L. distinguendus females avoided the odor of moldy wheat in an four-chamber olfactometer irrespective of the Aspergillus species used to inoculate the grain. In both Aspergillus species the odor of uninfested wheat was significantly preferred over inoculated grain (Figure 1a). Parasitoids were strongly arrested by the odor of larval feces from weevil cultures without mold (Figure 1b). However, larval feces from moldy weevil cultures were no longer preferred by the parasitoids when tested against filter paper as a control. The odor of feces from both the A. sydowii and A. versicolor cultures was neither significantly preferred nor avoided (A. sydowii: P = 0.76, A. versicolor: P = 0.058) (Figure 1c). However, when feces from moldy and non-moldy cultures were offered simultaneously, females clearly avoided the feces from the moldy cultures in both cases (Figure 1d).

Arrestment time of female parasitoids was significantly decreased on wheat grains from A. sydowii- and A. versicolor-infested weevil cultures when compared to the non-moldy control grains (Figure 2a). Moreover, characteristic elements of the host recognition behavior (drumming and drilling) were shown less often on grains from the moldy weevil cultures than on control grains (Figure 2b–c). No significant differences were found between grains from weevil cultures infested by A. sydowii and A. versicolor, respectively (arrestment time: U = 145.5, P = 0.14; drilling: U = 153, P = 0.20; drumming: U = 138, P = 0.09).

Females of L. distinguendus parasitized hosts from both moldy and non-moldy weevil-cultures. However, the number of offspring was significantly lower when host larvae originated from moldy weevil cultures infested with either A. sydowii or A. versicolor as compared to non-moldy cultures (Figure 3). There was no significant difference in the number of offspring between the two mold species (P = 0.43). Body size as measured by the length of the tibiae was significantly affected by the infestation with mold. Parasitoids that had developed on hosts from non-moldy weevil cultures had longer tibiae (female: 0.579 mm ± 0.003; male: 0.5058 mm ± 0.005) than individuals from A. sydowii- (female: 0.571 mm ± 0.003; male: 0.5058 mm ± 0.003) or A. versicolor-moldy cultures (female: 0.573 mm ± 0.003; male: 0.500 mm ± 0.003). Moreover, females of L. distinguendus had generally longer tibiae than males. There was no interaction observed between treatment and sex (Table 1).

Fecal volatiles from moldy and non moldy weevil cultures were collected by closed loop stripping and analyzed by coupled gas chromatography-mass spectrometry (GC-MS). The major eight-carbon fungal volatile detected in larval feces from moldy weevil cultures was 1-octen-3-ol accompanied by lower amounts of 3-octanone, and 3-octanol (Table 2). In larval feces from non-moldy host cultures all compounds were found only in traces. The headspace of A. sydowii contained additionally some sesquiterpenes as minor compounds.

L. distinguendus females avoided significantly synthetic 1-octen-3-ol at doses between 1500 to 300 ng (Figure 4). At a dose of 30 ng, however, parasitoids were neither repelled nor attracted to the synthetic chemical (P = 0.62).

Insect cultures were kept at constant conditions of 25°C, 65% relative humidity (RH), and a daily light:dark cycle of 16:8 h. The L. distinguendus strain used in the experiments was collected on Stegobium paniceum L. in a flour mill in Uzwill, Switzerland. In our laboratory the parasitoid was reared on larvae (3^rd and 4^th instar) of the granary weevil Sitophilus granarius in wheat grains (Triticum aestivum L., var. Batis), as described elsewhere 20.

For all experiments except for experiment 2, naïve L. distinguendus females were used, i.e., individuals without mating and oviposition experience. For this purpose, parasitoids were collected immediately after emergence and held in groups of at most 15 individuals in Petri dishes lined with moistened filter paper. To obtain experienced parasitoids, freshly emerged females were kept together with males in a Petri dish containing host-infested grains for parasitization. After three days, females were removed and kept without hosts in Petri dishes on moistened filter paper. One hour before experiments, L. distinguendus were isolated in 1.5 ml microcentrifuge tubes for acclimation at room temperature. Behaviors shown by the parasitoids during experiments were recorded by using the computer software The Observer 3.0 (Noldus, Wageningen, The Netherlands).

The mold species A. versicolor (Strain no.: BBA 72388) and A. sydowii (Strain no.: BBA 72389) used in the experiments were obtained from the culture collection of the Biologische Bundesanstalt für Land- und Forstwirtschaft, Institute of Plant Virology, Microbiology, and Biosafety, Berlin, Germany. The isolates originated from intense weevil- and fungi-infested grain cultures of our institute.

Stock cultures of A. sydowii and A. versicolor were freeze-dried and stored at 4°C. Subcultures were cultivated on nutrient deficient agar SNA at room temperature for 1 to 2 weeks. A 10 ml portion of sterile water was added to the subculture and the surface was scraped off until conidia were suspended. The suspension was filtered through a folded filter paper (S&S 595, Schleicher & Schuell, Dassel, Germany) and the conidia concentrations of the Aspergillus species were determined by counting in a Bürker counting chamber and adjusted with sterilized water to 1.5 × 10⁶ conidia per ml.

Three batches of wheat grain (750 g each) were autoclaved for 30 min. Two batches were treated with 45 ml conidia suspension of A. sydowii and A. versicolor, respectively. For control, the third batch of grain was treated with 45 ml sterilized water. The three batches of grain were incubated separately for four weeks in desiccators at 25°C. The humidity within the desiccators was adjusted to 65% by a saturated solution of ammonium nitrate 54. After this incubation period, mold was visible in the two treated batches of wheat but not in the control. Adult S. granarius (75 ml) were added to each batch of grain, allowed to lay eggs into the grains for seven days and subsequently removed. Larval feces used for the olfactometer bioassays and the chemical analysis were obtained by sieving the grain of treated and control cultures 21 to 28 days after the adults had been removed. After removal of the larval feces, weevil-infested grains of mold-treated and control cultures were used to investigate the influence of mold on the host recognition behavior of L. distinguendus in experiment 2 and fitness of the parasitoids in experiment 3.

The response of L. distinguendus females to different odor samples was examined using a static four-chamber olfactometer 2021. The olfactometer consisted of a cylinder made of acrylic (4 cm high, ∅ 19 cm) divided into four chambers by crosswise-arranged vertical plates. The top of the cylinder was covered by a plastic gauze (mesh 0.1 mm) functioning as a walking arena for the parasitoids. A lid consisting of a plastic ring (4 mm high, 19 cm ∅) and a glass plate was placed on top of the cylinder to prevent the parasitoids from escaping. Odor samples were placed in one of the chambers (test chamber) using a Petri dish (5.5 cm ∅). The opposite chamber was used as control chamber and the remaining two chambers adjacent to the test chamber were considered as buffer zones. A single female was released into the center of the walking arena and the time the parasitoid spent in the field above the test or control chamber was recorded for 10 minutes. Parasitoids that were motionless more than 50% of the total observation time were assumed to be unmotivated and excluded from statistical analysis. To avoid biased results due to possible side preferences of the parasitoids, the position of the samples and the controls was rotated clockwise after each test. Walking arena and glass plate were regularly cleaned with ethanol and demineralized water. Odor sources were exchanged after five individuals had been tested.

The following odor samples were offered in the olfactometer: (A) 10 g grain infested with A. sydowii or A. versicolor vs. 10 g untreated grain (N = 20 for each treatment), (B) 200 mg larval feces from the non-moldy control culture vs. a piece of brown filter paper (N = 21), (C) 200 mg larval feces from the mold-infested weevil cultures (A. sydowii: N = 21, A. versicolor: N = 31) vs. brown filter paper, and (D) 200 mg larval feces from the mold-infested weevil cultures (A. sydowii: N = 21, A. versicolor: N = 21) vs. 200 mg larval feces from the non-moldy control culture.

The host-recognition behavior of experienced L. distinguendus females was examined in a bioassay chamber (10 mm ∅ × 3 mm high) made from acrylic 55. Single grains from weevil cultures infested with A. sydowii or A. versicolor or those from the control culture were presented to individual females. Arrestment time on the grains and characteristic elements of the host-recognition behavior (number of drumming series and drilling) were recorded for 10 min using a stereomicroscope under illumination of a microscope light. Grains were exchanged after every parasitoid tested (N = 20 for each treatment).

This experiment was done to estimate the offspring production of L. distinguendus females in a no-choice situation with hosts from moldy and non-moldy weevil cultures. Virgin females were allowed to copulate with a male and subsequently placed individually in Petri dishes containing weevil-infested grains from the moldy weevil-cultures (A. sydowii: N = 25, A. versicolor: N = 23) or the control culture (N = 25). Weevil-infestation of grains was equal in moldy (A. sydowii: 82%, A. versicolor: 76%) and non-moldy (78%) weevil cultures. Parasitoids were allowed to oviposit until their death. Offspring was assessed by counting the emerged parasitoids after 30 days. In addition, hind tibia length 5657 was used to measure the body size of randomly selected individuals from the offspring of the three treatments (A. sydowii: N = 102, A. versicolor: N = 91, control: N = 143).

The response of L. distinguendus females to the typical fungal volatile 1-octen-3-ol (98%, Sigma-Aldrich, Steinheim, Germany) was investigated in another olfactometer experiment. The chemical was applied at different doses [1500 ng (N = 24), 300 ng (N = 25), and 30 ng (N = 19)] to filter paper discs (∅ 4 cm, Melitta, Minden, Germany) and offered in the test chamber of the olfactometer. Paper discs treated with the pure solvent (30 μl dichloromethane) were used as control.

Volatiles emitted from the larval feces of the three different weevil-cultures (N = 3 for each treatment) were collected by closed loop stripping (CLS) as described elsewhere 21. Volatiles from 5 g larval feces were collected for 4 h on a 1 mm charcoal layer (5 mg) of a CLS-adsorption tube (65 mm length × 5 mm ∅) (Gränicher & Quartero, Daumazan, France). The volatiles were eluted with 25 μl dichloromethane containing 5 ng/μl methyl undecanoate as an internal standard and used for chemical analysis by GC-MS.

Volatile extracts were analyzed by GC-MS using a Fisons 8060 GC (Fisons Instruments) equipped with a 30 m × 0.32 mm ID × 0.25 μm film thickness DB-5 ms column (J & W Scientific, Folsom, CA, USA) and coupled to a Fisons MD800 quadrupole MS operated in the electron impact (EI) mode at 70 eV. Helium was used as carrier gas at a head pressure of 10 kPa. The oven temperature was 40°C for 4 min and then rose to 280°C at a rate of 5°C/min. The final temperature was held for 10 min for thermal cleaning of the column. Chemicals were identified by comparison of mass spectra and retention times with those of synthetic reference compounds (Sigma-Aldrich, Steinheim, Germany). For quantification of selected compounds, the peak area of each volatile was related to the peak area of the internal standard.

The arrestment times of female parasitoids spent in test and control field of the olfactometer (experiments 1 and 4) were analyzed by a general estimator equation-generalized linear model (GEE-GLM, SAS version 9.1, PROC GENMOD). Model fitting was carried out with a Poisson error distribution and log link function. Residuals were inspected for normality. Since odor samples in these experiments were renewed after every five insects, we tested not only the factor treatment but also the factors sample and order as explanatory variables for the dependent variables (= arrestment times) to make sure that neither the order of the parasitoids nor the individual sample had a significant influence on their response. However, in none of the experiments a significant effect of order or sample was found indicating that multi-used odor sources elicited constant responses in the parasitoids during the experimental period. The arrestment time on the weevil-infested grains and the number of drumming series and drillings in experiment 2 were analyzed by the Kruskal-Wallis H-test followed by multiple Bonferroni-corrected Mann-Whitney U-tests for individual comparisons (data partially not normally distributed). Number of offspring (experiment 3) was compared by a one-way ANOVA and subsequent least significant difference (LSD) tests for post hoc comparison. Tibia lengths were analyzed by a two-way ANOVA (factor 1: treatment, factor 2: sex). Statistical analyses except for the GEE-GLM were done using Statistica 4.5 scientific software (StatSoft, Hamburg, Germany).