A heterogeneous distribution of parasitism risk has been proclaimed as an indicator of the stability of a host-parasitoid interaction. To date, there have been almost no experimental field studies that have attempted to assess the mechanisms that generate this heterogeneity and the environmental factors that influence it. In this study, I examined the interaction between a planthopper Prokelisia crocea (Hemiptera: Delphacidae) and its egg parasitoid Anagrus columbi (Hymenoptera: Mymaridae) that are distributed among discrete patches of prairie cordgrass, Spartina pectinata. In particular, I examined how patch geography (size or isolation of a patch) and host distribution within a patch (edge versus interior) influenced the density distribution of adult parasitoids, oviposition behavior, the proportion of hosts parasitized, and an index of the aggregation of parasitism (CV2) and adult parasitoids (1/knb) both within and among cordgrass patches. High values of CV2 and 1/knb indicate strongly aggregated distributions.
Based on a census that spanned 6 planthopper generations and 7-12 prairie sites, I found that the distribution of parasitism was strongly aggregated among cordgrass leaves within a patch (CV2 = 3.58), but weakly aggregated among cordgrass patches (CV2 = 0.58). Parasitism was also spatially and temporally density independent. To determine what influences the distribution of parasitism risk, I collected data on A. columbi colonization and oviposition behavior in 26 discrete cordgrass patches that had planthopper egg clutches added to them and all sources of A. columbi removed. I found that the number of immigrants significantly increased with patch size and decreased with patch isolation (a function of the distance to, and size of, neighboring patches). Patch size had no influence on the per-capita hosts parasitized per leaf, but there was a significant two-fold increase in per-capita attacks from the least to the most isolated patches (14 to 28 hosts parasitized). In a separate experimental study, I found that this oviposition response was affected primarily by the distance between source and target patches. The cause for this distance-dependent response was likely due to either an optimal ovipositional response to dispersal distance or a positive correlation in the A. columbi population between oviposition rate and dispersal ability. Regardless of its cause, increased ovipositions with dispersal distance may significantly increase patch-occupancy rates and facilitate the spatial spread of a parasitoid.
In accordance with census data, substantial within-patch aggregation of parasitism (CV2 = 1.63) did not translate into strong among-patch aggregation (CV2 = 0.13) for the 26 experimental patches. Searching adult parasitoids were randomly distributed within and among patches, and thus did not explain the high CV2 within patches. Interestingly, the aggregation of parasitism risk within a patch was significantly negatively correlated with patch size and positively correlated with patch isolation. The distribution of parasitism risk could be divided into two general classes, a within-parasitoid component and an among-parasitoid component. The within-parasitoid component was attributable to individuals engaging in multiple ovipositions within a cordgrass leaf and the distance-dependent oviposition response. The oviposition response was likely the causal agent behind the variation in CV2 with patch size and isolation. This source of aggregation is expected to contribute nothing to host-parasitoid stability. The among parasitoid component of aggregation appears to have been due to heterogeneity in the vulnerability of hosts and an edge effect (60% more heterogeneous distribution of parasitism along the patch edge than the patch interior). In theory, the among-parasitoid component of aggregation can contribute to the stability of a host-parasitoid interaction. Consequently, a change in landscape structure that leads to an increase in cordgrass edge habitat may promote a more stable host-parasitoid interaction.