The Forever Experiment

What 80,000 Generations of E. coli Teach Us About the Creativity of Evolution

In February 1988, the landscape of evolutionary biology was forever changed not by a fossil dig or a deep-sea expedition, but by the swirling of twelve glass flasks each inoculated with cells from a single bacterial clone. In a laboratory at the University of California, Irvine, Richard Lenski set in motion a deceptively simple ritual: every twenty-four hours, 100 microliters from each of twelve Escherichia coli cultures would be transferred into 9.9 milliliters of fresh medium. This medium contained a meager ration of just 25 milligrams of glucose per liter—a subsistence diet that forced the bacteria to reach a stationary density of roughly 5 x 10⁷ cells per milliliter.

I played a small role in the initiation of this project. I had met Rich Lenski in 1982, when he was a post-doc in Bruce Levin’s lab in Amherst, Massachusetts. Bruce had got Rich into the ways of experimental evolution back in those days, working with bacteria. But a point of contention among the three of us was the use of replication. I have long been a strong advocate for as much replication of evolving populations as possible. When Rich successfully recruited me to join him at the University of California, Irvine, in 1987, our primary topic of contention was how much replication to use in studies of long-term experimental evolution. I was trying to get him to use 30 to 50 E. coli populations, instead of the three-fold replication he had previously used. Rich decided to use relentlessly swirled Erlenmeyer flasks for the project, and his swirling machine could only handle twelve such flasks. So twelve E. coli populations were what he started with.

What began as a modest study of adaptation has since morphed into the Long-Term Evolution Experiment (LTEE), a forever laboratory project that has now surpassed 80,000 generations. It is a profound attempt to “play the tape” of evolution by natural selection in parallel twelve times over, challenging evolution by natural selection to reveal whether it is a predictable machine or a chaotic dance of historical accidents. But this was with one central qualification: any type of sexual recombination was absolutely forestalled in the design of the LTEE. This reduced the evolutionary genetic process to the interaction of mutation and selection solely.

Microbial Time Travel

The LTEE grants us a front-row seat to the theater of deep time through its “frozen fossil record.” Unlike the mineralized remains of the Paleolithic, these bacterial fossils are still alive. Every 500 generations, Lenski and his team preserve samples of the twelve populations in a glycerol-based suspension at -80°C. These ghosts of ancestors past can be thawed, revived, and placed in direct competition with their own descendants.

This capability transforms the study of evolution from a descriptive historical account into a rigorously measurable quantitative process of experimental evolution. In the LTEE, we can literally race the past against the present to quantify fitness gains. Furthermore, because these strains are preserved, we can interrogate them using modern genomic tools—such as whole-genome sequencing—that were nothing more than science fiction when the first flasks were seeded in 1988. In the LTEE, the past is never truly dead; it is merely waiting in the freezer to be called back into the light.

The One-in-Twelve Miracle: Breaking the Citrate Barrier

For 31,000 generations, the twelve populations lived in parallel, essentially ignoring the citrate that sat alongside their glucose in the DM25 medium. Citrate was intended only as a chelating agent, as E. coli is defined by its inability to consume the substance in the presence of oxygen. Then, in a single flask, the impossible happened. One lineage evolved the ability to grow aerobically on citrate (the Cit+ trait), leading to a massive spike in population density as it tapped into this previously untouched reservoir of energy.

This breakthrough provided a masterclass in historical contingency. Analysis revealed that the Cit+ trait was not the result of a single, lucky mutation. Instead, it required a rare tandem duplication that rearranged structural and regulatory sequences—but this innovation only functioned because the lineage had been primed by a specific sequence of prior mutations that had swept through that particular population.

Reflecting on this “one-in-twelve” event forces us to look beyond the flask to the very nature of life in the cosmos. If microbial evolution is so deeply contingent on a unique history of priming mutations, then the path of microbial evolution on Earth might be a singular, unrepeatable fluke, at least until the advent of Mendelian sex. [See, for example, the evolutionary genomics of sexual evolution in the earlier post, “The Tenet Effect.”] The citrate miracle suggests that the arrival of innovation in microbial evolution is not just about the final key turning the lock, but about the slow, invisible work of a succession of previous mutations forging the lock itself.

The Never-Ending Climb: Adaptation Has No Ceiling

Early in the experiment, it was assumed the bacteria would eventually reach “static optimization”—a state of perfection where they were so well-adapted to their simple environment that no further improvement was possible. Indeed, early data showed a hyperbolic curve: a frantic rush of adaptation followed by a rough plateau.

However, as the generations ticked into the tens of thousands, a different pattern emerged. Fitness, it turns out, continues to increase indefinitely. Even in a microcosm that never changes, evolution by natural selection finds ways to climb higher. In these populations, the rate of beneficial mutations is estimated at 1.7 x 10⁻⁶ per genome, which translates to roughly 850 daily miracles across each population. While most are lost to the daily 100-fold dilution, 5 to 10 survivors persist each day as potential fuel for the upward climb of adaptation. Even in a stationary environment, the potential for adaptation is seemingly infinite. Life never finishes adapting; it finds ever-finer margins to exploit.

Twelve Parallel Universes: The Tension Between Pattern and Chaos

The LTEE’s twelve replicates function as parallel universes, allowing us to see when evolution follows stable laws and when it indulges in idiosyncrasy. We see striking parallelism in cell size; all twelve populations evolved cells significantly larger than their ancestor. Yet, a fascinating paradox emerged: while the cells grew larger, their genomes actually shrank by an average of 1.4%. This leaner yet larger pattern suggests that evolution is a master of structural efficiency, at least efficiency by its own standards.

However, beneath these parallel trends lie undertales of divergence. At the genetic level, while selection often targeted the same genes (gene-level parallelism), the exact mutations were rarely the same. Microbial experimental evolution is a disciplined process that can arrive at similar functional solutions, but it does so through a variety of specific genetic substitutions, ensuring that no two populations—as parallel universes—are ever truly identical.

Genetic Speed Demons: When Evolution Goes into Overdrive

By generation 50,000, some of the populations had evolved into “mutators.” These strains suffered from defects in DNA proofreading, resulting in hypermutability. These speed demons accumulated point mutations at a staggering rate, often driven by specific transition or transversion biases depending on which part of the DNA repair machinery failed.

This led to chaotic genetic draft, or hitchhiking. In these “noisy” genomes, beneficial “driver” mutations were so frequent that they pulled along hundreds of neutral or even slightly harmful “passenger” mutations. This created a crowded evolutionary race defined by “Clonal Interference”, where multiple beneficial lineages compete so fiercely that they actually slow each other down. For the scientist, these genomes are a headache—the genomic noise of hitchhikers obscures the genomic signal of adaptation. Eventually, a subset of these lineages evolved compensatory changes to slow their mutation rates, proving that while speed can provide a short-term edge, it can impose a long-term genomic cost.

The Myth of the “Simple” Environment

We often assume that a simple environment—such as a glass flask with a single sugar—leads to a simple ecosystem. The LTEE proves otherwise. Despite the meager 25 mg/L glucose concentration, some populations refused to remain monolithic. Through negative frequency-dependent selection, their flasks became tiny, thriving ecosystems.

Through cross-feeding, divergent lineages began to coexist stably. One lineage might specialize in the primary glucose, while another lineage in the same flask evolves to thrive on the metabolic byproducts secreted by the first. Even when we give organisms almost no environmental complexity, they can find ways to create niches and ecotypes, demonstrating that ecological complexity is an emergent property of life itself.

The Elegance of the Long View

The Long-Term Evolution Experiment with E. coli is a profound testament to the infinite creativity of the biological process. It shows us that evolution is both a disciplined sculptor, carving repeatable patterns of increasing fitness across independent lineages, and a restless innovator, capable of shattering metabolic barriers.

To grasp the true scale of this work, one must look past the glass and the glucose. 80,000 generations represents a temporal journey of roughly two million years in human terms. Our own ancestors from 80,000 generations ago, Homo habilis, were striking rocks together to create the earliest stone tools on the African savannah. Over the span of generations it took for us to build civilizations, split the atom, and sequence the genome, the LTEE bacteria have been performing their own quiet miracles in a few milliliters of salt and sugar.

LTEE Redux: Back to their Origin

On Friday May 8, 2026, Rich Lenski returned to the University of California, Irvine, by Zoom, to teach a class in my graduate course on experimental evolution. He got to tell the long story of his Forever Experiment to old colleagues and friends, like myself and Joseph Graves, as well as introducing his work to new students in the field of experimental evolution. His work is a landmark in the history of evolutionary biology, and a pointer for the future of our field. In some respects, I believe that his work is like Galileo’s pioneering experiments on terrestrial mechanics. Someday, I contend, it will be seen as a starting point for the long project of making biology into a properly scientific field.

Postscript

Also this May, Rich Lenski and I — together with Margarida Matos and Joseph Graves — have published a book together: A Primer for Experimental Evolution (World Scientific). It’s a relative rarity in that it introduces both microbial and Mendelian experimental evolution, chiefly E. coli and Drosophila, respectively. All four of us spent time together at the University of California, Irvine, in the late 1980s and early 1990s, along with other important experimental evolutionists like Larry Mueller and Adam Chippindale. So in some ways the book embodies what could be called “The Irvine School” of experimental evolution, now scattered across other institutions in multiple countries.