Use of nitrogen to phosphorus ratios in plant tissue as an indicator of nutrient limitation and nitrogen saturation

Authors

  • Jack T. Tessier,

    Corresponding author
    1. State University of New York College of Environmental Science and Forestry, 350 Illick Hall, 1 Forestry Dr, Syracuse, NY 13210, USA
    Search for more papers by this author
  • Dudley J. Raynal

    1. State University of New York College of Environmental Science and Forestry, 350 Illick Hall, 1 Forestry Dr, Syracuse, NY 13210, USA
    Search for more papers by this author

*Present address and correspondence: Jack T. Tessier, Biological Sciences, Central Connecticut State University, 1615 Stanley St, New Britain, CT 06050 USA (fax +860 832 2594; e-mail TessierJ@ccsu.edu).

Summary

  • 1Ratios of nitrogen to phosphorus (N:P) in plant foliage have been used to assess nutrient limitation in wetland ecosystems and to indicate nitrogen saturation. Extension of this application to ecosystems other than wetlands remains to be evaluated.
  • 2We compared published N:P ratios as thresholds of nutrient limitation with published accounts from nutrient-addition experiments and the N:P ratios of understorey vegetation (Acer spp., Dryopteris intermedia, Erythronium americanum, Lycopodium lucidulum, Oxalis acetosella and Viola macloskeyi) from the Catskill Mountains of New York State, USA. We also performed a nutrient-addition experiment to test the response of these understorey plant species to inputs of N and P.
  • 3N:P ratios of Catskill understorey species indicated they were at or near P limitation relative to N:P ratios from other upland ecosystems. Our experiment supported this finding in that none of the species responded to N addition but all increased in P concentration and one increased in biomass with added P. Collectively, these results suggest that the understorey vegetation of the Catskill Mountains is not nitrogen limited, providing further evidence that hardwood forests in this area are nitrogen-saturated.
  • 4Synthesis and applications. This study demonstrates that N:P ratios can be effective predictors of nutrient limitation in upland ecosystems. Therefore N:P ratios can be used for management and monitoring purposes in considering the nutrient status of upland ecosystems with particular relevance to the continued deposition of elevated atmospheric N and to the diagnosis of nitrogen saturation.

Introduction

Although many forest ecosystems of the north-eastern USA have historically been considered (Vitousek & Howarth 1991) and experimentally determined (de Visser et al. 1994; Hurd, Brach & Raynal 1998; Rainey et al. 1999) to be nitrogen-limited, symptoms of nitrogen saturation in forests (Aber et al. 1989; Stoddard 1994) are increasingly being observed across North America (Fenn, Poth & Johnson 1996; Lovett, Weathers & Sobczak 2000; Spoelstra et al. 2001). In nitrogen-saturated forests, elevated nitrogen (N) export to runoff occurs throughout the year (Stoddard 1994) rather than only occurring during periods of flushing (Creed et al. 1996). Nitrogen saturation can be alleviated by disturbances large enough to stimulate subsequent growth increment in vegetation (Vitousek et al. 1998) or a reduction of atmospheric N deposition over an extended period.

Lovett, Weathers & Sobczak (2000) reported that stream water nitrate concentration patterns were indicative of nitrogen saturation in forested watersheds of the Catskill Mountains of New York, USA, a region that receives high rates of atmospheric N deposition (Ollinger et al. 1993). Stages of nitrogen saturation (Stoddard 1994) observed in the Catskills (Lovett, Weathers & Sobczak 2000) ranged from stage 0 (continuously low stream nitrate concentrations) to stage 3 (continuously high stream water nitrate concentrations). Other indications of nitrogen saturation include declines in net primary production, foliar biomass and fine root mass; an increase in foliar N concentrations and nitrate assimilation (Aber et al. 1989); and soil acidification and micronutrient depletion leading to a disruption between carbon (C) and N cycling (Asner, Seastedt & Townsend 1997).

Nitrogen saturation has many implications for ecosystems (Fenn et al. 1998), including altered stream and drinking water quality, altered plant community composition, decreased ecosystem health and modified nutrient cycling rates, and is therefore a serious concern for natural resource managers.

To characterize ecological problems, ecological stoichiometry is being used with increasing frequency (Chaneton, Lemcoff & Lavado 1996; Elser et al. 2000). For example, nitrogen to phosphorus (N:P) ratios have been used as diagnostic indicators of nitrogen saturation (Fenn, Poth & Johnson 1996) and limitation of vegetative growth by these nutrients (Penning de Vries, Krul & van Keulen 1980). More recently, N:P ratios have been applied to identify thresholds of nutrient limitation (Koerselman & Meuleman 1996; Verhoeven, Koerselman & Meuleman 1996; Aerts & Chapin 2000; Güsewell & Koerselman 2002). Based on studies of European wetland plants, thresholds of foliar N:P ratios were found to be < 14 for N limitation and > 16 for phosphorus (P) limitation. Recently, N:P ratios have been used to detect nitrogen saturation in western USA ecosystems (Fenn, Poth & Johnson 1996; Williams et al. 1996; Fenn et al. 1998). Foliar N:P ratios may be sensitive indices of nutrient limitation of vegetation growth, and thus may serve as effective tools for natural resource managers concerned about the effects of N deposition on ecosystem health. Assessments of plant N:P ratios using experimental N and P additions to upland vegetation are rare. We are aware of only two such studies in North America (Valentine & Allen 1990; Herbert & Fownes 1995), although work in alpine tundra has been conducted (Bowman et al. 1993; Bowman 1994; Theodose & Bowman 1997; Seastedt & Vaccaro 2001).

We predicted that ratios of N:P in understorey vegetation in Catskill Mountain forests exceed those of other upland vegetation because of high rates of N deposition in the region. Our objectives were to (i) compare the published N:P ratio thresholds for wetland ecosystems with N:P ratios from upland ecosystems; (ii) compare N:P ratios of Catskill understorey plant taxa with the thresholds and those of other upland systems to assess indications of limitation by P and possible nitrogen saturation in the ecosystem; and (iii) assess experimentally limitation by N or P in Catskill understorey vegetation.

Materials and methods

literature search on n:p ratios and nutrient limitation

We searched the literature for information on N:P ratios in ecosystems other than European wetlands to determine the universality of the thresholds set by Koerselman & Meuleman (1996). Online database searches for scientific literature were performed using OCLC FirstSearch and Cambridge Scientific with the following key-words: nitrogen content, nitrogen concentration, nitrogen deficiency, nitrogen limitation, phosphorus content, phosphorus concentration, phosphorus deficiency, phosphorus limitation, N:P, N/P, N:P ratio, N/P ratio, N to P ratio, nutrient concentration, nutrient content, nutrient deficiency, nutrient limitation, nitrogen fertilization and phosphorus fertilization. Subsequently, citation sections of primary literature located were used as a secondary source of pertinent studies. Studies were included if they involved fertilization with N and/or P separately sufficient to indicate limitation by the two nutrients. Studies were excluded if they did not provide N:P ratio or foliar nutrient concentration data sufficient to calculate the N:P ratios or if fewer than three replicates were included in the experimental design of the study.

Results were tabulated for a generalized comparison of N:P ratios and N or P limitation. We created plots of foliar N and P concentrations (sensuKoerselman & Meuleman 1996) from upland and wetland studies to assess visually thresholds of limitation by N and P. An N:P ratio exclusion line was placed on each plot to separate the systems that were solely P-limited from those that were limited by N alone or in tandem with P. This exclusion line was placed to minimize the number of outliers in the same way that Koerselman & Meuleman (1996) initially assessed N:P ratios as they relate to nutrient limitation in European wetlands. Not all of the studies found in the literature search could be included in the graphical display because some studies did not provide foliar nutrient concentration data.

assay of field plant n:p ratios

The study site for this project was a second-growth northern hardwood forest in Ulster County, New York, USA, within the Catskill State Park. The overstorey was dominated by Acer rubrum L., Acer saccharum Marshall, Fagus grandifolia Ehrh. and Betula alleghaniensis Britton. The understorey was dominated by Dryopteris intermedia (Muhl.) A. Gray, Lycopodium lucidulum Michx., Oxalis acetosella L. and seedlings of Acer saccharum, Acer rubrum L. and Acer pensylvanicum L. Nomenclature follows Gleason & Cronquist (1991). The 10-year mean for total N deposition near this site is 6·40 kg ha−1 (National Atmospheric Deposition Program (NSRP-3)/National Trends Network 2002).

As part of a study on vernal nutrient uptake by understorey vegetation, a total of 48 randomly located 1-m square plots was sampled during spring in 1999 and 2000. From a haphazardly chosen starting point, additional plots were located by walking a random number of paces in a random direction. Six plots were sampled at each harvest period. This replicate number was judged to be an adequate sample size to minimize estimates of error, while also preventing excessive damage to the forest understorey. Harvesting occurred periodically from snowmelt through to full canopy leaf-out in both years. All understorey vegetation within each quadrat was harvested to determine above- and below-ground biomass. Samples were divided by taxon: Lycopodium lucidulum, Dryopteris intermedia, Oxalis acetosella, Erythronium americanum Ker Gawler, or seedlings of Acer, collectively A. saccharum Marshall, A. rubrum L., and A. pensylvanicum L. Hereafter, these taxa will be referred to by their generic names.

Plant tissue samples were cleaned of soil with deionized water, dried for 1 week at 60 °C, weighed, and ground in a Wiley Mill to pass through a 1-mm screen. Total N was determined using Kjeldahl analysis (Bickelhaupt & White 1982). Total P was determined by microwave digestion using a CEM MDS 81D (CEM Corp., Matthews, NC, USA) followed by inductively coupled plasma spectroscopy using a Perkin-Elmer Optima 3300 DV ICP (Perkin-Elmer, Wellesley, MA, USA). Standard reference materials (NIST Apple leaves 1515 and Citrus leaves 1572; National Institute of Scientific Standards, Gaithersburg, MD, USA) were analysed along with the sample tissues to ensure accuracy within 10% of known N and P concentrations.

We calculated N:P ratios by dividing N concentration by P concentration (both as percentage nutrient in the sample) within each plant tissue sample. The community N:P ratio was calculated by determining the average N:P ratio of all of the tissue samples. N:P ratios were analysed with a five × eight (taxa × harvest date) factorial analysis of variance (anova). As there were no significant interactions, specific differences among the main effects were isolated using a Tukey's HSD with α= 0·05. All statistical analyses were performed using SAS version 7·0 (SAS Institute Inc. 1990).

experimental assessment of nutrient limitation

We collected plants from the aforementioned Catskill forest site for the experimental portion of this study. Erythronium americanum plants were collected in the spring of 1999. Dryopteris intermedia, Oxalis acetosella, Acer saccharum and Viola macloskeyi F. Lloyd plants were collected in the autumn of 1999. Lycopodium lucidulum was not used in this part of the study due to transplanting difficulties. Viola macloskeyi was included because it occurs in abundance in moist locations within the stand. Both above- and below-ground components were harvested.

We sought to homogenize initial biomass of the plants within each species so that changes in biomass that existed at the end of the experiment could be evaluated adequately. We used a screening process to ensure maximal consistency in initial biomass within each species. Ten plants of consistent height were collected and graded for inclusion in each experimental container (see below) for Erythronium, Acer and Viola. Dryopteris plants that had two to three fronds were chosen. Oxalis is a clonal species and thus patches of the species were collected that represented maximal cover over a 100-cm2 area. This procedure ensured that the initial biomass of plants within a species was as consistent as possible.

Plants were cleaned of soil using deionized distilled water and planted in 12·7-cm diameter plastic pots with an inert, N- and P-free, diatomaceous earth medium (PlayBall!, manufactured by Eagle-Picher Minerals Inc., Reno, NV, distributed by Agro-tech 2000 Inc., Plainsboro, NJ). This approach eliminated possible confounding effects from soil and soil microbes that would have occurred if forest soil had been used as a medium. A total of 35 experimental units of each of the five species was potted separately for use in the experiment, for a total of 175 pots.

Initially, these plants were placed in a greenhouse and treated three times per week with a complete nutrient solution designed to provide N and P at ambient concentrations found in soil water collected with lysimeters in northern hardwood forests of the north-eastern USA (Shepard et al. 1990; Zhang & Mitchell 1995). N and P concentrations corresponded to the ‘NP’ treatment (Table 1). This treatment was maintained until November 1999, when plants were covered with plastic and placed in a dark, cold room (temperature = 4·5 °C) for vernalization. While ambient levels of nutrients in soil do not account for the supply rate of nutrients, they do integrate nutrient input with the remaining nutrients available following plant and microbial uptake. Quantification of ‘plant-available’ nutrients can be extremely tenuous because it requires knowledge of rates of nutrient input from deposition and soil water flow, as well as biotic and abiotic nutrient uptake. Therefore, concentrations of nutrients in ambient extractable pools provide an index of what nutrients are available on an extended basis pending further pulses of nutrient input and subsequent biotic and abiotic sequestration, and were therefore used in this study as a surrogate for actual plant-available nutrient levels.

Table 1.  Contents of experimental growth solutions
 Treatment
1/2NNP2N5N1/2P2P5P
  • *

    One litre of micronutrient solution contained 3·73 g KCl, 1·55 g H3BO4, 0·34 g MnSO4 × H2O, 0·58 g ZnSO4 × 7H2O, 0·13 g CuSO4 × 5H2O, and 0·12 g Na2MoO4 × 2H2O.

NH4NO3 (g l−1)0·00120·00240·00480·0120·00240·00240·0024
0·5 m KH2PO4 (ml l−1)0·01360·01360·01360·01360·00680·02720·0680
0·5 m K2SO4 (ml l−1)5555555
1·0 m MgSO4 (ml l−1)2222222
Micronutrient solution (ml l−1)*1111111
Fe EDTA solution (ml l−1)1111111
CaSO4 (g l−1)0·3440·3440·3440·3440·3440·3440·344

In March 2000, plants were removed from the cold room, returned to the greenhouse, and placed in a randomized complete block design for two simultaneously run nutrient input experiments. In the first, N was provided to each of the species at each of four N input levels adjusted to 1/2N, NP (ambient), 2N and 5N of ambient field soil solution levels of N availability (Table 1). In the second, P was provided to each species at each of four P input levels adjusted to 1/2P, NP (ambient), 2P and 5P of ambient soil solution levels of P availability (Table 1). Nutrient solutions were standardized to pH = 6 and the growth medium had a pH = 6 as measured using a Corning 245 pH meter (Corning Inc., Corning, NY, USA). This design involved two factorial experiments of five species × four nutrient treatments with five replicates (blocks) of each experimental unit. A single set of five replicates for the NP treatment was used as the ambient supply rate for both experiments.

The plants were treated with the nutrient solutions three times per week for 8 weeks, a duration encompassing the length of the vernal nutrient flush (Creed et al. 1996) when plants normally receive high inputs of nutrients from snowmelt and nutrient mineralization. Plants were then removed from the pots, cleaned with deionized distilled water of all medium, dried for 1 week at 60 °C, weighed, and ground in a Wiley Mill to pass through a 1-mm screen. All plants were analysed for total N and P as described above.

In the N-addition experiment, biomass, N concentration and P concentration were analysed using anova in a five species × four nutrient treatments factorial blocked design. Because there was no interaction between species and nutrient treatment, a Tukey's HSD means separation test was used to isolate specific differences. The same statistical procedures were used for the P-addition experiment, with the exception of the biomass response. In that case there was an interaction between species and nutrient treatment and that interaction was isolated using LSMEANS and DIFF commands within proc mixed. This approach compares each data point with every other data point to determine specific differences and control for the experiment-wise error rate. The N:P ratio of each plant was determined by dividing the N concentration by the P concentration. The N:P ratios of all plants in the greenhouse study were compared using a five species × seven nutrient treatment factorial blocked anova. There were no significant interactions between species and nutrient treatment, therefore specific differences were isolated using a Tukey's HSD means separation test. All statistical procedures were performed in SAS version 7·0 (SAS Institute Inc. 1990) at α= 0·05.

Results

literature search on n:p ratios and nutrient limitation

Reports of nutrient limitation in the literature indicate that, in general, upland vegetation is limited solely by P at lower N:P ratios than wetland vegetation (Table 2). Graphical display of these data suggests that P limits vegetation growth in uplands at lower N:P ratios than in wetlands (Fig. 1). Vegetation in estuaries, in contrast, remained solely limited by N at higher N:P ratios than other wetland vegetation (Table 2), although insufficient data were available for graphical analysis of estuarine plants. Several studies offered thresholds for N vs. P limitation and areas in-between where N and P were co-limiting (Table 2).

Table 2.  Results of the literature search on N:P ratios and nutrient limitation in vegetation. The columns ‘N limitation’ and ‘P limitation’ indicate the N:P ratio at which the authors of the cited studies suggest that their respective ecosystem is limited by the indicated nutrient
StudySystemLocationAdditionsN:PLimited ByN LimitationP Limitation
Estuaries
Doering et al. (1995)EstuaryLaboratoryN & P15·6N  
Murray, Dennison & Kemp (1992)EstuaryVirginiaN & P14–18N & P  
Doering et al. (1995)EstuaryLaboratoryN & P20·4P  
Doering et al. (1995)EstuaryLaboratoryN & P25·9P  
Doering et al. (1995)EstuaryLaboratoryN & P28·9P  
Shores
Koerselman (1992)Coastal dunesThe Netherlands   < 16> 25
Wetlands
Vermeer (1986a)Wet grasslandThe NetherlandsN & P5·4N  
Aerts, Wallén & Malmer (1992)Sphagnum bogSwedenN & P6N  
Boeye et al. (1997)Wet meadowBelgiumN & P7·6–9·0N  
Verhoeven & Schmitz (1991)FenThe NetherlandsN & P11·95N  
Boeye et al. (1997)FenBelgiumN & P13·1–14·7N  
Vermeer (1986b)FenThe NetherlandsN & P13·3N  
Boyer & Wheeler (1989)FenEnglandP13·3P  
Boyer & Wheeler (1989)FenEnglandP14·6P  
Loach (1968)BogEnglandN & P21·4P  
Boyer & Wheeler (1989)FenEnglandP22·0P  
Tamm (1954)BogSwedenN & P23·0N  
Boeye et al. (1997)FenBelgiumN & P23·3–30·7P  
Verhoeven & Schmitz (1991)FenThe NetherlandsN & P23·85P  
Loach (1968)BogEnglandN & P25·3P  
Loach (1968)BogEnglandN & P30·0P  
Aerts, Wallén & Malmer (1992)Sphagnum bogSwedenN & P34P< 10> 14
Boyer & Wheeler (1989)FenEnglandP54·0P  
Wassen, Olde Venterink & de Swart (1995)MirePoland   < 14·3> 25
Verhoeven, Koerselman & Meuleman (1996)Wetland reviewEurope   < 14> 16
Koerselman & Meuleman (1996)Wetland reviewEurope   < 14> 16
Uplands
de Visser et al. (1994)Coniferous forestEuropeN7·0–14·5Not N  
Alan, Taylor & Dicks (2000)Carica papayaLaboratoryN & P7N & P  
Jacobson & Pettersson (2001)Picea abies standSwedenN & P7·54N  
Mohren, van den Burg & Burger (1986)Douglas fir forestThe NetherlandsN & P8N  
Jacobson & Pettersson (2001)Pinus sylvestris standSwedenN & P8·96N  
Clarholm & Rosengren-Brinck (1995)Picea abies plantationSwedenN & P9·8N & P  
Valentine & Allen (1990)Loblolly pine plantationNorth CarolinaN & P10·42N & P  
Bowman (1994)Alpine dry meadowColoradoN & P13N  
Herbert & Fownes (1995)Montane forestHawaiiN & P13·83P  
Bowman (1994)Alpine wet meadowColoradoN & P14P  
Ljungstrom & Nihlgård (1995)Beech forestEuropeP14·17At least P  
Bobbink (1991)Chalk grasslandThe NetherlandsN & P16·08N & P  
Present studyForest understoreyNew York 17·71P  
Mohren, van den Burg & Burger (1986)Douglas fir forestThe NetherlandsN & P22–25P  
Aerts & Berendse (1988)HeathThe NetherlandsN & P29·41P  
Aerts & Berendse (1988)HeathThe NetherlandsN & P27·5P  
Wall, Hellsten & Huss-Danell (2000)Alnus and TrifoliumLaboratoryN & P  > 7 
Ericsson et al. (1993)Picea abies plantationSweden   > 12·5 
Penning de Vries, Krul & van Keulen (1980)GrasslandWestern AfricaN & P  < 6·67> 26·32
Figure 1.

Comparison of N:P ratios among ecosystems. N:P ratio threshold of N limitation among (a) upland ecosystems and (b) wetland ecosystems based on literature search. (c) N:P ratio of Catskill understorey vegetation. Means with different letters are significantly different at α= 0·05. Error bars represent one standard error above and below the mean.

assay of field plant n:p ratios

Catskill understorey vegetation had relatively high N:P ratios compared with other upland ecosystems (Table 2). There were no significant differences among the collection dates. The pteridophytes Lycopodium and Dryopteris had the highest N:P ratio (Fig. 1c) and, based on many published thresholds (Aerts, Wallén & Malmer 1992; Ericsson et al. 1993; Koerselman & Meuleman 1996; Verhoeven, Koerselman & Meuleman 1996), were limited by P and not N. The other understorey taxa had N:P ratios placing them near or above the upper limits of N limitation without concomitant P limitation (Fig. 1c and Table 2).

experimental assessment of nutrient limitation

In the N-addition experiment there were no interactions between species and nutrient input level, therefore only the main effects of all species regardless of treatment and all treatments regardless of species are combined in Fig. 2. Dryopteris had the greatest biomass in all treatments and Erythronium the least (Fig. 2a). Oxalis and Viola had the highest concentrations of N and P and Acer had the lowest in all treatments (Fig. 2b,c). There were no significant differences in biomass, N concentration and P concentration among the different nutrient treatments in the N-addition experiment (Fig. 2d,e,f).

Figure 2.

Main effects for the N-addition experiment. Differences among species for all treatments combined in (a) biomass, (b) N concentration and (c) P concentration. There were no significant differences in (d) plant biomass, (e) N concentration and (f) P concentration among nutrient treatments for all species combined. Means with different letters are significantly different at α= 0·05. Error bars represent one standard error above and below the mean.

There was a significant interaction between species and nutrient input level in the P-addition experiment. Dryopteris was the only species to show a biomass response, reaching its greatest biomass at the 2P level and declining at the 5P level (Fig. 3a). None of the other species showed a biomass response (Fig. 3a). There were no interactions between species and treatment for nutrient concentration in the P experiment, therefore only main effects are shown in Fig. 3 as described above. Similar to the results of the N-addition experiment, Oxalis and Viola had the highest N and P concentrations in the P-addition experiment and Acer was consistently among the species with the lowest N and P concentrations (Fig. 3b,c). There was no significant response in N concentration in the P-addition experiment (Fig. 3d) but plants in the 5P treatment had a higher P concentration than plants in the NP treatment (Fig. 3e).

Figure 3.

Main effects and interaction for the P-addition experiment. (a) Biomass–species interaction. Differences among species for all treatments combined in (b) N concentration and (c) P concentration. Differences among nutrient input levels for all species combined in plant (d) N concentration and (e) P concentration. Means with different letters are significantly different at α= 0·05. Error bars represent one standard error above and below the mean.

There was no interaction between species and nutrient input treatment for N:P ratios of plants in the N-addition and P-addition experiments combined. Dryopteris and Erythronium were among the species with the highest N:P ratios and Viola had the lowest (Fig. 4a). The N:P ratio for plants in the 5N nutrient treatment was greater than that of the plants in the 5P nutrient treatment (Fig. 4b).

Figure 4.

N:P ratio results for nutrient-addition experiments. Main effects on N:P ratio of (a) species and (b) nutrient input levels. Means with different letters are significantly different at α= 0·05. Error bars represent one standard error above and below the mean.

Discussion

literature search on n:p ratios and nutrient limitation

Results of the literature search revealed extensive differences among studies regarding thresholds of N and/or P limitation of vegetative growth. Published N:P ratio thresholds of N limitation ranged from 6·7 to 16, while those for P limitation ranged from 12·5 to 26·3. Fewer published thresholds exist for upland systems than for wetland systems. Clearly there is a need for more experimental work to define accurately the N:P ratios for N and P limitation in specific upland systems (Koerselman & Meuleman 1996) in order for N:P ratios to be used for management and monitoring purposes. Continued assessment of N:P ratios among ecosystem types may also help to reveal the ecological significance of the few outliers in Fig. 1.

A spectrum of N:P ratio thresholds may exist for nutrient limitation given the general trend from estuarine to upland systems found in this literature search. Nitrogen limitation occurs at higher N:P ratios in wetlands (Koerselman & Meuleman 1996) than in uplands (Penning de Vries, Krul & van Keulen 1980; Fenn, Poth & Johnson 1996) and at still higher N:P ratios in estuaries (Murray, Dennison & Kemp 1992; Doering et al. 1995) than in wetlands. This gradient of N:P ratios indicative of N or P limitation may be related to nutritional adaptations of plants that result in the gradient of N typically limiting growth in upland settings (Vitousek & Howarth 1991) to P typically limiting growth in freshwater settings (Schindler 1974).

assay of field plant n:p ratios

Lycopodium and Dryopteris in the Catskill forest had higher N:P ratios (25·50 and 18·23, respectively) than other reported understorey N:P ratios in upland forest ecosystems (Fenn, Poth & Johnson 1996; T.M. Hurd & D.J. Raynal, unpublished data). Bracken fern had an N:P ratio of 14·8 (Fenn, Poth & Johnson 1996) in a nitrogen-saturated forest site in southern California. T.M. Hurd & D.J. Raynal (unpublished data) found N:P ratios of 9·98–11·78 in Dryopteris intermedia in the Adirondack Mountains of New York. The angiosperms (Acer, Oxalis and Erythronium) in the Catskill understorey had N:P ratios (12·96, 11·96, and 11·61, respectfully) intermediate to these levels but they were still high relative to other reported upland systems (Valentine & Allen 1990; de Visser, Krul & van Keulen 1994; Williams et al. 1996). Thus, values of Catskill understorey vegetation are near the P-limited end of the range of N:P ratios in upland understorey vegetation.

Such high N:P ratios may be a result of chronically high N deposition rates in the region (Ollinger et al. 1993). The grand mean N:P ratio for Catskill understorey vegetation of 17·71 indicates that this layer may have changed to P limitation from its historic condition of N limitation (Vitousek & Howarth 1991). N:P ratios have been used previously as an index of nitrogen saturation in southern California mixed conifers (Fenn, Poth & Johnson 1996) and in the alpine zone of the Colorado Front Range (Williams et al. 1996). The mean N:P ratio of the Catskill understorey vegetation is greater than the N:P of species of Pinus and Quercus in Fenn, Poth & Johnson (1996) and the N:P of Pinus aristata Engelm. in Williams et al. (1996). The high N:P ratios for the Catskill understorey vegetation support the interpretation of Lovett, Weathers & Sobczak (2000) that forests of the Catskill region are increasingly nitrogen-saturated (Stoddard 1994).

The 10-year mean for total N deposition was 2·31 kg ha−1 near the southern California site of Fenn, Poth & Johnson (1996), 3·16 kg ha−1 near the Colorado Front Range site of Williams et al. (1996), 4·70 kg ha−1 near the Adirondack Mountain site of T.M. Hurd & D.J. Raynal (unpublished data), and 6·40 kg ha−1 near the Catskill Mountain site of this study (National Atmospheric Deposition Program (NSRP-3)/National Trends Network 2002). Despite being exposed to elevated rates of N deposition, Dryopteris in the Adirondacks (T.M. Hurd & D.J. Raynal, unpublished data) has lower N:P ratios than vegetation in the southern California sites of Fenn, Poth & Johnson (1996). The understorey plants in the Catskill forest of this study have the highest N:P ratios of the studies surveyed and have also experienced the highest rates of N deposition. This lack of consistent effect of N deposition on N:P ratios among varying ecosystems indicates that plant species from different ecosystems may not respond uniformly to N deposition. Outlying data points in Fig. 1 may reflect these differences among ecosystem types in their response to N deposition and resultant N:P ratios. These plant species may also not respond uniformly to declines in N deposition brought about by legislation. Clearly, future monitoring of ecosystems recovering from chronic N deposition will be vital if we are to gauge the representativeness of these forests to future regulations governing N emissions. If research is performed within several upland ecosystem types to identify the thresholds of N:P ratios indicative of limitation by N or P, then N:P ratios have the potential to be a valuable monitoring tool in the future.

The absence of sites within the range of these understorey species that have not received elevated rates of N deposition precludes comparison of N:P ratios in these species in areas of high and low rates of N deposition. However, higher N:P ratios found in Dryopteris in the Catskills compared with those in the Adirondacks (T.M. Hurd & D.J. Raynal, unpublished data) may be indicative of effects associated with greater N deposition in the Catskill region. If high N:P ratios in understorey vegetation are the result of elevated rates of N deposition, then we would predict that N:P ratios would decrease if N deposition is experimentally decreased and would increase further if N deposition rates are experimentally increased in a given area. Future studies could address this hypothesis.

Differences in N:P ratio among taxa may be used to predict community compositional change. Lycopodium and Dryopteris are clearly more P-limited than are Acer, Oxalis and Erythronium. Given continued rates of N deposition, those taxa that are still limited by N (possibly Acer, Oxalis and Erythronium) should experience higher relative growth rates compared with those taxa that are no longer limited by N, assuming that N or P are limiting in the ecosystem. If the different taxa belong to different functional groups (however they are defined) then altered community composition may result in changes in the functional properties of the understorey layer, as observed in nutrient-addition experiments in alpine tundra communities (Bowman et al. 1993; Theodose & Bowman 1997). N:P ratios can then be helpful in predicting future ecosystem species composition and subsequently ecosystem function. Further studies are needed to evaluate rates of change in understorey composition related to atmospheric deposition of nutrients in order to predict the time span over which compositional and functional changes may occur. Specifically, data are needed from additional systems and locations as well as long-term data relating changes in N:P ratios of vegetation to plant community compositional change.

Aerts & Chapin (2000) caution against using N:P ratios of individual species rather than of the site because species can differ from the community N:P ratio (as we observed) and therefore indicate a less than direct relationship between nutrient limitation on a species level and net primary productivity. As the species with greatest mass are likely to dominate community biogeochemical properties (Grime 1998), net primary production should be related to the community-level N:P ratio and be driven by the N:P ratios of the dominant species. Therefore, an assessment of individual taxa can provide an indication of the response of each one to nutrient addition (Mohren, van den Burg & Burger 1986; Valentine & Allen 1990; Albaugh et al. 1998) and, by extension, the response of community biomass production as driven by the dominant species (Grime 1998). Thus, in this case, it is appropriate to consider the individual responses of plants to continued N deposition.

Aerts & Chapin (2000) also observed that the N:P ratio thresholds apply only when either N or P is limiting. Other factors that could limit plant growth include light availability, soil water and suboptimal temperatures. Because the plants in this study were harvested in spring when the canopy leaves are absent and melt water is abundant, it seems unlikely that light or water limits plant growth and nutrient uptake during this period. Temperature may have been limiting but as N:P ratios did not change over time (see the Results) and these taxa were clearly undergoing phenological development (J. Tessier, personal observation) it seems unlikely that cool temperatures in the early growing season were affecting the N:P ratios in these taxa. Therefore, we assume that either N or P was limiting growth in these plants, as is found in many cases (Vitousek & Howarth 1991).

experimental assessment of nutrient limitation

Oxalis and Viola have a lower growth rate relative to nutrient uptake than Dryopteris, Acer and Erythronium. If Oxalis and Viola were more efficient at nutrient capture than the other species, they would have exhibited higher N and P concentrations at the higher nutrient input levels than the other species. The lack of this difference indicates that they were capturing nutrients at the same rate as the other species but were not growing as rapidly per unit nutrient as the other species. However, the lack of a consistent biomass accumulation among species at the initiation of the experiments precludes the testing of this hypothesis.

The understorey species used in these experiments are not N-limited. If these species were N-limited they would be expected to respond to N addition with increases in N concentration or biomass, as observed by Hurd, Brach & Raynal (1998) in Adirondack hardwood forests. Dryopteris was the only species to exhibit both a biomass increase and an increase in P concentration in response to P addition, suggesting that it is the most P-limited species. The perennial life history of these species and the fact that only Erythronium completed a growth period to senescence during the study may have limited a biomass response in the other species. Additionally, all species responded to P addition with increasing P concentrations of tissue, indicating that they are responsive to P availability and therefore are probably P-limited (Bowman 1994). It is likely that more of the species would have had a biomass response to P addition (like Dryopteris, which was the most P-limited species in the experiment based on its N:P ratio in the field) had the experiment been carried out on a longer term basis. As calcium has the ability to precipitate P (Noe, Childers & Jones 2001), the high calcium content of the growth medium may have diminished the availability of P in solution, making the response of the plants to high P input levels a conservative one. This finding of an N surfeit of understorey vegetation in the Catskills is similar to that of Rainey et al. (1999), who found low levels of N uptake by understorey vegetation in a 15N tracer study in a Massachusetts red pine plantation and suggested that the understorey vegetation of that system was ceasing to be N-limited.

The decline in biomass of Dryopteris at the 5P input level suggests that the species decreased below-ground tissue allocation in response to a greater supply of a limiting nutrient (Ericsson & Ingestad 1988). The trend of increasing biomass from 1/2P to 2P indicates that the plants grew larger in response to increased availability of a limiting nutrient, as demonstrated in other studies (Valentine & Allen 1990; Albaugh et al. 1998). Dryopteris has determinate growth in its above-ground biomass (above-ground growth only occurs during a short time period in spring and was completed early in the duration of the experiment), indicating that the change in biomass occurred primarily in rhizome and root tissue. Hence a decline in total biomass at high nutrient availability implies that the plants were able to divert resources from below-ground mass towards other functions. What those other functions might be deserves further investigation but may include photosynthetic activity and resource capacitance for times of lower nutrient availability.

The N:P ratios of experimental plants were affected by nutrient availability. The combined non-significant increase in N concentration and decrease in P concentration in response to the 5N nutrient treatment led to a significant difference between the N:P ratio of plants at the 5N level compared with the 1/2P and 5P levels. The lowest N:P ratio was the 13·37 of Viola. As all species responded similarly to P availability by increasing P concentration at the 5P nutrient level, even plants with a N:P ratio of 13·37 are P-limited and not N-limited. This finding further supports evidence from the literature that upland vegetation is solely P-limited at lower N:P ratios than wetland vegetation (Koerselman & Meuleman 1996). Dryopteris was the most P-limited species and had the highest N:P ratio in the greenhouse experiment, further substantiating the use of N:P ratios as indicators of nutrient limitation in vegetation instead of more logistically difficult soil nutrient indicators (Smethurst 2000).

Further progress on the use of N:P ratios as indicators of nutrient limitation continues to provide evidence for the generalized use of the thresholds of N:P ratios that indicate nutrient limitation (Güsewell & Koerselman 2002). Our results support the previously reported thresholds in wetland systems and extend these findings to consider upland ecosystems. Güsewell & Koerselman (2002) report restrictions on the use of N:P ratios as indicators of nutrient limitation that include variation in thresholds based on temporal variation in weather patterns, the necessity that either N or P be the limiting factor, variations in thresholds depending on altitude, and that thresholds should not be applied to individual species. We would add to this list the potential for differences in thresholds among ecosystem types, as the upland systems that we reviewed tended to be N-limited at lower N:P ratios than wetland ecosystems. Further research should seek to quantify this variation among a range of ecosystem types.

In contrast to the findings of Güsewell & Koerselman (2002), our results indicate that N:P ratios may be used to indicate nutrient limitation in individual species. Dryopteris had the highest N:P ratio in the field assay and also was the most P-limited of the species considered in the experimental portion of this work. Within the upland northern hardwood forest studied here, N:P ratios can be effective predictors of nutrient limitation within species as well as the ecosystem as a whole. Further work should consider the extent to which this pattern holds true among ecosystem types.

The results of the experimental assessment of nutrient limitation support those of the assay of field plant N:P ratios, that all of the species investigated are not N-limited. These results indicate that N:P ratios are useful predictors of nutrient limitation in this northern hardwood forest. Therefore, N:P ratios can be used for monitoring of nutrient limitation in forests that have received high levels of atmospheric N deposition, as well as for assessing recovery of these systems as N deposition decreases with the advent of emissions controls. Within managed northern hardwood forests, N:P ratios may also be used to determine the utility, composition and effectiveness of fertilization treatments designed to maximize growth.

In conclusion, thresholds of N:P ratios indicative of N or P limitation vary among ecosystem types. More experimental work needs to be done in upland systems to isolate critical N:P ratios to indicate N or P limitation for use in ecosystem and natural resource management. High N:P ratios in Catskill understorey vegetation and its response to experimental N and P fertilization indicate phosphorus limitation and provide further evidence of nitrogen saturation of the ecosystem. Differences among N:P ratios of taxa in the understorey could lead to plant community compositional and ecosystem functional change given continued high rates of N deposition.

Acknowledgements

The authors thank the following for their assistance with this study. The Frost Valley YMCA permitted use of their forested property. Funding was provided by the New York City Department of Environmental Protection and the USDA Forest Service in co-operation with the US Geological Survey. Field and laboratory assistance was provided by Lisa Tessier, Kim Anderson, Tim Schreiber, Karl Didier, Steve Fuller, Dave Kubek, Scott Jones, Kim Kerinan, Tom Touchet, Greg McGee, Thad Yorks, Jodi Forrester, Dave Conner, Stephanie Kroll, Cyndi Boesse, Larry Whelpton, Don Bickelhaupt, Deb Driscoll and Marlene Braun. Dale Tuttle at Northern Nurseries, in Cicero, NY, donated the PlayBall! Helpful comments on earlier drafts of the manuscript were provided by Doug Burns, Don Leopold, Sam McNaughton, Myron Mitchell, Ruth Yanai and three anonymous referees.

Ancillary