Contrasting selection at multiple life stages maintains divergent adaptation between sunflower ecotypes

Conspecific populations living in adjacent, but contrasting, microenvironments represent excellent systems for studying natural selection. These systems are valuable because gene flow maintains genetic homogeneity except at loci experiencing strong, divergent selection. A history of reciprocal transplant and common garden studies in such systems, and a growing number of genomic studies, have contributed to understanding how selection operates in natural populations. While selection can vary across different fitness components and life stages, few studies have investigated how this ultimately affects allele frequencies and persistence of divergent populations. Here, we study two sunflower ecotypes in distinct, adjacent habitats by combining demographic models with genome-wide sequence data to estimate fitness components, absolute fitness, and allele frequency change at multiple life stages. This framework allows us to demonstrate that only local ecotypes experience positive population growth (lambda>1) and that the maintenance of divergent adaptation is mediated via habitat- and life stage-specific selection. We identify genetic variation, significantly driven by loci in chromosomal inversions, associated with different life history strategies in neighbouring ecotypes that optimize different fitness components and contribute to the persistence of each ecotype in its respective habitat.