Madeline A. Peters ^1,2 Aaron A. King ^2,3 Nina Wale ^1*

• ¹ Michigan State University, East Lansing, United States
• ² University of Michigan, Ann Arbor, Michigan, United States
• ³ Santa Fe Institute, Santa Fe, New Mexico, United States

For decades, mathematical models have been used to understand the course and outcome of malaria infections (i.e., infection dynamics) and the evolutionary dynamics of the parasites that cause them. The extent to which this conclusion holds will in part depend on model assumptions about the host-mediated processes that regulate RBC availability, i.e., removal (clearance) of uninfected RBCs and supply of RBCs. Diverse mathematical functions have been used to describe host-mediated RBC supply and clearance in rodent malaria infections; however, the extent to which these functions adequately capture the dynamics of these processes has not been quantitatively interrogated, as in vivo data on these processes has been lacking. Here, we use a unique dataset, comprising time-series measurements of erythrocyte (i.e., mature RBC) and reticulocyte (i.e., newly supplied RBC) densities during Plasmodium chabaudi malaria infection, and a quantitative data-transformation scheme to elucidate whether RBC dynamics conform to common model assumptions. We found that RBC supply and clearance dynamics are not well described by mathematical functions commonly used to model these processes. Indeed, our results suggest said dynamics are not well described by a single-valued function at all. Furthermore, the temporal dynamics of both processes vary with parasite growth rate in a manner again not captured by existing models. Together, these finding suggest that new model formulations are required if we are to explain and ultimately predict the within-host population dynamics and evolution of malaria parasites.