Soil rhizosphere food webs, their stability, and implications for soil processes in ecosystems

John C. Moore1, Kevin McCann2 and Peter C. de Ruiter3

1Department of Biological Sciences, University of Northern Colorado, Greeley, CO80639USA,

2Department of Zoology, University of Guelph, Guelph, OntarioN1G 2W1Canada,

3Department of Environmental Studies, University of Utretch, 3508 Utretch, The Netherlands

Introduction

As students of biology we design and are exposed to a variety of caricatures to convey complex interactions and relationships. We are all acquainted with the first caricature, what for a better term can be referred to a the fundamental equation of life:

6CO2 + 6H20  C6H12O6 + 6O2(1)

Like Euler’s equation in mathematics, Equation 1 maps several concepts. It embodies the conservation of matter, as each side of the equation possesses different molecules but equal numbers of atoms and mass. It depicts the inter-dependence between life and death processes operating at different scales, photosynthesis and respiration, the autotroph and heterotroph, the interaction between a plant and herbivore, and the immobilization of inorganic matter into organic matter and the mineralization of organic matter to inorganic matter. If we add nitrogen to the equation a similar set of processes emerges, and the interdependence of elements in shaping rates and life processes is evident (Reiners 1986, Sterner and Elser 2001). As with carbon, nitrogen is immobilized into organic matter and mineralized into inorganic matter, but we see an added dimension of a tight coupling of the compartmentalized aboveground and belowground processes as organisms from within each realm has perfected the biogeochemical pathways to immobilize the immobilize the inorganic metabolic wastes of the other. Nowhere is this more apparent than within the rhizosphere, the region of soil influenced by the roots of plants.

Students are also familiar with a set of caricatures used to depict trophic interactions. The figures include a plant an herbivore and a predator, and if the vignette is of a terrestrial systems, an arrow points below the soil surface to ‘nutrients’ and/or ‘microbes’, followed by an arrow point to plant roots. The clear emphases of these depictions is on the aboveground realm, even though the interactions occurring belowground within the rhizosphere may be as or more significant in scope, complexity and overall importance to the system. Part of the reason the aboveground system receives greater attention is purely for heuristic reasons, and soils and soil processes are given short shrift stems from the obscure nature of soil biota and processes.

Mathematical models represent a third type of caricature. On the one hand, effective models are internally consistent, simple in design and assumption, and thought provoking. On the other hand, they can be devoid of the details that make them biologically interesting and lead to biologically counterintuitive results. A good example of the later being the unstable mathematical representations of mutualisms and the ubiquitous nature of what appear to be stable symbiotic mutualisms that occur within the rhizosphere that have evolved over time.

The objectives of this chapter are to present an approach that incorporates the three types of caricatures described above: 1) the reciprocal transfer of nutrients that are essential for plant growth and heterotrophic life depicted in Equation 1, 2) the trophic interactions among organisms aboveground and belowground, and 3) the mathematical representations of these. We demonstrate that the rhizosphere possesses a distinct trophic structure that is important to mathematical stability, and that human activities can alter the structure that are mathematically unstable and in ways that alter key ecological process.

The Rhizosphere

We define the rhizosphere as plant roots and the surrounding soil that is influenced by plant roots. This definitionencompasses not only the roots and region of nutrient uptake by the roots, but extends into soils by action of root products and the trophic iteractions that are affected by these products or by roots (Coleman et al. 1983, Van der Putten et al. 2001, Moore et al. 2003). This definition is more inclusive than other definitions that includes roots and the soils that adhere to them, but operationally allows for a richer discussion.

Significant quantities of photosynthetic products produced by plants are diverted to roots for root growth, which provides a carbon base for the soil species. The rhizosphere is characterized by rapid and prolific root growth, the sloughing of root cells, root death, and the exudation of simple carbon compounds. The size and dynamic of the rhizosphere relative to the aboveground component of plants differs by plant species and ecosystem type. For example, in grasslands, the ratio of shoot to root (S:R) production is roughly 1:1, contrasting sharply with forests, where far more photosynthate is allocated aboveground (Jackson et al. 1996), while Arctic tundra is characterized by a rhizosphere that turns over slowly resulting in an accumulation of root materials (Shaver et al. 1990). Interestingly, the range in S:R is narrowly conserved between .1 and 5 (Farrar et al. 2003); significant when contrasted with the range in plant sizes. The reasons offered for the constancy in S:R is due to the constraints on plant imposed by limitations and invariance in C:N and C:P ratios and the selective pressure to acquire just enough of the soil-based resources to balance aboveground carbon fixation. The constancy in the S:R and the dependence on elemental ratios greatly simplifies and strengthens our ability to generalize any models that we may develop.

Detailed studies of the rhizosphere reveal that a growing root can be subdivided into a continuum of zones of activity from the root tip to the crown where different microbial populations have access to a continuous flow of organic substrates derived from the root (Trofymow and Coleman 1982). The root tip represents the first and lowest root zone. It is the site of root growth and is characterized by rapidly dividing cells and secretions or exudates that lubricate the tip as it passes through the soil. The exudates and sloughed root cells provide carbon for bacteria and fungi which in turn immobilize nitrogen and phosphorous. Farther up the root is the region of nutrient exchange, characterized by root hairs and lower rates of exudation. The birth and death of root hairs stimulates additional microbial growth (Bringhurst et al. 2001). The upper zones have been characterized as the region of remineralization of nutrients by predators, the region of symbiotic mutualistic relations, and the structural region (Coleman et al. 1983). Within each of the zones there is an infusion of carbon into the rhizosphere by plants which stimulates the growth and activity of microbes (Foster 1988, Grayston et al. 1996, Bardgett et al. 1998) and their invertebrate grazers (Lussenhop and Fogel 1991, Parmelee et al. 1993).

Rhizosphere food web

Hunt et al. (1987) presented a model of the rhizosphere food web for the North American shortgrass steppe in Colorado based on the three descriptions of food webs proposed by Paine (1980) and on the subdivisions of activities described above: 1) the connectedness web depicts the trophic interactions among organisms, 2) the energy flow web represents the flow of nutrients among organisms, and 3) the interaction web depicts the influences of the dynamics of one group on another (Figure 1). This approach has been adopted by several research groups that have attempted to link the structure of soil food webs in relation to the decomposition of organic matter and the mineralization of nutrients (Andrén et al. 1990, Brussaard et al. 1988, Brussaard et al. 1997, de Ruiter et al. 1993a, de Ruiter et al. 1993b, Hendrix et al. 1986, Hunt et al. 1987, Moore et al. 1988).

The connectedness web defines the model’s basic structure (Figure 1). The diagram simplifies the high complexity and diversity by defining the web in terms of functional groups of organisms that shared similar prey and predators, feeding modes, life history attributes and habitat preferences (Moore et al. 1988). At the base of the web are plant roots, labile (C:N ratio < 30:1) and resistant (C:N ratio > 30:1) forms of detritus, and an inorganic nitrogen source. These basal resources are utilized microbes and invertebrates, terminating with predatory mites.

The energy flow web expresses food web structure in quantitative measures, i.e. population sizes (biomass) and feeding rates (Figure 1). The estimates of flow are derived indirectly using a simple food web model (Figure 1), that used estimates of population sizes, turn-over rates, consumption rates, prey preferences and energy conversion parameters (Table 1, see de Ruiter et al. 1993b; Hunt et al. 1987; O'Neill 1969).

Feeding rates were estimated using the procedures presented by Hunt et al. (1987). Consumed matter is divided into a fraction that is immobilized into consumer biomass (assimilation) and a fraction that is returned to the environment as feces, orts, and unconsumed prey, and of the assimilated fraction, material that is incorporated into new biomass (production) and materials that is mineralized as inorganic material. The estimates begin with top predators with the assumptions that the amount of material required to maintain the predators steady state biomass must equal the sum of its steady state biomass and loss due to death divided by its ecological efficiency:

F = (DnatB + P)/easseprod(2)

where F is the feeding rate (biomass time-1), Dnatis the specific death rate (time-1) of the consumer, B (biomass) is the population size of the consumer, P is the death rate to predators (biomass time-1), and eassand eprodare the assimilation (%) and production (%) efficiencies, respectively. For a top predator the death due to predator is zero. For predators that consume multiple prey-types the fluxes are weighted by the predators feeding preferences for the respective prey. The estimation procedure moves downward through the prey to the basal resources with fluxes to each prey taking into account the biomass lost to predation. A dynamic version can be constructed by taking into account changes in the biomasses over an interval of time t, i.e., adding ∆B/t to the numerator of Equation 2.

The interaction web emphasizes the strengths of the interactions among the functional groups (Figure 1). Moore et al. (1993) and de Ruiter et al. (1995) developed a means to estimate interactions strengths from the energy flow web (Hunt et al. 1987) and the series of differential equations used to describe the dynamics of each functional group (the elements of the Jacobian matrix). The Jacobian matrix consists of zeros to denote no direct interaction (feeding) between functional groups and non-zeros representing the interaction strengths, as determined by the partial derivatives of the equations describing the growth and dynamics of the functional groups at or near equilibrium (May 1973):

ij = [(dXi/dt)/j](3)

Where ijrefers to theinteraction strengths that designate the per capita effects (in the case of the present energy flow webs per biomass effects) of the functional groups upon one another. Interaction strengths can be derived directly from equations used to model the population dynamics the functional groups if the population densities of the functional groups and the feeding rates (Equation 2) are known. The key assumptions behind the estimation procedure are as follows: 1) the equilibrium biomass of functional group i (X*i in Equation 3) in the rate equations are approximated from long-term seasonal averages of biomass of functional group i (Bi), and 2) the consumption terms in the rate equations for prey i to predator j the can be estimated from the flux rates estimated as described in Equation 2, i.e., Fij = cij X*i X*j , where cijis the coefficient of consumption of functional group j on functional group i. Hence, if derived from rate equations based on Lokta-Volterra rate equations, the interaction strengths are ij = Fij/Bj for the per capita effect of predator j on prey i, and ji = ajpjFij/Bj for prey i on predator j (Moore et al. 1993, de Ruiter et al. 1994, de Ruiter et al. 1995). The diagonal elements of the matrix cannot be derived from field data or estimates of energy fluxes, but can be scaled to the specific death rates (de Ruiter et al. 1995) or can be set a levels that ensure stability (Neutel et al. 2002) depending on your aims.

Patterns in Structure

The connectedness and energy flux descriptions reveal two patterns in the distribution of energy, nutrients and biomass within the system that are important to its stability. The first pattern deals with the flow of energy from roots to top predators. The food web that develops within the rhizosphere is complex (Figure 1), consisting of multiple assemblages of species that originate directly from roots and root by-products (Hunt et al. 1987). The system possesses three distinct pathways or energy channels (Table 2) originating from living plant roots, resistant detritus through fungi, and labile detritus through bacteria (Coleman 1976, Coleman et al. 1983, Hunt et al. 1987, Moore and Hunt 1988). Moore et al. (1988) described these assemblages as the root, bacterial and fungal energy channels, the organisms within which share distinct physiological and behavioral attributes (Table 3) in terms of resource utilization that lend themsleves to this type of compartmentalization (Schoener 1974). Root feeding insects and nematodes, root pathogens, and microbes that engage in symbiotic relationships with plant roots (e.g., mycorrhizal fungi, Rhizobium, Frankia) form the base of the root energy channel. The bacterial energy channel is composed of bacteria, protozoa, nematodes, and a few arthropods. The fungal energy channel largely consists of saprobic fungi, nematodes and arthropods. Soil bacteria compose most of the microbial biomass in the rhizosphere, are aquatic organisms and are more efficient in using the more labile root exudates than saprophytic fungi (Curl and Truelowe 1986). In contrast, fungi are more adapted to utilize more resistant root cells and substrates than are bacteria. Moreover, fungi and their consumers occupy air filled pore spaces and water films, and possess longer generation times. Moreover, nutrients within each channel are processed at different rates given the differences in the recalcitrance of the materials that bacteria and fungi utilize and the physiologies of the fauna within each channel. Coleman et al. (1983) reconized these differences and refered to what became known as the bacterial energy channel as a "fast cycle", while what came to be known as the fungal energy channel represented a ”slow cycle.” Mathematical representations of the type of compartmentalized architecture whose subsystems differ in dynamic properties as described above have been shown to be more dynamically stable than random constructs possessing the same diversity (number of groups) and complexity (number of linkages among groups) (May 1972, May 1973, Yodzis 1988, Moore and Hunt 1988).

The second pattern deals with the distribution of biomass with increased trophic level. The systems we have described to date possesses a steep trophic pyramids of biomass and feeding rates (Table 1). For example, bacteria and fungi account for upwards of 95% of consumer biomass, leaving less than 5 % for the protozoa and invertebrates occupying the upper trophic positions (Hunt et al. 1987, de Ruiter et al. 1993). Recent work linking the propositions by Oksanen et al. (1981) on the role of increased productivity adding trophic levels to systems in a stair-step manner, and the ‘paradox of enrichment’ of Rosenzweig (1971) that increased levels of productivity lead to oscillations and dynamic instability suggest that the pyramidal structure is more stable than alternatives with higher biomass maintained at upper trophic levels (Moore and de Ruiter 2000, Neutel et al. 2002, Moore et al. 2003).

The interaction web possesses a trophic level dependent pattern where interaction strength is relatively strong negative effects of predators on prey at the lower trophic levels and relatively strong positive effects of prey on predators at the higher trophic levels. This patterning in the interaction strengths in coincident with food web stability, as the Jacobian matrices of the soil food webs with the pattern are more stable than matrices without the pattern (de Ruiter et al. 1995, Moore et al. 1996, de Ruiter et al. 1998). Moreover, the pattern repeats itself in a fractal-like manner for each of the energy channels (Figure 3) reinforcing the evidence that the system is compartmentalized along energy utilization (Moore and de Ruiter 1997). In retrospect, these results are not surprising given that the patterning of interaction strength is also intimately linked with the distribution of biomass and the feeding rates discussed above given that both are components of the estimates of interaction strength. Redistributing interaction strengths in the analysis concomitantly redistributes biomass and feeding rates.

Shifts among Energy Channels within the Rhizosphere

The relative dominance of the root, bacterial, and fungal energy channels is important to key processes and stability. Studies of grasslands, forests, arctic tundra, and agricultural systems indicate that the linkages among the energy channels tend to be weak at the trophic levels occupied by roots, bacteria and fungi, and strongest at the trophic levels occupied by predatory mites. The strength of the linkages among energy channels and the dominance of a given energy channel varies by the type of ecosystem, can change with disturbance, and affects nutrient turnover rates (Figure 4). The fungal energy channel tends to be more dominant in systems where the ratio of carbon to nitrogen is high (e.g., forests, no-till agriculture) while the bacterial channel is more dominant in systems with narrow ratios (e.g., grasslands, conventional tillage agriculture). Regardless of the relative dominace of either channel, disturbances that either added labile nitrogen through fertilization, disrupt soil aggregates, or remove vegetation have been shown to induce shifts in the linkage between the fungal and bacterial energy channels that favor the bacterial channel (Figure 4). Accelerated rates of decomposition, increases in the rates of nitrogen mineralization, and increases in nitrogen loss have been associated with an increase in the dominace of the bacteria energy channel (Hendrix et al. 1986, Andrén et al. 1990, Moore and de Ruiter 1991, Doles 2000).