Associational resistance (AR) is an emergent property of ecological communities and may play an important role in their assembly and structuring. Gall densities of the midge Asphondylia borrichiae Rossi & Strong (Diptera: Cecidomyiidae) on the coastal plant Iva frutescens L. (Asteraceae) are reduced in the presence of a second host, Borrichia frutescens (L.) (Asteraceae). In this system associational resistance is mediated by parasitoid natural enemies that emerge from Borrichia galls and attack galls on Iva, thereby reducing gall densities on Iva. We quantified distances between patches of Iva and Borrichia and the relative abundances of both plant species and predicted that gall densities would be reduced and parasitism rates elevated on Iva closer to Borrichia compared with more distant patches and on Iva occurring with relatively greater Borrichia abundance in comparison to reduced ratios of Borrichia to Iva abundance. Although gall densities were elevated on Iva more distant from Borrichia, as compared with Iva adjacent to Borrichia, parasitism rates were unaffected by patch distance in this system. Increasing relative abundance of Borrichia was found to significantly reduce gall densities on Iva, though parasitism rates on Iva galls were unaffected by Borrichia abundance. These results suggest other factors, e.g., environmental quality, host plant genotype, etc., may swamp out the effects of parasitoid—mediated AR in this system as Iva becomes more distant from Borrichia or as the abundance of Borrichia relative to Iva is reduced.
Associational resistance (AR) occurs when there is a reduction in herbivore load or attack rate on a plant species growing in a mixed community relative to the herbivore load on that species when grown in monoculture. The converse phenomenon is known as associational susceptibility (AS) and has been found to occur more frequently in plant-insect interactions than AR (Barbosa et al. 2009). A number of mechanisms for AR have been hypothesized, among them the resource concentration hypothesis, the enemies hypothesis, the semiochemical diversity hypothesis, the repellent plant hypothesis and the attractant decoy hypothesis (Tahvahnainen & Root 1972; Root 1973; Atsatt & O'Dowd 1976; Russell 1989; Jactel et al. 2011). Parasitoid—mediated AR has previously been observed in the interactions of the gall midge Asphondylia borrichiae Rossi & Strong (Diptera: Cecidomyiidae) with 2 of its host species, Borrichia frutescens (L.) DC. (Asteraceae) and Iva frutescens L. (Asteraceae) (Stiling et al. 2003). The relatively larger A. borrichiae galls on B. frutescens produce relatively larger parasitoids. The parasitoids produced in such galls are more capable of parasitizing the relatively smaller A. borrichiae galls produced on I. frutescens, which are susceptible by dint of a size effect.
On small offshore spoil islands near the west coast of central Florida, I. frutescens gall parasitism rates were higher and galling rates lower in the presence of neighboring Borrichia, but the converse effect was not observed (Stiling et al. 2003). A follow—up study on the same islands, subsequent to large—scale additions and reductions of B. frutescens from different spoil islands supported earlier observational studies (Stokes & Stiling, in press). Because Iva gall densities were reduced and parasitism rates were higher where B. frutescens was added, while the reverse was observed where B. frutescens was removed, spillover of parasitoids from Borrichia to Iva is likely the most important mechanism of AR in this system. The parasitoid Torymus umbilicatus is likely the member of the parasitoid guild most responsible for the AR effect observed, given that this species' specific parasitism rate increases in the presence of or increased abundance of Borrichia, which is facilitated by its possession of an ovipositor over 2X longer than those of other guild members, permitting hyperparasitism later in gall development (Stiling & Rossi 1994). The attractant decoy hypothesis and other explanatory mechanisms requiring host switching by the gall midge are unlikely, given that host choice experiments showed minimal cross-genus oviposition by A. borrichiae (Rossi et al. 1999) and a molecular population genetic study supported the results of the host choice experiment and provided evidence in favor of genetic divergence among host-species-associated populations at two locales (Stokes et al. 2012). Thus, I. frutescens and B. frutescens are host to distinct populations with little migration between hosts.
Here we present data from 3 additional studies where we seek to quantitatively model response variables (i.e., galling rate and parasitism rate on I. frutescens) with factors that are likely to affect parasitoid-mediated AR: distance separating host plant species, relative abundance of host plant species, and gall densities and parasitism rates on B. frutescens. We expect gall densities to increase and parasitism rates to decrease on Iva with increasing distance from Borrichia. Further, we predict the relationship between the ratio of Borrichia to Iva abundance and Iva gall densities to be negative, while the abundance ratio should have a positive relationship on Iva gall parasitism rates.
MATERIALS AND METHODS
Effect of Host Patch Distance on Coastal Islands
This study was conducted on 6 spoil islands, part of a series of spoil islands spaced at intervals no less than 0.8 km off the Gulf Coast of Florida, near Clearwater and Tarpon Springs (Pinellas County) (Fig. 1). The islands are the result of a dredging operation by the U.S. Army Corps of Engineers in the 1960s and have subsequently been naturally colonized by coastal plants and their insect herbivores. All islands had two patches of I. frutescens, one adjacent to (< 1 m) Borrichia in island centers and one distant from (typically > 10 m) patches of B. frutescens, on island edges, near water and susceptible to tidal input. Monthly counts of galls were conducted from Apr 2009 through Jan 2010 to estimate galling rates in patches of I. frutescens adjacent to and distant from patches of B. frutescens, as well as on B. frutescens. During each sampling period, gall counts were taken of 100 haphazardly-selected leaf buds on each of 5 different I. frutescens bushes in patches near and far from B. frutescens patches, as well as gall counts on 400 haphazardly-selected B. frutescens leaf buds. Mature galls were collected from near and distant I. frutescens patches on all islands and returned to the lab for culturing of the gall community to determine parasitism rates.
Aggregate mean differences in galling rate, total parasitism rate, and T. umbilicatus-specific parasitism rates between patches of I. frutescens near to and far from B. frutescens were compared using paired-sample t-tests, using islands as replicates. We predicted galling rates on Iva close to Borrichia to be lower, and parasitism rates higher, than on Iva distant from Borrichia.
Effect of B. frutescens Abundance on AR-Courtney Campbell Causeway East
The Courtney Campbell Causeway (CCC) is a roadway which extends across northern Tampa Bay from Tampa, Hillsborough County, Florida to Clearwater, Pinellas County, Florida, On the east side of the CCC bridge is an extensive area of B. frutescens forming a continuous ground cover matrix with I. frutescens shrubs emerging intermittently, such that the distance between the host species is zero; moreover, relative host plant abundances vary with varying Borrichia stem density (Fig. 2). Monthly gall counts were performed from Jan 2009 to Sep 2009 to determine gall densities on I. frutescens and B. frutescens. Additionally, areal density of potential oviposition sites (leaf buds) on Borrichia on both sides of each I. frutescens shrub in the study was measured and the local average determined. Emergence holes were counted concurrently on both species and classified according to whether they were the result of emerging gall midges or parasitoids (Stiling & Rossi 1997). Linear regression analysis was performed using I. frutescens gall density and parasitism rate as response variables and B. frutescens gall density, oviposition site areal density and parasitism rate as independent variables influencing Iva gall density. We predicted Iva gall density to be lower and parasitism rate higher as Borrichia gall density, oviposition site density and parasitism rate increased.
Effect of Host Patch Distance and Relative Abundance — Courtney Campbell Causeway West
On the western side of the CCC bridge, discrete patches Borrichia and Iva of various relative abundances occur in a narrow band at random intervals, in effect a linear transect, approximately 4 km distant at the nearest sampling point to the CCC-East density study (Fig. 1). These patches were used to study both the effects of relative host species abundance and the effects of inter— patch distance on gall densities and parasitism rates on I. frutescens. Monthly censuses were performed from Feb 2009 to Oct 2009 to determine gall densities on I. frutescens and B. frutescens. Gall parasitism rates were determined by counting emergence holes and classifying them according to whether they were the result of emerging gall midges or parasitoids. Additionally, both Iva and Borrichia patch sizes and areal densities of potential oviposition sites (i.e., leaf buds) were determined to estimate the total number of potential oviposition sites per patch (Borrichia) or per bush (Iva) and the ratios of the abundances were log-transformed to standardize the scale as the host which is more abundant switches from one host species to the other. Distances between edges of adjacent patches were measured to obtain a minimum inter-patch distance for each pair of patches. To quantitatively model the effects of Iva-Borrichia inter-patch distance and relative oviposition site abundance on AR, linear regressions were conducted on 3 I. frutescens response variables for parasitoid-mediated AR: galling rate, total parasitism rate and T. umbilicatuss-pecific parasitism rate.
Effect of Host Patch Distance on Coastal Islands
Although differences in mean galling rates between near and distant patches of I. frutescens were statistically significantly different, with galling rates much higher on Iva bushes distant from Borrichia (t = 4.261, df = 5, P < 0.01; Fig. 3a), overall parasitism rates were not significantly different (t = 1.816, df = 5, P > 0.05; Fig. 3b). Furthermore, T. umbilicatus-specific parasitism rates were significantly greater on Iva patches distant from Borrichia in comparison to near Iva patches (t = 3.548, df = 5, P < 0.01; Fig. 3c).
Effect of B. frutescens Abundance on AR — Courtney Campbell Causeway East
Of the 4 comparisons performed by linear regression (B. frutescens galling rate on I. frutescens galling rate, B. frutescens leaf bud areal density on I. frutescens galling rate, B. frutescens parasitism rate on I. frutescens parasitism rate, and B. frutescens parasitism rate on I. frutescens galling rate), none were statistically significant (Table 1).
Effect of Host Patch Distance and Relative Abundance — Courtney Campbell Causeway West
Linear regressions were performed to measure the effect of each of 3 variables (B. frutescens galling rate, distance between host species patches, and relative host abundance) on 3 response variables: I. frutescens galling rate, overall parasitism rate on Iva and T. umbilicatus-specific parasitism rate on Iva. Of these 9 comparisons (Table 2), the only significant relationship was a negative correlation between I. frutescens galling rate and log-transformed relative host abundance, which explained approximately 50% of the observed variation (F = 13.017, r = 0.7073, P < 0.01; Fig. 2).
RESULTS OF LINEAR REGRESSIONS IN THE DENSITY STUDY AT COURTNEY CAMPBELL CAUSEWAY-EAST. NO SIGNIFICANT RESULTS WERE DETECTED.
In a prior study, Stiling and colleagues (2003) found a strong effect of B. frutescens galls on I. frutescens gall densities, which they attributed to parasitoids, especially T. umbilicatus, from B. frutescens spilling over and attacking A. borrichiae galls on I. frutescens. A follow-up study involving large scale experimental additions and removals of Borrichia on coastal spoil islands found additional support for the existence of parasitoid-mediated AR in this system (Stokes & Stiling, in press). However, the data presented here cannot further clarify the effects of interpatch distance or patch sizes on the strength of AR operating in the system. Although Iva galling rates were significantly lower on spoil islands when Iva was adjacent (< 1 m) to Borrichia in comparison to when Iva was located distant (> 10 m) from it, parasitism rates showed no such pattern and T. umbilicatus parasitism rates were significantly higher on patches of I. frutescens distant from B. frutescens. One possible explanation for these observed patterns is that there may be a high dispersal rate of both gall makers and parasitoids from islands. Following a flight of at least 0.8 km over open water, all immigrating insects, gall makers and parasitoids alike, may oviposit extensively on the first appropriate plants encountered, which happen to be the Iva bushes occurring at a distance from Borrichia, on island edges.
RESULTS OF LINEAR REGRESSIONS BETWEEN 3 PREDICTOR AND RESPONSE VARIABLES AT COURTNEY CAMPBELL CAUSEWAY-WEST. VALUES PROVIDED ARE THE REGRESSION COEFFICIENT AND THE P-VALUE OF THE MODEL, RESPECTIVELY. BOLD—FACE TYPE INDICATES A SIGNIFICANT RELATIONSHIP.
It was expected that distance between patches would affect the strength of AR along the western side of the Courtney Campbell Causeway. Such an effect would possibly weaken quickly, with AR rapidly dropping off after short distances, as in the case of the repellent plant hypothesis because of diffusion of noxious odors (Tahvanainen & Root 1972; Atsatt & O'Dowd 1976). Surprisingly, no effect of distance between patches on AR was detected in the study, possibly because the study design did not exceed the cruising range of the natural enemies, which is unknown. However, B. frutescens gall densities were generally low in the area examined, reducing the ability to detect distance effects on parasitoid—mediated AR.
Despite the inability to detect an effect of inter—patch distance on AR mediated by natural enemies, Iva galling rates strongly declined (by >5×) with increasing relative abundance of Borrichia, corresponding to general predictions of AR and accounting for 50% of the variation in Iva gall density. However, overall parasitism rates and T. umbilicatus—specific parasitism rates were unaffected by relative abundance of Borrichia. An increase in Borrichia oviposition site availability may indicate conditions favorable to Borrichia and, possibly, less favorable for Iva. Decreased microenvironmental favorability for Iva may, in turn, decrease Iva plant quality and, thereby, depress galling rates.
In summary, these results shed little light on the effects of inter-patch distance and relative host abundance on the strength of parasitoid-mediated AR in the Asphondylia borrichiae-Borrichia frutescens-Iva frutescens system. Plant genotype of Borrichia and local environmental conditions are both known to affect gall densities on Borrichia (Stiling & Rossi 1996). Iva gall densities are also likely to be influenced by environment and Iva genotype. A mosaic of Iva and Borrichia genotypes on coastal islands and along the Courtney Campbell Causeway, together with a varied range of microenvironmental factors could obscure distance and relative abundance effects in this system. More detailed studies are needed to disentangle the relative strengths of the effects of Iva and Borrichia relative abundance, plant quality, genotypic susceptibility and environmental quality to galling and inter-patch distance on parasitoid-mediated AR in this system.