SPECIAL FEATURE

Underground Processes in Plant Communities1

This set of papers represents the second Special Feature this year offering new insights into interactions in the rhizosphere. From a historical perspective, early ecologists saw the need and importance of underground research. Nearly 25% of the classic plant ecology textbook by John Weaver and Frederic Clements (1929. Plant Ecology. McGraw-Hill, New York, New York, USA) was devoted to soil, roots, and their interactions with other organisms. Although most modern ecologists recognize the important role of underground processes, they nd the rhizosphere difcult to study, with the organisms ranging in size from tiny (microbes) to gigantic (roots). Unfortunately, out of sight usually means out of mind, and strong programs linking the rhizosphere to plant community processes are few and far between. The fact that more than half of the plant biomass in grasslands is underground is not reected in the types of studies most commonly conducted. The bulk of this underground biomass is involved in various symbioses, antagonistic assaults, and mutualisms. For example, fungal root pathogens form some of the largest known living organisms and are extensively studied by applied and mechanistic biologists but are rather infrequently the focus of ecological studies. The authors of this Special Feature make strong arguments for a prominent role of underground dynamics in driving competitive processes between plants, interactions between plants and animals, and community and ecosystem level processes such as succession and nutrient cycling. This set of papers represents a fresh combination of conceptual and empirical approaches that matches the state-of-the-art in ecological research. The opening papers (Bargett and Wardle, van der Putten) introduce the concept of aboveground belowground linkages in plant communities and apply well developed ideas on plantanimal interactions from aboveground to underground processes. For example, although little ecological research has been conducted on the manifold defensive strategies of roots, it is quite likely that roots employ both direct defense by producing toxins and indirect defense by attracting enemies of root herbivores. Root exudates may cause benecial microbes to ourish and nourish plants in times of aboveground defoliation. The authors pose a novel mixture of potential interactions that have consequences extending from physiological to ecosystem ecology. Following the two introductory papers, the Special Feature next focuses on the role of microbes in community processes (Reynolds et al., Klironomos). Although a single milliliter of soil can contain over 100 million individuals of bacteria and other microbes, their specicity in interactions, functional roles, and consequences have been little explored. Mycorrhizal fungi are robbing some plants and feeding others, and we are only beginning to scratch the surface in determining what factors inuence the continuum of parasitic to mutualistic interactions. This gets more and more complicated as we begin to see that some mycorrhizae can act as conduits allowing for the ow of resources between roots of taxonomically distant plants. Of course, the circle of life is not completed until the roots themselves are recycled, and mycorrhizal fungi may play an important role here as well (Langley and Hungate). Indeed, microbes are simultaneously feeding, infecting, and recycling the phyto-ecosystem. The latter part of the Special Feature focuses on plantplant interactions underground and how these may scale up to community and ecosystem processes. Casper et al. mine data from a large database and merge this with experimental results and a model that helps us dene the root neighborhood that may ultimately determine plantplant competition. In addition to the proximity and overlap in roots underground, the foraging and proliferation of roots in heterogeneous soils may determine the levels of competition between plants at the individual level and also may have consequences for populations and communities (Hutchings et al.).
1 Feature accepted 31 October 2002. Reprints of this 79-page Special Feature are available for $12.00 each, either as pdf les or as hard copy. Prepayment is required. Order reprints from the Ecological Society of America, Attention: Reprint Department, 1707 H Street, Suite 400, Washington, D.C. 20006.

2256

Rhizo-ecologists have already been successful in unraveling some of the poorly understood complexities of real communities. The authors of the papers in this Special Feature have now challenged ecologists to come to further grips with how a variety of both aboveground and belowground organisms, from arthropods and microbes to interacting plants themselves, interact via the rhizosphere. Although these processes will remain out of sight, they will increasingly be studied and revealed. ANURAG AGRAWAL Special Features Editor
Key words: belowground competition; community dynamics and succession; multi-trophic interactions; mycorrhizae; plant defense and herbivory; soil and nutrient heterogeneity.
2003 by the Ecological Society of America

Special Feature

Ecology, 84(9), 2003, pp. 22582268 2003 by the Ecological Society of America

HERBIVORE-MEDIATED LINKAGES BETWEEN ABOVEGROUND AND BELOWGROUND COMMUNITIES

RICHARD D. BARDGETT1,4
1Institute

AND

DAVID A. WARDLE2,3

of Environmental and Natural Sciences, Lancaster University, Lancaster, LA1 4YQ UK 2Landcare Research, P.O. Box 69, Lincoln, New Zealand 3Department of Forest Vegetation Ecology, Swedish University of Agricultural Sciences, SE901-83 Umea, Sweden

Special Feature

Abstract. Understanding how terrestrial ecosystems function requires a combined abovegroundbelowground approach, because of the importance of feedbacks that occur between herbivores, producers, and the decomposer subsystem. In this paper, we identify several mechanisms by which herbivores can indirectly affect decomposer organisms and soil processes through altering the quantity and quality of resources entering the soil. We show that these mechanisms are broadly similar in nature for both foliar and root herbivory, regardless of whether they operate in the short term as a result of physiological responses of individual plants to herbivore attack or long-term following alteration of plant community structure by herbivores and subsequent changes in the quality of litter inputs to soil. We propose that a variety of possible mechanisms is responsible for the idiosyncratic nature of herbivore effects on soil biota and ecosystem function; positive, negative, or neutral effects of herbivory are possible depending upon the balance of these different mechanisms. However, we predict that positive effects of herbivory on soil biota and soil processes are most common in ecosystems of high soil fertility and high consumption rates, whereas negative effects are most common in unproductive ecosystems with low consumption rates. The signicance of multiple-species herbivore communities is also emphasized, and we propose that if resource use complementarity among herbivore species or functional groups leads to greater total consumption of phytomass, and thus greater net herbivory, then both positive and negative consequences of increasing herbivore diversity for belowground properties and processes are theoretically possible. Research priorities are highlighted and include a need for comparative studies of herbivore impacts on above- and belowground processes across ecosystems of varying productivity, as well as a need for experimental testing of the inuence of antiherbivore defense compounds on complex multitrophic interactions in the rhizosphere and the signicance of multiple herbivore species communities on these plantsoil interactions.
Key words: decomposition; ecosystem function; herbivores; multitrophic interactions; nutrient mineralization; plant litter; root herbivores; soil biota; soil fauna.

INTRODUCTION It is increasingly becoming recognized that to understand how terrestrial ecosystems function requires a combined abovegroundbelowground approach, because of the importance of feedbacks that occur between the producer and the decomposer subsystems (Van der Putten et al. 2001, Wardle 2002). These feedbacks between plants and decomposer organisms involve a variety of interactions, but they are also strongly inuenced by herbivores that consume plant material (Bardgett et al. 1998b). Many studies on impacts of herbivores on terrestrial ecosystems have focused exclusively on aboveground responses, such as changes in vegetation diversity and community structure (e.g., Collins et al. 1998). Further, there is currently evidence that herbivores have substantial indirect effects on soil
Manuscript received 3 May 2002; revised 20 July 2002; accepted 11 September 2002; nal version received 7 October 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256. 4 E-mail: r.bardgett@lancaster.ac.uk

organisms and their activities including soil mineralization processes, and this may be expected to have signicant consequences for plant productivity and community structure (Bardgett et al. 1998b). Recent studies also show that root herbivores can exert substantial indirect effects on soil biota and cycling of nutrients, with important aboveground effects (Bardgett et al. 1999a, b, Denton et al. 1999). However, much remains unknown about the roles of root herbivores in regulating soil biological properties, but as with foliar herbivores, their effects on soil biota and processes appear to be related to qualitative and quantitative changes in resource supply to soil resulting from their consumption of plant material. Here we evaluate how both foliar and root herbivory may inuence belowground organisms and processes, and how these changes in soil biological properties may in turn affect aboveground primary production. In doing this, we explore the different mechanisms for such effects over various temporal and spatial scales, ranging from the individual plant to the plant community.

2258

September 2003

UNDERGROUND PROCESSES

2259

Given that several types of herbivores may be simultaneously operating in the same plant community, we then discuss and make predictions on the issue of belowground effects of herbivory in the context of multiple herbivore species communities. The ultimate goal of this article is to predict the main ways that herbivores affect soil biota and processes, and ultimately the functioning of terrestrial ecosystems, and identify some future research questions that are central to furthering our understanding in this eld. PRODUCERDECOMPOSER FEEDBACK MECHANISMS Very little is known about the precise way that soil organisms and processes are affected by foliar and root herbivory. This is because of the existence of a wide variety of mechanisms that operate at both the individual plant and the plant community scale, which can involve either positive or negative effects on both the quantity and quality of resources that enter the decomposer subsystem (Bardgett et al. 1998b). Consistent with this variety of effects, studies that have used fenced exclosure plots to investigate the ecological consequences of grazing and browsing by mammals have found overall effects on decomposer organisms that can be positive (e.g., Stark et al. 2000, Bardgett et al. 2001) or negative (e.g., Suominen 1999), effects on C mineralization that are positive (e.g., Bardgett et al. 1997, Kielland et al. 1997) or negative (e.g., Pastor et al. 1988, Stark et al. 2000), and effects on net N mineralization that are positive (e.g., McNaughton et al. 1997a, Frank and Groffman 1998) or negative (e.g., McNaughton et al. 1988, Stark et al. 2000; reviewed in Wardle 2002). Further, in a study of 30 long-term (1336-yr-old) fenced plots aimed at excluding introduced deer and goats in a range of New Zealand rain forests, Wardle et al. (2001) showed that effects on soil biota, C mineralization, and C and N sequestration were highly idiosyncratic, with roughly equal numbers of sites showing strong positive and strong negative effects. There are no comparable studies looking at the effects of excluding root herbivores on soil biological properties at the eld scale, but we predict that the variety of feeding strategies of root-feeding organisms and the wide range of responses of plants to infestation will yield equally idiosyncratic responses at the ecosystem level. Despite the existence of the wide variety of ways that herbivores can inuence producerdecomposer feedbacks, we identify three key, but interrelated, mechanisms that we believe are consistently important. First, herbivores regulate the quantity of organic matter that is returned to soil through inducing physiological responses at the individual plant level, such as changes in biomass production and resource allocation, ne root dynamics, and root exudation (mechanism 1). Second, herbivores alter the quality of resource inputs to the decomposer subsystem either directly, through the return of feces and urine to soil, or indirectly by altering

concentrations of nutrients and secondary metabolites in tissues, with implications for plant litter quality, decomposition, nutrient supply, and ultimately ecosystem function (mechanism 2). Third, at the plant community scale, the effects of foliar herbivory on soil biota and their functioning occur over much longer time scales through altering the functional composition of vegetation which may either enhance or reduce overall plant litter quality and hence its decomposability (mechanism 3). Herbivores are also likely to inuence soil processes through physical disturbance and alteration of soil abiotic factors such as temperature and moisture status. Here, we focus on the effects of herbivory on producerdecomposer feedbacks that involve changes in resource quality and quantity.

Mechanism 1: alteration of resource quantity

There are two ways in which herbivory, both aboveground and belowground, can alter the quantity of organic matter that is returned to soil. In the short term, the quantity of resources supplied to soil can be altered through effects of herbivory on plant C allocation and root exudation patterns, whereas in the long term, the amount of organic matter returned to soil is affected by herbivore-induced shifts in net primary productivity (NPP). Both these mechanisms are intimately linked and collectively inuence plant productivity at the community level (Fig. 1). In the short term, herbivory alters the quantity of resources that are transferred to soil due to changes in root exudation patterns. Plants allocate large proportions of their assimilated C to root exudation (Bokhari 1977), which may stimulate the growth and activity of heterotrophic microbes in the rhizosphere. Both foliar (e.g., Holland et al. 1996) and root (Yeates et al. 1998) herbivory have been shown to increase root exudation in grassland plants, which in turn stimulates microbial biomass and activity in the rhizosphere (Denton et al. 1999, Hamilton and Frank 2001), and microbe-feeding fauna (Mikola et al. 2001a). These facilitating effects of herbivory on rhizosphere microbes, and presumably higher level consumers in the soil food-web (Mikola et al. 2001a), have been shown to positively feedback on soil N availability, plant N acquisition, leaf N content, and photosynthesis of a grazing tolerant grass (Poa pratensis), ultimately beneting plant productivity in grassland of high soil fertility (Hamilton and Frank 2001). Such benets of these herbivoreplant microbial interactions might not only be restricted to the individual plant that has been subjected to herbivory, but are also likely to benet neighboring plants. This is yet to be demonstrated with regard to foliar herbivory. However, root herbivory of N-xing white clover (Trifolium repens) by clover cyst nematode (Heteroderii trifoli) was shown to not only increase root exudation and microbial activity in the rhizosphere, but also increase the root growth and N content

Special Feature

2260

RICHARD D. BARDGETT AND DAVID A. WARDLE

Ecology, Vol. 84, No. 9

FIG. 1. Schematic of the effects of foliar and root herbivory on producerdecomposer feedbacks that result from changes in the quantity of resources returned to the soil. These mechanisms are most common in nutrient-rich grasslands, where dominant plant species benet from herbivory through positive feedbacks between herbivores, plants, and soil biota, and through preventing colonization by later successional plants which produce poorer litter quality. In unproductive and nutrientpoor ecosystems, compensatory plant growth responses to foliar herbivory will not occur, and declines in NPP caused by foliar herbivory should cause reductions in soil organisms.

Special Feature

of a neighboring grass (Lolium perenne) (Bardgett et al. 1999b). There is evidence that these stimulatory effects of herbivory on rhizosphere processes vary with the intensity of foliar herbivory and with plant species identity (Guitian and Bardgett 2000, Mikola et al. 2001a), with the level of root infestation of the host plant (Denton et al. 1999) and with the identity of the root-feeding organism (Bardgett et al. 1999a). However, on the basis of the above ndings, we might speculate that in grasslands of high soil fertility at least, such responses to herbivory have the potential to stimulate rhizosphere processes that ultimately feedback positively on plant nutrition and photosynthesis, and hence plant productivity at the community level. In the longer term, herbivory inuences the quantity of organic material that is returned to soil, and hence to the decomposer food web, through shifts in NPP. There are many instances where foliar herbivory by large mammals has negatively affected aboveground NPP (Milchunas and Lauenroth 1993). There is also some experimental (e.g., McNaughton 1985) and theoretical (e.g., De Mazancourt et al. 1998) evidence that aboveground grazing enhances aboveground NPP at the community level, especially in grasslands of high soil fertility that are grazed at intermediate intensity. Direct linkages between grazer-induced stimulation of aboveground NPP and soil food webs have not been made. However, in a eld study of gradients of sheep (Ovis aries) grazing intensity on British hill grasslands, Bardgett et al. (2001) reported that soil microbial biomass was maximal at intermediate levels of grazing

intensity, indicating that compensatory responses to intermediate levels of herbivory were probably related, in part, to enhanced circulation of nutrients by soil microbes (De Mazancourt et al. 1998) and are probably intimately linked to the stimulation of rhizosphere processes by herbivore-induced root exudation (Bardgett et al. 1998b), as well as other direct plant physiological responses to herbivory. Similarly, there is evidence that moderate densities of invertebrate herbivory (i.e., grasshoppers) can promote NPP through enhancing nitrogen cycling (Belovsky and Slade 2000). It is of note that herbivory generally has an adverse effect on mycorrhizal associations (Gehring and Whitham 1994), suggesting that the optimization of decomposer organisms at intermediate intensities of herbivory does not usually apply to mutualists that are intimately associated with plant roots. Further, these patterns of optimization of NPP by herbivores that are linked to the stimulation of rhizosphere processes appear to be especially common in productive grassland sites, where they typically reinforce conditions of high soil fertility. They do not appear to be widespread in slower growing forest ecosystems of relatively low soil fertility; Wardle et al. (2001) found no evidence of grazing optimization of plant biomass at any of 30 fenced exclosure plots in New Zealand forests. There is mixed evidence about how foliar herbivory affects root productivity. Pot experiments consistently show that repeated defoliation reduces root biomass (e.g., Guitian and Bardgett 2000, Mikola et al. 2001a), and eld studies have shown that moose and hare browsing reduces ne root productivity in Alaskan Tai-

September 2003

UNDERGROUND PROCESSES

2261

ga forest (Ruess et al. 1998). However, exclusion studies on Serengeti grasslands show that mammalian grazers do not necessarily inhibit root biomass and productivity (McNaughton et al. 1998), and in a global literature synthesis Milchunas and Lauenroth (1993) reported both enhancements and reductions in root biomass as a result of herbivore exclusion. There are numerous reports of root herbivores negatively affecting NPP in a range of ecosystems (e.g., Brown and Gange 1990, Ingham and Detling 1990), and also incidences where infestation of roots by insects has led to proliferation in the growth of lateral roots (Brown and Gange 1990). However, examining the direct effects of root herbivores on NPP in the eld is extremely difcult since plants are frequently damaged simultaneously by foliar and root herbivores (Muller-Scharer and Brown 1995). Root feeding organisms may also indirectly affect NPP by increasing the susceptibility of plants to soil-borne pathogenic microorganisms (Van der Putten et al. 2001), and mycorrhizal fungi might also be involved in these multitrophic interactions, for example by competing with plant pathogenic fungi or through plant defense (Van der Putten et al. 2001). The knock-on effects of herbivore-induced changes in NPP to the soil food web and nutrient cycling are not at all clear. For example, increasing NPP has been shown to have both positive and negative effects on both microbial biomass and higher trophic levels of the soil food web (reviewed in Wardle 2002). There are two reasons for idiosyncratic responses of decomposers to NPP. First, the relative importance of top-down and bottom-up forces in regulating soil food web components may be context dependent. Second, plants not only provide carbon resources for microbes but also compete with them for nutrients (Kaye and Hart 1997). Therefore, the direction of effects of herbivore-induced changes in NPP on decomposer organisms and nutrient cycling may be governed by which of two opposing effects (stimulation of microbes by carbon addition, inhibition of microbes by resource depletion) dominates. However, despite this uncertainty, we propose that positive effects of herbivory on soil biota and nutrient mineralization dominate in sites of high soil fertility, especially in grasslands where carbon addition to soil from plants typically stimulates soil biological activity since it alleviates carbon limitation of microbes.

Mechanism 2: changes in resource quality Dung and urine return.Ungulate waste and its facilitative effect on soil biological processes is commonly promoted as one of the main driving mechanisms for grazers stimulating N availability and plant nutrient uptake at the plant community scale, especially in grasslands of high fertility that carry large numbers of gazing animals (e.g., McNaughton et al. 1997a, b, Frank and Groffman 1998). Herbivorous mammals can return large quantities of undigested and nonassimi-

lated nutrients to the soil as dung and urine. This effectively shortcuts the litter-decomposition pathway, providing highly decomposable resources that are rich in labile nutrients, and which can stimulate microbial biomass and activity (Bardgett et al. 1998a), net C and N mineralization (Molvar et al. 1993), and ultimately plant nutrient acquisition and growth. Although the facilitative effect on plantsoil interactions of this form of nutrient redistribution is unquestionable, it has been argued that it cannot explain the widely documented positive effects of herbivores at large spatial scales (Hamilton and Frank 2001). In agricultural and natural ecosystems, waste patches typically only inuence a relatively small proportion of the surface area (Augustine and Frank 2001) and, in some cases, the positive effects of animal wastes on soil processes are insufcient to negate other, adverse effects of browsing (Pastor et al. 1988). Remarkably little is known about the effects of waste from invertebrate herbivores on soil processes, although in microcosms frass from larvae of the gypsy moth (Lymantria dispar), which had fed on foliage of Quercus velutina, has been shown to increases in microbial growth and N mineralization in soil (Lovett and Ruesink 1995). As with larger herbivores, the effects of invertebrate waste products on soil processes are probably highly localized, and may not be able to explain herbivore effects on soil processes at large spatial scales. Further, since ecosystems have a high degree of spatial structure with local variability in soil resources (Schlesinger et al. 1996), these localized effects of herbivores on soil nutrient supply will potentially exert strong effects on this variability, often increasing the heterogeneity of resources entering the soil (Augustine and Frank 2001). Such heterogeneity within ecosystems may create hotspots of productivity, which presumably will alter the magnitude of herbivore effects on soil biota and soil processes at the local scale. Leaf and root litter quality.Foliar herbivory in grasslands of high soil fertility often enhances leaf nutrient concentration (e.g., Holland and Detling 1990, Hamilton and Frank 2001) either directly through reallocation of nutrients within individual plants, or indirectly by stimulating soil mineralization processes, as discussed above. Likewise, in forest ecosystems, there are incidences where browsing by mammals has been reported to enhance nutrient concentrations and reduce levels of carbon-based secondary metabolites, such as phenolics, in foliage (Bryant and Reichart 1992). This has been shown to improve the quality of leaf litter being returned to soil, thereby enhancing decomposition and mineralization processes in forest ecosystems (Keilland et al. 1997). Stimulation of soil biota and soil processes by foliar herbivory may also be due, in part, to roots of grazed plants producing litter of a higher quality. Seastedt et al. (1988) showed that trimming of shoots of the grass Andropogon gerardii increased the N concentration of roots, and proposed that

Special Feature

2262

RICHARD D. BARDGETT AND DAVID A. WARDLE

Ecology, Vol. 84, No. 9

this would stimulate soil organisms through improved resource quality. Likewise, root herbivory has been shown alter the root N content of plants attacked by nematodes (Bardgett et al. 1999a, b). Such herbivoreinduced increases in foliage and root nutrient concentration will invariably feedback positively on soil biota and soil mineralization processes. Herbivores can also induce the production of secondary plant compounds in foliage and roots, which negatively impacts on soil biota due to reduced litter quality. For example, severe defoliation of trees, such as that caused by periodic invertebrate attack, often results in reduced concentrations of N and increased concentrations of certain secondary metabolites (e.g., phenolics) in subsequently produced foliage. It is unclear whether this response is induced for plant defense against browsing (Rhoades 1985) or whether phenolic production is a physiological response of trees recovering from the nutritional deciency associated with defoliation (Bryant et al. 1993). Whatever the mechanism responsible, the net result is likely to be the production of leaf litter with characteristics that are less favorable for decomposer organisms. Direct evidence of this is scarce, but Findlay et al. (1996) showed that cellular damage caused by spider mites to seedlings of Populus deltoides increased the concentration of polyphenols in foliage, resulting in a 50% reduction in the decomposition rate of subsequently produced leaf litter. The importance of these events in nature is not known, but presumably they would negatively inuence soil mineralization processes and plant productivity at the individual plant and plant community scale. Numerous other antiherbivore defenses are produced in shoots of herbaceous plants in response to herbivore attack (Agrawal 1998) and several defense compounds involved in aboveground defense are produced in roots, and then transported aboveground (Van der Putten et al. 2001). Almost nothing is known about how these induced antiherbivore defense compounds persist in litter produced by herbivore-attacked plants, or how they may affect components of the soil biota and soil mineralization processes. However, their impact on producerdecomposer feedbacks is complex and therefore difcult to predict because their induction following herbivore attack might also affect plant-pathogen interactions (Paul et al. 2000) or act as a cue for parasitoids and predators of the herbivore, thereby reducing herbivore attack. This latter mechanism has recently been demonstrated for both foliar herbivore (Kessler and Baldwin 2001) and root herbivore (Van Tol et al. 2001) attack. These multitrophic and nutritional interactions will inuence the net effect of herbivory on plant growth and make the outcome of producerdecomposer feedbacks difcult to predict.

Mechanism 3: functional composition of vegetation

Over long time scales, herbivory often leads to signicant changes in the functional composition of the

plant community, which in turn alters not just the quantity, but also the quality of litter inputs to soil, and hence affects soil biota and soil nutrient cycling. Across plant species, palatability of foliage and decomposability of plant litter are governed by similar suites of ecophysiological traits, and for this reason palatable plant species generally produce litter that is of a higher quality for decomposers than do unpalatable species (Grime et al. 1996). The inuence of herbivores on plant species replacement and vegetation succession, and associated changes in the functional composition of vegetation, can therefore be important in driving the belowground subsystem. Several examples in the literature exist both of retardation and acceleration of plant succession by herbivores (reviewed by Davidson 1993), and these effects are likely to impact upon the decomposer subsystem (Fig. 2). Retardation of succession occurs when dominant plant species benet through herbivory, for example through compensatory growth and positive feedbacks between plants and herbivores (Augustine and McNaughton 1998). This mechanism is most common in grasslands of high soil fertility, and here herbivory has positive effects on the decomposer subsystem through preventing colonization by later successional plants which produce poorer litter quality, as well as through returning carbon and nutrients to the soil in labile forms as dung and urine. Since herbivore diversity is positively related to soil nutrient availability on global scales (Olff et al. 2002) it is likely that existence of these positive feedbacks that involve soil biota are instrumental in sustaining global hot spots of herbivore diversity. Acceleration of succession tends to occur in lowproductivity ecosystems of low soil fertility and results from browsed plant species being disadvantaged by herbivory, and results in replacement of palatable plant species by species which are better defended against herbivory but produce litter of poorer quality. For example, Pastor et al. (1988) found that browsing by moose on deciduous tree species with nutrient-rich foliage increased the dominance of a Picea spp., which is of low palatability and produces litter of poorer quality. This shift in vegetation composition resulted in reduced soil microbial biomass and slower rates of litter decomposition and soil N mineralization, thereby reducing soil N availability and plant productivity. Similarly, insect and mammalian herbivory in N-limited Oak savanna was shown by Ritchie et al. (1998) to greatly decrease the abundance of plant species with N-rich tissue (especially the legume Lathyrus venosus), thereby reducing the positive contribution of these plants on N cycling in these ecosystems. While several studies have shown effects of aboveground herbivory on plant succession, information on the effects of root herbivores on vegetation change and resulting producerdecomposer feedbacks is scarce. What evidence is available suggests that root herbivory

Special Feature

September 2003

UNDERGROUND PROCESSES

2263

Special Feature

FIG. 2. Mechanistic basis of how herbivores affect the decomposer subsystem at a plant community level, through altering successional trajectories.

can accelerate plant succession, and is therefore likely to also indirectly affect the soil biota and soil processes through changes in the quality of resource inputs to soil. For example, root herbivory by plant-feeding nematodes was found to induce the degeneration of Marram grass (Ammophila arenaria), leading to its replacement by the grass Festuca rubra (Van der Putten and Peters 1997). Likewise, in grassland, root-feeding nematodes might have contributed to small-scale shifts in vegetation composition of grassland (Olff et al. 2000), and to accelerated succession of grass species in agricultural grassland, especially when N becomes limiting after fertilization is stopped and plants become more susceptible to nematode attack (Verschoor 2001). Root herbivores can also alter secondary succession in grassland by selectively feeding on N-rich seedlings, thereby increasing seedling mortality (Brown and Gange 1992). All these herbivore-induced changes in plant community composition will alter the quality of litter entering the soil, thereby inuencing soil biota and soil processes. However, the direction and nature of these potential producerdecomposer feedbacks, and their interrelationship with other indirect effects of root herbivory (e.g., increased susceptibility to pathogen at-

tack), remains unknown and is extremely difcult to predict. An additional role for root herbivores in plant succession and producerdecomposer feedbacks concerns their effects on N transfers from early successional N xing plants. It has been reported that nematode feeding on a legume increased the transfer of N from this plant to a neighboring grass species, and that the transfer was facilitated by increased microbial activity in the rhizosphere of the infested plant (Bardgett et al. 1999b). If these responses are common in legume-herbivore associations, then it is likely that they would contribute to facilitative effects of N-xers in plant succession. For example, root-herbivore induced enhancements in nutrient ux in the rhizosphere of naturally occurring N2-xers, such as alder (Alnus species), may benet other successional plant species, thus contributing to species replacement (Fig. 3). These ideas remain untested, but the widespread occurrence of N-xers in primary succession and their strong association with host-specic root herbivores suggests that such interactions might have an important role in determining vegetation change and soil nutrient cycling in natural ecosystems.

2264

RICHARD D. BARDGETT AND DAVID A. WARDLE

Ecology, Vol. 84, No. 9

Special Feature

FIG. 3. Proposed facilitative role for root herbivores in plant succession through their positive effect on N transfers from early-successional N-xing plants to late-successional plant species, contributing to plant species replacement. These facilitative effects are likely to occur in association with other inhibitory effects of root herbivores and pathogen attack on the early-successional N xer, thereby contributing to the displacement of early-successional species.

THE SIGNIFICANCE OF MULTIPLE-HERBIVORE-SPECIES COMMUNITIES The vast majority of studies on herbivore impacts in ecosystems have focused on a single species of herbivore. However, in real ecosystems, several species of herbivore usually coexist, and may indirectly interact with one another (Masters and Brown 1997, Van der Putten et al. 2001). It is therefore plausible that multiple-species herbivore communities may exert very different effects on community- or ecosystem-level properties (including those relevant to the decomposer subsystem) to what would be observed for single herbivore species systems. The ecological consequences of increasing herbivore diversity (at either the species or functional group level) may be expected to operate through the same types of mechanisms that have been proposed for ecosystem effects of plant diversity, in which greater resource partitioning among species and greater difference between species in key functional attributes maximizes the net effect of diversity (Nijs and Roy 2000, Daz and Cadibo 2001). At the within-plant level, resource partitioning, and therefore resource-use complementarity among herbivore species, is maximized when different herbivore species belong to different functional groups or guilds (cf. Root 1973) and utilize different tissues of the same plant. Similarly, at the level of the plant community, resource partitioning among herbivore species is most likely to occur when the herbivore community is dominated by specialists which selectively feed on different

plant species, and least likely when the community is dominated by generalists that are in competition for the same food resource. Greater resource partitioning through increased herbivore diversity would in turn be expected to lead to greater net resource utilization (through greater net consumption of plant material) by the herbivore community. It is well recognized that the effects of increased plant species diversity on production-driven processes, through greater total resource utilization, can only operate in a positive direction (Fig. 4a). In contrast, in the case of herbivores, greater net resource utilization through increased diversity can theoretically result in a range of possible effects on ecosystem processes (Fig. 4b). This is because, as explained in the earlier discussion, there are a variety of mechanisms through which increasing intensities of herbivory can affect the quantity and quality of resources entering the decomposer subsystem, and both positive and negative effects may arise depending upon the relative importance of different mechanisms. Further, it is possible that, within a system, increasing net consumption of plant material (through increasing herbivore density) may not only lead to just montonically increasing or decreasing ecosystem process rates, but also unimodal responses, because plant-driven processes can be optimized by intermediate levels of herbivory. Although Fig. 4 provides a theoretical framework about how herbivore diversity may inuence ecosystem processes, there are few studies that have provided rel-

September 2003

UNDERGROUND PROCESSES

2265

FIG. 4. (a) The types of responses that may be expected for plant productivity and related ecosystem functions to increasing plant diversity, and (b) the response of ecosystem functions that can theoretically occur as a result of increasing herbivore diversity through resource use complementarity. Both positive and negative responses of belowground processes are possible because increasing net herbivory (resulting from greater resource use complementarity) can have either positive or negative effects on decomposer organisms depending upon which mechanisms are dominating. Further, unimodal responses can theoretically occur if plant productivity and nutrient cycling rates are optimized by intermediate levels of herbivory.

Special Feature

evant data. At the within-plant level, it is recognized that resource partitioning among herbivore species may occur, for example, where different species utilize different tissues produced by the same plant (Masters and Brown 1992) or utilize tissues of that plant at different times of the year (see Masters et al. 1993). Further, Masters and Brown (1992) and Muller-Scharer and Brown (1995) both considered the net impact of aboveground and belowground herbivores, both singly and in combination, on plant production; neither study found consistent interactive effects of both herbivore groups together as opposed to each one operating singly. However, the issue of how resource partitioning among different herbivore species or functional groups consuming the same plant inuences the quality and quantity of resources produced by the plant, and the consequences of this for the decomposer subsystem, remains essentially unexplored. At the level of the plant community, some degree of specialization among herbivore species (and therefore resource partitioning) can be inferred by the frequently observed pattern of herbivore species diversity being correlated with plant species diversity (Southwood et al. 1979). There is a dearth of studies addressing the belowground functional signicance of herbivore diversity, although relevant data is presented by the experimental work of Mikola et al. (2001b). Here, synthetic plant communities of three grassland plant species were established, and treatments consisted of defoliation of the three species in all one-, two-, and three-way combinations. Indirect effects of defoliation were found to differ across treatments for some components of the soil food web, soil respiration and soil

nitrate levels, and these effects were determined by which combinations of species were defoliated rather than how many species were defoliated. Further, Wardle et al. (2000) found for synthesized plantherbivore grassland communities that the effects of two invertebrate herbivore species in combination on various above- and belowground properties and processes generally did not differ from that of each of the two herbivore species considered singly. This points to the two species being able to substitute for the effects of one another. Some evidence for the possible ecosystemlevel importance of resource use complementarity vs. competition among herbivore species emerges from the browsing mammal literature. For example, Bowers (1993) used fenced exclosure plots in which four different types of exclosures were used to selectively exclude different components of the mammalian herbivore community based on animal body size, and found that different subsets of the mammalian herbivore biota differed in their overall effects on the plant community. However, comprehensive experimental tests of the ecosystem impacts of terrestrial herbivore diversity, at either the functional group level or species level (e.g., along the lines of those recently performed for estuarine herbivore systems by Duffy et al. [2001]), remain to be conducted. CONCLUSIONS In this paper, we have identied and proposed three key mechanisms by which herbivores can indirectly affect decomposer organisms and soil processes through altering the quantity and quality of resources that enter the soil. These mechanisms appear to be

2266

RICHARD D. BARDGETT AND DAVID A. WARDLE

Ecology, Vol. 84, No. 9

broadly similar in nature for both foliar and root herbivory, and mechanisms resulting from both operate at a range of spatial and temporal scales. These can operate either in the short term, through involving physiological responses of individual plants to herbivore attack, or in the long term, involving alteration of plant productivity and community structure and subsequent changes in the quantity and quality of litter inputs to soil. Due to the variety of possible mechanisms, the effects of foliar and root herbivores on soil biota and ecosystem function are idiosyncratic and hence difcult to predict, with positive, negative, or neutral effects of herbivory being possible depending upon the balance of these different mechanisms. The variety of mechanisms is consistent with the idiosyncratic and heterogeneous nature of herbivore impacts that are commonly observed in the eld. What is clear is that the magnitude of herbivore effects on soil biota and soil processes, and the consequence of this for producerdecomposer feedbacks, differs greatly across ecosystems, and this appears to depend largely on soil fertility, a determinant of herbivore diversity (Olff et al. 2002), and on the proportion of NPP that is consumed by herbivores. The proportion of NPP that is consumed by herbivores varies considerably across ecosystems (McNaughton et al. 1989), and in more productive ecosystems, which are dominated by palatable, nutrient-rich plants and support a greater diversity and level of herbivory, positive effects on soil biota and soil processes appear to dominate. Here, a key mechanism for these positive effects of herbivory is that nutrients are returned to the soil in labile forms (e.g., dung and urine, and high quality litter) leading to positive effects on soil biota and mineralization processes, and hence plant productivity (mechanism 2). Additionally, plant physiological responses to herbivory, such as enhanced root exudation, also stimulate soil biota and nutrient mineralization processes in productive ecosystems (mechanism 1), leading to enhanced nutrient supply to plants. These mechanisms, together, reinforce soil fertility and ultimately benet plant productivity at the ecosystem scale. In contrast, in unproductive ecosystems of low soil fertility, with low herbivore consumption rates, negative effects on soil biota and soil processes appear to dominate. Here, selective grazing leads to changes in the functional composition of vegetation (mechanism 3), and especially the dominance of defended plant species that produce litter that is of poor nutritional quality to decomposers. The net effect of this is low levels of soil biotic activity, nutrient mineralization, and supply rates of nutrients from soil, and hence a reduction in plant productivity at the ecosystem scale. While positive effects on soil biota may occur in these unproductive ecosystems, for example due to fecal return, these effects are likely to be highly localized and hence of relatively little importance to plant productivity at the ecosystem scale.

We highlight some directions for future research on the effects on herbivory and producerdecomposer feedbacks. First, and perhaps most importantly, there is a need for comprehensive eld studies that compare the nature, and determine the underlying mechanisms, of herbivore impacts on aboveground and belowground processes in ecosystems of varying productivity and soil fertility. Second, more studies are required to explore the various ways that root herbivores impact on soil biota and soil processes. In particular, there is a need for experimental testing of their roles in relation to how root herbivores with differing feeding strategies and degrees of host specicity indirectly inuence the soil subsystem, and how their impacts are modied by complex multi-trophic interactions (for example between plants, root herbivores, and pathogens). A related question that requires testing is how predation of both foliar and root herbivores inuences soil biota and soil processes by causing trophic cascades which could affect the quantity and quality of resources produced by plants. Third, a key challenge is to further our understanding of how plants and soil biota respond to simultaneous attack from root and foliar herbivores, and to simultaneous attack from the diversity of herbivores that can exist both above and belowground. To this end, we have presented various predictions that require experimental investigation on the range of responses of soil biota and soil processes that can theoretically occur as a result of increasing herbivore diversity through resource use complementarity. Finally, virtually nothing is known about how induced antiherbivore defense compounds may directly affect components of the soil biota and soil mineralization processes, or how they might indirectly inuence the soil subsystem through altering multitrophic interactions between herbivores, parasitoides and predators; understanding of these multitrophic interactions is clearly a key research challenge.
LITERATURE CITED Agrawal, A. A. 1998. Induced responses to herbivory and increased plant performance. Science 279:12011202. Augustine, D. J., and D. A. Frank. 2001. Effects of migratory ungulates on spatial heterogeneity of soil nitrogen properties on a grassland ecosystem. Ecology 82:31493162. Augustine, D. J., and S. J. McNaughton. 1998. Ungulate effects on the functional species composition of plant communities: herbivore selectivity and plant tolerance. Journal of Wildlife Management 62:11651183. Bardgett, R. D., R. Cook, G. W. Yeates, and C. S. Denton. 1999a. The inuence of nematodes on below-ground processes in grassland ecosystems. Plant and Soil 212:2333. Bardgett, R. D., C. S. Denton, and R. Cook. 1999b. Belowground herbivory promotes soil nutrient transfer and root growth in grassland. Ecology Letters 2:357360. Bardgett, R. D., A. C. Jones, D. L. Jones, S. J. Kemmitt, R. Cook, and P. J. Hobbs. 2001. Soil microbial community patterns related to the history and intensity of grazing in sub-montane ecosystems. Soil Biology and Biochemistry 33:16531664. Bardgett, R. D., S. Keiller, R. Cook, and A. Gilburn. 1998a. Dynamic interactions between soil fauna and microorgan-

Special Feature

September 2003

UNDERGROUND PROCESSES

2267

isms in upland grassland soils: a microcosm experiment. Soil Biology and Biochemistry 30:531539. Bardgett, R. D., D. K. Leemans, R. Cook, and P. J. Hobbs. 1997. Seasonality of soil biota of grazed and ungrazed hill grasslands. Soil Biology and Biochemistry 29:12851294. Bardgett, R. D., D. A. Wardle, and G. W. Yeates. 1998b. Linking above-ground and below-ground food webs: how plant responses to foliar herbivory inuence soil organisms. Soil Biology and Biochemistry 30:18671878. Belovsky, G. E., and J. B. Slade. 2000. Insect herbivory accelerates nutrient cycling and increases plant production. Proceedings of the National Academy of Sciences (USA) 97:1441214417. Bokhari, U. G. 1977. Regrowth of western wheatgrass utilizing 14C [carbon isotope]-labeled assimilates stored in belowground parts. Plant and Soil 48:115127. Bowers, M. A. 1993. Inuence of herbivorous mammals on an old-eld plant communityyears 14 after disturbance. Oikos 67:129141. Brown, V. K., and A. C. Gange. 1990. Insect herbivory belowground. Advances in Ecological Research 20:158. Brown, V. K., and A. C. Gange. 1992. Secondary plant succession: how is it modied by insect herbivory? Vegetatio 101:313. Bryant, J. P., and P. B. Reichart. 1992. Controls over secondary metabolite production in arctic woody plants. Pages 377390 in F. S. Chapin, R. L. Jefferies, J. F. Reynolds, G. R. Shaver, and J. Svoboda, editors. Arctic ecosystems in a changing climate: an ecophysiological perspective. Academic Press, New York, New York, USA. Bryant, J. P., P. B. Reichart, T. P. Clausen, and R. A. Werner. 1993. Effects of mineral nutrition on delayed inducible resistance in Alaska paper birch. Ecology 74:20722084. Collins, S. L., A. K. Knapp, J. M. Briggs, J. M. Blair, and E. M. Steinauer. 1998. Modulation of diversity by grazing and mowing in native tallgrass prairie. Science 280:745 747. Davidson, D. W. 1993. The effects of herbivory and granivory on terrestrial plant succession. Oikos 68:2335. De Mazancourt, C., M. Loreau, and L. Abbadie. 1999. Grazing optimization and nutrient cycling: potential impact of large herbivores in a savanna system. Ecological Applications 9:784797. Denton, C. S., R. D. Bardgett, R. Cook, and P. J. Hobbs. 1999. Low amounts of root herbivory positively inuences the rhizosphere microbial community of a temperate grassland soil. Soil Biology and Biochemistry 31:155165. Daz, S., and M. Cadibo. 2001. Vive la difference: plant functional diversity matters to ecosystem processes. Trends in Ecology and Evolution 16:646655. Duffy, J. E., K. S. Macdonald, J. M. Rohde, and J. D. Parker. 2001. Grazer diversity, functional redundancy, and productivity in seagrass beds: an experimental test. Ecology 82:24172434. Findlay, S., M. Carreiro, V. Krischic, and C. J. Jones. 1996. Effects of damage to living plants on leaf litter quality. Ecological Applications 6:269275. Frank, D. A., and P. M. Groffman. 1998. Ungulate vs. landscape control of soil C and N processes in grasslands of Yellowstone National Park. Ecology 79:22292241. Gehring, C. A., and T. G. Whitham, 1994. Interactions between aboveground herbivores and the mycorrhizal mutualists of plants. Trends in Ecology and Evolution 9:251 255. Grime, J. P., J. H. C. Cornelissen, K. Thompson, and J. G. Hodgson. 1996. Evidence of a causal connection between anti-herbivore defense and the decomposition rate of leaves. Oikos 77:489494. Guitian, R., and R. D. Bardgett. 2000. Plant and soil microbial responses to defoliation in temperate semi-natural grassland. Plant and Soil 220:271277.

Hamilton, E. W., and D. A. Frank, 2001. Can plant stimulate soil microbes and their own nutrient supply? Evidence from a grazing tolerant grass. Ecology 82:23972402. Holland, J. N., W. Cheng, and D. A. Crossley, Jr. 1996. Herbivore-induced changes in plant carbon allocation: assessment of below-ground C uxes using carbon-14. Oecologia 107:8794. Holland, E. A., and J. K. Detling. 1990. Plant response to herbivory and belowground nitrogen cycling. Ecology 71: 10401049. Ingham, R. E., and J. K. Detling. 1990. Effects of root-feeding nematode son aboveground net primary production in a North American grassland. Plant and Soil 121:279281. Kaye, J. P., and S. C. Harte, 1997. Competition for nitrogen between plants and soil microorganisms. Trends in Ecology and Evolution 12:139143. Kessler, A., and L. T. Baldwin. 2001. Defensive function of herbivore-induced plant volatile emissions in nature. Science 291:21412144. Kielland, K., J. P. Bryant, and R. W. Ruess. 1997. Moose herbivory and carbon turnover of early successional stands in interior Alaska. Oikos 80:2530. Lovett, G. M., and A. E. Ruesink. 1995. Carbon and nitrogen mineralization from decomposing gypsey moth frass. Oecologia 104:133138. Masters, G. J., and V. K. Brown. 1992. Plant-mediated interactions between two spatially separated insects. Functional Ecology 6:175179. Masters, G. J., and V. K. Brown. 1997. Host-plant mediated interactions between spatially separated herbivores: effects on community structure. Pages 217236 in A. C. Gange and V. K. Brown, editors. Multitrophic interactions in terrestrial ecosystems. Blackwell Science, Oxford, UK. Masters, G. J., V. K. Brown, and A. C. Gange. 1993. Plant mediated interactions between above- and below-ground insect herbivores. Oikos 66:148151. McNaughton, S. J. 1985. Ecology of a grazing system: the Serengeti. Ecological Monographs 655:259294. McNaughton, S. J., F. F. Banyikwa, and M. M. McNaughton. 1997a. Promotion of the cycling of diet-enhancing nutrients by African grazers. Science 278:17981800. McNaughton, S. J., F. F. Banyikwa, and M. M. McNaughton. 1998. Root biomass and productivity in a grazing ecosystem: the Serengeti. Ecology 79:587592. McNaughton, S. J., M. Osterheld, D. A. Frank, and K. J. Williams. 1989. Ecosystem level patterns of primary productivity and herbivory in terrestrial habitats. Nature 341: 142144. McNaughton, S. J., R. W. Ruess, and S. W. Seagle. 1988. Large mammals and process dynamics in African ecosystems. BioScience 38:794800. McNaughton, S. J., G. Zuniga, M. M. McNaughton, and F. F. Banyikwa. 1997b. Ecosystem catalysis: soil urease activity and grazing in the Serengeti ecosystem. Oikos 80: 467469. Mikola, J., G. W. Yeates, G. M. Barker, D. A. Wardle, and K. I. Bonner. 2001a. Effects of defoliation intensity on soil food-web properties in an experimental grassland community. Oikos 92:333343. Mikola, J., G. W. Yeates, D. A. Wardle, G. M. Barker, and K. I. Bonner. 2001b. Response of soil food web structure to defoliation of different plant species combinations in an experimental grassland community. Soil Biology and Biochemistry 33:205214. Milchunas, D. G., and W. K. Lauenroth. 1993. Quantitative effects of grazing on vegetation and soils over a global range of environments. Ecological Monographs 63:327 366. Molvar, E. M., R. T. Bowyer, and V. van Ballenberghe. 1993. Moose herbivory, browse quality, and nutrient cycling in an Alaskan treeline community. Oecologia 94:472479.

Special Feature

2268

RICHARD D. BARDGETT AND DAVID A. WARDLE

Ecology, Vol. 84, No. 9

Muller-Scharer, H., and V. K. Brown. 1995. Direct and in direct effects of above- and belowground insect herbivory on plant density and performance of Tripleurospermum perforatum during early plant succession. Oikos 72:3641. Nijs, I., and J. Roy, 2000. How important are species richness, species evenness and interspecic differences to productivity? A mathematical model. Oikos 88:5766. Olff, H., B. Hoorens, R. G. M. De Goede, W. H. Van der Putten, and J. M. Gleichman. 2000. Small-scale shifting mosaics of two dominant grassland species: the possible role of soil borne plant pathogens. Oecologia 125:4554. Olff, H., M. E. Ritchie, and H. T. Prins. 2002. Global environmental controls of diversity in large herbivores. Nature 415:901904. Pastor, J., R. J. Naiman, B. Dewey, and P. Mcinnes. 1988. Moose, microbes and the boreal forest. BioScience 38:770 777. Paul, N. D., P. E. Hatcher, and J. E. Taylor. 2000. Coping with multiple enemies: an integration of molecular and ecological perspectives. Trends in Plant Science 5:220225. Rhoades, D. F. 1985. Offensivedefensive interactions between herbivores and plants: their relevance in herbivore population dynamics and ecological theory. American Naturalist 125:3205238. Ritchie, M. E., D. Tilman, and J. M. H. Knops. 1998. Herbivore effects on plant and nitrogen dynamics in oak savanna. Ecology 79:165177. Root, R. B. 1973. Organization of a plantarthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecological Monographs 43:95124. Ruess, R. W., R. L. Hendrick, and J. P. Bryant, 1998. Regulation of ne root dynamics by mammalian browsers in early successional Alaskan taiga forests. Ecology 79:2706 2720. Schlesinger, W. H., J. A. Raikes, A. E. Hartley, and A. F. Cross. 1996. On spatial pattern of soil nutrients in desert ecosystems. Ecology 77:276282. Seastedt, T. R., R. A. Ramundo, and D. C. Hayes. 1988. Maximisation of densities of soil animals by foliage herbivory: empirical evidence, graphical and conceptual models. Oikos 51:243248.

Southwood, T. R. E., V. K. Brown, and P. M. Reader. 1979. The relationships of plant and insect diversities in succession. Biological Journal of the Linnean Society 12:327 348. Stark, S., D. A. Wardle, R. Ohtonen, T. Helle, and G. W. Yeates. 2000. The effect of reindeer grazing on decomposition, mineralization and soil biota in a dry oligotrophic Scots pine forest. Oikos 90:301310. Suominen, O. 1999. The impact of cervid browsing and grazing on the terrestrial gastropod fauna in the boreal forests of Fennoscandia. Ecography 22:651658. Van der Putten, W. H., and B. A. M. Peters. 1997. How soilborne pathogens may affect plant competition. Ecology 78: 17851795. Van der Putten, W. H., L. E. M. Vet, J. A. Harvey, and F. L. Wackers. 2001. Linking above- and belowground multitrophic interactions of plants, herbivores, pathogens and their antagonists. Trends in Ecology and Evolution 16:547 554. Van Tal, R. W. H. M., A. T. C. van der Sommen, M. I. C. Boff, J. van Bezooijen, M. W. Sabelis, and P. H. Smits. 2001. Plants protect their roots by alerting the enemies of grubs. Ecology Letters 4:292294. Verschoor, B. C. 2001. Nematodeplant interactions in grasslands under restoration management. Dissertation. University of Wageningen, Wageningen, The Netherlands. Wardle, D. A. 2002. Communities and ecosystems: linking the aboveground and belowground components. Princeton University Press, Princeton, New Jersey, USA. Wardle, D. A., G. M. Barker, G. W. Yeates, K. I. Bonner, and A. Ghani. 2001. Impacts of introduced browsing mammals in New Zealand forests on decomposer communities, soil biodiversity and ecosystem properties. Ecological Monographs 71:587614. Wardle, D. A., K. I. Bonner, and G. M. Barker. 2000. Stability of ecosystem properties in response to above-ground functional group richness and composition. Oikos 89:1123. Yeates, G. W., S. Saggar, C. S. Denton, and C. F. Mercer. 1998. Impact of clover cyst nematode (Heterodera trifolii) infection on soil microbial activity in the rhizosphere of white clover (Trifolium repens)a pulse labelling experiment. Nematologica 44:8190.

Special Feature

Ecology, 84(9), 2003, pp. 22692280 2003 by the Ecological Society of America

PLANT DEFENSE BELOWGROUND AND SPATIOTEMPORAL PROCESSES IN NATURAL VEGETATION

WIM H. VAN
DER

PUTTEN1

Netherlands Institute of Ecology NIOO-KNAW, Multitrophic Interactions Department, P.O. Box 40, 6666 ZG Heteren, The Netherlands

Abstract. Root herbivores and pathogens play an important role in driving plant abundance, species diversity, and succession in natural vegetation. Subterranean plant feeders and pathogenic microorganisms interfere with basic functions of plant roots, such as resource uptake, storage of reserves, and anchoring of plants in the soil, but concepts and theories on control of herbivores and pathogens, such as the Green World Hypothesis, have been developed and applied almost exclusively for the aboveground subsystem. Root herbivores and pathogens affect spatial and temporal patterns in natural plant communities, and whether these patterns are cyclic or irreversible depends on characteristics of the root feeders, interactions with other soil or aboveground organisms, and the rate of changes in the abiotic environment. Established plants can tolerate root herbivores and pathogens at densities that are lethal to their offspring. Dispersal by seeds or rhizomes allows new cohorts to become established before root herbivores or root pathogens colonize and develop, but it provides only temporal release of plants from their natural enemies. Permanent release from root herbivores and pathogens contributes to plant invasiveness. I propose to expand the concept of plantsoil feedback by including plant defense belowground. Many plant secondary compounds are synthesized in the roots, and these chemicals could affect root herbivores, root pathogens, and their natural enemies. Plants may exert direct and indirect defense, resistance, tolerance, or dispersal to move away from the herbivores and pathogens belowground, and I propose that acknowledging trade-offs and life history strategies will enhance our capacity to predict spatiotemporal patterns in natural vegetation. Further studies in this area will enhance our understanding of plant abundance, succession, and invasions in natural communities, as well as the evolution of plant dispersal and other defensive strategies against root herbivores and pathogens in natural communities.
Key words: direct and indirect defense; dispersal; diversity; enemy release; invasiveness; multitrophic; nematode; plant community; resistance; root pathogen; subterranean herbivore; succession.

Special Feature

INTRODUCTION Many soil organisms have the potential to inuence the composition of plant communities in space and time, either through changing the availability of resources, or by direct feeding on plant roots (Wardle 2002). Plants affect soil organisms and soil organisms reciprocally affect plants, leading to a feedback that drives changes in plant communities over space and time (Bever et al. 1997). The feedback between plants and soil organisms can be positive, neutral, or negative for the plants involved, for their offspring, and for other plant species in the community (Reynolds et al. 2003). In order to enhance our understanding of plantsoil feedback in relation to plant life histories and plant community processes, interactions between plants and the various functional groups of organisms inhabiting the root zone need to be explored in more detail. I will focus mainly on interactions between plant roots, invertebrate herbivores, microbial pathogens,
Manuscript received 8 May 2002. revised 4 October 2002. accepted 12 October 2002; nal version received 11 November 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256. 1 E-mail: putten@nioo.knaw.nl

mutualistic symbionts, and natural enemies of the herbivores and pathogens. These interactions below ground may contribute to the relative abundance of plant species (Gange et al. 1993, Klironomos 2002), plant species diversity (Bever 1994, Holah and Alexander 1999, Packer and Clay 2000, De Deyn et al. 2003), primary succession (Van der Putten et al. 1993), secondary succession (Brown and Gange 1992, Holah and Alexander 1999, Callaway et al. 2000, Verschoor et al. 2001, De Deyn et al. 2003), and possibly also to plant invasiveness (Klironomos 2002). I propose that these plant community processes are strongly affected by plant defense belowground. In general, plant defensive strategies may include tolerance, resistance, direct and indirect defense, or moving away from their above- and belowground herbivores or pathogens (Burdon 1987, Karban and Baldwin 1997). Plant defense theory that includes the natural enemies of herbivores (Price et al. 1980), has been almost solely developed and tested for aboveground interactions (Van der Putten et al. 2001). In soil, most studies have focused on plantherbivore or plantpathogen interactions, and in spite of a vast amount of biocontrol research on belowground invertebrate her-

2269

2270

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

bivores and pathogens in crop systems, very little is known on the role of natural enemies of root herbivores and pathogens in the feedback between plant and soil communities. Since the number of examples of root herbivores and pathogens in natural plant communities is steadily increasing, we now can start examining plant defense from a belowground perspective in order to further understand spatiotemporal processes in natural vegetation and plant life history evolution. Here, I focus on direct and indirect interactions between living plant roots and soil organisms, explore examples of plantherbivore/pathogen interactions in natural ecosystems, and discuss current ideas investigating how plants may avoid, tolerate or resist, or defend themselves against root herbivores and root pathogens. ROOT HERBIVORES, ROOT PATHOGENS, AND HOST SPECIFICITY Most root herbivores are invertebrates (insects, micro-arthropods, and nematodes), though vertebrates, mainly small mammals, also feed on plant roots (Mortimer et al. 1999). As far as microorganisms are concerned, the majority of the examples of soil pathogens in natural vegetation concern fungal diseases (Jarosz and Davelos 1995). Detailed reviews have highlighted the effects of root-feeding insects (Andersen 1987, Brown and Gange 1990), root-feeding nematodes (Stanton 1988, Mortimer et al. 1999), and root pathogens (Jarosz and Davelos 1995), and how these affect the productivity and composition of natural plant communities. I will focus on host specicity and, briey, on dispersal of root herbivores and root pathogens, since these are important features for plant defense belowground. Host specicity varies widely between belowground insect herbivores (Mortimer et al. 1999). The larval stages of root herbivorous insects, as well as the other root herbivores and pathogens have limited active dispersal capacity, usually ranging from a few centimeters up to a meter per year (Mortimer et al. 1999, Gormsen 2001). However, active dispersal, such as of the aboveground stages of some root-feeding insect larvae, may facilitate host location (Nordenhem and Eidmann 1991). Microarthropods, nematodes and soil microorganisms are dispersed passively by wind (Orr and Newton 1971, Grifn et al. 2001), which does not favor the development of host specicity. Specic root feeding nematodes are mainly known in agricultural ecosystems, where crop rotation leads to the predictable presence of host plants. Specicity also occurs in temperate coastal foredunes, where natural plant species typically occur in monospecic stands. Some nematode species, such as Meloidogyne duytsi on Elymus farctus (sand twitch) and M. maritima on Ammophila arenaria (marram grass) occur on a single host-plant species (Van der Putten and Van der Stoel 1998). Probably, the predictability and reliability of the presence of host plant

roots favors selection for specialization in root-feeding nematodes in sand dunes. Microorganisms have poorer active dispersal abilities, for example, through hyphal growth, than most invertebrates (De Boer et al. 1998b). Specialists might also develop in the case of facultative saprotrophic growth of soil pathogens, such as oomycete fungi, which survive the absence of their host plants by feeding on dead organic matter (Jarosz and Davelos 1995). An example of this group of facultative saprotrophs is Pythium sp., which kills seedlings of Prunus serotina (black cherry) in North American forests (Packer and Clay 2000). However, facultative saprotrophic species are inferior competitors for dead organic matter and will, sooner or later, be outcompeted by obligatory saprophytes unless host plants become re-established. Other strategies that might favor the development of host specicity of nematodes and soil microorganisms are the formation of survival structures (cysts, dauer larvae, spores) which enables these organisms to survive periodic absence of the hosts. PLANT DEFENSE STRATEGIES AGAINST ROOT HERBIVORES AND ROOT PATHOGENS After publication of the Green World Hypothesis (Hairston et al. 1960), a major debate arose as to whether herbivores were controlled more effectively by resources or predators. Alternatively, it was argued that there is an arms race between plants and herbivores leading to the evolution of plant chemical defense (Ehrlich and Raven 1964). Nowadays, it is generally accepted that evolutionary processes and resulting adaptation from interactions between plants and herbivores cannot be fully understood without including the antagonists of the herbivores in a multitrophic framework (Price et al. 1980), although such approach has rarely been applied to belowground interactions. Interestingly, studies on aboveground plant pathogens have typically focused on disease resistance and coevolution of plants and pathogens (Burdon 1987), whereas studies examining control of natural plant pathogens by their natural enemies are rare (Yang et al. 1993). Studies on aboveground plantherbivore and plantpathogen interactions have pointed to a number of defense responses by plants and possible trade-offs. I will apply and develop these ideas for belowground defense in relation to plant life history strategies and plant community processes.

Special Feature

Avoidance and escape

Avoidance and escape have been studied in relation to both aboveground and belowground herbivores and pathogens. Plants may avoid their herbivores or pathogens through processes known as phenological escape, whereby they ower early (or late) in the growth season, or exhibit spatiotemporal unpredictability (Van der Meijden et al. 1988). Phenological escape also includes the postponement of, for example, owering un-

September 2003

UNDERGROUND PROCESSES

2271

til a reliable cue has indicated low risk of future attack (Agrawal 2000). The capacity of plants to produce roots in late autumn and early spring is an example of avoiding root herbivores and soil pathogens, as growth continues when the herbivores and pathogens may be relatively inactive. Seed dispersal (Packer and Clay 2000), clonal growth (De Rooij-Van der Goes et al. 1995), and dormancy enable plants to avoid or escape from root herbivores and pathogens. Similar to some plants that respond positively to aboveground herbivores (Agrawal 2000), exposure to mycorrhizal fungi increases the intensity of rhizome branching (Streitwolf-Engel et al. 1997), whereas pathogens stimulate unidirectional rhizome growth (DHertefeldt and Van der Putten 1998).

Tolerance and resistance

The capacity of plants to regrow and reproduce following herbivory is called tolerance (Strauss and Agrawal 1999). Similarly, tolerance for diseases is expressed as the performance of individuals in the presence compared to their performance in the absence of pathogens (Burdon 1987). Herbivory may even lead to overcompensation, which can increase tness (Agrawal 2000). Overcompensation occurs especially at low densities of the grazers (Crutcheld and Potter 1995), and it also expressed by fungal hyphae in soil when exposed to low densities of enchytraeids (Hedlund and Augustsson 1995). Plant roots may also exert such a compensatory response at low nematode densities. Whereas tolerant plants allow herbivores and pathogens to develop and reproduce, resistant plants reduce the performance of their enemies (Karban and Baldwin 1997). In agricultural systems, tolerance and resistance have been studied for above- and belowground herbivores, as well as for shoot and root pathogens. For natural systems, on the other hand, tolerance and resistance have been mainly studied in relation to aboveground herbivores and pathogens. Tolerance and other defense mechanisms, such as resistance, are considered alternative plant traits in plant-herbivore interactions (Van der Meijden et al. 1988). High growth rates enhance the competitive ability of plants, but it may come at the expense of their capacity to defend against herbivores (Herms and Mattson 1992). Tolerance might be favored when resistance is not possible, or very costly (Agrawal 2000). Empirical support to trade-offs between tolerance and defense are idiosyncratic (Strauss and Agrawal 1999). Probably, experimental conditions are crucial for the results obtained. In controlled conditions, a large fraction of plant root systems can be removed without negatively affecting plant production (Van der Veen 2000), however, selective root feeding reduces production of susceptible plants when competing with resistant species (Van der Putten and Peters 1997). Tolerance may evolve when, for example, plant attack by multispecies communities of pathogens is in-

evitable. In this case, avirulence is less likely to develop (Roy et al. 2000). Soil communities will usually consist of multispecies communities, so that root pathogens can be expected to be moderately aggressive. Some soil pathogens, however, are so aggressive that they kill all establishing plants (Packer and Clay 2000). There is no evolutionary penalty for such aggressiveness, as the pathogens concerned are facultative saprophytes, which enables feeding on organic matter and dead plant parts (Jarosz and Davelos 1995). Tolerance, therefore, might not be the ideal solution for plant defense against facultatively saprotrophic root pathogens. In model calculations, resistant species should often coexist with other, less resistant competing species (Chase et al. 2000). In the same models, tolerant species could not coexist with other, less tolerant competitors. According to these models, species-rich plant communities are expected to be mixtures of plant species that resist root herbivores and soil pathogens, and others that are tolerant.

Direct defense
A number of secondary plant compounds are produced, or presynthesized, in roots and then transported to other plant organs (Karban and Baldwin 1997). Examples are pyrrolizidine alkaloids of Senecio jacobea (Vrieling and Van Wijk 1994). Host plant secondary defense chemicals are able to affect the performance of root-feeding insect larvae (Brust and Barbercheck 1992), as well as entomopathogenic nematodes that feed on the insect larvae (Jaworska and Ropek 1994). These effects might be positive or negative for secondand third-trophic-level organisms, although studies on effects of plant chemicals on natural enemies of the belowground herbivores or pathogens are scarce and causal effects are difcult to establish. Root chemicals that affect root-sucking insects have been rarely reported (Brown and Gange 1990). There is also little information on direct effects of secondary plant chemicals on root-feeding nematodes. There is very little information on the effects of plant defensive compounds on root pathogens and their natural enemies. Fungal pathogens (Fusarium oxysporum) collected from roots of Senecio jacobea (common ragwort) performed better on alkaloids from S. jacobea than the same fungi which had been collected from other plant species (Hol and Van Veen 2002). Possibly, plant defense compounds select for certain rhizosphere microorganisms, but the feedback effects to plant performance and consequences for plant community processes have not yet been studied.

Special Feature

Indirect defense
A main question in relation to indirect defense is whether plants may be actively involved in recruiting the enemies of their enemies (Dicke and Vet 1999, Thaler et al. 1999). Entomopathogenic nematodes can be recruited by plant roots through unknown attractants

2272

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

that are released from the roots upon larval feeding (Van Tol et al. 2001, Boff et al. 2002). However, there are comparatively few other examples of active recruitment of third-level soil organisms by plants. Belowground signals will operate over much smaller scales than is the case above ground, because poor diffusion and transport rates of volatiles and soluble chemicals inhibit the spread of these compounds. Antagonists of root herbivores and pathogens are well known in agriculture, but the development of effective population densities lags behind their hosts, sometimes requiring years of monocropping before effective control levels have been achieved (Weller et al. 1995, Kerry and Crump 1998). In natural systems, therefore, short-lived plants might experience little benet from such antagonists, because they will be replaced by more resistant plant species before antagonists have even developed an effective population density. Nevertheless, antagonists of soil pathogens have been observed in natural soils (Holah and Alexander 1999), but these also directly affect root pathogens without any involvement of plant roots (De Boer et al. 1998a, b). Arbuscular mycorrhizal fungi have been reported to act as antagonists of root pathogens (Carey et al. 1992, Newsham et al. 1994, 1995), plant-feeding nematodes (Little and Maun 1996, Roncadori 1997), and rootfeeding insects (Gange 2000). Similarly, endophytic fungi are able to reduce plant sensitivity to herbivorous nematodes (Clay 1991). Pathogenic rhizobacteria can induced systemic acquired resistance (SAR) and nonpathogenic rhizobacteria can induce systemic resistance (ISR) against plant pathogens (Van Loon et al. 1998). The rhizosphere bacteria that induce resistance, for example uorescent Pseudomonas spp., need to be present in sufciently high densities in order to be effective (Van Loon et al. 1998). Most of this work has been done on crop plants and Arabidopsis thaliana. Theoretically, indirect defenses against root herbivores and pathogens might be operating in many natural systems, but their importance for spatiotemporal processes in natural plant communities has received little attention. PLANT COMMUNITY PROCESSES IN RELATION BELOWGROUND DEFENSE STRATEGIES
TO

tween plant defense belowground and the plant community processes that are observed in the eld.

Primary succession
In coastal dunes of Europe, the nitrogen-xing shrub Hippophae rhamnoides (sea buckthorn) is highly vig orous in outer dunes, whereas, along a transect from foredunes towards the inland, the shrubs die back and gradually disappear from the vegetation. Initially, it was supposed that root-feeding nematodes, such as Tylenchorhynchus microphasmis, were causing the dieback patterns (Oremus and Otten 1981). Subsequent studies, however, have shown that nematodes alone cannot cause the observed growth reduction (Maas et al. 1983), and that plant pathogenic fungi may be involved as well (Zoon et al. 1993). The precise mode of interaction still needs to be resolved. Similarly, both root-feeding nematodes and root pathogens are able to reduce vegetative growth of a pioneer dune grass Ammophila arenaria (Marram grass; Van der Putten et al. 1988). Single nematode or root pathogen species cannot account for the growth reduction observed in the eld soil, so it has been proposed that degeneration of A. arenaria is facilitated through a complex of rootfeeders and root pathogens (De Rooij-Van der Goes 1995). Colonizing A. arenaria produces new roots in spite of a negative feed-back from the soil community, which develops with a time lag of less than one month (Van der Stoel et al. 2002). Avoidance may, therefore, only partially explain vigorous plant growth in mobile sand dunes. Specialist nematodes can be controlled by the plant (Van der Stoel 2001), whereas generalist nematodes are limited by unknown factors (De Rooij-Van der Goes 1995). Plant defense against root-feeding nematodes, therefore, appears to depend on the type of nematode (generalist or specialist). In the eld, a range of direct and indirect effects such as mycorrhizal fungi and natural enemies of the nematodes appear to be involved in the belowground defense of A. arenaria (Fig. 1). It is difcult to establish how tolerance acts as a defense of A. arenaria against root-feeding nematodes. Root formation in mature plants continues in soil where seedlings are severely growth reduced (Van der Stoel et al. 2002), but this might be due to indirect defense, for example caused by mycorrhizal fungi (Little and Maun 1996). These symbiotic mutualists are prominent in zones with vigorous A. arenaria, whereas they are less signicant in degenerate stands (Kowalchuk et al. 2002). Later successional plant species are less susceptible to the soil community of A. arenaria than their predecessors (Van der Putten et al. 1993). Specialist rootfeeding nematodes of A. arenaria have not been found in roots of later succession plant species suggesting that successors are resistant against the nematodes of their predecessors (Van der Putten and Peters 1997). Therefore, avoidance and resistance are important be-

Special Feature

The question of what drives changes in the composition of natural plant communities has been a recurrent theme in ecology, but it has only recently been acknowledged that root herbivores and root pathogens can inuence these processes. Root herbivores and pathogens affect primary (Van der Putten et al. 1993) and secondary succession (Brown and Gange 1992, Verschoor et al. 2001, De Deyn et al. 2003), as well as plant dispersal (Packer and Clay 2000), plant species diversity (Bever 1994, De Deyn et al. 2003), and abundance or invasiveness of alien plant species (Klironomos 2002). I propose that there is a relationship be-

September 2003

UNDERGROUND PROCESSES

2273

Special Feature

FIG. 1. Factors that may control nematodes in the root zone of Ammophila arenaria (marram grass) in coastal sand dunes. Specialist nematodes (coinciding with endoparasitic feeding types) can be controlled by the plant, whereas generalists (coinciding with ectoparasites) are controlled by other factors. Arbuscular mycorrhizal fungi and endophytes, as well as rhizosphere fungi (Verticillium chlamydosporium) and bacteria (Pasteuria penetrans), may act as predators of nematode eggs and juveniles, respectively. The arrows point in the assumed direction of control. This scheme is used by the EU-EcoTrain project in order to determine the relative contribution of the various nematode control mechanisms.

lowground defense mechanisms that drive succession. Probably, tolerance, direct and indirect defense may counteract, or slow down succession. Both primary and secondary succession proceed in an environment where biotic and abiotic conditions change rapidly (Fig. 2). SECONDARY SUCCESSION In early secondary succession on previously cultivated land, soil insects are a major cause of seed predation and seedling mortality of annual and perennial forb species (Gange et al. 1991). Based on insecticide experiments (Brown and Gange 1989b), it was concluded that soil-dwelling insects negatively inuence the survival and competitiveness of early successional forbs in these secondary succession gradients and that these insects enhance the rate of succession (Gange et al. 1993). Foliar insecticides, on the other hand, increased the dominance of grasses, so that aboveground insects increase plant species diversity within succession stages (Brown and Gange 1989a). Aboveground plant-feeding insects indirectly reduce the performance of root-feeding insects due to limitation of carbohydrate investments into the roots following defoliation (Masters and Brown 1992). Arbuscular mycorrhizal fungi also affect aboveground plant-feeding insects, which leads to complex abovegroundbelowground tri-

trophic interactions (Brown and Gange 2002). With successional age, the root-feeding insect community contains more specialized insects (Mortimer et al. 1999), which will inuence plant defense strategies belowground throughout a successional sequence. Root-feeding nematodes might affect secondary succession, but it has been proposed that their effects are most obvious in local hot spots where the nematodes are abundant (Verschoor 2002). Outside these hot spots, nematode abundance seems too low to affect plant species interactions (Verschoor et al. 2002). High local abundance of nematodes might reect stochasticity of colonization and multiplication. Alternatively, it could be that nematode abundance increases when released from control by predators. There are no clear patterns of specialist and generalist root-feeding nematodes in relation to secondary successional stages in grasslands (Verschoor et al. 2001). Soil pathogens are involved in the secondary succession of Douglas r (Pseudotsuga menziesii) and true rs in western Oregon. These trees experience extensive root rot caused by the native fungal pathogen Phellinus weirii, which spreads from one tree to another via root contacts. The subsequent death of the trees may lead to the increased cover of herbaceous species and therefore enhances species diversity. At some sites,

2274

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

FIG. 2. Plantsoil feedback patterns according to Bever et al. (1997): every plant develops a soil community that may have a neutral ( ), negative ( ), or positive ( ) net effect on the productivity and tness of the plant or its offspring. Depending on the effects of the soil community of one plant species on the performance of other plant species, the vegetation development may be directed, leading to primary or secondary succession (left; redrawn from Van der Putten and Van der Stoel 1998), or cyclic (right). The thicknesses of the arrows indicate the strengths of interactions; arrows point in the direction of the effect, and straight lines indicate direct effects whereas dashed lines indicate indirect effects. However, these plant soil feedback schemes lack a temporal component, and they also do not elucidate the role of competitive and top-down components within the soil community. They also do not elucidate possible abovegroundbelowground interactions (see also Bardgett and Wardle 2003).

Special Feature

the reduction of Douglas r due to pathogenic activity enhances forest succession by the western hemlock (Tsuga heterophylla) (Holah et al. 1997). The presence of dead trees also leads to aggregate re incidence after which succession may be reset. The cycles of forest succession indicate at avoidance, but other defense types may be active as well, albeit not yet recognized.

Plant abundance and species diversity

Forces that drive plant abundance or plant species diversity and the possible consequences of species loss for the functioning of natural ecosystems are major issues in ecology (Schulze and Mooney 1993, Gaston 1994). Root herbivores and root pathogens undoubtedly contribute to plant species abundance and species diversity. In coastal dunes of the western USA, the abundance of the coastal shrub lupine (Lupinus arboreus) depends on the root-boring larvae of ghost moths (Hepialus californicus) (Strong 1999). The insect larvae cause large temporal and spatial changes in the shrub cover due to patchy dieback of L. arboreus (Strong et al. 1995, 1999) and they can reduce cumulative seed production over three years by up to 31% (Maron 1998). Shrub lupine is indirectly defended against the ghost moth larvae by entomopathogenic nematodes (Jaffee 1996). The nematodes are in turn subjected to predators, such as nematophagous fungi and predaceous nematodes (Koppenhoffer et al. 1997). Indirect defense does not prevent the local disappearance of lupine, so that herbaceous vegetation may take over until shrubs become reestablished (Maron and Jefferies 1999). The presence and abundance of shrub lupine, therefore, depends on a combination of indirect defense and avoidance. In seminatural grasslands, the ground cover of codominant plant species (Festuca rubra and Carex arenaria) oscillates over time. These oscillations coin-

cide with the occurrence of root herbivorous nematodes and other plant growth-reducing inhabitants of the root zone (Olff et al. 2000). In these grasslands, aboveground vertebrate herbivores (cows) affect the activity of yellow ants (Lasius avus), which use fresh subsoil to produce mounds. This subsoil contains less plantgrowth-reducing biota than the original top soil and therefore favors the ability of formerly dominant clonal plants to regrow. Because of its rapid vegetative expansion rate, C. arenaria is the rst to recolonize ant mounds, whereas F. rubra is the subsequent colonizer (Blomqvist et al. 2000). Ants therefore create patches where plants may avoid root-feeding nematodes and root pathogens. Mills and Bever (1998) observed that the accumulation rate of fungal soil pathogens of the genus Pythium was faster on roots of plant species that were sensitive to the pathogen than on less sensitive plants. Klironomos (2002) related plant abundance in a grassland community to the development rate of a negative feedback with the soil community and found that rare species accumulated root pathogens at a faster rate than dominant plant species. Fungi that were most abundant on the roots of rare plants contributed to this negative feedback. Both studies suggest that resistant plants may become dominant, whereas susceptible plants remain rare. Dominant and rare plants may coexist when there are ample potential regeneration sites acting as enemyfree spaces for rare plants. Moreover, the abiotic conditions of the system need to be relatively invariable in order to enable re-establishment of the rare species (Fig. 3). Coexistence may also be due to one-sided effects, which is illustrated by the pathogenic root fungi of the annual legume Chamaecrista fasciculata, that have a negative effect on the perennial prairie grass Andropogon gerardii (Holah and Alexander 1999).

September 2003

UNDERGROUND PROCESSES

2275

FIG. 3. Two theoretical examples of root herbivore or soil pathogen persistence. The persistent type (dashed line) may lead to directional succession, especially when the environment changes fast, so that after some time, conditions for plants have changed unfavorably (as in coastal dunes; Van der Putten et al. 1993). The non-persistent type (solid line) may allow cyclic succession and such type may contribute to plant species diversity in natural vegetation (as suggested by Bever [1994] and shown by Olff et al. [2000], Blomqvist et al. [2000] and Klironomos [2002]). The horizontal line represents the threshold herbivore or pathogen density that determines whether or not plants may become established.

In summary, dominant plants appear to control their root herbivores and pathogens whereas rare plants fail to do so. Most studies examining the relationship between plant abundance and species diversity, however, provide inconclusive evidence on the relative importance of tolerance, direct defense, and indirect defense.

Plant dispersal and vegetative expansion

According to the Janzen-Connell hypothesis, tree species diversity in neotropical rain forests depends on the presence of specialist herbivores (Janzen 1970, Connell 1978). New tree seedlings can establish only at a distance from parent trees, where the seedlings are out of reach of specialist herbivorous insects (Janzen 1970). This hypothesis has also been tested for soil pathogens in tropical rain forests (Augspurger 1983, Augspurger and Kelly 1984) and in a temperate forest (Packer and Clay 2000). In all cases, soil pathogens accumulating in the root zone of parent trees caused seedling mortality of the tree offspring. In the temperate forest, Pythium species caused the mortality of saplings of Prunus serotina, whereas other tree species did not appear to be affected with equal severity (Packer and Clay 2000). Prunus seedlings need effective dispersal at sufcient distance from older trees of the same species in order to establish. The mature trees appear to tolerate the root pathogen, whereas the seedlings cannot establish because of pathogen accumulation under the trees. Other tree species can establish underneath P. serotina, which indicates that coexisting forest species are resistant against Pythium of Prunus trees. Many clonal plants disperse mainly through the vegetative growth of rhizomes or stolons. In the case of A. arenaria (Van der Putten et al. 1988, De Rooij-Van der Goes et al. 1995), Carex arenaria (DHertefeldt

and Van der Putten 1998), and Trifolium repens (Ennik et al. 1965), clonal expansion leads to avoidance. However, in some cases the escape window seems so narrow (Van der Stoel et al. 2002) that other factors, such as tolerance, direct defense, or indirect defense may play an important role in maintaining plant growth when the newly formed roots are colonized by herbivores and pathogens. Exposure of A. arenaria to pathogen-infested patches reduced root development (De Rooij-Van der Goes et al. 1995). Root pathogens reduced local patch exploitation of C. arenaria because apical rhizome buds developed while lateral buds, from which lateral rhizomes develop, remained dormant (DHertefeldt and Van der Putten 1998). These examples, as well as the example showing an effect of mycorrhizal fungi on rhizome branching patterns (Streitwolf-Engel et al. 1997) reveal that plant defense belowground needs to be incorporated in studies on clonal growth. INVASIONS
OF

Special Feature

EXOTIC PLANTS

Plant invasions are among a suite of major factors worldwide that threaten ecosystem integrity and ecosystem functioning (Dukes and Mooney 1999). In their new territories, invasive plants might have escaped from their specic root herbivores and soil pathogens (Keane and Crawley 2002). Alternatively, local root herbivores, root pathogens, and other parasites provide biotic resistance against the invaders, which is one of the explanations why so few exotics become pests. Thus far, few studies have focused on how soil communities respond to invasive plant species. The invaders may cause major changes in the soil, especially when these plants have specic attributes such as the ability to x atmospheric nitrogen (Vitousek et al. 1987). Some invading plant species affect soil quality

2276

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

Special Feature

for native species, e.g., by changing the soil microbial composition (Callaway et al. 2000), whereas the mycorrhizal fungi will allow the invader to establish symbiotic relationships, because these symbiotic mutualists are relatively nonspecic (Richardson et al. 2000). In a study by Klironomos (2002), ve plant species that have invaded North America and Canada from Eurasia were compared with ve rare native species in their rate of developing a negative feedback to the soil community. Overall, the invaders had a neutral or positive feedback, whereas the rare native species developed a negative feedback. Isolation of spores of arbuscular mycorrhizal fungi and of the remaining microbial component of the soil and reinoculation of those fungal species showed that all ten plant species responded either neutrally or positively to the presence of mycorrhizal fungi. Although none of the invasive plant species were negatively affected by the addition of the remaining microbial component collected from pots with cultures from their own species, all of the rare species performed worse under these conditions (Klironomos 2002). This result suggests that the invasive species are free of soil pathogens from their native ecosystems while beneting from the (aspecic) mycorrhizal symbionts in their new territory. However, we need comparisons of the response of invasive plants to their soil communities in both their native and new territories in order to determine in which cases escape from native soil pathogens indeed might explain plant invasiveness as the dominant native species also had a neutral to positive soil feedback. DISCUSSION
AND

CONCLUSIONS

Plant defense belowground strongly affects spatiotemporal processes in natural vegetation. Dispersal through seeds or rhizomes enables plants to avoid root herbivores and pathogens that develop in the root zone of established plants. Plant dispersal has a number of functions, such as colonizing new resources and encountering new mating partners, but root herbivores and soil pathogens could be a major factor selecting for plant dispersal. A dilemma of plant dispersal, however, may be that plants escape from their belowground bodyguards as well. I assume that there are tradeoffs belowground similar to above ground between dispersal and defense (Van der Meijden et al. 1988) and fast growth and defense (Herms and Mattson 1992). Plants that develop clonally have a distinct disadvantage in that they cannot easily escape from their root feeders, although a benet is that their resistance might remain systemically induced and that both root feeders and their predators keep track of the newly formed roots. The ultimate escape from root herbivores and pathogens is achieved when plants reach new territories where no similar root feeders occur. Both P. serotina and A. arenaria have been introduced in new areas where they have become important pests. Studies that

examine the presence and sensitivity of invasive plant species to root feeders in their original and new territories will elucidate to what extent escape from belowground enemies may contribute to invasiveness. By contrast, the biotic resistance hypothesis argues that native plant enemies present in the new territories may switch over to the introduced species thereby suppressing their invasive capacity (Keane and Crawley 2002). Biotic resistance may be due to absence of plant resistance, but the exotics might also miss effective secondary chemicals, or fail to recruit the natural enemies of the plant feeders in the new territories. Soil pathogens or root herbivores can also be introduced into new territories. Phytophthora cinnamomi that has been introduced into Australia causes the die back of many native species thereby changing the species composition of native forest ecosystems (Weste et al. 2002). The native plants were obviously not resistant against the introduced soil pathogens, but it might also be (partly) due to the escape of the introduced pathogen from its own natural enemies. The consequence of avoidance depends on the rate of changes in the environment. In coastal foredunes, where strong gradients occur in the abiotic environment (sand deposition, salt spray, leaching of calcium carbonate, and accumulation of organic matter), avoidance results in directional succession. Plants that have accumulated root herbivores and pathogens are replaced by resistant plants (Van der Putten and Van der Stoel 1988). By the time that the abundance of root feeders of the predecessor has dropped below threshold densities the abiotic environment has changed, so that the predecessors cannot return (Fig. 3). This is in contrast with cyclic succession (Fig. 2), which can only occur in an environment that is maintained in a relatively constant state by, for example, mowing or grazing, or that is reset by, for example, soil disturbance (Blomqvist et al. 2000) or re (Holah et al. 1997). In the eld, many plants will be protected by mycorrhizal fungi, rhizosphere bacteria, or other defensive organisms, so that it will be very difcult to provide evidence for tolerance. Nevertheless, whether tolerance for root feeders is directly or indirectly achieved, it may largely affect plant community structure, for example, because seedlings cannot establish when exposed to the density of root herbivores or pathogens as supported by the roots of their parents (Packer and Clay 2000). Therefore, tolerance of mature plants for root feeders will contribute to plant species diversity in a similar way as predicted by the Janzen-Connell hypothesis (Janzen 1970, Connell 1978) for aboveground specialist insect herbivores. Resistance is an important mechanism driving directional or cyclic succession. Later succession plants are resistant against the root feeders of their predecessors which provides an indirect advantage in competition (Van der Putten and Peters 1997). In a mixed vegetation, resistant plants may also become dominant,

September 2003

UNDERGROUND PROCESSES

2277

TABLE 1. Summary of proposed consequences of plant defense strategies against belowground herbivores and pathogens for spatiotemporal processes in natural plant communities. Spatiotemporal process in natural vegetation Plant defense Avoidance Succession Leads to directional succession in a fastchanging environment and to cyclic succession in a relatively stable environment. Ultimate escape is not possible in a successional sequence. Abundance and diversity Dispersal and vegetative expansion Invasion Plant life histories selected for avoidance might not be as effective in new territories, because of no or different natural enemies. Escape from natural enemies to new territories without their enemies, and no native enemies present, contributes to invasiveness of exotics. Tolerance patterns in native and new territories may differ due to exposure to new communities of root feeders.

Abundance is only tem- Avoidance selects for poral; spatiotemporal dispersal and vegetadiversity is stimulattive expansion. ed.

Escape

Plants remain local, so that there is no longterm escape possible.

Duration of escape depends on dispersal capacity and unpredictability of newly founded populations. Tolerance of parents enforces the need of dispersal for establishment of offspring (e.g., P. serotina).

Tolerance

Resistance

Slows down succession, May increase abunalbeit that it may dance, but on the constrain the estabother hand, tolerance lishment of offspring, of parents and suswhen susceptible in ceptibility of offgermination and esspring increases ditablishment phase. versity (e.g., P. serotina). Later successional Resistant plants can beplants resist root come abundant, feeders of their prewhich might negadecessors. Resistant tively affect plant plants may be more species diversity. persistent (also their offspring), so that they could slow down succession. Increases plant persistence; effectiveness might depend on early (less defended plants) vs. late successional (well defended plants) stage. May increase plant vigor (e.g. of A. arenaria by mycorrhizal fungi). Effects on successional process have not been investigated. Stimulates abundance, but acts against diversity. Information on direct defense against belowground root is rare.

Resistant plants have lower need for dispersal because there is no real need to escape from their enemies.

Direct defense

Resistance patterns in native area may not be indicative of resistance in new territories. Therefore, susceptible plants may become as invasive as resistant plants when released from exposure to native root feeders. Dispersal does not afDirect defense may fect direct defense, evolve in the new and vice versa, since territories, especially defense mechanism is when they provide intrinsic. costs and no benets. Data are lacking. Mycorrhizal fungi may provide indirect defense in new territories, since they are fairly nonspecic. Other indirect defenses have not yet been properly explored.

Special Feature

Indirect defense

Abundance may be en- There is a possible hanced, but also spetrade-off between cies diversity (dedispersal and indirect pending on which defense. This may be species [the rare or less of a problem for abundant ones] the clonal plants, since indirect defense is rethey remain connectsponding to). ed to the established plant, which might lead to, for example, induced defense.

whereas susceptible plants remain relatively rare (Klironomos 2002). Therefore, plant species diversity and the consequences of biodiversity loss need to be examined from the perspective of belowground interactions as well (see also Reynolds et al. 2003). Root feeders are susceptible for secondary plant compounds (Van Dam et al. 2003), but almost all studies on the evolution of this type of plant defense concerned aboveground plantherbivore interactions (Van der Putten et al. 2001). Root feeders may affect the distribution of plant defense compounds, and therefore

herbivore damage, over the entire plant (Bezemer et al. 2003), but it remains to be studied to what extent direct defense is under selection from belowground herbivores. There is compelling evidence that indirect defense belowground does exist, although very little information exists on the consequence of this defense trait for spatiotemporal processes in natural plant communities. Mycorrhizal fungi may protect plants against root feeders, as well as aboveground herbivores (Newsham et al. 1995, Brown and Gange 2002) and plant roots pro-

2278

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

duce unknown signals that attract entomopathogenic nematodes (Van Tol et al. 2001). However, the reliability of these defense strategies needs to be further explored. I conclude that avoidance and resistance as plant defense strategies against belowground herbivores and pathogens are major drivers of change in the species composition of natural plant communities (Table 1). The type of changes (directional vs. cyclic) depends on other abiotic and biotic environmental factors (Fig. 3). On the other hand, the persistence of plants in a community may depend on their defense belowground (Table 1). Effective defense (or resistance) will increase the capacity of plants to become dominant, and it may increase the length of the successional trajectory in which certain plant species can maintain themselves. Future studies on plant ecology, plant population dynamics, and plant community ecology need to include plant defense belowground in order to get a better understanding of plant community processes and tradeoffs between plant life history strategies.
ACKNOWLEDGMENTS

Special Feature

I thank Jeffrey A. Harvey, Arjen Biere, Wietse de Boer, Nicole M. van Dam, T. Martijn Bezemer, Anurag A. Agrawal, and two anonymous referees for their useful comments on previous versions of the manuscript. Ideas for the manuscript emerged from discussions with the members of the Multitrophic Interactions Department and the EU-EcoTrain project team (contract HPRN-CT-2002-00210 with the European Commission). This is NIOO 3157. LITERATURE CITED Agrawal, A. A. 2000. Overcompensation of plants in response to herbivory and the by-product benets of mutualism. Trends in Plant Science 5:309313. Andersen, D. C. 1987. Below-ground herbivory in natural communities: a review emphasising fossorial animals. Quarterly Review of Biology 62:261285. Augspurger, C. K. 1983. Offspring recruitment around tropical trees: changes in cohort distance with time. Oikos 40: 189196. Augspurger, C. K., and C. K. Kelly. 1984. Pathogen mortality of tropical tree seedlings: experimental studies of the effects of dispersal distance, seedling density, and light conditions. Oecologia 61:211217. Bardgett, R. D., and D. A. Wardle. 2003. Herbivore-mediated linkages between aboveground and belowground communities. Ecology 84:22582268. Bever, J. D. 1994. Feedback between plants and their soil communities in an old eld community. Ecology 75:1965 1977. Bever, J. D., K. M. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: the utility of the feedback approach. Journal of Ecology 85:561573. Bezemer, T. M., R. Wagenaar, N. M. Van Dam, and F. L. Wackers. 2003. Interactions between above- and below ground insect herbivores as mediated by plant defense system. Oikos 101:555562. Blomqvist, M. M., H. Olff, M. B. Blaauw, T. Bongers, and W. H. Van der Putten. 2000. Interactions between aboveand belowground biota: importance for small-scale vegetation mosaics in a grassland ecosystem. Oikos 90:582 598.

Boff, M. I. C., R. H. W. M. Van Tol, and P. H. Smits. 2002. Behavioural response of Heterorhabditis megidis towards plant roots and insect larvae. Biocontrol 47:6783. Brown, V. K., and A. C. Gange. 1989a. Differential effects of above- and below-ground insect herbivory during early plant succession. Oikos 54:6776. Brown, V. K., and A. C. Gange. 1989b. Herbivory by soildwelling insects depresses plant species richness. Functional Ecology 3:667671. Brown, V. K., and A. C. Gange. 1990. Insect herbivory below-ground. Advances in Ecological Research 20:158. Brown, V. K., and A. C. Gange. 1992. Secondary plant successionhow is it modied by insect herbivory. Vegetatio 101:313. Brown, V. K., and A. C. Gange. 2002. Tritrophic below- and above-ground interactions in succession. Pages 197222 in T. Tscharntke and B. A. Hawkins, editors. Multitrophic level interactions. Cambridge University Press, Cambridge, UK. Brust, G. E., and M. E. Barbercheck. 1992. Effect of dietary cucurbitacin-C on southern corn rootworm (Coleoptera, Chrysomelidae) egg survival. Environmental Entomology 21:14661471. Burdon, J. J. 1987. Diseases and plant population biology. Cambridge University Press, Cambridge, UK. Callaway, R., T. DeLuca, and W. Belliveau. 2000. Biologicalcontrol herbivores may increase competitive ability of the noxious weed Centaurea maculosa. Ecology 81:3259 3259. Carey, P. D., A. H. Fitter, and A. R. Watkinson. 1992. A elds-study using the fungicide benomyl to investigate the effect of mycorrhizal fungi on plant tness. Oecologia 90: 550555. Chase, J. M., M. A. Leibold, and E. Simms. 2000. Plant tolerance and resistance in food webs: community-level predictions and evolutionary implications. Evolutionary Ecology 14:289314. Clay, K. 1991. Fungal endophytes, grasses, and herbivores. Pages 199226 in P. Barbosa, V. A. Krischik, and C. G. Jones, editors. Microbial mediation of plantherbivore interactions. John Wiley and Sons, New York, New York, USA. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:13021310. Crutcheld, B. A., and D. A. Potter. 1995. Tolerance of coolseason turfgrasses to feeding by Japanese beetle and southern masked chafer (Coleoptera: Scarabaeidae) grubs. Journal of Economic Entomology 88:13801387. De Boer, W., P. J. A. Klein Gunnewiek, P. Lafeber, J. D. Janse, B. E. Spit, and J. W. Woldendorp. 1998a. Anti-fungal properties of chitinolytic dune soil bacteria. Soil Biology and Biochemistry 30:193203. De Boer, W., P. J. A. Klein Gunnewiek, and J. W. Woldendorp. 1998b. Suppression of hyphal growth of soil-borne fungi by dune soils from vigorous and declining stands of Ammophila arenaria. New Phytologist 138:107116. De Deyn, G. B., C. E. Raaijmakers, H. R. Zoomer, M. P. Berg, P. C. De Ruiter, H. A. Verhoef, T. M. Bezemer, and W. H. Van der Putten. 2003. Soil invertebrate fauna enhances grassland succession and diversity. Nature 422:711 713. De Rooij-van der Goes, P. C. E. M. 1995. The role of plantparasitic nematodes and soil-borne fungi in the decline of Ammophila arenaria L. Link. New Phytologist 129:661 669. De Rooij-Van der Goes, P. C. E. M., W. H. Van der Putten, and B. A. M. Peters. 1995. Effects of sand deposition on the interaction between Ammophila arenaria, plant parasitic nematodes and pathogenic fungi. Canadian Journal of Botany 73:11411150.

September 2003

UNDERGROUND PROCESSES

2279

DHertefeldt, T., and W. H. van der Putten. 1998. Physiological integration of the clonal plant Carex arenaria and its response to soil-borne pathogens. Oikos 81:229237. Dicke, M., and L. E. M. Vet. 1999. Plantcarnivore interactions: evolutionary and ecological consequences for plant, herbivore and carnivore. Pages 483520 in H. Olff, V. K. Brown, and R. H. Drent, editors. Herbivores between plants and predators. Blackwell Science, Oxford, UK. Dukes, J. S., and H. A. Mooney. 1999. Does global change increase the success of biological invaders? Trends in Ecology and Evolution 14:135139. Ehrlich, P. R., and P. H. Raven. 1964. Butteries and plants: a study in coevolution. Evolution 18:586608. Ennik, G. C., J. Kort, and C. F. Van de Bund. 1965. The clover cyst nematode (Heterodera trifolii Goffart) as the probable cause of white clover dieback in a sward. Journal of the British Grassland Society 20:258262. Gange, A. C. 2000. Arbuscular mycorrhizal fungi, Collembola and plant growth. Trends in Ecology and Evolution 15:369372. Gange, A. C., V. K. Brown, and L. M. Farmer. 1991. Mechanisms of seedling mortality by subterranean insect herbivores. Oecologia 88:228232. Gange, A. C., V. K. Brown, and G. S. Sinclair. 1993. Vesicular-arbuscular mycorrhizal fungia determinant of plant community structure in early succession. Functional Ecology 7:616622. Gaston, K. J. 1994. Rarity. Chapman and Hall, London, UK. Gormsen, D. 2001. Colonization processes of soil fauna and mycorrhizal fungi. Dissertation. University of Lund, Lund, Sweden. Grifn, D. W., C. A. Kellogg, and E. A. Shinn. 2001. Dust in the wind: long range transport of dust in the atmosphere and its implications for global public and ecosystem health. Global Change and Human Health 2:2033. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control and competition. American Naturalist 94:421425. Hedlund, K., and A. Augustsson. 1995. Effects of enchytraid grazing on fungal growth and respiration. Soil Biology and Biochemistry 27:905909. Herms, D. A., and W. J. Mattson. 1992. The dilemma of plants: to grow or defend. The Quarterly Review of Biology 67:283335. Hol, W. H. G., and J. A. Van Veen. 2002. Pyrrolizidine alkaloids from Senecio jacobaea affect fungal growth. Journal of Chemical Ecology 28:17631772. Holah, J. C., and H. M. Alexander. 1999. Soil pathogenic fungi have the potential to affect the co- existence of two tallgrass prairie species. Journal of Ecology 87:598608. Holah, J. C., M. V. Wilson, and E. M. Hansen. 1997. Impacts of a native root-rotting pathogen on successional development of old-growth Douglas r forests. Oecologia 111: 429433. Jaffee, B. A. 1996. Soil microcosms and the population biology of nematophagous fungi. Ecology 77:690693. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. American Naturalist 104:501528. Jarosz, A. M., and A. L. Davelos. 1995. Effects of disease in wild plant-populations and the evolution of pathogen aggressiveness. New Phytologist 129:371387. Jaworska, M., and D. Ropek. 1994. Inuence of host-plant on the susceptibility of Sitona lineatus (L.) (Coleoptera: Curculionidae) to Steineronema carpocapsae Weiser. Journal of Invertebrate Pathology 64:9699. Karban, R., and I. T. Baldwin. 1997. Induced responses to herbivory. University of Chicago Press, Chicago, Illinois, USA. Keane, R. M., and M. J. Crawley. 2002. Exotic plant invasions and the enemy release hypothesis. Trends in Ecology and Evolution 17:164170.

Kerry, B. R., and D. H. Crump. 1998. The dynamics of the decline of the cereal cyst nematode, Heterodera avenae, in four soils under intensive cereal production. Fundamental and Applied Nematology 21:617625. Klironomos, J. N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature 417:6770. Koppenhofer, A. M., B. A. Jaffee, A. E. Muldoon, and D. R. Strong. 1997. Suppression of an entomopathogenic nematode by the nematode-trapping fungi Geniculifera paucispora and Monacrosporium eudermatum as affected by the fungus Arthrobotrys oligospora. Mycologia 89:557. Kowalchuk, G. A., F. A. De Souza, and J. A. Van Veen. 2002. Community analysis of arbuscular mycorrhizal fungi associated with Ammophila arenaria in Dutch coastal sand dunes. Molecular Ecology 11:571581. Little, L. R., and M. A. Maun. 1996. The Ammophila problem revisited: a role for mycorrhizal fungi. Journal of Ecology 84:17. Maas, P. W. T., P. A. I. Oremus, and H. Otten. 1983. Nematodes (Longidorus sp. and Tylenchorhynchus microphasmis Loof) in growth and nodulation of sea buckthorn (Hippophae rhamnoides L.). Plant and Soil 73:141147. Maron, J. L. 1998. Insect herbivory above- and belowground: individual and joint effects on plant tness. Ecology 79: 12811293. Maron, J. L., and R. L. Jefferies. 1999. Bush lupine mortality, altered resource availability, and alternative vegetation states. Ecology 80:443454. Masters, G. J., and V. K. Brown. 1992. Plant-mediated interactions between two spatially separated insects. Functional Ecology 6:175179. Mills, K. E., and J. D. Bever. 1998. Maintenance of diversity within plant communities: soil pathogens as agents of negative feedback. Ecology 79:15951601. Mortimer, S. R., W. H. Van der Putten, and V. K. Brown. 1999. Insect and nematode herbivory below-ground: interactions and role in vegetation development. Pages 205 238 in H. Olff, V. K. Brown, and R. H. Drent, editors. Herbivores between plants and predators. Blackwell Science, Oxford, UK. Newsham, K. K., A. H. Fitter, and A. R. Watkinson. 1994. Root pathogenic and arbuscular mycorrhizal fungi determine fecundity of asymptomatic plants in the eld. Journal of Ecology 82:805814. Newsham, K. K., A. H. Fitter, and A. R. Watkinson. 1995. Arbuscular mycorrhizae protect an annual grass from root pathogenic fungi in the eld. Journal of Ecology 83:991 1000. Nordenhem, H., and H. H. Eidmann. 1991. Response of the pine weevil Hylobius abietis L (Col, Curculionidae) to host volatiles in different phases of its adult life-cycle. Journal of Applied Entomology 112:353358. Olff, H., B. Hoorens, R. G. M. de Goede, W. H. Van der Putten, and J. M. Gleichman. 2000. Small-scale shifting mosaics of two dominant grassland species: the possible role of soil-borne pathogens. Oecologia 125:4554. Oremus, P. A. I., and H. Otten. 1981. Factors affecting growth and nodulation of Hippophae rhamnoides L. ssp. rham noides in soils from two successional stages of dune formation. Plant and Soil 63:317331. Orr, C. C., and O. H. Newton. 1971. Distribution of nematodes by wind. Plant Disease Reporter 55:6163. Packer, A., and K. Clay. 2000. Soil pathogens and spatial patterns of seedling mortality in a temperate tree. Nature 404:278281. Price, P. W., C. E. Bouton, P. Gross, B. A. McPheron, J. N. Thompson, and A. E. Weis. 1980. Interactions among three trophic levels: inuence of plants on interactions between

Special Feature

2280

WIM H. VAN DER PUTTEN

Ecology, Vol. 84, No. 9

insect herbivores and natural enemies. Annual Review of Ecology and Systematics 11:4165. Reynolds, H. L., A. Packer, J. D. Bever, and K. Clay. 2003. Grassroots ecology: plantmicrobesoil interactions as drivers of plant community structure and dynamics. Ecology 84:22812291. Richardson, D. M., N. Allsopp, C. M. DAntonio, S. J. Milton, and M. Rejmanek. 2000. Plant invasionsthe role of mutualisms. Biological Reviews of the Cambridge Philosophical Society 75:6593. Roncadori, R. W. 1997. Interactions arbuscular mycorrhizas and plant parasitic nematodes in agro-ecosystems. Pages 101114 in A. C. Gange and V. K. Brown, editors. Multitrophic interactions in terrestrial systems. Blackwell Science, Oxford, UK. Roy, B. A., J. W. Kirchner, C. E. Christian, and L. E. Rose. 2000. High disease incidence and apparent disease tolerance in a North American Great Basin plant community. Evolutionary Ecology 14:421438. Schulze, E. D., and H. A. Mooney. 1993. Biodiversity and ecosystem function. Springer-Verlag, Berlin, Germany. Stanton, N. L. 1988. The underground in grasslands. Annual Review of Ecology and Systematics 19:573589. Strauss, S. Y., and A. A. Agrawal. 1999. The ecology and evolution of plant tolerance to herbivory. Trends in Ecology and Evolution 14:179185. Streitwolf-Engel, R., T. Boller, A. Wiemken, and I. R. Sanders. 1997. Clonal growth traits of two Prunella species are determined by co-occurring arbuscular mycorrhizal fungi from a calcareous grassland. Journal of Ecology 85:181 191. Strong, D. R. 1999. Predator control in terrestrial ecosystems: the underground food chain of bush lupine. Pages 577602 in H. Olff, V. K. Brown, and R. H. Drent, editors. Herbivores between plants and predators. Blackwell Science, Oxford, UK. Strong, D. R., J. L. Maron, P. G. Connors, A. Whipple, S. Harrison, and R. L. Jefferies. 1995. High mortality, uctuation in numbers, and heavy subterranean insect herbivory in bush lupine, Lupinus arboreus. Oecologia 104:85 92. Strong, D. R., A. V. Whipple, A. L. Child, and B. Dennis. 1999. Model selection for a subterranean trophic cascade: root-feeding caterpillars and entomopathogenic nematodes. Ecology 80:27502761. Thaler, J. S., A. L. Fidantsef, S. S. Duffey, and R. M. Bostock. 1999. Trade-offs in plant defense against pathogens and herbivores: a eld demonstration of chemical elicitors of induced resistance. Journal of Chemical Ecology 25:1597 1609. Van Dam, N. M., J. A. Harvey, F. L. Wackers, T. M. Bezemer, W. H. Van der Putten, and L. E. M. Vet. 2003. Interactions between aboveground and belowground induced responses against phytophages. Basic and Applied Ecology 4:6377. Van der Meijden, E., M. Wijn, and H. J. Verkaar. 1988. Defense and regrowth: alternative plant strategies in the struggle against herbivores. Oikos 51:355363. Van der Putten, W. H., and B. A. M. Peters. 1997. How soilborne pathogens may affect plant competition. Ecology 78: 17851795. Van der Putten, W. H., and C. D. Van der Stoel. 1998. Plant parasitic nematodes and spatio-temporal variation in natural vegetation. Applied Soil Ecology 10:253262. Van der Putten, W. H., C. Van Dijk, and B. A. M. Peters. 1993. Plant-specic soil-borne diseases contribute to succession in foredune vegetation. Nature 362:5356.

Van der Putten, W. H., C. Van Dijk, and S. R. Troelstra. 1988. Biotic soil factors affecting the growth and development of Ammophila aenaria. Oecologia 76:313320. Van der Putten, W. H., L. E. M. Vet, J. A. Harvey, and F. L. Wackers. 2001. Linking above- and belowground multi trophic interactions of plants, herbivores, pathogens and their antagonists. Trends in Ecology and Evolution 16:547 554. Van der Stoel, C. D. 2001. Specicity, pathogenicity and population dynamics of the endoparasitic nematode Heterodera arenaria in coastal foredunes. Dissertation. Wageningen University, Wageningen, The Netherlands. Van der Stoel, C. D., W. H. Van der Putten, and H. Duyts. 2002. Development of a negative plantsoil feedback in the expansion zone of the clonal grass Ammophila arenaria following root formation and nematode colonisation. Journal of Ecology 90:978988. Van der Veen, A. 2000. Competition in coastal sand dune succession cause or mechanism? Dissertation. University of Groningen, Groningen, The Netherlands. Van Loon, L. C., P. A. H. M. Bakker, and C. M. J. Pieterse. 1998. Systemic resistance induced by rhizosphere bacteria. Annual Review of Phytopathology 36:453483. Van Tol, R., A. T. C. van der Sommen, M. I. C. Boff, J. van Bezooijen, M. W. Sabelis, and P. H. Smits. 2001. Plants protect their roots by alerting the enemies of grubs. Ecology Letters 4:292294. Verschoor, B. C. 2002. Carbon and nitrogen budgets of plantfeeding nematodes in grasslands of different productivity. Applied Soil Ecology 20:1525. Verschoor, B. C., R. G. M. De Goede, and L. Brussaard. 2002. Do plant parasitic nematodes have differential effects on the productivity of a fast- and a slow-growing grass species? Plant and Soil 243:8190. Verschoor, B. C., R. G. M. De Goede, F. W. De Vries, and L. Brussaard. 2001. Changes in the composition of the plant-feeding nematode community in grasslands after cessation of fertiliser application. Applied Soil Ecology 17: 117. Vitousek, P. M., L. R. Walker, L. D. Whitaker, D. MullerBombois, and P. A. Matson. 1987. Biological invasions by Myrica faya alters ecosystem development in Hawaii. Science 238:802804. Vrieling, K., and C. A. M. Van Wijk. 1994. Estimating costs and benets of the pyrrolizidine alkaloids of Senecio jacobea under natural conditions. Oikos 70:449454. Wardle, D. A. 2002. Communities and ecosystems: linking the aboveground and belowground components. Princeton University Press, Princeton, New Jersey, USA. Weller, D. M., L. S. Thomashow, and R. J. Cook. 1995. Biological control of soil-borne pathogens of wheat: benets, risks and current challenges. Pages 149160 in H. M. Hokkanen and J. M. Lynch, editors. Biological control: benets and risks. Plant and Microbial Biotechnology Research Series 4. Cambridge University Press, Cambridge, UK. Weste, G., K. Brown, J. Kennedy, and T. Walshe. 2002. Phytophthora cinnamomi infestationa 24-year study of vegetation change in forests and woodlands of the Grampians, Western Victoria. Australian Journal of Botany 50:247 274. Yang, D., F. Plante, L. Bernier, Y. Piche, M. Dessureault, G. Laamme, and G. B. Ouellette. 1993. Evaluation of a fungal antagonist, Phaeotheca dimorphosphora, for biological control of tree diseases. Canadian Journal of Botany 71: 426433. Zoon, F. C., S. R. Troelstra, and P. W. T. Maas. 1993. Ecology of the plant-feeding nematode fauna associated with Sea buckthorn (Hippophae rhamnoides L. ssp. rhamnoides) in different stages of dune succession. Fundamental and Applied Nematology 16:247258.

Special Feature

Ecology, 84(9), 2003, pp. 22812291 2003 by the Ecological Society of America

GRASSROOTS ECOLOGY: PLANTMICROBESOIL INTERACTIONS AS DRIVERS OF PLANT COMMUNITY STRUCTURE AND DYNAMICS
HEATHER L. REYNOLDS,1 ALISSA PACKER, JAMES D. BEVER,
AND

KEITH CLAY

Department of Biology, Indiana University, Bloomington, Indiana 47405 USA

Abstract. A growing body of research on plantmicrobe interactions in soil is contributing to the development of a new, microbially based perspective on plant community ecology. Soil-dwelling microorganisms are diverse, and interactions with plants vary with respect to specicity, environmental heterogeneity, and tness impact. Two microbial processes that may exert key inuences on plant community structure and dynamics are microbial mediation of niche differentiation in resource use and feedback dynamics between the plant and soil community. The niche differentiation hypothesis is based on observations that soil nutrients occur in different chemical forms, that different enzymes are required for plant access to these nutrients, and that soil microorganisms are a major source of these enzymes. We predict that plant nutrient partitioning arises from differential associations of plant species with microbes able to access different nutrient pools. Feedback dynamics result from changes in the soil community generated by the specicity of response in plant microbe interactions. We suggest that positive feedback between plants and soil microbes plays a central role in early successional communities, while negative feedback contributes both to species replacements and to diversication in later successional communities. We further suggest that plantmicrobe interactions in the soil are an important organizing force for large-scale spatial gradients in species richness. The relative balance of positive feedback (a homogenizing force) and negative feedback (a diversifying force) may contribute to observed latitudinal (and altitudinal) diversity patterns. Empirical tests of these ideas are needed, but a microbially based perspective for plant ecology promises to contribute to our understanding of long-standing issues in ecology, and to reveal new areas of future research.
Key words: diversity; mycorrhizal fungi; N-xing bacteria; niche partitioning; plant ecology; soilborne pathogens; succession.

Special Feature

INTRODUCTION Ecological theory recognizes a variety of abiotic and biotic factors that shape the structure and dynamics of terrestrial plant communities. Climate, age, environmental harshness, area, isolation, disturbance, environmental heterogeneity, and biotic interactions may all play key roles in determining plant community composition and diversity in space and time (Tilman and Pacala 1993, Rosenzweig 1995). Much of plant ecological theory has been dominated by debate over the relative importance of these factors, manifest in controversy over plant strategy theory and the relationships between competition intensity, species diversity, and habitat productivity (Reynolds 1999, Aarssen 2001). Among the biological forces structuring plant communities, plantmicrobe interactions have received little attention (Thompson et al. 2001). This is not surprising, given the overwhelming diversity of soil microbes (Torsvik et al. 1994), the technical challenges involved in their study, the tendency to view microbial associates as simply extensions of the plant and their outright exclusion from experimental designs through
Manuscript received 15 May 2002; revised 26 August 2002; accepted 11 September 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256. 1 E-mail: hreynold@bio.indiana.edu

use of articial soil mixes or soil amendments that result in depleted microbial communities. In recent years, however, a growing body of ecological research on plantmicrobe interactions in soil has developed. A new, microbially based perspective on plant community structure and dynamics is emerging from this work. In this paper, we review the diversity of plantmicrobe interactions in soil and discuss the costs and benets to each partner. We discuss the role of environmental heterogeneity and host specicity in generating a continuum of positive and negative effects, both direct and indirect. Two processes, mediation of niche differentiation in plant resource use and feedback dynamics between the plant and soil community, emerge from these fundamentals as key microbial inuences on plant community structure and dynamics. We discuss the evidence for these processes and their applications to understanding temporal and spatial vegetation patterns. We conclude with a look toward how a microbial perspective can inform future research directions in plant ecology. BASICS
OF

PLANTMICROBE INTERACTIONS

Diversity and function

Soil-dwelling microorganisms encompass a diversity of phylogenetic groups and all three major functional

2281

2282

HEATHER L. REYNOLDS ET AL.

Ecology, Vol. 84, No. 9

groups (producer, consumer, and decomposer). Genetic diversity can be exceptionalseveral thousand genomes per gram of soil (Torsvik et al. 1990). Yet our knowledge of microbial species diversity is still in its infancy (Tate 1997). Interactions between soil microbes and plants span the range from mutualistic to pathogenic. As decomposers, soil microbes are indirectly responsible for the bulk of terrestrial vegetations annual nutrient demand (Schlesinger 1991). In turn, plant matter is the major source of photosynthetically xed carbon for decomposers. Yet microbes and plants also compete for soil nutrients (Hodge et al. 2000 a), making their relationship simultaneously mutualistic and competitive (Harte and Kinzig 1993). In contrast, nitrogenxing bacteria and mycorrhizal fungi enhance host plant tness by direct provisioning of mineral resources or by providing protection against other pests. Nonmycorrhizal fungi, various rhizosphere bacteria, protozoa, and nematodes have also been shown to protect plants from soilborne enemies like fungi, bacteria, actinobacteria, protozoa, nematodes, and viruses (Lavelle and Spain 2001). Such pathogens are responsible for damping-off diseases, vascular wilts, and root rots (Katan 1991).

Special Feature

Continuum of effects: environmental dependence

Many of the effects of microbes on plants are dependent upon environmental conditions. For example, the benets that plants receive from nutritional mutualisms with soil microbes (e.g., nitrogen-xing bacteria and mycorrhizal fungi) depend upon the availability of essential soil minerals. Available phosphorus (P) is often the primary determinant of plant dependence on arbuscular mycorrhizal (AM) fungi, such that the effect of the interaction can be negative under high P conditions (Bethlenfalvay et al. 1982). The benet that plants receive from protective mutualisms is also environment dependent, as the benet depends upon the presence of enemies. A similar environmental dependence is inherent in the interactions of plants and soil pathogens. Major groups of soil pathogens (e.g., Pythium) have high pathogenesis under a limited range of environmental conditions (e.g., wet and warm). The ability of plants to tolerate or ward off pathogens is also dependent upon environmental factors such as nutrient and light availability. Thus, the impact of soil communities on plant community processes will also be a function of environmental factors, presenting a particular challenge to investigations of the community consequences of plantsoilmicrobe interactions.

of association (i.e., the ability to form specic associations) and the specicity of the plant and microbe responses to their association (i.e., the dependence of relative tness on specic associations, as estimated by specicity of growth responses). While these two aspects of specicity are not completely independent, they are not necessarily collinear either. For example, interactions that have relatively high specicity of association, such as the association between ectomycorrhizal host plants and ectomycorrhizal fungi, might also be expected to show relatively high specicity of response. However, this does not imply that plant-microbe interactions that show low specicity of association also show low specicity of response. The interaction of plants and AM fungi shows relatively low specicity of association, but the response of plants to individual species of AM fungi can vary greatly depending on the plantAM fungal combination (Van der Heijden et al. 1998). Similarly, the relative growth rates of AM fungi also depend greatly on the identity of the plant with which they are associated (Eom et al. 2000; Bever 2002). Specicity of infection and of plant response is also a well-known characteristic of soil pathogens (Bruehl 1987, Agrios 1997). In fact, when measured, soil microbes commonly show specicity of response to plant species and specicity of growth effects on those hosts. While the specicity of association is important for establishing the potential relationships, the specicity of response determines the community dynamics resulting from these associations (Bever 1999). MICROBIAL MECHANISMS FOR THE MAINTENANCE OF DIVERSITY Given a local species pool, the species diversity of a given area depends on the existence of mechanisms that prevent a competitive dominant from excluding all other species. Recognizing that competition occurs over resources, coexistence mechanisms may be categorized on the basis of whether (Aarssen 1984) (1) competition is avoided in the rst place, through resource partitioning (e.g., rooting depth differences), or (2) competitive exclusion is avoided, despite overlap in resource use (e.g., competitive reversal). Here, we discuss new facets to these coexistence mechanisms that emerge from a microbially based perspective.

Microbial mediation of resource partitioning

We focus here on soil nitrogen (N) and P, the two most common limiting nutrients in terrestrial ecosystems (Chapin et al. 1986). These nutrients exist in a variety of inorganic (e.g., ammonium, phosphate) and organic (e.g., amino acids, nucleic acids) pools made available to plants through the action of soil enzymes (e.g., proteases, ribonucleases), the bulk of which are thought to come from bacteria and fungi (Tabatabai and Dick 2002). If microbes or their enzymatic activities are differentially associated with plant species, a

Specicity of association vs. specicity of response

Mutualistic and pathogenic associations between plants and soil microbes are known to range in their specicity from highly specic associations between orchids and their mycorrhizal fungal symbionts to the cosmopolitan associations of plant roots with rhizosphere bacteria. It is useful to distinguish the specicity

September 2003

UNDERGROUND PROCESSES

2283

Special Feature

FIG. 1. (a) N partitioning. Different plant species access different forms (e.g., NO3 vs. NH4 vs. amino acids) or sources (e.g., NH4 from N2, proteins, urea or chitin) of N. (b) A similar model can be envisioned for P nutrition.

kind of resource partitioning may arise in which different plant species access different pools, thus avoiding competition for the same pools (Fig. 1). In this way, plants may experience a single nutrient axis (e.g., N), as a diversity of axes, thereby vastly increasing the number of niches available for competing plant species. On the surface, this is simply the standard niche-partitioning model, but the niche partitioning is enabled by soil microbes. Under the simplest form of the microbial mediation hypothesis, a single microorganism is able to produce the entire range of enzymes and different plant species cause this super microbe to express different enzymes. However, inevitable physiological trade-offs and the diversity of microorganisms with distinct enzymatic capabilities make this super microbe hypothesis untenable. A more realistic hypothesis is that plant

species associate with different groups of microbes, so gaining differential access to nutrient pools. Conrmation of this hypothesis would have the important implication that plant species coexistence depends upon soil microbial diversity and microbial dynamics. Evidence for microbially mediated resource partitioning.Evidence for less obvious forms of microbially mediated nutrient partitioning than that between plants with N-xing symbionts and plants that rely on soil N awaits future studies. Evidence for partitioning of plant species into ammonium, nitrate, and amino acid specialists is accumulating in a range of ecosystems (McKane et al. 2002, Miller and Bowman 2002), although a specic role for microbes has not been established. Nor do we know whether plant species, via microbial associations, differentially access complex organic pools, such as protein, chitin, or nucleic acids.

2284

HEATHER L. REYNOLDS ET AL.

Ecology, Vol. 84, No. 9

We do know that microbial populations are highly elevated in the rhizospheres surrounding plant roots (Paul and Clark 1996), where they may be inuenced by the quality and quantity of substances that plant roots continuously exude (Rovira 1969, Hamilton and Frank 2001). Thus, the possibility arises that specic plantmicrobe and/or plantmicrobial-enzyme associations could develop. Differential plantmicrobe associations have been observed (Grayston et al. 1998, Westover and Bever 2001). Yet, with the obvious exception of N xation, the potential relevance of such associations for plant N and P partitioning has not been investigated. Mycorrhizal mediation of resource partitioning. Mycorrhizae are potentially key players in this model of microbially mediated N and P partitioning. All forms of mycorrhizal fungi exhibit phosphatase activity (Marschner 1995). In otherwise sterile soil, AM grasses showed increased growth and P uptake compared to noninfected plants when supplied with such organic P sources as ribonucleic acid (RNA) (Jayachandran et al. 1992) and phytate (Jayachandran et al. 1992, Tarafdar and Marschner 1994). Ecto- and ericoid mycorrhizae are capable of breaking down complex organic N and taking up dissolved organic N such as amino acids (Ahmad and Hellebust 1991), allowing plants direct access to organic N (Turnbull et al. 1995, 1996). A number of studies have now also demonstrated enhanced acquisition of inorganic and organic N sources by AM plants (Michelsen et al. 1996, Hawkins et al. 2000). A key issue is whether, like ecto- and ericoid mycorrhizae, AM fungi are able to directly access complex organic N forms (e.g., chitin) or whether they simply enhance the ability of plants to compete with free-living heterotrophic microbes for the products of organic matter mineralization. Recent studies have been conicting (Hodge et al. 2000b, 2001). Might mycorrhizae facilitate differential access of plant species to different pools of N and P? Prior research with AM fungi has focused almost exclusively on a few AM fungal species in a single genus (Glomus). Yet substantial host specicity of AM fungi with plant hosts has been demonstrated (Bever et al. 1996, Van der Heijden et al. 1998). This host specicity, in combination with AM fungal diversity, has been implicated in the maintenance of plant community diversity (Van der Heijden et al. 1998, Bever et al. 2001), although the specic mechanism is not known. The potential for microbially mediated niche partitioning may be even greater in ecto- and ericoid mycorrhizae, which are thought to be more diverse than AM fungi and to exhibit greater host specicity. For example, isolates of four ectomycorrhizal species from northern Australian forests had different abilities to use a range of amino acids, protein and inorganic N sources and this was affected by host plant identity (Turnbull et al. 1995). More studies of this sort are needed for mycorrhizae in general.

Positive vs. negative feedback dynamics

A second, qualitatively distinct manner in which interactions with soil microbes can contribute to the maintenance of diversity in plant communities is through feedbacks on plant growth resulting from changes in the composition of the soil community. While feedbacks may be involved in microbially mediated resource partitioning, we focus here on feedback dynamics involving plants with overlapping resource use. Feedbacks result from the community dynamics generated by the specicity of response in plantmicrobe interactions. Because growth rates of microbes are known to be host dependent, the composition of the soil community will likely change due to the identity of the local host plant. Moreover, as the growth promotion of soil microbes is also host-specic, the change in the composition of the soil microbes will likely alter the relative performance and local abundance of different plant species. The feedback then represents the dynamics of the net direct effects of the microbial community on individual plant types. Expected dynamics of feedback.The growth consequences of soil community feedback can be positive or negative. For example, if the presence of plant species A causes an accumulation of microbes that specically benet that host, then the relative performance of plant A with its soil community will improve over time. This positive feedback dynamic will be expected to reinforce the initial abundance of plant A and lead to the loss of other plant types from the local community, at least on a local scale (Bever et al. 1997, Bever 1999). In the case of negative feedback, a given host performs relatively worse with its own soil community over time. This can result from host-specic degradation of the soil community, such as might result from the accumulation of species-specic pathogens. However, negative feedback can also result from host-specic changes in the composition of the soil community that improve the growth of a second plant species (i.e., an indirect facilitation mediated by the change in the microbial community). The composition of the mycorrhizal fungal community, for example, might change in a manner that improves the growth of a competing plant species (Bever 1999, 2002). Negative feedback will prevent any single plant species from dominating a site and thereby contribute to the coexistence of competing plant species (Bever et al. 1997). Evidence for positive feedback.Positive feedback is known to play an important role in plantsoil community interactions. Positive feedback dynamics are evident in the limited success of plant establishment in the absence of particular symbionts, as illustrated by the history of pine forests in the tropics. Attempts at growing pine failed until trees were inoculated with compatible ectomycorrhizal fungi, and now these pines are increasingly perceived as a potentially invasive spe-

Special Feature

September 2003

UNDERGROUND PROCESSES

2285

cies (Rejmanek and Richardson 1996). A similar pattern has been found with invasive Russian olive (Elaeagnus angustifolia) and its N-xing Frankia symbiont (Richardson et al. 2000). Similarly, the success of plants that are dependent upon AM fungi or N-xing bacteria are dependent upon the initial abundance of their symbionts (Medve 1984, Larson and Siemann 1998). Local-scale positive feedback is expected to maintain sharp boundaries between patches (Molofsky et al. 2001), and contribute to the maintenance of boundaries between vegetation dominated by plants associated with arbuscular vs. ectomycorrhizal fungi. Evidence for negative feedback.There is accumulating evidence of negative feedbacks important role in structuring plant communities. In agricultural settings, the accumulation of species-specic soil pathogens drives the rotation of crops. Much of the evidence for negative feedback in unmanaged communities has come from efforts to test the Janzen-Connell hypothesis that high tropical forest tree diversity results from negative density-dependent mortality resulting from species-specic seed predation or herbivory (Janzen 1970, Connell 1971). Seedling mortality in tropical forests has been repeatedly found to increase with the density of conspecics (e.g., Wills et al. 1997, Harms et al. 2000) and with proximity to mature conspecics (Augspurger 1992, Condit et al. 1994). Current evidence suggests accumulation of soilborne pathogens as the causal mechanism for these effects (Augspurger and Kelly 1984, Augspurger 1988). In temperate forests of eastern North America, negative feedbacks may drive the reciprocal replacement and coexistence of beech and maple codominants (Fox 1977, Woods 1979), although the soil microbial mechanism has not been demonstrated. The high mortality of black cherry (Prunus serotina) seedlings near adult conspecics and at high seedling densities has been shown to result from accumulation of soil pathogens in the genus Pythium (Packer and Clay 2000). There is also accumulating evidence of negative soil community feedbacks playing an important role in grassland communities. In greenhouse assays of soil community feedback within an old eld community in North Carolina, negative feedback was found in nine of the 14 pairwise comparisons tested (Bever 1994, Bever et al. 1997). In this system, several complementary microbial mechanisms of negative feedback have been identied, including the accumulation of hostspecic soil pathogens in the genus Pythium (Mills and Bever 1998, Westover and Bever 2001), host-specic shifts in the composition of rhizosphere bacteria (Westover and Bever 2001), and host-specic changes in the composition of the AM fungal community (Bever 2002). Negative feedbacks have also been found to drive plant community dynamics within sand dune communities (Van der Putten et al. 1993). Again, complementary microbial mechanisms have been identied, with accumulation of root feeding nematodes

working synergistically with pathogenic fungi to decrease plant growth (Van der Putten et al. 1990, Van der Putten and Troelstra 1990). Soil communities have also been suggested as a driver of plant community dynamics in tallgrass prairies (Holah and Alexander 1999) and European grazed grasslands (Olff et al. 2000). In the latter system, the negative soil community feedbacks have been implicated as a cause of a shifting mosaic of grassland dominants (Olff et al. 2000), a pattern predicted by spatial simulations of this process (Molofsky et al. 2002). LARGE-SCALE VEGETATION PATTERNS In the previous section, we discussed how microbially mediated resource differentiation and feedback have implications for the maintenance of plant community diversity at a local scale. Here, we discuss how these processes may contribute to large-scale vegetation patterns. We hypothesize that the relative importance of positive versus negative feedback processes shifts over temporal and spatial gradients, with positive feedback processes dominating earlier in succession and at high latitudes and altitudes and negative feedback processes dominating later in succession and at low latitudes and altitudes. This hypothesis could be tested in natural communities by examining plantsoil feedback over successional and spatial gradients.

Special Feature

Temporal patterns
Vegetation change over time, or succession, is characterized by sequential species replacements, typically accompanied by changes in community diversity. Classic successional models focus on vegetation-based mechanisms of species replacement, differing in emphasis on recruitment limitation, life history strategy, environmental modication, and competitive exclusion (Connell and Slatyer 1977, Tilman 1988). The role of plantmicrobe interactions in such models is principally limited to that of N-xing associations in facilitation. Yet evidence for microbial inuences on succession can be found in early work, which established that many soil microbes vary predictably with vegetation through succession (Tresner et al. 1954, Wohlrab et al. 1963). More recently, mycorrhizae (Janos 1980, Read 1991) and soil pathogens (Van der Putten et al. 1993, Clay and Van der Putten 1999) have been recognized as potentially important drivers of succession. Here, we consider plant-microbe interactions in species replacements and changes in community diversity through succession, using the context of feedback and resource differentiation.

Species replacements Positive feedback.Soil mutualists and pathogens can be limiting in the very beginning stages of both primary and secondary successions, favoring asymbiotic, ruderal, life history strategies (Janos 1980, Sprent 1993), particularly in secondary successions,

2286

HEATHER L. REYNOLDS ET AL.

Ecology, Vol. 84, No. 9

FIG. 2. A conceptual model of how the relative strength of positive and negative feedback changes over successional time and over latitudinal gradients. The exact shape of the curves does not matter. The important point is that there is a switchover point at some intermediate successional time or latitude where the strength of negative feedback exceeds that of positive feedback.

where soil nutrient availability is comparatively higher. As mutualists and their plant hosts arrive at a site, positive feedback dynamics cause populations of both to build (Fig. 2). The importance of positive feedback dynamics between plants and N-xing microbes in primary successions is well recognized. In the classic example of primary succession at Glacier Bay, N-xers such as Dryas are among the rst vascular plants to colonize (Crocker and Major 1955). Similar dynamics are expected between host plants and mycorrhizal fungi. Janos (1980) envisioned a shift from nonmycorrhizal to obligately mycorrhizal plants from early to late secondary succession, with low levels of facultatively mycorrhizal associations throughout. In sand dune and many other successional communities there is also a shift later in succession from herbaceous plant species involved in obligate symbioses with AM fungi to woody species involved in obligate symbioses with ecto- or ericoid mycorrhizae (Smith and Read 1997), coincident with a shift in predominance of inorganic versus organic N (Read 1993). In the case of N-xing species, death and decomposition of N-xing species results in soil building over time. Thus, the benets of N-xation are not restricted to the plant host (Chapin et al. 1994), and a positive feedback dynamic transitions to facilitation of additional plant species. N-xing associations may play a similar role in secondary successions on infertile soil. In contrast to N-xing associations, mycorrhizal associations are relatively ubiquitous and act to increase access to rather than add nutrients to soil. The availability of P and other nutrients often decrease in late succession (Walker and Syers 1976). The expectation might therefore be of a strong role for mycorrhizae throughout much of succession, after an initial recruitment-limited (and total N-limited, for primary succession) phase. Negative feedback.Negative feedback dynamics are expected to be less important during the initial stage

of succession, which is characterized by harsher conditions and lower host densities than are favorable to most disease organisms. As plant host densities build and modify the abiotic environment, conditions should become more favorable to soil pathogens, leading to an increasing role for negative feedback in driving species replacements in succession (Fig. 2). Van der Putten et al. (1993) showed that soilborne diseases can drive successional change in a foredune community; individuals showed reduced biomass in the soil of their successors, but not in the soil of their predecessors. The traits of early successional species themselves may make them particularly vulnerable to negative feedback by soil pathogens. Rapid growth is a well-known trait of ruderal species, and it is also established that rapid growth trades off with allocation to herbivore defense (Coley et al. 1985, Poorter 1990). Given the assumption that belowground patterns of growth and defense mirror aboveground patterns, ruderal species, by their own success, increase the likelihood that they will be replaced by slower growing species, better defended against pathogens. Early and late successional species differ in quality as well as quantity of antiherbivore defense; defenses of ruderals tend to be against generalist herbivores while defenses of climax species tend to be against specialists (Coley et al. 1985). Extrapolating belowground once again, this suggests that specialization of plant defenses against soil pathogens will increase over successional time. Microbially mediated resource differentiation.Microbially mediated resource partitioning might play a role in species replacements if the forms of N or P change through succession. Gorham et al. (1979) implicated fungal symbionts in such sequential partitioning. Indeed, the example given above, of a shift from arbuscular to ecto- or ericoid mycorrhizal plant species with a change in inorganic to organic forms of N is an illustration of this process. Other shifts in forms of nutrients over succession (e.g., nitrate to ammonium, or protein to chitin) would provide the opportunity for

Special Feature

September 2003

UNDERGROUND PROCESSES

2287

additional species replacements on the basis of soil resource specialization.

Community diversity Feedback.Positive feedback, operating alone, would be expected to lead to monocultures, or even, in the absence of soil building or temporal shifts in forms of nutrients (e.g., inorganic to organic N), arrested successions. It has been suggested, for example, that higher host specicity of ecto- compared to endomycorrhizae leads to dominance by ectomycorrhizal species, to the exclusion of endomycorrhizal species (Connell and Lowman 1989). Even mycorrhizal networks (where benets are shared by many hosts) can lead to lower diversity if one species in the network is the dominant sink for nutrients (Connell and Lowman 1989, Allen and Allen 1990). In contrast, the process of negative feedback through the soil community can generate a pattern of increased species diversity over successional time. Plant life histories and pathogen colonization opportunities vary with succession and offer two explanations for the increases in plant diversity during early succession. As soil pathogens remove weakly defended plants, only better defended plants remain. Later successional species with greater investment in defense are less susceptible to generalist pathogens and only vulnerable to attack by more specialized pathogens that are able to overcome their more sophisticated plant defenses. Increased specialization of pathogens should lead to reciprocal negative feedback that promotes plant species coexistence. Furthermore, opportunities for pathogen dispersal to a site increase over succession, leading to greater pathogen diversity and increasing the chances for reciprocal negative feedback. Microbially mediated resource differentiation. Community complexity increases during succession as plant, microbial, and animal species diversity increase, and this leads to increases in the quantity and kinds of nutrient pools. This is expected to be particularly true for primary succession, but could also operate to a lesser extent in certain kinds of secondary succession, such as those initiating after severe re, where much of the above- and belowground resources may be volatilized and homogenized. We suggest that as the diversity of microbes and nutrient inputs increase over succession, the opportunity for microbially mediated differentation in resource use increases, promoting increased plant community diversity over succession. Supporting this idea, ectomycorrhizal infectiveness and diversity has been found to increase over a successional gradient (Boerner et al. 1996). Microbe-mediated niche diversication will also be affected by the specicity of the associations between plants and microbes. In the case of soil pathogens and mycorrhizal fungi, associations typically exhibit higher levels of specicity as succession proceeds. If the same pattern is true for the microorganisms involved in me-

diating resource differentiation, then we would predict greater opportunities for species coexistence (i.e., greater specialization of resource niches) as succession proceeds. SPATIAL PATTERNS Large-scale spatial patterns in the organization and diversity of ecological communities are well known, and the underlying causes are much debated (Rosenzweig 1995). Species diversity, primary productivity, and environmental equitability all tend to increase with decreasing latitude (or altitude), corresponding to increasing mean temperature and precipitation (Waide et al. 1999, Mittelbach et al. 2001). Correlated changes in multiple environmental variables make the underlying mechanisms difcult to assign, but a variety of abiotic and biotic factors have been proposed as key drivers (Waide et al. 1999). We suggest that plant microbe interactions in the soil, particularly feedback processes, may also be an important driver of largescale spatial gradients. We predict that the relative balance of positive and negative feedback shifts towards negative feedback as productivity increases with decreasing latitude or altitude (Fig. 2). Relatively few empirical studies have attempted to assess the relative strength of positive and negative feedback in the soil over large-scale spatial gradients but there is a rich conceptual and theoretical literature. Further, results from a number of studies indirectly contribute to our understanding of microbial interactions with plants over large spatial scales.

Special Feature

Mutualism and positive feedback

Positive interactions may be particularly advantageous in extreme environments, such as those at high latitude or altitude (Sanders 1968), and facilitation has been associated with habitats of high abiotic stress (Greenlee and Callaway 1996, Choler et al. 2001). Adaptations to physical stresses often arise from mutualistic symbioses with microbes, and these symbioses are often characterized by the host-specicity required to generate positive feedback. It has been hypothesized that as terrestrial productivity increases and light becomes increasingly limiting, the high energetic costs of N xation may outweigh its benets (Gutschick 1981). If this is generally true of mutualistic symbioses with microbes (at least those requiring signicant carbon inputs from the plant host) then it would promote a gradient of increasing importance of positive feedback between plants and soil microbes with increasing latitude or altitude (and decreasing primary productivity).

Pest pressure and negative feedback

Insofar as latitude and altitude are surrogates for temperature and precipitation, pest and pathogen pressure should increase as productivity increases with decreasing latitude or altitude (Weltzien 1972, Givnish 1999).

2288

HEATHER L. REYNOLDS ET AL.

Ecology, Vol. 84, No. 9

For example, damping-off fungi, which are major sources of seedling mortality for tropical tree species (Augspurger and Kelly 1984), thrive in warm, moist habitats. Levin (1975) concluded, based on a review of much evidence, that pathogen pressure is greater in the tropics. Additional evidence for a latitudinal gradient in pathogen pressure comes from agriculture. It is well established that plant species diversity, which generally decreases with latitude, may reduce the incidence and severity of disease (Elton 1958, Knops et al. 1999), but, when diversity is held constant, a latitudinal effect is evident. For example, disease diversity of soybean, the United States second largest cash crop and grown over a wide latitudinal range, was shown to be significantly correlated with latitude, with the largest number of soybean diseases occurring further south (Yang and Feng 2001). On a worldwide basis Septoria tritici blotch of wheat decreased linearly with increasing distance from the equator, although S. nodorum blotch increased with increasing latitude (Leath et al. 1993). In order to lead to diversity-maintaining negative feedback, pest pressure must operate in a host-specic way. There is evidence that soil pathogens generate negative feedback in the tropics (Augspurger 1988). Consistent with the mechanism of negative feedback, density-dependence in the survival and growth of many tropical tree species has also been reported (Wills et al. 1997, Harms et al. 2000), although a recent study found that similar proportions of tree species are affected by density-dependent mortality in temperate vs. tropical forests (HilleRisLambers et al. 2002). The correlation between latitude and outbreeding reproductive systems lends further support to the idea that negative feedback is greater at lower latitudes. Sexual reproduction is a mechanism by which host populations can generate and regenerate rare host genotypes favored in the face of pathogen attack (Clay and Kover 1996). Plants in tropical regions are more frequently characterized by outbreeding mechanisms than plants in temperate areas, and increased pest pressure in tropical systems may be responsible for this pattern (Levin 1975). Results from animal systems also support this conclusion (Ghiselin 1974, Glesener and Tilman 1978).

Special Feature

2001). In temperate forests, host-specic soilborne pathogens cause seedling mortality of black cherry (Prunus serotina) in a distance and density-dependent manner (Packer and Clay 2000). Soilborne pathogens may affect vegetation composition and dynamics in temperate grasslands as well (Mills and Bever 1998, Holah and Alexander 1999, Matthews and Clay 2000, Olff et al. 2000). However, particular studies of positive or negative feedback in any one habitat may not reect large-scale geographical (or temporal) patterns of the relative strength of feedback processes. Appropriate empirical tests of the hypothesis illustrated in Fig. 2 could manipulate microbial communities (i.e., the home vs. away culture approach of Bever [1994, 1997]) or their presence or absence (i.e., the soil sterilization approach of Van der Putten et al. 1993) across latitudinal and altitudinal (or successional) gradients. Plantsoil feedback, and plantmicrobe interactions in general, are likely to interact with other mechanisms, including other biotic interactions that are themselves correlated with spatial or temporal gradients. Nonetheless, the concept of plantsoil microbe interactions as drivers of large-scale vegetation gradients deserves more attention, and has the potential to enrich our understanding of vegetation diversity and succession.

Microbially mediated resource partitioning

It is difcult to imagine that the diversity of N, P, or other soil nutrient pools changes much over largescale spatial gradients. However, we expect microbially mediated partitioning of soil nutrients to occur most strongly in low productivity habitats, where soil resources are relatively more limiting than light (Tilman 1988) and the benets of partitioning correspondingly higher. Unproductive habitats can occur anywhere, but the general trend of increasing productivity from poles to equator or high to low altitude leads to the expectation that microbially-mediated resource partitioning will be relatively less important in the tropics or at low altitudes. Empirical tests of this hypothesis are needed. CONCLUSIONS Plant interactions with soil microorganisms have become more central to a variety of ecological research efforts, in spite of the fact that most ecologists lack training in microbiology, the techniques of microbial ecology are challenging to apply in the eld, and the diversity of microbes and their interactions with plants are to a large extent still unknown. In this paper, we have focused on microbial mediation of niche differentiation in resource use and feedback dynamics between the plant and soil community, but soil microbes may inuence many other important ecological processes. We have posed hypotheses that are consistent with large bodies of existing data, but few studies have been conducted that explicitly test these ideas. With respect

The case for feedback

Positive feedback is certainly not limited to stressful habitats, nor is negative feedback limited to the tropics. For example, ectomycorrhizal seedlings strongly benet from proximity to parent trees in certain temperate oak forests (Dickie et al. 2002). Positive feedback with ectomycorrhizae may also be responsible for anomalous low-diversity tropical forests (Connell and Lowman 1989, Torti and Coley 1999). Likewise, soilborne plant pathogens create openings in high latitude coniferous forests, affecting forest composition and successional patterns (Holah et al. 1993, McLaughlin

September 2003

UNDERGROUND PROCESSES

2289

to microbial mediation of resource differentiation, there is rst a need to test the basic premise. Useful approaches range from observational studies of plant nutrient use and microbial associates in greenhouse or eld to highly controlled greenhouse and eld experiments, where the identity of nutrient resources and the composition of the microbial community are manipulated, and host plant responses are monitored. Advances in molecular and stable isotope techniques make these approaches feasible. Subsequent research might then focus on comparison of the phenomenon in earlyvs. late-successional communities or in communities along latitudinal or altitudinal gradients. Such comparisons, using techniques of measuring plant growth in sterilized and cultured soil, are also needed to evaluate the relative balance of positive and negative feedback over temporal and spatial gradients. These relatively simple techniques have much to offer, although a disadvantage is that the microbial community is treated as a black box and only their net effect on plant growth is revealed. The growing array of molecular techniques will enhance our perspective on plant-soil feedback by allowing the manipulation of individual microbial groups, as well as quantication of microbial populations and their individual effects on plants in diverse natural communities.
ACKNOWLEDGMENTS Thanks to the Plant Interactions Group (PIG) at Indiana Universitys Department of Biology and two anonymous reviewers for helpful feedback. Partial nancial support was provided by NSF DEB-0090056 to K. Clay and A. Packer, NSF DEB-0049080 to J. D. Bever, and an Andrew W. Mellon Foundation grant to H. L. Reynolds, J. D. Bever, and P. A. Schultz. LITERATURE CITED Aarssen, L. W. 1984. On the distinction between niche and competitive ability. Acta Biotheoretica 33:6783. Aarssen, L. W. 2001. On correlations and causations between productivity and species richness in vegetation: predictions from habitat attributes. Basic and Applied Ecology 2:105 114. Agrios, G. N. 1997. Plant pathology. Academic Press, San Diego, California, USA. Ahmad, I., and J. A. Hellebust. 1991. Enzymology of nitrogen assimilation in mycorrhiza. Methods in Microbiology 23:181202. Allen, E. B., and M. F. Allen. 1990. The mediation of competition by mycorrhizae in successional and patchy environments. Pages 367389 in J. B. Grace and G. D. Tilman, editors. Perspectives on plant competition. Academic Press, New York, New York, USA. Augspurger, C. K. 1988. Impact of plant pathogens on natural plant populations. Pages 413433 in A. J. Davy, M. J. Hutchings, and A. R. Watkinson, editors. Plant population ecology. Blackwell Scientic, Oxford, UK. Augspurger, C. K. 1992. Experimental studies of seedling recruitment from contrasting seed distributions. Ecology 73:12701284. Augspurger, C. K., and C. K. Kelly. 1984. Pathogen mortality of tropical tree seedlings: experimental studies of the effects of dispersal distance, seedling density, and light conditions. Oecologia 61:211217.

Bethlenfalvay, G. J., M. S. Brown, and R. S. Pacovsky. 1982. Relationships between host and endophyte development in mycorrhizal soybean. New Phytologist 90:537543. Bever, J. D. 1994. Feedback between plants and their soil communities in an old eld community. Ecology 75:1965 1977. Bever, J. D. 1999. Dynamics within mutualism and the maintenance of diversity: inference from a model of interguild frequency dependence. Ecology Letters 2:5261. Bever, J. D. 2002. Negative feedback within a mutualism: Host-specic growth of mycorrhizal fungi reduces plant benet. Proceedings of the Royal Society of London 269: 25952601. Bever, J. D., J. B. Morton, J. Antonovics, and P. A. Schultz. 1996. Host-dependent sporulation and species diversity of arbuscular mycorrhizal fungi in a mown grassland. Journal of Ecology 84:7182. Bever, J. D., P. Schultz, A. Pringle, and J. Morton. 2001. Arbuscular mycorrhizal fungi: more diverse than meets the eye and the ecological tale of why. BioScience 51:923 931. Bever, J. D., K. M. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: the utility of the feedback approach. Journal of Ecology 85:561573. Boerner, R. E. J., B. G. DeMars, and P. N. Leicht. 1996. Spatial patterns of mycorrhizal infectiveness of soils long a successional chronosequence. Mycorrhiza 6:7990. Bruehl, G. W. 1987. Soilborne plant pathogens. Cambridge University Press, Cambridge, UK. Chapin, F. S., III, P. M. Vitousek, and K. van Cleve. 1986. The nature of nutrient limitation in plant communities. American Naturalist 127:4858. Chapin, F. S., III, L. R. Walker, C. L. Fastie, and L. C. Sherman. 1994. Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska. Ecological Monographs 64:149175. Choler, P., R. Michalet, and R. M. Callaway. 2001. Facilitation and competition on gradients in alpine plant communities. Ecology 82:32953308. Clay, K., and P. Kover. 1996. The Red Queen hypothesis and plant/pathogen interactions. Annual Review of Phytopathology 34:2950. Clay, K., and W. H. van der Putten. 1999. Pathogens and plant life histories. Pages 275302 in T. O. Vuorisalo and P. K. Mutikainen, editors. Life history evolution in plants. Kluwer Academic Publishers, Dordrecht, The Netherlands. Coley, P. D., J. P. Bryant, and F. S. Chapin III. 1985. Resource availability and plant antiherbivore defense. Science 230: 895899. Condit, R., S. P. Hubbell, and R. B. Foster. 1994. Density dependence in two understory tree species in neotropical forest. Ecology 75:671680. Connell, J. H. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rainforest trees. Pages 294310 in P. J. den Boer and G. R. Gradwell, editors. Dynamics of populations. Center for Agricultural Publishing and Documentation, Wageningen, The Netherlands. Connell, J. H., and M. D. Lowman. 1989. Low-diversity tropical rainforests: some possible mechanisms for their existence. American Naturalist 134:88119. Connell, J. H., and R. O. Slatyer. 1977. Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111:1119 1144. Crocker, R. L., and J. Major. 1955. Soil development in relation to vegetation and surface age at Glacier Bay, Alaska. Journal of Ecology 43:427448.

Special Feature

2290

HEATHER L. REYNOLDS ET AL.

Ecology, Vol. 84, No. 9

Dickie, I. A., R. T. Koide, and K. C. Steiner. 2002. Inuences of established trees on mycorrhizas, nutrition, and growth of Quercus rubra seedlings. Ecological Monographs 72: 505521. Elton, C. S. 1958. The ecology of invasions by plants and animals. Methuen, London, UK. Eom, A. H., D. C. Hartnett, and G. W. T. Wilson. 2000. Host plant species effects on arbuscular mycorrhizal fungal communities in tallgrass prairie. Oecologia 122:435444. Fox, J. F. 1977. Alternation and coexistence of tree species. American Naturalist 111:6989. Ghiselin, M. T. 1974. The economy of nature and the evolution of sex. University of California Press, Berkeley, California, USA. Givnish, T. J. 1999. On the causes of gradients in tropical tree diversity. Journal of Ecology 87:193210. Glesener, R. R., and D. Tilman. 1978. Sexuality and the components of environmental uncertainty: clues from geographical parthenogenesis in terrestrial animals. American Naturalist 112:659673. Gorham, E., P. M. Vitousek, and W. A. Reiners. 1979. The regulation of chemical budgets over the course of terrestrial ecosystem succession. Annual Review of Ecological Systems 10:5384. Grayston, S. J., S. Wang, C. D. Campbell, and A. C. Edwards. 1998. Selective inuence of plant species on microbial diversity in the rhizosphere. Soil Biology and Biochemistry 30:369378. Greenlee, J. T., and R. M. Callaway. 1996. Abiotic stress and the relative importance of interference and facilitation in montane bunchgrass communities in western Montana. American Naturalist 148:386396. Gutschick, V. P. 1981. Evolved strategies in nitrogen acquisition by plants. American Naturalist 118:607637. Hamilton, E. W., III, and D. A. Frank. 2001. Can plants stimulate soil microbes and their own nutrient supply? Evidence from a grazing tolerant grass. Ecology 82:2397 2402. Harms, K. E., S. J. Wright, O. Calderon, A. Hernandez, and E. A. Herre. 2000. Pervasive density-dependent recruitment enhances seedling diversity in a tropical forest. Nature 404:493495. Harte, J., and A. P. Kinzig. 1993. Mutualism and competition between plants and decomposers: implications for nutrient allocation in ecosystems. American Naturalist 141:829 846. Hawkins, H. J., A. Johansen, and E. George. 2000. Uptake and transport of organic and inorganic nitrogen by arbuscular mycorrhizal fungi. Plant and Soil 226:275285. HilleRisLambers, J., J. S. Clark, and B. Beckage. 2002. Density-dependent mortality and the latitudinal gradient in species diversity. Nature 417:732735. Hodge, A., C. D. Campbell, and A. H. Fitter. 2001. An arbuscular mycorrhizal fungus accelerates decomposition and acquires nitrogen directly from organic material. Nature 413:297299. Hodge, A., D. Robinson, and A. Fitter. 2000a. Are microorganisms more effective than plants at competing for nitrogen? Trends in Plant Science 5:304308. Hodge, A., D. Robinson, and A. H. Fitter. 2000b. An arbuscular mycorrhizal inoculum enhances root proliferation in, but not nitrogen capture from, nutrient-rich patches in soil. New Phytologist 145:575584. Holah, J. C., and H. M. Alexander. 1999. Soil pathogenic fungi have the potential to affect co-existence of two tallgrass prairie species. Journal of Ecology 84:598608. Holah, J. C., M. V. Wilson, and E. M. Hansen. 1993. Effects of a native forest pathogen, Phellinus weirii, on Douglasr forest composition in western Oregon. Canadian Journal of Forestry Research 23:24732480.

Janos, D. P. 1980. Mycorrhizae inuence tropical succession. Mycorrhizae 12:5664. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. American Naturalist 102:592595. Jayachandran, K., A. P. Schwab, and B. A. D. Hetrick. 1992. Mineralization of organic phosphorus by vesicular-arbuscular mycorrhizal fungi. Soil Biology and Biochemistry 24: 897903. Katan, J. 1991. Interactions of roots with soil-borne pathogens. Pages 823836 in Y. Waisel, A. Eshel, and U. Kafka, editors. Plant roots: the hidden half. Marcel Dekker, New York, New York, USA. Knops, J. M. H., D. Tilman, N. M. Haddad, S. Naeem, C. E. Mitchell, J. Haarstad, M. E. Ritchie, K. M. Howe, P. B. Reich, E. Siemann, and J. Groth. 1999. Effects of plant species richness on invasion dynamics, disease outbreaks, insect abundances and diversity. Ecology Letters 2:286 293. Larson, J. L., and E. Siemann. 1998. Legumes may be symbiont-limited during old-eld succession. American Midland Naturalist 140:9095. Lavelle, P., and A. V. Spain. 2001. Soil ecology. Kluwer Academic, Boston, Massachusetts, USA. Leath, S., A. L. Scharen, and M. E. Dietz-Holmes. 1993. Factors associated with global occurrences of Septoria nodorum blotch and Septoria tritici blotch of wheat. Plant Disease 77:12661270. Levin, D. A. 1975. Pest pressure and recombination systems in plants. American Naturalist 109:437451. Marschner, H. 1995. Mineral nutrition of higher plants. Academic Press, San Diego, California, USA. Matthews, J., and K. Clay. 2000. Inuence of fungal endophyte infection on plantsoil feedback and community interactions. Ecology 82:500509. McKane, R. B., L. C. Johnson, G. R. Shaver, K. J. Nadelhoffer, E. B. Rastetter, B. Fry, A. E. Giblin, K. Kielland, B. L. Kwiatkowski, J. A. Laundre, and G. Murray. 2002. Resource-based niches provide a basis for plant species diversity and dominance in arctic tundra. Nature 415:68 71. McLaughlin, J. A. 2001. Impact of Armillaria root disease on succession in red pine plantations in southern Ontario. The Forestry Chronicle 77:519524. Medve, R. J. 1984. The mycorrhizae of pioneer species in disturbed ecosystems in western Pennsylvania. American Journal of Botany 71:787794. Michelsen, A., I. K. Schmidt, S. Jonasson, C. Quarmby, and D. Sleep. 1996. Leaf 15N abundance of subarctic plants provides eld evidence that ericoid, ectomycorrhizal and non- and arbuscular mycorrhizal species access different sources of soil nitrogen. Oecologia 105:5363. Miller, A. E., and W. D. Bowman. 2002. Variation in nitrogen-15 natural abundance and nitrogen uptake traits among co-occurring alpine species: do species partition nitrogen by form? Oecologia 130:609616. Mills, K. E., and J. D. Bever. 1998. Maintenance of diversity within plant communities: soil pathogens as agents of negative feedback. Ecology 79:15951601. Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross, H. L. Reynolds, R. B. Waide, M. R. Willig, S. I. Dodson, and L. Gough. 2001. What is the observed relationship between species richness and productivity. Ecology 82: 23812396. Molofsky, J., J. D. Bever, and J. Antonovics. 2001. Coexistence under positive frequency dependence. Proceedings of the Royal Society of London 268:273277. Molofsky, J., J. D. Bever, J. Antonovics, and T. J. Newman. 2002. Negative frequency-dependence and the importance of spatial scale. Ecology 83:2127.

Special Feature

September 2003

UNDERGROUND PROCESSES

2291

Olff, H., B. Hoorens, R. G. M. de Goede, W. H. van der Putten, and J. M. Gleichman. 2000. Small-scale shifting mosaics of two dominant grassland species: the possible role of soil-borne pathogens. Oecologia 125:4554. Packer, A., and K. Clay. 2000. Soil pathogens and spatial patterns of seedling mortality in a temperate tree. Nature 404:278281. Paul, E. A., and F. E. Clark. 1996. Soil microbiology and biochemistry. Academic Press, San Diego, California, USA. Poorter, H. 1990. Interspecic variation in relative growth rate: on ecological causes and physiological consequences. Pages 4568 in H. Lambers, M. L. Cambridge, H. Konings, and T. L. Pons, editors. Causes and consequences of variation in growth rate and productivity of higher plants. SPB Academic Publishing, The Hague, The Netherlands. Read, D. J. 1991. Mycorrhizas in ecosystems. Experientia 47:376391. Read, D. J. 1993. Mycorrhiza in plant communities. Advances in Plant Pathology 9:131. Rejmanek, M., and D. M. Richardson. 1996. What attributes make some plant species more invasive? Ecology 77:1655 1661. Reynolds, H. L. 1999. Plant interactions: competition. Pages 649675 in F. I. Pugnaire and F. Valladares, editors. Handbook of functional plant ecology. Marcel Dekker, New York, New York, USA. Richardson, D. M., N. Allsopp, C. M. DAntonio, S. J. Milton, and M. Rejmanek. 2000. Plant invasions- the role of mutualisms. Biological Reviews of the Cambridge Philosophical Society 75:6593. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge University Press, New York, New York, USA. Rovira, A. D. 1969. Plant root exudates. Botanical Review 35:3556. Sanders, H. L. 1968. Marine benthic diversity: a comparative study. American Naturalist 102:243282. Schlesinger, W. H. 1991. Biogeochemistry: an analysis of global change. Academic Press, San Diego, California, USA. Smith, S. E., and D. J. Read. 1997. Mycorrhizal symbiosis. Academic Press, London, UK. Sprent, J. I. 1993. The role of nitrogen xation in primary succession. Pages 209220 in J. Miles and D. W. H. Walton, editors. Primary succession on land. Blackwell Scientic Publications, Boston, Massachusetts, USA. Tabatabai, T., and W. A. Dick. 2002. Enzymes in soil. Pages 567596 in R. G. Burns and R. P. Dick, editors. Enzymes in the environment. Marcel Dekker, New York, New York, USA. Tarafdar, J. C., and H. Marschner. 1994. Phosphatase activity in the rhizosphere and hyphosphere of VA mycorrhizal wheat supplied with inorganic and organic phosphorus. Soil Biology and Biochemistry 26:387395. Tate, R. L., III. 1997. Soil microbial diversity: whither to now? Soil Science 162:605606. Thompson, J. N., et al. 2001. Frontiers of ecology. BioScience 51:1524. Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton, New Jersey, USA. Tilman, D., and S. Pacala. 1993. The maintenance of species richness in plant communities. Pages 1325 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological

communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA. Torsvik, V., J. Goksyr, and F. L. Daae. 1990. High diversity in DNA of soil bacteria. Applied Environmental Microbiology 56:782787. Torsvik, V., J. Goksyr, F. L. Daae, R. Srheim, J. Michelsen, and K. Salte. 1994. Use of DNA analysis to determine the diversity of microbial communities. Pages 3948 in K. Ritz, J. Dighton, and K. E. Giller, editors. Beyond the biomass: compositional and functional analysis of soil microbial communities. Wiley, Chichester, UK. Torti, S. D., and P. D. Coley. 1999. Tropical monodominance: a preliminary test of the ectomycorrhizal hypothesis. Biotropica 31:220228. Tresner, H. D., M. P. Backus, and J. T. Curtis. 1954. Soil microfungi in relation to the hardwood forest continuum in Southern Wisconsin. Mycologia 46:313333. Turnbull, M. H., R. Goodall, and G. R. Stewart. 1995. The impact of mycorrhizal colonization upon nitrogen source utilization and metabolism in seedlings of Eucalyptus grandis Hill ex Maiden and Eucalyptus maculata Hook. Plant, Cell and Environment 18:13861394. Turnbull, M. H., S. Schmidt, P. D. Erskine, S. Richards, and G. R. Stewart. 1996. Root adaptation and nitrogen source acquisition in natural ecosystems. Tree Physiology 16:941 948. Van der Heijden, M. G. A., J. N. Klironomos, M. Ursic, P. Moutoglis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and I. R. Sanders. 1998. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396:6972. Van der Putten, W. H., P. H. T. Maas, W. J. M. Van Gulik, and H. Brinkman. 1990. Characterization of soil organisms involved in the degeneration of Ammophila arenaria. Soil Biology and Biochemistry 22:845852. Van der Putten, W. H., and S. R. Troelstra. 1990. Harmful soil organisms in coastal foredunes involved in degeneration of Ammophila arenaria and Calammophila baltica. Canadian Journal of Botany 68:15601568. Van der Putten, W. H., C. Van Dijk, and B. A. M. Peters. 1993. Plant-specic soil-borne diseases contribute to succession in foredune vegetation. Nature 362:5356. Waide, R. B., M. R. Willig, C. F. Steiner, G. Mittelbach, L. Gough, S. I. Dodson, G. P. Juday, and R. Parmenter. 1999. The relationship between productivity and species richness. Annual Review of Ecological Systematics 30:257300. Walker, T. W., and J. K. Syers. 1976. The fate of P during pedogenesis. Geoderma 15:119. Weltzien, H. C. 1972. Geophytopathology. Annual Review of Phytopathology 10:277298. Westover, K. M., and J. D. Bever. 2001. Mechanisms of plant species coexistence: complementary roles of rhizosphere bacteria and root fungal pathogens. Ecology 82:32853294. Wills, C., R. Condit, R. B. Foster, and S. P. Hubbell. 1997. Strong density- and diversity- related effects help to maintain tree species diversity. Proceedings of the National Academy of Sciences (USA) 94:12521257. Wohlrab, G., R. W. Tuveson, and C. E. Olmsted. 1963. Fungal populations from early stages of succession in Indiana dune sand. Ecology 44:734740. Woods, K. D. 1979. Reciprocal replacement and the maintenance of codominance in a beech maple forest. Oikos 33: 3139. Yang, X. B., and F. Feng. 2001. Ranges and diversity of soybean fungal diseases in North America. Phytopathology 91:769775.

Special Feature

Ecology, 84(9), 2003, pp. 22922301 2003 by the Ecological Society of America

VARIATION IN PLANT RESPONSE TO NATIVE AND EXOTIC ARBUSCULAR MYCORRHIZAL FUNGI

JOHN N. KLIRONOMOS1
Department of Botany, University of Guelph, Guelph, Ontario, Canada N1G 2W1

Special Feature

Abstract. High variability in plant-growth response to the presence of different mycorrhizal fungi can be a major determinant of local plant species diversity. Multiple species of arbuscular mycorrhizal fungi can coexist in terrestrial ecosystems, and co-occurring plants can differ in their response to colonization by these different fungi. However, the range of mycorrhizal plant-growth responses that can occur within communities has not been determined. In the present study, I crossed a large number of plant and fungal species that co-occur to determine the range of responses that can exist within an ecosystem. I also crossed exotic fungal isolates vs. local plant isolates and local fungal isolates vs. exotic plant isolates to determine whether the range of plant growth responses differs when using foreign genotypes. The data indicate that plant growth responses to mycorrhizal inoculation within an ecosystem can range from highly parasitic to highly mutualistic. In this study, the direction and magnitude of the response depended on the combination of plant and fungal species. No plant did best with the same fungal isolate. The range of responses was greatest when using local plants and fungi. Whereas parasitic and mutualistic responses were also detected when using foreign plant or fungal genotypes, the range of responses was signicantly reduced, as was the relative frequency of positive responses. Overall, this study suggests that, within ecosystems, arbuscular mycorrhizal fungi can function along a continuum from parasitism to mutualism, and that extreme responses are more common when using locally adapted plants and fungi. This high variation in plant growth response may be a large contributor to plant species coexistence and the structure of plant communities.
Key words: exotic; mycorrhiza; native; plantmicrobe interactions; soil ecology.

INTRODUCTION Arbuscular mycorrhizal fungi (AMF) (Phylum Glomeromycota) are symbionts that infect the roots of most terrestrial plants, and provide their hosts with a wide variety of soil resources in exchange for photosynthate (Smith and Read 1997). The interaction between plants and AMF has traditionally been regarded as a mutualism, in that both partners benet from the association. AMF are obligate biotrophs, so they typically gain by interacting with plants (Smith and Read 1997). However, recent evidence indicates that the costs and benets of maintaining a symbiosis with AMF can differ signicantly for plants, and the resulting plant responses can vary widely (Powell et al. 1982, Jensen 1984, Haas et al. 1987, Raju et al. 1990, Streitwolf-Engel et al. 1997, van der Heijden et al. 1998a, Klironomos 2000). Since it is possible for AMF to confer high costs on plants, it is not appropriate to assume that AMF always stimulate the growth of plants in natural ecosystems. Plant growth responses may range from positive (mutualism) to neutral (commensalism) and even to negative (parasitism), so the symbiosis should more accurately be dened as a continManuscript received 11 July 2002; revised 14 October 2002; accepted 16 October 2002. Corresponding Editor: M. Parker. For reprints of this Special Feature, see footnote 1, p. 2256. 1 E-mail: jklirono@uoguelph.ca

uum from parasitism to mutualism (Johnson et al. 1997). A better understanding of the factors that inuence the positioning of AMF along the parasitismmutualism continuum would lead to a better appreciation of the roles of AMF in natural ecosystems. Whether plants benet from the association depends on a number of factors, including the genotypes of the organisms involved, and the environmental conditions under which they interact. A signicant amount of research has focused on the inuence of environmental conditions, mainly the availability of nutrients (Habte and Manjunath 1991, Koide 1991, Hetrick et al. 1992). Also, we have made signicant progress in our understanding of what plant taxa are likely to form the symbiosis (Harley and Harley 1987, Newman and Reddell 1987), and to some degree, which will most likely require AMF to successfully establish and compete for nutrient resources (Janos 1980, Allen and Allen 1984, 1990, Miller 1987). However, very little is known about the effects of AMF genotype on plant growth, even though it has been recognized that, everything else being equal, AMF taxa can differ signicantly in their growth strategies (Hart and Reader 2002) and in their inuence on plant growth and development (Sanders and Fitter 1992, Streitwolf-Engel et al. 1997, van der Heijden et al. 1998a).

2292

September 2003

UNDERGROUND PROCESSES

2293

Although AMF differ in their abilities to inuence plant growth, it is currently not possible to categorize them along the parasitemutualist continuum because not all infected plants seem to do best with the same AMF (Sanders and Fitter 1992, van der Heijden et al. 1998a, Kiers et al. 2000, Klironomos 2000). Even when a few co-occurring plant and AMF taxa are studied, there is a strong interaction between plant and AMF taxa in plant growth response (Sanders and Fitter 1992, van der Heijden et al. 1998b). Terrestrial ecosystems contain many AMF and plant species that coexist in communities (Johnson et al. 1991, Allen et al. 1995, Sanders et al. 1996, Helgasson et al. 1998, Picone 2000, Ergeton-Warburton and Allen 2000), so the inuence of AMF on growth of plants may be very complex. To date, there is little appreciation of the range of plant-growth responses that can occur within any AMF community. Yet, such information would be valuable for the development of plant community models because the variation in AMFhost response within a community is believed to be a driving force behind the structuring of AMF and plant communities (Grime et al. 1987, Bever et al. 1996, van der Heijden et al. 1998b, Hartnett and Wilson 1999, Marler et al. 1999, Smith et al. 1999). Ecological studies on plantAMF interactions have typically taken one of three approaches when choosing AMF inoculum for experimentation. The rst approach has been to use AMF inoculum from foreign soils, usually isolates that have been stored in a culture collection. The benet of this approach is that AMF can be accessed very quickly, but unfortunately interactions between plants and foreign AMF may have little relevance to the interactions that occur locally, and observed responses may not represent the suite of responses that plants would normally encounter at home. A second approach has been to use local, but whole-soil inoculum. Whereas this is more appropriate than the rst option, it does not provide information on the effects and responses of individual fungal taxa. The third approach has been to use multiple local AMF, but this rarely occurs because of the time and resources required to successfully isolate and culture individual AMF from the eld. Even when this third approach is taken, a small number of isolates are typically compared. The present study was designed to answer a number of questions related to variation in plant-growth response to AMF. The research was conducted with plants and AMF from an old-eld plant community within the Long-Term Mycorrhiza Research Site (LTMRS), which is located at the University of Guelph, Ontario, Canada. This site was chosen because a large number of AMF have been isolated and maintained in culture (van der Heijden et al. 1998b, Klironomos 2000). First, I wanted to determine whether mutualistic interactions are more common than parasitic interaction between plants and AMF that coexist in communities. This was done by comparing the growth of mycorrhizal plants with that

of nonmycorrhizal controls. I selected 10 plant species that coexist within the old-eld community and crossed each one with one of 10 AMF species that also coexist at the same eld site. To determine the magnitude of variation in growth response that is possible with a single AMF species, I also inoculated one AMF isolate onto each of 64 plant species. Finally, my nal objective was to assess whether the source of AMF is an important determinant of host response, because it is currently not known if exotic AMF are as effective as locally adapted isolates. To determine this I grew (a) home plants with home AMF, (b) home plants with foreign AMF, and (c) foreign plants with home AMF. MATERIALS
AND

METHODS

Plant material
The 64 plant species that were used in this study are listed in Table 1. These plants coexist in an old-eld community at the LTMRS, University of Guelph (43 32 30 N, 80 13 00 W). Seeds were collected during the growing seasons of 19971999 and they were then stored at 4 C until ready to be used in experiments. Only 10 of these species were used in Experiment 1 (see list in Table 1). For these 10 species, additional seeds were also collected from various other locations throughout southern Ontario and Quebec to test for differences among home and foreign plants. Home plants are those that originated from the LTMRS. Foreign plants are those that originated from other locations. The remaining plant species listed on Table 1 were used only in Experiment 2, which required only home material.

Special Feature

Fungal material
AMF species that were used in this study are listed in Table 2. The 10 AMF species were isolated from the LTMRS (referred to as home AMF), and also from various other locations (referred to as foreign AMF). Home AMF were isolated from eld soil by rst allowing the AMF to sporulate in trap cultures containing Allium porrum L. cv. Giant Musselburgh, and then using single AMF spores to start single species cultures (Brundrett 1996). Some of the foreign AMF were also isolated in the same way from foreign meadow or grassland ecosystems, whereas others purchased from the INVAM culture collection. For a period of two years prior to setting up the experiments described below, all AMF were grown in dual pot culture with the host Allium porrum under similar greenhouse conditions. During that time, AMF were subcultured at three month intervals to keep the cultures clean and viable. All AMF are stored in the University of Guelph (LTMRS) culture collection.

Experimental set-up and design Experiment 1.This experiment was set up using a 10 plants 3 sources factorial design of 11 AMF

2294

JOHN N. KLIRONOMOS
TABLE 1.

Ecology, Vol. 84, No. 9

of origin. The AMF factor consisted of one of 10 AMF species or the non-AMF control (Table 2). The plant factor consisted of one of 10 plant species (Table 1). The source of origin factor consisted of one of either (a) home AMF and home plants, (b) foreign AMF and home plants, or (c) home AMF and foreign plants. Each treatment combination consisted of 10 replicated units, for a grand total of 3300 experimental units. Each unit was positioned in a completely randomized design on benches in a greenhouse. The experiment ran from May through August 1999 under ambient light conditions, 23.7:18.4 C mean day:night temperatures, and 50.2: 71.4% mean day : night relative humidity. Each experimental unit consisted of a pot (15 cm diameter 60 cm length) containing a sterile soil/silica sand mix, AMF inoculum, and an individual plant. The sandy loam soil was collected from the LTMRS (total N 83.1 mmol/kg; total P 6.2 mmol/kg; percentage 6.1). It was mixed together with of organic matter silica sand at a 1:1 ratio, autoclaved, and then added to individual pots. At a depth of 2 cm below the surface of the soil, we added a band of AMF inoculum with a mass of 1 g. This inoculum was composed of sheared Allium porrum roots (precolonized by one of the AMF isolates) and 100 spores. The nonmycorrhizal experimental units received an equal total amount of uncolonized Allium porrum roots. To correct for possible differences in microbial communities, each experimental unit received a 50-mL ltered washing comprised of microbial extract from every AMF isolate used (Koide and Li 1989). Plant seeds were germinated in a growth chamber at 20 C on moist lter paper. Individual seedlings were then transferred to the pots. We initially added two seedlings, but after one week we removed one plant. The remaining plant in each pot was left to grow for a period of 16 weeks. All plants were watered every two days or as needed with deionized water. They were also fertilized once per week with a modied Long-Ashton Nutrient solution (half strength P; Hewitt 1966). Experiment 2.This experiment was set up using a factorial design of 2 AMF 64 plant species. The AMF factor consisted of either the addition of Glomus etunicatum from the LTMRS or a nonmycorrhizal control. The 64 plant species are listed in Table 1, and they all originated from the LTMRS. Each treatment combination consisted of 10 replicated experimental units. There was a total of 1280 experimental units, each randomly positioned in a greenhouse for the duration of the experiment. The experiment ran from May through August 2000 under ambient light conditions, 24.3:19.5 C mean day : night temperatures, and 53.1: 74.5% mean day : night relative humidity. The construction of experimental units was similar to those described for Experiment 1. Plants were allowed to grow for 16 weeks. They were watered every two days or as needed with deionized water, and fertilized once

Plant species used in Experiments 1 and 2. Used in experiment no. 2 1, 2 2 2 2 1, 2 2 1, 2 2 2 2 2 2 2 1, 2 2 2 2 2 2 1, 2 2 2 1, 2 2 2 2 2 2 2 2 2 2 1, 2 2 2 1, 2 2 2 2 2 1, 2 2 2 1, 2 2 2 2 2 2 2 2 2 2 2

Plant species 1. Achillea millefolium L. 2. Agrostis gigantea Roth. 3. Agrostis scabra L. 4. Apocynum cannabinum L. 5. Asclepias syriaca L. 6. Asparagus ofcinalis L. 7. Aster novae-angliae L. 8. Aster lanceolatus Willd. 9. Aster lateriorus (L.) Britton. 10. Bromus inermis Leysser 11. Carex aurea Nutt. 12. Carex ava L. 13. Carex acca Schreber. 14. Carex granularis Muhl. 15. Centaurea jacea L. 16. Cerastium vulgatum L. 17. Chenopodium ambroisioides L. 18. Chrysanthemum leucanthemum L. 19. Cichorium intybus L. 20. Circium arvense (L.) Scop. 21. Circium vulgare (Savi) Tenore. 22. Convolvulus arvensis L. 23. Coronilla varia L. 24. Dactylis glomerata L. 25. Daucus carota L. 26. Echium vulgare L. 27. Erigeron philadelphicus L. 28. Erigeron strigosus Muhl. 29. Fragaria virginiana Duchesne 30. Galium mollugo L. 31. Galium palustre L. 32. Geum aleppicum Jacq. 33. Hieracium auranticum L. 34. Hieracium pilosella L. 35. Hieracium lachenalii C. Gmelin. 36. Hypericum perforatum L. 37. Juncus dudleyi (Weieg.) F. J. Herm. 38. Linaria vulgaris Miller. 39. Medicago lupulina L. 40. Oenothera biennis L. 41. Oenothera perennis L. 42. Panicum lanuginosum Elliott. 43. Phleum pratense L. 44. Plantago lanceolata L. 45. Poa compressa L. 46. Poa pratensis L. 47. Potentilla recta L. 48. Prunella vulgaris L. 49. Ranunculus acris L. 50. Rudbeckia hirta L. 51. Rudbeckia serotina Farw. 52. Satureja vulgaris (L.) Fritsch. 53. Silene vulgaris (Moench) Garcke. 54. Solidago canadensis L. 55. Solidago graminifolia L. 56. Solidago nemoralis Aiton. 57. Solidago rugosa Miller. 58. Taraxacum ofcinale Weber ex Wiggers. 59. Tragopogon pratensis L. 60. Trifolium pratense L. 61. Trifolium repens L. 62. Veronica ofcinalis L. 63. Vicia cracca L. 64. Viccia sativa L.

2 2

Special Feature

September 2003 TABLE 2.

UNDERGROUND PROCESSES

2295

Fungal species used in Experiments 1 and 2. Experiment no. 1 1 1 1 1 1, 2 1 1 1 1

Fungal species

Home isolate Guelph011 Guelph020 Guelph035 Guelph064 Guelph004 Guelph009 Guelph017 Guelph041 Guelph022 Guelph051

Foreign isolate INVAM: CL 139-3 INVAM: NC 148 INVAM: NC 175 INVAM: KS 885 Guelph001 Guelph005 INVAM: PA 126 Guelph004 INVAM: NC 153 INVAM: WV848B-1

Acaulospora denticulata Sieverding and Toro Acaulospora morrowiae Spain & Schenck Gigaspora margarita Becker & Hall Gigaspora rosea Nicolson & Schenck Glomus intraradices Schenck & Smith Glomus etunicatum Becker & Gerdemann Glomus geosporum (Nicolson & Gerdemann) Walker Glomus mosseae (Nicolson & Gerdemann) Gerdemann & Trappe Scutellospora calospora (Nicol. & Gerd.) Walker & Sanders emend. Koske & Walker Scutellospora pellucida (Nicolson & Schenck) Walker & Sanders emend. Koske & Walker

per week with a modied Long-Ashton Nutrient solution (half strength P; Hewitt 1966). Response variables measured and statistical analysis.For both experiments, plant shoots and roots were harvested at the end of the 16-wk period. Plant material was then dried at 60 C for 36 h and then weighed to determine biomass. Prior to drying, a subsample of roots was taken from each pot and stored in 50% ethanol. This subsample of roots was then cleared in 10% potassium hydroxide, and stained with Chlorazol Black E (Brundrett et al. 1984) to conrm the presence of AMF structures. The abundance of AMF structures in plant roots was not assessed in this study because in an earlier study (Klironomos 2000) I found no relationship between percentage root colonization and plant growth response for a subset of the plant species tested here. The data in this study are presented as the percentage difference in plant growth between mycorrhizal and nonmycorrhizal treatments. To calculate 95% condence intervals of these estimates, I took a bootstrap sample of the plant biomass values for a particular plant/AMF combination, and a second bootstrap sample of the nonmycorrhizal controls, calculated the percentage difference, and repeated 999 times. The data in both experiments were analyzed with factorial ANOVA to determine the signicance of main factor effects and interactions. Chi-square analyses were used to compare the frequencies of positive and negative interactions. Comparisons of response frequency distributions between native and exotic fungi was assessed using Kolmogorov-Smirnov goodness-of-t tests. Mycorrhizal dependency typically refers to the difference in plant growth between mycorrhizal and nonmycorrhizal treatments. Mycorrhizal dependency was calculated for each plant species following the formulas by van der Heijden (2002). When the mean mass of mycorrhizal plants was greater than nonmycorrhizal controls, then mycorrhizal dependency (1 [bn/ a]) 100%, where b is the mean dry mass of the nonmycorrhizal treatment, n is the number of treatments containing AMF, and a is the mean dry mass of a treatment containing AMF. When the mean mass of my-

corrhizal plants was less than nonmycorrhizal controls, ( 1 [ a/bn]) then mycorrhizal dependency 100%. Mycorrhizal species sensitivity refers to the variation in plant growth response when associated with different AMF (van der Heijden 2002). For each plant species it was calculated as the coefcient of variation on the dry mass in response to different AMF treatments (van der Heijden 2002). RESULTS

Special Feature

Experiment 1
At the time of harvest, all plants growing in the AMF treatments were infected by fungal hyphae and the vast majority of plants were infected by one or both of the AMF-specic structures (arbuscules, vesicles). In only the following plant/fungal combinations did we not detect any AMF structures: A. gigantea (home)/G. geosporum (home), A. novae-angliae (home)/A. morrowiae (home), O. biennis (home)/A. morrowiae (home), A. gigantea (home)/A. denticulata (foreign), S. canadensis (home)/ S. pellucida (foreign), B. inermis (foreign)/G. geosporum (home), C. leucanthemum (foreign)/G. mosseae (home), C. leucanthemum (foreign)/ S. pellucida (home), S. canadensis (foreign)/A. morrowiae (home). Furthermore, arbuscules or vesicles were not detected in any of the plants grown in the non-AMF treatments. The presence of AMF often signicantly affected the growth of plants, but the direction and magnitude of this effect was dependent on the specic combination of AMF species and plant species, and also on their source of origin (Figs. 13). AMF species and plant 0.05), species main effects were not signicant (P however the source of origin main effect was signi0.014). Also there were signicant interaccant (P plant species (P tion effects between AMF species 0.0001), AMF species source of origin (P 0.0001), plant species source of origin (P 0.0001), plant species source of origin and AMF species 0.0001). (P No plant or AMF species was consistently associated with either positive or negative plant growth responses.

2296

JOHN N. KLIRONOMOS

Ecology, Vol. 84, No. 9

FIG. 1. The inuence of different home arbuscular mycorrhizal fungi on the growth of different home plants (Experiment 1). Bars represent the percentage change in biomass of mycorrhizal plants compared to nonmycorrhizal controls. The mean 95% condence interval for all the treatments is 18%.

Special Feature

Growth responses to interactions between home plants and home fungi are found in Fig. 1. The number of negative and positive responses was approximately equal (53 and 47, respectively, Fig. 4a), and the magnitude of plant growth response ranged from 49% to 46%. When foreign plants or AMF were used,

negative and positive responses were also detected, but some notable differences occurred. First, the direction and magnitude of each response differed, even though the same species was used. Second, the shape of the frequency distributions (Fig. 4) differed (foreign AMF/ home plants, Kolmogorov-Smirnov Z 1.20, P

FIG. 2. The inuence of different foreign arbuscular mycorrhizal fungi on the growth of different home plants (Experiment 1). Bars represent the percentage change in biomass of mycorrhizal plants compared to nonmycorrhizal controls. The mean 95% condence interval for all the treatments is 12%.

September 2003

UNDERGROUND PROCESSES

2297

FIG. 3. The inuence of different home arbuscular mycorrhizal fungi on the growth of different foreign plants (Experiment 1). Bars representage the percentage change in biomass of mycorrhizal plants compared to nonmycorrhizal controls. The mean 95% condence interval for all the treatments is 14%.

Special Feature

0.111; home AMF/foreign plants, Kolmogorov-Smir1.49, P 0.024). The ratio of negative:posnov Z itive responses was shifted from 53:47 (home AMF/ 0.64, P 0.424) to 55:45 (foreign home plants, 2 1.00, P 0.317) and 61:39 AMF/home plants, 2 (home AMF/foreign plants, 2 7.84, P 0.005). The range of responses was smaller ( 38% to 24% when using foreign AMF/home plants; 30% to 19% when using home AMF/foreign plants). The high variation in plant growth response to different AMF resulted in high mycorrhizal species sensitivity for the plant species tested (Fig. 5). Mycorrhizal species sensitivity was highest when home plants were associated with home fungi. However, regardless of the source of the plant and AMF isolates, overall mycorrhizal dependency was low. No signicant relationship was detected between mycorrhizal dependency and mycorrhizal species sensitivity (Fig. 5).

but the range ( 46 to 48) was similar to that observed in experiment 1, where 10 plants were crossed with 10 fungi. DISCUSSION These data clearly show that plant responses to colonization by AMF can range from highly positive to highly negative. It was proposed by Johnson et al. (1997) that mycorrhizal associations could be considered symbioses that functionally range along a continuum of parasitism to mutualism, and that environmental condition, particularly the abundance of soil nutrients, could determine the position of AMF along that continuum. The present data supports the parasitism mutualism hypothesis, and furthermore, it indicates that the frequency of parasitic vs. mutualistic interactions can be equal even with communities of plants and AMF that grow in a common environment. Clearly, it is inaccurate to refer to arbuscular mycorrhizal symbioses only as mutualisms. This ignores the prevalence of parasitism within this symbiosis. Parasitism may be more frequent under certain environmental conditions, such as in agricultural soils that have been repeatedly fertilized (Johnson 1993). However, it is also frequent in more natural systems. The present site (LTMRS) has not been disturbed for 35 years (it is an abandoned farmland), and has low to moderate nutrient levels. Fungi isolated from this site and tested for host response under low nutrient conditions seemed equally likely to be parasites or mutualists. To complicate matters even more, it does not seem possible to place AMF taxa into one of two cat-

Experiment 2
The majority of plants (home) that co-occur at the LTMRS were successfully infected by Glomus etunicatum (home). Of the 64 plants tested, only the following plants were not found to contain any arbuscules or vesicles at time of harvest: Apocynum cannabinum, Carex aurea, Carex ava, Carex acca, Carex granularis, Chenopodium ambrosioides, Convolvulus arvensis, Juncus dudleyi. Overall, there was a large variation in plant growth response among the plants when grown in the presence of this one AMF isolate (Fig. 6). The majority of responses were small, and not signicantly different from the nonmycorrhizal controls,

2298

JOHN N. KLIRONOMOS

Ecology, Vol. 84, No. 9

Special Feature

mean biomass for each plant species across all AMF treatments was similar to that of their nonmycorrhizal counterparts. This indicates that high mycorrhizal species sensitivity does not necessarily translate to high overall mycorrhizal dependency, even though a positive relationship has been shown between these two variables in at least one other ecosystem (van der Heijden 2002). The composition of AMF communities has been shown to greatly inuence plant diversity and ecosystem functioning at the LTMRS (van der Heijden et al. 1998b), the same site where plant and fungi were collected for the present experiments. Based on the present results, the effect of AMF on plant diversity and productivity reported earlier is not likely a result of differences in mycorrhizal dependency among plant species, since the plant species tested had similar overall mycorrhizal dependency. Mycorrhizal dependency was calculated as a mean response to the various AMF, using a random draw of AMF taxa from the site. However, we know that AMF are not randomly or even homogenously distributed at the site (Hart and Klironomos 2002). Rather they follow different and patchy spatial distributions. Furthermore, there is evidence that plants select for AMF that benet them most (Klironomos 2002), creating a positive feedback between species of plants and AMF in local environments. Thus, at small spatial scales, distinct AMF communities are formed, and certain plants benet more than others, depending on their responses to the local AMF. At larg-

FIG. 4. The frequency distribution of plant-growth responses in Experiment 1 (10 plant species crossed with 10 fungal species).

egories (mutualist or parasite). AMF do have a taxonomic basis to their own growth and development (Hart and Reader 2002), but their inuence on plant growth is highly dependent on the plant genotype with which they are associated. For example, whereas the AMF, Acaulospora morrowiae, stimulated the growth of Rudbeckia hirta by 45%, the same AMF isolate reduced the growth of Plantago lanceolata by 47%. Conversely, the AMF, Gigaspora rosea, stimulated Plantago lanceolata by 41% and reduced the growth of Rudbeckia hirta by 40%. The frequency curve of host responses observed with a single individual AMF (as demonstrated with Glomus etunicatum crossed with 64 plant species) follows a normal distribution (Kolmogorov0.056, df 64, P 0.020). Smirnov Also it is difcult to determine mycorrhizal dependency (the extent to which a plant benets from the presence of AMF) for individual plant species. All plants tested in this study varied widely in their response to individual AMF taxa (high mycorrhizal species sensitivity (see van der Heijden 2002), yet the

FIG. 5. The relationship between mycorrhizal dependency and mycorrhizal species sensitivity (Experiment 1): (a) home arbuscular mycorrhizal fungi vs. home plants, y 0.09x 24.8, r2 0.01, P 0.824; (b) foreign arbuscular mycorrhizal fungi vs. home plants, y 0.32x 13.6, r2 0.17, P 0.724; (c) home arbuscular mycorrhizal fungi vs. foreign 2 plants, y 0.24x 12.0, r 0.71, P 0.792.

September 2003

UNDERGROUND PROCESSES

2299

Special Feature

FIG. 6. The inuence of Glomus etunicatum (home) on the growth of 64 home plants (Experiment 2). Bars represent the percentage change in biomass of mycorrhizal plants compared to nonmycorrhizal controls. The mean 95% condence interval for all the treatments is 17%. The plant species names are in Table 1.

er scales within the plant community, no plant species can dominate because no plant does best with all AMF at the site. By this logic, mycorrhizal species sensitivity may promote plant species coexistence and diversity within the LTMRS, even though overall mycorrhizal dependency is very low. If mycorrhizal species sensitivity is an important determinant of diversity within plant communities, then native AMF will be more inuential on plant communities than exotic AMF. In this study, mycorrhizal species sensitivity was signicantly reduced when exotic AMF or plants were used. This suggests that plant and AMF communities are locally adapted, and that similar, but exotic, AMF taxa affect plants to a lesser degree. This data has implications for restoration practitioners that may wish to add AMF into the soil to help plant establishment and growth (Haselwandter 1997). If the plan is to incorporate AMF inoculum comprised of a single AMF isolate, then the present study

offers little direction. There is no evidence that a single local or foreign AMF isolate can effectively promote growth of all plant species in a community. However, if a goal of restoration is to establish and maintain a diverse plant community, then the present results would suggest the addition of a diverse, and locally adapted, AMF community to the restoration protocol. In summary, AMF can inuence the structure of plant communities, but not because they always stimulate the growth of mycorrhizal plants. It is the parasitismmutualism nature of plant/AMF interactions and the high mycorrhizal species sensitivity of individual plant species that may be the principal factors in the maintenance of plant community structure. Also, even though any one individual exotic AMF may not function any differently from a native counterpart, exotic AMF communities offer less variation in plant response than native AMF. Finally, the present results and conclusions may be of general signicance to all terrestrial

2300

JOHN N. KLIRONOMOS

Ecology, Vol. 84, No. 9

ecosystems, or they may only be applicable to herbaceous plant communities with generally low mycorrhizal dependency. This can only be determined after similar research is conducted in a variety of plant growth forms and ecosystem types.
ACKNOWLEDGMENTS The author would like to thank A. Jones, G. LaPierre, G. Lewis, D. MacIntosh, J. Martin, J. Mosquin, P. Moutoglis, and A. Shaw for lab assistance. This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada. LITERATURE CITED Allen, E. B., and M. F. Allen. 1984. Competition between plants of different successional stages: mycorrhizae as regulators Canadian Journal of Botany 62:26252629. Allen, E. B., and M. F. Allen. 1990. The mediation of competition by mycorrhizae in successional and patchy environments. Pages 367389 in J. B. Grace and D. Tilman, editors. Perspectives on plant competition. Academic Press, New York, New York, USA. Allen, E. B., M. F. Allen, D. J. Helm, J. M. Trappe, R. Molina, and E. Rincon. 1995. Patterns and regulation of mycorrhizal plant and fungal diversity. Plant and Soil 170:47 62. Bever, J. D., J. B. Morton, J. Antonovics, and P. A. Schultz. 1996. Host-dependent sporulation and species diversity of arbuscular mycorrhizal fungi in a mown grassland. Journal of Ecology 84:7182. Brundrett, M., N. Bougher, B. Dell, T. Grove, and N. Malajczuk. 1996. Working with mycorrhizas in forestry and agriculture. ACIAR Monograph 32, Canberra, Australia. Brundrett, M. C., Y. Piche, and R. L. Peterson. 1984. A new method for observing the morphology of vesicular-arbuscular mycorrhizae. Canadian Journal of Botany 62:2128 2134. Ergeton-Warburton, L. M., and E. B. Allen. 2000. Shifts in arbuscular mycorrhizal communities along an anthropogenic nitrogen deposition gradient. Ecological Applications 10:484496. Grime, J. P., J. M. L. Mackey, S. H. Hillier, and D. J. Read. 1987. Floristic diversity in a model system using experimental microcosms. Nature 328:420422. Haas, J. H., B. Bar-Yosef, J. Krikun, R. Barak, T. Markovitz, and S. Kramer. 1987. Vesicular-arbuscular mycorrhizal fungus infestation and phosphorus fertilization to overcome pepper stunting after methyl bromide treatment. Agronomy Journal 79:905910. Habte, M., and A. Manjunath. 1991. Categories of vesiculararbuscular mycorrhizal dependency of host species. Mycorrhiza 1:312. Harley, J. L., and E. L. Harley. 1987. A check-list of mycorrhiza in the British Flora. New Phytologist 105:1102. Hart, M. M., and J. N. Klironomos. 2002. Diversity of arbuscular mycorrhizal fungi and ecosystem functioning. Pages 225242 in M. G. A. van der Heijden and I. Sanders, editors. Mycorrhizal ecology. Springer-Verlag, Berlin, Germany. Hart, M. M., and R. J. Reader. 2002. Taxonomic basis for variation in the colonization strategy of arbuscular mycorrhizal fungi. New Phytologist 153:335344. Hartnett, D. C., and G. W. T. Wilson. 1999. Mycorrhizae inuence plant community structure and diversity in tall grass prairie. Ecology 80:11871195. Haselwandter, K. 1997. Soil micro-organisms, mycorrhiza, and restoration ecology. Pages 6580 in K. M. Urbanska, N. R. Webb, and P. J. Edwards, editors. Restoration ecology and sustainable development. Cambridge University Press, Cambridge, UK.

Helgason, T., T. J. Daniell, R. Husband, A. H. Fitter, and J. P. Y. Young. 1998. Ploughing up the wood-wide web? Nature 394:431. Hetrick, B. A. D., G. W. T. Wilson, and D. C. Hartnett. 1992. Relationship of mycorrhizal symbiosis, rooting strategy and phenology among tall grass prairie forbs. Canadian Journal of Botany 70:15211528. Hewitt, E. 1966. Sand and water culture methods used in the study of plant nutrition. Commonwealth Bureau Technical Communication 22. Farnham Royal, Bucks, UK. Janos, D. P. 1980. Mycorrhizae inuence tropical succession. Biotropica 12:5664. Jensen, A. 1984. Responses of barley, pea and maize to inoculation with different vesicular-arbuscular mycorrhizal fungi in irradiated soil. Plant and Soil 78:315323. Johnson, N. C. 1993. Can fertilization of soil select less mutualistic mycorrhizae? Ecological Applications 3:749 757. Johnson, N. C., J. H. Graham, and F. A. Smith. 1997. Functioning of mycorrhizal associations along the mutualismparasitism continuum. New Phytologist 135:575586. Johnson, N. C., D. R. Zak, D. Tilman, and F. L. Peger. 1991. Dynamics of vesicular-arbuscular mycorrhizae during old eld succession. Oecologia 86:349358. Kiers, E. T., C. E. Lovelock, E. L. Krueger, and E. A. Herre. 2000. Differential effects of tropical arbuscular mycorrhizal fungal inocula on root colonization and tree seedling growth: implications for tropical forest diversity. Ecology Letters 3:106113. Klironomos, J. N. 2000. Host-specicity and functional diversity among arbuscular mycorrhizal fungi. Pages 845 851 in C. R. Bell, M. Brylinsky, and P. Johnson-Green, editors. Microbial biosystems: new frontiers. Proceedings of the Eighth International Symposium on Microbial Ecology. Atlantic Canada Society for Microbial Ecology, Halifax, Canada. Klironomos, J. N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature 417:6770. Koide, R. 1991. Nutrient supply, nutrient demand and plant response to mycorrhizal infection. New Phytologist 117: 365386. Koide, R. T., and M. Li. 1989. Appropriate controls for vesicular arbuscular mycorrhizal research. New Phytologist 111:3544. Marler, M. J., C. A. Zabinski, and R. M. Callaway. 1999. Mycorrhizae indirectly enhance competitive effects of an invasive forb on a native bunchgrass. Ecology 80:1180 1186. Miller, R. M. 1987. Mycorrhizae and succession. Pages 205 219 in W. R. Jordan III, M. E. Gilpin, and J. D. Aber, editors. Restoration ecology, a synthetic approach to ecological research. Cambridge University Press, Cambridge, UK. Newman, E. I., and P. Reddell. 1987. The distribution of mycorrhizas among families of vascular plants. New Phytologist 106:745751. Picone, C. 2000. Diversity and abundance of arbuscular-mycorrhizal fungus spores in tropical forest and pasture. Biotropica 32:734750. Powell, C. L., G. E. Clark, and N. J. Verberne. 1982. Growth response of four onion cultivars to isolates of VA mycorrhizal fungi. New Zealand Journal of Agricultural Research 25:465470. Raju, P. S., R. B. Clark, J. R. Ellis, and J. W. Maranville. 1990. Effects of species of VA-mycorrhizal fungi on growth and mineral uptake of sorghum at different temperatures. Plant and Soil 121:165170. Sanders, I. R., J. P. Clapp, and A. Wiemken. 1996. The genetic diversity of arbuscular mycorrhizal fungi in natural

Special Feature

September 2003

UNDERGROUND PROCESSES

2301

ecosystemsa key to understanding the ecology and functioning of the mycorrhizal symbiosis. New Phytologist 133: 123134. Sanders, I. R., and A. H. Fitter. 1992. Evidence for differential responses between hostfungus combinations of vesicular-arbuscular mycorrhizas from a grassland. Mycological Research 96:415419. Smith, M. D., D. C. Hartnett, and G. W. T. Wilson. 1999. Interacting inuence of mycorrhizal symbiosis and competition on plant diversity in tallgrass prairie. Oecologia 121:574582. Smith, S. E., and D. J. Read. 1997. Mycorrhizal symbioses. Second edition. Academic Press, London, UK. Streitwolf-Engel, R., T. Boller, A. Wiemken, and I. R. Sanders. 1997. Clonal growth traits of two Prunella species are

determined by co-occurring arbuscular mycorrhizal fungi from a calcareous grassland. Journal of Ecology 85:181 191. Van der Heijden, M. G. A. 2002. Arbuscular mycorrhizal fungi as a determinant of plant diversity. Pages 243265 in M. G. A. van der Heijden and I. Sanders, editors. Mycorrhizal ecology. Springer-Verlag, Berlin, Germany. Van der Heijden, M. G. A., T. Boller, A. Wiemken, and I. R. Sanders. 1998a. Different arbuscular mycorrhizal species are potential determinants of plant community structure. Ecology 79:20822091. Van der Heijden, M. G. A., J. N. Klironomos, M. Ursic, P. Moutolis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and I. R. Sanders. 1998b. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396:6972.

Special Feature

Ecology, 84(9), 2003, pp. 23022312 2003 by the Ecological Society of America

MYCORRHIZAL CONTROLS ON BELOWGROUND LITTER QUALITY

J. ADAM LANGLEY1
AND

BRUCE A. HUNGATE

Department of Biological Sciences, Merriam-Powell Center for Environmental Research, Northern Arizona University, Flagstaff, Arizona 86011 USA

Special Feature

Abstract. Plant productivity and ecosystem productivity are strongly inuenced by nutrient availability, which is largely determined by the decomposition rate of plant litter. Belowground litter inputs (dead roots, mycorrhizae, and exudates) are larger than aboveground litterfall in many systems. Chemical quality and diameter primarily control decomposition for coarse roots, but these patterns do not hold for ner classes of roots, which are frequently colonized by mycorrhizae. Though mycorrhizal status is known to drastically alter root chemistry, morphology, life span, and exudation, it has never been explicitly considered as a factor affecting root decomposition. We hypothesize that mycorrhizal status substantially inuences ne root decomposition rates. Both ectomycorrhizal (EM) and arbuscular mycorrhizal (AM) fungi can change root properties but do so in different ways. Dominant tree species of most cold and temperate forests rely heavily on EM associations. EM fungi form massive structures that envelop ne roots. Roots infected by ectomycorrhizae have higher nitrogen concentrations than nonmycorrhizal roots, which would be expected to increase decomposition rates, but much of this nitrogen is bound in recalcitrant forms, such as chitin, so the net effect on decomposition is difcult to predict. AM fungi lack elaborate, macroscopic structures and may not alter root chemistry as profoundly. In addition to mycorrhizal roots, external fungal hyphae can contribute signicantly to ecosystem carbon budgets and thereby inuence rates of soil carbon turnover. Hyphae have commonly been considered a rapidly decomposing carbon pool, though this has never been demonstrated experimentally. If hyphae are produced at the expense of rapidly decomposing root exudates, then the net effect of hyphal litter production might be to reduce soil microbial activity and overall carbon cycling rates. Based on known differences in morphology and chemistry, EM hyphae may be more recalcitrant than AM hyphae. In summary, we submit that mycorrhizal status could substantially inuence ne root decomposition and soil carbon processing rates, potentially explaining some of the variation observed within and among individual plant species and ecosystems.
Key words: arbuscular mycorrhiza; carbon cycling; decomposition; ectomycorrhiza; feedbacks; ne roots; hyphae.

INTRODUCTION The rst widely appreciated mycorrhizal functions were those that most clearly and directly benet individual plants, such as enhanced uptake of nutrients and water and inhibition of root pathogens. More recently, ecologists have recognized ecosystem-level inuences of mycorrhizal colonization, most notably, the maintenance of soil structure (Miller and Jastrow 1990), the potential for active degradation of organic substrates (Harley 1971, Read 1991) and possible facilitation of plant succession (for review, see Hart et al. 2001). Recent realization of rising atmospheric [CO2] has underscored the importance of understanding ecosystem carbon (C) cycling, yet much of the research on mycorrhizal response to elevated CO2 has focused on the mutualistic benets to plant growth. The direct
Manuscript received 6 may 2002; revised 10 September 2002; accepted 16 September 2002; nal version received 31 October 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256. 1 E-mail: Adam.Langley@nau.edu

contributions of mycorrhizal products to ecosystem C and nutrient cycles have yet received only cursory experimental attention (Treseder and Allen 2000). We contend that the mycorrhizal associations, by inuencing belowground litter quality, may directly contribute to the C cycling characteristics of host plants and the ecosystems in which they occur. We will demonstrate how mycorrhizal colonization affects the quality of root tissue and extramatrical products and then survey the implications for ecosystem processes.

Belowground products
The majority of energy and nutrients in soils is derived from detritus of plants. The inputs of root and root-associated turnover and exudation can be as great as aboveground litter inputs in magnitude and may have unique feedbacks to plants (Hendricks et al. 1993). Belowground products are deposited near active roots, creating a close coupling between decomposition and plant nutrient uptake (Gordon and Jackson 2000). The technical challenge of quantifying belowground products, and tracking the fates thereof, has hindered ap-

2302

September 2003

UNDERGROUND PROCESSES

2303

FIG. 1. Representation of carbon (C) uxes into, and out of, the ve primary belowground litters using an ectomycorrhizal example. The relative decomposability of each is represented by the thickness of lines 15. Dotted lines indicate mycorrhizal products. Root litter (1) has a mean residence time on the order of one year while exudates (3) reside on the order of days; the inuence of mycorrhizal colonization on each of these (2, 4) is unknown. The residence time of hyphal litter (5) also remains unknown but is believed to be from weeks to months. The net effect of mycorrhizal colonization on overall belowground litter quality depends on these unknown decay coefcients and on the distribution of C allocation to each type of litter (6 10).

preciation of the importance and complexity of these inputs. Whereas ecologists typically consider only root turnover, large portions of plant-derived C may enter the soil as root exudates or mycorrhizal tissue (Fig. 1). Though the structure, chemistry and decomposition environment of belowground and aboveground detritus differ greatly, chemical parameters successful for predicting aboveground litter decomposition (primarily N and lignin; Melillo et al. 1982) have been used for root litter, often with less success either within or among ecosystems (King et al. 1997, Robinson et al. 1999, Silver and Miya 2001). The ubiquitous, intimate, and often prolic presence of mycorrhizal fungi within and around root systems strongly alters root chemistry, architecture, and rhizoplane biology. Observed inconsistencies in ne root decomposition rates suggest that mycorrhizal status may play an important role in determining root litter quality. Ectomycorrhizal (EM) fungi envelop roots and cortical cells with intensive layering of hyphae. Arbuscular mycorrhizal (AM) fungi form less massive, internal structures. Mycorrhizal fungi divert great portions of plant photosynthate away from the root to support extramatrical hyphae. The decomposition rate and exudant properties of these hyphae remain largely unknown. A number of studies have investigated mycorrhizal inuence on rhizosphere activity, and though the precise mechanism is not always clear, mycorrhizae appear play a role in regulating rhizospheric C processing rates (Rygiewicz and Anderson 1994, Olsson et al. 1996a, b). We hypothesize that the type and intensity of mycorrhizal colonization, by altering quality of roots and rhizosphere products, could substantially inuence ecosystem level decomposition rates.

Differences between mycorrhizal types

Different types of mycorrhizae have distinct morphologies and functions that depend on both plant and

fungal characteristics. The two most abundant types of mycorrhizal fungi (arbuscular and ectomycorrhizal) come from divergent phyla (Kendrick 2000). While there is considerable phenotypic variation within each type of mycorrhiza, we will focus on the differences between AM and EM associations. Such a comparison is useful for generalizing about mycorrhizal roles in ecosystems, where mycorrhizal species may not be known but the mycorrhizal type of dominant plants frequently is known. The microscopic structure of hyphae differs according to fungal phylogeny (Table 1). EM fungal hyphae have thicker walls, pigmentation, and septa between cells, and are generally believed to live longer than AM hyphae (Harley 1971). The same features that may afford a longer lifespan to EM hyphae also appear to render them more resilient to degradation upon senescence. Because their effects on root and hyphal litter quality are likely to differ, we hypothesize that the two major types of mycorrhiza could substantially differ in their effects on C processing rates. The rst section of our argument addresses mycorrhizal alterations of plant roots and the decomposability of the fungusroot. The second section discusses the fate extramatrical mycorrhizal products. The third describes the importance of these mycorrhizal inuences for ecosystem processes. For clarity we will use the term mycorrhizal root to denote the fungus-root symbiosis, consisting of fungal organs that are inseparable from plant roots (mantle, Hartig net, in ectomycorrhizae; vesicles, arbuscules, intramatrical hyphae in arbuscular mycorrhizae). Any extramatrical structures (e.g., hyphae, rhizomorphs) will be referred to specically. This distinction is technically appropriate because root decomposition studies often have not differentiated mycorrhizal roots, and likely include the fungus-root organs. Conversely, extramatrical structures are largely left behind when roots are removed

Special Feature

2304 TABLE 1. Association type AM

J. ADAM LANGLEY AND BRUCE A. HUNGATE

Life history habits of the two most abundant types of mycorrhizal associations. Morphology Hyphae thin-walled, coenocytic Root microscopic internal hyphae, vesicles, arbuscules, coils within root cells macroscopic fungal layering outside roots, hartig net internally between cells Fungal division Glomeromycota Saprotrophic potential entirely dependent on host plant C Plant associates most herbaceous species, many angiosperms and primitive gymnosperms

Ecology, Vol. 84, No. 9

Distribution dominant in tropics and grasslands, interspersed in EMdominated systems dominant in cold coniferous and many temperate forests, spotty in tropics

EM

thick-walled, pigmented, septated

Basidio- and can live freely, enAscomycozymatic capabilita ty varies with species

advanced gymnosperm, many angiosperm trees, dipterocarps

from the soil, and so are not included in conventional root decomposition studies. MYCORRHIZAL INFLUENCES ROOT LITTER QUALITY
ON

Mycorrhizal roots can make up a large portion of belowground production depending on the mycotropism (dependence and type of mycorrhizae) of dominant plant species. Most temperate conifers and many broadleaf tree species host EM fungi. Though direct estimates are rare, EM roots make up as much as 78% of ne root production in coniferous forests of the northwestern United States (Fogel and Hunt 1983). Studies of other EM systems yielded similar large estimates of mycorrhizal production (e.g., Vogt et al. 1981, Markkola et al. 1995). Most herbaceous plants and many trees form arbuscular mycorrhizal associations, though the extent of colonization varies widely (Smith and Read 1997). AM fungi have a less intensive but more extensive presence than EM fungi. Estimates from young potted plants suggest that AM hyphal biomass constitutes around 3% root weight externally (Jakobsen and Rosendahl 1990) and 517% root weight internally (Hepper 1977). In this section, we review the architectural and chemical properties of EM and AM roots compared to nonmycorrhizal roots and present some data on how those traits may affect root decomposition.

Special Feature

from root pathogens and herbivores (Harley and Smith 1983) and could potentially decrease decomposability of dead roots as well, depending on the chemical resilience of the surrounding fungal tissue. The thickness of the Hartig net varies greatly among EM types. In EM angiosperms, hyphal inltration is often restricted to the root epidermal cells; the underlying cortical cells are highly lignied (Brundrett 2002). AM fungi can induce subtle changes in root architecture that do not likely have a strong inuence on root decomposability.

Chemistry
Globally, root chemistry appears to have a stronger inuence on root decomposition than does climate (Silver and Miya 2001). Because plant and fungal tissues differ in chemical composition, the amount of mycorrhizal fungal tissue on the surface of, and inside, roots could be as important as any single chemical parameter in predicting root decomposition. Fungal tissue often contains higher concentrations of nitrogen (N) and phosphorus (P) than plant root tissue (Table 2). Higher concentrations of these nutrients, especially N, in litter are predicted to accelerate decomposition according to aboveground paradigms (Aber and Melillo 1991). However, fungal cell walls contain up to 60% chitin, which is recalcitrant to decomposition (Swift et al. 1979), but also is 6.9% N, much more than most plant tissues (Table 2). Roots with mycorrhizal colonists could have relatively high [N] and actually be less decomposable than uncolonized roots with lower [N]. Such a suppression of decomposition in N-rich roots would be expected in roots that have higher masses of fungi, and thus, more chitin. More massive fungal structures apparently lead to higher chitin contents in EM roots than AM roots (Table 2). AM colonization may not cause a large, chitin-mediated depression of root decomposition. In fact, in some (typically tropical) systems, litter [Ca] is found to correlate well with root decomposition (e.g., Bloomeld et al. 1993). AM colonization may increase root [Ca] by accumulating calcium phosphates resulting in increased decomposition

Architecture
Root architecture can affect decomposition, partly by changing the ratio of surface area to mass of root litter (Berg 1984), thereby minimizing the relative mass of root subjected to direct attack by decomposers. EM colonization can profoundly change root architecture. For example, roots colonized by EM fungi tend to slow or halt apical growth (Smith and Read 1997). EM roots lack ne root hairs, have a short, rounded shape, and are usually no thicker than the nest root class. This change in shape, along with the protection of the EM hyphal sheath, greatly diminishes the surface area of plant tissue exposed to free-living heterotrophs. This sheathing in EM roots is believed to protect live roots

September 2003

UNDERGROUND PROCESSES

2305

TABLE 2. Chemical composition of roots in unknown innate mycorrhizal condition (coarse and ne) and of known mycorrhizal status (arbuscular and ectomycorrhizal), and that of mycorrhizal hyphae. Chemical component (%) N P Ca K Lignin Chitin Roots Coarse ( 2 mm) 0.59 0.071 0.291 0.262 22.41 0
1

Fine ( 2 mm) 0.94 0.051 0.231 0.262 23.61 0

1

Hyphae AM 1.34 0.283 0.49 0.0143 20.35 0.574

5

EM 1.71 0.1510 0.1212 0.2110 33.25 2.566

5

AM 1.58 0.337 0.918 0.908 0 7.734

7

EM 2.2210 0.089 0.099 0.209 0 611

Sources (superscript numbers in table): 1, Silver and Miya (2001), review of 33 root decomposition studies; 2, Gordon and Jackson (2000), review of 56 studies; 3, Giddens and Todd (1984); 4, Frey et al. (1994), means from Greenhouse specimens of Glomus intraradices on red clover; 5, J. Langley and B. Hungate, unpublished data from Pinus edulis (EM) and Helianthus annuus (AM); 6, Wallander et al. (1997), means of ve EM species throughout one year; 7, Smith and Gianinazzini-Pearson (1987), internal hyphae; 8, Weiersbye et al. (1999), mean values from spores of unidentied AM fungi on grass Cynodon dactylon; 9, Wallander et al. (2002), median values from PIXE analysis of 21 eld-collected Rhizopogon spp. and Pinus muricata rhizomorphs; 10, Fogel and Hunt (1983), mean from mixed-species EM roots in Douglas r stand, hyphae likely to be EM mycelium; 11, Markkola et al. (1995); 12, Kottke et al. (1998), median from 17 EM species in German Pinus abies stands. Fine roots from graminoid species (Silver and Miya 2001) which are likely to host arbuscular mycorrhizae.

Special Feature

rates where Ca availability limits decomposition (Silver and Miya 2001). Secondary chemical components, both plant and fungal, could strongly alter the decomposition of roots colonized with fungi. There is some evidence that plants mobilize secondary compounds selectively to mycorrhizal roots, perhaps as a defensive response to initial colonization, or in order to regulate the internal penetration of the fungal hyphae (Brundrett 2002). For example, Marks (1968) reported accumulation of plantderived tannins in the cortex of dead mycorrhizal roots. Moreover, the fungi forming mycorrhizal associations can produce their own secondary compounds that may affect decomposition. The range of potential compounds produced differs among mycorrhizal types. EM fungi, like many dikaryomycetes, can produce antibiotics (Olsson et al. 1996b) and antifungals (Kope et al. 1991) which inhibit saprotrophic soil microbes and could slow root decomposition. Glomeromycetous AM fungi are not known to produce many specialized secondary metabolites, but do store large amounts of lipids in vesicles, dramatically increasing root lipid content (Harley and Smith 1983, Finlay and So derstrom 1991), which should increase root carbon quality. Despite these striking physical, chemical, and functional differences, no studies have included mycorrhizal parameters (colonization, chitin content, etc.) in initial litter quality assessment nor have any studies examined decomposition in light of mycorrhizal status. Based on chemical and architecture parameters, we predict that EM colonization deters decomposition of dead roots compared to uncolonized roots of the same species, while AM colonization does not have a large effect on root decomposition.

Root decomposition studies

No published studies that explicitly measure root decomposition have considered mycorrhizal roots as a separate class of roots. Sequential coring studies have made inferences about turnover of mycorrhizae (e.g., Vogt et al. 1982, Fogel and Hunt 1983), but these are difcult to translate to estimates of decomposition. With accurate live/dead distinctions, one can estimate the turnover rate of live roots, but independent estimates of root decay constants are necessary for compartment ow models that can predict decomposition (Publicover and Vogt 1993). Litterbag studies have measured long term decomposition rates in eld settings for a number of plant species, typically comparing two or more classes of roots dened by diameter. Several such studies of EMdependent tree species tend to nd anomalous patterns for the nest class of roots, which commonly associate with ectomycorrhizae. McClaugherty et al. (1984) surveyed roots from several hardwood and pine species and consistently found that coarse roots decomposed much more rapidly than ner ones (from 22 to 191% more mass loss) despite lower nutrient concentrations ( 50% lower [N] in coarse roots). The authors invoked the modestly higher percentages of nonstructural carbohydrates in the ner roots to explain the pattern. Burke and Raynal (1994) found a similar relationship between mass loss and diameter of roots from a northern hardwood forest, with the nest roots decomposing most slowly. Again, the pattern could not be explained with initial litter chemistry; the nest class of roots had higher concentrations of N and P than intermediate (0.51.5 mm) and larger roots (1.53.0 mm). Camire

2306

J. ADAM LANGLEY AND BRUCE A. HUNGATE

Ecology, Vol. 84, No. 9

et al. (1991) found similar depressions of decomposition in the nest root classes in both Alnus and Populus species, as did Lohmus and Ivask (1995) in Abies, all of which frequently form EM associations. Camire et al. (1991) hypothesized that high N content in these ne roots could stimulate the formation of relatively stable Nlignin complexes and suppress decomposition rates. In loblolly pine, King et al. (1997) found that the 510 mm diameter root class decomposed slightly quicker than the 05 mm root class (k value 0.0018 d 1 vs. 0.0016 d 1) despite having C:N ratios that were consistently twice as high throughout two years of decomposition. In all of these cases, root decomposition appears to be stunted in the smallest diameter class despite having the highest [N]. We suggest that ectomycorrhizae, which occur in ner classes of roots and increase [N], decelerate decomposition rates in these litterbag studies and contribute to the decoupling of [N] and root decomposition rates. Robinson et al. (1999) used microcosms to measure decomposition of mixtures of roots from arctic plants. They found decomposition to be unpredictable by any initial chemical parameter measured (C, N, P, soluble carbohydrates, cellulose, lignin). Based on their qualitative mycorrhizal assessment, the most labile roots were nonmycorrhizal, followed by an AM species, with one lightly EM and one heavily EM species being more recalcitrant. However, mycorrhizal status was neither quantied nor invoked to explain decomposability. In a comprehensive review of root decomposition studies, Silver and Miya (2001) concluded that the indices of litter quality useful for predicting foliar litter decomposition rates (lignin, N) were not accurate predictors of root decomposition. Nor were climatic variables as successful for root litter as for litter deposited on top of the soil. While ne ( 2 mm diameter) roots of broadleaves and conifers have mean concentrations of N, P, Ca, lignin (and various ratios thereof) that are not statistically separable, conifer roots decomposed much more slowly. The conifer roots examined (spruce, r, and pine) are heavily colonized by ectomycorrhizae under most eld conditions (Smith and Read 1997). Within conifer studies, root decomposition had a signicant negative relationship with [N], consistent with a suppression of root decomposition by EM colonization. We compared the initial decomposability of ectomycorrhizal and nonmycorrhizal roots of Pinus edulis in microcosms (Fig. 2; see Plate 1). Regardless of the treatment of the tissue, nonmycorrhizal ne roots decomposed more quickly than mycorrhizal roots despite 0.07% having less than one-third as much N, 0.63 0.12%, respectively (df 6, t 11.1, P and 2.20 0.001). When the tissues were ground, the differences between EM and nonmycorrhizal root decomposition narrowed, suggesting that the mycorrhizal root structure, perhaps the chitinous shell, may contribute to the relative recalcitrance of EM roots. The above

Special Feature

FIG. 2. Cumulative respiration (means 1 SE) after two weeks of decomposition of ectomycorrhizal and nonmycorrhizal roots of Pinus edulis. Fine roots ( 1 mm) were collected near Flagstaff, Arizona, in June 2002, and sorted according to presence of ectomycorrhizal status. A 25-mg sample of each root type (EM and Non) was placed in sterile soil and reinoculated with a slurry of soil from the pine woodland. One group (N) of each root type was allowed to decompose with its native rhizoplane and internal fungi, which could affect decomposition. To control for differing rhizoplane microbial communities, a subset of each type of root was surface sterilized (SL); a sterile inoculum with equivalent organic content was given to half of the surface sterilized samples (SS) in order to compare the relative decomposition potential of internal fungi. To control for differing internal fungi, a subset of ecto- and nonmycorrhizal roots was autoclaved (A). To estimate the importance of architecture on decomposition a set of autoclaved groups was nely ground (AG). Results of t tests (assuming unequal variances) comparing each group of EM and nonmycorrhizal roots (n 5) are as follows: P 0.10; **P 0.01; NS, no signicant difference.

roots were collected in early summer, the very driest part of an extreme drought. Previously, a similar incubation (but without the above sterilization controls) was performed on P. edulis roots collected in the winter, a much wetter time. In this case, the EM roots, with native associated community intact, decomposed more quickly than nonmycorrhizal roots, implying that the seasonal viability of the fungus could contribute to the root decomposition rate. In a laboratory culturing study, Kerley and Read (1998) found that more N was mineralized from substrate consisting of ericoid mycorrhizal than nonmycorrhizal root necromass, but only in the presence of viable ericoid mycorrhizae and a host plant. Similarly, the decomposition of EM root litter may in some cases depend on presence of viable fungal hyphae (see Feedbacks). EXTRAMATRICAL PRODUCTS We have discussed the apparent mycorrhizal role in root decomposition, but their greatest inuence may be manifested outside the root. Up to one-fth of anabolized photosynthate may enter the soil in the form of mycorrhizal hyphae (Finlay and So derstrom 1991). The direct inputs to the rhizosphere merit particular attention (Table 3). In this section we describe the qual-

September 2003

UNDERGROUND PROCESSES

2307

Special Feature

PLATE 1. Comparative architecture of (A) an EM root and (B) an uncolonized ne root of Pinus edulis. We placed equivalent masses of each root type in litter bags and buried them under pinyon canopies at Sunset Crater National Monument, near Flagstaff, Arizona. Subsets of litterbags were placed in decomposition cores (C and D). The cores have open windows covered with 45- m nylon mesh to allow hyphal inltration and exclude growing roots (C). One core from each pair (D) is periodically rotated to sever and thus exclude ingrowing hyphae as well. This experiment allowed us to examine the interaction between EM status of decomposing roots and the presence of viable, connected EM hyphae on root decomposition. Photo in panel (A) courtesy of K. Haskins; panels (BD), A. Langley.

TABLE 3.

The inuence of extramatrical inputs on rhizosphere activity.

Mycorrhizal type Arbuscular

Effect on microbial activity Three times the bacterial CFUs in rhizoplane of AM plants.

Reference Meyer and Linderman (1986) Jakobsen and Rosendahl (1990) Olsson et al. (1996a) Johnson et al. (2002) Katznelson et al. (1962) Neal et al. (1964) Rygiewicz and Anderson (1994) Olsson et al. (1996b) Setala et al. (1999)

Ectomycorrhizae

Proportion of assimilate in belowground respiration twice as high in AM as AM . No AM effect after 30 d with live hyphae. Subsequent plant-free soil incubations from AM treatment had increased bacterial activity through 17 d. Approximately 5% of xed C was transported through AM fungi and respired within 21 h. Could not distinguish between AM hyphal and heterotrophic CO2 production. EM roots supported bacterial types with more complex nutritional requirements. Bacterial counts of EM roots could be higher or lower than EM roots depending on morphotype. Increased root/ectomyocrrhizal respiration. Fate of exudates and mycorrhizal products not examined. EM presence decreased bacterial numbers and reduced rates of bacterial activity. No change in nematode abundance with EM addition despite greater root mass. Considered EM structures to be recalcitrant C sink.

2308

J. ADAM LANGLEY AND BRUCE A. HUNGATE

Ecology, Vol. 84, No. 9

ities of EM and AM extramatrical products, and their inuence on organic matter decomposition, highlighting studies that have isolated the inuence of mycorrhizal inputs on rhizosphere activity.

Hyphae
Both AM and EM hyphae (Finlay and Soderstrom 1991) can comprise the largest portion of soil microbial biomass depending on the type of system. The quality of hyphae as a microbial substrate is not known, but morphological differences between fungal phyla may indicate patterns of relative decomposability among mycorrhizal fungi. AM-forming fungi, which are glomeromycetes, produce large-diameter, thin-walled hyphae that have no cross-wall barriers to stem the leakage of cytoplasm if a cell wall is breached by soil organisms or disturbance. Ectomycorrhiza-forming fungi, which are dikaryomycetes, typically have thinner hyphae with thicker walls that protect cytoplasm with septa between cells (Kendrick 2000). They commonly form rhizomorphs, aggregates of hyphae with radial specialization, that decrease their surface area to volume ratio. Based on these morphological characteristics, EM hyphae would seem more resistant to microbial degradation. AM and EM hyphae may differ in chemical quality as well. Extracting hyphae purely from one type of fungus in quantities large enough to perform conventional chemical analyses is extremely difcult. Recent advances in particle-induced x-ray emission (PIXE) in conjunction with microscopic imaging techniques have afforded rened estimations of chemical composition of individual eld-grown mycorrhizal hyphae (Weiersbye et al. 1999, Wallander et al. 2002). These few analyses have reported extremely variable values for some elements and are highly dependent on the substrate in which hyphae were grown. The available results suggest that EM hyphae may have lower concentrations of nutrients such as Ca and K than AM hyphae (Table 2), which may indicate a relatively high ratio of recalcitrant structural to labile cytoplasmic components in EM fungi.

Special Feature

and 9, in Fig. 1) by increasing photosynthetic rates and directing higher proportions of photosynthate belowground. It is not known how much exudation originates from the plant and how much is from associated fungi (Grayston et al. 1996). Much less work has been done regarding the effects of EM fungi on exudation from tree roots. Many of the proteinaceous exudates from EM fungi are in the form of proteolytic, cellulytic and lignolytic enzymes (Harley and Smith 1983), compounds that could deter other potential decomposers. A tremendous advance in the recognition of the inuence of external mycorrhizal products on soil C has been the discovery and study of glomalin, a glycoprotein produced by AM hyphae (Wright and Upadhyaya 1996). In addition to the indirect consequences of soil structure for soil C cycling, Rillig et al. (2001) have recently acknowledged the direct inputs of recalcitrant C inputs in the form of glomalin. Using 14C over a Hawaiian chronosequence they estimated the mean turnover time to be roughly on the scale of a decade. This exudate represents a recalcitrant pool that builds up and disappears slowly and has primarily been implicated in maintaining soil structure. Whereas AM hyphae may be readily labile, production of glomalin may retard turnover of soil carbon. The balance of the two apparently opposing inuences on total soil C processing is not known.

Mycorrhizal inuence on rhizosphere C processing

Photosynthate transported belowground can immediately enter the soil in the form of exudates and fungal tissues, both of which may be subject to rapid heterotrophic degradation. The distinction between this pathway of soil respiration and root/mycorrhizal respiration may be a semantic one, and teasing the two apart presents a daunting technical obstacle, particularly in the eld. Using root-free hyphal compartments and controlled inoculations, it is possible to make inferences about the inuence of mycorrhizal colonization on the quantity and quality of these ephemeral inputs. Laboratory microcosms employing such techniques have contributed evidence about the direction of mycorrhizal inuence on C processing. Because of the short observation period of these experiments, the contribution of root input is negligible. More rapid pathways of photosynthetic input into soil, hyphae and exudation, drive patterns of energy ow into soil. Olsson et al. (1996a, b) measured microbial thymidine and leucine incorporation to assess microbial activity. Inoculation with each of six EM fungi suppressed bacterial numbers and activity in Pinus contorta pots. The pattern was attributed to a diversion of C from rich exudates into recalcitrant EM structures. In a similar experiment, they found that live AM associations had no effect. However, rhizospheric soil around colonized roots did support greater activity when removed from the plant, than uncolonized roots. These results suggest that dead AM hyphae and as-

Hyphal and root exudates

Along with root and hyphal products, root exudation is an important pathway of organic matter deposition into soil, though they are infrequently considered a component of plant litter. Root systems can exude a large portion of photosynthate in the form of simple compounds like sugars, amino acids, and organic acids (Grayston et al. 1996). Arbuscular mycotropic plants tend to exude more rich sugars than nonmycorrhizal plants under mycorrhizal-sterile conditions, but reduce upon colonization (Schwab et al. 1984), indicating a potential tradeoff between root exudation (ux 8 in Fig. 1) and mycorrhizal C sink activity (mycorrhizal C pathways 7, 9, 10). Mycorrhizal colonization, however, can increase total the amount of exudates (both uxes 8

September 2003

UNDERGROUND PROCESSES

2309

sociated products decompose easily compared to other belowground sinks such as EM hyphae and root exudates. Setala et al. (1999) inoculated Pinus sylvestris microcosms with fungi and different guilds of nematodes. The presence of EM fungi greatly increased the ow of C below ground, but the energetic increase did not cascade up to fungivores and predators. This study, too, concluded that EM fungi tied up C in relatively inaccessible pools. Ecologists have contrastingly regarded dead hyphae as labile (e.g., Fitter et al. 2000, Lindahl et al. 2002) and recalcitrant (e.g., Setala et al. 1999, Treseder and Allen 2000) despite the lack of published reports of hyphal decomposition rates. Taken together, the studies above demonstrate that mycorrhizae can redistribute soil C within the rhizosphere, and the ultimate fate of that C may depend on mycorrhizal type. Based on their chemistry, architecture, and inuence on microbial activity, we propose that EM products are less labile than AM ones. Equally as important and elusive as hyphal decomposition are the effects of mycorrhizal status on C allocation to, and fate of, alternative C pools. For instance, if EM inoculation depresses rhizospheric microbial activity in a short-term incubation (in which root turnover is negligible) as in Setala et al. 1999, then either EM hyphae or EM exudates must be recalcitrant compared to the C exuded from roots of nonmycorrhizal plants (in Fig. 1, the sum of uxes 4 and 5 must ux 3). be Translating the effect of hyphal inputs from microcosms to intact ecosystems is further complicated by the other inuences imposed by mycorrhizal hyphae, which can alter soil factors and host physiology, indirectly affecting soil C processing (Olsson et al. 1996b). The production, functions and fates of mycorrhizal hyphae and mycospheric exudates remain difcult to investigate, particularly in intact ecosystems. Controlled inoculum and substrate laboratory studies like those above have outlined the parameters of mycorrhizal potential to inuence C quality and processing in the rhizosphere. The use of isotope tracers and gas exchange techniques in tandem with soil meshes that allow ingrowth of hyphae but exclude roots has recently provided promise for investigating the effects of mycorrhizal status on belowground C processing rates even in the eld (Johnson et al. 2002). FEEDBACKS Ecologists have recently recognized the relationship of plant mycorrhizal status and plant C cycling traits (Cornelissen et al. 2001), as well as the importance of senescent mycorrhizal hyphae in ecosystem N cycling (Lindahl et al. 2002). It is commonly hypothesized that plant species can inuence C and nutrient cycling via inputs of aboveground litters (Hobbie 1992) which can feed back to inuence plant competition (Berendse 1994). We contend that mycorrhizal inuence on be-

lowground litter quality may contribute to such plant littersoil feedbacks. Belowground litter production often exceeds that aboveground (Fogel and Hunt 1983, Hendricks et al. 1993) and may have an inordinate importance because it is deposited below the soil surface where microenvironments are often more favorable for rapid nutrient release and uptake (Berg 1984). Unlike most aboveground tissues, different portions of root systems can grow and die simultaneously in close proximity and so, plant root litter decomposition dynamics could have important feedbacks directly to the nutrient availability for an individual plant. We hypothesize that inputs of mycorrhizal litter can directly inuence nutrient availability in the soil environment where they occur and these inuences feed back to the nutrition and productivity of the host plants and possibly to the composition of plant communities.

Nitrogen cycling
Based on laboratory and eld decomposition studies, EM fungi, through their alteration of roots and rhizosphere products, appear to render belowground litters less accessible to N mineralizing microorganisms. In actual ecosystems, intact hyphal networks may partially compensate for the poor litter quality apparent in litterbag studies, which disturb the soil, by rapidly mobilizing hyphae with specialized enzymes to the site of degradation (Leake et al. 2001). It has long been recognized that mycorrhizal fungi may directly mineralize organic N and transfer it to host plants (Went and Stark 1968). Direct mineral cycling by mycorrhizal fungi will be signicant in systems where litter quality is poor and nutrients are not readily available to freeliving heterotrophs (Read 1991). For instance, Northrup et al. (1994) proposed that pines produce litter with high phenolic content that may be relatively more accessible to the ectomycorrhizae that are associated with the pine than to other microbes and plants. We believe this direct mycorrhizal mineralization could be much tighter and quite common in litters deposited underground. EM fungi appear to produce recalcitrant litters and are well positioned to rst access the nutrients therein. They have close proximity to senescing root tissues as well as an enzymatic specialization in degrading plant cell walls (Cairney and Burke 1994), and some ability to digest fungal ones like chitin (Leake and Read 1990). Accordingly, intramatrical EM hyphae, unlike other types of mycorrhizal hyphae, appear to remain viable throughout root senescence (Harley and Smith 1983). If EM hyphae remain connected to viable roots, a host plant could subsidize the fungus with C and energy that free-living organisms must obtain from soil substrates (Harley 1971). Indeed, hyphal connections from live to dead roots are commonly observed (Went and Stark 1968). Live EM hyphae are also well positioned to rst colonize and efciently exploit other mycorrhizal hy-

Special Feature

2310

J. ADAM LANGLEY AND BRUCE A. HUNGATE

Ecology, Vol. 84, No. 9

Special Feature

phae as a nutrient source (Lindahl et al. 2002). The sum of this evidence indicates that ectomycorrhizae may largely contribute to their own mineralization in soil. This internalized mycorrhizal recycling could have implications for N competition among plants and soil heterotrophs (Kaye and Hart 1997). In systems where N is limiting, conservative plants can retranslocate up to half of foliar N upon leaf senescence, but are not believed to retranslocate significant amounts of N or P from senescing roots (Nambiar and Fife 1991, Gordon and Jackson 2000). Fine roots of perennial plants grow and die simultaneously. EM associations, by directly accessing nutrients in dead roots, may allow an individual plant to reallocate crucial nutrients from inefcient or dying roots/mycorrhizae to support new tissues incurring only an energetic cost. Likewise, EM hyphal networks could transfer nutrients from dead roots of one plant, to other live plants of the same or different species (Harley and Smith 1983). As well as preserving N for individual plants, the accumulation of N-rich recalcitrant EM structures could represent a mechanism for N retention in forest ecosystems. Temperate coniferous forests, where EM fungi occur in great abundance, tend to retain higher proportions of N inputs than temperate broadleaf forests (broadleaf systems have 7.8 times higher leaching rates than coniferous forests but only 2.7 times the inputs; summarized from Johnson 1992 and references therein). Though other mechanisms are certainly active, we believe that EM immobilization can contribute to the retentive properties of EM dominated systems.

growth may give way to those that can access the nutrients bound in EM structures. Mycorrhizal inuence on belowground litter quality, coupled with mycorrhizal control of plant ability to access that litter, could represent an important feedback pathway inuencing plant community diversity and dynamics.

Consequences for ecosystems

EM root and fungal structures, owing to chemical and structural qualities, appear to be more recalcitrant than AM structures. These differences could have important consequences for explaining ecosystem-level C cycling among system types. Forests with dominant trees reliant on EM fungi, tend to have higher proportions of litter from belowground sources (Vogt et al. 1986) and lower rates of total soil respiration than other forest types when controlling for covarying soil and climatic factors (Raich and Tufekcioglu 2000). Large amounts of mycorrhizal litter could contribute to slower heterotrophic respiration rates in EM dominated ecosystems. Such systems also have higher proportions of soil respiration from autotrophic sources, namely roots and mycorrhizae (reviewed in Raich and Tufekcioglu 2000), also suggestive of mycorrhizal domination of soil activity. Mycorrhizal processing of belowground litter that is relatively inaccessible to free-living heterotrophs could contribute to these global patterns in soil processes. Effects of mycorrhizae on decomposition could also be important in light of ongoing global environmental change. For both AM and EM species, the degree of mycorrhizal infection is sensitive to a number of anthropogenic perturbations, including N deposition, elevated CO2, ozone, and land-use change, to name a few (reviewed in Cairney and Meharg 1999). The implications of such changes in mycorrhizal infection for plant nutrient acquisition, water relations, and pest resistance has been studied extensively (Harley 1971, Smith and Read 1997). Additionally, however, through the mechanisms outlined here, such changes in mycorrhizal status could also modulate the effects of these environmental changes on decomposition and soil C processing. Shifting allocation to AM structures under elevated CO2, for example, has been predicted to increase soil C sequestration in fungal products (Treseder and Allen 2000). Though glomalin appears to sequester C because of its recalcitrant nature (Rillig et al. 2001), enhanced AM hyphal production could stimulate soil microbial activity causing a net acceleration of CO 2 return to the atmosphere (as seen in Jakobsen and Rosendahl 1990, Olsson et al. 1996a). Similarly, we have argued above that morphological and chemical characteristics of EM structures and associated hyphae may retard their decomposition compared to uncolonized roots or root exudates. If this is the case, increased allocation to EM structures in response to elevated CO2 could slow soil carbon turnover. Other factors that inuence mycorrhizal status, such as herbivory (Gehring

Plant community dynamics

A plant species, by altering abiotic resources and soil communities, can render the affected soil more or less suitable for its own propagation, thereby potentially affecting plant succession and maintenance of plant diversity within a system (Bever et al. 1997). For example, it has been suggested that the changes in the quality of aboveground litter inputs, from herbaceous to woody litters, during succession may lead to the shift from AM to EM plants, the latter of which can access more recalcitrant organic compounds, in temperate forest systems (Pankow et al. 1991). Belowground plant products enter the soil in close spatial and temporal proximity to nutrient uptake and could drive such feedbacks. The differences between belowground litters produced by AM and EM plants could also contribute to such patterns observed throughout ecosystem development. EM roots can account for up to 5% of organic mass in soil, but the turnover of EM roots can account for 50% of organic matter throughput in coniferous forests; hyphae (mostly EM) may contribute another 31% of total throughput (Fogel and Hunt 1983). The accumulation of EM detritus could sequester large portions of ecosystem N in recalcitrant organic forms. Plants that rely on readily available mineral N for

September 2003

UNDERGROUND PROCESSES

2311

and Whitham 1991) and natural disturbance (Harley 1971), could also elicit indirect effects on decomposition and soil carbon processing. To our knowledge, the hypotheses outlined here have neither been tested nor articulated previously, yet anomalous patterns of ne root decomposition, suggest that mycorrhizae could be important for ne root decomposition in many terrestrial ecosystems. Likewise, the fate of extramatrical mycorrhizal products (hyphae, exudates) in soils remains unknown but warrants further experimental attention. Each belowground product must be examined a context that allows for identication of complex feedbacks to communities and ecosystems.
ACKNOWLEDGMENTS This research was supported by the Andrew W. Mellon Foundations program in Conservation and the Environment. We thank A. Agrawal, S. Chapman, C. Gehring, and anonymous reviewers for essential advice and assistance. LITERATURE CITED Aber, J., and J. M. Melillo. 1991. Terrestrial ecosystems. Saunders College Publishing, San Francisco, California, USA. Berendse, F. 1994. Litter decomposabilitya neglected component of plant tness. Journal of Ecology 82:187190. Berg, B. 1984. Decomposition of root litter and some factors regulating the process: long-term root litter decomposition in a Scots pine forest. Soil Biology and Biochemistry 16: 609618. Bever, J. D., K. M. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: the utility of the feedbacks approach. Journal of Ecology 85:561573. Bloomeld, J., K. A. Vogt, and D. J. Vogt. 1993. Decay rate and substrate quality of ne roots and foliage of two tropical tree species in the Luquillo Experimental Forest, Puerto Rico. Plant and Soil 150:233245. Brundrett, M. C. 2002. Coevolution of roots and mycorrhizas of land plants. New Phytologist 154:275304. Burke, M. K., and D. J. Raynal. 1994. Fine root growth phenology, production, and turnover in a northern hardwood forest ecosystem. Plant and Soil 162:135146. Cairney, J. W. G., and R. M. Burke. 1994. Fungal enzymes degrading plant cell walls: their possible signicance in the ectomycorrhizal symbiosis. Mycological Research 98: 13451356. Cairney, J. W. G., and A. A. Meharg. 1999. Inuences of anthropogenic pollution on mycorrhizal fungal communities. Environmental Pollution 106:169182. Camire, C., B. Cote, and S. Brulotte. 1991. Decomposition of roots of black alder and hybrid poplar in short-rotation plantings: nitrogen and lignin control. Plant and Soil 138: 123132. Cornelissen, J. H. C., R. Aerts, B. Cerabolini, M. J. A. Werger, and M. G. A. van der Heijden. 2001. Carbon cycling traits of plant species are linked with mycorrhizal strategy. Oecologia 129:611619. Finlay, R., and B. Soderstrom. 1991. Mycorrhiza and carbon ow to soil. Pages 134162 in M. Allen, editor. Mycorrhizal functioning. Routledge, Chapman and Hall, New York, New York, USA. Fitter, A. H., A. Heinemeyer, and P. L. Staddon. 2000. The impact of elevated CO2 and global climate change on arbuscular mycorrhizas: a mycocentric approach. New Phytologist 147:179188.

Fogel, R., and G. Hunt. 1983. Contribution of mycorrhizae and soil fungi to nutrient cycling in a Douglas-r ecosystem. Canadian Journal of Forest Research 13:219232. Frey, B., A. Vilarino, H. Schuepp, and J. Airnes. 1994. Chitin and ergosterol content of extraradical and intraradical mycelium of the vesiculararbuscular mycorrhizal fungus Glomus intraradices. Soil Biology and Biochemistry 26:711 717. Gehring, C. A., and T. G. Whitham. 1991. Herbivore-driven mycorrhizal mutualism in insect-susceptible pinyon pine. Nature 353:556557. Giddens, J., and R. L. Todd. 1984. Rhizosphere microorganismsoverview. Pages 5158 in R. L. Todd and J. E. Giddens, editors. Microbialplant interactions. ASA Special Publication number 47. American Society for Agronomy, Madison, Wisconsin, USA. Gordon, W. S., and R. B. Jackson. 2000. Nutrient concentrations in ne roots. Ecology 81:275280. Grayston, S. J., D. Vaughan, and D. Jones. 1996. Rhizosphere carbon ow in trees, in comparison with annual plants: the importance of root exudation and its impact on microbial activity and nutrient availability. Applied Soil Ecology 5: 2956. Harley, J. L. 1971. Fungi in ecosystems. Journal of Ecology 59:653668. Harley, J. L., and S. E. Smith. 1983. Mycorrhizal symbiosis. Academic Press, New York, New York, USA. Hart, M. M., R. J. Reader, and J. N. Klironomos. 2001. Lifehistory strategies of arbuscular mycorrhizal fungi in relation to their successional dynamics. Mycologia 93:1186 1194. Hendicks, J. J., K. J. Nadelhoffer, and J. D. Aber. 1993. Assessing the role of ne roots in carbon and nutrient cycling. Trends in Ecology and Evolution 8:174178. Hepper, C. M. 1977. A colorimetric method for estimating vesiculararbuscular mycorrhizal infection in roots. Soil Biology and Biochemistry 9:1518. Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336339. Jakobsen, I., and L. Rosendahl. 1990. Carbon ow into soil and external hyphae from roots of mycorrhizal cucumber roots. New Phytologist 115:7783. Johnson, D. W. 1992. Nitrogen retention in forest soils. Journal of Environmental Quality 21:112. Johnson, D., J. R. Leake, N. Ostle, P. Ineson, and D. J. Read. 2002. In situ 13CO2 pulse-labelling of upland grassland demonstrates a rapid pathway of carbon ux from arbuscular mycorrhizal mycelia to the soil. New Phytologist 153: 327334. Katznelson, H., J. W. Rouatt, and E. A. Peterson. 1962. The rhizosphere effect of mycorrhizal and non-mycorrhizal roots of yellow birch seedlings. Canadian Journal of Botany 40:377382. Kaye, J. P., and S. C. Hart. 1997. Competition for nitrogen between plants and soil microorganisms. Trends in Ecology and Evolution 12:139143. Kendrick, B. 2000. The fth kingdom. Focus Publications, R. Pullins, Newburyport, Massachusetts, USA. Kerley, S. J., and D. J. Read. 1998. The biology of mycorrhizal in the Ericacea XX. Plant and mycorrhizal necromass as nitrogenous substrates for the ericoid mycorrhizal fungus Hymenoscyphus ericae and its host. New Phytologist 139: 353360. King, J. S., H. L. Allen, P. Dougherty, and B. R. Strain. 1997. Decomposition of roots in loblolly pine: effects of nutrient and water availability and root size class on mass loss and nutrient dynamics. Plant and Soil 195:171184. Kope, H. H., Y. S. Tsantrizos, J. A. Fortin, and K. K. Ogilvie. 1991. p-Hydroxybenoylformic acid and R-(-)-p-hydroxymandelic acid, two antifungal compounds isolated from the

Special Feature

2312

J. ADAM LANGLEY AND BRUCE A. HUNGATE

Ecology, Vol. 84, No. 9

liquid culture of the ectomycorrhizal fungus Pisolitus arhizus. Canadian Journal of Microbiology 37:258264. Kottke, I., X. M. Qian, K. Pritsch, I. Haug, and F. Oberwinkler. 1998. Xerocomus badisPicea abies, an ectomycorrhiza of high activity and element storage in acidic soil. Mycorrhiza 7:267275. Leake, J. R., D. P. Donnelly, E. M. Saunders, L. Boddy, and D. J. Read. 2001. Rates and quantities of carbon ux to ectomycorrhizal mycelium following 14C pulse labeling of Pinus sylvestris seedlings: effects of litter patches and interaction with a wood-decomposer fungus. Tree Physiology 21:7182. Leake, J. R., and D. J. Read. 1990. Chitin as a source for mycorrhizal fungi. Mycological Research 94:993995. Lindahl, B. O., A. F. S. Taylor, and R. D. Finlay. 2002. Dening constraints on carbon cycling in boreal forests towards a less phytocentric perspective. Plant and Soil 242:123135. Lohmus, K., and M. Ivask. 1995. Decomposition and nitrogen dynamics of ne roots of Norway spruce (Picea abies) at different sites. Plant and Soil 168169:8994. Markkola, A. M., R. Ohtonen, O. Tarvainen, and U. AhonenJonnath. 1995. Estimates of fungal biomass in Scots pine stands on an urban pollution gradient. New Phytologist 131:139147. Marks, G. C., N. Ditchburne, and R. C. Foster. 1968. Quantitative estimates of mycorrhiza populations in radiata pine forests. Australian Forestry 32:2638. McClaugherty, C. A., J. D. Aber, and J. M. Melillo. 1984. Decomposition dynamics of ne roots in forested ecosystems. Oikos 42:378386. Melillo, J. M., J. D. Aber, and J. F. Muratore. 1982. Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecology 63:621626. Meyer, J. R., and R. G. Linderman. 1986. Selective inuence on populations of rhizosphere or rhizoplane bacteria and actinomycetes by mycorrhizas formed by Glomus fasciculatum. Soil Biology and Biochemistry 18:191196. Miller, R. M., and J. D. Jastrow. 1990. Hierarchy of roots and mycorrhizal fungal interactions with soil aggregation. Soil Biology and Biochemistry 5:579584. Nambiar, E. K. S., and D. F. Fife. 1991. Nutrient retranslocation in temperate conifers. Tree Physiology 9:185207. Neal, J. L., W. B. Bollen, and B. Zak. 1964. Rhizosphere microora associated with mycorrhizae of Douglas r. Canadian Journal of Microbiology 10:259265. Northrup, R. R., Z. Yu, R. A. Dahlgren, and K. A. Vogt. 1994. Polyphenol control of nitrogen release from pine litter. Nature 377:227229. Olsson, P. A., E. Baath, I. Jakobsen, and B. Soderstrom. 1996a. Soil bacteria respond to presence of roots but not to mycelium of arbuscular mycorrhizal fungi. Soil Biology and Biochemistry 28:463470. Olsson, P. A., M. Chalot, E. Baath, R. D. Finaly, and B. Soderstrom. 1996b. Ectomycorrhizal mycelia reduce bac terial activity in a sandy soil. FEMS Microbiology Letters 21:7786. Pankow, W., T. Boller, and A. Wiemken. 1991. The signicance of mycorrhizas for protective ecosystems. Experientia 47:391394. Publicover, D. A., and K. A. Vogt. 1993. A comparison of methods for estimating forest ne root production with respect to sources of error. Canadian Journal of Forest Research 23:11791186. Raich, J. W., and A. Tufekcioglu. 2000. Vegetation and soil respiration: correlations and controls. Biogeochemistry 48: 7190.

Read, D. J. 1991. Mycorrhizas in ecosystems. Experientia 47:376391. Rillig, M. C., S. F. Wright, K. A. Nichols, W. F. Schmidt, and M. S. Torn. 2001. Large contribution of arbuscular mycorrhizal fungi to soil carbon pools in tropical forest soils. Plant and Soil 233:167177. Robinson, C. H., J. B. Kirkham, and R. Littlewood. 1999. Decomposition of root mixtures from high arctic plants: a microcosm study. Soil Biology and Biochemistry 31:1101 1108. Rygiewicz, P. T., and C. P. Anderson. 1994. Mycorrhizae alter quality and quantity of carbon allocated belowground. Nature 369:5860. Schwab, S. M., R. T. Leonard, and J. A. Menge. 1984. Quantitative and qualitative comparison of root exudates of mycorrhizal and nonmycorrhizal plant species. Canadian Journal of Botany 62:12271231. Setala, H., P. Kulmala, J. Mikola, and A. M. Markkola. 1999. Inuence of ectomycorrhiza on the structure of detrital food webs. Oikos 87:113122. Silver, W. L., and R. K. Miya. 2001. Global patterns in root decomposition: comparisons of climate and litter quality effects. Oecologia 129:407419. Smith, S. E., and V. Gianninazi-Pearson. 1987. Enzymic separation of VA mycorrhizal fungus from root: characteristics of the fungus. Proceedings of the Seventh North American Conference on Mycorrhizae, Gainesville, Florida, USA. Smith, S. E., and D. J. Read. 1997. Mycorrhizal symbiosis. Academic Press, San Diego, California, USA. Swift, M. J., O. W. Heal, and J. M. Anderson. 1979. Decomposition in terrestrial ecosystems. University of California Press, Berkeley, California, USA. Treseder, K. K., and M. F. Allen. 2000. Mycorrhizal fungi have a potential role in soil carbon storage under elevated CO2 and nitrogen deposition. New Phytologist 147:189 200. Vogt, K. A., R. L. Edmonds, and C. C. Grier. 1980. Seasonal changes in biomass and vertical distribution of mycorrhizal and brous-textured conifer ne roots in 23- and 180-yearold subalpine Abies amabilis stands. Canadian Journal of Forest Research 11:223229. Vogt, K. A., C. C. Grier, C. E. Meier, and R. L. Edmonds. 1982. Mycorrhizal role in net primary production and nutrient cycling in Abies amabilis ecosystems in western Washington. Ecology 63:370380. Vogt, K. A., C. C. Grier, and D. J. Vogt. 1986. Production, turnover, and nutrient dynamics of above- and belowground detritus of world forest. Advances in Ecological Research 15:303377. Wallander, H., L. Johansson, and J. Pallon. 2002. PIXE analysis to estimate the elemental composition of ectomycorrhizal rhizomorphs grown in contact with different minerals in forest soil. FEMS Microbiology Ecology 39:147156. Wallander, H., H. B. Massicotte, and J. E. Nylund. 1997. Seasonal variation in protein, ergosterol and chitin in ve morphotypes of Pinus sylvestris L. ectomycorrhizae in a mature Swedish forest. Soil Biology and Biochemistry 29(1):4553. Weiersbye, I. M., C. J. Straker, and W. J. Przybylowicz. 1999. Micro-PIXE mapping of elemental distribution in arbuscular mycorrhizal roots of the grass, Cynodon dactylon, from gold and uranium mine tailings. Nuclear Instruments and Methods in Physics Research B 158:335343. Went, F. W., and N. Stark. 1968. The biological and mechanical role of soil fungi. Proceedings of the National Academy of Sciences (USA) 60:497504. Wright, S. F., and A. Upadhyaya. 1996. Extraction of an abundant and unusual protein from soil and comparison with hyphal protein of arbuscular mycorrhizal fungi. Soil Science 161:575586.

Special Feature

Ecology, 84(9), 2003, pp. 23132321 2003 by the Ecological Society of America

DEFINING A PLANTS BELOWGROUND ZONE OF INFLUENCE

BRENDA B. CASPER,1 H. JOCHEN SCHENK,2
2

AND

ROBERT B. JACKSON3

1 Department of Biology, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6018 USA Department of Biological Science, California State University Fullerton, P.O. Box 6850, Fullerton, California 92834-6850 USA 3Department of Biology and Nicholas School of the Environment and Earth Sciences, Duke University, Durham, North Carolina 27708 USA

Abstract. The concept of a zone of inuence, the area over which a plant alters the environment, forms the basis of many models of plant competition. Because of logistical difculties, we actually know little about the sizes and shapes of zones of inuence belowground. Here we advocate obtaining data on plants belowground zones of inuence, including the length and distribution of lateral roots, in order to understand better how plants respond to their abiotic soil environment and to other plants. We provide several examples from recent work. First, we present an analysis of a large global data set which shows that maximum lateral root spread correlates with canopy size but that, for a given canopy size, maximum lateral root spread is greater in arid environments and in coarse textured soils. Second, we use an experiment with the weedy annual Abutilon theophrasti to show how using nutrient analogs as tracers yields information about lateral root distributions within populations. In our experimental populations, the belowground zone of inuence extended well beyond the closest neighboring plants. Overlap in zones of inuence increased in nutrient patches. Third, we propose a new conceptual model of belowground zones of inuence based on these and other data sets. The model assumes that the probability of resource uptake or competing with a particular neighbor declines with distance from the stem but that considerable uptake at great distances from the stem is still possible. It also allows for plasticity in root distributions as might occur in spatially heterogeneous soils. Finally, we suggest how better information on the shapes and sizes of belowground zones of inuence will help develop a more predictive framework for understanding plant competition.
Key words: soil resources. belowground competition; canopy volume; lateral root spread; neighborhood models;

Special Feature

INTRODUCTION Because plants are sessile, most of their environmental interactions take place within a restricted space. The area over which a plant takes up resources, produces chemical exudates, casts shade, sheds litter, or otherwise alters its environment is considered its zone of inuence (Uranov 1965, Bonan 1993, Stohlgren 1993). Characterizing the zone of inuence is important both because its size and shape determine the total pool of resources available to an individual and because overlap in zones of inuence is necessary for facilitation or competition to occur (Czaran and Bartha 1992). The zone of inuence concept has been applied previously to belowground processes (Uranov 1965, Yastrebov 1996), and spatially explicit models based on this concept have examined competition for belowground resources (Mou et al. 1993, Biondini and Grygiel 1994, Huston and DeAngelis 1994, Biondini 2001). This approach should prove useful for predicting the outcome of competition as a function of the soil
Manuscript received 10 May 2002; revised 6 September 2002; accepted 11 September 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256.

environment, but it has been developed with little understanding of the actual spatial distribution of root systems, their activity under different abiotic conditions, or their architectural plasticity in the presence of neighboring plants. To predict the outcome of plant interactions as a function of local soil conditions, neighborhood models have a long history aimed at understanding the strength and consequences of plant competition as a function of the spatial arrangement of aboveground plant parts. In some early models, the neighborhood simply consisted of the number of plant stems within a circular area of arbitrary radius, while later models incorporated information about neighbor sizes, proximity or angular dispersion (Ek and Monserud 1974, Mack and Harper 1977, Weiner 1982, 1984, Silander and Pacala 1985, Lindquist et al. 1994) or searched for the aboveground neighborhood size that best explains plant size variation (Pacala and Silander 1985). Distance-based predictors used in these models typically assume that plants are more effective at taking up resources closer to their stems, and the strength of neighbor interactions is often modeled to decline according to a predetermined function of distance (e.g., linear, hyperbolic;

2313

2314

BRENDA B. CASPER ET AL.

Ecology, Vol. 84, No. 9

Weiner 1984, Pacala and Silander 1987, Yastrebov 1996, Zhang and Hamill 1997). These modeling approaches reect the fact that it is much easier to dene a zone of inuence aboveground than below. The size and spatial distribution of shoots are easily measured, and sources of temporal variation in aboveground zones of inuences, such as changes in sun angle, are predictable. In contrast, belowground zones of inuence, delimited by the spatial distribution of roots and their associated mycorrhizae, are likely to differ in both size and shape from their aboveground counterparts and are much more difcult to measure. Excavating roots carefully enough to reveal their connection to a particular plant is logistically challenging (Caldwell et al. 1996, Jackson et al. 1999). Excavations tend to miss many ne roots (Smit et al. 2000) that are most important for resource uptake (Passioura 1988, Robinson 2001). Excavations also may vary in efciency as a function of root density (Cahill 2002), rarely reveal the spatial distribution of fungal hyphae, and do not provide information on which roots are currently active. Such data are better supplied through the application of tracers, as described below. Here, we call for more comprehensive measures of root distributions and root activity among competing plants and present data illustrating how belowground zones of inuence may be affected by environmental conditions such as climate, soil texture, and the spatial distribution of nutrients. First, we enlarge and analyze a global data set of maximum lateral root spread and canopy size across climate and soil types to establish potential limits on the size of the belowground neighborhood (Schenk and Jackson 2002). Second, we describe our own work using nutrient analogs as tracers to examine functional root overlap and root system responses to spatial nutrient heterogeneity within populations of the weedy annual Abutilon theophrasti. Finally, we propose a new conceptual model for dening belowground zones of inuence, based on understanding root distributions and the mechanisms by which plants interact with their soil environment, and use data from the tracer study to parameterize the model. The model does not assume that the uptake of belowground resources declines as a xed, simple function of distance to plant stem as do traditional neighborhood models, and it incorporates plasticity in root distributions in response to the abiotic environment and other plants. METHODS

Special Feature

and maximum lateral root spread as affected by climate and physical soil characteristics. For this analysis, the data set of Schenk and Jackson (2002) was expanded (n 782, usually one observation per species) to include observations from humid environments as well as water-limited systems. The data set includes measurements of lateral root spreads (Lmax,i), canopy heights (Hi), and canopy widths (Wi) for individual plants i. The lateral root spread of a plant is dened as the maximum linear distance (one-sided) reached by its roots, as measured from the stem base. Canopy volumes were estimated assuming an ellipsoid Hi(Wi)2/6. Climatic conditions were charshape: Vi acterized by mean annual precipitation (MAP) and an aridity index (MAP/PET), with PET being the mean potential evapotranspiration as calculated by the Penman-Monteith method, taken from the global 0.5 gridded data set of Choudhury (1997). The data were grouped into two aridity classes, arid (MAP/PET 0.2, as dened by United Nations Environment Program [UNEP, 1992]) and semiarid to humid (MAP/PET 0.2). Data were also grouped into two soil texture classes, coarse (gravel to loamy sand) or ne (sandy loam or ner), where such information was available 688). (n Allometric relationships between aboveground plant sizes and lateral root spreads were examined by reduced major axis (RMA) regression (Niklas 1994, Sokal and Rohlf 1995) of Lmax,i against canopy volumes (Vi). The allometric equation used was log(Lmax,i)

b(log[Vi])

(1)

where a is the allometric constant and b is the allometric scaling factor. RMA regressions were calculated using the program PAST, version 0.65 (yvind Hammer, Paleontological Museum, University of Oslo, Norway). Allometric constants and scaling exponents were compared between aridity classes using ANCOVA. Effects of climate and soil texture on allometric relationships were analyzed in generalized linear models with Lmax,i (log transformed) as the dependent variable, and with log-transformed Vi, aridity class, soil texture class, and all their possible interactions as independent variables using SYSTAT version 9.0 (SPSS, Chicago, Illinois, USA).

Measuring the distribution of root function using tracers

Maximum lateral root spread is only part of the size or shape of a plants zone of inuence, as additional information is needed on the placement and activity of roots in different directions. Because physically uncovering roots and determining individual plant ownership where they intermingle is difcult, we used uptake of stable nutrient analogs to (1) estimate a plants belowground zone of inuence in the presence of competing plants, and (2) identify changes in the belowground zone of inuence with changes in the spatial

Analyzing maximum lateral root spreads globally

Maximum lateral root spread determines the potential horizontal extent of a belowground zone of inuence. Previous analyses of root distributions taken from root excavations reported in the literature have shown maximum lateral root spread to vary directly with canopy size (Schenk and Jackson 2002). Our analysis examined the allometric relationship between canopy size

September 2003

UNDERGROUND PROCESSES

2315

distribution of nutrients. Isolated plants often proliferate roots in nutrient patches (Drew and Saker 1975, Fitter 1994, Casper and Jackson 1997), and we wanted to know whether similar root system plasticity occurs among potentially competing individuals. The information we acquired on root distributions enabled us to parameterize our model by estimating the probability that a plants roots intercept a point at a given distance from the stem. Tracers are ideally suited to measure root function in different locations and to identify overlap in zones of inuence among co-occurring plants. Various types of tracers have been used extensively to examine vertical and horizontal root extensions (Hall et al. 1953, Dansgaard 1964, Fox and Lipps 1964, Fitter 1986, Mamolos et al. 1995) but have been used less to illuminate the belowground structure of populations or communities. Stable nutrient analogs, as applied in our study of A. theophrasti, can be especially useful because they are present in much lower concentrations in plant tissues and in the soil than many stable isotopic tracers (such as 15N), and therefore have lower detection limits and higher resolution. Different tracers can be used in the same population or community, allowing lateral root spread to be measured simultaneously in multiple directions, and thus determine shape parameters of the belowground zone of inuence. Populations of A. theophrasti were established in an outdoor garden by transplanting one-week-old seedlings into regular grids of nine rows of nine plants spaced 6 cm apart. Plants were omitted in locations 2 and 6 in rows 3 and 6, and after four weeks, one of four tracers, consisting of 0.2 M chloride salts of Cs, Li, Rb, and Sr, was randomly assigned to each of these four locations. Cs and Sr are analogs for K, and Li and Rb are analogs for Na and Ca, respectively. For each location, a syringe was used to inject 2 mL of tracer at 2-cm incremental depths (a total of 10 mL) to create a nearly continuous vertical column. In our experiment, plants took up only Sr and Rb. Spatially paired populations were assigned either of two nutrient treatments. This was done by preparing the soil before planting to a depth of 15 cm and adding 33.6 g of slowrelease fertilizer either evenly throughout the 60 60 cm plotthe homogeneous nutrient treatmentor divided among four 6 cm wide, 15 cm deep cylindrical patches centered on the points of tracer injectionthe heterogeneous nutrient treatment. Two weeks after tracer injection, all plants except those around the perimeter of each population were harvested, dried, and weighed. The portion of each plant produced since the time of tracer application, as estimated based on growth in a marked subset of the plants, was analyzed for tracer concentrations. Within a set of paired plots, uptake of a particular tracer was indicated by levels higher (onetailed t test; P 0.01) than those found in the 10 plants that were most distant from a point where that tracer had been injected ( 38.4 cm). Tracers moved hori-

zontally in the soil less than 3 cm. Differences between nutrient treatments in the number or locations of plants taking up tracer indicate root system responses to nutrient patches. Results are based on four replicate populations of each nutrient treatment (see Casper et al. 2000 for methodological details). RESULTS
AND

DISCUSSION

Analyzing maximum lateral root spreads globally

Maximum lateral root spreads Lmax,i were strongly 0.001) related to aboveground canopy sizes Vi (P and yielded the following regression: log(Lmax,i) 0.458 0.451(log[Vi]) (r2 0.528). The slopes, i.e., the allometric scaling factors, were different ( P 0.0001) between the two aridity classes. Plants from arid environments tended to have larger lateral root spreads relative to canopy size than plants from semiarid and humid environments. This difference is most pronounced in larger, woody plants (Fig. 1). Soil characteristics also had an effect on the allometric scaling factor, with plants growing on soils of coarse texture tending to have larger lateral root spreads for a given canopy size than those growing on soils of ner textures. This is indicated by the signicant interaction ( P 0.05) between canopy size and soil texture in the generalized linear model (Table 1). These results suggest the prediction that, all else being equal, belowground zones of inuence will be bigger in coarse than in ne textured soils. Longer maximum lateral root spread found for arid climates may be related to the low plant density typically found there. Wide-spreading roots enable plants to take up water from bare interspaces between individual plants (Walter 1963). The observation that lateral roots extend farther from the stem base in coarse than in ner soils lends support to the suggestion made by Sperry et al. (1998) and Jackson et al. (2000) that plants of a given canopy size need larger root systems in coarse textured soils because such soils offer larger resistance to water ow and have smaller water-holding capacities. Coarse soils may also offer less impedance to root growth, enabling plants to explore a larger soil volume.

Special Feature

Measuring the distribution of root function using tracers

Examining results from the experiment with A. theophrasti, several features of belowground population structure were described by the capacity of plants to acquire tracers from discrete injection points. Roots of many plants extended well beyond immediate neighbors, with a maximum spread of 32 cm (Fig. 2). Root systems also responded to nutrient patches; twice as many plants took up tracer when nutrient patches were centered on the points of tracer injection (Fig. 2). Although the introduction of nutrient patches increased the probability that a given plant intercepted the soil

2316

BRENDA B. CASPER ET AL.

Ecology, Vol. 84, No. 9

FIG. 1. Allometric relationships between aboveground canopy volume and maximum lateral root spread. The regression lines and equations are based on reduced major axis regressions performed on log-transformed data, using the general equation log(Li) a b(log[Vi]), with Li expressed in meters and Vi in cubic meters.

Special Feature

space marked by tracer and consequently increased the number of overlapping root systems in that space, the overall spatial distribution of plants taking up tracer was similar in the two treatments (Fig. 2). In both the homogeneous and heterogeneous nutrient treatments, roughly 55% of all plants taking up tracer were located within 10 cm of the tracer. The fact that plants equidistant from Sr and Rb sometimes took up one but not the other suggests that the root systems are spatially asymmetric (Casper et al. 2000). Despite the concentration of roots in nutrient patches, plants growing in heterogeneous soil did not apparently acquire more nutrients than plants in homogeneous soil. Soil treatment affected neither mean plant biomass (F1,3 0.859, P 0.422) nor the sizes of the largest plants as measured by the biomass of each of the four largest plants in each population (F1,3 5.568, P 0.099). However, the sizes of the largest plants varied considerably among populations (7.6813.57 g for the single largest plant), and both block (repre-

senting the pairs of homogeneous and heterogeneous soil treatment plots; F3,24 3.664, P 0.026) and the interaction between soil treatment and block ( F3,24 5.21, P 0.007) were signicant in ANOVA. Results are consistent with prior studies with A. theophrasti, where the spatial distribution of nutrients did not affect population level productivity but had subtle effects on plant size hierarchies within the populations (Casper and Cahill 1996, 1998).

A conceptual model of belowground zones of inuence

Taken together, the two data sets discussed above indicate that a plants zone of inuence, and the number of neighbors with which it interacts belowground, depend on local and global environmental conditions. Here, we propose a conceptual model that accounts for such environmental inuences and their effects on the probability that given locations in the soil are within the zone of inuence of one or more roots. In this

TABLE 1. Statistical parameters of a generalized linear model (GLM) of lateral root spread (log transformed) as a function of aboveground canopy size (log transformed), aridity class (arid vs. semiarid to humid), and soil texture class (coarse vs. medium to ne). Source Canopy size Aridity Soil texture Canopy size Canopy size soil Aridity Canopy size Error
SS

df 1 1 1 1 1 1 1 680

F
353.465 58.736 0.447 6.851 3.992 0.090 0.131

P
0.0001 0.0001 0.504 0.009 0.046 0.765 0.718

aridity soil texture texture soil texture

aridity

48.728 8.097 0.062 0.944 0.550 0.012 0.018 93.743

Note: The r2-value of the GLM was 0.582.

September 2003

UNDERGROUND PROCESSES

2317

FIG. 2. Proportion of all Abutilon theophrasti plants that took up Sr tracer as a function of distance from the injection point. Filled circles represent plants in populations where a patch of fertilizer was centered on the points of tracer injection. Open circles represent plants in populations where the same quantity of fertilizer was evenly distributed. Curves were t using Eq. 2. Parameters from that equation for the best ts are listed in the legend (homogeneous treatment, adjusted r2 0.818, df 6, 6, F 36.26; heterogeneous treatment, adjusted r2 0.982, df 6, 6, F 403.49). Parameter c was signicantly different between the two ts (P 0.05), while parameter d was not.

model, roots are treated as discrete entities. In contrast to previous models (e.g., Weiner 1984, Pacala and Silander 1987, Zhang and Hamill 1997), this model does not assume that the absolute amount of resource uptake within an individual root system declines as a simple function of distance from the stem, but assumes instead that the probability of resource uptake from a particular soil location and the probability of competing with a particular neighbor is a function of the distance from the stem. This model accounts for the fact that plants are capable of taking up large quantities of resources at great distances from the stem and allows for patchiness and stochasticity in root placement. The following statistical approach provides a theoretical framework for estimating the belowground zone of inuence: For any given location l in the soil one determines the probability Pl that it is within the zone of inuence of a plant. For an individual plant (i), the density of roots typically declines with distance from the stem and will reach zero beyond the maximum lateral root spread Lmax,i. Therefore, the probability for any location in the soil to be within the zone of inuence of a root from that plant will also be a function of distance from the stem. Such probability functions can be determined horizontally and vertically; we illustrate the approach for the horizontal dimension and project the zones of inuence around roots within the upper layers of the soil onto a horizontal plane. Zoneof-inuence probability functions may have different shapes, but exponential declines with distance from the stem are likely. In consequence, the following model may be used to calculate the probability Pl,i for any location l to be part of the belowground zone of inuence of a plant i. Similar calculations could in principle be applied to hyphae of mycorrhizal fungi. Pl,i is a function of the distance Li from the stem base of the plant within the maximal lateral extent Lmax,i of the root system:

Pl,i P

max[1, c (exp[ Li / f ])] 0 for Li

Lmax,i

(2)

where c is a constant characterizing overall root density within the zone of inuence, and f is a shape parameter characterizing the relationship between root length density and distance to the stem (see Fig. 3). Note that the probability distribution for the placement of roots is symmetric around the stem base, but this does not mean that actual root systems will typically be symmetric. Other variables may further inuence Pl,i, including soil characteristics at that location (e.g., structure, nutrient concentration, water content, presence of toxins, presence of roots from other plants, or presence of other soil organisms). Such factors may be incorporated into the model by multiplication with additional variables, as demonstrated here for a general parameter characterizing nutrient availability Nl:

Special Feature

Pl,i P

max[1, c (exp[ Li / f ]Nl )] 0 for Li

Lmax,i.

(3)

This additional parameter accounts for the fact that root placement is often nonrandom and can respond to resource gradients, thereby increasing the probability for nutrient-rich soil locations to become part of a plants belowground zone of inuence. Furthermore, the model can be extended to account for spatial correlation within root systems. Living roots are connected, which means that locations adjacent to those known to be part of a belowground zone of inuence will have a higher probability of also being inside this zone. In addition to these factors, Pl,i may also be decreased (Schenk et al. 1999) or increased (Gersani et al. 1998, Robinson et al. 1999) if roots from a competitor j are already present in the location, which could be ac-

2318

BRENDA B. CASPER ET AL.

Ecology, Vol. 84, No. 9

FIG. 3. Models of the probability for a point within the upper 10 cm of soil and at a given distance from the stem to be within the resource uptake zone of a root (curves calculated by Eq. 2), depicting different shapes of belowground zones of inuence. In (a), the shape parameter f in Eq. 2 is held constant, while c varies; in (b), c is held constant, while f varies.

counted for by an additional parameter in Eqs. 2 or 3. Such competitive effects could be studied by testing the hypothesis that the probability of any soil location to contain roots of two plants, i and j, is equal to Pl (i and j) Pi Pj, which is true if Pi Pj are independent of each other. We applied this conceptual model to the Abutilon system described above, by using the tracer experiments to parameterize Eq. 2 (Fig. 2). In this experiment, soil locations were represented by tracer injection points, which were surrounded by Abutilon plants spaced at various distances from these points. All plants for the homogeneous and heterogeneous nutrient treatments, respectively, were divided into groups by their distance to the injection points, with a group containing all plants within the same 6-cm distance increment (0 6 cm, 612 cm, 1218 cm, . . . , 3642 cm). We determined the proportion of plants within each group that took up tracer. These proportions were plotted against the distances between tracer injection points

and plant stems, resulting in a graph of Pl,i as a function of Li, as in Eq. 2. We then tted the equation to these data. Note that these data characterize the average belowground zone of inuence for the tested Abutilon populations, because they were gathered by measuring many plants spaced at various distances from a given injection point. In the homogeneous nutrient treatment, there was a 0.11 for any given soil lototal probability of Pl,i cation within the maximum root spread of an individual Abutilon plant (32 cm) to be within the belowground zone of inuence of that plant, or, in other words, the zone of inuence of an average Abutilon plant covered 11% of the area within its maximum root spread. Pl,i was 0.3 for distances 06 cm and declined to P 0 at distances 24 cm from the stem (Fig. 2). Tracer within the small area close to the stem was thus most likely to be taken up, but the distribution of tracer uptake within the population revealed that the majority of tracer ( 75%) was taken up from the much larger area between 12 and 30 cm from the stems of individual plants. Nutrient-enriched locations in the heterogeneous soil treatment were about twice as likely to supply tracer to the plants than the nonenriched locations in the homogeneous soil treatment (Fig. 2). This information could, in principle, be used to estimate the magnitude of parameter Nl in Eq. 3, but overall soil nutrient availability in these experiments was not quantied. Competitive effects on root placements could not be quantied, because all plants in these regularly spaced arrays were equally subjected to intraspecic competition. In conclusion, the probabilistic model outlined can be used to formulate and test explicit hypotheses about shapes and sizes of belowground zones of inuence and the variables that affect them, including interactions with resources and other plants. Once the factors that determine the shapes and sizes of belowground zones of inuence are quantied, they could be incorporated into predictive models of plant performance.

Special Feature

Belowground zones of inuence and competition

Maximum lateral root spread denes the boundary of the zone of inuence or the soil space from which a plant gathers resources, but the distribution of roots within this space greatly affects the degree of overlap in zones of inuence among competitors and the extent of competition and resource partitioning. Overlapping zones of inuence determine the set of plants that make up the competitive neighborhood. The amount of overlap in the lateral distribution of roots in the upper soil layers, where nutrient concentrations are usually highest (Sposito 1989, Jobbagy and Jackson 2001), may be particularly important for root competition. Co-occurring plant species can sometimes reduce competition for water by partitioning soil space vertically (Fernandez and Caldwell 1975, Sala et al. 1989, Ehleringer et

September 2003

UNDERGROUND PROCESSES

2319

al. 1997), but this option may be less available in competition for nutrients (but see McKane et al. 2002). Temporal partitioning of root activity may also occur even if root systems physically overlap (Gebauer and Ehleringer 2000, McKane et al. 2002). For these reasons, it is particularly important to measure root functionnot just the physical presence of roots. For models of belowground competition it is clearly insufcient to assume that belowground zones of inuence are of a xed, circular shape. For example, in a sparse, natural, herbaceous community in Florida shrubland, plants took up tracer over distances of up to 97 cm, with a mean of 51 cm, but in many cases more distant plants took up tracer while closer plants did not (Hawkes and Casper 2002). Excavations of all but the nest roots in woody species often reveal root systems shaped as irregular polygons that overlap less with neighbors than would circular systems with the same areas (Brisson and Reynolds 1994, Mou et al. 1995). Such shapes may result from responses of root growth to soil resource availability or from direct root avoidance mechanisms, which to date have been found mostly in water-limited environments (Mahall 1998). There is ample evidence that belowground competition can affect the shape of root systems (Schenk et al. 1999), and, conversely, that asymmetric shapes of root systems inuence the magnitude of competition (Brisson and Reynolds 1997).

Future directions
More information on the plasticity of belowground zones of inuence is needed before we can apply our probabilistic model to other systems and compare its predictive power to more traditional models that assume xed zones of inuence. Better data on how root systems respond to different resource conditions and under different neighbor densities are necessary to determine the probability of root placement at different distances from the stem. For example, at the moment we are unable to say how belowground zones of inuence and neighborhood sizes should change with even simple changes in population structure, such as plant density. Plant size decreases with increases in density, but how do root features and the zone of inuence change with plant size? Within a population, do smaller, subordinate plants produce shorter lateral roots as suggested by the interspecic comparisons in our global data set? Or do they produce fewer laterals without altering their length? Much recent work on the design of whole root systems (Einsmann et al. 1999, Johnson and Biondini 2001, Wijesinghe et al. 2001) has been motivated by Campbell et al.s (1991) study showing a tradeoff in the scale and precision of root foraging. They demonstrated that species with larger root systems, as measured by biomass, were less able to proliferate roots in nutrient patches. Whenever the subject of interest is really the spatial area over which a plant forages, we

suggest that the zone of inuence is a more appropriate measure than biomass. This is because the storage function of roots may contribute greatly to biomass, and storage biomass may vary independently of ne root biomass (Casper et al. 1998). Moreover, depending on the scale of nutrient heterogeneity, the size and shape of the zone of inuence may be as important as precision in enabling plants to access nutrient patches. Many studies examining plant responses to nutrient patches have worked with isolated, potted plants, and we need to know whether their ndings extrapolate to competing individuals (Hutchings and de Kroon 1994, Robinson et al. 1999, Fransen et al. 2001). At least in populations of A. theophrasti, more individuals actively taking up nutrients from the same vertical column of soil where nutrients are locally elevated, but productivity does not increase. The results are consistent with Gersani et al.s (1998, 2001) idea that higher nutrient levels are balanced by the production of more roots until nutrient availability is roughly the same in the different microhabitats. If this commonly occurs, nutrient heterogeneity could alter the shape of the zone of inuence but little affect the outcome of competition among plants with the ability to proliferate roots in patches. In summary, we think that characterizing the size, shape, and plasticity of the belowground zone of inuence, using methods that directly assess root activity, such as tracers, holds much more promise than measures of root biomass, root/shoot ratio, or even total root surface area (Gleeson and Tilman 1990, Aerts et al. 1991, Cahill and Casper 2000, Johnson and Biondini 2001) in helping us understand the strength of belowground interactions in plant communities. For plants with the same root/shoot ratio, for example, many short lateral roots may make plants stronger belowground competitors with nearest neighbors, while more sparse systems with longer lateral root spread may enable plants to search over longer distances and locate pockets where nutrients are elevated or competing roots are absent. Characterizing the belowground zone of inuence should be done in a eld setting, both with and without neighbors, where lateral root growth is not constrained by the size of a pot. Armed with such information, our approach should have numerous applications, enabling us to make predictions about the nature of plant interactions under different abiotic conditions and with changes in the number and identities of neighboring plants.
ACKNOWLEDGMENTS The research and ideas presented here were supported by NSF (DEB-9708165 and DEB-9733333) and the Andrew W. Mellon Foundation. J. Cahill, M. Hutchings, and S. Wilson provided helpful comments on the manuscript. LITERATURE CITED Aerts, R., R. G. A. Boot, and P. J. M. van der Aart. 1991. The relation between above- and belowground biomass al-

Special Feature

2320

BRENDA B. CASPER ET AL.

Ecology, Vol. 84, No. 9

location patterns and competitive ability. Oecologia 87: 551559. Biondini, M. 2001. A three-dimensional spatial model for plant competition in a heterogeneous soil environment. Ecological Modelling 142:189225. Biondini, M., and C. E. Grygiel. 1994. Landscape distribution of organisms and the scaling of soil resources. American Naturalist 143:10261054. Bonan, G. B. 1993. Analysis of neighborhood competition among annual plants: implications of a plant growth model. Ecological Modelling 65:123136. Brisson, J., and J. F. Reynolds. 1994. The effect of neighbors on root distribution in a creotebush (Larria tridentata) population. Ecology 75:16931702. Brisson, J., and J. F. Reynolds. 1997. Effects of compensatory growth on population processes: a simulation study. Ecology 78:23782384. Cahill, J. F., Jr. 2002. What evidence is necessary in studies which separate root and shoot competition along productivity gradients? Journal of Ecology 90:201205. Cahill, J. F., Jr., and B. B. Casper. 2000. Investigating the relationship between neighbor root biomass and belowground competition: eld evidence for symmetric competition belowground. Oikos 90:311320. Caldwell, M. M., J. H. Manwaring, and S. L. Durham. 1996. Species interactions at the level of ne roots in the eld: inuence of soil nutrient heterogeneity and plant size. Oecologia 106:440447. Campbell, B. D., J. P. Grime, and J. M. L. Mackey. 1991. A trade-off between scale and precision in resource foraging. Oecologia 87:532538. Casper, B. B., and J. F. Cahill, Jr. 1996. Limited effects of soil nutrient heterogeneity on populations of Abutilon theophrasti (Malvaceae). American Journal of Botany 83:333 341. Casper, B. B., and J. F. Cahill, Jr. 1998. Population level responses to nutrient heterogeneity and density by Abutilon theophrasti (Malvaceae). American Journal of Botany 85: 16801687. Casper, B. B., J. F. Cahill, Jr., and L. A. Hyatt. 1998. Aboveground competition does not alter biomass allocated to roots in Abutilon theophrasti. New Phytologist 140:231 238. Casper, B. B., J. F. Cahill, Jr., and R. B. Jackson. 2000. Plant competition in heterogeneous habitats. Pages 111130 in M. J. Hutchings, E. A. John, and A. J. A. Stewart, editors. The ecological consequences of environmental heterogeneity. Blackwell Scientic, Oxford, UK. Casper, B. B., and R. B. Jackson. 1997. Plant competition underground. Annual Review Ecology and Systematics 28: 545570. Choudhury, B. J. 1997. Global pattern of potential evaporation calculated from the Penman-Monteith equation using satellite and assimilated data. Remote Sensing of Environment 61:6481. Czaran, T., and S. Bartha. 1992. Spatiotemporal dynamic models of plant populations and communities. Trends in Ecology and Evolution 7:3842. Dansgaard, W. 1964. Stable isotopes in precipitation. Tellus 16:436468. Drew, M. C., and L. R. Saker. 1975. Nutrient supply and the growth of the seminal root system in barley. II. Localized, compensatory increases in lateral root growth and rates of nitrate uptake when nitrate supply is restricted to only part of the root system. Journal of Experimental Botany 26:79 90. Ehleringer, J. R., R. D. Evans, and D. Williams. 1997. Assessing sensitivity to change in desert ecosystemsa stable isotope approach. Pages 223237 in H. Grifths, editor. Stable isotopes. BIOS Scientic Publishers, Oxford, UK.

Einsmann, J. C., R. H. Jones, P. Mou, and R. J. Mitchell. 1999. Nutrient foraging traits in 10 co-occurring plant species of contrasting life forms. Journal of Ecology 87:609 619. Ek, A. R., and R. A. Monserud. 1974. FOREST: a computer model for simulating the growth and reproduction of mixed species forest stands. Pages 114 in University of Agricultural and Life Sciences Research Report A2635. University of Wisconsin, Madison, Wisconsin, USA. Fernandez, O. A., and M. M. Caldwell. 1975. Phenology and dynamics of root growth of three cool semi-desert shrubs under eld conditions. Journal of Ecology 63:703714. Fitter, A. H. 1986. Spatial and temporal patterns of root activity in a species-rich alluvial grassland. Oecologia 69: 594599. Fitter, A. H. 1994. Architecture and biomass allocation as components of the plastic response of root systems to soil heterogeneity. Pages 305323 in M. M. Caldwell and R. W. Pearcy, editors. Exploitation of environmental heterogeneity by plants: ecophysiological processes above- and belowground. Academic Press, San Diego, California, USA. Fox, R. L., and R. C. Lipps. 1964. A comparison of stable strontium and 32P and tracers for estimating alfalfa root activity. Plant and Soil 20:337350. Fransen, B., H. de Kroon, and F. Berendse. 2001. Soil nutrient heterogeneity alters competition between two perennial grass species. Ecology 82:25342546. Gebauer, R. L., and J. R. Ehleringer. 2000. Water and nitrogen uptake patterns following moisture pulses in a cold desert community. Ecology 81:14151424. Gersani, M., Z. Abramsky, and O. Falik. 1998. Density-dependent habitat selection in plants. Evolutionary Ecology 12:223234. Gersani, M., J. S. Brown, E. E. OBrien, G. M. Maina, and Z. Abramsky. 2001. Tragedy of the commons as a result of root competition. Journal of Ecology 89:660669. Gleeson, S. K., and D. Tilman. 1990. Allocation and the transient dynamics of succession on poor soils. Ecology 7: 11441155. Hall, N. S., W. F. Chandler, C. H. M. van Bavel, P. H. Reid, and J. H. Anderson. 1953. A tracer technique to measure growth and activity of plant root systems. Technical Bulletin. No. 101. North Carolina Agricultural Experimental Station, Raleigh, North Carolina, USA. Hawkes, C. V., and B. B. Casper. 2002. Lateral root function and root overlap among mycorrhizal and nonmycorrhizal herbs in a Florida shrubland, measured using rubidium as a nutrient analog. American Journal of Botany 89:1289 1294. Huston, M. A., and D. L. DeAngelis. 1994. Competition and coexistence: the effects of resource transport and supply rates. American Naturalist 144:954977. Hutchings, M. J., and H. de Kroon. 1994. Foraging in plantsthe role of morphological plasticity in resource acquisition. Advances in Ecological Research 25:159238. Jackson, R. B., L. A. Moore, W. H. Hoffmann, W. T. Pockman, and C. R. Linder. 1999. Ecosystem rooting depth determined with caves and DNA. Proceedings of the National Academy of Sciences (USA) 96:1138711392. Jackson, R. B., J. S. Sperry, and T. E. Dawson. 2000. Root water uptake and transport: using physiological processes in global predictions. Trends in Plant Science 5:482488. Jobbagy, E. G., and R. B. Jackson. 2001. The distribution of soil nutrients with depth: global patterns and the imprint of plants. Biogeochemistry 53:5177. Johnson, H. A., and M. E. Biondini. 2001. Root morphological plasticity and nitrogen uptake of 59 plant species from the Great Plains grassland, USA. Basic and Applied Ecology 2:127143.

Special Feature

September 2003

UNDERGROUND PROCESSES

2321

Lindquist, J., D. Rhode, and K. J. Puettmann. 1994. The inuence of plant population spatial arrangement on individual plant yield. Ecological Applications 4:518524. Mack, R. N., and J. L. Harper. 1977. Interference in dune annuals: spatial pattern and neighborhood effects. Journal of Ecology 65:345363. Mahall, B. E. 1998. Inter-root communications and the structure of desert plant communities. Pages 265280 in H. E. Flores, J. P. Lynch, and D. Eissenstat, editors. Radical biology: advances and perspectives on the function of plant roots. Current Topics in Plant Physiology 18. American Society of Plant Physiologists, Rockville, Maryland, USA. Mamolos, A. P., G. K. Elisseou, and D. S. Veresoglou. 1995. Depth of root activity of coexisting grassland species in relation to N and P additions, measured using nonradioactive tracers. Journal of Ecology 83:643652. McKane, R. B., L. C. Johnson, G. R. Shaver, K. J. Nadelhoffer, E. B. Rastetter, B. Fry, A. E. Giblin, K. Kielland, B. L. Kwiatkowski, J. A. Laundre, and G. Murray. 2002. Resource based niches provide a basis for plant species diversity and dominance in arctic tundra. Nature 415:68 71. Mou, P., R. H. Jones, R. J. Mitchell, and B. R. Zutter. 1995. Spatial distribution of roots in sweetgum and loblolly pine monocultures and relations with aboveground biomass and soil nutrients. Functional Ecology 9:689699. Mou, P., R. J. Mitchell, and R. H. Jones. 1993. Ecological eld theory model: a mechanistic approach to simulate plantplant interactions in southeastern forest ecosystems. Canadian Journal of Forest Research 23:21802193. Niklas, K. J. 1994. Plant allometry: the scaling of form and process. The University of Chicago Press, Chicago, Illinois, USA. Pacala, S. W., and J. A. Silander, Jr. 1985. Neighborhood predictors of plant performance. Oecologia 66:256263. Pacala, S. W., and J. A. Silander, Jr. 1987. Neighborhood interference among velvet leaf, Abutilon theophrasti, and pigweed, Amaranthus retroexus. Oikos 48:214224. Passioura, J. B. 1988. Water transport in and to roots. Annual Review of Plant Physiology and Plant Molecular Biology 39:245265. Robinson, D. 2001. Root proliferation, nitrate inow and their carbon costs during nitrogen capture by competing plants in patchy soil. Plant and Soil 232:4150. Robinson, D., A. Hodge, B. S. Grifths, and A. H. Fitter. 1999. Root proliferation in nitrogen-rich patches confers competitive advantage. Proceedings of the Royal Society of London 266:431435.

Sala, O. E., R. A. Golluscio, W. K. Lauenroth, and A. Soriano. 1989. Resource partitioning between shrubs and grasses in the Patagonian steppe. Oecologia 81:501505. Schenk, H. J., R. M. Callaway, and B. E. Mahall. 1999. Spatial root segregation: are plants territorial? Advances in Ecological Research 28:145180. Schenk, H. J., and R. B. Jackson. 2002. Rooting depths, lateral spreads, and belowground/aboveground allometries of plants in water-limited ecosystems. Journal of Ecology 90:480494. Silander, J. A., Jr., and S. W. Pacala. 1985. Neighborhood predictors of plant performance. Oecologia 66:256263. Smit, A. L., A. G. Bengough, C. Engels, M. van Noordwijk, S. Pellerin, and S. C. van de Geijn. 2000. Root methods: a handbook. Springer-Verlag, Berlin, Germany. Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principle and practice of statistics in biological research. Third edition. W. H. Freeman, New York, New York, USA. Sperry, J. S., F. R. Adler, G. S. Campbell, and J. P. Comstock. 1998. Limitation of plant water use by rhizosphere and xylem conductance: results from a model. Plant, Cell and Environment 21:347359. Sposito, G. 1989. The chemistry of soils. Oxford University Press, New York, New York, USA. Stohlgren, T. J. 1993. Intra-specic competition (crowding) of giant sequoias (Sequoiadendron giganteum). Forest Ecology and Management 59:127148. UNEP (United Nations Environment Program), editors. 1992. World atlas of desertication. Edward Arnold, London, UK. Uranov, A. A. 1965. Fitogennoe pole. Pages 251254 in E. M. Lavrenko, editor. Problemy Sovremennoj Botaniki. Volume 1. Nauka, Moscow, Russia. Walter, H. 1963. The water supply of desert plants. Pages 199205 in A. J. Rutter, and F. H. Whitehead, editors. The water relations of plants. The British Ecological Society Symposium No. 3. John Wiley and Sons, New York, New York, USA. Weiner, J. 1982. A neighborhood model of annual plant interference. Ecology 63:12371241. Weiner, J. 1984. Neighborhood interference amongst Pinus rigida individuals. Journal of Ecology 72:183195. Wijesinghe, D. K., E. A. John, S. Beurskens, and M. J. Hutchings. 2001. Root system size and precision in nutrient foraging: responses to spatial pattern of nutrient supply in six herbaceous species. Journal of Ecology 89:972983. Yastrebov, A. B. 1996. Strength of tree phytogeneic elds in lichen-green moss pine forests. Russian Journal of Ecology 27:17. Zhang, J., and A. S. Hamill. 1997. Seed weight, intraspecic competition, and plant performance in Abutilon theophrasti. Canadian Journal of Botany 75:16141620.

Special Feature

Ecology, 84(9), 2003, pp. 23222334 2003 by the Ecological Society of America

TOWARD UNDERSTANDING THE CONSEQUENCES OF SOIL HETEROGENEITY FOR PLANT POPULATIONS AND COMMUNITIES
MICHAEL J. HUTCHINGS, ELIZABETH A. JOHN,
AND

DUSHYANTHA K. WIJESINGHE

School of Biological Sciences, University of Sussex, Falmer, Brighton, Sussex, BN1 9QG, UK

Abstract. Several recent studies demonstrate that yield of individual plants, and their allocation of biomass between roots and shoots, can be profoundly affected by the pattern of supply of soil-based resources. Patchy provision of soil-based resources can affect the location of root biomass, as roots often proliferate in nutrient-rich patches. Root system size is important in determining whether plants access nutrient-rich patches, and the proportion of root systems located within such patches. This proportion will alter as growth proceeds. Species with small root systems have a limited ability to place roots in nutrientrich patches even when they are very close. Of four species with different root system sizes, the growth of the species with the smallest root system was signicantly limited by being located in nutrient-poor substrate even when nutrient-rich substrate was only 3.5 cm away, whereas three species with larger root systems were not disadvantaged. Both in the laboratory and in the eld, root density is higher in nutrient-rich patches, and such patches can contain roots of many plants. Recent work showing that plants can respond to nonself roots sharing the same nutrient supply suggests that competition will be more severe in nutritionally patchy substrates than in homogeneous environments with the same overall nutrient supply. Taken together, these facts lead to the prediction that inter- and intraspecic plant interactions will be inuenced by patterns of nutrient supply. We present evidence supporting this prediction, and indicating that population and community structure are also affected by patterns of nutrient supply. Signicant differences in population yield, plant size distribution, and mortality have been recorded between populations growing under patchy and uniform conditions. Plant communities grown from identical seed inocula, with the same overall nutrient supply, provided in different spatial and temporal patterns, differed by up to 44% in total biomass, up to 70% in root biomass, and differed in species composition. These signicant effects of heterogeneous resource supply on plants merit further detailed study. We present a framework of predictions of the impacts of different types of spatial heterogeneity in nutrient supply on the performance of single plants, and on plant interactions, plant populations, and plant communities.
Key words: belowground competition; nutrient heterogeneity; patch scale; plant competition; plant growth; root proliferation; soil heterogeneity.

Special Feature

INTRODUCTION A number of ecological investigations carried out in recent years have established that spatial heterogeneity in the availability of soil-based resources can strongly inuence the growth and patterns of biomass allocation of single plants. Heterogeneity in supply might improve resource uptake by purely passive means, but many species also exhibit strong physiological and/or morphological responses to soil-based heterogeneity (e.g., Jackson et al. 1990, Jackson and Caldwell 1996, Einsmann et al. 1999, and Robinson 1994 for a review). For plants to respond to heterogeneity, its scale must either match or be less than the spatial scale over which the root systems of individual plants spread. Importantly in the context of the present paper, heterogeneity can elicit both localized and plant-wide responses.
Manuscript received 13 May 2002; revised 5 October 2002; accepted 17 October 2002. Corresponding Editor: A. A. Agrawal. For reprints of this Special Feature, see footnote 1, p. 2256.

Consequently we focus on the morphological responses of plants when patches of different quality are smaller than, or of a similar size to, the root systems of individual plants. We begin our examination of this subject, and of the effects of heterogeneity on the interactions between plants, by reviewing some of the relevant behavior of single plants. A good understanding of individual plant responses to soil resource heterogeneity is essential if we are to make realistic predictions about the impact of heterogeneity on plant interactions. We then develop the argument that plant interactions will differ under heterogeneous and homogeneous conditions, using evidence from studies on competing plants, plant populations, and plant communities. We propose that our current limited knowledge of the effects of heterogeneity on plant behavior signicantly hinders our understanding of plant competition, and community composition. Our attention in this contribution focuses almost entirely upon spatial heterogeneity in the distribution of

2322

September 2003

UNDERGROUND PROCESSES

2323

soil-based resources. With a few exceptions, we discuss studies in which the same total quantity of resources is provided in homogeneous and heterogeneous patterns, so that the effects of supplying nutrients in patches are not confounded with the effect of supplying different amounts of nutrients. This approach to the study of heterogeneity has not always been adopted, some authors preferring experimental designs in which nutrient patches are provided by adding nutrients (e.g., Robinson 1994) to an otherwise homogeneous background supply. THE EFFECTS
OF

ON

SPATIAL HETEROGENEITY PLANTS

FIG. 1. Mean ( 1 SE) root/shoot ratio of clones of Glechoma hederacea grown in large- and small-patch environments with different degrees of contrast between patches. Patch scale had a signicant effect on root/shoot ratio (F1,48 14.75, P 0.0004), whereas neither contrast (F5,48 1.23, P 0.3078) nor the interaction between scale and contrast (F5,48 0.56, P 0.7305) signicantly affected root/shoot ratio. Data are from Wijesinghe and Hutchings (1999).

Proliferation responses within root systems

There is evidence that many plant species can preferentially select nutrient-rich soil patches for root placement when grown in heterogeneous conditions (reviewed in Robinson 1994). Although most of the early work on this topic examined the proliferation responses of crop species, we now know that noncrop species with a wide range of life histories and ecological preferences also proliferate roots in nutrient-rich patches. For example, Einsmann et al. (1999) measured precision of root placement in 10 species with a range of life forms. Precision was measured as the difference in ne root mass in fertilized and unfertilized patches in heterogeneous substrate and in equivalent substrate patches in homogeneous conditions. Values were converted to relative precision measurements by dividing by the total ne root mass produced by plants in each treatment. Accurate measures of precision require that root growth is not constrained by pot size, and care was taken to ensure that this was the case. Several species exhibited a high degree of precision. In a study on eight herbaceous species, Campbell et al. (1991) also observed selectivity in placement of new root biomass in nutrient-rich patches in a heterogeneous substrate, and reported a signicant negative relationship between root system size, measured as root mass, and the level of precision, measured as the proportion of new roots placed in nutrient-rich patches of habitat. Einsmann et al. (1999) failed to conrm this relationship using their measure of precision. Wijesinghe et al. (2001) observed a signicant negative relationship between root system mass and precision (calculated as root biomass in enriched patches/total root biomass) in six herbaceous species, but also showed that precision was not constant for any species. It was affected both by the location of the plant with respect to nutrientrich patches and by the spatial arrangement of those patches. Although at rst sight the benets of root proliferation in nutrient-rich patches seem obvious, several authors (Robinson 1996, Leyser and Fitter 1998) have pointed out that proliferation of roots in ephemeral nutrient-rich patches may not be an energetically efcient way of acquiring resources. This response to patchily

distributed resources may be more easily explained as a response to competition for a nite resource, an idea to which we return below. Nutrients and other resources are unevenly distributed in soils (e.g., Jackson and Caldwell 1996, Cain et al. 1999, Farley and Fitter 1999, Lister et al. 2000), and proliferation of roots in nutrient-rich patches has been recorded in several natural environments (e.g., Caldwell et al. 1991, Mou et al. 1995). For instance, Mordelet et al. (1996) report a strong correlation between palm tree root density and soil nutrient status, with root density up to 10 times higher in areas of nutrient-rich soil than in the surrounding nutrient-poor savannah. Thus, it is clear that both nutrients and roots are unevenly distributed in natural environments, and that roots tend to be concentrated in areas of nutrient enrichment. This will inevitably create hot spots for plant interactions.

Special Feature

Root and shoot biomass

An illustration of the impacts of soil-based heterogeneity on plant production and allocation to roots and shoots is provided by the stoloniferous clonal herb Glechoma hederacea (Lamiaceae). This plant has a remarkable ability to match the placement of its roots in different soil patches to the relative resource level in each patch. Growth of G. hederacea was examined in several treatments that provided the same total quantity of nutrients in environments consisting of high and low quality patches. There were different levels of contrast in nutrient supply between high and low quality patches and two patch scales (Wijesinghe and Hutchings 1999). In the treatment labeled 50/50 (see Fig. 1) there was no contrast; all of the substrate was a mixture of 50%

2324

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

sand and 50% compost (i.e., the environment was homogeneous). In treatment 0/100, the substrate in poor and good patches was respectively either 100% sand or 100% compost. Other treatments were made up of patches with intermediate contrast (e.g., patches in treatment 10/90 were either 90% sand, 10% compost or 10% sand, 90% compost). When contrast was higher, the total biomass of roots produced by the clone was greater, and the proportion of them located in nutrientrich patches was greater. The plants also responded to patch size; although the total area and volume of all of the high quality patches was the same in treatments with patches of different sizes, more root biomass was produced when patches were larger. These root growth and root placement responses resulted in signicant impacts on total clone growth. Growth was greater in treatments with large, highly contrasting patches than in homogeneous conditions with the same nutrient supply, but less when patches were small and highly contrasting. The size of patches provided in this experiment also altered the pattern of allocation of biomass to roots and shoots. Whole clone root/shoot (R/S) ratio was significantly affected by patch size (see Fig. 1, showing previously unpublished results from the experiment of Wijesinghe and Hutchings 1999). In an earlier experiment (Birch and Hutchings 1994), G. hederacea clones were provided with the same total nutrient supply either homogeneously or heterogeneously, with half of the nutrients conned to a patch occupying only 10% of the area available for growth. Whole-clone R/S ratio was nearly twice as high in the heterogeneous treatment as in the homogeneous treatment (Birch and Hutchings 1994), because of earlier and far greater root growth by the ramets in the nutrient-rich patch. We do not currently know enough to generalize across species about the impact of heterogeneity in soil resources on R/S ratio; many studies on the impact of heterogeneity on plant growth have not measured whole plant R/S ratio. It is clear, however, that heterogeneity can have substantial effects. Change in the allocation of resources between roots and shoots could have profound implications for competition, and ultimately for tness, between plants in situations where above- or belowground resources are limiting. The sensitivity of R/S ratio to patchiness in soil resources must also change our view of R/S ratio as a measure of plant response to limiting resources. The normally accepted view in physiology and ecophysiology is that when light is limiting, a higher proportion of biomass is allocated to photosynthetic machinery, reducing R/S ratio, whereas when soil-based resources are scarce, more biomass is allocated to roots, increasing R/S ratio (Aung 1974, Hunt and Nicholls 1986). This may not reect the situation in heterogeneous environments (Fitter and Hay 1987, Muller et al. 2000). More im portantly in the context of the current discussion, generalizations about the inuence of environmental con-

ditions on allocation to roots and shoots are far more difcult to make when plants experience heterogeneity. Furthermore, for clonal plants, several studies (e.g., Stuefer et al. 1994, 1996) show that plants grown under heterogeneous conditions exhibit important variations in R/S ratio at levels below that of the whole plant. When the scale of heterogeneity is smaller than that of the whole plant, clonal species not only locate a higher proportion of their roots in nutrient-rich patches, but they also develop higher R/S ratios locally in these patches (e.g., Birch and Hutchings 1994). This behavior (reviewed by Alpert and Stuefer [1997] and Hutchings and Wijesinghe [1997]) is the opposite of that reported at the whole plant level for many nonclonal and clonal species in homogeneous conditions. It indicates strong functional rather than allometric shifts in resource partitioning. Whether such functional shifts occur in nonclonal plants does not appear to have been investigated. In addition to affecting the spatial distribution of plant biomass and allocation of biomass to different plant parts, the studies reviewed above, and others, show that total plant biomass can be signicantly increased or even decreased, when only the pattern of resource delivery, rather than the quantity of resources, is changed. For example, Einsmann et al. (1999) demonstrated signicant effects of patch scale on the yield of several species, and Wijesinghe et al. (2001) showed that distance to nutrient patches can affect yield. Cahill and Casper (1999) report a signicant effect of heterogeneity on above ground yield of Phytolacca americana Both scale of heterogeneity and contrast in patch quality affected yield of Glechoma hederacea (Wijesinghe and Hutchings 1997, 1999). Different types of heterogeneity either increased or decreased growth compared with that achieved under homogeneous conditions with the same supply of resources. Lower growth under heterogeneous conditions could be caused by inability to match root distribution to pattern of resource supply, or by high or low resource concentrations limiting growth more than moderate concentrations. If the pattern of root distribution can be matched to soil resource supply, and biomass allocation to roots and root growth enhanced where soil-based resources are more abundant, more resources may be acquired and total growth may be greater. As indicated above, Birch and Hutchings (1994) also demonstrated that the onset of growth of the root initials on different ramets of Glechoma hederacea was extremely sensitive to local soil conditions, occurring far earlier in ramet development where nutrient supply was high, and later where it was low. This allows earlier and greater resource acquisition from resource-rich patches, and results in lower investment in resource-poor patches. Altogether, therefore, the plant acquires resources more efciently, and achieves greater growth in appropriately scaled heterogeneous conditions. It should also

Special Feature

September 2003

UNDERGROUND PROCESSES

2325

FIG. 2. Effects of spatial scale of heterogeneity in relation to the size of plant root systems, and contrast in patch quality, on the performance of individual plants. It is assumed that, in the boxes at the right of the gure, the overall level of resource supply is the same under both levels of contrast.

Special Feature

be recognized, however, that such benets of heterogeneity may be transient. For example, plants may grow faster in heterogeneous than homogeneous conditions because of more efcient resource acquisition. However, growth may ultimately be no greater under either condition when resource supply is exhausted and differential uptake from patches of different quality has rendered the environment more homogeneous. A recent study has even indicated that a species that begins growth at an advantage under heterogeneous conditions can ultimately be suppressed by a competitor with slower initial growth that continues to accumulate biomass over a longer period (Fransen and de Kroon 2001). The evidence presented so far demonstrates that both scale of patches and contrast in nutrient supply between patches can profoundly affect plant responses to heterogeneity (Fig. 2). We propose that changes in biomass and patterns of biomass allocation caused by heterogeneity in resource supply will inevitably affect interactions between plants, particularly as evidence suggests that responses to heterogeneity depend both on species characteristics and plant size, and on the form and scale of heterogeneity (Fig. 2, Box 1). Speciesspecic responses to heterogeneity reect evolutionary history, and imply that there must have been variable selection in the past for particular responses to specic kinds of heterogeneity (Oborny 1994, Alpert and Simms 2002). ACCESS
TO

NUTRIENT-RICH PATCHES

Plant roots initially tend to grow somewhat radially, although probably far from symmetrically under het-

erogeneous conditions (Casper et al. 2003), from a point source. Consequently, unless a plant germinates within a resource-rich patch, or is immediately next to one, none of its root system will initially be in resourcerich conditions. As growth proceeds, some roots may access resource-rich patches and the plant may begin to display precision in locating its roots in such patches. Precision therefore depends on plant size and on the plants initial distance from resource-rich patches. Plants that can access a nutrient-rich patch from an early stage of growth may initially show strong precision in locating their roots in the patch, but as they grow, and perhaps deplete the patch of nutrients, root distribution throughout the substrate is likely to become more even. Species of plants with widely spreading root systems, and roots with rapid lateral extension, are more likely than species with smaller or slower growing root systems to access laterally distant nutrient patches. That the initial location of a plant can have lasting consequences for its subsequent ability to access patchily distributed resources was shown by a previously unpublished experiment conducted on four herbaceous species with root systems of different masses ( Arrhenatherum elatius, Campanula rotundifolia, Papaver rhoeas, and Plantago lanceolata). Circular pots 15 cm in diameter were set up containing substrate arranged as two opposing quadrants of nutrient-rich soil separated by two opposing quadrants of nutrient-poor soil. There were no barriers between the quadrants, allowing roots to grow freely between them. A single seedling of one of the species was grown in each pot, planted in either a nutrient-rich or nutrient-poor quadrant, 3.75

2326

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

BOX 1. Known and predicted effects of small- and large-scale spatial heterogeneity in the availability of soil-based resources, on pairwise competition between plants, on plant populations, and on plant communities. It is assumed that (i) within each habitat considered the sizes of patches are relatively constant, (ii) at least one patchily distributed resource is limiting to growth, and (iii) overall resource supply is the same in the different habitat types at each of the three levels of complexity. Predicted effects are set in italics. Pairwise competitionSmall scale, high contrast: Roots of both plants cross patch boundaries. Higher quality patches selected for root placement by both plants, resulting in more intense competition than in equivalent homogeneous environments. Small patches rapidly dominated by fast-growing plants, denying resources to slower-growing neighbors. Competitive asymmetry greater than in homogeneous environments. Selection favors plants that respond rapidly to resource-rich patches. Location of each competitor with respect to patch quality is critical for success. These effects are increasingly important as the limiting resource is conned to smaller patches, and as patch contrast increases. PopulationsSmall patches, high contrast: Resource-rich patches can become occupied by roots of plants located in both rich and poor patches. This will produce intensied competition involving many plants. There will be mortality in both rich and poor patches. However, overall mortality may be less than in equivalent homogeneous environments, as the intensity of competition is not as spatially uniform and the probability of survival will be higher in resource-poor patches. Resource-poor patches will be little utilized by roots, but plants in these patches may suffer from competition for light from large plants growing nearby in resourcerich patches. CommunitiesSmall patches, high contrast: Fast-growing species with high resource demands access many resource-rich patches by growing across resource-poor patches, eliminating slow-growing species. Overall, species richness lower, and the community more uniform, than in heterogeneous habitat with large patches. Communities less diverse than in equivalent homogeneous habitats, because resource-rich patches favor fast-growing species, increasing the intensity of competition throughout the community. Colonization by new species less likely than in large scale or homogeneous habitats, as there are few refuges from intense competition for young seedlings. Pairwise competitionLarge scale, high contrast: The environment is perceived as uniformly good or bad, except when plants are located at patch edges. Under such conditions, plants will select the higher quality patch for root placement, resulting in intensied local root competition. Selection for rapid root proliferation in resource-rich patches will be less intense than at smaller scales of heterogeneity, because it is less likely that the root system of a single plant can completely dominate a resource-rich patch.

Special Feature

PopulationsLarge patches, high contrast: For plants with roots located wholly within a single patch, population processes will operate as they would in a homogeneous environment of the same quality. Rates of growth and mortality, and intensity of competition and biomass accumulation will differ dramatically between patches. However, plants at patch boundaries will mainly place their roots in the more resource-rich patch, resulting in more intense competition in resource-rich locations near patch boundaries than in equivalent parts of homogeneous environments.

CommunitiesLarge patches, high contrast: Little effect of heterogeneity on plant interactions except at patch boundaries where large species with fast root proliferation eliminate slow-growing species. Intensity of competition greater in resource-rich patches, and fast-growing species eliminate slow-growing species. Rich patches support lower plant density and diversity and a different sub-community than poor patches. Few fast-growing species survive in poor patches. In poor patches, plant density and diversity are greater, and slow-growing species dominate. Overall diversity higher than in equivalent homogeneous habitats. Colonization by new species more common in resource-poor patches.

September 2003

UNDERGROUND PROCESSES

2327

FIG. 3. The mean proportion of total root biomass distributed in the four quadrants within pots for each of the species Arrhenatherum elatius, Papaver rhoeas, Plantago lanceolata, and Campanula rotundifolia. The quadrant into which each plant was transplanted at the beginning of the experiment is designated as home. The home and opposite quadrants were of similar quality. In (a) the home quadrant was nutrient rich; in (b) the home quadrant was nutrient poor.

cm from the intersection of the four quadrants. The total quantity of nutrients in each pot was the same in each treatment. After 12 weeks of growth, aboveground biomass was dried and weighed, and roots were extracted separately from each quadrant, dried, and weighed. Analyses were carried out to determine whether the species displayed selectivity in placement of roots between the four quadrants (described as home, opposite, and anking quadrants). Nonparametric Friedman tests (with chi-square approximation with df 3) were carried out for each species, to compare the proportions of the total root biomass recovered from the four quadrants. The null hypothesis, that roots are evenly distributed, would be supported if the median root distributions were the same for all quadrants. Two (or more) quadrants having different median values would indicate uneven root distribution. Separate tests were carried out for plants starting growth in rich and poor home quadrants. Differences between species in the degree of selectivity of the four quadrants for root placement were analyzed by xed-model multivariate analysis of variance of the proportion of root biomass located in each quadrant. Data were angular transformed before analysis. All species tended to place roots unevenly between the four quadrants (Fig. 3, Table 1). A signicantly greater proportion of roots was recovered from the rich home quadrant, irrespective of its quality, than from other quadrants in all cases except for P. lanceolata in rich home quadrants. The overall pattern of root distribution between the quadrants did not differ signicantly between A. elatius, P. rhoeas, and P. lanceolata, but for all of these species the pattern differed significantly from that of C. rotundifolia (Bonferroni multiple means comparison test at P 0.05). Roots of C. rotundifolia were signicantly more conned to the home quadrant than were the roots of the other species. About 60% of C. rotundifolia roots were located in the home quadrant, regardless of its quality (Fig. 3), whereas only 3035% of the root biomass of the other three

Special Feature

TABLE 1. Analyses of selectivity of root placement patterns in four pot quadrants of either high or low nutrient content, for four species, when originally planted in either a nutrientrich or nutrient-poor quadrant. Plant in rich/poor quadrant rich poor rich poor rich poor rich poor

Species

Fr
26.43 19.48 5.88 17.88 36.85 27.40 49.77 55.83

P
0.0001 0.001 not signicant 0.001 0.0001 0.0001 0.0001 0.0001

Arrhenatherum elatius Plantago lanceolata Papaver rhoeas Campanula rotundifolia

Notes: Analysis was by the Friedman test. The values tabulated are the test statistic (Fr) and the probability of accepting the null hypothesis of equality of root distribution among the four 20 per species for plants in both rich and poor quadrants. quadrants of the pot. n

2328

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

FIG. 4. The relationship between total root biomass (natural log transformed) and the proportion of total root biomass distributed in nutrient-rich quadrants for species grown in isolation. Correlations between these variables were as follows: for plants that began growth in rich quadrants, Pearson product-moment correlation r 0.48, P 0.0001, n 80; for plants that began growth in poor quadrants, r 0.70, P 0.0001, n 80. Data were pooled across species for each correlation.

species was located in the home quadrants. The more extensive exploration of the substrate observed in the species with larger root systems would enable them to access nutrients from non-home quadrants to a considerable extent, whereas greater connement to its home quadrant would be more likely to cause a small species such as C. rotundifolia to deplete accessible resources quickly to levels that limit further growth. Species with larger root systems may avoid local resource depletion.

Strikingly, there were strong correlations between the biomass of individual plant root systems and the proportion of roots in nutrient-rich quadrants (Fig. 4). In rich home quadrants, those plants with the smallest root systems (mostly C. rotundifolia) tended to be strongly conned to their home quadrant, whereas those with larger root systems projected a high proportion of their roots into other quadrants including the nutrient-rich quadrant opposite. In poor home quadrants, plants with large root systems also projected a large proportion of their roots into other quadrants, whereas the roots of most C. rotundifolia plants were once again mostly conned to their home quadrant (Fig. 4). There was a large (45%) and signicant reduction in yield of C. rotundifolia caused by beginning life in poor rather than rich home quadrants (Table 2), but no such decrease for the other species. Thus the impact of soil heterogeneity on plant performance was crucially dependent both on whether a plant began growth in a rich or poor patch and on the size of the plant. It is clear that, particularly for species of small stature, starting life in a poor patch in a heterogeneous environment can signicantly limit performance, even when nutrient-rich conditions are close by, and, as in this experiment, even in the absence of competition. Adding competition to this handicap might be expected to result in greater suppression of the growth of such plants, or even in their death, because plants that acquire access to resource-rich patches faster than their neighbors are advantaged in competition (Box 1; Hodge et al. 1999). Few studies of the effects of heterogeneity on plants have directly examined the impact on plant performance of the distance to nutrient patches. Although the outcome of such experiments may be affected by the size of plant root systems and the amount of growth prior to harvest, Wijesinghe et al. (2001) found signicant effects of distance to nutrient patches on precision of root placement in two out of six species, and smaller (nonsignicant) effects on another two species. Some roots of some plants may gain access to nutrient patches a considerable distance away, having bypassed numerous intervening plants to reach these patches (Casper et al. 2000). Thus, for some species, distance

Special Feature

TABLE 2. Mean ( 1 SE, n 20) plant mass and results of t tests comparing mass of plants beginning life in nutrient-rich or nutrient-poor patches within 15 cm diameter pots containing the same overall supply of nutrients in each treatment. Mean ( 1 Species
SE )

plant mass (g) Nutrient-poor quadrants 4.18 0.15 1.07 2.45 0.32 0.04 0.09 0.26

Nutrient-rich quadrants 4.75 0.28 1.10 2.40 0.33 0.05 0.05 0.15

t
1.383 2.499 0.654 0.358

df 38 38 38 38

P (twotailed)
0.175 0.008 0.516 0.722

Arrhenatherum elatius Campanula rotundifolia Papaver rhoeas Plantago lanceolata

Notes: The t tests were performed on ln-transformed data. The means presented here are untransformed.

September 2003

UNDERGROUND PROCESSES

2329

from nutrient patches may limit growth less than expected. Casper et al. also found that many different plants may place roots in, and acquire resources from, the same patch of soil, and suggested that the number of plants interacting in this way might be greater in nutrient-rich patches in heterogeneous soils than in equivalent volumes of homogeneous soils, a prediction supported by a limited amount of data from natural environments (Caldwell et al. 1991, Mordelet et al. 1996). This leads to the prediction that heterogeneity would fundamentally change competitive interactions between plants, with greater intensity of competition being experienced within nutrient-rich patches (Box 1). We now address this prediction. MODE
AND I NTENSITY OF COMPETITION HETEROGENEOUS CONDITIONS IN

The ability of plants to locate roots in resource-rich soil patches with precision, the greater root biomass found in these patches, and the possibility that larger numbers of plants access such patches, suggest that competitive interactions may differ in uniform and heterogeneous soils with the same overall nutrient supply. It has been predicted (Schwinning and Weiner 1998) that whereas belowground competition is size symmetric when soil-based resources are homogeneously distributed, spatial heterogeneity may produce sizeasymmetric competition, with the plants that reach resource-rich patches rst gaining an advantage over their competitors that is disproportionate to the difference in their sizes (see also Casper and Jackson 1997). Despite the intuitive appeal of the prediction that competition will differ under homogeneous and heterogeneous conditions, the relevant information is conicting, and in some cases difcult to interpret. Cahill and Casper (1999) grew the herbs Phytolacca americanum and Ambrosia artemisiifolia alone and together, in homogeneous and heterogeneous conditions. When grown alone, both species produced more shoot biomass (signicantly more for P. americanum) under heterogeneous conditions. The presence of neighbors did not reduce mean shoot biomass of either species signicantly in any treatment. Casper et al. (2000) have cited this study as providing no evidence for an effect of heterogeneity on the interaction between plants. However, when grown alone, the mean shoot weight of P. americanum was signicantly greater (by 56%) in heterogeneous conditions than in homogeneous conditions. With neighbors, there was no difference between shoot weights in heterogeneous and homogeneous conditions (their Fig. 2). This suggests that P. americanum suffered greater competitive suppression under heterogeneous than homogeneous conditions. Results were less clear cut for A. artemisiifolia. Blair (2001) looked for evidence of greater shoot size variation among competing plants of Ipomoea tricolor under heterogeneous conditions than under homogeneous conditions with the same total level of re-

source supply. The absence of a signicant difference was regarded as evidence that competitive interactions were unaffected by the pattern of nutrient supply, although such a difference would only provide indirect evidence for size-asymmetric competition under heterogeneous conditions (Schwinning and Weiner 1998). Fransen et al. (2001) studied the impact of the spatial pattern of nutrient supply on plant competition between the grasses Festuca rubra and Anthoxanthum odoratum. Two different heterogeneous treatments and one homogeneous treatment were used. All treatments provided the same total nutrient supply. The two grasses were sown in monocultures or as 50:50 mixtures. Plant density was the same in all treatments. Under homogeneous conditions, the relative competitive ability of F. rubra exceeded that of A. odoratum, but both species were equally competitive under heterogeneous conditions. In the heterogeneous treatment with small, concentrated patches of nutrients, both species produced more root biomass in the enriched patches, but the extent of this proliferation was the same for both (neither showed such proliferation in the treatment with larger, less concentrated patches). Whereas root activity (estimated by measuring uptake of strontium, a calcium analogue) of F. rubra was higher than that of A. odoratum under homogeneous conditions, both species had similar root activities under heterogeneous conditions. Thus, the change in relative competitiveness under heterogeneous conditions appeared to have been caused by greater physiological responsiveness of A. odoratum to the pattern of nutrient provision. It would be predicted that plants with the ability to forage with precision for resources would be at a competitive advantage under heterogeneous conditions. Bliss et al. (2002) found that precise foragers were advantaged in competition with less precise foragers when growing in heterogeneous environments, although this was only signicant for 2 out of 6 species pairs tested. Clearly this prediction merits further investigation. Other studies indicate that the ability of plants to exploit nutrient-rich patches depends on whether or not interspecic neighbors are also sharing the patch (Huante et al. 1998, Hodge et al. 1999). Huante et al. (1998) demonstrated that for a fast growing tropical tree, the presence of a competitor of a different species reduced its ability to place its roots in a nutrient-rich patch. A recent study by Day (2001) is one of the rst to demonstrate that the intensity of competition is increased by heterogeneity. He examined the impact of competitors on the growth of the grass Briza media under heterogeneous and homogeneous conditions in which the total supply of nutrients was the same. Single target plants of B. media were either grown alone, or with a single conspecic neighbor, or a neighbor of the grass Festuca ovina. In the heterogeneous treatment the plants were placed in nutrient-poor locations within

Special Feature

2330

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

pots containing small cores of nutrient-rich soil, including one directly between, and equidistant from, the neighbor and the target plant. The impact of competition in each treatment was estimated by calculating the mean reduction in growth of target plants under competition compared with their mean growth when grown alone with the same nutrient supply. The biomass of B. media was signicantly reduced by competition in the heterogeneous treatment, but was not signicantly reduced by either intra- or interspecic competition in the homogeneous treatment. This constitutes strong direct evidence for a difference in the intensity of plant competition under heterogeneous and homogeneous conditions when nutrient supply remains the same. Both intra- and interspecic competitors caused similar reductions in the growth of B. media in heterogeneous conditions. Contrast between rich and poor patches was high in this experiment, and the size of the nutrient-rich patches was very small compared with the size of the nutrient-poor patches. This rather marked form of heterogeneity would be very likely to affect the intensity of competition (Box 1), because roots are competing for a resource that is highly restricted in its spatial distribution. One other study has demonstrated changes in competitive interactions in response to supplying a soilbased resource in different heterogeneous patterns (Novoplansky and Goldberg 2001). Water, rather than nutrients, was provided in pulses to three perennial grass species, producing heterogeneity in time rather than space. The performance of single plants was affected more by the total quantity of water supplied than by pulse frequency, but pulse frequency had far greater effects on the relative performance of plants growing in competition. When pulses were frequent, the species characteristic of the most productive eld conditions, which also had the fastest growth, caused the greatest suppression of competitors and was best at tolerating competition. When pulses were less frequent, the species from the least productive eld conditions, which had the slowest growth, became better at both suppressing and withstanding the effects of competitors. Although this experiment did not compare growth under heterogeneous and homogeneous conditions, it indicates clearly that providing soil-based resources in different heterogeneous congurations changes competitive relations. As the authors state, these results add to the growing evidence that we need to analyze not only the effects of mean resource conditions on plants, but also the effects of spatial and temporal variation in resource availability. Although limited in quantity, the available evidence supports the view that both the scale and pattern of heterogeneity in resource supply will have consequences for the way in which plants compete (Box 1). Moreover, it is clear that assessments of the relative competitive ability of species made under homogeneous conditions may be an inadequate basis from which to

predict the outcome of competition between the same species when grown under heterogeneous situations. THE VALUE TO PLANTS OF ROOT PROLIFERATION RESOURCE-RICH PATCHES OF SOIL
IN

Proliferation and selective placement of roots in nutrient-rich patches of soil is common, but by no means universally exhibited by plants (Robinson 1994). Robinson (1994, 1996) and others have calculated that, for many species, root proliferation may not occur until after most of the nutrients (particularly those with high mobility) in patches have been taken up. If this is the case, root proliferation would probably be an inefcient way of acquiring resources, being too slow and energetically costly for the resources gained to exceed the cost of the investment. Although nutrient-rich patches in natural soil are certainly transient (e.g., Farley and Fitter 1999), there is a shortage of data about their duration. Nutrient patches have remained detectable for rather long periods in several experimental studies, even when the concentration of mobile nutrients, such as nitrate ions, was measured. For example, nutrient patches were still detectable 610 wk after the start of glasshouse-based pot experiments involving competing plants (Casper and Cahill 1996, 1998, Cahill and Casper 1999). Many observations of root proliferation in soil patches with a high nutrient content have involved plants growing in isolation. Under natural conditions, however, plants usually grow with competing neighbors. Robinson et al. (1999) have demonstrated that, under these conditions, a plant that increases root growth in a nutrient-rich patch more rapidly than its competitor gains a higher proportion of the nutrients from the patch and, consequently, also gains a competitive edge. Root proliferation may be a deeply embedded response to competition that can not be avoided even when plants grow alone. It may even be advantageous for plants growing in isolation, as they are likely to have to compete with the soil microbial community for available nutrients. However, there may be important costs associated with root proliferation in resource-rich patches. Recent research has shown that when plants compete intraspecically for nutrients at a given concentration within a shared volume of soil, each plant produces more roots, but has a lower reproductive yield, than when given sole access to half as much soil containing nutrients at the same concentration (Gersani et al. 2001). Thus, although root proliferation denies resources to competitors, it imposes a cost in individual tness. Although nutrient acquisition may be faster, it is less efcient, because more roots are produced than necessary to harvest the nutrients from the patch. Ultimately, however, the cost to each plant is lower than it would be if it ceded resources to its competitor (the tragedy of the commons; Hardin 1968). When a plant grows alone, speed of acquisition becomes less important, as there are no neighbors to capture the nu-

Special Feature

September 2003

UNDERGROUND PROCESSES

2331

trients, and efcient nutrient harvesting can be achieved with less root, permitting greater biomass allocation to functions such as reproduction. The increase in root proliferation recorded by Gersani et al. (2001) when resources were contested by more than one plant indicates a highly developed capacity in plants to distinguish between their own roots and those of other plants. Further studies should be undertaken to examine the generality of the results of Gersani et al. (2001). Their study suggests that plants display highly aggressive behavior when faced with the problem of acquiring resources from a shared space. This strengthens the prediction that competition will be more intense, and more asymmetric, when resources are concentrated in a smaller proportion of the rooting volume than when they are uniformly distributed. One nal consideration to bear in mind is that species with the most marked root proliferation responses in nutrient-rich soil may well be those that forage most extensively (Crick and Grime 1987, Hutchings and de Kroon 1994). Such species tend to inhabit resourcerich environments, where nutrient-rich patches can be frequently encountered in time and space. It would be predicted that these species will produce low-cost, short-lived roots for rapid exploration and exploitation of resource-rich substrate patches. Their roots will either be jettisoned quickly, after patches are depleted of resources, or used as bridgeheads to reach more distant patches. In such species, the costs and benets of root proliferation cannot realistically be estimated on the basis of what happens in single patches, since proliferation serves both to increase resource acquisition in the plants immediate vicinity and to enable wider exploration of the soil environment. THE POPULATION-LEVEL CONSEQUENCES OF HETEROGENEITY IN NUTRIENT SUPPLY Despite convincing demonstrations of the impact of heterogeneity on individual plants, to date, only four studies have directly examined the effects of spatial heterogeneity in nutrient supply on plant populations. Once again results are contradictory. When populations of Abutilon theophrasti were grown with a xed quantity of nutrients supplied either uniformly or in a checkerboard pattern of high or low nutrient concentration patches, there were no signicant differences in yield or size hierarchy (Casper and Cahill 1996, 1998). The effect of heterogeneity on the performance of individual plants growing at the same locations within populations was also examined. Plants at only two out of 10 randomly chosen locations were affected by pattern of nutrient supply (Casper and Cahill 1998). However, mortality was twice as high in populations under homogeneous conditions as under heterogeneous conditions, and this difference was signicant (Casper and Cahill 1996).

In contrast, Day et al. (2001, 2003) found that the total yield, and yield of both below- and aboveground parts, of populations of Cardamine hirsuta was signicantly higher under heterogeneous conditions with a range of patch scales, than under homogeneous conditions providing the same overall nutrient supply. Variation in shoot size was signicantly greater in heterogeneous conditions, primarily because of large differences in the sizes of plants located in nutrient-rich and nutrient-poor soil patches, and increased (though not signicantly) as the scale of patches increased. As would be expected in heterogeneous conditions, the proportion of the populations biomass in nutrient-rich patches (calculated from the biomass of roots in nutrient-rich patches plus shoot biomass supported by plants rooted in nutrient-rich patches), was signicantly above 50%, although both rich and poor patches were equal in area and soil volume. Similar results are reported by Facelli and Facelli (2002), who found that size inequality and shoot biomass were both higher in environments in which phosphorus was supplied patchily than in otherwise identical environments where it was supplied homogeneously. The concentration of biomass in nutrient rich locations reported by Day (2001) and Facelli and Facelli (2002) suggests that nutrient-poor patches could be refuges from severe competition in heterogeneous conditions. This conclusion is reinforced by Days (2001) observation that, despite populations being planted at identical densities and in identical patterns, overall mortality was signicantly greater in homogeneous conditions. Homogeneous conditions may generate greater mortality because there are no such refuges. The majority of the plants that died in populations in patchy substrates were in nutrient-rich patches (Day 2001). The results of Day (2001) and Day et al. (2003) clearly demonstrate that, both at the level of individual patches, and at the level of the whole population, there are important differences in competitive processes and their outcome between populations growing under heterogeneous and homogeneous conditions. This is probably largely due to greater variation in growth and ensuing intensity of competition between plants in nutrient-rich and nutrient-poor soil patches in heterogeneous conditions. The differences observed by Day (2001) in mortality between populations in heterogeneous and homogeneous conditions parallel those reported by Casper and Cahill (1996). These studies of the impact of heterogeneity on populations lead to the conclusion that heterogeneity has major effects on several aspects of population performance (Box 1). THE COMMUNITY-LEVEL CONSEQUENCES OF HETEROGENEITY IN NUTRIENT SUPPLY To date, little quantitative information has been published about the effects on communities of supplying nutrients heterogeneously rather than homogeneously.

Special Feature

2332

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

It is often assumed that, because heterogeneous environments possess much variation in quality between habitat patches, the diversity of the communities they support is also high, because different species preferentially occupy patches of different quality (Fitter 1982, Pacala and Tilman 1994, Huston 1999). However, data do not always support this prediction (Kleb and Wilson 1997, Collins and Wein 1998, Wilson 2000). When the scale of heterogeneity is small enough for parts of the root systems of single plants to occupy several different quality patches simultaneously, the relationship between the number of different types of patches in the environment and the number of species that can be supported might be far from simple (Box 1). If some species are particularly well adapted to dominate patchily distributed resources, then heterogeneity at some scales could lead to reductions in species richness compared to otherwise identical homogeneous environments. For this reason, experimental studies of the effects on community structure of smallscale soil-based heterogeneity are urgently required. A recent study by Wijesinghe et al. (unpublished manuscript) demonstrates some consequences of such heterogeneity. The effect on community structure of providing the same quantity of nutrients in different spatial and temporal patterns was examined. A eld experiment was carried out over two years in boxes into which a standard inoculum of seeds of 20 species was sown. The substrate in the boxes was divided into 64 patches of identical size, without partitions separating them, so that roots could grow freely between them, and each patch was either enriched by adding a single tablet of a slow-release fertilizer or left without nutrient addition. Two scales of heterogeneous treatments were established, in which either 32 or 16 of the 64 patches were fertilized at each of two or four prescribed dates, respectively, in both years of the experiment. These treatments were labeled HET50 and HET25 because either 50% or 25% of the patches were fertilized at each feeding. There was also a homogeneous treatment (HOM), in which all 64 patches were enriched with a single fertilizer tablet at the start of each of the two growing seasons. All treatments received the same total amount of nutrients over the course of each of the two years. A destructive harvest of all above- and belowground plant parts, carried out at the end of the second year, showed that neither the Shannon Index of community diversity nor community evenness were affected at all by treatment. However, there was a 44% difference in biomass produced by the communities in the most (HET25) and least (HOM) productive of the treatments. Treatment affected root biomass more strongly (root biomass was 71% greater in the most productive than the least productive treatment) than aboveground biomass (there was a 20% difference in aboveground biomass between the most and least productive treatment).

The effects of treatment on both root and aboveground biomass were signicant. Analysis of the effect of homogeneity vs. heterogeneity, and of the HET50 vs. the HET25 treatments showed signicant effects on community biomass. Despite the lack of effects of treatment on community measures of diversity, there were effects of treatment on individual species. For instance, the biomass of Rumex crispus in the homogeneous treatment was at least twice that in any of the four heterogeneous treatments. Thus, this experiment demonstrates that heterogeneity in nutrient supply exerts important effects, both at the community level and at the level of individual species. However, this study was not able to distinguish between the effects of heterogeneity on individual species performances or its effects on species interactions. One group of species that was differentially affected by the different treatments was those that colonized the communities despite not being in the sown mixture. Such species were significantly more abundant in the nal community in the HOM and HET50 treatments than in the HET25 treatments. That some patterns of resource supply make a community more invasible than others has also been demonstrated recently by Davis and Pelsor (2001). It is not surprising that local variation in the intensity of competition, caused by spatial and temporal variation in resource supply, affects the ease of invasion by new species into plant communities (Davis and Pelsor 2001, Davis et al. 2000). We are only at the beginnings of understanding the impact of patchiness in resource supply on community structure. However, we are condent in predicting that future research will conrm that heterogeneity can both increase and decrease diversity, depending on the scale of patches, and change community biomass, the distribution of biomass between below- and aboveground parts, and species composition. Our rationale is summarized in Box 1. CONCLUSIONS There is a growing body of ecological evidence indicating important effects of heterogeneity in the availability of soil-based resources on individual plants, on the interactions between plants, on plant populations and on plant communities. Our knowledge of these effects, and the level of agreement over what they are, is greatest at the individual plant level. We still know very little about population- and community-level effects. Even the question of whether heterogeneity in the supply of soil-based resources alters the mode and intensity of competition is still subject to some debate. Our progress in understanding the impacts of heterogeneity on plant interactions will in part depend on progress in related areas, such as the understanding of belowground plantplant recognition mechanisms, and on our ability to describe belowground interactions accurately.

Special Feature

September 2003

UNDERGROUND PROCESSES

2333

Homogeneous environments rarely exist outside the laboratory and glasshouse. Even in agriculture, spatial heterogeneity (and, just as importantly, temporal heterogeneity) abounds. The impacts of heterogeneity on plant growth can be so great that it seems almost unbelievable that plant physiologists and ecologists have given it so little attention until recently. There is still much to be learned and understood about many fundamental issues, such as how plants behave, how much they grow, how they interact, and how their populations and communities develop, when their resources are patchily congured, rather than spatially or temporally uniform in supply. Further research is urgently need to ll the fundamental gaps that still remain in our knowledge in these areas.
ACKNOWLEDGMENTS We gratefully acknowledge guidance from Anurag Agrawal, and the constructive comments of two anonymous referees on an earlier draft of this manuscript. Part of the work reported was supported by grant GR3/11069 from the Natural Environment Research Council, UK. LITERATURE CITED Alpert, P., and E. L. Simms. 2002. The relative advantages of plasticity and xity in different environments: when is it good for a plant to adjust? Evolutionary Ecology 16: 285297. Alpert, P., and J. F. Stuefer. 1997. Division of labour in clonal plants. Pages 137154 in H. de Kroon and J. van Groenendael, editors. The ecology and evolution of clonal plants. Backhuys Publishers, Leiden, The Netherlands. Aung, L. G. 1974. Rootshoot relationships. Pages 2961 in W. Carson, editor. The plant root and its environment. University Press of Virginia, Charlottesville, Virginia, USA. Birch, C. P. D., and M. J. Hutchings. 1994. Exploitation of patchily distributed soil resources by the clonal herb Glechoma hederacea. Journal of Ecology 82:653664. Blair, B. 2001. Effect of soil nutrient heterogeneity on the symmetry of below-ground competition. Plant Ecology 156:199203. Bliss, K. M., R. H. Jones, R. J. Mitchell, and P. P. Mou. 2002. Are competitive interactions inuenced by spatial nutrient heterogeneity and root foraging behavior? New Phytologist 154:409417. Cahill, J. F., and B. B. Casper. 1999. Growth consequences of soil nutrient heterogeneity for two old-eld herbs, Ambrosia artemisiifolia and Phytolacca americanum grown individually and in combination. Annals of Botany 83:471 478. Cain, M. L., S. Subler, J. P. Evans, and M. J. Fortin. 1999. Sampling spatial and temporal variation in soil nitrogen availability. Oecologia 118:397404. Caldwell, M. M., J. H. Manwaring, and S. L. Durham. 1991. The microscale distribution of neighboring plant roots in fertile soil microsites. Functional Ecology 5:765772. Campbell, B. D., J. P. Grime, and J. M. L. Mackey. 1991. A trade-off between scale and precision in resource foraging. Oecologia 87:532538. Casper, B. B., and J. F. Cahill. 1996. Limited effects of soil nutrient heterogeneity on populations of Abutilon theophrasti (Malvaceae). American Journal of Botany 83:333 341. Casper, B. B., and J. F. Cahill. 1998. Population level responses to nutrient heterogeneity and density by Abutilon theophrasti (Malvaceae): an experimental neighborhood approach. American Journal of Botany 85:16801687.

Casper, B. B., J. F. Cahill, and R. B. Jackson. 2000. Plant competition in spatially heterogeneous environments. Pages 111130 in M. J. Hutchings, E. A. John, and A. J. A. Stewart, editors. The ecological consequences of environmental heterogeneity. Blackwell Science, Oxford, UK. Casper, B. B., and R. B. Jackson. 1997. Plant competition underground. Annual Review of Ecology and Systematics 28:545570. Casper, B. B., H. J. Schenk, and R. B. Jackson. 2003. Dening a plants belowground zone of inuence. Ecology 84:23132321. Collins, B., and G. Wein. 1998. Soil resource heterogeneity effects on early succession. Oikos 82:238245. Crick, J. C., and J. P. Grime. 1987. Morphological plasticity and mineral nutrient capture in two herbaceous species of contrasted ecology. New Phytologist 107:403414. Davis, M. A., J. P. Grime, and K. Thompson. 2000. Fluctuating resources in plant communities: a general theory of invasibility. Journal of Ecology 88:528534. Davis, M. A., and M. Pelsor. 2001. Experimental support for a resource-based mechanistic model of invasibility. Ecology Letters 4:421428. Day, K. J. 2001. Studies of the effects of spatial heterogeneity in nutrient supply on plants and their populations. Ph.D. Dissertation. University of Sussex, Brighton, UK. Day, K. J., M. J. Hutchings, and E. A. John. 2003. The effects of spatial pattern of nutrient supply on the early stages of growth in plant populations. Journal of Ecology 91:305 315. Einsmann, J. C., R. H. Jones, M. Pu, and R. J. Mitchell. 1999. Nutrient foraging traits in 10 co-occurring plant species of contrasting life forms. Journal of Ecology 87:609619. Facelli, E., and J. M. Facelli. 2002. Soil phosphorus heterogeneity and mycorrhizal symbiosis regulate plant intra-specic competition and size distribution. Oecologia 133:54 61. Farley, R. A., and A. H. Fitter. 1999. Temporal and spatial variation in soil resources in a deciduous woodland. Journal of Ecology 87:688696. Fitter, A. H. 1982. Inuence of soil heterogeneity on the coexistence of grassland species. Journal of Ecology 70: 139148. Fitter, A. H., and R. K. M. Hay. 1987. Environmental physiology of plants. Academic Press, London, UK. Fransen, B., and H. de Kroon. 2001. Long-term disadvantages of selective root placement: root proliferation and shoot biomass of two perennial grass species in a 2-year experiment. Journal of Ecology 89:711722. Fransen, B., H. de Kroon, and F. Berendse. 2001. Soil nutrient heterogeneity alters competition between two perennial grass species. Ecology 82:25342546. Gersani, M., J. S. Brown, E. E. OBrien, G. M. Maina, and Z. Abramsky. 2001. Tragedy of the commons as a result of root competition. Journal of Ecology 89:660669. Hardin, G. 1968. The tragedy of the commons. Science 162: 12431248. Hodge, A., D. Robinson, B. S. Grifths, and A. H. Fitter. 1999. Why plants bother: root proliferation results in increased nitrogen capture from an organic patch when two grasses compete. Plant, Cell and Environment 22:811820. Huante, P., E. Rincon, and F. S. Chapin III. 1998. Foraging for nutrients, responses to changes in light, and competition in tropical deciduous tree seedlings. Oecologia 117:209 216. Hunt, R., and A. O. Nicholls. 1986. Stress and the coarse control of growth and rootshoot partitioning in herbaceous plants. Oikos 47:149158. Huston, M. A. 1999. Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. Oikos 86:393401.

Special Feature

2334

MICHAEL J. HUTCHINGS ET AL.

Ecology, Vol. 84, No. 9

Special Feature

Hutchings, M. J., and H. de Kroon. 1994. Foraging in plants: the role of morphological plasticity in resource acquisition. Advances in Ecological Research 25:159238. Hutchings, M. J., and D. K. Wijesinghe. 1997. Patchy habitats, division of labour and growth dividends in clonal plants. Trends in Ecology and Evolution 12:390394. Jackson, R. B., and M. M. Caldwell. 1996. Integrating resource heterogeneity and plant plasticity: modelling nitrate and phosphate uptake in a patchy soil environment. Journal of Ecology 84:891903. Jackson, R. B., J. H. Manwaring, and M. M. Caldwell. 1990. Rapid physiological adjustment of roots to localised soil enrichment. Nature 344:5860. Kleb, H., and S. Wilson. 1997. Vegetation effects on soil resource heterogeneity in prairie and forest. American Naturalist 150:283298. Leyser, O., and A. H. Fitter. 1998. Roots are branching out in patches. Trends in Plant Science 3:203204. Lister, A. J., P. Mou, R. H. Jones, and R. J. Mitchell. 2000. Spatial patterns of soil and vegetation in a 40-year-old slash pine (Pinus elliottii) forest in the coastal plain of South Carolina, USA. Canadian Journal of Forest Research 30: 689699. Mordelet, P., S. Barot, and L. Abbadie. 1996. Root foraging strategies and soil patchiness in a humid savanna. Plant and Soil 182:171176. Mou, P., R. H. Jones, R. J. Mitchell, and B. Zutter. 1995. Spatial distribution of roots in sweetgum and loblolly pine monocultures and relations with above-ground biomass and soil nutrients. Functional Ecology 9:689699. Muller, I., B. Schmid, and J. Weiner. 2000. The effect of nutrient availability on biomass allocation patterns in 27 species of herbaceous plant. Perspectives in Plant Ecology, Evolution and Systematics 3:115127. Novoplansky, A., and D. E. Goldberg. 2001. Effects of water pulsing on individual performance and competitive hierarchies in plants. Journal of Vegetation Science 12:199 208.

Oborny, B. 1994. Growth rules in clonal plants and predictability of the environment: a simulation study. Journal of Ecology 82:341351. Pacala, S. W., and D. Tilman. 1994. Limiting similarity in mechanistic and spatial models of plant competition in heterogeneous environments. American Naturalist 143:222 257. Robinson, D. 1994. The responses of plants to non-uniform supplies of nutrients. New Phytologist 127:635674. Robinson, D. 1996. Resource capture by localized root proliferation: why do plants bother? Annals of Botany 77:179 183. Schwinning, S., and J. Weiner. 1998. Mechanisms determining the degree of size asymmetry in competition among plants. Oecologia 113:447455. Stuefer, J. F., H. de Kroon, and H. J. During. 1996. Exploitation of environmental heterogeneity by spatial division of labour in a clonal plant. Functional Ecology 10:328 334. Stuefer, J. F., H. J. During, and H. de Kroon. 1994. High benets of clonal integration in two stoloniferous species, in response to heterogeneous light environments. Journal of Ecology 82:511518. Wijesinghe, D. K., and M. J. Hutchings. 1997. The effects of spatial scale of environmental heterogeneity on the growth of a clonal plant: an experimental study with Glechoma hederacea. Journal of Ecology 85:1728. Wijesinghe, D. K., and M. J. Hutchings. 1999. The effects of environmental heterogeneity on the performance of Glechoma hederacea: the interactions between patch contrast and patch scale. Journal of Ecology 87:860872. Wijesinghe, D. K., E. A. John, S. Beurskens, and M. J. Hutchings. 2001. Root system size and precision in nutrient foraging: responses to spatial pattern of nutrient supply in six herbaceous species. Journal of Ecology 89:972983. Wilson, S. D. 2000. Heterogeneity, diversity and scale in plant communities. Pages 5370 in M. J. Hutchings, E. A. John, and A. J. A. Stewart, editors. The ecological consequences of environmental heterogeneity. Blackwell Science, Oxford, UK.