The primary aim of the current study was to determine methods for sampling temperatures within transmission environments and to propose these as a framework for better understanding local transmission ecology. The empirical data presented serve as a pilot study for a more extensive longitudinal monitoring programme and so are relatively limited in scope (i.e. they are not themselves intended to provide an exhaustive evaluation of the transmission environment in the urban slums in Chennai). Nonetheless, the study illustrates the benefits of examining the variation in temperature between different potential mosquito resting habitats and potential implications for malaria transmission.
Most studies that consider the role of temperature in malaria transmission use temperatures reported by local weather stations. The current study revealed that the temperatures within the local transmission sites were warmer and more varied than those recorded by the weather station at the airport. The temperature loggers were located in a densely packed urban environment that showed less cooling at night (Figure 2B). Such ‘urban heat islands’ have been reported elsewhere in the literature [41, 42] and tend to experience higher average daily temperatures than surrounding areas via effects on night-time temperatures.
Within the transmission sites, indoor temperatures remained warmer and were more stable than those recorded for outdoor environments. This type of thermal buffering has been reported in other studies, both in general terms  and with specific reference to malaria transmission [3, 44]. With the exception of the overhead tank, which proved to be something of an extreme environment, differences in mean temperatures between microhabitats were relatively small. Even so, variation between environments led to differences in the predicted EIPs of 1–4 days for both P. falciparum and P. vivax. Adding the effects of daily temperature variation had relatively little effect because, in contrast to other parts of the world , the DTRs themselves were relatively small and consistent across habitats.
Anopheles stephensi exhibits both endophilic (indoor resting) and exophilic (outdoor resting) behaviour. The current study took a broad snap shot of temperatures available for adult mosquitoes, but it is possible that An. stephensi only utilizes a subset of these environments. Further, temperatures will likely vary within structure types (i.e. at different positions within a single house type) and so the temperatures experienced by mosquitoes could depend on the subtleties of the precise resting position. There is some indication that adult anopheline mosquitoes can avoid the warmest locations , but there is little evidence for precise behavioural thermoregulation . Additionally, it has been shown that expression of mosquito heat-shock proteins increase in response to thermal stress  and these proteins have been shown to interact with Plasmodium development . The extent to which these interactions would affect EIP in these environments is unclear, but should be considered in future studies. Further study on the activity patterns of An. stephensi in this urban environment and resting preferences within and between structure types is clearly required. For the current study there was no systematic mosquito sampling, although case data from the malaria clinics indicated there was active malaria transmission in these areas during the study period and immature forms of An. stephensi were found in the field sites, indicating that An. stephensi was utilizing some portion of the environments sampled. Precise behavioural data coupled with appropriate high resolution environmental data would refine understanding of the temperatures mosquitoes experience during parasite development. However, very few studies characterize mosquito resting behaviour in detail or couple entomological measures with site-specific estimates of microclimate.
Local meteorological station data provide a single measure of temperature and hence, generate a single estimate of temperature-dependent traits such as EIP. In contrast, the multiple data loggers placed within the transmission environment provide a measure of the temperature envelope in which mosquitoes live. In the current study, the EIPs based on the weather station data did not consistently align with the EIPs based on the ensemble mean temperatures derived from the different microhabitat types and clearly failed to capture the enormous amount of potential variation that exists within the transmission setting. The use of a distribution of temperatures does not make predictions of life history traits more ‘precise’, but it does make them more accurate in the sense that they represent the actual environment where mosquitoes rest and transmission occurs. If a single measure is required, the use of an ‘ensemble mean’ based on the average temperatures recorded across the local microhabitats should provide a more robust characterization of the transmission environment than the remote meteorological station data.
This study emphasizes the complexity of the thermal environment and its equally complex interaction with parasite development. Understanding transmission dynamics requires some consideration of this complexity. Even small absolute changes in EIP of 1–3 days can have marked impacts on transmission risk. The average daily probability of survival for An. stephensi has been estimated as 0.810 . With an EIP of 10 days this means that approximately 12% of adult mosquitoes would live long enough to be able to transmit malaria. Increasing EIP to 12 days reduces this percentage to 8.5%, while reducing EIP to 8 days increases the value to about 18.5%. All else being equal, these changes would alter transmission intensity by approximately −45% to +50% .
Gaining knowledge of the temperatures experienced by mosquito vectors in the field is an important step towards a better understanding of temperature as a driver of malaria transmission. Data on the natural thermal environments of vectors are also valuable for contextualizing laboratory work on mosquito-parasite interactions. The majority of laboratory studies on vector competence and host-parasite interactions are conducted at a constant temperature of around 26°C. While only a limited study, the data from the urban transmission sites in Chennai suggest mosquitoes rarely encounter such temperatures and are subject to both higher mean temperatures and daily temperature fluctuations.