Ecosystems are inherently complex and contain a mind-boggling array of biotic and abiotic interactions. The studies mentioned in the previous section estimated effects of perturbations on functional groups within a soil food web. Many authors then attempted to scale the results up to gain information about the response of the soil food web as a whole; several interpret effects on consumer abundance as an indirect response to effects on resource abundance and, sometimes, lack of an effect on resource abun- dance as an indirect response to stimulatory effects on consumer abundance.
However, interpretation of functional group abundances becomes difficult when time lags mask the manifestation of consumer responses to increased resource availability (Ettema et al., 1999; Wardle et al., 1999), especially when the sampling design does not take temporal variation into account.
Alternatively, responses in any one functional group may be independent of the trophic interactions involving that functional group. Instead, functional groups may respond to the direct effects of the perturbation or indirect effects on abiotic factors. Therefore, it is not always valid to interpret results in the context of food web interactions, or to extrapolate changes in population estimates to ecosystem-level responses.
Fortunately, soil ecologists have developed two approaches to collapse large datasets on organism abundance and trophic status into interpretable estimates of the structure and function of the food web itself. The first, nematode faunal analysis, is an empirical approach that incorporates infor- mation regarding life history characteristics (e.g., rate of population growth) and trophic status of a subset of soil organisms to estimate emergent proper- ties of the existing soil food web, such as stability and productivity. The second is a modeling approach that attempts to (1) predict how localized changes within a food web (i.e., within a functional group) will influence
the overall stability and productivity of the food web and (2) deter- mine what properties of food webs make them resistant or resilient to perturbations.
4.1. Nematode faunal analysis 4.1.1. Theory
It is typically difficult to quantify the condition of an ecosystem, which is dependent on many factors (e.g., nutrient status, disturbance history). The nematode faunal analysis concept attempts to gain information that surrogates for ecosystem-level factors by estimating components of food web structure from nematode communities. Nematodes are particularly suited as environ- mental indicators since they contain more trophic complexity than other taxonomic groups of soil organisms (Fig. 1); nematodes represent multiple trophic levels and occupy energy pathways based on all three resource-types (roots, bacteria, fungi). Nematodes are also important as their trophic activities influence nutrient cycling in natural and managed systems (Anderson et al., 1983; Ingham et al., 1985). An analogous system for estimating food web structure does not exist for any other group of soil organisms.
Various indices are used to interpret nematode community shifts at a relatively high level of taxonomic resolution (family/genus); the most frequently used are the maturity index (MI), channel index (CI), enrich- ment index (EI), and structure index (SI). The indices combine information regarding the trophic guild (bacterivore, fungivore, herbivore, carnivore, or omnivore) and life history of the sampled nematodes. Life history is scored along a colonizer-persister scale; colonizer taxa have high population growth rates and are typical of nematode communities following a recent disturbance. Persister taxa are slower growing and typical of nematode communities in environments with low frequency of disturbance. The maturity index (MI; Bongers, 1990)
MI ẳXn
iẳ1
kcpx ncp n
ð1ị
accounts for the relative proportion (nc–p/n) of nematodes in a sample (exclud- ing plant feeders) that fit into categories (c–p) along the colonizer-persister scale, with k representing the weighting for any particular c–p category.
A sample with a low MI indicates that the sample is dominated by opportunist taxa; as the MI approaches the maximum (5), the sample becomes increasingly dominated by slower growing, disturbance-sensitive taxa. An analogous index exists for plant-feeding nematodes, the plant-parasite index (Bongers, 1990), and the weighted MI (Yeates, 1994) includes plant-feeding and free-living taxa.
Nematologists proposed additional indices that incorporate life history characteristics and trophic behavior of nematodes to a greater extent. The channel index (CI; Ferriset al., 2001)
CIẳ100 0:8Fu2
3:2Ba1þ0:8Fu2
ð2ị
estimates the relative weighting of the bacterial and fungal pathways of the soil food web by measuring the relative abundances of opportunitistic, free- living nematodes in these guilds. A CI that approaches 0 indicates domi- nance by the bacterial energy pathway, while an index approaching 100 indicates dominance by the fungal pathway. The perceived benefit of employing the CI, as opposed to estimating the ratio of bacterial- or fungal-feeding nematodes to all microbivorous nematodes (the nematode channel ratio), is that the CI focuses on the faster-growing, opportunistic bacterial- and fungal-feeding species that respond rapidly to enrichment, while attempting to correcting for differences in the rate at which energy flows through the two pathways.
The EI (Ferriset al., 2001), which estimates responses associated with the nutrient status of a system, is calculated
EIẳ100
Pn
iẳ1kene
Pn
iẳ1keneỵPn
iẳ1kbnb
ð3ị
and the SI (Ferriset al., 2001), which estimates the degree to which trophic interactions within food webs have developed, is calculated
SIẳ100
Pn
iẳ1ksns
Pn
iẳ1ksnsỵPn
iẳ1kbnb
ð4ị
wherenrepresents abundance andkrepresents the weightings for feeding guilds associated with enrichment (e), structure (s), and basal (b) components of the food web. Both indices scale on a range from 0 to 100. A high EI indicates greater availability of labile nutrients in the system, which stimu- lates the more rapidly cycling bacterial pathway. A high SI indicates the greater abundance of carnivorous and omnivorous nematodes, presumably due to a lack of disturbance in the system or greater resilience/resistance of the food web as structured.
Estimates from the enrichment and structure indices can be calculated from the same sample and graphed together (Fig. 2); the placement of data points in one of the four quadrats in the bivariate plot space suggests certain functional properties of the ecosystem within which the food web resides (Table 1).
4.1.2. Application
Several recent studies have employed this version of the nematode faunal analysis concept. Most of these studies were conducted in agricultural systems, estimating soil food web responses to soil and crop management practices. In a series of papers, Wanget al. (2003, 2004, 2006b) evaluated the main effects of amendments on nematode trophic structure and their interactive effects with other management practices. Compost amendment (269 Mg ha1 year1, derived from sticks, lawn clippings, and wood fragments) for 5 years increased nutrient availability (higher EI: 31.8 vs 23.9 in the absence of compost) and the relative contribution of the bacterial energy pathway (low CI: 18.5 vs 59.4);
the SI (38.4–52.2) indicated an intermediate level of trophic organization but was not significantly affected by compost amendment (Wang et al., 2004).
Amending soil from compost-incorporated and control plots with sunn hemp (Crotalaria juncea) hay (1 g per 100 g soil) resulted in a greater MI in one of two greenhouse experiments (2.02–2.12 vs 1.97–2.00 in theC.junceaunamended soil) but no effects on the structure, enrichment, or channel indices (Wang et al., 2003). In a field experiment, amendment withC.junceahay resulted in a greater reduction in the maturity and channel indices, suggesting increased abundance of opportunitistic, bacterial-feeding nematodes, and a greater increase in the EI, indicating more rapid nutrient cycling, than ammonium nitrate application (Wanget al., 2006b).
Basal
Basal condition
Structure trajectory
Structure index
Quadrat D Quadrat C
Quadrat B Quadrat A
Fu2 (0.8)
Ca2 (0.8) Om4 (3.2) Ca3 (1.8) Ca4 (3.2) Fu3 (1.8) Fu4 (3.2) Ba3 (1.8) Ba4 (3.2)
Om5 (5.0) Ca5 (5.0) Fu5 (5.0) Ba5 (5.0) Fu2
(0.8)
Ba2 (0.8)
Ba1 (3.2)
Structured Enriched
Enrichment trajectory
Enrichment index
Figure 2 Functional groups of soil nematodes characterized by trophic group and life history characteristics. Groups belonging to basal, enriched, or structured food webs are included and their weightings for calculation of structure and enrichment indices indicated. Reprinted from Ferriset al. (2001), with permission from Elsevier.
In another study, Lianget al. (2005) observed reduction in the CI following fertilization with urea, associated with increased NO3and NH4levels; how- ever, the slow-release urea formulation resulted in a higher value for the SI, indicating greater trophic diversity. In a comparison of long-term organic, low-input, and conventional management systems, Berkelmans et al. (2003) observed that the organic and low-input systems, relative to the conventional system, were frequently associated with higher enrichment and SI, indicating higher fertility and greater trophic structure, and lower basal and channel indices, reflecting reduced abundance of opportunistic nematodes and rapid nutrient cycling through the bacterial pathway of the soil food web. Ferris et al. (2004) manipulated the trophic structure of nematode communities (and presumably, other microbial feeders) through a combination of fall irrigation and carbon input, following which they observed greater nitrogen mineralization in the subsequent cropping season.
The type of amendment used will play a role in determining the overall effect on nutrient availability. Ferris and Matute (2003) observed structural and functional succession of the nematode community in response to sub- strates of differing C/N ratios. The EI declined over time at a rate regardless of the substrate added. Progression toward fungal domination of energy flow was faster for wheat straw (C/N ẳ 75.9) than for alfalfa (C/N ẳ 10.6), but not faster than for compost (C/Nẳ10.6), indicating that factors in addition to C/N are also important. There was also a succession from enrichment opportunist bacteriovores to general opportunist bacteriovores, but the rate of succession did not differ among the types of amendments (Ferris and Matute, 2003).
Other studies have incorporated the nematode faunal analysis concept into estimates of soil biodiversity in grasslands and pastures, the advantage
Table 1 Soil nutrient status and food web condition inferred from combined calculation of nematode community structure and enrichment Indicesa
General
diagnosis Quadrat A Quadrat B Quadrat C Quadrat D
Disturbance High Low to
moderate
Undisturbed Stressed Enrichment N-enriched N-enriched Moderate Depleted Decomposition
channels
Bacterial Balanced Fungal Fungal
C:N ratio Low Low Moderate to
high
High Food web
condition
Disturbed Maturing Structured Degraded
a Quadrats refer to those presented in Fig. 2. Reprinted from Ferriset al. (2001), with permission from Elsevier.
being that functional components of the ecosystem are also measured with potential implications for nutrient cycling and grassland productivity. For example, Zolda (2006) studied the nematode fauna of grazed and ungrazed grasslands in Austria, Stirling and Lodge (2005) estimated the relationships among climatic and plant species factors and nematode communities in Australian pastures, Bell et al. (2005) studied the nematode fauna of New Zealand tussocks, and De Deynet al. (2004) employed the nematode faunal analysis concept to address the effects of plant diversity on nematode taxonomic and functional diversity.
Hoeksemaet al. (2000) and Sonnemann and Wolters (2005) used the MI in their evaluations of the effects of elevated CO2on nematode community structure. Hoeksema observed an increase in the MI associated with ele- vated CO2in a low-N soil, indicating greater abundance of slower-growing nematode taxa; however, this result was not observed in the high-N soil, nor in the study by Sonnemann and Wolters (2005).
Nematode faunal analyses suggest that nematode communities are quite susceptible to disturbance. For example, Berkelmanset al. (2003) observed that 1 year of a common crop and tillage undid the effects of several years of divergent management practices (organic/low input/conventional). How- ever, some analyses suggest that nematode communities are also resilient to some disturbances. Wanget al. (2006a) observed only short-term effects of solarization or cowpea cover cropping on the SI, disappearing by the end of the experiment (5–6 months); methyl bromide fumigation, however, had persistent effects. Wanget al. (2004) observed little difference in the trophic structure of nematode fauna when comparing untilled plots versus plots undergoing multiple roto-tilling events for 25 years; the tilled plots had been left fallow for 1.5 years prior to sampling, leaving the possibility open that the nematode community recovered quickly once frequent manual disturbance was removed from the system. The time required to recover from disturbance provides additional information regarding ecosystem recovery and should be a focus of future research.
Further research should improve the utility and sensitivity of nematode faunal analysis. Debate continues regarding the placement of taxa into c–p groups (Bongers, 1990) and the generalities of genera and family-level resolution of trophic groups (Yeates et al., 1993). Both are based largely on observations of nematode behavior on agar media, which may not be representative of behavior in nature. Tylenchid nematodes, classified as plant-, algal-, and lichen feeders but possibly also fungal feeders (Yeates et al., 1993), can constitute 30% or more of a sample (Ferris and Bongers, 2006). Furthermore, an evaluation of nematode community indices in three different ecosystem types (wetland, forest, and agricultural) indicated that the indices were differentially sensitive to disturbance in the different ecosystems and that variance within community composition at the genus level within families was more sensitive than the community indices to ecosystem type and disturbance (Neheret al., 2005). Fiscus and Neher
(2002) used multivariate statistical techniques to evaluate the sensitivity of nematode taxa to particular agricultural disturbances, suggesting that indi- vidual analyses could be tailored to have greater sensitivity by selecting particular taxa relevant to the disturbance(s) under study.
4.2. Modeling food web dynamics 4.2.1. Theory
The modeling approach to studying food webs highlights properties of the system emerging from the individual interactions occurring within. Early models focused on connectivity food webs, in which linkages between two interacting groups indicate where trophic interactions occur but all linkages are assigned equal weight. Models by May (1972, 1973) arrived at the conclusion that complex food webs (i.e., those containing many interacting species) are less likely to be stable than simple webs; increases in species richness (S) must be accompanied by a decrease in either connectance,
C ẳ L
S2 ð5ị
where Lis the proportion of all possible linkages that are realized, or the average strength of the interactions (per capita effect of one species on another) occurring in the system. May observed, however, that the presence of com- partments in food webs, within which species interact readily with each other but very little with species in other compartments, increased the feasibility of constructing large food webs (May, 1972, 1973). Lower richness within individual compartments allowed for more and stronger interactions among species without risking instability.
In the 1980s and early 1990s, soil ecologists conducted surveys of soil food webs whereby they represented interactions as quantifiable flows of material cascading through the web. These surveys, and subsequent model- ing exercises, are built upon available descriptions of connectivity webs (Fig. 1) by assigning weights to these linkages. Weights represented either the amount of energy present within and moving between pools (energy webs), or the per capita effects of one functional group on another (func- tional or interaction strength webs). From these models it became clear that, in determining the stability of food webs, the number of interactions within a food web is less important than how those interactions are structured. As a result, these researchers were capable of addressing questions related to the emergent properties of the food web, emphasizing properties associated with trophic diversity and structure (how many trophic levels/linkages are supported at any particular level of productivity?) and ecosystem sta- bility (how resistant/resilient is food web structure to environmental perturbation?).
Huntet al. (1987) derived equations for modeling the flux of energy, as carbon and nitrogen, through food webs. For the consumption rate,F, of a consumer,j
Fj ẳ ðdjBjỵPjị ajpj
ð6ị
where P and d represent the predatory and nonpredatory death rates, respectively, B represents biomass, and aand p represent the assimilation (ingested and not lost in feces) and production (retained as biomass) effi- ciencies, respectively, of the consumer. For consumers that feed on more than one prey, the consumption rate of prey, i, is a function of the preference,w, and biomass ofirelative to that of all prey,k, so that
Fij ẳ wijBi Pn
kẳ1wkjBk
Fj ð7ị Energy webs are particularly useful for estimating how sensitive the length and reticulation of the food web is to the amount of energy entering and moving within the web. Moore and Hunt (1988) demonstrated that energy channeled through a soil food web largely via compartments (roots, bacteria, and fungi as basal resources in each pathway) with little movement of energy between pathways at intermediate trophic levels. The authors’
analysis of published connectivity webs of trophic relationships showed that the number of energy pathways (resource richness) in a food web correlated positively with the richness of consumers and negatively with connectance.
This result supports resource compartmentation, reducing the proportion of species that directly interact, as a mechanism allowing stable species rich food webs to persist (May, 1972, 1973; Moore and Hunt, 1988).
Functional webs represent the dynamic effects of trophic interactions, with a change in abundance at one trophic level eliciting a quantifiable change at another. DeAngelis (1992) and Moore et al. (1993) derived the dynamics of producer, consumer, and detritus density. Biomass density,X, of producerichanges over time in relation to growth (at both the individual and population levels combined) and consumption, such that
dXi
dt
ẳriXiXn
jẳ1
cijXiXj ð8ị whererrepresents the specific growth rate of the producer andcrepresents the consumption coefficient for consumer j. Biomass density of detritus, d,
changes over time in relation to the amount of detritus entering the system from allochthonous inputs, autochthonous inputs due to unassimilated and unconsumed prey, and autochthonous inputs due to nonpredatory death of consumers, and the consumption of detritus, such that
dXd dt
ẳRdỵXn
iẳ1
Xn
jẳ1
ðð1aiịcjiXjXiị ỵXn
iẳ1
diXiXn
jẳ1
cdjXdXj
ð9ị where Rd represents the rate of allochthonous input. Biomass density of consumer,j, changes over time in relation to decline due to nonpredatory death, decline due to being consumed by n consumers, l, and growth associated with consumption, such that
dXj
dt ẳ djXjXn
lẳ1
cjlXiXlþXn
iẳ1
ajpjcijXiXj ð10ị
functional webs are particularly useful for estimating how perturbations of the web, such as the removal of one or more trophic groups, will affect the abundance of other trophic groups.
de Ruiteret al. (1995) linked the functional and energy web models by assuming that feeding rates (Fij) and biomass (Bi,Bj) in the energy model equal consumption rates (cijXiXj) and biomass density (Xi,Xj) in the func- tional model, respectively, in order to estimate the consumption coefficient
cij ẳ Fij BiBj
ð11ị
from nutrient flux data and estimate interaction strengths, a, as the per capita effects of consumerjon preyi,
aij ẳ Fij
Bj ð12ị
and vice versa,
ajiẳ ajpjFij Bi
ð13ị
in soil food webs. Strong interactions occur when per capita effects of consumers on prey or vice versa are large. In an analysis of several soil food webs by de Ruiteret al. (1995), complex interactions, both strong and weak, had strong effects on stability. Varying the interaction strengths of most pathways in the root-pathway and at intermediate and higher trophic levels (secondary consumer and up) had strong impacts on food web stability, while varying the strengths of interactions among fungi or bacteria and their consumers had very little impact on stability.
Rooneyet al. (2006) further linked the models by relating interaction strength to the speed of energy flow, v, represented by the rate that consumer biomass is turned over, such that
vj ẳ Pn
iẳ1aij
Bj
ð14ị
for energy flux into consumerjand vj ẳ Pn
lẳ1alj Bj
ỵdj ð15ị
for energy flux out of consumerj, suggesting that fast energy flux webs are composed of strong interactions and slow energy flux webs contain weak interactions. Rooneyet al. (2006) observed similar asymmetrical partition- ing of energy to pathways in six marine (pelagic vs benthic) and terrestrial (bacterial vs fungal) food webs, with higher-order consumers deriving energy from both pathways and coupling the pathways. By varying the energy flowing through one pathway relative to a second constant pathway, they observed that stability (associated with both resilience and resistance) was lowest when the two were equal and increased with increasing differ- ence between the variable and constant pathways. Temporal asynchrony in the flux of energy through different pathways means that consumers at higher trophic levels, where the soil food web is much more reticulate, may be less likely to encounter highly variable resource availability (McCannet al., 2005).
Moore et al. (2005) modeled the stability of a two-channel food web, containing a single resource base, two primary consumers, two secondary consumers, and a single top predator and using parameters from the Colorado shortgrass steppe food web (Huntet al., 1987), and varied the proportion of energy partitioned to each pathway; they found that the system demonstrated stability when 20–60% of energy was partitioned to the fast (bacterial) pathway, the optimum being40%. Simulated patterns of allocation outside of this range result in unstable dynamics in food web structure. Stability is