Summary
Pathogens compete for hosts through patterns of cross-protection conferred by immune responses to antigens1. To evade the immune system, several pathogens possess hypervariable multi-copy gene families encoding large pools of antigenic variants2. In the malaria parasite Plasmodium falciparum, var genes encoding the major blood-stage antigen, PfEMP1, are one such family, with repertoires of 50-60 genes in an individual parasite3, and tens of thousands of gene variants in local populations of high transmission regions4 generated through ectopic recombination5,6 and mutation. Deep sampling of asymptomatic children in a West African population has recently revealed non-random structure in this enormous diversity, with extremely low genetic overlap among var gene repertoires even in multi-genome P. falciparum isolates7. This is consistent with previous strain theory, postulating co-existence of discrete non-overlapping pathogen strains8–10 as a result of selection against recombinants due to cross-immunity. However, the combinatorial complexity of the var system in high transmission regions remains beyond the reach of existing strain theory, and neutral models do not yet exist to differentiate signatures of immune selection from those of pure transmission dynamics. Here, we present theory to identify signatures of immune selection that reveal non-neutral structures both in simulated systems and in an extensively sampled population in Bongo District (BD), Ghana. We develop two neutral models that encompass malaria epidemiology but exclude competitive interactions between parasites. We then present an analytical framework based on genetic similarity networks appropriate for the recombination mechanisms that generate diversity. Network patterns harbor distinctive signatures of selection structuring antigenic diversity in this highly recombinant gene family through frequency-dependent competition for hosts. This unique population structure created by non-neutral forces, likely immune selection, underlies the ability of the parasite to multiply infect individual hosts with long lasting chronic infections11, constituting a large reservoir of transmission in highly endemic regions of Africa. To be successful, elimination strategies must move beyond prevalence of infection and target this diversity which is at the heart of malaria transmission and pathology.