It’s a small world: How leaf damage and soil fungi can affect bee disease

Methods

Summary

For each experiment, we will randomly assign 100 plants to the treatment (herbivory or fungal addition) and 100 plants to the control. For the herbivory experiment, leaves will be damaged by adding tobacco hornworm in bags at the bolting stage. Prior work has found that damage at this stage will induce floral defenses. Control plants will have bags without caterpillars. For the soil fungi experiment, fungi will be added to sterilized soil and pots when seeds are planted; colonization will be confirmed via root samples one month later. Sterilized fungal inoculum will be added to control plants. Nectar and pollen will be collected and used daily. Worker bees will be inoculated with Crithidia and randomly assigned to one of four treatment combinations: treatment/control nectar and treatment/control pollen. Each day we will collect and pool nectar and pollen within treatment, and feed these to bees in the appropriate treatment. If there is insufficient nectar, we will mix it with sucrose solution to augment; we will use the same amount of tobacco nectar added to sucrose solution in each treatment. Insufficient pollen will be combined with a standard multifloral pollen mix. After one week, bees will be dissected and pathogen cells counted. We will continue to test bees until flowering has ended. 

Challenges

The methods for damaging tobacco leaves with tobacco hornworm are straightforward, and the Adler lab has experience with this and with experiments asking how diet affects bumble bee disease. We have done less working manipulating soil fungi, and there are various challenges with maintaining sterile conditions and avoiding contamination. If these challenges prove too difficult, we may alter that project to instead ask how different fertilizers, which are also ubiquitously used in agriculture, affect floral traits and bee disease. Another possible challenge is whether flowers produce adequate nectar and pollen. Dr. Adler has data on nectar production from prior tobacco work, and we have planned for sample sizes that should be more than adequate. 

Pre Analysis Plan

For each experiment, we will analyze the effect of nectar treatment (e.g., herbivory vs. control) and pollen treatment (e.g., herbivory vs. control) on pathogen cell count in experimental bees using generalized linear models with appropriate error distributions. We will include source colony as a random factor, and measures of bee size as covariates. We will also analyze the effect of treatment on nectar and pollen production and nicotine concentrations. We predict that leaf herbivory will reduce pathogen load via increased nicotine, and that this effect will be stronger for pollen than nectar due to higher nicotine concentrations. We know much less about how beneficial fungi will affect floral defenses. If fungi also increase nicotine in nectar and pollen, we predict this will also reduce bee pathogen infection. However, if fungi increase nectar or pollen nutrients but not nicotine, this could increase pathogen infection.