Plants display a variety of specialist adaptations to particular environments. Beach ecotypes are especially common due to a suite of abiotic stressors that act as environmental drivers of adaptation. These stressors include low water availability, salt deposition, low nutrient soil, burial by sand, and strong wind; associated phenotypes include dwarfism, succulence, cuticles and salt excluding glands, rapid growth, light coloration, reflective leaves, transpirational cooling, long hypocotyls, a high density of trichomes, and large seed size. The model organism Arabidopsis thaliana is found in a variety of environments across the globe, including coastal environments. As a predominantly selfing species, A. thaliana is a good candidate for the production of local ecotypes. Additionally, high-resolution genomic data and ease of phenotyping make A. thaliana an attractive choice for the experimental exploration of adaptation to beach environments. I use several complementary approaches to investigate coarse sand beach populations of A. thaliana on the Baltic Sea coast of southern Sweden. In the first chapter, I investigate the genetic architecture of adaptation to beaches in A. thaliana using two types of genomic scans. First, I perform an unbiased genome wide scan using Population Branch Statistic (PBS) to identify regions of the genome under selection in the beach population of interest. I apply additional population genetic metrics to the top-scoring regions in order to identify putative genes involved in adaptation. Second, I focus on three phenotypes that differentiate beach and inland populations and are characteristic of beach plants: hypocotyl length, trichome density, and seed weight. I perform genome wide association mapping using 298 Swedish lines for each phenotype and identify significant associations with each phenotype, including a region of chromosome 4 near the candidate gene SPA2 significantly associated with hypocotyl length and a region on chromosome 2 strongly associated with trichome density and the candidate genes TCL1/TCL2/ETC2. Combining the two approaches described above, I ask whether the most closely associated SNPs for each phenotype have a higher average PBS value than expected. This test asks whether SNPs associated with phenotypes of interest show evidence of adaptation in the beach population, something I find to be true for hypocotyl length and seed weight, but not trichome density. Thus, hypocotyl length and seed weight appear to be under selection in the beach population. In the second chapter, I investigate hypocotyl length more deeply. I demonstrate that plants from coarse sand beaches have significantly longer hypocotyls than inland and fine sand beach conspecifics and identify a genetic variant accounting for 18% of the variance in hypocotyl length in the mapping population. I confirm the correlation between this genetic variant and hypocotyl length in two additional populations, a global set of lines from the RegMap panel and in a set of Bay x Sha RIL lines, and demonstrate that this genetic variant increases seedling emergence frequency when seeds are buried. I conclude that this genetic variant is in linkage with a gene involved in hypocotyl elongation, possibly SPA2 (a suppressor of phytochrome A important for growth in darkness) and that this gene confers a fitness advantage when seeds are buried. In the third chapter, I test the hypothesis that beach lines are more tolerant of drought and salt stress than inland conspecifics. I use total seed production as a proxy for lifetime fitness and quantify reductions in fitness under several levels of drought and salt spray stress. I detect no significant response to salt spray. While drought has a negative effect on fecundity, beach lines do not consistently show different responses to drought than inland lines. This set of complementary studies approaches adaptation to beaches from genotypic, phenotypic, and environmental perspectives. I identify phenotypes and genetic regions important for adaptation in this population of A. thaliana, as well as an environmental driver of selection. Additionally, I introduce a new method of combining types of genomic data to yield information about phenotypes under selection in populations of interest.