Quantifying environmentally induced variation using 10^5 replicates of Luria and Delbruck

Variation is the fuel for Darwinian natural selection. The role of the environment in shaping variation has been debated and studied quantitatively since at least Luria and Delbrück’s 1943 experiment. Through an 87 replicate assay, they produced a statistical distribution and argued that variation is preexisting and not induced by the environment. Yet, in the years since, we have learned of many cases of environmentally induced variation such as CRISPR-based bacterial immunity, and stress induced mutagenesis in microbes and in cancer. It has been shown that 87 replicates of Luria and Delbruck’s experiment do not have the statistical power needed to resolve induced variation that is known to coexist with pre-existing variation in these biological examples. Here we benchmark a quantitative and sensitive method, HiDenSeq, to test for induced variation that may coexist with pre-existing variation. We leverage barcodes to carry out ~ 10^5 replicates of the LD in one pot and we measure small deviations from the LD distribution to measure any induced variation mechanism. Through selection using a drug (canavanine) during UV exposure in S. cerevisae, we find that the heavy tailed distribution is modified, and quantify the amount of pre-existing vs induced variation. Our results open up a phenomenological route to quantifying modes of variation simply from the statistics of adapting lineages.