Accelerating invasion potential of disease vector *Aedes aegypti* under climate change

Vector-borne diseases represent a significant global health challenge, causing over 1 billion infections and 1 million deaths annually. Insect vectors, being ectothermic, are highly sensitive to climatic fluctuations, making understanding their responses to climate change crucial for predicting disease risk. The distribution and abundance of vectors often directly influence the incidence and spread of vector-borne diseases. Estimating a species' environmental suitability, which reflects how favorable a location is for a species based on environmental factors (temperature, rainfall, habitat), is a common approach. This can be done through statistical models correlating species occurrences with environmental covariates or mechanistic models utilizing physiological responses to environmental parameters. Mechanistic models offer advantages, especially for invasive species, by identifying specific biophysical pathways linking life-history traits (development rates, mortality) to the environment, thus avoiding biases inherent in correlative approaches. Previous studies have developed mechanistic models for invasive vectors like *Aedes albopictus* and *Aedes aegypti*, but these have rarely been applied at large scales to assess long-term responses to climate change. This study addresses this gap by using a phenology model to examine environmental suitability for *Aedes aegypti* development. Phenology models explicitly model an organism's physiological development across life stages based on empirically derived responses to environmental conditions. They estimate development rates and thresholds, often from controlled experiments, and calculate the number of successful life-cycle completions (LCC) per time period. This approach is particularly valuable for assessing the impact of climate change on vector populations because it incorporates the fine-scale effects of daily temperature variation on development rates. The study aims to develop and validate a phenology model for *Ae. aegypti*, applying it to explore changes in LCC intensity in response to past and projected climate changes globally, using daily climate inputs and considering RCP 4.5 and RCP 8.5 climate change scenarios.

The literature extensively documents the impact of climate change on vector-borne diseases. Studies have highlighted the potential for global climate change to facilitate the expansion or re-establishment of mosquito vector populations and the diseases they transmit into new or previously occupied regions. However, many of these studies rely on correlative methods, which may have inherent limitations when considering potentially invasive disease vectors. Mechanistic models, offering a more robust approach, are less common in this context despite their advantages in assessing responses to novel environments. Previous research on *Ae. aegypti* has used mechanistic models, focusing on temperature-sensitive population dynamics across life stages. While informative, these have rarely been integrated into large-scale distribution estimates to assess long-term environmental change. The use of phenology models, well-established in agriculture for invasive pest predictions, is novel in the context of human disease vectors, offering a unique perspective on assessing the impact of climate change on vector populations. Therefore, this study builds upon existing research by employing a phenology model to incorporate the fine-scale effects of daily temperature variation on key development rates and project LCC intensity over a long time period.

This study developed a spatially explicit phenology model for *Aedes aegypti*, incorporating each life stage from egg to oviposition. The model utilized daily minimum and maximum near-surface air temperature and precipitation data from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) dataset, with a spatial resolution of 0.25° and spanning 1950-2100. The model calculates the number of life-cycle completions (LCC) per time step based on empirically derived temperature thresholds and development rates obtained from the literature. The model incorporates temperature-dependent development rates for each stage (egg hatching, immature development, blood feeding, oviposition), considering lower and upper temperature thresholds (cold-kill and heat-kill conditions, respectively) that affect development success. A precipitation constraint was also included to reflect the water requirements of the aquatic stages. The model's output is a development intensity index representing the theoretical probability of occurrence, abundance, and establishment potential given successful dispersal. Four commonly used GCMs from CMIP5 were used for each RCP scenario (RCP 4.5 and RCP 8.5), and outputs were averaged to account for inter-model variability. Model validation was performed at two levels: global-scale validation against a large dataset of *Ae. aegypti* occurrence records and a local-scale validation against abundance data from a study in Mexico. The global-scale validation used AUC, Kappa, and Pearson correlation to assess the model's ability to predict the species' presence. The local-scale validation correlated model outputs with observed abundances. Global spatial and temporal trends in LCC intensity were investigated by averaging LCC over various spatial (continental, climate type, latitudinal bands) and temporal scales. Invasion frontiers were tracked by identifying the contour line representing a threshold LCC (≥10), reflecting areas suitable for population establishment. Finally, seasonal trends in LCC intensity were assessed using monthly LCC estimates across latitudinal bands, with seasonal Kendall trend tests used to analyze long-term trends and Sen's slope estimator used to quantify the rate of change.

The phenology model successfully reproduced the known spatial patterns of *Ae. aegypti* occurrence and local-level abundance. Global LCC increased from 7.08 per year in the 1950s to 7.62 per year at the turn of the century (2000-2004 average), indicating a 7% increase in global suitability. Future projections show this trend accelerating, with a predicted increase of 17.1% (RCP 4.5) and 24.3% (RCP 8.5) by the 2050s. This translates to an acceleration of suitability increase from 1.5% per decade (1950s-2000s) to 3.2% (RCP 4.5) and 4.4% (RCP 8.5) per decade (2000s-2050s). The model showed that tropical areas historically had the highest LCC, and these are predicted to see the greatest increases in the future. In contrast, temperate regions show marked gains while arid, polar, and boreal regions show low suitability with little change. Analysis of invasion frontiers demonstrated that in the USA, the invasion frontier has already advanced in southeastern states, with further rapid expansion predicted for the future. In China, historically slow expansion is predicted to accelerate significantly under future climates, particularly impacting regions with recent dengue outbreaks. In Europe, while current suitability remains low, southern margins are projected to become suitable for *Ae. aegypti* by the 2030s. Seasonal analysis revealed that both peak LCC and the duration of favorable periods will increase under future climate change scenarios. The strongest gains in peak LCC will occur in the tropics and subtropics, while middle latitudes will show more significant increases in the duration of favorable periods. This increase in the duration of favorable periods may be crucial for mosquito establishment in historically less favorable regions.

The findings strongly suggest that climate change is increasing the suitability of environments for *Ae. aegypti* development, leading to accelerated invasion potential and increased disease transmission risk globally. The significant increases in LCC, particularly the accelerated increase projected for the future, highlight the potential for widespread expansion of this vector. The analysis of invasion frontiers indicates that regions like the southeastern USA, and parts of China and Southern Europe, are likely to experience rapid increases in *Ae. aegypti* suitability. The observed increases in both peak LCC and the duration of favorable periods further amplify the risk, indicating that not only will mosquito populations potentially be higher in favorable periods, but they may also be present for a longer time of the year. While the model's predictions are robust and well-validated, some discrepancies exist between predictions and observations, particularly in regions with historically low LCC but currently occupied areas. This could be attributed to factors such as management interventions, environmental stochasticity, dispersal constraints, microclimates, species interactions, and limitations in the model's representation of mosquito life-history responses. Future research should focus on integrating these additional factors into the model to refine its accuracy. The study underscores the critical need to consider the effects of climate change on vector-borne diseases, highlighting the urgency for proactive mitigation and adaptation strategies. The rapid expansion of *Ae. aegypti* and the increasing suitability of previously unsuitable regions will likely lead to increased global arboviral disease transmission risks, requiring both focused public health intervention and addressing underlying climate change.

This study demonstrates the accelerating invasion potential of *Aedes aegypti* under climate change using a novel phenology model. The model's predictions show significant increases in environmental suitability and invasion speed in multiple regions globally, highlighting the growing risk of arboviral diseases. Future research should focus on refining the model by incorporating additional factors like management interventions, dispersal limitations, and microclimates. The findings emphasize the urgent need for proactive strategies to mitigate the health impacts of climate change on vector-borne disease transmission.

The model, while robust, has some limitations. It may underestimate establishment thresholds in certain regions due to factors not explicitly included in the model. For instance, management interventions, environmental stochasticity, dispersal constraints, microclimates, species interactions, and finer details of mosquito life-history responses, such as variability in growing degree days requirements or responses to fluctuating temperature regimes, were not fully accounted for. These represent opportunities for future model improvement. Additionally, the model focuses on the suitability for mosquito development, not necessarily disease transmission. While increased mosquito populations increase the potential for disease outbreaks, additional modeling of transmission dynamics is required to fully assess the health impacts of this expanding vector population.