Special Reviews

Interactions and Self-Organization in the Soil-Microbe Complex

See allHide authors and affiliations

Science  11 Jun 2004:
Vol. 304, Issue 5677, pp. 1634-1637
DOI: 10.1126/science.1097394


Soil is the most complicated biomaterial on the planet. As with any material, the physical habitat is of prime importance in determining and regulating biological activity. However, until recently the opaque nature of soil has meant that any interrogation of its interior architecture has been relatively rudimentary, restricted to simple qualitative expressions of the physical heterogeneity that fail to relate to any specific function. However, new techniques and insights into the biophysical and biochemical processes of this inner space are leading to the developments of theoretical frameworks and experimental approaches that will allow us to sustainably manage Earth's most important resource. We introduce the concept that the soil-microbe system is self-organized and suggest new priorities for research based on an integrative approach that combines biochemistry and biophysics.

The Mars Rover is looking for evidence that life existed in martian soil. This is perhaps the only extraterrestrial foothold for life we will be able to directly observe for generations to come. If signs of life are indeed discovered, it promises to be one of the most notable scientific findings of all time. It is appropriate that soil is afforded this special status, underpinning as it does all forms of life on our own planet. Given its importance, it is surprising how little we know about our most important natural resource. Indeed, much about soil remains a mystery, yet it probably presents us with the most important clues as to how complex ecosystems become capable of self-organization and sustaining functionality. Pick up a handful of soil and ask the question “what is in it?” and an exciting new journey into inner space begins.

Biological Diversity

In fertile garden or organic soil, there will be more individual organisms than the total number of human beings that have ever lived: 1012 bacteria, 104 protozoa, 104 nematodes, 25 km of fungi, and countless other species. Depending on clay composition and amount, the soil's total surface area could cover a full-sized football pitch. So, although the soil is teeming with life, the fraction of the surface area covered by soil microbes is only about 10–6 %, which, coincidentally, is the same percentage of land area on Earth that humans cover. The important point here is not the absolute coverage, which can vary with clay content, moisture, and substrate availability, but the fact that it is considerably less than 1%, even with the most optimistic balance of inputs.

The soil-microbe complex is particularly important with regard to the services it provides for agriculture, waste management, and the water industry, as well as the natural and seminatural environment (1). There is now a large body of evidence showing that heterogeneity at these scales is highly ordered, and the resulting spatial clustering of the matrix gives soil its characteristic aggregated structure (2). This aggregated state gives rise to a broad range in pore sizes that permits the coexistence of air and water essential to the biological functioning of soil. This structure determines the ease with which plants may extract water, the rate of flow in unsaturated conditions, and the rate of diffusion of compounds and gases into and out of the soil matrix. Small changes in soil moisture can change the associated rate coefficients by orders of magnitude, sensitively dependent on the fine-scale structure.

The processes governing the survivability and functioning of above-ground communities are the same as those below ground. We now know that an understanding of ecosystem function requires an integration of both biotic and abiotic factors (3). Nowhere is ecosystem functioning more important to sustaining life than in soil. Here, however, although much work has concentrated on developing techniques that are able to measure the diversity of soil biology, little effort has been spent in connecting these measures with habitat and then function. This is despite the fact that the physical structure of soil is likely to have a major impact on the diversity of biophysical microenvironments for soil microbes. A major impediment to progress is the structural complexity of soil that presents major challenges in understanding its role as a habitat.

Diversity in Microenvironments

The physical habitat of soil is characterized by physical and temporal heterogeneities across all measured scales, from nm to km (4). The geometrical complexity of the pore pathways determines the biochemical processes that govern life on Earth, such as plant productivity (5), water retention (6), and greenhouse gas emissions (7), and offer an unrivaled buffering capacity against potential pollutants entering the waterways (8, 9). A key distinction between older geological features, such as sandstone and granite, and younger soil systems is that the latter exhibit pore structures that are defined not only by the chemical nature of the material but by life itself and thus experience significant biophysical and biochemical changes over relatively short spatial and temporal scales. This distinction appears to dominate all fertile soils and is possibly a key diagnostic for the health of soil ecosystems because it represents an important functional bridge between the physics and biology of soil (Fig. 1).

Fig. 1.

(A and B) Two soil thin sections taken from the same core. Note the high degree of spatial variability within one undisturbed soil sample. Each thin section is ∼30 μm thick and 2 cm in length. To visualize the pore space, we captured images on a high-resolution digital camera first under transmitted light (left-hand images) and then with the use of cross polar light. The latter allows us to distinguish pores from quartz grain. A binary image is then made of the solid (black) and the pore (white) space (right-hand images). (C) High-resolution biological thin section (∼30 μm thick and 600 μm in length). Illumination under ultraviolet light reveals the location of fluorescently labeled microbes, which can be segmented with the use of a series of image processing steps to reveal their location relative to the structure (yellow spots on right-hand image).

Quantifying the physical habitat of soil systems has been an area of intense research with relatively little reward. Most of the techniques at large (m) and small (μm to mm) scales have revealed a staggering variability in pore structures but have failed to relate such variability to function in any meaningful way (2). Over the past decade, research has moved from highly qualitative descriptors of pore shape to more quantitative measures of pore connectivity and tortuosity. More importantly, increasing emphasis has been given to understanding the consequences of structure for the physical and chemical properties of soil with a range of modeling techniques (6).

The capacity of soil to sorb chemicals from solution (the cation exchange capacity) regulates the movement of pollutants to the atmosphere and waterways. It controls the sorptive properties of nutrients through interaction between cations and clay particles, which tend to have net negative charges, and thus the productivity of the soil in terms of plant biomass and the activity of organisms and movement of viruses through the soil. The techniques used to measure this involve the homogenization of the soil below a specific particle size range, which is then saturated. Under such circumstances accessibility of reactive sites to any chemical in solution is maximized. This test is in widespread use throughout the world, the results of which may bear little relation to what is really happening. The accessibility of the reactive sites and retention time of solution near an exchange site is dictated by the fine-scale structure and the hydraulic connectivity and conductivity of the habitat. Comparisons between cation exchange capacities in homogenized and structured soils show that the amount of exchangeable cations of the structured soil is reduced by 10% in the structured system, where mass flow dominates. Where diffusion processes becomes more important, over longer time scales and/or under desaturated conditions, the reduction is over 50% (10). In other areas such as designing porous catalysts for chemical reactors, the links between reactive surfaces and exchange processes have been explained in terms of fractal processes (11, 12), an approach that has had a growing following in application to soil (13) and in some of our recent work on the impact of irregular surfaces in predator-prey dynamics in soils (14).

An important step toward integrating the biology and physics of the soil ecosystem has been the development of techniques that allow the microbes to be observed in their natural physical habitat. Foster, with the use of ultrathin sections, investigated the organization of soil organic matter and minerals in the context of bacteria (15) and suggested that the polar nature of some organisms allows physical adhesion between organisms and charged clay platelets (Fig. 1). More recent advances in sample preparation, image processing, and analysis have now allowed us to physically fix the location of fungi and bacteria and investigate their spatial distribution in the context of a spatially heterogeneous habitat (16). The use of such techniques in combination with powerful geostatistical analysis reveals a hitherto unknown variability in microbial populations. Some of our recent work has shown that, whereas the distribution of individual bacterial cells was organized at two separate scales in the subsoil, within the topsoil the distribution was random (17). Other work examining wholecommunity DNA and amplified fragment length polymorphism DNA fingerprinting showed that, even within a habitat deemed homogeneous at the plot scale, bacterial distributions were highly structured, with individual microbial communities being aggregated in specific ways in response to the spatial heterogeneity imposed by the habitat (18). The importance of such aggregation in terms of function and how we understand our environment is summarized in an excellent paper by Ettema and Wardle (19) (Movie S1).

Any technique based on sectioning is limited by the difficulty in inferring the properties of the original three-dimensional (3D) structure. Some of these limitations will be overcome by the advancements in computer-aided tomography, pioneered in soils research by Crestana et al. (20). New desktop systems promise advanced tomography capabilities including resolutions of 1 μm, dual energy facilities to decouple solid, water, and pore, as well as (in the higher specification systems) a facility to quantify the 3D distribution of some organics and minerals (Fig. 2). However, technical difficulties associated with distinguishing microbes will have to be overcome before the technique can be used to investigate structure-microbe interactions in situ (Movie S2).

Fig. 2.

A 3D visualization of a repacked dried soil core (3.5 cm by 1 cm), with a resolution of 42 μm Voxel. Computer tomography reconstruction produced by X-Tek System Limited (Tring, England) for SIMBIOS Centre. The range in photon densities of the inorganic and organic materials in many soils systems means that strict image-capture protocols and sophisticated image analysis must be used to resolve the pore space from the surrounding components. New generation machines will be able to distinguish water from pore and identify specific mineral components in three dimensions.

In parallel with the development of these imaging techniques, sophisticated modeling tools have been developed with the aim of understanding the consequences of soil structure and microbial distribution for various functions of the soil-microbe complex. Perhaps the most important of the generic approaches is the use of fractals as models of structure. These have provided a sufficiently simple representation to allow some potentially important generalizations to be drawn. First, with the use of these models it has been shown that assumptions behind the traditional measures of soil structure, such as the particle size distribution and the water retention characteristic, are invalid and that these measures on their own are unreliable. It has also been possible to show that the fractal-like structure of soil means that coefficients determining diffusion and convection rates will be scale-dependent over the length scales where the models are appropriate. These length scales are of importance to microbial processes, and, as we shall see, scale dependency in the rate coefficients might play a defining role in the organization of the soil-microbe complex and the sustainability of soil function. Second, fractal models have shown how natural length scales can emerge from the interaction between biological and physical processes in soil. With the use of the distribution of oxygen in soil as an example, Rappoldt and Crawford (21) have shown that simple models for denitrification that model the geometry of soil as a packing of aggregates obeying a log-normal size distribution require an artificial (and arbitrary) cutoff in upper aggregate size to perform well. A fractal model produces the same result, but the rarity of large-scale structures emerges naturally as a consequence of the interaction between biological activity and physical complexity of soil (Fig. 3).

Fig. 3.

The predicted distribution of oxygen in structured soil modeled as a fractal and its dependence on microbial respiration rate. Each box represents a 2D layer of soil open to the atmosphere at the upper and lower boundary. The structure in each box is the same, whereas the potential respiration rate per unit volume decreases from a maximum (top) to a minimum (bottom). Red denotes low oxygen concentration, yellow denotes atmospheric concentration, and light blue is the soil matrix. The pore-scale spatial complexity and diversity of oxygen environments is obvious in all boxes, as is the spatial proximity of high and low oxygen concentration regimes. Even where potential microbial respiration is low, regions of low oxygen concentration prevail. [See (21) for further details. Images courtesy of K. Rappoldt.]

Perhaps the most important lesson from fractal models is the emphasis on the power-law correlations that are observed in the physical structure of soil. These on their own cannot be regarded as proof that the structure is fractal, and indeed the measurement of the fractal properties of soil is fraught with difficulties (22). Further evidence is required, and this relates to the origins of the underlying power-law structural correlations. Therefore progress requires that we move from the static picture of soil structure that has predominated in the literature to one that embraces the dynamic properties. Interestingly, this picture points to the fundamental importance of the interaction between the physical and biological processes in soil.

System Dynamics

The prevailing understanding of the origin of soil structure emphasizes the different forces that dominate at different scales. At the molecular scale, aggregated structures such as clay platelets are dominated by electrostatic and van der Waal forces. At progressively larger scales, aggregation is dominated by soil microbes, such as fungi and bacteria either exuding glue-like substances or acting as reinforcement rods through the soil matrix, or roots acting to bind soil aggregates and particles together. This view formed the conceptual model of Tisdall and Oades (23) and has been consistently misinterpreted as providing evidence for the existence of discreet, experimentally manipulative aggregates rather than as a qualitative description of the aggregated hierarchical nature of the soil system in terms of the linkages between the architecture of the habitat and biological functioning. Over the past decade, this conceptual model has been used as an excuse to develop a wide variety of tests that purport to quantify the stability of soil ecosystems (24, 25) but in reality tell us little about the functioning of soil and more about the tests used (26). Nevertheless it does implicate microbial activity in the genesis of soil structure and provides us with a basis of a conceptual model for the dynamics of the soil-microbe complex.

Aggregation processes have been studied in a wide range of contexts in physics, chemistry, and biology (27). These approaches have been exploited in a simple model for soil aggregation and in particular to study the role of microbial activity in affecting structural characteristics (28). This model incorporated the impact of microbial activity on structural genesis as a scale-dependent binding probability representing the relative weakness and smaller probability of microbially mediated binding at larger scale. Fractal-like power-law correlations in the aggregated structure were predicted. The results showed that, whereas mechanical restructuring and the particle size distribution of the primary particles are major determinants of soil structure as measured by the fractal dimension, microbial activity affected only the range over which fractal-like power-law correlations in structure emerged. This, combined with the scale dependency in the rate coefficients that is predicted by fractal models for soil structure, means that increased microbial activity would lead to changes in soil structure that would increase local oxygen diffusion rates. Thus changes in the environmental context that act to increase local microbial activity (e.g., the arrival of a biologically available solute) will result in a reorganization of the soil-microbe complex that facilitates further exploitation (Fig. 4). Structure continues to build up in this way until the context is changed (e.g., the resource is used up), whereupon the structure changes to one with correspondingly lower diffusion rates.

Fig. 4.

Conceptual model for self-organization in the soil-microbe complex. Substrate arrives at a location in soil, and the potential microbial respiration rate increases, leading to local depletion of oxygen (left-hand image, with oxygen concentration represented as in Fig. 3). Increased microbial activity changes the local structure, creating a more open aggregated state. This leads to enhanced rates of oxygen supply (right-hand image). As substrate is used up, activity declines and the structure collapses to a more closed state. The open and closed states may represent optimal configurations for oxygen supply in a high potential activity regime (open) and protection from desiccation and predation in a low potential activity regime (closed).

In this picture, the functionality and dynamical behavior of soil emerges as a consequence of the interaction between the physical and biological processes as mediated by the structure of soil. The soil-microbe complex can be viewed as a self-organizing system capable of adapting to prevailing conditions (29, 30). Therefore, by ignoring the interaction between physical and biological processes in soil, we might be missing its most essential feature.


Soil is the most diverse and important ecosystem on the planet. Myriad biophysical and biochemical processes persist in parallel that are required to sustain all of the other trophic levels in the biosphere. A key to this is the physical complexity of the soil physical structure that provides the habitat for soil organisms and the conduit for essential resources.

Although ecologists studying aboveground systems are beginning to realize the important role of interactions between biotic and abiotic factors, soil science is still progressing with soil biology, soil chemistry, and soil physics as largely independent fields. This is doomed to failure, because no one discipline will be able to understand the most complex biomaterial on the planet. There is mounting evidence that the essential features of soil will emerge only when the relevant physical and biochemical approaches are integrated. Some progress has been made, but there remains much to be done. For example, a theory linking microbial population dynamics to biodiversity and function in terms of the soil microenvironment is more or less absent. This presents a major challenge and will require molecular biologists, microbiologists, soil physicists, and theoreticians to work closely together. A means of functionally classifying microbes in terms of essential traits that relate growth and competition to environmental conditions will be required. This will have to be accompanied by better molecular tools that enable accurate determination of abundances of different functional types in soil and determine the rate and consequences of gene flow. A 3D dynamical model of soil structural genesis is also absent but is likely to be needed to study the potential for self-organization in soil.

The representation of the diversity of microenvironments in soil depends on an ability to model the distribution of water, unsaturated solute flow, and diffusion in porous media. Techniques for some of these are relatively well advanced, but the modeling of multiphase flow is still problematic. An integrated approach to studying soil with emphasis on the interactions between physical and biological processes is required to bring the science up to comparable levels with above-ground ecology. This would greatly benefit from contributions by scientists working in related topics in application areas outside soil. Not only will soil science benefit, but, by having functionality, ecosystem dynamics, and evolution at its core, soil could become the paradigm system for exercising our understanding of sustainable ecological processes and reveal insights into how the inner space of a wide range of materials impacts function.

Supporting Online Material


Movies S1 and S2


View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article