- Top of page
- Materials and methods
Theory on intraguild predation (IGP) predicts possible coexistence of all three species only if the intermediate consumer (intraguild prey, IG-prey hereafter) is superior to the other consumer (IG-predator hereafter) in competing for the shared resource (Holt & Polis 1997). If so, the equilibrium community structure varies along a productivity gradient (Polis & Holt 1992; Mylius et al. 2001; Holt & Polis 1997; Diehl & Feißel 2000). At low productivity levels, only the IG-prey persists with the resource because it is the superior competitor and because the IG-prey is not abundant enough to sustain a population of IG-predators. At high productivity, the IG-prey is always excluded, independent of initial conditions. At intermediate productivity levels, predictions depend on the type of functional response (Holt & Polis 1997). As our test animals have type II functional responses (Sabelis 1992; van Rijn et al. 2005), we present predictions only of models with such functional reponses. These models show bi-stability at intermediate productivity levels, with either all three species coexisting or the IG-prey being excluded (Holt & Polis 1997; Mylius et al. 2001).
Empirical tests of these patterns of coexistence and exclusion at different productivity levels are scarce. Experiments with aquatic systems involving protists showed that community structure indeed depends on productivity, although not always as predicted by theory (Lawler & Morin 1993; Morin 1999; Diehl & Feißel 2000, 2001). These deviations from theory were attributed to the limited parameter space for coexistence in this particular experimental system, stochastic extinction during the phase of transient dynamics or accumulation of toxic waste products. Furthermore, the basal resource consisted of several species, which could give rise to complex dynamics at the resource level and thereby could affect higher trophic levels (Diehl & Feißel 2001; Janssen et al. 2007).
Here, we test IGP theory in a terrestrial system composed of two predatory mite species: Iphiseius degenerans (Berlese) and Neoseiulus cucumeris (Oudemans). These two species are used currently as biological control agents of thrips in greenhouses (van Houten & van Stratum 1995), and co-occur in the Mediterranean area (De Moraes et al. 2004). In this system, I. degenerans, the IG-predator, preys on all mobile juvenile stages of N. cucumeris, the IG-prey, whereas eggs and adults of N. cucumeris are invulnerable, and both species are capable of feeding and reproducing on several types of pollen (van Rijn & Tanigoshi 1999). In the experiments presented here, we used cattail pollen (Typha sp.) as the shared resource. The amount of pollen was varied to create different productivity levels and it was supplied at regular intervals to simulate resource dynamics.
We first tested whether our system matched the conditions under which theory predicts such patterns of coexistence and exclusion, i.e. that the IG-prey (N. cucumeris) is superior to the IG-predator (I. degenerans) at exploitative competition for the shared resource (pollen). Secondly, we tested whether intraguild predation occurred in presence of the shared resource. This is carried out because many species engage in intraguild predation only the in absence of other food sources, which reduces the system effectively to one of resource competition. Finally, we conducted competition experiments at different resource densities to assess whether the patterns of exclusion or coexistence in our experimental system matched those predicted by IGP theory.
- Top of page
- Materials and methods
Although our system meets the assumptions of theoretical models of intraguild predation (i.e. the IG-prey is the superior competitor for the shared resource; Holt & Polis 1997; Diehl & Feißel 2000; Mylius et al. 2001), we failed to find the coexistence and exclusion patterns predicted by theory. In agreement with theory, the IG-predator excluded the IG-prey at the highest resource density, except for one replicate in which the IG-predator was excluded. At intermediate resource levels, at which the IG-predator cannot persist on the resource alone, we would expect either the coexistence of IG-prey and IG-predator or the extinction of the IG-predator (Holt & Polis 1997; Diehl & Feißel 2000; Mylius et al. 2001). Exclusion of the IG-predator was, however, observed in only one of four replicates, whereas the other replicates showed unexpected exclusion of the IG-prey or extinction of both species.
How to explain the fact that we found multiple outcomes of the interaction for all productivity levels? This is unlikely to arise from the existence of alternative equilibria. Although theory predicts such alternative equilibria in systems with IGP at intermediate productivity levels, these predictions do not include the possibility of IG-predators excluding IG-prey at productivity levels that are insufficient for IG-predators to persist.
The pattern found here, that either the IG-predator or the IG-prey excluding each other, is predicted by more recent theory in which the IG-predator shows ontogenetic diet shifts (van de Wolfshaar, de Roos & Persson 2006). These authors show mutual exclusion of the IG-predator and IG-prey, with initial conditions determining which species will be excluded. The region in which this mutual exclusion occurs is largely independent of productivity, a pattern that we also observe in our experiments. However, the IG-predator in our system does not show such ontogenetic diet change (M. Montserrat and A. Janssen, personal observation). Other recent theory shows that coexistence of IG-prey and IG-predators is limited when the two predators are engaged in reciprocal intraguild predation, with IG-prey adults feeding on IG-predator juveniles (HilleRisLambers & Dieckmann 2003). Indeed, predation can be reciprocal in our system, as in many other systems of IGP (cf. Polis et al. 1989). This factor could lead to mutual exclusion, with priority effects determining which species will persist. In disagreement with this latter theory, however, is the fact that the IG-predator in our system is capable of excluding the IG-prey at productivity levels that are insufficient for the IG-predator to persist.
Hence, it is clear that our experimental results do not support model predictions. We suggest three possible explanations for this discrepancy. First, stochastic rather than deterministic dynamics during the initial, transient phase of the interaction are an important determinant of our experimental results. There is no theory on stochastic transient dynamics of systems with intraguild predation (Briggs & Borer 2005). In the system studied here effects of stochasticity are unavoidable; indeed, the densities of mites were rather high relative to natural densities. Perhaps experiments with systems that reach much higher densities, such as protists (Lawler & Morin 1993; Morin 1999; Warren, Law & Weatherby 2003) would be more suitable to study effects of stochasticity.
Secondly, our data are on transient dynamics, whereas model predictions are based on equilibrium dynamics. Indeed, it has been argued that transient dynamics in ecological systems are more likely to be the rule than an exception (Hastings 2004; references therein). Such transient dynamics may be highly dependent on initial conditions. Indeed, there is experimental evidence that initial conditions determine which stable community is reached (Warren et al. 2003).
Thirdly, it is important to realize that the system studied here differs from most theoretical systems because of the stage structure of the populations of IG-predators and IG-prey. This may be crucial because not all stages of the IG-predator feed on all stages of the IG-prey. Models considering stage-structured populations of both IG-prey and IG-predators are still lacking. Also, other aspects of stage structure such as role reversals (Janssen et al. 2002; Magalhães et al. 2005) are still not covered by IGP theory and, indeed, many IG-prey also feed on small IG-predators (Polis et al. 1989). We suspect that the interplay of initial conditions and stage structure of populations is important for an understanding of system dynamics, because different initial conditions and reciprocal IGP are likely to occur in most IGP systems.
We suggest that the outcome of the interaction between populations of the IG-predator and the IG-prey depend upon initial conditions and stage structure of the populations. Indeed, the IG-prey was always eradicated by the IG-predator when the initial numbers of the IG-predator were high (Figs 4f,c, 5a,b). Hence, the high initial IG-predator populations may have killed all IG-prey during the transient phase, even if IG-predators were expected to be absent at equilibrium under a deterministic scenario (Fig. 1). As shown in Figs 1 and 2, a single IG-predator female killed around 12 IG-prey juveniles per day (irrespective of pollen density), whereas the oviposition rate of the IG-prey at low pollen density was in the order of 1·6 eggs per female per day (unpublished data). Together with a high initial density of the IG-predator, this increases the likelihood of eradication of the IG-prey during the initial phase of the interaction. Indeed, when the initial numbers of the IG-predator were low, the IG-prey excluded the IG-predator in most cases (Figs 4a, 5c,d and 6). Furthermore, comparison of the results at intermediate pollen supply rate show that the outcome of the interaction was more variable with a high initial density of the IG-predator (Figs 4 and 6). At high densities, the stage structure of the IG-predator population was more variable, as was the outcome of the interaction, whereas when the density and the stage structure was kept constant (i.e. one adult female, Fig. 6), the outcome of the experiments was always the same.
Three replicates showed different dynamics from other replicates with similar initial conditions, and they require further explanation: one with the IG-prey being excluded by IG-predator even though initial numbers of the IG-predator were low (Fig. 4d), and two with both the IG-predator and IG-prey going extinct simultaneously (Fig. 4e,f). We suspect that these replicates differed in stage structure from the other replicates, perhaps with different densities of invulnerable IG-prey stages (i.e. adults) and non-predatory IG-predator stages (i.e. juveniles). Such stage structure is predicted to affect coexistence and exclusion patterns (Mylius et al. 2001). Unfortunately, our data do not allow further analysis.
In conclusion, our experimental system fulfilled the necessary conditions for the productivity-dependent community shifts that are predicted by theory to be observed, yet our experimental results did not confirm such shifts. In particular, our results suggest that the initial size structure of IG-predator and IG-prey populations may, to a large extent, determine the transient dynamics and patterns of coexistence and exclusion. We suggest that transient dynamics, stochastic events and stage structure of the populations are responsible for this discrepancy and that theory and experiments should focus on this.