Authors

Discussion

Climate change is reshaping the global landscape of vector‑borne diseases by altering the environmental conditions that determine vector abundance, survival, and pathogen development. Rising temperatures, increased variability, and more frequent extreme events are expected to shift the geographical distribution and seasonality of many diseases, including malaria, with complex and sometimes unpredictable consequences for transmission. Understanding the ecological mechanisms mediating these effects is therefore crucial for anticipating future risks and designing robust control strategies.

A key determinant of transmission is adult mosquito survival, which governs the probability that vectors live long enough to support the parasite’s extrinsic incubation period (EIP). Through an Africa‑wide survey of malaria‑control stakeholders, we showed that although survival and age structure are central to vectorial capacity, they remain largely absent from operational surveillance. While >70% of institutions conduct routine vector monitoring, <50% assess mosquito age, largely due to the labour‑intensive nature of dissections, limited technical expertise, and resource constraints. Age‑grading was consistently deprioritised relative to abundance, species identification, biting rates, and insecticide resistance, highlighting a critical operational gap.

We further demonstrated that existing age‑grading methods, despite their long use, show variable accuracy under realistic conditions. In a blinded validation of the ovarian-dissection based Polovodova method on >3000 Anopheles mosquitoes, we found high reliability for parity (κ ≈ 0.8) but steep declines in accuracy for later gonotrophic cycles, with systematic underestimation of older females. These insights provide empirical performance benchmarks and emphasise limitations when these methods are used for demographic inference or intervention evaluation.

To address these challenges, we developed mid‑infrared spectrometry coupled with machine learning (MIRS‑ML), a rapid and scalable tool that predicts mosquito species and age based on cuticular spectral signatures. Trained on >40,000 genetically and ecologically diverse specimens from three major African vectors, the model accurately resolved age classes in wild populations from Tanzania and Burkina Faso, demonstrating strong potential for integration into routine surveillance.

Building on this, we are developing MIRS‑ML approaches capable of estimating biological age—capturing the ageing process itself rather than environmental artefacts. This work is motivated by our experiments showing that realistic environmental variation strongly shapes mosquito development and survival. Manipulating temperature means, diurnal ranges, and humidity, we observed pronounced species‑ and stage‑specific responses: fluctuating temperatures and higher humidity consistently reduced adult survival in both An. gambiae