Charles Darwin saw evidence of evolution in all the animals and plants he studied. He boxed, bottled, and pressed thousands of specimens: mockingbirds and iguanas from the Galápagos, corals from the Indian Ocean, lichens and seaweeds from Tierra del Fuego, and even the skull of an extinct giant sloth from Argentina. He turned the gardens and greenhouses at Down House into personal laboratories for breeding pigeons, fertilizing orchids, and dissecting barnacles.

But one form of life was conspicuously absent from Darwin’s work: microorganisms.

Get the print magazineSubscribe for $100 to receive six beautiful issues per year.Subscribe

Darwin might be impressed at how these invisible creatures have come to reinforce and shape his theory. Over the past four decades, scientists have used microbes to reproduce evolution under laboratory conditions and test hypotheses about how natural selection works, in evolution experiments that have led to surprising new discoveries.

Natural selection is driven by a few basic processes: births, deaths, mutation, selection, and competition. Individuals better adapted to their environment tend to survive and reproduce more successfully than their competitors, passing their advantageous traits down to their offspring. Over generations, beneficial traits increase in frequency in the population until traits that were once found in just a few are shared by many.

In the large organisms Darwin studied, natural selection happens over many thousands of years. This is because these species’ reproductive cycles are long and those with the most adaptive genes do not have many more children than those with the least. More time between generations means more time is needed for genes to spread through a population. However, in species with shorter generation times, it can progress much more quickly.

For example, elephants have a generation time of 20 to 30 years, while the fruit fly grows from egg to adult in a mere ten days. Tens of generations of fruit flies can pass in a month or two, and one elephant takes up as much space in a lab as millions of fruit flies. Geneticists have a long and successful history using flies to study natural selection, but large-scale evolutionary changes, such as the origin of new functions and new species, typically require thousands of generations. Even with fruit flies, experiments attempting to capture this spread of evolutionary time could take centuries.

William Dallinger, a minister and amateur scientist, hypothesized in 1880 that microorganisms might make evolution experiments possible. A population of E. coli can experience seven to ten generations overnight. An individual E. coli bacterium is about 100 times smaller than the width of a human hair, and half a cup of broth can hold billions of them. Larger populations contain more genetic diversity than smaller ones, thanks to random mutations during the DNA replication process. They are therefore more likely to include at least one individual with mutations that make them better able to survive a changing environment.

Dallinger’s incubator.

Image

Journal of the Royal Microscopical Society.

Dallinger tested his hypothesis by growing water-dwelling microbes called flagellates in a custom-built copper incubator, exposing them to progressively hotter and hotter water. After much trial and error, his evolved flagellates were thriving at temperatures lethal to ordinary ones (and when they were put back into lukewarm water, they died). Today’s evolution experiments operate on similar principles.

The longest-running and most celebrated of modern evolution experiments is the appropriately named Long-Term Evolution Experiment (LTEE). Started by Richard Lenski in 1988 at the University of California, Irvine, and continuing in the hands of Jeffrey Barrick at the University of Texas at Austin, the LTEE has been running nearly continuously for 80,000 generations of E. coli over nearly 40 years. This is equivalent to two million years of human evolution.

The experiment began when 12 genetically identical populations of E. coli were grown in liquid medium. Every day since then, one percent of the previous day’s culture has been transferred into fresh medium. The medium is a dilute sugary solution limited in glucose, which E. coli uses as its primary carbon source. After about seven generations the glucose runs out and the bacteria stop growing until the next day, when they are transferred into fresh medium. Like Dallinger’s warm water, glucose-limited media is a selective pressure on the microbes, spurring the evolution of adaptations that compensate for a lack of their preferred food source.

Every 75 days (about 500 generations), a portion of LTEE’s cloudy soup of bacteria is stored in a minus-80-degree-centigrade freezer. These remain as frozen fossil records that can be used for direct comparison to their descendants.

Experimental design of the LTEE.

Image

Georgia Tech.

The LTEE has shed light on many unanswered questions about the dynamics of evolution, and experimentally validated long-running speculations. Do species improve indefinitely in a constant environment or will they stop at some maximum level? By comparing evolved E. coli with their ancestors, LTEE found that the rate of adaptation to the environment slows over time, but doesn’t plateau. Even after tens of thousands of generations in a stable laboratory environment, natural selection seems to be able to continuously eke out improvements.

Another major finding was that not all replicate populations follow the same evolutionary trajectory. In one replicate, named Ara-2, the population diverged into two coexisting lineages: one that rapidly consumes glucose and another that feeds on a byproduct of glucose metabolism called acetate. From a single population came a community of two.

But the most surprising finding was the observation that after about 31,000 generations, a different replicate, Ara-3, gained the ability to grow on citrate. Natural E. coli can’t metabolize citrate—in fact, it’s one of the defining features of the species—so the emergence of a strain which thrives on this carbon source could represent an entirely new species.

Simple, long-running experiments like the LTEE show what can be learned about evolution if given enough time. Other scientists, such as Roy Kishony, at Technion-Israel Institute of Technology, further accelerate the rate of evolution in their experiments by combining the intense selective pressure of antibiotics with cleverly designed experimental apparatuses. Kishony’s group has designed many selection devices, often with elaborate names like ‘morbidostat,’ but the most well-known is the Microbial Evolution and Growth Arena (MEGA) plate.

The MEGA plate is a two-by-four-foot rectangular petri dish that is divided into nine bands of gelatinous nutrient agar. The outermost bands have no antibiotics; the next inward has an antibiotic concentration barely high enough to kill E. coli; the bands inward have ten times that concentration, then a hundred times; and the middle band has the same antibiotic concentrated to 1,000 times. Spread over the top of each layer of agar, the bacteria can swim in.

At the start of the experiment, a drop of E. coli is placed on the outermost band. The bacteria multiply and spread right up to the boundary with the first antibiotic concentration band, but no further. The starting E. coli isn’t antibiotic resistant, and neither are most of its descendants. But thanks to random mutation a select few are, and these are able to cross the barrier and fan out into the next band. The descendants of these mutants themselves spread and mutate, crossing bands—or getting stuck behind them—until some reach the band that’s 1,000 times more concentrated than the first. The whole process takes just ten to twelve days, and can be recorded by a camera mounted above the plate to track the spread of each lineage.

We already knew that higher levels of resistance usually come at the cost of slower growth. With the MEGA plate, this evolutionary trade-off has immediate consequences: more-resistant but slower-growing strains get boxed in and trapped behind less-resistant but faster-moving ones. This is one reason why evolution may not always favor the most antibiotic-resistant bacteria.

No individuals in the initial population of E. coli could jump directly from the antibiotic-free agar to 1,000-times-concentration antibiotics. Adaptation to such a different environment is possible, but requires several specific mutations in multiple genes and these are extremely unlikely to occur together by chance under normal conditions. But when evolution is steered in that direction by a gradual gradient of intermediate concentrations, the E. coli reliably get there in a matter of days.

Evolution experiments are easily adaptable to new technologies. Whole-genome sequencing, for example, has been an enormously powerful tool for evolutionary biologists because it enabled them to connect changes in phenotype (observable characteristics, like antibiotic resistance) with genotype (underlying genetic mutations). New technologies can not only be used with future experiments, but also the frozen stocks from previous generations. Whole-genome sequencing may not have been feasible when the LTEE began, but 20 years later the researchers are able to sequence the genomes of the initial frozen *E. coli *generations just as easily as they could the latest generation. Whatever the next great technological revolution in biology may be, freezer shelves full of bacteria stand ready to be studied in a whole new light.

Today, labs around the world are running evolution experiments of all shapes and sizes, each using microbes to understand a specific facet of evolution. Some study predation by mixing predator and prey species, and observing how each adapts to the other. Other groups have studied starvation by growing bacteria for long periods of time without the addition of any nutrients, nor the removal of dead cells. And by selecting yeasts for increased size, others have directed the evolution of macroscopic multicellularity from single-celled ancestors.

Evolution by its nature takes time. With microbes we’ve been able to condense it down to more manageable timescales, but even 80,000 generations is a blip on the evolutionary clock. As these experiments continue to run, the more we’re sure to learn from them.