Response of the herbaceous layer of forest ecosystems to excess nitrogen deposition

Authors


Frank S. Gilliam (tel. +1 304 696 3636, fax +1 304 696-3243, e-mail gilliam@marshall.edu).

Summary

  • 1This review brings into focus what is known about the response of the herbaceous layer of forest ecosystems to increasing nitrogen deposition. The emphasis on forests in general is important for two reasons. First, forests often occupy areas receiving high rates of atmospheric deposition of N. Second, compared with herb-dominated communities, about which much is known regarding response to excess N, forests generally display greater biological and structural complexity. The more specific focus on the herbaceous layer – here defined as all vascular (herbaceous and woody) plants ≤ 1 m in height – is warranted because most of the species diversity of forests occurs in the herb layer.
  • 2Most responses of forest ecosystems to N saturation, defined here as ecosystem supply of N exceeding ecosystem demand for N, can be characterized by two complementary hypotheses, each emphasizing different facets of ecosystem structure and function. One focuses on ecosystem processes, such as net primary productivity, whereas the other addresses seasonal patterns of concentrations of inline image in stream water. Although neither hypothesis considers the effects of chronically elevated N deposition on forest herbs, both share a prediction – a dramatic increase in availability of inline image– that is relevant to forest herb response. Such a shift towards inline image dominance has important implications for changes in herbaceous layer dynamics. For example, increases in soil inline image pools can increase invasibility by exotic species. Also, because preferential use of inline image vs. inline image can be highly species-specific among forest herbs, predominance of inline image availability can bring about further changes in herb layer species composition. Also important to forest herbs is the tendency for enhanced mobility of inline image to increase mobility and decrease availability of essential cations, such as Ca2+.
  • 3The response of plant species of the herbaceous layer of forest ecosystems to excess N availability can arise from N-mediated changes in several processes. Here I identify six – competition, herbivory, mycorrhizal infection, disease, species invasions and exotic earthworm activity – and discuss how they respond to excess N and how this response affects the herb layer of forests. With the exception of the activity of exotic earthworms (which is correlated with, but not necessarily caused by, high N deposition), there is compelling evidence that all processes respond sensitively to N deposition and that this response generally leads to drastic shifts in species composition and decreases in biodiversity of forest herb communities.
  • 4A hypothesis is suggested – the nitrogen homogeneity hypothesis – that predicts loss of biodiversity in forest ecosystems experiencing chronically elevated N deposition. This hypothesis is based on the decrease in spatial heterogeneity of N availability that is typically high in forest soils under N-limited conditions.

Introduction

The human-caused increase in biologically reactive nitrogen (N) in the biosphere is a biogeochemical phenomenon that has consequences for plant biodiversity and thus embraces two distinct facets of plant ecology: biogeochemistry and vegetation science. Biogeochemists who understand the mechanisms of how excess N alters nutrient cycling in terrestrial ecosystems may be less aware of the processes that determine the composition of plant communities. Conversely, vegetation scientists who understand plant community dynamics may be less aware of the biogeochemical consequences of adding excess N to the environment. A primary impetus of this review, thus, is an attempt to bridge the gap between these two prominent ecological subdisciplines.

The main purpose of this review is to bring into focus what is known about the response of the herbaceous layer of forest ecosystems to increasing N deposition. The emphasis on forests in general is important for two reasons. First, forests occupy a high proportion of areas receiving high rates of atmospheric deposition of N, e.g. eastern North America and western Europe (Holland & Lamarque 1997). Second, compared with herb-dominated communities, about which much is known regarding response to excess N (Pennings et al. 2005; Suding et al. 2005), forests generally display greater biological and structural complexity. The specific focus on the herbaceous layer is warranted because most of the species diversity of forests occurs in the herb layer (Gilliam & Roberts 2003a).

My review will be organized into four sections. The first section will introduce the phenomenon of N saturation. This is not intended to provide an exhaustive review of the phenomenon; indeed, I would encourage readers to consider previous publications that do an excellent job reviewing N saturation, e.g. Fenn et al. (1998) or Aber et al. (2003). Rather, I will focus primarily on biogeochemical processes with a particular focus on aspects germane to forest plant species, e.g. the relationship between soil inline image pools and invasibility, and inline image vs. inline image use by herb layer species.

The second section will be a synthesis of published work on the effects of N additions on the forest herbaceous layer. Because the nature of the problem of excess N differs notably between North America and Europe, the way in which the effects of N on herbaceous vegetation has been studied also varies between these two regions. Consequently, I will distinguish between European and North American literature.

The third section will be devoted to identifying specific mechanisms – such as increased disease or decreased competitive ability – to explain observed responses of forest herbaceous layers to excess N availability. These will be posed as potentially complementary, rather than mutually exclusive, hypotheses to help understand and predict the effects of increasing N supply on forest herb communities.

The final section will be used to develop a general hypothesis of the negative effects of excess N on forests, one which has particular relevance to the herbaceous layer. Biodiversity is more than just the number and balance of different species occurring in an area; it also has a functional process component (Roberts & Gilliam 1995). Spatial and temporal heterogeneity in these processes (e.g. nutrient cycling, energy flow and species dynamics) contribute to the overall biodiversity of an ecosystem. This hypothesis will make a connection between excess N-mediated decreases in heterogeneity and decreases in species diversity of impacted forests, while predicting further decreases in the future in areas where N deposition remains unmitigated.

Note that for the sake of consistency I will use the term ‘herbaceous/herb layer’ throughout this review to refer to the lowest stratum of forest vegetation. As reviewed by Gilliam & Roberts (2003b), this layer is often, though not always, defined as comprising all vascular plants – woody and herbaceous – that are ≤ 1 m in height. Many aspects vary, however, across this facet of the plant ecological literature, including the specific height used to delimit the stratum and exclusion/inclusion of non-vascular plants (e.g. bryophytes). Gilliam & Roberts (2003b) also surveyed synonymous terms found in the literature, including ‘herbaceous/herb stratum’, ‘herbaceous understorey’, ‘ground layer’, ‘ground vegetation’ and ‘ground flora’. Again, for the sake of simplicity and consistency I will use ‘herbaceous/herb layer’, despite alternative terms used within given studies.

Nitrogen saturation

The N2 gas that makes up 78% of our atmosphere is inert to virtually all living organisms, other than a small group (< 60 genera –Postgate 1982) of prokaryotes (i.e. symbiotic and non-symbiotic N-fixers). Furthermore, other than the high energy associated with lightning strikes that combines N2 and O2 ultimately to form NO2, N2 enters into few non-anthropogenic, abiotic chemical reactions. As a result, N2 is often referred to as non-reactive N in contrast to other forms, such as inline image, inline image and NO2, that are collectively referred to as reactive N (Nr). Although the most inclusive sense of Nr (see Galloway et al. 2004 for a recent comprehensive review) comprises all forms of N that readily enter into reactions, including biochemical reactions (biologically active), photochemical reactions (photochemically active) and electromagnetic absorption (radiatively active), I confine my discussion here to biologically active Nr, particularly the predominantly plant-labile forms of inline image and inline image.

Increases in anthropogenic N-fixation, including fossil-fuel combustion and the Haber–Bosch process for production of N fertilizers, have added increasing amounts of Nr to the biosphere. Galloway et al. (2004) estimated that total atmospheric deposition of inline image and inline image to terrestrial ecosystems has increased from 17 Tg N year−1 in 1860 to 64 Tg N year−1 in the early 1990s, and projected further increases to 125 Tg N year−1 by 2050. [Note: for simplicity I am presenting these as inline image and inline image rather than the more inclusive formulas NOy and NHx used by Galloway et al. (2004).] Thus, there will have been a nearly 10-fold increase in atmospheric deposition of Nr within the period of the last 200 years.

Although atmospheric deposition of N clearly has increased over this period, it is still quite spatially heterogeneous, with higher levels of N deposition generally being closely associated with high densities of human populations. For example, annual N deposition can vary 3–4-fold across the north-eastern USA, increasing from eastern Maine to western New York (Ollinger et al. 1993). Terrestrial ecosystems that receive high levels of N deposition are often predisposed to a phenomenon known as N saturation, which develops when the availability of N exceeds the total biological demand for N (Aber et al. 2003). Other factors, however, such as topography, vegetation and geology (i.e. parent materials), can also predispose ecosystems to N saturation to varying but important degrees (Williard et al. 2005). In parts of North America (particularly the eastern USA) and Europe there is a high degree of spatial coincidence of forests and chronically elevated atmospheric N deposition (Holland & Lamarque 1997). Consequently, most N-saturated ecosystems are forests.

Responses of forest ecosystems to N saturation can be generally characterized by two complementary hypotheses, each of which emphasizes different facets of ecosystem structure and function. The first, originally articulated by Aber et al. (1989), focuses on ecosystem processes, such as net primary productivity (which is predicted to decline at later stages of N saturation), whereas the other, described by Stoddard (1994), addresses seasonal patterns of concentrations of inline image in stream water. Notably, neither hypothesis considers the effects of chronically elevated N deposition on forest biodiversity, particularly that of herb layer communities. They do, however, share a prediction that is relevant to forest herb response: a dramatic increase in availability of inline image accompanied by change in N dynamics toward rapid movement of inline image in N-saturated forest soils. Such a shift towards inline image dominance has important implications for changes in herbaceous layer dynamics. For example, increases in soil inline image pools can increase invasibility by exotic species (Fargione & Tilman 2005). Also, because preferential use of inline image vs. inline image can be highly species-specific among forest herbs (Marschner 1995; Muller 2003), predominance of inline image availability can bring about further changes in herb layer species composition. Also important to forest herbs is the tendency for enhanced mobility of inline image to increase mobility and decrease availability of essential cations, such as Ca2+ (Gilliam et al. 2005).

The second hypothesis of N saturation (Stoddard 1994) predicts that there should be elevated concentrations of stream inline image as N saturation progresses, accompanied by decreases in the magnitude of seasonal fluctuations of stream inline image found in N-limited forests, in which there are measurable levels of inline image during the dormant season that decline to nearly undetectable levels during the growing season. Peterjohn et al. (1996) clearly demonstrated this response for a mixed hardwood forest watershed at the Fernow Experimental Forest (FEF), West Virginia, experiencing chronically elevated ambient inputs of N. Analysing data from 1971 to 1994, they found that annual variability in stream inline image concentrations decreased from a maximum coefficient of variation (which reflects the degree of seasonal fluctuation) of > 100% in 1975 to a minimum consistently < 20% from 1984 to the end of the sampling period. Thus, because inline image leaching is pronounced during the growing season at N-saturated sites, ecosystem loss of cations can be substantial at a time when plant demand for cations is at a maximum. Furthermore, dormant-season losses of inline image can facilitate loss of essential cations at a time when metabolic activities of plants are negligible and thus virtually incapable of mitigating such losses.

In addition to these effects of excess N on temporal dynamics of N processing, Gilliam et al. (2001) showed at this same site (FEF) that experimental additions of N to an entire watershed significantly decreased spatial heterogeneity of N processing, including net N mineralization and nitrification and uptake of N by prominent herb species, compared with a high degree of natural spatial heterogeneity on untreated reference watersheds. Decreases in spatial heterogeneity of soil N processing have important implications for the response of forest herbs – including composition, cover and biodiversity – to excess N deposition.

Response of the herbaceous layer to excess N

Field experiments investigating the effects of N availability on species composition and diversity of plant communities date back to the mid-19th century, with the first being the Park Grass Experiment in Rothamsted, England, established in 1856 (Richardson 1938). Although that experiment originally was designed to determine the effect of a variety of inorganic nutrient fertilizers on plant yield (and to compare it with the effect of manure), the early researchers also observed rapid and notable effects on species composition, particularly in response to additions of inorganic N. One of the more significant findings was that excess N greatly shifted species composition and decreased species diversity of herb communities. This response has been supported by even more extensive recent studies at the Cedar Creek Natural Area, Minnesota, that have focused specifically on the effects of a wide range of levels of N addition (Fig. 1). Regrettably, in contrast to numerous studies on effects of N on herb-dominated ecosystems (see Suding et al. 2005 for a meta-analysis of nine such sites in the USA), fewer have been done on forested ecosystems (Bobbink et al. 1998). This is a significant discrepancy because forest ecosystems generally display far greater structural and biological complexity than their herb-dominated counterparts, thus precluding extrapolation of findings from herb-dominated systems to forests, at least without empirical data to support such extrapolation.

Figure 1.

Change in relative abundance (%) in response of important grass species (Schizachyrium scoparium and Agropyron repens) to experimental nitrogen (N) additions for 3 years, which were the following: Low N = 1 kg N ha−1 year−1, High N = 27.2 kg N ha−1 year−1. Figure based on data taken from Tilman (1987).

synthesis of recent research on stand-level responses of forest herb layers to n

Problems associated with chronically elevated N deposition leading to N saturation were addressed earlier in Europe than in North America (e.g. Ågren & Bosatta 1988), in part because these problems arose earlier and were relatively more widespread in Europe. Excess wet deposition of N in Europe has largely been in the form of inline image arising from NH3 volatilization associated with agricultural practices (Asman et al. 1998; van Tol et al. 1998; Erisman & de Vries 2000; Krupa 2003). By contrast, anthropogenically elevated wet N deposition in North America has occurred more in the form of inline image, resulting from emissions of NOx associated with high-energy combustion, including internal combustion engines of motor vehicles (USEPA 2002). Not surprisingly, then, more research on the biogeochemical effects of N deposition has been carried out in Europe than in North America, particularly in the form of co-ordinated networks such as NITREX (Wright & vanBreemen 1995), and more work on the effects of N on forest vegetation has also been done in Europe than in North America. For example, when Taylor et al. (1994) reviewed the effects of air pollutants on forests of North America they summarized numerous studies of biogeochemical and tree physiological responses of forest ecosystems to N deposition, but provided no data on the response of the herb layer. However, when Boxman et al. (1998) summarized vegetation and soil biota responses for the NITREX network, they presented no data on the effects of N manipulations on the herb community of European forests. A summary of results reviewed here is given in Table 1.

Table 1.  A summary of studies examining the effects of excess N on the herbaceous layer of contrasting forest ecosystems
Ecosystem typeN treatment/depositionTimeResponseReference
  • *

    Time since termination of treatment, which was carried out as a tree growth fertilization experiment from 1971 to 1990.

Boreal forest, Sweden50 kg N ha−1 year−1/NH4NO31 yearno change in biomassNordin et al. (1998)
Oak forests, Sweden6–20 kg N ha−1 year−1/gradientNAincreases in nitrophilous, acid-tolerant speciesBrunet et al. (1998)
Spruce/fir forest, Switzerland30 kg N ha−1 year−1/NH4NO33 yearsno significant responseSchleppi et al. (1999)
Scots pine plantation, Sweden108 kg N ha−1 year−1/NH4NO39 years*long-lived effects, significant decline in diversityStrengbom et al. (2001)
Boreal forest, Sweden< 3 to > 12 kg N ha−1 year−1/gradientNAdecline in occurrence/cover of ericaceous shrubsStrengbom et al. (2003)
Hardwood forest, USA35 kg N ha−1 year−1/(NH4)2SO42 yearsno significant responseGilliam et al. (1994)
Hardwood forest, USA14 and 28 kg N ha−1 year−1/(NH4)2SO43 yearssignificant decline in coverHurd et al. (1998)
Red pine, USA50 and 150 kg N ha−1 year−1/NH4NO37 years80% decline in density/90% decline in biomassRainey et al. (1999)
Hardwood forest, USA35 kg N ha−1 year−1/(NH4)2SO46 yearsno significant responseGilliam et al. (2006)

European research

Although some research on effects of N on the herb layer of European forests has employed experimental additions of N (Nordin et al. 2005), most published work has been based on sampling forest herb communities along gradients of atmospheric deposition of N or by utilizing forest fertilization, which was originally part of management practices to enhance tree growth, as surrogates for experimental treatments (e.g. Brunet et al. 1998; Strengbom et al. 2001, 2003). Furthermore, because excess N deposition in Europe is particularly pronounced as inline image and because nitrification of added inline image acidifies soil (Barber 1995), European studies often have emphasized the general effects of soil acidification on vegetation, in addition to the more specific effects of N (Falkengren-Grerup 1986, 1989; Økland et al. 2004).

Nordin et al. (1998) added N as 15N double-labelled NH4NO3 to the herb layer of a boreal forest in northern Sweden at doses up to 50 kg N ha−1 year−1 and found no short-term response of biomass of species including Vaccinium myrtillus, Deschampsia flexuosa and bryophytes. They did find increases in concentrations of free amino acid N in plant tissue, suggesting a treatment-mediated increase in storage of N. The increase in storage of N has implications for effects of N on other facets of plant health and survival, such as susceptibility to herbivory and pathogenic fungal infection, that will be discussed later.

Brunet et al. (1998) employed multivariate methods to assess potential effects of N deposition on herb layer species of oak forests by sampling along a gradient of N deposition in southern Sweden. Species demonstrating increases in importance (cover) with increased N deposition were generally both nitrophilous and acid-tolerant. They concluded that N deposition influenced vegetation directly by increasing N availability and indirectly by increasing soil acidity.

Schleppi et al. (1999) treated experimental plots in Alptal, Switzerland, with NH4NO3 added at a rate of 30 kg N ha−1 year−1. Although the primary focus of that study was the response of the dominant tree species Norway spruce [Picea abies (L.) Karst] and silver fir (Abies alba Miller) to N additions, they also examined the herb layer and found no significant change in cover or composition following 3 years of treatment. They ascribed the lack of response to the complete dominance of the herb layer by perennial species, allowing N to be stored efficiently over indefinite periods in roots, stolons, bulbs and shoots (Schleppi et al. 1999).

Strengbom et al. (2001) used a creative approach to assess the response of herb layer species to increased N inputs by taking advantage of two separate forest fertilization experiments in northern Sweden. In one experiment, initiated in 1971, N had been added as NH4NO3 granules at varying rates up to 108 kg N ha−1 year−1, with that particular treatment having been terminated in 1990, 9 years prior to their study. The authors carried out point-intercept measurements of herb layer plants in all treatment areas, thus allowing assessment of legacy effects of N deposition. Species composition, as measured via principal components analysis, was significantly different between control and treatment areas, and Simpson's diversity was significantly lower on treatment areas 9 years after cessation of N treatments in the first experiment. These responses arose to a great extent from sharp declines in ericaceous species, such as Vmyrtillus and Vvitis-idaea, and increases in the nitrophilous grass, Dflexuosa (Fig. 2). The authors concluded that effects of increased N deposition on the herb layer of managed Swedish forests can be long-lived (Strengbom et al. 2001).

Figure 2.

Change in species abundance (measured as number of hits on a line transect) of dominant herbaceous layer species of Swedish boreal forests (Vaccinium vitis-idaea and Deschampsia flexuosa) in response to experimental nitrogen (N) additions 9 years following cessation of treatment, which were the following: N1 = 34 kg N ha−1 year−1, N2 = 62 kg N ha−1 year−1, N3 = 108 kg N ha−1 year−1. Figure drawn from data taken from Strengbom et al. (2001).

Strengbom et al. (2003) conducted a synoptic survey of presence or absence of these same species (V. myrtillus, V. vitis-idaea and D. flexuosa) within 557 boreal forest stands throughout Sweden. Using results of an N-deposition model (MATCH –Langner et al. 1995), they divided Sweden into five classes of N deposition from < 3 to > 12 kg N ha−1 year−1. The ericaceous species V. myrtillus and V. vitis-idaea, both regarded as keystone species of the herb layer of forests throughout Sweden (Niemeläet al. 1982), declined along the gradient, even at the lower rates of N deposition. In contrast to increases from experimentally applied N reported in Strengbom et al. (2001), D. flexuosa exhibited no significant pattern in response to the deposition gradient. They concluded that, given the importance of the two species negatively affected by increases in N, N deposition has contributed to a restructuring of the herb layer of much of the boreal forest of Sweden (Strengbom et al. 2003). In a review of many of these studies throughout Sweden, Nordin et al. (2005) echoed these conclusions, adding that small but chronic increases can cause profound changes in composition and structure of herb layer communities.

Despite the statistical limitations inherent in gradient studies (e.g. lack of true controls), these studies have the advantage of reflecting the actual conditions of N deposition to which European forests are exposed. Although there is some understandable variation among forest types, vegetation scientists in Europe have clearly demonstrated that herb layer communities are sensitive to the range of anthropogenically elevated levels of N deposition. Furthermore, the gradient of N deposition is clearly related to spatial patterns of human populations across the European landscape.

North American research

Studies examining the response of the herbaceous layer to increases in N availability in North America generally have not employed the gradient analysis so common in Europe. Studies in North America more often use experimental additions of various forms of available N [e.g. NH4NO3, (NH4)2SO4, HNO3]. In addition to determining effects of N on species composition and diversity of the herb layer, these studies also include effects on nutrient uptake (e.g. N and cations).

Beginning in 1989, N has been added to an entire watershed at FEF, West Virginia. Treatments are administered via three aerial applications (via helicopter) of (NH4)2SO4 per year. March and November applications are made at approximately 7 kg N ha−1, and July applications are approximately 21 kg N ha−1, for a total of 35 kg N ha−1 year−1. Gilliam et al. (1994) reported results for the first year of this study, concluding that there was no response of the herbaceous layer following 3 years of treatment, including total herb cover and biomass, species richness and species diversity. In addition, there were no significant changes in tissue concentrations of plant nutrients for the herb layer as a whole (i.e. pooled across all species). However, working on a subset of sample plots used by Gilliam et al. (1994) and focusing on a single species –Viola rotundifolia– that occurred on all plots of the subset, Gilliam et al. (1996) found significantly higher foliar N in V. rotundifolia from the treatment watershed than in untreated control watersheds, accompanied by lower foliar Ca2+ and Mg2+, in response to 4 years of treatment. They suggested that the N additions to the treatment watershed had simultaneously increased N availability and decreased nutrient cation availability to herb layer species (Gilliam et al. 1996).

Hurd et al. (1998) added N in two forms at varying rates at three sites in the Adirondack Mountains (Adirondack Park, New York). N was added biweekly as dissolved HNO3 spray at the rate of 14 kg N ha−1 year−1 (2× ambient), whereas N was added as single applications in May of crystalline (NH4)2SO4 at 14 and 28 kg N ha−1 year−1 (2× and 4× ambient). They found that cover of prominent herbaceous species, Oxalis acetosella, Maianthemum canadense and Huperzia lucidula, in hardwood forests declined significantly after only 3 years of treatment, particularly in response to additions of (NH4)2SO4. This decline appeared to have resulted from increased shading by the ferns, such as Dryopteris intermedia and Dennstaedtia punctilobula, which can reduce light levels to the forest floor at peak growing season by as much as 70% (George & Bazzaz 2003). The observed response to N additions was more pronounced at sites experiencing lower inputs of atmospheric N.

Rainey et al. (1999) added N as NH4NO3 at rates of 50 and 150 kg N ha−1 year−1 to the forest floor of red pine stands of the Harvard Forest of western Massachusetts. Treatments began in 1988 with fertilizations applied as NH4NO3 solutions in six equal monthly applications. In 1991 and 1992 15N-labelled NH4NO3 was used in a tracer study on the low N (50 kg N ha−1 year−1) treatment (Nadelhoffer et al. 1999). Recoveries of labelled N were low in more prominent herb layer species, namely Mcanadense and Trillium borealis. However, N concentrations in these species were significantly higher in treatment vs. control plots, whereas cation concentrations were generally lower in treatment plots. This supported the earlier findings of Gilliam et al. (1996) at FEF that increased nitrification in forest soils under N-saturated conditions can decrease availability of nutrient cations, especially Mg2+ and Ca2+. Following 7 years of N additions at Harvard Forest, density and biomass of herb layer species had declined by 80% and ∼90%, respectively. This was especially pronounced for the dominant species, M. canadense, which decreased in density from 216 in control plots to 44 and 14 stems m−2 in low N and high N treatments, respectively. Similarly, biomass of M. canadense decreased from 9 g m−2 in control plots to 1.5–0.5 g m−2, respectively (Rainey et al. 1999).

Thus, both Hurd et al. (1998) and Rainey et al. (1999) found rapid responses (i.e. as short as 3 years following N treatments) of species of the herbaceous layer to experimental additions of N in hardwood forests of the eastern USA. In sharp contrast, Gilliam et al. (2006) concluded that 6 years of aerial applications of N to a central Appalachian hardwood-dominated watershed at FEF had no significant effects on the herb layer. This latter study essentially extended by an additional 3 years the sample period initially reported in Gilliam et al. (1994). They examined herb cover, composition and several diversity indicators, including the Shannon–Wiener index, species richness and species evenness, none of which was significantly different between treatment and control watersheds for any of the years sampled, 1991, 1992 and 1994. Consistent with the conclusion of Hurd et al. (1998) that sensitivity of the herb layer to N additions was largely a function of ambient N deposition, Gilliam et al. (2006) interpreted the lack of observed response as arising from the extremely high levels of N deposition at FEF – as much as 20 kg N ha−1 year−1 for total (wet plus dry) deposition (Gilliam & Adams 1996) – and that watershed soils were N-saturated prior to initiation of N treatments. Interestingly, meta-analysis of decomposition studies similarly revealed differing effects of N deposition on litter decomposition in relationship to ambient N. That is, experimental additions of N were found to stimulate decomposition at field sites with low ambient N deposition (< 5 kg N ha−1 year−1), but inhibit decomposition at sites with high ambient N deposition (Knorr et al. 2005).

Specific effects of N on herb layer species – mechanisms of response

As suggested by these studies, the general patterns of response of the forest herbaceous layer to increasing N often, though not always, include the following: (i) initial increases in cover of the herb layer as a whole; (ii) decreases in species richness from loss of relatively numerous N-efficient species; (iii) decreases in species evenness from increasing dominance of relatively few high N-requiring species and (iv) declines in forest biodiversity as a result of decreases in both richness and evenness. In addition, these studies suggest that the response time of herb layer species to increases in N availability is inversely related to ambient N deposition. Thus, a rapid response should occur in areas receiving low levels of atmospheric deposition of N, such as that found for Hurd et al. (1998), whereas a more delayed response would be expected in areas receiving high N deposition, such as that found by Gilliam et al. (2006). Finally, explicit (and implicit) in many, if not most, of these studies is that increasing N supply enhances inline image production and mobility, decreasing availability of essential soil cations.

Although some plants display predictable responses to N additions (Pennings et al. 2005), the response of plant species of the herbaceous layer of forest ecosystems to excess N availability can arise from several mechanisms. Suding et al. (2005) used a meta-analysis of the responses of over 900 plant species in 34 N-fertilization experiments carried out in herb-dominated ecosystems throughout the USA to examine mechanisms to explain losses in species diversity from N fertilization. They distinguished between functional-based and abundance-based mechanisms, where functional-based mechanisms operate at the level of plant functional groups (e.g. life history, photosynthetic pathway) and abundance-based mechanisms operate on an individual species’ abundance (i.e. from rare to common), independent of functional type, wherein locally rare species may be more susceptible to local extinction from N treatment.

Here I take a different approach, and have identified six processes that have been shown both to respond to varying levels of N availability and to influence composition and diversity of forest herb communities. These are interspecific competition, herbivory, mycorrhizal infection, pathogenic fungal infection, species invasions and exotic earthworm activity. It is important to note that these are not mutually exclusive and that some may indeed interact synergistically to explain herb layer response to increasing N availability. For example, increasing N may increase foliar N concentrations, enhancing herbivory, which in turn may increase the likelihood of infection by pathogenic fungi. Furthermore, others may, in some specific instances, be inseparable in their response to increasing N, i.e. changes in one of these processes may result from the response of another one. For example, increasing N may alter the outcome of interspecific competition indirectly because of its more direct effect on mycorrhizal infection. Nonetheless, I address these individually to assess their potential importance in determining the response of the forest herb layer to increasing N availability. The last mechanism – earthworm activity – is not a phenomenon that directly responds to N deposition. I include it here, however, because (i) there is often a close spatial correlation of N deposition and earthworm activity linked by high human population densities (e.g. north-eastern USA) and (ii) recent studies have demonstrated a profound influence of exotic earthworms on the herb layer of forests via alteration of soil properties. The general responses of these processes to excess N, as well as the relevance of those responses to the herb layer, are summarized in Fig. 3.

Figure 3.

Conceptual model of the direct effects of excess N on the herbaceous layer and soil N in forest ecosystems.

interspecific competition

There is little consensus as to mechanisms explaining outcomes of interspecific competition (Aerts 1999). However, regardless of the actual mechanisms involved, most of the results of N additions to herb-dominated plant communities, such as those begun in Rothamsted in 1856 and that depicted in Fig. 1, have been suggested to arise from N-mediated changes in success of dominant species. For example, Wedin & Tilman (1993) found that Schizachyrium scoparium (an N-efficient species) out-competed Poa pratensis and Agropyron scabra (high N-requiring species) under conditions without added N, and did so in large part by taking up inorganic N in a way that decreased its availability to the high-N species. This supported what they called R*, the resource reduction model for competition (Wedin & Tilman 1993).

The dynamics of the herbaceous layer of forest ecosystems sharply contrast with those of herb-dominated systems in ways that are relevant to understanding the nature of interspecific competition and its response to increasing N availability. Although nutrients are certainly important for herb layer species, light is generally the primary limiting factor for understorey species of closed-canopied forests (Coomes & Grubb 2000; Neufeld & Young 2003). Furthermore, light directly reaching the forest understorey does so in discrete patches called ‘sunflecks’. Such a light environment was described by Gilliam & Roberts (2003c) as a ‘dynamic mosaic’, to emphasize the variation in size and distribution of sunflecks at temporal scales from diurnal to seasonal and even successional (Brown & Parker 1994). Availability of nutrients such as N can display a great deal of patchiness in which plant roots not only compete interspecifically but also with soil microbes (Hodge et al. 2000). Although the degree of temporal variability of nutrients is generally much less than that of light, the typically high spatial heterogeneity of soil resources contributes measurably to species diversity of the herb layer (Beatty 2003).

It is important to note that the herbaceous layer in the broadest sense includes both herbaceous and woody species, which Gilliam & Roberts (2003b) referred to as ‘resident’ species (e.g. annual, biennial and perennial herbs that generally remain ≤ 1 m in height) and ‘transient’ species (e.g. seedlings and saplings of trees that have the ability to exceed this height), respectively, to highlight differences among species in time spent in this lowest vegetation stratum of forest communities. Tree seedlings, in particular, exhibit a wide range of tolerance among species to the spatially and temporally dynamic light environment of the forest floor (Kobe et al. 1995). Thus, the success of transient species to emerge from the herb stratum depends to a large extent on competitive interactions with resident species.

Both light and N have been shown to limit herb layer species synergistically. Based on an extensive review, Coomes & Grubb (2000) concluded that light alone limits plant growth under high moisture/nutrient conditions. In support of this observation, Anderson (2003) showed that N uptake by a variety of herb species increased with N supply, but did so to a greater degree in unshaded vs. shaded conditions. The initial response of the herbaceous layer to increases in deposition of N, then, would be to lessen limitation by N and increase limitation by light. Eventually, however, further excess N would have a direct effect on competitive abilities of herb layer species. The resource reduction model for competition (Wedin & Tilman 1993) predicts that plants gain competitive superiority by utilizing essential resources in such a way that decreases availability of those resources to other plants. Thus, chronic inputs of high N to a forest would give a competitive disadvantage to N-efficient plants (e.g. Schizachyrium scopariumFig. 1) because they are unable to draw down available soil N that is constantly re-supplied via atmospheric deposition.

Admittedly, the high light and low N conditions at such herb-dominated communities (e.g. Cedar Creek Natural History Area) are in sharp contrast to conditions typical of a forest understorey (Gilliam & Roberts 2003c). However, Price & Morgan (2006) demonstrated similar results for a herb-rich forest in Australia. They concluded that increased soil N (along with phosphorus and potassium as N–P–K fertilizer) increased competitive exclusion of species adapted to low N availability, resulting in declines in species initially dominant prior to treatment.

herbivory

Consumption of plant material, whether by vertebrate or invertebrate herbivores, is one of several disturbances that can have a profound impact on the herbaceous layer of forests (Roberts & Gilliam 2003; Roberts 2004). Furthermore, herbivory has been shown to interact with nutrient availability to regulate species composition of the herb layer in some forest ecosystems (John & Turkington 1995). The degree to which herbivores consume plant material for their energy and nutritional demands is a function of several factors, the more important of which involve the nutritional value and palatability of plant tissue. Bison, for example, selectively graze in burned tallgrass prairie, even though unburned areas are of equal access, largely because leaves of dominant prairie grasses are higher in N and lower in unpalatable compounds in burned compared with unburned prairie (Knapp et al. 1999).

Turkington et al. (2002) summarized results of a 10-year study on effects of N fertilization on herbivory in herb communities of boreal forests Canada. Their primary impetus was to test hypotheses of ‘bottom-up control’ of vegetation composition, wherein resource (e.g. N) availability is the primary determinant, vs. ‘top-down control’, wherein herbivores themselves are predominant. They concluded that soil resources, especially N availability, were more important in determining herb community composition, and that herbivore activity was more of a response variable, rather than an additional driver, in these forests.

Herb layer species in forests receiving chronically elevated N deposition will undergo luxury uptake of N (Muller 2003), increasing foliar N concentrations and creating the potential to alter herbivore activity. Hartley & Mitchell (2005) found that additions of N increased grazing consumption of species adapted to low-N soils and did so in a way that shifted herb community composition. Anderson (2003) found that most herb layer plants receiving experimental N additions were more susceptible to vertebrate herbivory than those in control plots, but that this was not consistent for all species, suggesting a species-specific response.

Throop & Lerdau (2004) provided perhaps the most extensive review of work on the effects of N deposition on insect herbivory. They concluded that, despite finding an expected degree of variability among studies, the most probable mechanisms to explain such effects were related to deposition-mediated changes in quality and availability of forage plant tissue, especially those related to tissue chemistry, including increases in amino acid (e.g. glutamine and arginine) content and simultaneous decreases in carbon (C)-based defensive secondary compounds (e.g. condensed tannins and phenolics). Using the CENTURY model (Parton et al. 1988), Throop et al. (2004) demonstrated that the response of herbivory to increased N can have major impacts on ecosystem-level processes, such as C cycling. Nordin et al. (1998) demonstrated similar results for low-growing ericaceous shrubs in Sweden, wherein Lepidoptera larvae were found to increase their consumption of foliar material with greater application rates of N.

May & Killingbeck (1992) demonstrated that simulated defoliation measurably decreases plant fitness and concluded that this occurs primarily because herbivory decreases the host plant's ability to reclaim essential nutrients (i.e. nutrient resorption), thus interfering with nutrient dynamics within the plant. Wise & Abrahamson (2005) further showed that herbivory-mediated reductions in plant fitness were more sensitive under conditions of high N availability. This suggests that herb layer species in forests receiving high levels of N deposition may suffer simultaneous reductions in biomass and fitness as a result of N-induced increases in herbivory. Ultimately, N-mediated increases in herbivory should lead to pronounced shifts in species dominance of the herb layer via preferential consumption by herbivores of those plants that demonstrate greatest sensitivity to increased N availability.

Although less investigated than what has been presented here, two other facets of herbivory merit mention in the context of excess N-mediated responses: below-ground herbivory and higher trophic levels. Stevens & Jones (2006) demonstrated that, although fine roots (< 1.0 mm diameter) responded sensitively to fertilizer treatments, root herbivore activity also responded to such treatments, suggesting that excess N not only stimulates above-ground, but also below-ground, herbivory. Focusing on C:N stoichiometry in plant and animal tissues (i.e. C:Nplants >> C:Nherbivores > C:Npredators), Denno & Fagan (2003) provided evidence that N limitation may promote omnivory among carnivorous arthropods. This suggests that excess N could have impacts on higher trophic levels of forest ecosystems.

mycorrhizae

Mycorrhizal associations compose a well-known mutualistic relationship between vascular plant root and fungal symbionts, and are quite common among species of the herbaceous layer of forests (Whigham 2004). Among the considerable benefits conferred to the plant by the fungus, including increased access to nutrients, protection against pathogens and enhanced water-use efficiency (Smith & Read 1997), mycorrhizal plants are particularly superior competitors compared with non-mycorrhizal plants in nutrient-limited soils, largely because of their fungal-enhanced abilities to take up nutrients, such as N (Read & Perez-Moreno 2003).

One particular type of endomycorrhizae – ericoid mycorrhizae – is associated with ericaceous plants, which are especially well adapted to characteristics of many older forest soils – highly acidic, weathered and infertile (Read 1996; Perotto et al. 2002). Ericoid mycorrhizae also allow host plants to take up amino acids and complex nitrogenous organic compounds (Grelet et al. 2005). These fungal symbionts, however, exhibit notable sensitivity to increases in N deposition. Yesmin et al. (1996) found that ericoid mycorrhizal infection declined significantly along an N deposition gradient, from 84.5 ± 3.1% at 2 kg N ha−1 year−1 to 14.5 ± 2.6% at 10 kg N ha−1 year−1.

Wallenda & Kottke (1998) reviewed the literature on field experiments investigating the effects of added N on ectomycorrhizae and found that there was little effect of N on below-ground populations. They did, however, find measurable shifts in above-ground (i.e. sporocarp) communities, as did Lilleskov et al. (2002), generally concluding that increased N decreases ectomycorrhizal diversity. Similarly, using a combination of N deposition gradient and glasshouse studies, Dighton et al. (2004) found that ectomycorrhizae declined significantly with increased N in soils from New Jersey pine barren forests. Other studies have found less notable effects of increasing N on ectomycorrhizae (Lilleskov & Bruns 2001).

Synthesizing the extensive results of the NITREX network in Europe, Brandrud & Timmermann (1998) envisaged the response of mycorrhizae to N enrichment as occurring in three phases. In Phase 1 there is a rapid decrease in above-ground diversity and sporocarp production by mycorrhizal fungi; below-ground mycorrhizae and fine-roots of host species are usually not affected. By Phase 2, generally sensitive fungi disappear, with below-ground structures variably affected. Finally, Phase 3 occurs with an essentially complete disappearance of sporocarp production, along with sharp declines in density of below-ground mycorrhizae and fine-roots of host species.

Although it lies largely beyond the scope of this review, Read & Perez-Moreno (2003) developed an ecosystem gradient model relating changes in mycorrhizal communities to changes in N form and availability in terrestrial ecosystems that is relevant to this discussion. They envisaged a gradient from (1) soils with little N mineralization and no nitrification dominated by organic N cycling and plants commonly supporting ericoid and ectomycorrhizal fungi with well-developed saprotrophic capabilities, to (2) soils with limited N mineralization dominated by a more diverse assemblage of plants supporting ericoid, ecto- and arbuscular mycorrhizae, to (3) soils with high nitrification potential and dominated by inline image availability and plants supporting arbuscular mycorrhizae with limited or no saprotrophic capabilities (Read & Perez-Moreno 2003). This gradient is consistent with findings of Eaton & Ayres (2002), who demonstrated that early-successional mycorrhizae were characterized by having low N mineralization efficiency, whereas late-successional fungal species exhibited high mineralization efficiency. Forests experiencing N saturation from excess N deposition can quickly change from (2) to (3), with one of the immediate impacts being a pronounced shift in species composition of the herbaceous layer. Superimposed on this scenario for forests of eastern North America is the paucity of old-growth forests (McCarthy 2003). Thus, forest of extremely late-successional stages, as envisaged by Eaton & Ayres (2002), are relatively rare in this region.

disease

Most work carried out on the effects of excess N on the frequency and occurrence of plant diseases, primarily pathogenic fungal infections, has been carried out in Swedish forests (Nordin et al. 1998, 2006; Strengbom et al. 2001, 2002, 2003, 2006). The herbaceous layer of these forests is generally dominated by low-growing ericaceous species of the genus Vaccinium. As previously discussed, these species are known to decline with increasing N loading as the result of the elimination of their fungal symbionts – ericoid mycorrhizae (Read & Perez-Moreno 2003). However, they also are impacted by excess N through significant and substantial increases in the rate of fungal infections. Strengbom et al. (2003) found this response by sampling Swedish forests along a gradient of N deposition. Similar patterns were found using experimental additions of N (Nordin et al. 1998; Strengbom et al. 2006). Indeed, Strengbom et al. (2002) found that increases in N-enhanced pathogenic fungal infections can be rapid and can further result in major shifts in herb layer species dominance within 3 years of treatment.

N-mediated changes in pathogenic fungal infections have also been shown to covary with other factors, both abiotic and biotic. Strengbom et al. (2006) found that individuals of host plant Vaccinium myrtillus were proportionally more susceptible to the parasitic fungus Valdensia heterodoxa when exposed to 12.5 and 50 kg N ha−1 year−1 added as granular NH4NO3 compared with untreated control plots. They also found that the incidence of disease was significantly and positively correlated with precipitation, with the higher summer precipitation enhancing the effects of added N. Nordin et al. (2006) demonstrated the contrast between vascular plant response vs. pathogenic response to added N in boreal forests. Plant species discriminated between the form of added N:N-efficient species exhibited no response, high-N-requiring species responded mostly to inline image and bryophytes responded to inline image. By contrast, occurrence of pathogenic fungi increased in response to all additions of N, regardless of the form in which it was added.

Mitchell et al. (2003) considered the additional interactions of elevated CO2 and decreased plant species diversity with the effects of excess N on pathogenic fungal infection of several herbaceous species. They developed a metric called ‘pathogen load’, which they defined as the relative (%) leaf area infected by fungi, to quantify these responses. They found that all three variables (elevated CO2, decreased diversity and increased N) increased pathogen load on several herb species, in some cases interactively. The response to N was particularly pronounced for N-efficient species, such as C4 grasses. The most profound response of pathogen load was not to N, but to diversity. Plots with one or four species had greater than twice the pathogen load as plots with nine or 16 species. This finding has important relevance to long-term, though indirect, effects of N on herb layer communities. That is, N-mediated declines in herb diversity may make forest herb communities more susceptible to pathogenic fungi.

invasive species

Invasion of non-native species in undisturbed forests with closed, intact canopies is generally rare (Luken 2003). Recent work, however, has shown that species invasions can be facilitated by increasing N supply. Although most work has been done in non-forest ecosystems such as grasslands, shrublands and deserts (Padgett & Allen 1999; Brooks 2003; Fenn et al. 2003), recent studies suggest that N limits the invasibility of the herbaceous layer of both conifer and hardwood forests (Cassidy et al. 2004). Thus, N-saturated forests would be particularly susceptible to introduction of exotic species.

Much of the observed, excess N-mediated increase in invasibility of forest herb layer communities appears to be related to one of the hallmark symptoms of N saturation: increases in net nitrification, resulting in increased availability and mobility of inline image in impacted soils. Fargione & Tilman (2005) found that increases in biomass of invasive species were significantly related to increasing soil inline image. They further concluded that soil inline image was the best predictor of success for species that invaded experimental monocultures.

Windham (2002) presented a compelling mechanism to explain this response. Because much of the N deposited from the atmosphere occurs as inline image, she undertook a comparative study to examine variation in the ability of native vs. exotic species to utilize inline image, measured as levels of nitrate reductase activity (NRA) in leaves and roots (Muller & Janiesch 1993), across a gradient in inline image deposition across North America and in southern Chile. Results demonstrated notably higher use of inline image by exotic species, as evidenced in ranges of potential NRA: 0–186 nmol inline image g−1 h−1 for native species and 145–1291 nmol inline image g−1 h−1 for exotic species. These results were consistent with earlier work by Kourtev et al. (1999), who found that NRA was generally higher in the invasive species Berberis thunbergii and Microstegium vimineum than in native species at adjacent sites in hardwood forests of northern New Jersey. Windham (2002) suggested that inline image deposition creates a ‘low light/high nitrate’ niche in forests that favours exotic herb layer species which are better adapted to produce NRA more quickly than native species. Accordingly, invasive exotic species are generally better able to take advantage of higher soil inline image, and do so at the expense of native species.

Furthermore, it is quite likely that positive feedbacks can develop to further exacerbate increases in invasions of forest herbaceous layers by exotic species that are caused by excess soil inline image. Ehrenfeld et al. (2001) examined invasion of the herb layer of hardwood stands of northern New Jersey by the two exotic species studied by Kourtev et al. (1999): B. thunbergii and M. vimineum. They found that soil in which these species grew had higher pH and net N mineralization and nitrification than soil occupied by native species, results which were found both in the field and under the more controlled conditions of the glasshouse. Higher pH in impacted soil was probably the result of preferential uptake of inline image by the exotic species (Marschner 1995; Ehrenfeld et al. 2001). In a review of studies of this phenomenon that included 56 invasive species, Ehrenfeld (2003) concluded that most invasives had the ability to alter the N cycle of forest soils drastically, particularly by increasing N availability, altering rates of N fixation, and producing litter of lower C : N ratios that decomposes more rapidly than that of co-occurring native species. Thus, a dangerous feedback can develop when (i) excess N deposition increases N availability in forest soils that (ii) enhances the success of exotic species invading the herbaceous layer that (iii) further alter N dynamics of forest soil by maintaining high levels of available inline image that (iv) simultaneously gives a competitive advantage to invasive species themselves and (v) a disadvantage to native species. The result is a further drastic shift in species composition and decrease in biodiversity of the herb layer.

earthworm activity

There is no apparent causal relationship between excess N deposition and introduction and activity of exotic earthworm species. Even so, Kourtev et al. (1999) found that introduced (European) earthworm populations were significantly and substantially higher in soil that supported invasive species. Thus, enhanced activity of exotic earthworms can add to the positive feedback mechanism just described. In addition, exotic earthworms have been shown to decrease colonization rates of mycorrhizae in forests of north-east USA (Lawrence et al. 2003). Activity of exotic earthworms also is often spatially correlated with excess N because both often arise from human activity (Gundale et al. 2005), especially in areas of high densities of human populations.

Considering the effect that earthworm activity can have on herbaceous communities (Bohlen et al. 2004), such a spatial coincidence can confound the direct effects of excess N with those of earthworms. Bohlen et al. (2004) provided stunning photographic evidence of the effects of exotic earthworms in forests of northern Minnesota. Among the apparent effects were: general lack of herb layer cover, areas of exposed mineral soil from a greatly reduced litter layer, and exposed roots of canopy trees from disappearance of forest floor. Gundale (2002) surveyed 28 populations of the rare fern Botrychium mormo and found that local extirpation of this species was directly associated with the introduced earthworm Lumbricus rubellus.

Much of the change in the herbaceous layer caused by exotic earthworms results from worm-mediated alterations in soil nutrient dynamics. Gundale (2002) found thinner O horizons in soils occupied by Lrubellus and suggested that the associated decline in B. mormo was related to the dependence of B. mormo on nutrients in the O horizon. Hale et al. (2005) examined soil characteristics of northern hardwood forests of Minnesota that had been invaded by European earthworms. Because glaciated soils such as those they studied lack native earthworm populations, they were able to assess effects directly by comparing invaded vs. non-invaded sites. They found that earthworm activity rapidly decreased the O horizon, but increased thickness, bulk density and total organic matter content of the A horizon; percentage organic matter and fine-root density of the A horizon decreased (Hale et al. 2005).

In addition to these indirect effects of exotic earthworms on herb communities via alteration in soil conditions, other effects can be more direct. For example, herbaceous layer species can experience direct mortality when earthworms consume forest floor from beneath them (James & Cunningham 1989). Herb layer species may also suffer reduced reproduction and survival from earthworm activity via consumption and deep burial of seeds, alteration of germination microclimate, and increased susceptibility to vertebrate herbivory (Bohlen et al. 2004).

Clearly, excess N deposition does not directly increase the introduction and activity of non-native earthworm species in forest ecosystems, and vice versa. However, these two anthropogenic disturbances display a high enough degree of spatial coincidence, and indeed their effects are often quite similar enough, that their impacts on the herbaceous layer must be considered in many instances to occur in tandem.

The nitrogen homogeneity hypothesis: a general hypothesis of herb response to excess N in forest ecosystems

This final section will be used to develop a general hypothesis of the negative effects of excess N on forests, one which has particular relevance to the herbaceous layer. Biodiversity can be seen as more than just the number of different species occurring in an area. It also has a functional process component (Roberts & Gilliam 1995). Thus, spatial heterogeneity in processes such as nutrient cycling contributes to the overall biodiversity of an ecosystem. This hypothesis will make a connection between excess N-mediated decreases in spatial heterogeneity of soil N processing and decreases in species diversity of impacted forests.

Spatial heterogeneity in essential resources contributes to maintenance of plant species diversity, although the degree to which this operates is still open to considerable debate (see Stevens & Carson 2002 for a recent discussion). Hutchings et al. (2003) argued that the spatial extent of patches (e.g. of high vs. low N) relative to the spatial extent of root systems of plants in the patches determines how sensitive this relationship is. They concluded that spatial heterogeneity of soil resources, such as N, plays an important role in influencing structure and composition of plant communities.

Intuitively, however, it is clear that increasing heterogeneity (i.e. patchiness) in availability of an essential nutrient, such as N, in forest soils would give rise to coexistence of greater numbers of species through minimizing or even obviating competitive exclusion. This is especially valid for herb layer communities, wherein the spatial distribution of species is influenced greatly by spatial heterogeneity of soil nutrient availability (Beatty 2003; Small & McCarthy 2003; Whigham 2004).

Spatial heterogeneity of N processing (e.g. N mineralization, nitrification) is typically high in forest soils (Laverman et al. 2002), the result of several factors, including effects of spatial variability in chemistry of tree litter (Aubert et al. 2005). However, watershed-scale research at the FEF has shown that one of the effects of excess N inputs on terrestrial ecosystems is to decrease the high ambient degree of spatial heterogeneity in soil N dynamics, including both soil and plant processing of N (Gilliam et al. 2001, 2005). Although not statistically significant, net nitrification was higher on a watershed (WS3) treated with 35 kg N ha−1 year−1 than two untreated control watersheds (Fig. 4). More important, however, spatial variability in net nitrification (measured as coefficient of variation of means based on seven sample plots within each watershed) was approximately four times higher in the control watersheds than in the treated watershed. Similarly, the available pool of inline image in soils was generally higher in WS3 than in the control watersheds, whereas spatial variability was 3–4 times higher in control watersheds than in WS3 (Fig. 4). Gilliam et al. (2001) reported similar contrasts for spatial variability in foliar N concentrations of herb layer species, litter N and soil N, all of which exhibited greater spatial heterogeneity on the untreated control watersheds.

Figure 4.

Spatial variation in N-treated (WS3) vs. untreated (WS4 and WS7) watersheds at the Fernow Experimental Forest, West Virginia. (a) Shaded bars represent mean net nitrification; means with the same letter are not significantly different at P < 0.05. Open bars are coefficients of variation associated with each mean. (b) Shaded bars represent mean soil inline image pools; means with the same letter are not significantly different at P < 0.05. Open bars are coefficients of variation associated with each mean.

From both a biogeochemical and a plant standpoint, one of the more serious results of N saturation in forest ecosystems is the profound shift in the nature of N cycling from one predominated by processing of organic N to one dominated by the mobility of inline image. Biogeochemically this is important because of the highly mobile, anionic nature of this inorganic form of N. Thus, when inline image is produced in excess of plant and microbial demand, it moves easily though the mineral soil, leaching essential cations, such as Ca2+, with it, decreasing their availability to forest plants. With respect to plant response, this increase in soil inline image availability is important for several reasons. Many of the specific responses of the herbaceous layer to excess N previously discussed – and summarized in Fig. 3– arise directly from increased inline image availability and uptake by plants, including: (i) increasing competitive abilities for a small number of nitrophilous species, (ii) increases in herbivory, (iii) decreases in mutualistic fungi, (iv) increases in pathogenic fungi and (v) enhanced invasibility by exotic species that are highly competitive under conditions of high soil inline image. All of these, alone and in combination, have the potential to cause net reductions in species richness of the herb layer and, ultimately, in the biodiversity of impacted forests.

I suggest that, in addition to these well-documented responses, an additional response be added: that of reducing the naturally high degree of spatial heterogeneity in soil N availability. In other words, not only does N saturation shift the N cycle toward inline image mobility, but it also results in more homogeneous availability of N in the soil. When increased spatial homogeneity in soil N availability is coupled with responses summarized herein (Fig. 3), excess N can be seen to be especially effective in decreasing the species diversity of the herb layer.

Such a response will certainly vary among ecosystem types, with some being more, and other less, sensitive to change. However, given the ubiquitous importance of N in controlling the structure and function of forest ecosystems (Vitousek & Howarth 1991) and the increasingly ubiquitous nature of anthropogenically enhanced N deposition (Galloway et al. 2004), all forests are potentially susceptible to the negative impacts of excess N. The nitrogen homogeneity hypothesis predicts that N saturation ultimately will decrease forest biodiversity (Fig. 3).

It should be clear from this review that whereas much is known about the biogeochemical responses of forests to excess N deposition (Fenn et al. 1998; Aber et al. 2003) and about the ecology of the herbaceous layer (Gilliam & Roberts 2003a; Whigham 2004), far less is known about the dynamics of herb layer communities in N-saturated forests. Certainly, this represents an area in need of further research. For example, among the processes affected by excess N discussed herein, mechanisms for N-mediated invasion of non-native species appear among the least understood, yet species invasions represent one of the more serious threats to forest biodiversity. Thus, future research on the ecology of the herbaceous layer in impacted forests will be at once interesting and essential.

Acknowledgements

I would like to thank the following individuals for their invaluable contributions in improving the quality of this paper: Mary Arthur, Jeffrey May, Nicole Turrill Welch, Dennis Whigham and Lisa Windham. I also thank David Gibson, Editor, Journal of Ecology, for his initial invitation to write this review and Robert Jones, Associate Editor (and two anonymous referees) for making excellent suggestions that greatly improved the quality of this paper.

Ancillary