Predicting the response of disease vectors to global change: The importance of allometric scaling.

The distribution of disease vectors such as mosquitoes is changing. Climate change, invasions and vector control strategies all alter the distribution and abundance of mosquitoes. When disease vectors undergo a range shift, so do disease burdens. Predicting such shifts is a priority to adequately prepare for disease control. Accurate predictions of distributional changes depend on how factors such as temperature and competition affect mosquito life-history traits, particularly body size and reproduction. Direct estimates of both body size and reproduction in mosquitoes are logistically challenging and time-consuming, so the field has long relied upon linear (isometric) conversions between wing length (a convenient proxy of size) and reproductive output. These linear transformations underlie most models projecting species' distributions and competitive interactions between native and invasive disease vectors. Using a series of meta-analyses, we show that the relationship between wing length and fecundity are nonlinear (hyperallometric) for most mosquito species. We show that whilst most models ignore reproductive hyperallometry (with respect to wing length), doing so introduces systematic biases into estimates of population growth. In particular, failing to account for reproductive hyperallometry overestimates the effects of temperature and underestimates the effects of competition. Assuming isometry also increases the potential to misestimate the efficacy of vector control strategies by underestimating the contribution of larger females in population replenishment. Finally, failing to account for reproductive hyperallometry and variation in body size can lead to qualitative errors via the counter-intuitive effects of Jensen's inequality. For example, if mean sizes decrease, but variance increases, then reproductive outputs may actually increase. We suggest that future disease vector models incorporate hyperallometric relationships to more accurately predict changes in mosquito distribution in response to global change.