11Genome Wide Association Studies (GWASs) have the potential to reveal the genetics of microbial 12 phenotypes such as antibiotic resistance and virulence. Capitalizing on the growing wealth of bacterial 13 sequence data, microbial GWAS methods aim to identify causal genetic variants while ignoring 14 spurious associations. Bacteria reproduce clonally, leading to strong population structure and genome-15 wide linkage, making it challenging to separate true "hits" (i.e. mutations that cause a phenotype) 16 from non-causal linked mutations. GWAS methods attempt to correct for population structure in 17 different ways, but their performance has not yet been systematically evaluated. Here we developed a 18 bacterial GWAS simulator (BacGWASim) to generate bacterial genomes with varying rates of 19 mutation, recombination, and other evolutionary parameters, along with a subset of causal mutations 20 underlying a phenotype of interest. We assessed the performance (recall and precision) of three 21 widely-used univariate GWAS approaches (cluster-based, dimensionality-reduction, and linear mixed 22 models, implemented in PLINK, pySEER, and GEMMA) and one relatively new whole-genome 23 elastic net model implemented in pySEER, across a range of simulated sample sizes, recombination 24 rates, and causal mutation effect sizes. As expected, all methods performed better with larger sample 25 sizes and effect sizes. The performance of clustering and dimensionality reduction approaches to 26 correct for population structure were considerably variable according to the choice of parameters. 27 Notably, the elastic net whole-genome model was consistently amongst the highest-performing 28 methods and had the highest power in detecting causal variants with both low and high effect sizes. 29 Most methods reached good performance (Recall > 0.75) to identify causal mutations of strong effect 30 size (log Odds Ratio >= 2) with a sample size of 2000 genomes. However, only elastic nets reached 31 reasonable performance (Recall = 0.35) for detecting markers with weaker effects (log OR ~1) in 32 smaller samples. Elastic nets also showed superior precision and recall in controlling for genome-33 wide linkage, relative to univariate models. However, all methods performed relatively poorly on 34 highly clonal (low-recombining) genomes, suggesting room for improvement in method development. 35 These findings show the potential for whole-genome models to improve bacterial GWAS 36 3 performance. BacGWASim code and simulated data are publicly available to enable further 37 comparisons and benchmarking of new methods. 38 39 Author summary: 40 Microbial populations contain measurable phenotypic differences with important clinical and 41 environmental consequences, such as antibiotic resistance, virulence, host preference and 42 transmissibility. A major challenge is to discover the genes and mutations in bacterial genomes that 43 control these phenotypes. Bacterial Genome-Wide Association Studies (GWASs) are family of 44 methods to statistically assoc...