Issue 193 – October 2022


Between Chaos and Order: The (Un)predictability of Evolution

When the crew of USS Enterprise encounters yet another alien species that looks exceptionally humanlike, we know to suspect the production budget rather than more profound reasons. But the underlying question is an important one: How humanlike, or not, would members of an extraterrestrial civilization be? How familiar would we find walking through a forest on a vaguely Earth-like planet? Are we facing a universe of unimaginably alien beings, or something closer to the depictions in popular SF franchises?

It all comes down to the question of how (un)predictable the origin and evolution of life are. This debate in the field of evolutionary biology has lasted for decades, and research so far points in both directions; it depends on what species (or indeed other levels of the biosphere), conditions, and timescales we’re looking at. Would we ever be able to say which direction is more important in the long term, and whether we can predict evolution in a given lineage and environment?

Paleontologist and evolutionary biologist Stephen Jay Gould famously asked what would happen if we were able to replay the tape of life from its origins; in his view, historical contingency—happenstance, put simply—would shape the outcome so vastly that we would not recognize the new present life-forms. It’s true that a lot of Earth’s life evolution was shaped by some major overhauls—both from within and without. In the former case, if ancient cyanobacteria hadn’t evolved oxygenic photosynthesis, would it have occurred in another group—and if so, when and with what consequences? In the latter case, if the large asteroid missed Earth approximately 65 million years ago, would dinosaurs still have become extinct (excepting birds), or would they have flourished and perhaps even evolved a civilization?

In epic fantasy, we often see monumental battles between order and chaos, destiny and accident. In evolution, there is convergence and contingency.

Convergence vs. Contingency

Perhaps, after having rewound time to life’s origin on Earth, we really wouldn’t recognize the Earth today; life’s evolution would have been too contingent on accidents. Did a bacterium with a “lucky mutation” die instead of another one? Whole new tree of life.

Except maybe not. Maybe life would have arrived to noticeably similar solutions, albeit likely on different paths. Maybe we’d even see the technologically advanced two-eyed intelligent bipeds of Star Trek screenwriters’ fantasies?

If we look at life on Earth, we can certainly see countless examples of what is called evolutionary convergence—two or more different paths arriving to a very similar solution. The body shape of a shark and a dolphin is very similar, though each belongs to a different group, separated by some 450 million years of distinct evolution since their last common ancestor. Bats evolved flight independently of birds, yet they converged on a very similar solution. Eyes evolved dozens of times, each time with a somewhat different architecture, but to a similar end. Cichlid fish settling new lakes evolved to occupy the same niches—and the list could go on and on.

Can we really quantify the relative occurrence of convergence, though? Aren’t the examples above just cherry-picking, not representative of the “usual” process of evolution? After all, we can readily notice these similarities, but what of the countless other traits where none occurred, and what of the countless species and countless genes barely looked at?

We can’t rewind back to ancient Earth to see what would happen. Nor can we run an experiment spanning millions of years. That’s where long-term microbial evolution experiments come in handy. Where a hundred human generations would take about two thousand years, it could be less than a week in many bacterial species.

Multiple experiments starting with the same bacterial strains, running for years or even decades, have been trying to “play the tape” several times at once to see how the results would look. Richard Lenski’s Escherichia coli long-term evolution experiment is the most famous of these, started in 1988 and still running after more than seventy-five thousand generations. Lenski initially divided a genetically uniform population of E. coli into twelve populations, growing them in identical environmental conditions and closely tracking the traits of each.

Many changes, such as an increasing cell size, mirrored each other in most or all of the twelve populations, but there were also unique changes, such as a single population evolving the ability to feed on citrate—a chemical present in the growth medium initially for different reasons. But the mutations allowing for this trait either never occurred in any of the other populations, or didn’t succeed there by chance (such as occurring in a bacterium also carrying a harmful mutation). However, when Lenski and his team tried to replay the history of this particular citrate-feeding population, they got interesting results. Because samples from each population were regularly frozen and stored, they could thaw these “frozen fossils” from different points in time and let them grow to see which ones would also evolve citrate-feeding. Going back about 10,000 generations, the bacterial samples were able to do that. Before that point, no; specific mutations had to have already occurred to enable evolution of this trait. Will this population remain unique, or will some of the other eleven converge on citrate-eating in time? We can only wait and see.

In another experiment, Oliver Tenaillon’s team decided to expose E. coli to a standing selective pressure: heat exposure. They cultured 115 populations for 2000 generations and traced the changes occurring in those cultures. On the level of the individual mutations, each population took a unique path—but these mutations often occurred and were selected for in the same genes and led to similar outcomes helping the bacteria become more heat resistant.

In fact, the earliest known long-term microbe evolution experiment was a much cruder version of this, understandably lacking the genetic sequencing part: In 1878, William Dallinger, an amateur scientist and Darwin admirer, began to grow microbes in a copper flask that he’d slowly heated up over the course of months and years. His microbial culture was able to tolerate the heat, while “wild-type” microbes placed into the same temperature died. He was able to run the experiment for eight years; alas, preceding the advent of genetics, it was mostly forgotten.

The present-day experiments and many, many more tend to reveal the diversity of mutations, but also the regulating role of interactions between genes—because any gene has to be able to “work with others” in the complex network of relationships between them. While mutations are random (mostly; they are more likely to occur in some regions of the DNA than others, as one nucleotide swap rather than another), once they occur, we can often predict pretty well what happens in the short term. The longer term is trickier. Although evolution arises from random events, it is not itself random: natural selection pushes it in some directions. But it is, to some extent, chaotic: a small change can have great cascading effects in the long term, making the fine details very much unpredictable on longer timescales (but not necessarily preventing very broad, generalized predictions).

Can we conclude anything about the predictability of the evolution of Earth’s life, then? With some caution, we can arrive to a few takeaways: We can predict quite well how life responds to a new selection pressure using already existing genetic variation. If the response is contingent on new mutations, though, it is much harder to predict if or when it happens.

In addition, we are able to predict convergence on a broader scale, such as in multiple lineages evolving eyesight or flight. Making evolutionary predictions also has very concrete practical significance: Being able to determine which strains of flu will dominate the next season enables us to tailor each year’s vaccine to them—and it usually works pretty well. The future of COVID vaccines will likely follow this path, with a recommended annual booster tailored to match the virus strain predictions (although new mutations can always upend our predictions).

What does it all tell us about the prospects of life on other planets, though? There’s the basic conclusions: we can expect body shapes to converge to some of the forms advantageous in a given medium, such as a streamlined fishlike shape for water column creatures that need to be able to move quickly. But what about the evolution of sensoria, intelligence, or even “details” such as the number of appendages? In other words, could four-armed, eight-eyed infrared-sensing beings who think vastly different from us have populated a “replay Earth,” and how likely would they have been elsewhere?

Paths toward Intelligence

Some scientists, such as the paleontologist Simon Conway Morris, are convinced that civilization-producing intelligence is both rare and confined to very humanlike beings on very Earth-like planets—a position taking evolutionary convergence and the Rare Earth theory (postulating that for complex life to evolve, a lot of improbable steps had to have happened, such as the impact leading to the formation of the Moon stabilizing Earth’s tilt) to the extreme. Many science fiction novels have been able to challenge that assumption (with life-forms such as the scramblers in Peter Watts’s Blindsight or Eridians in Project Hail Mary by Andy Weir), but coming up with an exotic intelligent species doesn’t equal tracing the whole evolutionary path to it.

What evolutionary paths did lead to high intelligence on Earth? Apes, including ourselves, cetaceans, birds, and cephalopods are usually cited as examples of animals with high generalized intelligence. Cetaceans are mammals like us—quite evolutionarily close, not a great analogy for contemplating aliens, even if they live in a markedly different environment than we do.

Despite their different neural architectures, birds such as corvids and parrots evolved complex social behavior and high intelligence independently of mammals; birds’ and mammals’ last common ancestor lived nearly 300 million years ago. Tool use, tactical deception—those are just some of their skills. And yet, from the point of view of trying to imagine evolution in a whole distinct biosphere, trying to use avian and mammalian intelligence as an example falls flat on its face. Within the scope of all of Earth’s biosphere, birds and mammals are practically the same group, and our last common ancestor must have already been fairly intelligent. We diverged—and then in some ways converged again.

Octopi might be a better example—our last common ancestor with cephalopods lived over 700 million years ago in the Precambrian, and must have had very little brain to speak of (although this rudimentary brain still shared the basic neuronal cell workings with us and present-day octopi, making claims of “alien intelligence” of cephalopods somewhat exaggerated). Indeed, they offer us insights such as that high intelligence does not necessarily mean very high centralization (they do have big brains, but it appears that a lot of their complex behavior is driven by the arms’ own enervation).

Cephalopods long seemed to be unlikely candidates for creating a civilization, though. It’s not their aquatic environment (one can think up even a fairly technological aquatic civilization), but they appeared to lack the social behavior shared by many vertebrates, especially among mammals and birds. Octopi were thought to live alone and die before or very shortly after their offspring hatch.

Notice the past tense, though; in recent years, small octopus “villages” have been discovered, and even shallow-water octopi were found to be able to spawn multiple times in their lifetime, something previously suspected in deep-sea cephalopods (which remain very much understudied). With so few long-term observations of octopi in the wild so far, there might yet be social complexity to be discovered in some species (a tantalizing science-fictional subject explored, e.g., by Ray Nayler in The Mountain in The Sea). Even with what we know so far, octopi show us yet another evolutionary path toward general intelligence, converging on it from a different vantage point.

Roads Not Taken

By now it’s probably clear that “convergence vs. contingency” is a false dichotomy; both play important roles in evolution, and one or the other can prevail on a given level or in a given lineage or environment. On a deeper level, the occurrence of convergence and degree of predictability will be shaped by the genetic architecture of a given biosphere. There’s no guarantee that the modes of selection will work the same everywhere. We can expect units that are better adapted to the current environment to survive. What would be those units, though? Would they be able to reproduce sexually (or would the distinction between asexual and sexual reproduction seem meaningless in that biosphere)? What would be the information storage medium, how readily would it change, and how would it be repaired? What would be the relationship between this information (genotype) and life’s appearance (phenotype)?

These questions are very relevant even on Earth, and we can imagine them growing to monumental proportions elsewhere, or perhaps even in the imaginary replay of Earth’s history. If we reran it twelve times, how many times would we even have ended up with DNA as the predominant information-storage molecule? If Earth had ever truly been an “RNA world,” how likely would it have remained one? In other words, do all roads on Earth lead to DNA, or is it an accident frozen in time?

Have there ever been competing genetic codes? Was ours plainly better, or did it prevail mostly by chance? Would others have been more or less prone to effects of mutations, misreading, shifts? What if the mechanism of getting the information to the resulting cell/body had consisted of different steps, more or less prone to error?

Even Earth, our singular biosphere sample, suggests many roads not taken that would have greatly changed the course of evolution. The million-dollar question: Were these roads not taken (or perhaps taken and abandoned via extinction), because they lead nowhere fruitful, or simply by chance? Are there “forbidden territories” in the imagined evolutionary landscape, or just places not visited in the semi-random walks through it?

We’ll never know for sure. We can only take a more educated guess after closely examining the (un)repeatability of evolution on current Earth, and, perhaps one day, comparing multiple biospheres—whether with truly endless forms most beautiful, remains to be seen.

Further Reading

  • Wonderful Life: The Burgess Shale and the Nature of History (Stephen Jay Gould; published by W. W. Norton, 1989)—note that it has become dated, but it’s still a great introduction to the topic of evolutionary contingency
  • Life’s Solution: Inevitable Humans in a Lonely Universe (Simon Conway Morris; published by Cambridge University Press, 2003) —a great account of evolutionary convergence, even if taken to more extreme conclusions
  • In The Light of Evolution: Essays from The Laboratory and Field (ed. Jonathan Losos; published by W. H. Freeman, 2011, Second Edition) —a collection of essays, some of which deal with the contingency/convergence debate and experimental evolution
  • Improbable Destinies: Fate, Chance, and the Future of Evolution (Jonathan Losos; published by Riverhead Books, 2018) —bridging the perceived gap between convergence and contingency

Author profile

Julie Nováková is a scientist, educator and award-winning Czech author, editor and translator of science fiction, fantasy and detective stories. She published seven novels, one anthology, one story collection and over thirty short pieces in Czech. Her work in English appeared in Clarkesworld, Asimov’s, Analog, and elsewhere. Her works have been translated into eight languages so far, and she translates Czech stories into English (in, Strange Horizons, F&SF, Clarkesworld, and Welkin Magazine). She edited or co-edited an anthology of Czech speculative fiction in translation, Dreams From Beyond, a book of European SF in Filipino translation, Haka, an outreach ebook of astrobiological SF, Strangest of All, and its more ambitious follow-up print and ebook anthology Life Beyond Us (Laksa Media, upcoming in late 2022). Julie’s newest book is a story collection titled The Ship Whisperer (Arbiter Press, 2020). She is a recipient of the European fandom’s Encouragement Award and multiple Czech genre awards. She’s active in science outreach, education and nonfiction writing, and co-leads the outreach group of the European Astrobiology Institute. She’s a member of the XPRIZE Sci-fi Advisory Council.

Share this page on: