Mechanisms and Mitigation: Effects of Light Pollution on West Nile Virus Dynamics

Light pollution, or the presence of unnatural light at night, is a pervasive and growing problem across the globe. While often pictured in urban centers, light pollution is far reaching and can affect seemingly safe and minimally developed environments. For example, agricultural communities with artificial lighting near facilities can generate such light pollution in rural areas. Further, streetlights and illuminated billboards along roads and highways can generate light pollution far from cities. Given how pervasive this anthropogenic stressor is, it is surprising that not much is known about how artificial light at night, or ALAN, affects humans or wildlife, especially those that harbor infectious diseases. Previously, the biomedical field uncovered many negative effects of ALAN exposure on the immune system. Laboratory rodents experience exaggerated fever responses and decreases in bacterial killing ability, among other consequences. This is altogether not surprising as components of vertebrate immune systems possess circadian rhythms. Additionally, other studies have found that exposure to ALAN induces hormonal dysregulation, leading to a mismatch in circadian and circannual timing. Ultimately, the consensus among the research community is that ALAN generates a myriad of negative effects on immunity and other components of organismal physiology. However, it is yet to be uncovered how ALAN may affect infectious disease dynamics. Here, I investigated for the first time how light pollution might affect infectious disease dynamics. To do so, I considered the mosquito-transmitted flavivirus, West Nile virus (WNV). WNV is among one of the most important arboviruses worldwide, and continues to affect human, horse, and bird populations in the United States following its introduction in 1999. WNV has also been described as a “peri-urban” disease-causing agent, as it often emerges in suburbs and other built environments. Lastly, as WNV is harbored by common and urban-residing avian reservoirs, light pollution has the potential to affect transmission through effects on these amplifying hosts. I used the ubiquitous house sparrow as the study species here because they are competent WNV reservoirs in nature, reside in almost all part of the continental United States, and are residents of light polluted areas. Overall, I asked how exposure to low-intensity light pollution may affect the host competence (i.e., ability to generate new infections) of house sparrows to WNV. In my first chapter, I considered how light pollution might affect multiple aspects of WNV dynamics in house sparrows, from molecular mechanisms to outbreak potential. I first discussed the circadian nature of immunity and viral defenses and outline how these may become dysregulated by exposure to light at night and other downstream effects. I then discussed the effects on host competence including infectious period, mortality, and vector-associated behavior. Lastly, I walked through multiple components of the R0 equation (i.e., the basic reproductive number) which essentially determines how many new infections one host can generate based on infectious period, probability of being infected after a bite (in hosts and vectors), biting rate, background mortality, disease-induced mortality, and other relevant parameters. Combined, this chapter outlined how ALAN might be altering vector-transmitted diseases. In my second chapter, I experimentally tested whether low-intensity ALAN affected WNV responses in House sparrows. My results first revealed that exposure to ALAN did not affect glucocorticoid regulation in sparrows, indicating that any such immune dysregulation would not have been mediated by this stress hormone. Importantly, I found that exposure to ALAN, no matter the duration (ranging from 7 to 21 days), allowed sparrows to maintain infectious levels of WNV for twice as long as controls. Although ALAN-exposed sparrows lost a significant amount of body mass, there were no differences in mortality rate among the groups. Additionally, transcriptomic analyses revealed that while ALAN-exposed birds upregulated WNV immune defenses earlier than controls, they displayed signs of immunopathology. After considering all of this, I used a simple single-host, single-vector R0 model, and found that extending the infectious period by two days increased outbreak potential by ~41%. In my third chapter, I explored whether spectral composition of ALAN affects host competence to WNV in house sparrows. I repeated a similar WNV infection experiment but substituted out an incandescent light for 3 different types of LED lights: cool white, warm white, and amber-hued (which are marketed as “wildlife-safe”). First, I found that exposure to warm white light at night significantly suppressed the circadian hormone, melatonin. Melatonin is known for regulating many circadian functions, but also plays a role in mediating immune responses and attenuating infection-induced damage. Additionally, I found that exposure to broad-spectrum (both cool and warm white) ALAN did not affect viremia (i.e., amount of virus in circulation), but interestingly, exposure to amber-hue ALAN marginally but significantly increased WNV resistance (i.e., decreases viremia). Alternatively, birds exposed to broad-spectrum ALAN did experience higher WNV-induced mortality and tended to die at lower viremias than control birds. Altogether, altering spectral composition of light at night has the potential to alleviate negative effects on wildlife. In my fourth and last chapter, I investigated whether these effects of ALAN observed in the lab manifest ecologically using Florida Department of Health (FDOH) sentinel chicken surveillance data. The FDOH monitors environmental circulation of vector-transmitted diseases by surveying chickens throughout the state weekly for the presence of WNV antibodies, indicating that they have been exposed to the virus. I extracted ALAN intensities from the World Atlas of Artificial Sky Brightness at these surveillance sites while accounting for weather and urbanization variables. Mixed effect model selection revealed that ALAN as a polynomial term and temperature of the month prior were the best predictors of WNV exposure risk across Florida. Exposure risk was lowest in non-light polluted areas, peaked in areas of low light pollution, and then tapered off in areas of moderate to high light pollution. Unlike previous results, I did not find support for parameters of urbanization here, but more work needs to be done to uncover what about ALAN exposure at ground-level may be driving WNV exposure risk