Files in this item



application/pdfJai_Holt.pdf (5MB)
(no description provided)PDF


Title:The effects of environmental factors on honey bee morbidity
Author(s):Holt, Jai
Advisor(s):Baylis, Katherine R.
Department / Program:Agr & Consumer Economics
Discipline:Agricultural & Applied Econ
Degree Granting Institution:University of Illinois at Urbana-Champaign
Honey Bee
Abstract:This thesis examines the relationship between environmental factors and honey bee health. I have three primary objectives in conducting this research. First, I examine whether being located near neonicotinoid seed treated crops increases honey bee morbidity. Second, I examine if forage availability and weather affect honey bee health. Third, I provide a demonstration of how the biological modeling of environmental factor effects on honey bee health can be used by beekeepers to make cost-effective treatment decisions. I take an interdisciplinary approach to tackling these objectives drawing on methods from epidemiology to test hypotheses informed by entomology. This research is the first to make use of geocoded data from the USDA APHIS Honey Bee Disease survey to examine environmental factors, including forage availability and weather, which affect honey bee disease. I use varroa destructor mite and nosema parasite infection levels as measures of honey bee health. Varroa is an important measure of honey bee health because it has been linked to increased overwintering colony losses (USDA 2013). Nosema has also been linked to higher overwintering losses and is also associated with decreased colony productivity (Botías, et al. 2013). To proxy for neonicotinoid exposure, I use corn, soy, cotton, canola, sorghum, rice, barley, spring wheat and winter wheat as my treated field crops because nearly all seed planted to grow these crops is treated with neonicotnoids. My hypothesis is that increasing the acreage of neonicotinoid seed treated field crops in the average forage range of two miles surrounding an apiary will lead to higher disease loads within the bee populations in those apiaries. I estimate pesticide exposure by examining key times bees are most likely to come in contact with neonicotinoids; when treated crops are planted and when they bloom. I build on a previous study by adding multiple years of apiary sample data and by using additional control variables including weather and forage availability. In addition, I isolate observations of apiaries that were non-migratory for the analysis. The results of the nosema analysis provide evidence that location near neonicotinoid treated field crops may be associated with higher levels of nosema in some cases. When examining the relationship between nosema levels and the area of treated field crops without consideration for pesticide timing, I find that canola is the only one of the 9 field crops that is positively correlated with the nosema level of the colony. When peak exposure timing controls are added to the analysis, I find evidence that nosema levels are higher during planting time for some crops. Corn, soy, cotton, canola and rice all have a positive and statistically significant relationship to nosema level during planting time in at least some of the model specifications. Location near treated crops during bloom timing does not appear to be correlated with higher nosema levels. The findings that being located near treated crops during planting time is correlated with increased nosema levels is consistent with findings from a lab study that found bee colonies exposed to low levels of the neonicotinoid imidacloprid had higher nosema levels compared with the control group (Pettis, et al. 2012). My findings support the argument that neonicotinoid exposure should not be ruled out as a factor that influences honey bees’ ability to fight of nosema infection. I also examine whether being located near treated crops increases the likelihood of being contaminated with threshold levels of infection. Examining threshold levels of infection is important because low levels of infection are less likely to lead to colony loss. When I consider the threshold level of nosema of 1 million spores per bee, canola is correlated with a higher probability of nosema infection in all model specifications. I also find that apiaries located near canola have an increased probability of having a non-zero amount nosema detected in the apiary sample. The consistent evidence that canola is correlated with more nosema is interesting because canola is typically considered a good forage crop for honey bees. Therefore, one might expect that non-treated canola would be correlated with lower nosema levels because bees foraging on canola would be healthier over all. The positive relationship between canola acreage and nosema level could be driven by pesticide exposure. However, without precise measures of neonicotinoids surrounding the apiaries it is not possible to determine if neonicotinoids are the cause of this relationship. It is possible that there is something about canola itself that leads to these higher nosema levels. Further studies are needed to disentangle the effects of neonicotinoids and canola on the level of nosema infestation. The findings from the varroa mite analysis suggest that an increase in the nearby acreage of treated field crops is not correlated with an increase in mite level. These results hold when I account for peak exposure periods during planting and blooming. Treated crops that do not provide comparative good bee forage including corn, barley and wheat are not correlated with higher mite levels. When considering the threshold level of mite infection of 3 mites per bee, I find that apiary locations near neonicotinoid treated crops are not correlated with an increased probability of having a threshold level of varroa mite infection (except for rice in some specifications). Therefore, there is little evidence to suggest that location near treated field crops will increase varroa mite levels. When pursuing my second objective of determining if forage availability and weather impact honey bee health, I expect that increased forage availability is correlated with morbidity levels. I investigate 5 different strategies for controlling for forage availability. Of these strategies, I find that using a Normalized Density Vegetation Index (NDVI) does the best job of explaining morbidity levels. My results suggest varroa mite level is negatively correlated with NDVI as expected. Nosema level, on the other hand, is positively correlated with NDVI. When examining the relationship between weather and morbidity, I find the minimum temperature in the month of the sample is an effective indicator of honey bee health. I expect that increasing the minimum temperature will be correlated with lower morbidity levels. The results support this hypothesis in the case of both nosema and varroa. I also examine the relationship between precipitation in the month of the sample and morbidity level. The results of my analysis suggest that precipitation is neither correlated with nosema level nor varroa level. Finally, I demonstrate how using a biological model of honey bee health that takes into consideration environmental factors can improve a beekeepers’ abilities to determine whether to treat their colonies. I use an example of an environmental model that predicts whether or not a colony has nosema parasite. This model is useful because nosema is difficult to detect. I show how a commercial beekeeper using the environmental model to choose which colonies to treat can theoretically save money by avoiding treating for nosema unnecessarily. The results of this research call attention to the complexity of honey bee health and the need for continued interdisciplinary research to solve the remaining honey bee health mysteries. Nonetheless, this research also highlights the usefulness of analyzing real world data as a compliment to existing lab-based scientific studies on honey bee health.
Issue Date:2014-09-16
Rights Information:Copyright 2014 Jai Holt
Date Available in IDEALS:2014-09-16
Date Deposited:2014-08

This item appears in the following Collection(s)

Item Statistics