deficiency affects the growth and nutrient use of wetland sedges.
• Five wetland Carex species were grown at nine N : P supply ratios (0.6–405) with two absolute levels of N and P. Biomass and nutrient concentrations were determined after one and two growing seasons.
• Shoot biomass was maximal at N : P supply ratios of 15–26 after one season but 5–15 after two seasons. Photosynthesis after the first season, second-year growth, leaf longevity, and the fraction of nutrient supply retained by plants over two seasons were all negatively related to N : P supply ratios, with small effects of absolute supply. The five Carex species responded similarly to N : P ratios but differed in nutrient resorption efficiency and biomass allocation.
• Plants treated with high N : P ratios appeared to lose nutrients below ground. Such losses may reduce plant performance in P-limited wetlands affected by high N deposition.
Physiological responses of wetland plants to excess nitrogen have been investigated extensively, but only a few studies have considered the specific effects of P deficiency as opposed to N deficiency. In short-term experiments P deficiency reduces the growth of young plants less than N deficiency (Keddy et al., 2001), probably because these plants can adjust to P deficiency through a reduction in internal P concentrations (Pérez-Corona & Verhoeven, 1996; Güsewell & Bollens, 2003; Güsewell, 2004). Conversely, P deficiency seems to have a more negative impact than N deficiency on long-term plant performance, for example through increased mortality (Gotelli & Ellison, 2002) and impaired reproduction (Brouwer et al., 2001). Field studies also suggested that P deficiency induced by N enrichment increases root turnover and decreases nutrient retention by the vegetation (Güsewell et al., 2002; El-Kahloun et al., 2003). Likewise in a growth experiment with 15 wetland species, plants receiving high N supply retained a smaller fraction of the supplied nutrients in their biomass over 2 yr than those receiving high P supply (Güsewell et al., 2003a). Available data therefore suggest that nutrient retention by wetland plants depends on the type of nutrient limitation; however, rigorous testing of this hypothesis requires growth experiments that vary N : P supply ratios (N vs P limitation) independently from variation in the total supply of N and P (Güsewell & Bollens, 2003; Güsewell et al., 2003a).
The observation that different species dominate N- and P-limited wetlands (Verhoeven & Schmitz, 1991), and that some species can expand in P-poor wetlands under high N deposition while many cannot (Tomassen et al., 2004), suggests that plant species respond differently to N : P supply ratios (Tomassen et al., 2003). Short-term experiments failed to demonstrate relationships between physiological responses of species to N vs P deficiency and their occurrence at N- or P-limited field sites (Pérez-Corona & Verhoeven, 1996; Güsewell, 2004), but the long-term responses to N : P supply ratios, which might be more decisive for species distributions in the field, have not yet been investigated. Knowledge about species-specific responses to N and P deficiency would help in evaluating whether under certain circumstances the negative impacts of high N deposition on species diversity in wetlands might be mitigated by P fertilisation (Geerts & Oomes, 2000; Roem et al., 2002; Aerts et al., 2003a).
In a 2-yr experiment I investigated how five Carex species from mesotrophic, N- or P-limited wetland communities respond to a broad range of N : P supply ratios to test three main hypotheses, derived from the earlier work described above: first that plants initially grow faster with high N : P supply ratios than with low N : P supply ratios, but that this reverses with time; second that plants retain nutrients less effectively when grown with high N : P supply ratios; and third that species that dominate at P-limited sites tolerate high N : P supply ratios better than species that dominate at N-limited sites.
The experiment involved five Carex species that are common in mesotrophic fens and wet grasslands of Central Europe: C. curta L., C. elata Curtis, C. flacca Schreber, C. flava s.str. L and C. panicea L. (nomenclature: Halliday & Beadle, 1983). Presently most of these sites are nature reserves and are mown regularly in summer or autumn for conservation; combined with high N deposition, this management has driven some sites to P or K limitation, while others are still N-limited (Verhoeven et al., 1996b; Olde Venterink et al., 2001, 2002). All Carex species can be found at N- or P-limited sites (e.g. Güsewell et al., 2003b). However, two of them (C. curta and C. elata) are most abundant in N-limited vegetation (Verhoeven & Schmitz, 1991; Bollens et al., 2001), whereas the three others (C. flacca, C. flava s.l. and C. panicea) are particularly common at sites with P-limited biomass production (Verhoeven & Schmitz, 1991; Boeye et al., 1999); at those sites, the three Carex species are sometimes N-limited while most other species are P-limited, suggesting a competitive advantage of these Carex species at high N : P ratios (e.g. Pauli et al., 2002; Van der Hoek et al., 2004).
Mother plants of the five species were collected in the field (C. curta, C. panicea and C. flacca) or cultivated from seeds (C. elata, C. flava) in spring 1998. They were grown outdoors for 1 yr in pots with garden mould to minimise any effects due to their sites of origin (one site per species). In April 1999, clones were split into individual shoots (each with two or three leaves) and kept on tap water until new roots had formed (5 wk). Each shoot was then planted in a pot with 400 g of quartz sand (Carlo Bernasconi, Zürich, CH, 0.1–0.7 mm) and trimmed to a length of 4 cm to reduce differences in initial size (200–300 mg dry mass per plant). These shoot stubbles died and decayed during the first few months of the experiment, and only newly formed shoots were included in later measurements. Carex flacca was planted 3 wk later than the other species to replace Carex disticha (a species of P-rich disturbed sites), which was included initially but did not survive the transplantation.
The experiment lasted from May 1999 until September 2000 in an open unheated glasshouse with c. 50% daylight and 5–30°C. Pots were placed on flat trays, which were regularly shifted on the glasshouse bench to compensate for heterogeneity in light and temperature. Between November 1999 and March 2000, the pots were exposed to winter conditions outdoors (−10 to +10°C). At the end of the experiment, only a few plants of C. elata and C. curta grown at high nutrient supply and intermediate N : P ratios had largely filled the pots with their roots; growth restriction by pot size was therefore probably unimportant for plant responses to treatments.
Eighteen nutrient treatments were applied, combining nine N : P supply ratios with two supply levels in a factorial design with three replicates. Treatments were defined by the total amounts of N and P (in mg) supplied per plant during the experiment, which were calculated as:
with L, the overall supply level (geometric mean of N and P supply), being either 28.7 mg for high supply or 9.6 mg for low supply, and N : P, the ratio of N to P supplied, ranging from 0.58 (strong N limitation) to 405 (strong P limitation). At a N : P supply ratio of 1.7, the treatments were comparable to flushing pots (100 ml water holding capacity) once per week with 1/4 strength (high supply) or 1/12 strength (low supply) Rorison nutrient solution; overall nutrient supply was therefore low compared with many other growth experiments (Garnier & Freijsen, 1994; Castro-Diéz et al., 2000). Details on treatments are given in Appendix 1, available online as supplementary material.
Nutrients were supplied as KNO3 and KH2PO4 twice weekly for 16 wk in the first year and once weekly for 16 wk in the second year. All other essential elements were supplied in constant, nonlimiting amounts (Appendix 1). For potassium, part of this amount was provided by the KNO3 and KH2PO4, and the rest by adding KCl, so that both K supply and the total ionic strength of the nutrient solutions were similar across treatments. During the first year, the weekly supply of nutrients increased three-fold in the course of the growth period to account for the increasing size of the plants. In the second year nutrient supply was constant over the season as plants were already established. Nutrient solutions were prepared by mixing together appropriate volumes of stock solutions and water; after watering the pots, 2 ml of nutrient solution was applied to the sand surface with a pipette.
To avoid an uncontrolled leaching of nutrients, pots were watered every 1–3 d by giving deionised water into the trays. The toxic accumulation of excess nutrients was avoided by leaching the sand every 3 wk, at least 4 d after the last fertilisation: pots were filled three times up to the rim with water and left to drain in-between.
To follow the growth of plants, morphological measurements were carried out 2 months after the start of each growth period (25 July 1999 and 20 May 2000). The total number of partially or fully expanded leaves was counted and the length of all shoots taller than 1 cm was measured for each. These measurements were converted into estimates of total leaf biomass using regression models established in an earlier experiment (Appendix 2, available online as supplementary material). The regression fit (r2) was 0.84 in C. elata and 0.92–0.95 in the other species.
The rate of photosynthetic assimilation was determined on 25–26 July 1999, after 3 months of growth, using a portable gas exchange chamber (PP-systems CIRAS-1, Hitchin, UK) with an internal CO2 concentration of 350 ppm and a flow of 150 µmol s−1. The measurements were carried out in full sunshine on the youngest fully expanded leaf of the longest shoot per plant. Photosynthesis could not be measured on C. flacca as many plants were too small to introduce their leaves in the gas exchange chamber.
Leaf senescence was monitored at the end of the second growth period as follows: on 29–30 August 2000, the oldest still entirely green leaf was selected on two shoots per pot. The length of the leaf blades was measured to the nearest mm, and maximal width to the nearest 0.1 mm; the product of both was an estimate of leaf area. One of the two leaves per pot was tagged with a small piece of plastic foil; the other leaf was cut off at the ligula, dried at 70°C and weighed. When tagged leaves had fully senesced, they were cut off, dried and weighed. Only a fraction of the tagged leaves senesced before the harvest, providing a crude measure of leaf turnover. Leaf mass loss during senescence was determined by comparing the mass-per-area of green and senesced leaves (Güsewell, 2005).
At the end of the first growth period (28 October 1999) all shoots (entirely consisting of leaves) were clipped c. 0.5 cm above the sand surface, sorted into living and senesced parts, dried at 70°C until constant mass and weighed. This first harvest simulated the mowing regime currently practised at typical sites of the study species. In the second growth period, inflorescences of flowering plants were cut away with their culms at the end of June, when seeds had ripened, to determine their dry mass before seed dispersal. Senesced leaves were collected monthly from the end of June to the end of September and pooled to determine litter production. At the end of the second growth period (21–25 September 2000, on account of the earlier start of growth compared with 1999), all shoots (i.e. leaves) were clipped at sand surface. Subsequently, the sand was washed from the below-ground parts, and roots were separated from rhizomes or the tussock basis. All fractions were dried at 70°C until constant mass and weighed.
All biomass fractions were analysed for total N and P concentrations. Nutrients in litter and inflorescences were determined after pooling the replicates to have enough material. After digestion with a modified Kjeldahl procedure (1 h at 200°C and 2 h at 340°C in a mixture of concentrated sulphuric acid, salicylic acid, copper and selenium), the N and P concentrations of the digests were analysed colorimetrically on a continuous-flow analyser (Skalar SA-40, Skalar, Breda, NL). Biomass and nutrient data per species and treatment are given in Appendix 5, available online as supplementary material.
Shoot production during each growth period was calculated as the sum of shoot biomass at harvest, leaf litter (corrected for mass loss during senescence) and inflorescences (if present). To analyse the seasonal distribution of growth, shoot biomass after 2 months (July 99 or May 00) was expressed as a percentage of each year's total shoot production. The rapidity of shoot senescence was measured by the percentage of shoot production that senesced before the harvest (litter mass as percentage of total shoot production). Note that this percentage may depend both on the seasonal distribution of growth and on leaf life span (Schippers & Olff, 2000). Shoot regeneration after the winter was calculated as the ratio between biomass in May 2000 and biomass in October 1999. Finally, to compare shoot growth in both seasons, ‘second-year growth’ was calculated as the ratio of shoot biomass in September 2000 to that in October 1999.
Nutrient resorption efficiency was derived from nutrient concentrations in living and senesced leaves, adjusted for leaf mass loss during senescence (Güsewell, 2005). As leaf mass loss differed significantly among species but not among nutrient treatments (tested with three-way anova), species means were used: 0% mass loss in C. elata, 4.7% in C. curta and C. flava, 14.4% in C. flacca and 16.8% in C. panicea.
Nutrient accumulation in plant biomass (mg N or P per plant) was calculated for each species and treatment from the mean dry mass of each biomass fraction (leaves of 1999 and 2000, leaf litter, roots, rhizomes or tussock basis, inflorescences) multiplied by the mean N and P concentrations of that fraction. This corresponds to the amounts of nutrients taken up by plants minus those lost through below-ground plant parts, since all above-ground parts, including litter, were taken into account. Calculations were based on means because nutrient concentrations were not available for all replicates (see Measurements). Dividing N or P accumulation by the total amounts of N or P supplied to plants during both growth periods yielded percentage nutrient recovery.
Biomass data, biomass ratios, nutrient concentrations and nutrient accumulation were log-transformed for statistical analysis, and the number of inflorescences was square-root transformed to obtain a normal distribution and homogenous error variance. Photosynthetic rates and percentages could be analysed without transformation. The effects of N : P supply ratios, supply level, species, and their interactions (all fixed) were analysed with three-way anova. Only the two–way interactions were included in the model because models with three–way interaction could not be fitted in all cases due to missing data or due to the pooling of replicates for some chemical analyses. Whenever the three–way interaction could be fitted, it was not significant, and models with or without this interaction yielded closely similar results. The model without three–way interaction was therefore used throughout for consistency.
Plant growth and senescence
Nutrient treatments significantly influenced nearly all traits describing plant growth (Table 1). The total supply of N and P (supply level) strongly influenced plant biomass, whereas the N : P supply ratio had a greater influence than total supply on the seasonal distribution of growth, the rapidity of shoot senescence, shoot regeneration after the winter, and second-year growth (compare F-ratios in Table 1). All growth variables also differed among the five Carex species (Table 1). Interactions between the effects of N : P ratio, supply level and species were partly significant but generally weaker than the main effects (much smaller F-ratios in Table 1). For most variables (exceptions are mentioned hereafter), N : P supply ratios therefore had similar effects across supply levels and species.
Table 1. anova results (F-ratios and significance levels) for the effects of the nitrogen (N): phosphorus (P) supply ratio, overall supply level and Carex species on plant growth, reproduction and senescence
Shoot biomass (mean of the five species) was maximal at a N : P supply ratio of 15 in both years (Fig. 1a,b), but responses to N or P deficiency (i.e. low or high N : P supply ratios) differed: in 1999, N deficiency reduced shoot growth equally or more than P deficiency (Fig. 1a), whereas in 2000, the effect of P deficiency was stronger (Fig. 1b). Photosynthesis in July 1999 and the number of inflorescences in June 2000 were maximal at N : P supply ratios of 15–26 and were reduced more by P deficiency than by N deficiency (Fig. 1c,d). For inflorescences, this asymmetry of responses was pronounced in C. curta and C. flava, whereas C. flacca and C. panicea hardly flowered at both low and high N : P ratios.
The timing of growth and senescence was influenced by N : P supply ratios (Fig. 1e–h). Plants treated with high N : P ratios produced a greater fraction of their shoot biomass in the first part of the 1999 season than those with low N : P ratios (Fig. 1e; linear contrast significant with P = 0.0014). This difference did not exist in 2000 (Fig. 1f). In both years, the fraction of shoot biomass that senesced before the harvest increased towards high N : P supply ratios (Fig. 1g,h). The earlier senescence of plants treated with high N : P ratios may have reflected the earlier growth in 1999 (compare Fig. 1e and 1g) but not in 2000 (Fig. 1f,h), which indicates a reduced life span in 2000. This was confirmed by the senescence of individual leaves: when plants were harvested in 2000, only 21.5% of the tagged leaves had senesced at N : P ≤ 5, but 46.6% at N : P 15–45, and 63.0% at N : P ≥ 78. Shoot regeneration after the winter (May 2000) and second-year growth both decreased towards high N : P supply ratios (Fig. 1i,j). This trend was more pronounced at low than at high overall supply (Fig. 1j, cf. significant N : P-by-supply level interaction in Table 1).
The five Carex species differed in maximal shoot biomass and produced this maximum at slightly different N : P ratios (Table 1); depending on species, shoot biomass was maximal at N : P 15 or 26 in 1999 (Fig. 2a), and at N : P 5 or 15 in 2000 (Fig. 2b). Nevertheless, the effects of N : P supply ratios on second-year growth (Fig. 2c) and on the fraction of senesced shoot biomass (Fig. 2d) were similar in all species. C. curta seemed to have a greater second-year growth at low N : P ratios than the other species (Fig. 2c), but the species-by-N : P ratio interaction was not significant (Table 1).
The fraction of total plant biomass harvested as living shoots in 1999 and 2000 and as tussock basis or rhizomes in 2000 varied little among N : P supply ratios (Fig. 3). By contrast, the fraction harvested as roots decreased (from 49.9% to 33.1%) and the fraction harvested as litter increased (from 6.4% to 24.8%) with increasing N : P supply ratio (Fig. 3). Allocation patterns differed little between the two supply levels: even the significant differences for root and shoot fractions were ≤ 3%, and there was no N : P ratio-by-supply level interaction (Table 2).
Table 2. Biomass allocation: species means (± SE) as well as anova results (F-ratios and significance levels) for the percentage of total plant mass harvested as roots, tussock basis or rhizomes, shoots (living leaves), litter (senesced leaves), and inflorescences
The five species differed in biomass allocation (Table 2). The root fraction was lower in the two rhizomatous species C. flacca and C. panicea than in the three others. The fraction of biomass harvested as litter (leaves that senesced during the growing season) was particularly large in C. curta and C. panicea. Species-level growth data indicate that this was due to short leaf life span in C. curta, whereas C. panicea grew earlier in the season than the other species and therefore also senesced earlier (details not shown). Inflorescences represented a significant fraction of biomass only in C. curta and C. flava (Table 2). In all species, increasing N : P ratios were associated with a decreasing root fraction and increasing litter fraction; only the effect size varied. The effect of N : P ratios was most pronounced in C. curta and C. panicea and least pronounced in C. elata and C. flacca (interactions in Table 2).
Nutrient concentrations, resorption, accumulation and recovery
The N and P concentrations of all plant fractions differed significantly among nutrient treatments and Carex species (Appendix 3, available online as supplementary material). N : P supply ratios influenced P concentrations more than N concentrations: the mean N concentration of biomass and litter increased approximately two-fold across the range of N : P supply ratios, whereas the P concentration decreased 10- to 6-fold (Appendix 4), available online as supplementary material. Nutrient concentrations varied even more for individual species-treatment combinations (Appendix 5): for example, the N concentration of leaves in 1999 ranged from 5.6 to 21.3 mg g−1 (mean = 10.9 mg g−1) and the P concentration from 0.21 to 5.38 mg g−1 (mean = 0.64 mg g−1). Of the five Carex species, C. elata generally had the lowest nutrient concentrations and C. flava was most plastic in response to N : P supply ratios (see species effects and interactions in Appendix 3 and species-level data in Appendix 5).
Nutrient resorption efficiency differed among nutrient treatments and species (Table 3). Overall, N was resorbed more efficiently at low nutrient supply than at high supply (Fig. 4a), whereas P resorption efficiency differed only slightly between the supply levels (Fig. 4b). C. curta and C. elata resorbed N and P less efficiently that C. flacca, C. flava and C. panicea (Fig. 4c,d). The five species also responded differently to treatments (interactions in Table 3): the resorption efficiency of C. curta and C. elata decreased towards high N : P supply ratios but this was not the case with the three other species (Fig. 4c,d). P resorption efficiency was particularly high (around 90%) in C. flacca and C. panicea at N : P supply ≥ 5 (Fig. 4d).
Table 3. anova results (F-ratios and significance levels) for the effects of the nitrogen (N): phosphorus (P) supply ratio, overall supply level and Carex species on nutrient resorption efficiency, total nutrient accumulation in plant biomass (mg N or P per plant) and nutrient recovery (% of supply)
Significance levels: ***, P < 0.001; **, P < 0.01; *, P < 0.05; no symbol, P > 0.05.
The total amounts of N and P harvested with plant biomass over both seasons, both in absolute terms (‘accumulation’) and relative to the amounts supplied (‘recovery’) differed significantly among nutrient treatments (Table 3). Differences among the five Carex species, though significant, were small in size and did not interact with treatment effects (Table 3). The greatest amount of N was accumulated by plants grown at N : P 15; plants grown at N : P 135 contained less N than those grown at N : P 1.7 despite a nine-fold greater N supply (Fig. 5a). P accumulation decreased more than 10-fold with increasing N : P supply ratio (Fig. 5b). Roughly half of the N and P was accumulated in below-ground parts; above-ground N and P was mostly contained in leaves, as the fraction of N and P contained in litter and inflorescences was very small (Fig. 5a–d).
N recovery decreased strongly with increasing N : P supply ratio and slightly with increasing supply level (Fig. 5c). P recovery was below 50% in all treatments and decreased slightly with increasing N : P ratio (Fig. 5d). Thus, with increasing N : P supply ratio, plants did not only receive a smaller supply of P, but also retained a smaller fraction of it in their biomass.
Time-dependent effects of N : P ratios on growth
The results of this study confirm that N : P supply ratios affect the growth of wetland sedges differently in the short term and in the long term: N supply mainly limited growth during the first weeks of the experiment, but P limitation became more important subsequently. This corresponds to the patterns found in earlier work (Güsewell et al., 2002, 2003a), but the present experiment additionally demonstrated that these time-dependent effects are directly related to N : P supply ratios and are largely independent of total nutrient supply, as long as growth remains nutrient-limited.
In the first year (1999), the shoot biomass of the five Carex species was maximal at N : P supply ratios of 15–26. This concurs with results from short-term growth experiments, where plant growth was limited by N up to relatively high N : P supply ratios, especially when total nutrient supply was low (Güsewell & Bollens, 2003; Güsewell, 2004). The initial P content of the young plants and internal re-allocation of P probably supported shoot growth during some time when P supply was low (Usuda, 1995; Köhler et al., 2001; Shen et al., 2003). After 2 months, however, P deficiency started to affect growth: the rate of photosynthesis in July 1999 was strongly reduced at N : P supply ratios ≥ 15, and shoot biomass increased relatively less between July and October in plants treated with high N : P ratios than in those with low N : P ratios (cf. Figure 1e). The P concentrations in leaves of plants grown at N : P 15 were much lower after 6 months here (0.3–0.6 mg g−1) than after 3 months in an earlier experiment (0.7–1.4 mg g−1; Güsewell, 2004), indicating that plants grown at N : P 15 became P-deficient in the course of the growing season (Marschner et al., 1996). Some plants treated with N : P ratios > 15 grew poorly and appeared stunted, suggesting fungal infection. Given the randomisation and the regular shifting of pots in the glasshouse, fungal infection of particular plants was unlikely due to local contamination. P-limited plants are more prone to mycorrhizal infection (Jayachandran et al., 1992; Titus & Leps, 2000; Treseder & Vitousek, 2001); they may also have been more vulnerable to infection by fungal disease in this experiment (Smith, 2002).
In the second year, shoot biomass was greater than in the first year at low N : P supply ratios, but smaller than in the first year at high N : P supply ratios. The negative effect of high N : P ratios was already present at the beginning of the second season, with reduced regeneration after the winter and fewer inflorescences, showing that nutrient treatments during the first season contributed to this effect. As this was more pronounced at the low than at the high supply level, P deficiency rather than N toxicity must have been responsible. N concentrations in plant biomass were generally below 20 mg g−1, that is similar to concentrations of Carex plants in the field (Güsewell & Koerselman, 2002), whereas P concentrations were lower than typically found in the field (around 1 mg g−1; Güsewell & Koerselman, 2002) at all N : P supply ratios above 5, supporting the assumption of P deficiency. Besides low internal P concentrations, reduced biomass allocation to roots (Andrews et al., 1999; De Groot et al., 2003), increasing energy costs of P acquisition (Schachtman et al., 1998) or small plant size (El-Kahloun et al., 2000) may also have contributed to reducing second-year growth at the highest N : P ratios.
The harvest of shoots in October 1999 before full senescence doubtless influenced growth in the second season, as it removed photosynthetic area and nutrients. Would high N : P ratios have reduced second-year growth if shoots had been allowed to senesce naturally? Indeed, the same negative effect of high N : P ratios on second-year growth was found in two subsequent experiments in which the plants were not harvested at the end of the first season (S. Güsewell & H. Olde Venterink, unpublished data). In still another experiment with four Carex species, clipping shoots at the end of the summer reduced the biomass remaining after the winter by 30% relative to nonclipped plants, but this effect was independent of N : P supply ratios (S. Güsewell, unpublished data). It therefore seems unlikely that the effects of N : P ratios on second-year growth were caused by the 1999 harvest.
Nutrient losses at high N : P supply ratios
The time-dependent growth responses to N : P supply ratios indicate a trade-off between faster initial growth (at high N : P ratios) and greater long-term performance (at low N : P ratios). Similar trade-offs have been found in interspecific comparisons (Aerts & van der Peijl, 1993; Ryser, 1996) and in response to variation in N supply (Fransen & de Kroon, 2001); they have been attributed to a faster biomass turnover and greater nutrient losses in fast-growing plants. In this experiment, accelerated leaf senescence at high N : P ratios was indicated by an increasing fraction of senesced biomass at harvest (especially in 2000) and by the senescence of individual tagged leaves. When P is re-mobilised from older leaves in P-deficient plants, a concomitant decrease in N content and the accumulation of starch or sugars can accelerate leaf senescence (Christensen et al., 1981; Sritharan et al., 1992; Chen & Lenz, 1997).
Consistent with the suggested trade-off, plants treated with high N : P ratios retained only a small fraction of the nutrients supplied to them, which confirmed the second hypothesis of this study. The poor P retention during this 2-yr experiment (Fig. 5d) contrasts with results from 3-month experiments (Güsewell & Bollens, 2003; Güsewell, 2004), where plants with N : P supply ratios ≥ 45 contained more than 100% of the P supplied to them. In those experiments, P-limited plants took up all available P (confirming direct uptake measurements in Güsewell, 2004) besides retaining their initial P content over the short growth period.
Nutrient accumulation, as calculated for the present experiment, included all above-ground fractions (also leaf litter and seeds). Causes for the poor P retention over two growth periods must therefore be searched below ground. Possible causes are low P uptake, root turnover, root leaching or exudation (cf. Aerts et al., 1992a). As P-deficient plants took up all P in short-term experiments (Güsewell, 2004), it is likely that a reduction in P uptake at high N : P supply ratios – if at all – occurred only in the second year of this experiment, as a consequence rather than a cause of P deficiency. Indeed, the root mass fraction decreased strongly towards high N: P ratios (Fig. 3), far more than it did after 3 months (Güsewell, 2004). This suggests that either root production was inhibited in the second year, or root turnover was accelerated. Increased root turnover has been reported to result from high N supply (Pregitzer et al., 1995; Fransen & de Kroon, 2001) and from severe P deficiency (Fisher et al., 2002; El-Kahloun et al., 2003). Root turnover causes important nutrient losses (Aerts et al., 1992a; Vásquez de Aldana et al., 1996) as there is little or no resorption during root senescence (Gordon & Jackson, 2000); it might therefore have contributed to the poor P retention in this experiment. Finally, root exudations are often stimulated by P deficiency, and their role for P uptake is well established (Dakora & Phillips, 2002), whereas possible implications for P conservation seem to be unexplored.
The fate of P lost from plant roots was not investigated here; probably part of it was leached from the pots during the winter, when the interruption of shoot growth and low temperatures reduced nutrient uptake. In nutrient-poor wetland soils, P lost from plant roots would likely be immobilised by soil microbes (Walbridge, 1991; Williams & Silcock, 2001), especially if the latter are additionally supplied with labile carbon, for example from root exudations (Moorhead & Linkins, 1997; Schmidt et al., 1997). Fertilisation experiments in such wetlands also showed reduced P retention by plants after N fertilisation (Güsewell et al., 2002).
Species-specific responses in relation to occurrence in the field
The five Carex species responded similarly to N : P supply ratios, that is high N : P ratios reduced second-year growth, root allocation, leaf longevity and nutrient recovery in all of them, which contrasts with the third hypothesis that species from N- and P-limited sites would respond differently. It has often been found that even Carex species from contrasting habitats are physiologically very similar (Choo & Albert, 1999; Visser et al., 2000). The consistently negative effects of high N : P ratios on second-year growth observed here are nevertheless remarkable given that Carex species tend to dominate P-poor wetlands, where they also often increase relative to other functional groups (grasses, forbs or bryophytes) after N fertilisation (Pauli et al., 2002; Aerts et al., 2003b; Van der Hoek et al., 2004). High N : P supply ratios might therefore be even more detrimental for grass or forb species that are less abundant at P-limited sites. Indeed, the forbs Cirsium palustre and Mentha aquatica had a large mortality between the first and second year at high-N-low-P supply (Güsewell et al., 2003a), and the grasses Alopecurus pratensis and Agrostis capillaris senesced rapidly already after 3 months of growth at high N : P supply ratio (H. Olde Venterink & S. Güsewell, unpublished data).
The five Carex species differed in biomass allocation, temporal pattern of growth and senescence, and nutrient resorption efficiency. Biomass allocation to roots was greater in C. curta and C. elata (most abundant at N-limited sites) than in C. flacca, C. flava and C. panicea (more abundant at P-limited sites). The same difference was found by Pérez-Corona & Verhoeven (1999) in a short-term experiment for the two species C. acutiformis (N-limited sites) and C. lasiocarpa (P-limited sites). Moreover, C. acutiformis tended to increase its biomass allocation to roots and reduce its biomass allocation to rhizomes under P deficient conditions (Pérez-Corona & Verhoeven, 1999), which is the opposite of the responses of C. flacca and C. panicea in this experiment. If indeed root turnover or exudation caused important nutrient losses from P-limited plants, a greater biomass allocation to roots might increase nutrient losses (Vásquez de Aldana et al., 1996). Thus, the differences in root allocation found here might be relevant at P-limited sites.
Interspecific differences in nutrient resorption efficiency revealed by this experiment correspond to those obtained for the same species at their field sites (Güsewell, 2005). The adaptive value of efficient nutrient resorption has been questioned (Escudero et al., 1992; Eckstein et al., 1999). The three Carex species from P-limited sites did not only have a high P resorption efficiency but also high resorption proficiency (Killingbeck, 1996), with litter P concentrations below 0.1 mg g−1 at all N : P supply ratios ≥ 15, and often even below 0.05 mg g−1. These results support the suggestion of Aerts & Chapin (2000) that highly efficient P resorption (> 80%) may be an adaptation to P-limited sites, whereas N resorption rarely exceeds 80%, as would be needed for a substantial role in nutrient conservation (Escudero et al., 1992). In the present experiment, the three species did not benefit greatly from their efficient P resorption since most leaf biomass was harvested before it had senesced. At unmanaged field sites, these species would not only resorb nutrients efficiently (Güsewell, 2005); C. flacca and C. flava keep part of their leaves until the next spring, when resorbed nutrients can be re-allocated directly to new growth, avoiding below-ground losses (Bausenwein et al., 2001). In C. panicea, the extensive rhizome system might improve nutrient storage. At managed field sites, the low stature of these three species would also allow them to retain part of their leaves, so that nutrient resorption would still afford some nutrient conservation.
Another difference was that numerous cluster roots were observed during the harvest on C. flacca, C. flava and C. panicea, but not C. curta and C. elata. The role of cluster roots and their exudations of organic acids for P acquisition in P-poor soils is well established (Lamont, 2003). In this experiment clusters seemed unimportant for P acquisition since all P was supplied as orthophosphate. In P-limited fens, however, the ability of plants to acquire calcium- or iron-bound phosphate might be important (Pérez-Corona et al., 1996).
In conclusion, this experiment has confirmed that high N : P supply ratios reduce second-year growth (and probably long-term performance) in wetland sedges, and that below-ground nutrient losses might contribute to this effect. The five Carex species responded similarly to N : P ratios despite their contrasting abundance at N- or P-limited sites. However, differences in root allocation and nutrient resorption suggest that the species would differ in their ability to avoid or to compensate nutrient losses at P-limited field sites. Although P limitation may limit the increase in biomass production and the spread of faster-growing species under high atmospheric N deposition (Gordon et al., 2001; Tomassen et al., 2004), the nutritional imbalance caused by high N and low P supply may be particularly detrimental to the growth and reproduction of individual species (Brouwer et al., 2001; Gotelli & Ellison, 2002). This possibility should be considered when assessing the impact of high N deposition on wetland vegetation.
Discussions with J.T.A. Verhoeven, W. Koerselman and R. Aerts stimulated and helped to focus this research. Constructive comments from H. Cornelissen, H. Olde Venterink, M. Suter, S. Toet and two referees improved the manuscript. U. Bollens, K. Edelkraut, T. Fotsch, M. Gaschen, A. Gerster, A. Hoenderboom, G. Rouwenhorst, R. Trachsler, and P. van Ven kindly assisted with experimental and analytical work. The research was supported by grant TMR grant ENV4-CT97-5075 from the European Union.
Appendix 1 Total amounts of nitrogen (N) and phosphorus (P) and of other elements supplied to each plant in the first and second year of the growth experiment.
Appendix 2 Estimation of shoot dry mass from morphological measurements.
Appendix 3anova results (F-ratios and significance levels) for the effects of the nitrogen (N) : phosphorus (P) supply ratio, overall supply level and Carex species on nutrient concentrations in shoot biomass (1999 and 2000), root biomass in 2000, and leaf litter in 1999.
Appendix 4 Nutrient concentrations (a–d, nitrogen (N), e–h, phosphorus (P)) in plant biomass (shoot biomass in 1999 and 2000, root biomass in 2000, leaf litter in 1999) in relation to the N : P supply ratio and overall supply level.
Appendix 5 Means per species and treatment (n= 3) for the biomass of all plant fractions (mg dry mass per plant), the number of inflorescences in May 2000, the rate of photosynthesis in July 1999, and nutrient concentrations in four of the biomass fractions.