Communities of insect herbivores are thought to be structured mainly by indirect processes mediated by shared natural enemies, such as apparent competition. In host–parasitoid interaction networks, overlap in natural enemy communities between any pair of host species depends on the realized niches of parasitoids, which ultimately depend on the foraging decisions of individuals. Optimal foraging theory predicts that egg-limited parasitoid females should reject small hosts in favour of future opportunities to oviposit in larger hosts, while time-limited parasitoids are expected to optimize oviposition rate regardless of host size. The degree to which parasitoids are time- or egg-limited depends in part on weather conditions, as this determines the proportion of an individual's lifespan that is available to foraging. Using a 10-year time series of monthly quantitative host–parasitoid webs, we present evidence for host-size-based electivity and sex allocation in the common secondary parasitoid Asaphes vulgaris. We argue that this electivity leads to body-size-dependent asymmetry in apparent competition among hosts and we discuss how changing weather patterns, as a result of climate change, may impact foraging behaviour and thereby the size-structure and dynamics of host–parasitoid indirect interaction networks.
Previous studies of the effects of climate change have generally focused on the responses of species in isolation in spite of the fact that interactions between species are a significant component of the response of communities to climate change . For instance, among three Drosophila fly species, simulated temperature changes resulted in the extinction of the least-adapted species through direct competition for resources; however, the presence of a single, shared, generalist natural enemy (the parasitoid wasp, Leptopilina boulardi) ameliorated competitive pressure, allowing the three fly species to coexist . The indirect effects of climate change will not only alter species physiology, but the way species interact; for example, climatic warming may generally increase disease burden among marine and terrestrial organisms , with consequences for community structure and dynamics. A general way in which climate change may affect ecological communities indirectly is through changes in the body-size structure, which in turn affects interactions among species . Various aspects of this are covered in this theme issue and here we focus on body-size-dependent indirect interactions in insect host–parasitoid networks (HPNs).
While direct interactions such as predation have a clear impact upon population dynamics, their consequences may further affect other members of a community. These complex, further effects are known as indirect interactions: the impact of one organism/species on another mediated through the actions of a third. Natural-enemy-mediated indirect interactions are thought to be a particularly important structuring force in herbivorous insect communities due to the typically low level of interspecies competition for resources in these systems and the strongly linked population dynamics between the herbivores and their parasitoid natural enemies . There is a particular need to consider climate change impacts on insect host–parasitoid communities, because of both their economic importance in agriculture and their prevalence within natural ecosystems [6,7]. For example, estimates suggest that impacts on the functionality of parasitoids as control agents of crop pests could cost the United States' agricultural industry up to $20 billion annually in crop damage .
One type of indirect interaction that has been hypothesized as being a particularly important structuring force in HPNs is apparent competition . This process is defined as a negative indirect effect between two species that share a natural enemy, and is so named because, like resource competition, it can lead to competitive exclusion . Indirect interactions may be density- (effects are driven by changes in abundance of the third party) or trait-mediated (effects are driven by changes in the morphology or behaviour of the third party) [7,10]. Apparent competition can lead to the exclusion of a species where one host species is affected by parasitism disproportionately more than the other species. This ‘parasitism asymmetry’ can result from the foraging decisions of a natural enemy; for instance, preference for one host over another by a parasitoid may result in the preferred host experiencing higher levels of mortality than the non-preferred host . Conversely, if the parasitoid exhibits no preference between hosts, the hosts may experience apparent mutualism; whereby, for both host species, the presence of the other dilutes parasitism-induced mortality, at least in the short term [9,11,12].
Which species in an ecological community can engage in apparent competition is determined by the host ranges of the concurrent parasitoid species, as this determines whether host species can share natural enemy populations. Within these limitations, however, the strength and degree of asymmetry of indirect interactions between two species will strongly depend on whether parasitoids attack the different hosts indiscriminately or exhibit a degree of preference. A likely important host trait in this context is body size. In solitary parasitoids, where only one individual develops on each host, there is a strong correlation between host body size and offspring fitness [13–15]. Further, it has been shown in a field experiment that populations of a host species that grow to a larger body size than nearby conspecific populations suffer higher parasitism rates from generalist parasitoids . If parasitoids do account for host body size when foraging, we would expect the realized niche of a parasitoid population to depend on the body size distribution of the available host species, with larger host species being over-represented within the realized niche relative to their abundance.
Another potentially important aspect of host body-size effects on parasitoid-mediated apparent competition is parasitoid sex allocation. Parasitoid sex-ratio is an important component of host–parasitoid population dynamics, as only female offspring go on to induce host mortality in the next generation . Parasitoid wasps, like all Hymenoptera, have a haplo-diploid sex determination system whereby males develop from unfertilized eggs and females from fertilized eggs. At each oviposition event, a female parasitoid can elect to lay a fertilized or an unfertilized egg. Because the effect of body size on fitness is typically greater for females, solitary parasitoids are more likely to lay fertilized eggs in larger hosts [16,18]. Sex ratio allocation behaviour could therefore mediate asymmetric apparent competition when the smaller host species receives mostly male and the larger mostly female parasitoid eggs. In this case, the larger host would be the main source of the population of female parasitoids attacking the small host, while the presence of the smaller host would either have no effect on the larger host, or even have a positive indirect effect through a dilution of parasitism .
Elective foraging behaviours, such as those described above, can determine the distribution of interaction strengths in HPNs. In predator–prey food webs, the size-structured distribution of weak and strong links between species is known to be an important aspect of ecosystem stability . However, the understanding of the role of size-structuring in predator–prey interactions is much deeper than that for HPNs, and there are fundamental differences between these systems . For example, the relationship between trophic level and body-size in HPNs is inverse to that of predator–prey networks . Further, unlike in predator–prey systems, handling time and capture rate are largely independent of body size in host–parasitoid interactions . The two types of foraging electivity described earlier may, however, still drive a size-structured distribution of the weak and strong links that drive direct and indirect interactions within host parasitoid networks. Therefore, it is likely that the various degrees of foraging electivity exhibited by parasitoids are an important component of network structure and stability [7,22].
The degree of electivity exhibited by parasitoids depends on whether individuals are time- or egg-limited. Egg-limited parasitoids run out of eggs while still having host available to attack, while time-limited parasitoids die while still having eggs to allocate to hosts [19,23,24]. Egg-limited parasitoids should exhibit strong preferences for higher quality hosts, as they experience a greater reduction in future fitness gains through the use of every egg than time-limited parasitoids. Parasitoids that are time-limited owing to shorter lifespans or to short, environmentally determined, foraging windows incur low costs to future fitness when attacking suboptimal hosts and are less likely to forage electively because these individuals are expected to optimize oviposition rate, not offspring fitness .
While the effects of climate change on parasitoid physiology are relatively well documented, there have been few considerations of how parasitoid behaviour will be altered [25–29]. Weather conditions can have strong effects on parasitoid lifespan and foraging windows, thereby influencing the degree to which individuals are egg- or time-limited [23,26]. We predict that climate change will lead to a shift towards time-limitation through constraining foraging windows (as a result of increased precipitation ) and reduced lifespans (as a result of increased temperature ). Such a shift should lead to less elective foraging and a breakdown of the size structure of host–parasitoid indirect interaction networks . Importantly, for the ordinarily highly egg-limited secondary parasitoids that are the focus of this study, a reduction in foraging electivity may result in a loss of the size-structured distribution of parasitism asymmetry that drives the indirect interactions that are considered important determinants of the dynamics of these systems.
In order to assess whether these considerations have a significant effect, patterns of parasitoid preference must be ascertained at the network level [10,31,32]. Here, we use a 10-year time series consisting of monthly host abundances and parasitism rates by a single parasitoid species to demonstrate that parasitoid electivity does indeed play a role in structuring the host–parasitoid interaction network. We show that large hosts are significantly over-represented within the realized niche of the parasitoid at any point in time and that there is a strong female-biased sex allocation in larger hosts. Furthermore, we show that the degree of parasitoid electivity is not constant, indicating an environmental impact on community structure and dynamics mediated by effects on parasitoid foraging behaviour.
(a) Study species and sample collection
We focused on the wasp Asaphes vulgaris (Hymenoptera: Pteromalidae). Asaphes vulgaris is a solitary (only one larva can develop per host individual) secondary parasitoid; i.e. the hosts of A. vulgaris are primary parasitoids that themselves parasitize sap-feeding aphids. The larvae of these primary parasitoids develop inside a living, growing aphid until the aphid reaches either its penultimate or final (adult) instar. At this stage, the primary parasitoid larvae kills its aphid-host and spins its cocoon inside the host's exoskeleton, resulting in what is known as a ‘mummy’. Asaphes vulgaris oviposits in these aphid mummies, placing its egg on the primary parasitoid (pre-) pupa. It is a common and abundant species, known to attack a wide range of primary parasitoid species through a wide range of aphid hosts that span a wide range of body sizes . As such, it is particularly likely to mediate indirect population interactions among its hosts species and has the opportunity to forage selectively .
The data used to determine host preference and sex ratio allocation choices made by secondary parasitoid wasps came from a long-term study of an aphid parasitoid food web [33,35]. Briefly, surveys were conducted between April and October, from 1994 through to 2003, within a single, mesohydric meadow (approx. 18 000 m2 in size) in Silwood Park, Berkshire, UK. Each month, densities of the mummies of each aphid species were estimated. Samples of 400 mummies of each aphid species (if present) were taken to the laboratory to rear out primary and secondary parasitoids, in order to obtain data on host–parasitoid associations.
(b) Host-quality metrics
Preliminary laboratory experiments with fresh Aphidius ervi (primary parasitoid) via Acyrthosiphon pisum (aphid) mummies (N = 30) showed that an approximation of mummy volume (calculated as mummy length × width × width) produced a strong linear relationship with mummy mass (corresponding coefficients = 0.90). Mummy length and width were measured, using a Leica M165C microscope and its associated image analysis software ‘Leica Application Suite v. 3’ from all of the samples of A. vulgaris mummies collected in the field between 1994 and 1999, and measurement followed simple guidelines using easily identified anatomical structures. The measurements were used to calculate mean mummy sizes for each aphid species, irrespective of the primary parasitoid species.
Different aphid species are abundant at different times of the year and it is likely that host choice occurs according to the relative sizes of the hosts available at any given time, rather than their absolute sizes . Therefore, in order to compare the sizes of host species present during different months, a host-quality metric called ‘size difference’ was calculated: 2.9where Sspecies was the mean size of the mummies of the focal aphid species, and Smonth was the mean of the mummy sizes of all the aphid species present in the field during a given month. Size difference was calculated for each aphid species for every month that it was present in the ecosystem, thus allowing a comparative metric of host-species quality across time.
(c) Statistical analysis of host-preference
We quantified the preference for each host species in each month using the Strauss Linear Index (SLI), which is defined as 2.9where, for each month within the experimental period, proportion in diet (PoD) was the density of A. vulgaris wasps emerging from the mummies of an aphid species as a proportion of total A. vulgaris density; and proportion in environment (PoE) was the mummy density of the aphid species as proportion of total mummy density (i.e. the sum of the mummy densities of every aphid species that can potentially be attacked) . The SLI has a value of zero for a species when its proportion in the diet matches its proportion in the environment. Values greater than zero indicate over-representation in the diet (i.e. preference), and values below zero indicate under-representation. SLI values were calculated for each host species during each month it was present in the field along with at least one other potential host. Only aphid species that appeared in the A. vulgaris host-range at least once were considered in the calculation of SLI values. Further, values were not calculated for a few very rare aphid species for which no reliable size data could be collected.
A linear mixed effect (LME) modelling analysis was carried out to determine whether variation in monthly SLI values obtained for each aphid could be explained by the relative sizes of the aphid mummies, while including year and month as a nested random variable.
(d) Statistical analysis of sex ratio allocation
The number of male and female A. vulgaris that eclosed from each host species were scored for each month. A binomial GLM was used to test for a relationship between the sex ratios of parasitoids reared from the mummies of different aphid species and the aphid species' corresponding size difference value for that month. A quasi-binomial distribution was used to account for overdispersion.
(e) Measures of changing electivity through the year
In order to test for an effect of environmental conditions on foraging electivity, we calculated the absolute deviance from random foraging exhibited by A. vulgaris during each month as the sum of the absolute SLI values of all aphid species present in the sample. This was the only measure of deviance that was not inherently associated with the number of aphid species present during each month (see the electronic supplementary material). We conducted a GLM analysis with sum deviance as the dependent variable, and with the following explanatory variables: season of the year (spring: April and May; summer: June, July and August; autumn: September and October), a metric of monthly competition for hosts (taken as the total mummy density divided by the estimated total density for A. vulgaris wasps for each month), and size range (the difference in volume between the smallest and the largest aphid species present during the month).
Our LME analysis indicated a significant positive relationship ( p < 0.001) between the relative mummy size of an aphid species and its corresponding SLI value for that month (figure 1).
(b) Sex ratio allocation
The sex ratio of A. vulgaris wasps eclosing from the mummies of an aphid species during a given month was significantly associated with the corresponding size difference value for that aphid species (F1,118 = 89.772, *p < 0.001), where relatively smaller species hosted more male secondary parasitoids (figure 2).
(c) Seasonal change in electivity
As a measure of changing electivity across time, we calculated the variable ‘sum deviance’ (the sum of the absolute difference from zero found for the SLI of every aphid species present during a month). According to our GLM analysis, variation in monthly sum deviance could not be explained by any of the measured biotic or abiotic factors that would be expected to alter electivity, nor the interactions between them: host availability per competitor (F1,27 = 1.016, p = 0.322; figure 3a), the size range of available aphid mummies (F1,27 = 0.277, p = 0.604; figure 3b) and season (F2,27 = 0.936, p = 0.405; figure 3c).
We predicted that the realized niche of a parasitoid at any point in time would not simply reflect the relative densities of the host species within its fundamental niche, but would be biased towards larger host species. Our analysis provided strong support that this is indeed the case for A. vulgaris. We made a further prediction that parasitoids would show a sex allocation bias placing more females in hosts that were relatively larger. Our analysis of A. vulgaris sex ratio across its host range (figure 2) does indeed show that this species predominately lays female eggs in the largest hosts while the smallest hosts predominantly produce male parasitoids. The numerical contribution to the female parasitoid population per host killed therefore increases with host size, as predicted.
Both these apparent host-size effects on parasitoid foraging behaviour have implications for natural-enemy-mediated indirect interactions among the hosts in the community. First, as our results show, the primary parasitoids of larger aphid species suffer higher mortality rates due to parasitism, presumably due to secondary parasitoid preference for higher quality hosts. This could lead to the negative effects of apparent competition to be biased towards larger host species. This could, however, be countered by the sex-ratio effect. Because it is only female adult parasitoids that ultimately cause host mortality, small hosts that produce mostly male parasitoids contribute little to the effective parasitoid population and therefore have reduced potential to indirectly affect mortality of the larger hosts [17,19]. In fact, in the short term, they may even have a positive dilution effect. In return, the majority of parasitoid-induced mortality of the smaller hosts is likely to be caused by females derived from larger hosts, leading us to predict asymmetric, negative indirect effects of larger hosts on smaller hosts. Therefore, both these effects lead us to suspect that the size-structure of the host community could significantly affect the distribution of indirect interaction strengths and thereby community dynamics. At this stage, it is difficult to predict what the outcome of these combined effects is for size-structured indirect interactions, and to answer that question will require a dynamic modelling approach and experimental manipulation .
For the effects of climate change in this context, we predicted that environmental conditions would affect the degree of electivity shown by A. vulgaris. Abiotic factors such as temperature [25,26], precipitation [29,30] and atmospheric CO2 concentration  have been shown to alter parasitoid lifespan, egg-load and foraging efficiency. Weather conditions can affect the cost of each oviposition event with regard to future reproductive success and, therefore, whether parasitoid females would be predicted to optimize oviposition rate (low electivity) or offspring quality (high electivity) . While our data indicate that the absolute degree of electivity exhibited by foraging A. vulgaris varied considerably across the time frame of this study, we could not find any definitive effects of the biotic and abiotic factors that have been predicted to drive host-choice behaviour: such as competition, the variety in size of available hosts and the season of the year. We expect that the temporal resolution of our analysis was too crude to pick up the variation in environmental effects and that experimentation is required to test hypotheses on climate effects on parasitoid electivity.
However, importantly for the theme of this special issue , our results suggest that the degree of electivity exhibited by foraging secondary parasitoids is not constant and cannot be explained merely by the variation in host quality at any given time, which suggests a possible effect of the external environment on parasitoid foraging behaviour. Owing to the evidence for a size-structured distribution of parasitism asymmetry driven by the foraging behaviour of secondary parasitoid wasps, we can suggest that climate change scenarios, particularly those that hinder foraging success, either by reducing foraging time (increased precipitation or shorter lifespan at higher temperatures) or by reducing host availability (e.g. drought), may result in an increase in non-elective foraging behaviour [19,23,24]. The loss of electivity by foraging secondary parasitoids could reduce the asymmetry of parasitism mortality between primary parasitoids, with knock-on effects on their aphid hosts. This would most likely alter the distribution of strong and weak links within HPNs, with a shift to greater evenness of link strength. Interestingly, studies in this issue on both theoretical predator–prey systems  and empirical litter feeding arthropod communities  predict changes in diet breadth as a result of changes in foraging behaviour in response to climate change, indicating the potential generality of these effects. While our predictions are based on weather variability, de Sassi et al.  demonstrate that changes in mean temperature and nitrogen deposition can also lead to changes in electivity. Both these factors affect host density, which should lead to a higher host encounter rate for parasitoids, making it more likely that they will be egg-limited and therefore more elective. Indeed, de Sassi et al. found stronger bias towards high quality hosts under these conditions. The implications of these various effects for network dynamics are difficult to predict at this stage, especially as this also has to take into account the process of sex-ratio allocation demonstrated here, and there is an urgent need to study these effects in multi-species host–parasitoid models.
A major strength of our study is that it uses a comprehensive dataset of a large natural community that was sampled quantitatively every month that insects were present for ten years. Field data of this quality are very rare and, in the context of parasitoid foraging behaviour in the field, so far non-existent. An important limitation, however, is that we infer parasitoid preference from the frequency of emerging offspring. An alternative explanation for our observation that A. vulgaris emerges more often than expected from larger hosts is that larval survival increases with host size. This could lead to the same pattern even if host size played no role in the oviposition and foraging decisions of parasitoid females. We think it is unlikely that this is the case, and behavioural studies of host acceptance rates in other parasitoid species have shown that these depend on the relative quality of the available host species [14,41]. Similar observational studies of secondary parasitoid foraging behaviour and larval survival will allow us to distinguish between these mechanisms in our study system. It is also worth noting that we assumed a purely density-dependent host-choice null model, which would be violated if not all hosts are consumed with exactly the same functional response . While this is likely to be an issue for predator–prey systems, it should be remembered that host–parasitoid interactions do not follow the same allometric relationships between body-size, handling time and capture rate that can be used to define alternative null models for predator–prey interactions. Further studies elucidating the role of allometry in parasitoid foraging decisions are required before we can apply nonlinear null-models to HPNs.
The direct effects of climate on parasitoid physiology are well studied [7,26], but little attention has been paid to the effect of abiotic conditions on parasitoid foraging behaviour and the implications for community-level interactions . Building on the work of Rosenheim [23,24], we have previously developed a framework linking the life-history characteristics of parasitoids to optimal foraging behaviour on the basis of the concepts of reproductive success being affected by either egg- or time-limitation . The analyses presented here suggest that the size structure of host communities, and the effect that this has on indirect interactions between species, may form an important further element in our mechanistic understanding of these complex systems and our ability to predict how they will be affected by climate change.
We thank the anonymous reviewers and the editor for insightful comments on an earlier draft of this paper. Joseph Faulks provided technical assistance. This project was funded by a Linnean Society of London SynTax grant to F.J.F.v.V. and D.C.H.'s studentship, funded by the UK Natural Environment Research Council (NE/I528326/1). The participation of F.J.F.v.V. at the European Science Foundation SIZEMIC workshop in Hamburg was supported by the German Research Foundation (JA 1726/3–1) and the Cluster of Excellence CliSAP (EXC177), University of Hamburg, funded through the DFG.
One contribution of 17 to a Theme Issue ‘Climate change in size-structured ecosystems’.
- This journal is © 2012 The Royal Society