Lenski's 33-Year Bacteria Study Proves Adaptation Never Plateaus

Veritasium

Data from 74,500 generations of E. coli demonstrate that natural selection in an unchanging environment follows an open-ended power law, continuously generating fitness gains without ever reaching an evolutionary ceiling.

AI-assisted summary of The Longest-Running Evolution Experiment — original source. Check the original for full context.

Source published: 2021-06-16

Key ideas

Richard Lenski's Long-Term Evolution Experiment (LTEE) has tracked twelve isolated lineages of E. coli since 1988, reaching over 74,500 generations—equivalent to roughly 1.5 million years of human evolution. (1:11)
After tens of thousands of generations, one lineage evolved the rare ability to metabolize citrate in an oxic environment, a breakthrough driven by both historical contingency and an intricate genetic rearrangement. (10:31)
Long-term fitness gains do not plateau as an asymptote; instead, they follow an open-ended power law model where adaptation slows progressively but never ceases, even in an unchanging environment. (13:14)
A daily 100-fold serial dilution discards 99% of bacterial cultures to prevent runaway growth that would mathematically exceed the volume of the observable universe within 42 days. (3:08)

Detailed notes

## Background Published on YouTube by Veritasium on June 16, 2021, this video examines the predictability and repeatability of natural selection by asking what would happen if life's tape were rewound and replayed. It highlights Prof. Richard Lenski's Long-Term Evolution Experiment (LTEE) with Escherichia coli alongside a sponsored kitchen demonstration investigating microbial contamination.

## TL;DR Derek Muller visits Richard Lenski's laboratory to explore the Long-Term Evolution Experiment, which has tracked twelve isolated lineages of E. coli over 33 years and more than 74,500 generations. By conducting daily serial dilutions and freezing viable generational 'fossils,' the researchers revealed how compounding beneficial mutations drove sustained fitness gains and enabled direct time-travel competition trials. The study's crowning discovery came when one population evolved the ability to metabolize citrate under oxic conditions, demonstrating how rare innovations depend on historical contingency and prior genetic architecture. Ultimately, the experiment overturned assumptions of an evolutionary plateau, proving that adaptation continues indefinitely along an open-ended power law even in an unchanging environment.

## Atmosphere & style The presentation blends playful, curious science communication with deep reverence for long-horizon academic dedication. Muller's dynamic hosting is grounded by Richard Lenski's thoughtful, modest reflections, turning the repetitive daily grind of pipetting and autoclaving into an epic narrative spanning the equivalent of 1.5 million years of evolution. Bookended by a lighthearted domestic demonstration using fluorescent dye, the viewing experience feels both casually engaging and philosophically profound as it contemplates chance, fate, and the boundless adaptability of life.

## Bulletin - [01:11](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=71s) Richard Lenski's Long-Term Evolution Experiment (LTEE) has tracked twelve isolated lineages of E. coli since 1988, reaching over 74,500 generations—equivalent to roughly 1.5 million years of human evolution. - [10:31](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=631s) After tens of thousands of generations, one lineage evolved the rare ability to metabolize citrate in an oxic environment, a breakthrough driven by both historical contingency and an intricate genetic rearrangement. - [13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) Long-term fitness gains do not plateau as an asymptote; instead, they follow an open-ended power law model where adaptation slows progressively but never ceases, even in an unchanging environment. - [03:08](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=188s) A daily 100-fold serial dilution discards 99% of bacterial cultures to prevent runaway growth that would mathematically exceed the volume of the observable universe within 42 days.

## Key takeaways - [00:00](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=0s) In a petri dish gradient, bacteria evolve resistance to a 1,000-fold antibiotic concentration in just 11 days through stepwise spontaneous mutations. - [01:11](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=71s) The Long-Term Evolution Experiment isolates twelve distinct E. coli lineages from a single common ancestor to guarantee that shared outcomes represent true independent evolution rather than preexisting genetic variation. - [01:11](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=71s) Rapid bacterial reproduction allows six to seven generations per day, accumulating vast evolutionary progress compared to slower-cycling species like corn. - [03:08](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=188s) Daily serial transfers keep populations sustainable by transferring exactly 1% of the broth to fresh minimal glucose medium while sterilizing the remaining 99%. - [05:26](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=326s) Despite a low per-cell mutation rate, a flask with approximately one billion bacteria generates around one million new genetic mutations daily. - [05:26](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=326s) Advantageous mutations can compound exponentially across transfers even after surviving the daily 99% culling process. - [07:36](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=456s) By freezing living samples every 500 generations, researchers create a physical fossil record enabling direct competition experiments between evolved strains and their ancestors. - [07:36](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=456s) Neutral red and white genetic markers allow direct visual counting to determine which bacterial generation outcompetes the other for limited resources. - [09:57](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=597s) Early fitness gains show an explosive adaptive phase followed by deceleration, initially leading scientists to expect a complete plateau in improvement. - [10:31](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=631s) In 2003, one lineage unexpectedly evolved the capacity to consume citrate, a compound wild E. coli cannot utilize under oxygenated conditions. - [11:48](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=708s) Replaying evolution from revived ancestral clones demonstrated that citrate utilization depended on potentiating background mutations as well as a specific, rare rearrangement. - [13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) Across decades, bacterial cell sizes increased significantly while overall population numbers per flask declined. - [13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) Several lineages evolved defective DNA repair systems that increased mutation rates a hundredfold before secondary mutations tempered the excessive mutational load. - [13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) Fitness data refutes the bounded rectangular hyperbola hypothesis, proving that adaptation in a constant environment fits an unbounded power law model. - [16:18](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=978s) A fluorescent household test demonstrates that reusable dishcloths rapidly cross-contaminate high-touch kitchen surfaces like sink faucets and appliance handles.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=0s) Evolution in Action and an Unsuspecting Household Test

A dish of bacteria encounters progressively higher concentrations of antibiotics. At first, the bacteria stop dead in their tracks upon reaching the initial antibiotic barrier. Soon, however, a spontaneous mutant emerges that can survive the dose. Another mutation follows, allowing the bacteria to tolerate 10 times the original concentration, then 100 times, until after just 11 days of rapid evolution, the surviving strain withstands an antibiotic concentration 1,000 times stronger than the dose that would have wiped them out at the start.

Noting that bacteria thrive everywhere—especially in damp environments like kitchen dishcloths—Derek Muller sets up a stealth experiment sponsored by Bounty paper towels. Without alerting anyone in his household, he coats a common dishcloth with fluorescent powder mimicking bacterial spread to see where the contamination eventually travels. Leaving that domestic petri dish behind, he flies out to Michigan to investigate the world’s longest-running laboratory evolution experiment.

[01:11](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=71s) The Long-Term Evolution Experiment: Twelve Isolated Lines

In 1988, evolutionary biologist Prof. Richard Lenski began an ambitious endeavor with twelve separate populations of live *Escherichia coli*, all derived from a single common ancestor. Over the past 33 years, Lenski and a dedicated team of researchers have kept the experiment running continuously, tending to the flasks every single day, including weekends.

```mermaid flowchart TD accTitle: Starting the Long-Term Evolution Experiment accDescr: Diagram showing how a single ancestor colony was diluted and split into twelve independent bacterial lineages. subgraph Isolation["Initial Plating & Isolation"] A["Single Common Ancestral Population"] --> B["Diluted onto Petri Dish"] B --> C["Individual Colonies Form"] end subgraph TwelveLines["Twelve Replicate Populations"] C --> D["6 Arabinose-Negative Lines (Red Marker)"] C --> E["6 Arabinose-Positive Lines (White Marker)"] end TwelveLines --> F["Continuous Independent Evolution (1988–Present)"] ```

While other evolutionary studies exist—such as the University of Illinois corn-breeding experiment running since 1896—corn produces only one generation per year. In contrast, Lenski’s *E. coli* cycle through six or seven generations every day, reaching generation 74,500 over 33 years. Scaled to human reproductive timescales, that corresponds to roughly 1.5 million years of hominid evolution. Lenski explains that each of the 12 lines began from an individual colony on a plate so that no shared variation was present from the outset.

> “And that's important because it means if we see the same thing happening in replicate populations, it's not because they started out with the same genetic variants and natural selection just fished out the same thing over and over.” — Richard Lenski

- Corn breeding experiment at the University of Illinois: ongoing since 1896 (1 generation per year). - Lenski's Long-Term Evolution Experiment (LTEE): started in 1988, spanning 33 years and over 74,500 generations. - Equivalent human timeframe: 1.5 million years of evolution.

[03:08](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=188s) The Daily Regimen, Exponential Growth, and the Autoclave Room

The artificial habitat designed for the bacteria is vastly simpler than the natural world. Kept strictly at 37 degrees Celsius with no competing species, they reside in a minimal liquid broth containing potassium phosphate, citrate, and a strictly limited amount of glucose as their sole carbon source. The evolutionary pressure is unambiguous.

> “Above all else, consuming that glucose and converting it to offspring, replicating as fast as possible has been essentially what we seem to be selecting for.” — Richard Lenski

Each day, the bacteria consume the available glucose, divide six to seven times, and multiply their population a hundredfold. To keep the cycle going indefinitely, the team performs a daily serial transfer: 0.1 milliliters (exactly 1% of the flask’s volume) is pipetted into a sterile flask of fresh medium. This dilutes the bacteria 100-fold, resetting their space and nutrients so they can expand again.

```mermaid flowchart LR accTitle: Daily Serial Transfer Cycle accDescr: The 24-hour cycle of dilution, exponential growth, and disposal in the LTEE. A["Flask at Full Density (After 6–7 Divisions)"] -->|"0.1 mL (1%)"| B["Fresh Medium Flask"] A -->|"9.9 mL (99%)"| C["The Autoclave Room ('Bacterial Crematorium')"] B -->|"24 Hours Growth at 37°C"| A ```

The remaining 99% of the culture heads straight to the autoclave room, an unceremonious sterilizing facility Derek humorously dubs a "bacterial crematorium." This daily purge is mathematically mandatory. If the researchers had accommodated the daily hundredfold expansion by increasing the broth volume rather than culling it, the culture would require one cubic meter of solution by day two, exceed 10 times the volume of the Earth by day 13, and outgrow the observable universe by day 42.

[05:26](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=326s) The Power of Numbers: How Rare Mutations Drive Natural Selection

Derek raises a fundamental paradox: if genetics is geared toward stability and replication fidelity, how does evolutionary novelty occur so readily in the lab? Lenski points out that per-cell mutation rates in bacteria are indeed conservative. Only about one in 100 to one in 1,000 bacterial cells carries a single mutation, whereas human offspring typically inherit 10, 20, or 50 new mutations. However, sheer population size turns that low rate into massive genetic variation.

With roughly a billion individual bacteria growing in a single flask, a one-in-a-thousand error rate generates approximately one million new mutations every single day. Lenski breaks down the fate of those daily million variants: - **Neutral mutations (~50%):** Many have no discernible impact on growth in minimal glucose broth, often affecting genes suited for the wild that are unexpressed in the lab. - **Deleterious mutations (~50%):** Flaws that make the cell an inferior competitor, dooming it to extinction. - **Beneficial mutations (10 to 1,000):** Rare variants that improve metabolic efficiency or reproduction rates compared to their parent cells.

Even if an advantageous cell divides 10% faster, it still faces the daily 99% cull. But once a favorable mutation makes it into the transferred 1%, that 10% edge compounds exponentially across subsequent flasks.

> “So the mutations, y'know, are really rare when they first occur and many of them are lost, but once they get common if they have that competitive advantage, they'll just sweep through the population.” — Richard Lenski

[07:36](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=456s) The Frozen Fossil Record and Bacterial "Fight Club"

Evaluating whether bacteria are truly improving over time relies on an extraordinary biological feature: *E. coli* can be frozen in suspended animation and revived decades later with full viability. Every 500 generations—roughly every 75 days—the team freezes representative samples of all 12 lines, creating a physical "frozen fossil record."

This setup allows Lenski to run direct "time travel" head-to-head competitions, matching generation 70,000 against its distant ancestors in a bacterial fight club. To track who wins, the experiment uses a neutral genetic color marker embedded from day one: - Six populations carry an arabinose-utilization trait that turns colonies **red** on indicator agar plates. - Six populations carry an alternative allele that turns colonies **white**.

By pitting a red evolved lineage against a white ancestor (or vice versa), researchers seed them together at an equal ratio in a single flask, let them compete for glucose over a 24-hour cycle, and then plate out the mixture. Counting the red and white colonies by hand reveals which strain divided faster and utilized the limited sugar more effectively. In head-to-head trials, the evolved lineages drastically outcompete their ancestral forebears.

[09:57](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=597s) Early Discoveries: Darwinian Adaptation and the Plateau Question

When asked about the earliest major findings from the long-term lines, Lenski notes that the data provided an unambiguous confirmation of natural selection in real time.

> “The first thing we found, not that there was any doubt about it, but it's one of the most direct demonstrations of Darwinian adaptation by natural selection, you can imagine.” — Richard Lenski

Early on, the bacterial fitness curves mimicked typical evolutionary trends: when introduced to a novel environment, adaptation starts with a dramatic, rapid surge in competitive fitness, followed by a noticeable deceleration as the easiest improvements are exhausted. Watching the adaptation rate slow down, Lenski initially expected the populations to eventually flatline entirely, even wondering early in the project whether it was time to bring the experiment to an end.

[10:31](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=631s) The Citrate Breakthrough: A "Lemony Dessert"

Richard Lenski had initially wondered when to wrap up the project, but he received wise counsel from colleagues and his wife, Madeline, who urged him to keep the experiment going. It was a fortuitous choice. In 2003, after tens of thousands of generations, one of the twelve populations began doing something extraordinary: it started consuming citrate. Citrate had been in the Davis minimal growth medium from day one as a chelating agent to bind trace metals, but wild *E. coli* is defined species-wide by its inability to transport and metabolize citrate under oxic conditions.

When researchers noticed that one of the culture flasks had developed an unexpected, dense turbidity, Lenski’s initial reaction was skepticism. He assumed foreign bacteria had contaminated the vessel. To verify the lineage, the team went back to the frozen fossil record to restart evolution and ran genetic checks on the unusually cloudy culture. The tests confirmed that the microbes were not contaminants; they were authentic descendants of the ancestral *E. coli* strain.

As Richard Lenski described the discovery: “It was very clear that one of our bacteria lineages had essentially I like to say sort of woken up one day, eaten the glucose, and unlike any of the other lineages discovered that there was this nice lemony dessert, and they'd begun consuming that and getting a second source of carbon and energy.”

[11:48](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=708s) Rewinding Life's Tape: Historical Contingency

The citrate innovation raised fundamental evolutionary questions: why did it take tens of thousands of generations to appear, and why did it happen in only one of the twelve flasks? Researcher Zachary Blount set out to test how this adaptation arose by pulling ancestral clones of that specific lineage out of the deep freeze at various generational time-points and attempting to re-evolve the trait. Lenski analogized the approach to rewinding a film to minute five, minute ten, or minute twenty to see whether the movie would end the same way if allowed to play forward again.

Two competing hypotheses had emerged to explain the extreme rarity of citrate utilization. The first held that the innovation required an exceedingly rare, intricate single mutation—such as a specific DNA inversion requiring precise genomic breakpoints—that could theoretically happen at any time but simply had astronomical odds. The second hypothesis argued for historical contingency: a stepwise series of background mutations that made an otherwise unremarkable mutation functional only after the genetic architecture had already changed.

When the replay experiments were completed, the findings bridged both schools of thought. As Richard Lenski summarized: “To make a long story short, it turns out it's such a difficult trait to evolve because both of those hypotheses are true.” The population required potentiating mutations that built the foundation, followed by a rare rearrangement event that finally unlocked the transport of citrate.

```mermaid flowchart TD accTitle: Rewinding the Tape of Evolution accDescr: Flowchart showing how Zachary Blount tested historical contingency by replaying evolution from frozen bacterial clones. subgraph Ancestry["Frozen Fossil Record"] A["Ancestral Clones (Minute 5, 10, 20)"] end subgraph Hypotheses["Competing Explanations"] H1["Hypothesis 1: Ultra-rare specific mutation"] H2["Hypothesis 2: Contingent series of mutations"] end subgraph Outcome["Experimental Conclusion"] R["Citrate Utilization (Cit+)"] end A --> H1 A --> H2 H1 -->|"Requires rare breakpoint"| R H2 -->|"Requires prior genetic changes"| R ```

[13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) Cell Size, Hypermutability, and the Power Law of Fitness

Beyond citrate utilization, the Long-Term Evolution Experiment revealed several surprising evolutionary dynamics:

- **Bacterial cell size vs. population size**: Instead of packing into higher cell counts over time, the bacteria actually decreased in total population count within the flasks while individual cells became significantly larger. - **Hypermutation and recovery**: Six of the twelve lineages evolved defective DNA repair systems, causing hypermutability where mutation rates spiked to one hundred times their ancestral level. However, because excessively high mutation loads produce deleterious mutations, several lines subsequently acquired secondary mutations that dialed their mutation rates back down.

The most profound realization concerned the trajectory of evolutionary fitness. Lenski had long assumed the bacteria would eventually hit an adaptive plateau, modeled as a rectangular hyperbola that climbs steeply before reaching a hard asymptote.

```mermaid flowchart LR accTitle: Comparison of Fitness Models accDescr: Diagram comparing the bounded rectangular hyperbola model with the open-ended power law model. subgraph Hyperbola["Rectangular Hyperbola"] H_Early["Rapid Initial Gains"] --> H_Late["Hard Asymptote (Plateau/Flatline)"] end subgraph PowerLaw["Power Law Model"] P_Early["Rapid Initial Gains"] --> P_Late["Ever-Slowing Improvement (No Upper Limit)"] end ```

That assumption proved wrong. As Richard Lenski admitted: “This view I had that they were flat lining turned out to be quite false.” A two-parameter power law model—where improvements decelerate continuously but never encounter an upper ceiling—matched the empirical data far better. When fed only the first 5,000 generations of data, the hyperbolic model predicted an asymptote that the bacteria subsequently blew past. In contrast, the power law model trained on just one-tenth of the early experimental history accurately projected fitness all the way out to 50,000 and 60,000 generations.

This result offers an unexpected conclusion about adaptation. Scientists often attribute continuous evolution to dynamic environments, red-queen races against pathogens, or mass extinction events. Yet Lenski's bacteria prove that even in a strictly uniform, unchanging flask of broth, evolution never truly stops. As Lenski reflected: “But what our experiment suggests is that even in the absence of environmental change, there are so many opportunities of smaller and smaller magnitude to continue to make progress that, in fact, progress probably would never stop even in a constant environment.”

[16:18](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=978s) Kitchen Experiment Conclusion

The scene returns to Derek Muller's home demonstration, where he had applied fluorescent powder to see how bacteria can be transferred around living spaces. Turning off the room lights and switching on a handheld ultraviolet torch, the glowing tracer dye reveals widespread contamination across the kitchen surfaces.

Bright streaks glow heavily inside the sink basin, across the faucet neck, and over the tap handles where someone had wiped down the fixtures with a shared dishcloth. Fluorescent traces and clear finger marks appear all over the handle and front panel of the dishwasher, illustrating how easily reusable dishcloths spread microbes to high-touch household surfaces. Emphasizing the challenge of keeping a house clean with young children, Derek concludes with a sponsorship message: “As a dad, I encounter lots of big messes and sticky hands and slimy faces.” He recommends using Bounty paper towels to clean spills before thanking the sponsor and viewers.

## Notable quotes > [13:14](https://www.youtube.com/watch?v=w4sLAQvEH-M&t=794s) what our experiment suggests is that even in the absence of environmental change, there are so many opportunities of smaller and smaller magnitude to continue to make progress that, in fact, progress probably would never stop even in a constant environment. — Richard Lenski

## Mentioned Derek Muller, Richard Lenski, Zachary Blount, Nkrumah Grant, Devin Lake, Kate Bellgowan, Minako Izutsu, Madeline Lenski, Escherichia coli, Long-Term Evolution Experiment, University of Illinois, Bounty, Veritasium

## Video description If you ran evolution all over again, would you get humans? How repeatable is #evolution? This video is sponsored by @BountyBrand. #biology #science #QPU #HygieneBeginswithBounty #BountyPartner

Special thanks to Prof. Richard Lenski and team for showing me around the lab — it is an honor to be able to witness and document such a historic science experiment. Thanks to Dr Zachary Blount for the help with research and setting up the competition time-lapse, Dr Nkrumah Grant for microscope images of the long-term line cells @NkrumahGrant Devin Lake, Kate Bellgowan, and Dr. Minako Izutsu for being part of this video. Long Live the LTEE!

LTEE website — http://myxo.css.msu.edu/ecoli/index.html Intro footage courtesy of the Kishony Lab — https://kishony.technion.ac.il ▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀ References: Lenski, R. E., & Travisano, M. (1994). Dynamics of adaptation and diversification: a 10,000-generation experiment with bacterial populations. Proceedings of the National Academy of Sciences, 91(15), 6808-6814. — https://ve42.co/Lenski1994

Lenski, R. E., Rose, M. R., Simpson, S. C., & Tadler, S. C. (1991). Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during 2,000 generations. The American Naturalist, 138(6), 1315-1341. — https://ve42.co/Lenski1991

Good, B. H., McDonald, M. J., Barrick, J. E., Lenski, R. E., & Desai, M. M. (2017). The dynamics of molecular evolution over 60,000 generations. Nature, 551(7678), 45-50. — https://ve42.co/Good2017

Blount, Z. D., Borland, C. Z., & Lenski, R. E. (2008). Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli. Proceedings of the National Academy of Sciences, 105(23), 7899-7906. — https://ve42.co/Blount2008

Blount, Z. D., Lenski, R. E., & Losos, J. B. (2018). Contingency and determinism in evolution: Replaying life’s tape. Science, 362(6415). — https://ve42.co/Blount2018

Wiser, M. J., Ribeck, N., & Lenski, R. E. (2013). Long-term dynamics of adaptation in asexual populations. Science, 342(6164), 1364-1367. — https://ve42.co/Wiser2013

N, Scharping. (2019). How a 30-Year Experiment Has Fundamentally Changed Our View of How Evolution Works. Discover — https://ve42.co/Scharping

▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀ Special thanks to Patreon supporters: Mike Tung, Evgeny Skvortsov, Meekay, Ismail Öncü Usta, Paul Peijzel, Crated Comments, Anna, Mac Malkawi, Michael Schneider, Oleksii Leonov, Jim Osmun, Tyson McDowell, Ludovic Robillard, Jim buckmaster, fanime96, Juan Benet, Ruslan Khroma, Robert Blum, Richard Sundvall, Lee Redden, Vincent, Marinus Kuivenhoven, Alfred Wallace, Arjun Chakroborty, Joar Wandborg, Clayton Greenwell, Pindex, Michael Krugman, Cy 'kkm' K'Nelson, Sam Lutfi, Ron Neal

▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀ Research and Writing by by Derek Muller, Petr Lebedev and Casey Rentz Animation by Ivy Tello Filmed by Derek Muller, Emily Zhang and Raquel Nuno Edited by Derek Muller Music by Jonny Hyman and from Epidemic Sound https://epidemicsound.com Additional video supplied by Getty Images Thumbnail image courtesy of the Kishony Lab Produced by Casey Rentz ▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀

[Source](https://www.youtube.com/watch?v=w4sLAQvEH-M)

中文简报

伦斯基长达33年的细菌研究证明适应永无止境来自74,500代大肠杆菌的数据表明，在不变环境中，自然选择遵循开放式幂律，持续产生适应度收益，从未达到进化上限。德里克·穆勒（Derek Muller）走访了理查德·伦斯基（Richard Lenski）的实验室，探访了其长期进化实验（LTEE）。该实验在33年间追踪了12个独立的大肠杆菌品系，繁衍超过74,500代。通过每日进行连续稀释并冷冻保存具活性的世代“化石”，研究人员揭示了复合有益突变如何推动持续的适应度提升，并实现了直接的跨时空竞争实验。该研究最重大的发现出现在某一群体进化出了在有氧条件下代谢柠檬酸盐的能力，这表明罕见的创新取决于历史偶然性与原有的遗传构架。最终，该实验推翻了进化存在停滞期的假设，证明即使在不变的环境中，适应也会沿着开放式幂律无限持续。理查德·伦斯基的长期进化实验（LTEE）自1988年以来一直追踪12个独立的大肠杆菌品系，已繁衍超过74,500代——大致相当于人类进化150万年。 (1:11) 经过数万代的繁衍，一个品系进化出了在有氧环境中代谢柠檬酸盐的罕见能力，这一突破是由历史偶然性与复杂的基因重排共同推动的。 (10:31) 长期的适应度收益并没有像渐近线那样陷入停滞；相反，它们遵循开放式幂律模型，其中适应虽然逐渐放缓，但即使在不变的环境中也从未停止。 (13:14) 每天进行100倍的连续稀释会丢弃99%的细菌培养物，以防止其失控生长——在数学计算上，这种失控生长会在42天内超过可观测宇宙的体积。 (3:08)

Source quotes

[13:14] Richard Lenski: what our experiment suggests is that even in the absence of environmental change, there are so many opportunities of smaller and smaller magnitude to continue to make progress that, in fact, progress probably would never stop even in a constant environment.

Original transcript

[0:00] - These are bacteria growing into increasingly concentrated antibiotics. [0:05] The bacteria stop growing [0:06] when they hit the first antibiotic strip, [0:09] but then a mutant appears capable of surviving [0:12] in the antibiotic. [0:13] Then another mutation occurs [0:15] and now the bacteria can survive 10 times the concentration, [0:19] then a hundred times, [0:22] and finally, after just 11 days of evolution, [0:25] these bacteria can survive antibiotics [0:27] a thousand times stronger [0:29] than what would have killed them at the start. [0:31] You're watching evolution in action. [0:34] This video was sponsored by Bounty, [0:36] reminding you that a hygienic way to clean up messes [0:38] and spills is with paper towel. [0:40] Now, bacteria are everywhere, [0:42] especially in damp places like dishcloths. [0:45] So I'm gonna do a little experiment [0:46] where I add some fluorescent powder to this dishcloth [0:49] to represent bacteria [0:50] without telling anyone in my household. [0:52] And then I'm gonna come back later [0:54] to see where all of this powder ends up. [0:58] But for now, I'm off to Michigan [1:00] to see the world's longest-running evolution experiment. [1:03] Let's go. [1:04] (plane engine roaring) [1:06] (bright upbeat music) [1:11] - This is Richard Lenski. [1:12] He started the experiment 33 years ago, [1:15] and along with a team of colleagues, [1:17] he has kept it going even on weekends ever since. [1:22] In this lab are 12 flasks of live E. coli bacteria. [1:27] They are the lucky few, the ones that have survived [1:31] over three decades of evolution in a lab. [1:34] - So there are 12 long-term lines. [1:38] - [Derek] In 1988, a single common ancestor [1:41] spawned these 12 separate populations. [1:43] And ever since that day, [1:45] they have been growing and dividing independently. [1:48] So how have they evolved? [1:50] There are other long running evolution experiments [1:53] like since 1896, [1:54] scientists at the University of Illinois [1:56] have selectively bred corn, [1:58] but they get only one generation per year, [2:01] whereas these bacteria go through six [2:04] or seven generations a day. [2:07] So after 33 years, [2:08] the bacteria in these flasks are generation 74,500. [2:13] If those were human generations, [2:15] it would represent 1.5 million years of hominid evolution. [2:21] - What we do to start the experiment [2:22] is we dilute a large population of bacteria [2:26] onto a Petri dish and each individual cell makes a colony. [2:30] And so then we take a little bit from one colony [2:33] to start one population, [2:34] a little bit from another colony [2:35] to start another population, [2:37] but effectively they all started from individual cells. [2:41] And that's important because it means [2:43] if we see the same thing happening in replicate populations, [2:47] it's not because they started out [2:50] with the same genetic variants [2:52] and natural selection just fished out [2:54] the same thing over and over. [2:55] You actually had to have independent mutations [2:58] that would give rise to these competitive advantages, [3:01] whatever it might be that would produce the repeatability [3:05] across the 12 replicate populations. [3:08] - [Derek] The lab environment is very different [3:10] from the one the bacteria are used to. [3:12] It's much simpler, [3:13] there are no other organisms present, [3:15] they're kept at 37 degrees Celsius, [3:17] and they live in the same solution, [3:20] a mixture including glucose, potassium phosphate, citrate, [3:23] and a few other things. [3:25] Their only carbon source is glucose, which is limited. [3:28] - Above all else, [3:29] consuming that glucose and converting it to offspring, [3:33] replicating as fast as possible [3:35] has been essentially what we seem to be selecting for. [3:38] - [Derek] Every day in each flask [3:40] the bacteria divide six or seven times, [3:43] which increases their numbers a hundred fold. [3:46] Normally we think of generations [3:48] as being limited by the time that has elapsed, [3:51] but in the case of these bacteria, [3:53] they're limited by the resources available to them. [3:56] If they had 10 times the solution, [3:57] they would increase their numbers an additional 10 times. [4:01] Then almost every day for the past 33 years [4:05] this transfer process has taken place. [4:08] 0.1 milliliters or 1% of the solution from each flask [4:12] is transferred to a sterile new flask [4:15] containing the same solution. [4:17] It essentially dilutes the bacteria a hundred fold. [4:21] This gives them the space and resources [4:23] they need to grow and divide again, [4:25] increasing their population a hundred times [4:28] before this same process is repeated the next day. [4:32] And then the day after that and the day after that, [4:35] even on weekends, [4:36] this process has been going on for over three decades. [4:41] What happens to bacteria that are not transferred [4:44] to the new flasks, the 99%? [4:47] - That's the autoclave room. [4:49] - [Derek] What happens in the autoclave room? [4:51] - Every day 99% of the E. Coli meet their demise [4:55] in this horrible room. [4:56] - [Derek] Is this like a bacterial crematorium? [4:58] - Yeah. Yeah, exactly. [5:01] - [Derek] You can imagine if the scientists didn't do this, [5:03] but instead gave all the bacteria [5:05] a hundred times more solution to grow on every day, [5:08] well, the experiment would soon become unmanageable. [5:11] On day two, you would need a cubic meter of solution, [5:15] but by day 13 the experiment would be [5:17] 10 times the volume of earth. [5:19] And by day 42, [5:21] the experiment would fill up the entire observable universe. [5:26] To me it seems like the whole idea of genetics [5:29] is for them to stay constant [5:30] and for mutations to be rare. [5:32] - Yes. [5:33] - Yet in your experiment it seems they're not rare. [5:36] - So we estimate that in our bacteria [5:40] only about one out of maybe a hundred [5:43] or one out of a thousand cells [5:45] will have even a single mutation. [5:46] So that's not very much. [5:49] By contrast, in humans it's estimated that each one [5:52] of our offspring has perhaps 10, 20, 50 new mutations. [5:57] So the bacteria are extremely conservative, [5:59] but they're also billions of them and even a tiny flask. [6:02] And so if you have a chance in a thousand of mutating [6:05] and you have a billion individuals, [6:07] then every day in one of these flasks [6:09] we might have a million new mutations. [6:12] So that's a lot of variation [6:14] on which natural selection can act. [6:15] Maybe half of those mutations have no effect whatsoever [6:20] on the bacteria's ability to grow [6:22] and thrive in that particular environment. [6:24] They might be mutations that would matter [6:26] out in the great outdoors, [6:27] but this was a very simple environment, [6:29] maybe those genes aren't even expressed. [6:32] Another half of the mutations speaking very, very roughly [6:34] might actually be deleterious mutations. [6:36] They make the bacteria an inferior competitor, [6:39] but there's maybe out of those million mutations [6:42] that occur every day, [6:43] maybe there's 10, maybe there's a hundred, [6:45] maybe there's a thousand of them [6:47] that actually change something in the cell [6:49] that gives the bacteria a competitive advantage [6:52] over their progenitors. [6:53] And those then grow over the course of that day. [6:56] Then every day 99% of the population is eliminated. [7:01] It's lucky 1% prevails, [7:03] and if those lucky 1% include one of the guys [7:06] from the previous day that was growing at 10% faster [7:09] than the other guys, [7:10] it has a higher probability of contributing [7:13] to that next flask, the next day's flask [7:16] and in the fresh medium, [7:18] and it will continue to grow faster at a 10% clip, [7:21] and that compounds by an exponential process. [7:24] So the mutations, y'know, are really rare when they first occur [7:28] and many of them are lost, [7:29] but once they get common [7:31] if they have that competitive advantage, [7:33] they'll just sweep through the population. [7:36] - [Derek] But how do you know that the bacteria [7:37] actually have a competitive advantage? [7:40] I mean, how can you tell that they're getting better suited [7:42] to their environment? [7:44] Well, this is where one of the unique properties [7:46] of bacteria come in. [7:48] They can be frozen for long periods of time [7:51] and then revived. [7:52] - And so they're stored in suspended animation here. [7:56] So these are the racks that contain the frozen samples [7:59] of the bacteria from the various generations. [8:03] - [Derek] Every 500 generations, roughly 75 days, [8:07] they freeze a sample of each population. [8:10] By freezing previous generations, [8:12] Lenski and his team have a frozen fossil record. [8:16] - So our samples from over 30 years ago [8:19] remain perfectly viable. [8:20] And so that gives us an ability to do [8:23] what I like to call time travel. [8:25] We can literally compare organisms that lived [8:27] at different points in time. [8:29] So we can compete bacteria [8:30] from generation 70,000 against their ancestor. [8:34] - [Derek] That's right. [8:36] The way they measure fitness is by competing [8:38] the current generation of bacteria [8:40] against older generations. [8:42] It's like a strange bacteria fight club. [8:45] - Hey! [8:46] - [Derek] They thaw out the old generation, [8:48] mix it into a flask with the current generation, [8:51] and then plate out a sample of the solution [8:53] to see the relative abundances [8:54] of the two populations at the start. [8:56] Then they incubate the flask for a day, [8:59] and then plate them out again. [9:01] And the point is to compare their relative growth rates, [9:04] which generation was better able to utilize the glucose [9:08] and divide faster? [9:11] - Well, how the heck do you tell the evolve bacterium [9:12] from its ancestor? [9:13] Do they wave little flags at you and say, [9:15] "Hey, I'm the evolved guy." [9:16] And of course they don't. [9:18] But what we have is this color marker [9:21] embedded in the experiment. [9:23] So six of our populations on a certain kind of agar plate [9:27] make red colonies and six of them make white colonies. [9:30] And we have one version of the ancestor [9:32] that makes red colonies and one that makes white colonies. [9:35] We can compete one of the red evolved populations [9:38] against the white ancestor, [9:40] or one of the white evolved populations [9:42] against the red ancestor and we can distinguish them. [9:45] - [Derek] In this case, [9:46] it's clear that the evolved red population [9:48] outcompeted their white ancestor. [9:51] Now, to determine the winner, [9:53] all of the colonies are counted by hand. [9:57] - So what was the earliest sort of big findings [9:59] from the experiment? [10:00] - The first thing we found, [10:02] not that there was any doubt about it, [10:04] but it's one of the most direct demonstrations [10:06] of Darwinian adaptation by natural selection, [10:08] you can imagine. [10:09] Yes, they're getting to be better competitors over time. [10:13] It's a common observation in other evolution experiments [10:16] that evolution in a new environment gets off [10:18] to a rip-roaring start [10:20] and then tends to slow down over time. [10:22] And so we repeated that observation [10:25] and I imagined that the longterm lines [10:27] would actually sort of flat line at some point. [10:29] And I actually thought about stopping the experiment, [10:31] but I got wise advice from colleagues [10:33] and from my wife, Madeline, let's keep it going. [10:36] And so I agreed to that. [10:38] - [Derek] And it's good thing he did because in 2003, [10:41] the bacteria started doing something remarkable. [10:44] - One of the 12 lineages suddenly began to consume [10:49] a second carbon source, citrate, [10:51] which had been present in our medium [10:53] throughout the experiment. [10:55] It's in the medium is what's called a chelating agent [10:58] to bind metals in the medium. [11:01] But E. coli going back to its original definition [11:03] as a species is incapable of that. [11:06] But one day we found one of our flasks had more turbidity. [11:09] I thought we probably had a contaminant in there. [11:11] Some bacterium had gotten in there [11:13] that could eat the citrate [11:14] and, therefore, had raised the turbidity. [11:17] We went back into the freezer and restarted evolution. [11:20] We also started checking those bacteria [11:22] to see whether they really were E. coli. [11:24] Yep, they were E. Coli. [11:25] Were the really E. coli [11:26] that had come from the ancestral strain? [11:28] Yep. So we started doing genetics on it. [11:31] It was very clear that one of our bacteria lineages [11:34] had essentially I like to say sort of woken up one day, [11:37] eaten the glucose, [11:38] and unlike any of the other lineages [11:40] discovered that there was this nice lemony dessert, [11:43] and they'd begun consuming that [11:44] and getting a second source of carbon and energy. [11:48] Zack was interested in the question [11:49] of why did it take so long to evolve this [11:52] and has only one population evolve that ability? [11:56] He went into the freezer [11:57] and he picked bacterial individuals or clones [12:00] from that lineage that eventually evolved that ability. [12:03] And then he tried to evolve that ability again [12:06] starting from different points. [12:08] So in a sense, it's almost like, [12:10] well, it's like rewinding the tape [12:11] and starting let's go back to the minute five of the movie. [12:14] Let's go back to a minute 10 of the movie, [12:16] minute 20 of the movie [12:17] and see if the result changes depending on when we did it, [12:21] because this citrate phenotype [12:23] there were essentially two competing explanations [12:25] for why it was so difficult to evolve. [12:29] One was that it was just a really rare mutation. [12:31] It wasn't like one of these just change one letter. [12:34] It was something where maybe you had to flip [12:36] a certain segment of DNA [12:37] and you had to have exactly this break point [12:39] and exactly that break point. [12:41] And that was the only way to do it. [12:42] So it was a rare event, [12:44] but it could have happened at any point in time. [12:47] The alternative hypothesis is that, [12:50] well, what happened was a series of events [12:54] that made something perfectly ordinary become possible [12:58] that wasn't possible at the beginning [13:00] because a mutation would only have this effect [13:03] once other aspects of the organism had changed. [13:07] To make a long story short, [13:09] it turns out it's such a difficult trait to evolve [13:11] because both of those hypotheses are true. [13:14] - [Derek] The experiment uncovered [13:16] other surprising findings. [13:17] Like instead of the bacteria [13:19] getting more numerous over time, [13:21] they actually decreased in number, [13:24] but each individual bacterium got larger. [13:28] Six of the 12 populations evolved hypermutability, [13:31] mutation rates a hundred times higher than their ancestors, [13:35] but these populations subsequently acquired [13:37] additional mutations [13:39] that brought the mutation rate back down. [13:41] I mean, it's advantageous [13:42] to be able to evolve faster than others, [13:44] but if the mutation rate is too high, [13:46] then offspring have too many deleterious mutations. [13:50] But maybe the most surprising finding of all [13:53] is what didn't happen. [13:55] - This view I had that they were flat lining [13:57] turned out to be quite false. [13:59] I had sort of imagined a very simple mathematical model. [14:03] You can create something called a rectangular hyperbola, [14:06] I guess, which has an initial high slope [14:08] and then reaches an asymptote, [14:10] but they're equally simple models. [14:13] There's a model that also has just two parameters [14:15] called a power law model that says things slow down, [14:19] but it doesn't have an upper bound. [14:20] It says, just keep going for time immemorial, [14:23] and things will just keep going faster, [14:25] but at a slower and slower rate of further improvement. [14:28] And it turned out that model actually fits our data [14:32] better than that original model I had imagined. [14:35] And not only does it fit it better, [14:37] okay, you say statistics, science, fitting curves, [14:40] it actually predicts the future. [14:41] And that's what's really cool because the original model, [14:45] if you give it say just 5,000 generations [14:47] worth of our fitness data [14:49] and ask it to predict into the future, [14:51] it says the asymptote is here. [14:53] But then when we get more data, no! [14:55] The bacteria are up here. [14:56] They've passed that asymptote. [14:58] Whereas this power law model, [15:00] which says things are slowing down, [15:01] but never reaching an asymptote? [15:03] We give it just 1/10 of our data from the past [15:06] and it projects very accurately out to 50,000 [15:09] and even 60,000 generations when we last looked. [15:12] It predicts sort of the future course [15:14] of the evolutionary trajectory. [15:16] And to me, that's kind of profound [15:17] and it's sort of changed the way I look at this experiment, [15:21] and even a little bit how I look at life on earth. [15:25] I mean, life on earth doesn't stop evolving. [15:27] We know that. [15:28] And we know that, [15:29] but we think that's, [15:30] oh, that's because they're asteroid impacts. [15:31] That's because of human impacts. [15:33] That's because they're viruses [15:35] that are attacking their hosts [15:36] and that the co-evolution is causing evolution [15:38] is causing evolution to never stop. [15:40] The world is always changing. [15:42] So of course, evolution never stops. [15:44] And that is 100% true. [15:46] But what our experiment suggests [15:49] is that even in the absence of environmental change, [15:53] there are so many opportunities of smaller [15:56] and smaller magnitude to continue to make progress [15:59] that, in fact, progress probably would never stop [16:03] even in a constant environment. [16:07] To me, it's one of the reasons [16:08] to keep this experiment going. [16:09] Does this model continue to predict the unfolding [16:12] of the future fitness trajectory? [16:18] - Okay. Now the conclusion of the experiment [16:21] where I look for that fluorescent powder [16:24] using this UV torch. [16:26] So let's hit the lights. [16:29] (laughs) [16:29] Whoa. There is a lot of fluorescent powder around here. [16:34] Obviously there's a lot in the sink, [16:36] but also here on the tap, [16:39] it looks like someone wiped down the handle and the faucet. [16:44] Oh, check out the dishwasher. [16:46] Yeah, and on the handle. [16:49] This is a great way of visualizing [16:51] how dishcloths can spread bacteria around the house. [16:54] What? (laughs) [16:58] It looks like finger marks. [17:02] As a dad, I encounter lots of big messes [17:04] and sticky hands and slimy faces. [17:07] So when there is a mess or spill in my kitchen [17:10] I choose to clean it up with Bounty [17:12] So I wanna thank Bounty for sponsoring this video, [17:14] and I wanna thank you for watching.