Fine-scale variation in microclimate across an urban landscape shapes variation in mosquito population dynamics and the potential of Aedes albopictus to transmit arboviral disease

Abstract

Most statistical and mechanistic models used to predict mosquito-borne disease transmission incorporate climate drivers of disease transmission by utilizing environmental data collected at geographic scales that are potentially coarser than what mosquito populations may actually experience. Temperature and relative humidity can vary greatly between indoor and outdoor environments, and can be influenced strongly by variation in landscape features. In the Aedes albopictus system, we conducted a proof-of-concept study in the vicinity of the University of Georgia to explore the effects of fine-scale microclimate variation on mosquito life history and vectorial capacity (VC). We placed Ae. albopictus larvae in artificial pots distributed across three replicate sites within three different land uses-urban, suburban, and rural, which were characterized by high, intermediate, and low proportions of impervious surfaces. Data loggers were placed into each larval environment and in nearby vegetation to record daily variation in water and ambient temperature and relative humidity. The number of adults emerging from each pot and their body size and sex were recorded daily. We found mosquito microclimate to significantly vary across the season as well as with land use. Urban sites were in general warmer and less humid than suburban and rural sites, translating into decreased larval survival, smaller body sizes, and lower per capita growth rates of mosquitoes on urban sites. Dengue transmission potential was predicted to be higher in the summer than the fall. Additionally, the effects of land use on dengue transmission potential varied by season. Warm summers resulted in a higher predicted VC on the cooler, rural sites, while warmer, urban sites had a higher predicted VC during the cooler fall season.

Fig 1. An impervious surface map of Athens-Clarke County, Georgia, U.S.

Spatial pixels (30 m²) were binned according to proportion of impervious surface, with high, intermediate, and low proportion of impervious surface corresponding to urban (red), suburban (blue), and rural (white) sites, respectively. From this map, we selected three sites (black dots, 30 m²) from each land use class for the artificial pot experiments.

Fig 2. Daily variation in temperature and relative humidity.

Ambient mean (solid lines), minimum (lower dotted lines), and maximum (upper dotted lines) daily temperature (A) and relative humidity (B) were recorded by data loggers across the duration of both experiments on urban (red), suburban (blue), and rural (black) sites and by the local weather station (green) on campus.

Fig 3. Local weather station data over or under-predict metrics of mosquito relevant microclimate.

Differences between daily mean, minimum, and maximum values for temperature and relative humidity recorded by data loggers on urban (red), suburban (blue), and rural (white) sites in the summer (A, C) and fall (B, D). Mean and standard errors associated with each land use category reflect estimated marginal means and standard errors from mixed effects models (random factor: pot nested within site) estimating the effects of land use on average daily mean, minimum, and maximum temperatures and relative humidity, while means and standard errors for the weather station data represent data collected from a local weather station at the University of Georgia, Athens GA U.S.A. over the course of each experiment conducted in the summer and fall 2015.

Fig 4. Season and land use both affect the probability of adult emergence.

The cumulative percentage of mosquito adults emerging across urban (red), suburban (blue), and rural (black) sites in both the summer (solid lines) and fall (dashed lines).

Fig 5. Season and land use have qualitatively different effects on mosquito per capita growth rates and dengue transmission potential.

The effects of land use on mosquito body size (A), per capita mosquito growth rates (B), and relative vectorial capacity, or transmission potential (C) in the summer (red bars) and fall (yellow bars). Means and standard errors represent estimated marginal means and standard errors from our mixed model analysis.