Questions on the processes structuring antigenic and associated genetic diversity in pathogens have been at the center of both strain theory and phylodynamics. Competition for hosts mediated by specific immunity to particular variants establishes a form of frequency-dependent competition that confers an advantage to the rare and is known as immune selection. Although this process which maps to stabilizing competition based on ‘niche differences’ is well understood in models of low to medium complexity, its role in generating strain structure remains an open question for hyper-diverse systems. This challenge applies to a number of pathogens such as Plasmodium falciparum and Trypanosoma brucei exhibiting enormous antigenic diversity locally as the result of multigene families and extensive genetic recombination. Here, we ask whether immune selection plays an important role in structuring P. falciparum populations of highly endemic regions into strains from the perspective of the multigene family known as var. We propose the combination of network analyses of genetic similarity with agent-based models that track immune memory in each host, to identify signatures of dominant processes underlying strain structure under reticulate evolution. We illustrate these methods with empirical molecular data from West African populations.
Results/Conclusions
Three kinds of networks of genetic similarity corresponding to the respective levels of genes, repertoires of genes, and the multilayer structure of repertoires over time, all reveal clear distinctive properties differentiating immune selection from neutral alternatives. Neutrality refers here to models that retain the demography of the transmission system but eliminate specific competition, including complete absence of immune memory and generalized protection from repeated infection. Application of these network analyses to empirical isolates from asymptomatic infections from local populations in Ghana provide unequivocal evidence for an important role of immune selection. Thus, this form of stabilizing competition enables the coexistence of a large number of gene variants and strains, with patterns of limiting similarity that differ from typical niches at lower dimensionality. Consequences are discussed for malaria control and for epidemiological models which currently do not include explicit consideration of antigenic diversity.