Development of an effective and economical pheromone mating disruption system for swede midge

Objective 1. Test the ability of pheromone blends to elicit non-competitive mechanisms of disruption.

Experiment 1.1. Male swede midge mate-seeking behavior following pheromone pre-exposure.

For some insect species, pre-exposure to large doses of pheromone desensitizes males and renders them incapable of responding normally to pheromones afterward. Synthetic pheromone blends that prevent males from responding normally to pheromone signals released by females can enhance the efficacy of mating disruption. In the laboratory, we pre-exposed male midges to either a low or high dose of three-component chiral or racemic pheromones (Table 2) for ten minutes, and then placed them into a y-tube olfactometer. A y-tube olfactometer is a device that allows an insect to choose between two different odor sources. We gave the males a choice between a solvent-only control and a female-mimicking attractive pheromone dose. We tested whether pheromone blend and dose influenced the likelihood that males were able to locate the female mimic using a binary logistic regression. 

Table 2. Pheromone treatments used in male desensitization experiment^z

^zThe natural S stereoisomers comprise either one quarter or one half of each racemic blend, and thus four times or twice as much racemic material was needed to deliver the same amounts of each S stereoisomer as in the chiral blend².

Experiment 1.2. Effect of pheromone exposure on reproductive behavior of females.

To date, few studies have documented whether or not female insects are capable of "autodetecting" and responding to pheromone mating disruption treatments. In some moth species, females increase the frequency of their own pheromone-releasing behavior and are less likely to mate after pheromone pre-exposure⁴. The objectives of these experiments were to determine whether a) swede midge females increase "calling" (pheromone-releasing) behavior when exposed to synthetic pheromone; and b) female propensity to mate changes following pheromone pre-exposure.

Calling 

For the first experiment, we took newly emerged, virgin females from a wild Ontario population and placed them singly into flasks containing pheromone wicks (Fig. 6) and observed their calling behavior in the laboratory for four hours. We first conducted a dose-response experiment to determine which pheromone dosages elicited responses from females using two doses of the three-component chiral blend (Table 3). We chose these doses because they are high enough to confuse males in a laboratory (pers. obs.). Because we found no differences in female calling behavior between low and high doses, we used the low dose for our blend experiment. In the blend experiment, the wicks released either the three-component chiral pheromone blend, the three-component racemic blend, the chiral main pheromone component, the racemic main pheromone component, or a solvent-only control (Table 3). Air circulated through the flasks using tubing connected to a small air compressor, and pheromone-laden air leaving the flasks was removed from the room via an overhead exhaust duct. We checked the females every fifteen minutes from 9:30 am through 1:30 pm, their peak period of calling, and recorded whether or not they were releasing pheromones by checking to see if their ovipositors and pheromone glands were extruded. 

Table 3. Pheromone blends used in autodetection experiments

We repeated the calling experiment in 52x52 ft plots of broccoli treated with mating disruption to better understand how females react to synthetic pheromone in realistic field settings. The plots were set up at the University of Guelph Elora Research Station in Elora, Ontario. Each plot contained 20 rows of 'Everest' broccoli with 80 mating disruption dispensers. We used reservoir-type semi-permeable polyethylene bag dispensers. Each bag contained a microcentrifuge tube with a small hole drilled in the top, containing 50 ug (2S,9S)-diacetoxyundecane, 100 ug (2S,10S)-diacetoxyundecane, and 1 ug (2S)-acetoxyundecane (the three-component chiral blend). The bags hung from wire stakes 25 cm from the ground.

We held newly-emerged unmated females individually in small glass vials with mesh covers to allow for air circulation, and placed half of the vials in untreated plots and half in plots treated with three-component chiral pheromone mating disruption dispensers. Vials of midges sat atop overturned pots so that they were approximately 25 cm above the ground at the same height as the dispensers. We observed the females in both treated and untreated plots concurrently from 9:30 am until 12:30 pm and recorded whether or not they were releasing pheromones every ten minutes. Although we were only able to observe female calling behavior in one treated and one untreated plot at a time, we repeated the experiment across six experimental periods and two plots from each treatment in late July and August in 2017 and 2018, during 6 – 12 weeks after installing dispensers. During each period, we observed the calling behavior of 4-10 midges.

To test the effect of pheromone mating disruption treatments on the probability of female calling in our laboratory and field experiments, we used generalized linear mixed models (GLM) with generalized estimating equations (GEE) extensions. Calling followed a binomial distribution. We also tested whether pheromone treatments influenced the duration of calling. We used a binary scoring system (yes/no) to record whether we observed midges calling for at least half of the time period, and analyzed the effects of pheromone treatments on the probability of midges "continuously calling" using a binary logistic regression.

Propensity to mate 

In the second part of our experiments, we took newly emerged, virgin females from our Swiss laboratory colony reared at the University of Vermont, pre-exposed them three-component chiral and racemic pheromone blends in flasks using the same dosages as in our blend-response calling experiments (Table 3, Fig 6), and then observed their propensity to mate with males. We omitted the single-component pheromone treatments from these experiments because they did not elicit any behavioral changes in the females in the first experiment (results detailed below). Using the same flask setup as the first experiment, we exposed the females to pheromone treatments for 10 minutes, removed them from the flasks, and then housed them in small containers with four males. We then observed the following behaviors: 1) Female calling every two minutes; 2) Female rejection of males following male courtship behavior; and 3) Successful copulation.

We used a binary logistic regression to determine whether pheromone treatments significantly influenced female mating success following pre-exposure. Similarly to our calling experiments, we used a GLM model with GEE extension to determine whether the probability of calling differed over time between pheromone treatments for midges that did not mate. Finally, we used another binary logistic regression model to determine whether the probability of females rejecting courting males differed between pheromone treatments. 

Objective 2. Test the effect of swede midge population density on the efficacy of pheromone blends for mating disruption in enclosures.

Experiment 2. Population density effects on pheromone mating disruption efficacy.

We tested the efficacy of three-component chiral and racemic pheromone blends to disrupt mating using two densities of swede midge held in cages with cauliflower plants. Two population densities of newly emerged, unmated midges, either 2 females with 3 males, or 10 females with 15 males, were held in each cage. Each cage contained four potted cauliflower plants and received either a dispenser releasing the three-component chiral blend (50 ug (2S,9S)-diacetoxyundecane, 100 ug (2S,10S)-diacetoxyundecane, and 1 ug (2S)-acetoxyundecane), the three-component racemic blend (200 ug 2,9-diacetoxyundecane, 400 ug 2,10-diacetoxyundecane, and 2 ug 2-acetoxyundecane) or no dispenser (control). We situated the cages at least 50 m apart and separated them by buildings to prevent the spread of pheromone treatments between cages. After allowing the midges three days to mate, we returned them to the laboratory, where we removed the plants and placed them in separate midge-free cages to allow the eggs to develop into larvae. After 10 days, we dissected the growing points of each plant and counted the number of larvae, our proxy for mating success, where few or no larvae indicate mating disruption success. 

Objective 3. Conduct open-field trials of single-component pheromone blends to disrupt mating and provide crop protection.

Experiment 3. Testing single-component blends to disrupt mating and reduce crop damage in the field.

We tested whether chiral and racemic single-component pheromone treatments (either 100 ug 2S,10S-diacetoxyundecane or 400 ug 2,10-diacetoxyundecane, respectively) successfully confused males and provided protection of a broccoli test crop. We used 80 polyethylene bag pheromone dispensers (Fig. 3) to distribute the pheromone in 52x52 ft plots of broccoli (described for field calling experiments). Plots were situated at least 1,640 ft away from each other to avoid contamination of pheromone treatments via wind. We installed the dispensers only once, and did not replace them during the season. The broccoli was managed organically without any insecticides and grown according to regional practices. Our field plots were located at the University of Guelph Elora Research Station, Pfenning's Organic Farm, a commercial organic vegetable operation in New Hamburg, ON, and at the Institut de recherche et de développement en agroenvironnement (IRDA) in St-Bruno, QC. We chose these locations because of their well-established populations of swede midge, which allowed us sufficient and dependable swede midge pressure to test our treatments. We had one block (one plot of each treatment) at Elora, Pfenning's, and IRDA in 2017, and one block at Elora in 2018.

We collected data throughout the broccoli's growing season on trap shutdown and swede midge damage to the broccoli. In addition to our pheromone dispensers, each plot also contained four commercial monitoring traps for swede midge. Each trap contained the commercial swede midge lure, which releases an attractive dose of the swede midge pheromone at much lower concentrations than the dispensers (Fig. 7). We checked the traps weekly for males. When pheromone mating disruption is successful, there should be few to no males caught in traps (trap shutdown), indicating that the males are confused and cannot locate attractive pheromone sources. At three and six weeks after transplanting we rated 60 random plants per plot for damage to vegetative parts using a three-point damage scale for presence of swollen petioles, stem scaring, leaf curling, and damage to the growing point. At the time of harvest, we rated 60 broccoli crowns per plot for marketability (either yes or no) and scarring and deformation of the head using a five-point scale of increasing severity. 

To determine whether pheromone blend was a predictor of the number of males caught in traps in the plots, we used a GLM with GEE model. We used ordinal logistic regression and Kruskal Wallis tests to compare damage ratings and numbers of marketable broccoli heads across pheromone treatments.