A number of factors may affect the robustness of using this model- based approach to soil food web dynamics for evaluating ecosystem-level effects. For example, food web connectance is dependent on the species richness of the food web. Amalgamation of soil food webs at the level of functional groups underestimates food web richness and, probably, biases connectance estimates (Wardle, 1995). Studies by Martinez and coworkers (Martinez, 1993; Martinez et al., 1999) indicate that food web structure predicted by theory is highly dependent on the resolution at which the interactions within the web are resolved. We also lack a complete under- standing of the roles that many soil organisms play in the food web. Here, I discuss some shortcomings of the current soil food web model and draw attention to techniques that should help to address some of these shortcomings.
5.1. Resolution
Functional groupings underestimate the diversity of organisms taking part in soil food webs. For example, carnivorous nematodes are represented in one or two trophic groups, as predatory or omnivorous, in most food web models.
However, researchers observe differences in feeding behavior of predatory nematodes, generally, depending on taxonomic affiliation, that may affect the prey items that they feed on (Khan and Kim, 2007; Yeates et al., 1993).
Predators in the order Mononchida eat nematode prey whole or cut nema- todes into smaller pieces before ingesting them. Those in the Diplogasterida can be omnivorous, feeding on microbes and microbivores, and possess a smaller buccal cavity than mononchs, limiting their ability to feed on large prey. Predators in the order Dorylaimida are also omnivorous but ingest prey contents after either piercing the prey with an odontostyle or slicing it with a mural tooth. Predators in the Aphelenchida feed in a fashion similar to dorylaims, piercing prey with a stylet and paralyzing it, followed by ingestion of body contents; aphelenchs are generally smaller than dorylaims but are able to enzymatically digest and consume large prey items. Khan and Kim (2007) summarize the results of feeding experiments and measurements of field populations from studies addressing the biological control potential of various predatory nematode species, revealing particular effects at finer levels of taxonomic resolution. Some predatory species (e.g., Mononchus aquaticus) feed on a wide range of plant-feeding nematodes, while others (e.g.,Discolai- mus arenicolus) have a narrow observed prey preference. Nematode species also differ in terms of the degree with which they suppress prey populations, ranging from no effect to complete elimination. The data summarized by Khan and Kim (2007) are limited to observations where prey items were plant- feeding nematodes, but these observations suggest that finer resolution will reveal greater trophic structure in other parts of the soil food web, as well.
Another simplification is that food web diagrams often depict unidirec- tional flows of energy through consumers. Real webs, however, contain omnivory (feeding at more than one trophic level) and cannibalism (feeding within one’s own functional group), which are difficult to detect unless stable isotope abundance within the different biomass pools is also estimated (Section 5.2). As a result, the extent of consumption in food web models is generally underrepresented. In addition, some organisms are grouped within lower trophic levels but actually consume organisms at higher trophic levels.
For example, nematophagous fungi feed on a variety of nematode species but, when quantifying soil food web dynamics, the abundance of nemato- phagous fungi is represented as fungal biomass, thus overestimating resource abundance and underestimating consumer abundance.
5.2. Integration of the detritivore and herbivore food webs A great deal of effort has gone into describing trophic interactions involving detritivores and root herbivores, mainly nematodes, to determine their roles in nutrient cycling and plant health, respectively. However, interactions involving consumers of root herbivores and symbiotic microorganisms are
often poorly integrated into food web models. Aboveground, pathogenic fungal biomass was similar to herbivore biomass in a grassland experiment (Mitchell, 2003), suggesting that pathogenic microorganisms, in theory, could form the base of a food web energy pathway through which a considerable proportion of energy flows. Pathogenic bacteria and fungi are likely similarly abundant belowground; however, the extent to which these pathogenic organisms are consumed directly, and thus their direct involvement in soil food webs, is unclear.
Many symbiotic organisms are susceptible to predation at some point in their life cycles. Some microbial antagonists (e.g., myxobacteria) kill bacte- rial and fungal pathogens and assimilate the contents of lysed cells (Dawid, 2000); further predation on these antagonists would constitute a food chain.
Protozoan predation regulates population density of nitrogen-fixing rhizo- bia (Danso and Alexander, 1975). Mycorrhizal fungi have extensive hyphal networks, extending into the soil from plant roots, that may be subject to grazing (Smith and Read, 1997). Ectomycorrhizal fungi are able to access carbon through their association with ectomycorrhizal plant hosts or via decomposition of recalcitrant carbon sources, suggesting that their consu- mers access carbon derived from both the detritus and living plants. AM fungi, on the other hand, are unable to access carbon other than that derived from the mycorrhizal host and, thus, the path of energy flow is less ambigu- ous. However, there is currently some debate as to how AM fungi fit into soil food webs. Laboratory experiments (Moore et al., 1985) and studies with field soil using hyphal in-growth cores (Johnson et al., 2005) suggest that mites and collembola represent significant sources of biomass loss for AM fungi, yet choice experiments and vertically structured microcosm experiments (Klironomos and Kendrick, 1996) suggest that AM fungi are less palatable to microarthropods and that the vertical distribution of fungi in litter and soil plays a significant role in determining whether trophic inter- actions occur among fungi and their consumers.
Root-feeding arthropods are attacked by a variety of parasites and predators. Entomopathogenic nematodes of the families Steinernematidae and Heterorhabditidae, as a group, can infect and kill a range of soil insects, herbivorous and otherwise (Poinar, 1979). The nematodes are essentially bacterivores, feeding on bacteria that they carry around with them and inoculate into the host hemocoel, and only occur outside of the host as an infective juvenile stage. As these infective juveniles have patchy distribu- tions (Stuart and Gaugler, 1994), they represent an ephemerally abundant food source containing energy derived largely, but not entirely, from living plants. Tracking the source of that energy, however, is complicated since taxonomic identification of infective juveniles is difficult, abundance is generally determined through indirect and imprecise measures, and, even in cases where relationships involving specific herbivores and nematodes in
the field are described (Parkmanet al., 1993; Stronget al., 1996), unknown alternative hosts may contribute significantly to energy flows.
In addition to their involvement within the herbivore food web, root herbivores and symbionts may indirectly influence the detritivore food web via their effects on the productivity of individual plant species (De Deyn et al., 2004; Klironomos, 2002, 2003) and entire plant communities (Burdon, 1987; De Deyn et al., 2003; van der Heijden et al., 1998), thus controlling the amount of detritus entering the soil food web. However, other factors complicate the relationship between plant productivity and activity in the detritivore food web. In a 3-year field experiment, Wardle et al. (1999) did not observe consistent associations between plant biomass and abundance within various functional groups of soil biota; the authors suggested that long-term litter and soil organic matter dynamics, resource quality, and regulation of consumer populations by predators structured soil communities to a greater extent than short-term changes in plant biomass.
More detailed descriptions of soil food web dynamics may help to address uncertainties regarding the linkages between the herbivore and decomposer food webs, and the indirect versus direct effects of herbivore activity on consumer activity in the decomposer food web.
Some invertebrate herbivores spend their entire life cycle belowground and actively disperse over relatively short distances (e.g., various plant- feeding nematodes), while others have both aboveground and belowground components of their life cycles and can actively disperse over long distances (e.g., various dipterans have larval stages described as ‘‘root maggots’’). This distinction may be of functional consequence since the consumption of invertebrates with aboveground dispersal stages prevents their emergence and dispersal and, thus, retains nutrients within the system. This process is analogous to that recently suggested for aquatic food webs contained within bromeliads (Ngai and Srivastava, 2006). Therefore, consumption of/by invertebrates with aboveground dispersal stages should have a greater per capita effect on local nutrient dynamics than consumption on/by those without aboveground dispersal stages.
5.3. Role of technology in resolving soil food webs
Initially, when describing trophic interactions, soil ecologists were limited to conducting feeding trials under artificial, laboratory conditions and examining gut contents of field-collected specimens. The adoption of stable isotopes was a significant technological advance that allowed soil ecologists to describe feeding behavior and energy flow through soil food webs in the field (Huntet al., 1987). Isotopic signatures (13C,15N) in consumer biomass vary predictably in response to signatures in resource biomass. The isotopic signature of a material is depleted each time that material passes through a consumer; therefore, an organism’s isotopic signature can be use to infer its
trophic level and whether it has engaged in omnivory (McNabb et al., 2001). Soil ecologists were thus able to validate the placement of functional groups within trophic levels and estimate the number of trophic levels present within the web, but were not able to detect specific consumer–
resource interactions. However, a number of recently developed techniques should facilitate the mapping of these interactions in much greater detail.
Analysis of dietary fatty acids may provide further resolution of trophic interactions to a level where the investigator can identify, broadly, the taxo- nomic group that the sampled organisms fed upon. Fatty acids present in cellular membranes of bacteria and fungi are assimilated into neutral (storage) lipids of their consumers and can be detected across at least three trophic levels (Ruess et al., 2005). Groups of bacteria (e.g., aerobic bacteria, anaerobic bacteria, cyanobacteria) and fungi (AM fungi, other fungi) can be resolved by the relative abundance of phospholipid fatty acids (PLFAs) that make up cellular membranes, and certain fatty acids are used as biomarkers to determine the presence and relative abundance of some groups in soil (Frostega˚rdet al., 1993; Olsson, 1999) or the diets of consumers (Ruesset al., 2005). Ruesset al.
(2005) found the technique powerful enough to discriminate the regions from which collembolans were sampled; some species (e.g.,Folsomia quadrioculata) had similar fatty acid compositions in the different regions, suggesting similar diets, while others (e.g.,Neanurum muscorum) differed in fatty acid composition at the different regions, suggesting a geographic pattern in diet. This technique is limited to detecting trophic interactions in which fatty acids are assimilated into consumer biomass; organisms that are consumed but whose fatty acids are not assimilated are not detected.
Recently developed approaches combine the analyses of stable isotopes and dietary fatty acids to gain simultaneous estimates of dietary preferences and food quality. Haubertet al. (2006) provided a variety of bacterial iso- lates to each of three different collembolan species, observing shifts in the neutral lipid fatty acid (NLFA; i.e., storage lipids) profiles of collembolans depending on the bacteria on which they fed. Three different surrogate variables, in addition to body mass and C/N ratio, were used to infer that one of the bacterial species represented poor food quality for the collembo- lans; collembolan NLFA:PLFA ratios were reduced and both13C and15N were enriched, suggesting metabolic mobilization of lipid reserves. Another approach for studying trophic interactions, stable isotope probing, involves the pulse-labeling of a resource and attempting to detect its presence in potential consumers; by looking for the presence of the isotopic signature within biomarker PLFAs, the consumer of the resource can be identified (Dumont and Murrell, 2005). Johnson et al. (2005) used stable isotope probing to estimate the extent to which grazing by collembola reduced AM fungal growth, indicated by the PLFA 16:1o5.
Another novel approach to resolving trophic interactions in soil is the analysis of DNA in the gut contents of predators. Soil contains a number of
materials (e.g., humic acids) that inhibit the polymerase chain reaction and can, therefore, lead to false negatives in detection of target DNA. Two recent studies describe protocols that allow for detection of prey DNA in the gut contents ( Juen and Traugott, 2006) and whole specimens (Pons, 2006) of predatory soil invertebrates. Juen and Traugott observed DNA degradation over two days in their experiment, suggesting that this method would be useful for determining the identity of prey that had been recently consumed. Clearly, the use of this method would be constrained to con- sumers that ingest their prey prior to digestion, as do many bacterial- and fungal feeders and organisms at higher trophic levels.
Many consumers occupying low levels in the soil food web, including bacteria and fungi, obtain nutrients following the secretion of extracellular enzymes. Advances in stable isotope probing of DNA sequences may allow for high resolution of trophic interactions at the base of the soil food web.
Dumont and Murrell (2005) review the use of stable isotope probing in environmental microbiology. Here, microorganisms in environmental sam- ples are exposed to a stable isotope-labeled substrate (e.g., 13C-labeled glucose). RNA is then extracted from the sample, labeled with a fluorescent probe, and hybridized to an oligonucleotide array to identify the RNA sequences extracted from the sample. The array is then viewed using autoradiography to determine which of the extracted RNA sequences were derived from organisms utilizing the labeled substrate (i.e., contain radioactive elements). A recent study adapted the technique to characterize microbial trophic interactions. Lueders et al. (2006) amended field soil with 13C-labeled Escherichia coli, separated labeled RNA from unlabeled RNA using equilibrium density gradient centrifugation, and then charac- terized the RNA sequence heterogeneity of the two fractions. Sequences belonging to fungi in the Microascaceae and bacteria in the Xanthomona- daceae, Myxococcales, and Bacteroidetes were associated specifically with the 13C-labeled RNA fraction, suggesting that some organisms in these groups assimilated nutrients derived from the amendedE.coli.