UC Berkeley

Environmentally mediated infectious diseases pose significant challenges to public health. Because these pathogens undergo key stages of their life cycles outside the human host, they are highly sensitive to environmental conditions, particularly climate variability and climate change. Soil-borne fungal infections are a notable example because temperature and precipitation can affect pathogen growth, dispersal, and human exposure. Incidence of coccidioidomycosis (Valley fever), an emerging fungal disease caused by inhalation of Coccidioides spores, has increased substantially in California in recent decades. While growing evidence indicates that coccidioidomycosis incidence is linked to climate conditions, important questions remain about how climate variability and extremes affect disease risk. In this dissertation, I examine how climate influences coccidioidomycosis incidence across multiple timescales in California, from seasonal transmission dynamics to the effect of short-term extreme events to generating near-term climate-based forecasts and long-term climate change projections. First, I examine how climate structures the seasonal timing of coccidioidomycosis transmission. Using a distributed-lag Markov state transition model, I characterize variation in the onset, end, and duration of transmission seasons across California and estimate the effects of temperature and precipitation on season timing. I find that dry spring conditions shift the onset of the transmission season 2.80 weeks earlier, dry fall conditions prolong the season by 0.69 weeks, and combined dry spring and fall conditions lengthen the season by a total of 3.70 weeks, suggesting that projected shifts toward longer dry seasons under climate change may extend periods of elevated disease risk. Second, I examine whether extreme weather events affect coccidioidomycosis risk above and beyond background climatic conditions. Using a time-stratified case-crossover design, I find that, compared to no extreme heat event, extreme heat events lasting 2-3 days increase the odds of infection by 3.75%-11.07%, whereas extreme heat events lasting 4 or more days reduce the odds of infection by 3.23%-14.34%. In contrast, extreme precipitation events consistently reduce infection odds, ranging from 6.62%-16.16% reductions for 2-day events to 25.06%-45.11% reductions for events lasting at least four days. Together, these results show that the epidemiology of coccidioidomycosis may be punctuated by extreme events that either amplify or interrupt background climatic influences. Third, I ask whether climate-disease relationships can be used prospectively to anticipate future incidence. I develop a climate-based ensemble forecasting framework that combines statistical and machine learning models to forecast near-term (1-2 years) coccidioidomycosis incidence across California. I show that interannual hydroclimatic variability, including multi-year drought cycling, combined with intra-annual meteorological patterns, can be used to accurately (R2 = 0.87) anticipate future increases in incidence. These findings demonstrate the practical value of climate-informed forecasting for public health preparedness, risk communication, and clinical awareness. Finally, I examine how climate change may reshape the future burden and distribution of coccidioidomycosis in California. By combining estimates of the relationship between temperature, precipitation, and incidence from distributed-lag non-linear models with high-resolution downscaled climate projections, I project that warming temperatures and shifting precipitation patterns will increase statewide incidence by 244-540 cases per year by 2100, corresponding to a 7.8%-18.4% increase over the historical baseline, depending on the climate scenario. I also project that climate change will amplify year-to-year variability in incidence, likely leading to larger incidence increases following wet winters, and will lead to a northward shift in disease incidence over the coming decades. Further, I find that projected incidence increases may disproportionately affect Hispanic populations, highlighting the potential for climate change to exacerbate incidence disparities and identifying a priority population for public health planning. Overall, this dissertation shows that climate is a fundamental driver of coccidioidomycosis transmission, ranging from near-term seasonal disease risk to shifting dynamics over the coming decades. By integrating flexible statistical and machine learning models with high-resolution disease surveillance and climate data, this work advances our understanding of how environmental change influences infectious disease risk and provides a framework for anticipating and preparing for the growing burden of climate-sensitive fungal diseases.