Evolution follows patterns at some scales but not others, and the honest answer is that biologists have been arguing about exactly this question for decades. Certain outcomes appear again and again across independent lineages, suggesting that natural selection funnels life toward predictable solutions. Yet the specific creatures that emerge, and the exact paths they take, depend heavily on historical accidents that no one could have forecast. The real picture is a mix of repeatable trends and genuine surprises.
The Case for Predictable Outcomes
One of the strongest arguments that evolution follows patterns comes from convergent evolution, where unrelated species independently arrive at the same solution to the same problem. The saber-toothed cat (Smilodon) and the marsupial Thylacosmilus both evolved enormous, blade-like canine teeth for killing large prey, despite being only distantly related. Smilodon is a placental mammal more closely related to housecats and elephants. Thylacosmilus is more closely related to kangaroos and koalas. Neither animal’s close relatives have saber teeth, which means this trait evolved from scratch in two separate lineages facing similar ecological pressures.
These parallels show up everywhere in nature. Dolphins and ichthyosaurs converged on the same streamlined body plan separated by hundreds of millions of years. Eyes evolved independently in vertebrates and mollusks. Flight arose separately in birds, bats, insects, and pterosaurs. The paleontologist Simon Conway Morris has argued that convergence is the dominant theme of evolution: replay the tape of life and you may not get the same species, but you will almost certainly get large, fast, filter-feeding ocean animals, reef-building organisms, and predators that hunt from the seafloor. In his view, evolution’s endpoints are largely inevitable because the physical world offers only so many workable designs.
The Case for Unpredictability
Stephen Jay Gould made the opposite argument. He proposed that chance events, what he called historical contingency, were so important that a smart observer watching early animal life in the Cambrian period could not have predicted which lineages would survive and which would vanish. A mass extinction that wipes out one group opens the door for another, and which group happens to survive can come down to luck rather than superiority.
Gould’s point becomes clearer with specific examples. Whales exist because certain mammalian adaptations related to suckling (the separation of the mouth passage from the airway) were later co-opted for high-speed, open-mouthed feeding underwater. A reptile trying the same feeding strategy would drown. So while large, fast marine filter-feeders might be “inevitable” in some abstract sense, the particular pathway that produced whales depended on a chain of prior adaptations that had nothing to do with ocean life. Human consciousness is another case: it depended on a string of unique evolutionary events and possibly even recent historical circumstances that favored consciousness as a style of mental function. These are not outcomes you could have predicted from first principles.
What Laboratory Evolution Reveals
One of the most powerful tests of evolutionary predictability comes from Richard Lenski’s Long-Term Evolution Experiment at Michigan State University. Twelve genetically identical populations of E. coli bacteria have been evolving independently in the same laboratory environment for over 60,000 generations across more than 25 years. If evolution is predictable, these populations should evolve similar changes. If it is contingent, they should diverge.
The answer, fittingly, is both. All 12 populations independently evolved mutations in the same gene (pykF), and different specific mutations in that gene spread through every single population. This level of parallelism is strong evidence that certain genetic changes are reliably beneficial in a given environment. At the same time, only one of the 12 populations evolved the ability to feed on a second nutrient in the growth medium, a dramatic innovation that the other 11 never stumbled upon. That single evolutionary event required a specific and unlikely sequence of prior mutations, making it a textbook example of contingency.
Physical and Developmental Constraints
Part of the reason evolution shows patterns is that organisms cannot evolve in just any direction. Physics, chemistry, and the way embryos develop all impose hard limits on what is possible, and these limits channel evolution down particular paths.
You cannot have a vertebrate with wheels instead of legs, for instance, because blood cannot circulate to a rotating organ. Five-foot-tall mosquitoes are impossible because the structural properties of insect bodies and the physics of fluid dynamics forbid it. The elasticity and tensile strength of biological tissues set boundaries on what body plans can function. These physical constraints eliminate vast regions of theoretical possibility before natural selection even gets involved.
Developmental constraints add another layer. When organisms depart from their normal development, they tend to do so in only a limited number of ways. If limb shape is determined by a particular chemical patterning process during embryonic growth, then limb forms that cannot be generated by that process simply will not appear. Closely related species also carry what biologists call phyletic constraints: the genetic toolkit they inherited from ancestors restricts the kinds of novelty they can produce. This is why all vertebrates share the same basic body plan at a critical stage of embryonic development, regardless of how different the adults look. That conserved stage constrains what evolution can build next.
Large-Scale Trends Over Deep Time
Zooming out to geological timescales, some patterns emerge that hold across entire groups of organisms. One of the most famous is Cope’s Rule: the tendency for body size to increase over evolutionary time. A study of brachiopods (shelled marine animals) found that their average body size increased by more than a hundredfold during their initial radiation from the Cambrian through the Devonian period, growing at a rate sufficient to increase size by an order of magnitude roughly every 77 million years. This increase occurred nearly in parallel across all major brachiopod groups.
But the pattern breaks down at smaller scales. When researchers looked at individual brachiopod families rather than broad classes, those smaller lineages showed random, unbiased size changes with no consistent direction. The large-scale trend was real, but it emerged from statistical patterns across many lineages rather than from a universal force pushing every species to get bigger. This scale-dependence is a recurring theme: evolution can look orderly from a distance while being messy and unpredictable up close.
Why Related Species Stay Similar
Another pattern in evolution is that closely related species tend to occupy similar ecological roles and environments, a phenomenon called phylogenetic niche conservatism. A study of mammals found that tropical species with small ranges and specialized diets tend to have more conserved thermal niches, meaning their temperature preferences change very slowly over evolutionary time compared to those of widespread, generalist species in temperate zones.
Interestingly, the same study found that a species’ current geography and recent evolutionary history had a greater influence on its ecological niche than its deep ancestry did. In other words, where an animal lives now and what it has been doing lately matters more than what its distant ancestors were doing millions of years ago. Niche conservatism is real, but it is not absolute, and it varies considerably depending on the type of organism and the environment it inhabits.
Predictability in Medicine
The question of whether evolution follows patterns is not just academic. In medicine, the ability to predict how bacteria evolve resistance to antibiotics could save lives. Researchers have built evolutionary models that reconstruct the history of resistance mutations in pathogens like E. coli and Streptococcus pneumoniae, and these models can predict the impact of interventions like changing antibiotic prescribing practices.
Research on fitness landscapes (maps of how different genetic combinations affect an organism’s survival) has found that even in complex, rugged landscapes with many possible outcomes, the fittest genotype tends to remain accessible from the starting point through multiple different pathways. This means that while the exact route a pathogen takes to resistance may vary, the destination is often reachable regardless of path, making the broad outcome somewhat predictable even when the details are not.
Still, the field has significant limitations. A meta-analysis found that only about 35% of resistance-modeling studies actually fitted their models to real epidemiological data, and just 5% validated predictions against a separate dataset. Researchers have acknowledged that they largely cannot yet predict the future frequency of antibiotic resistance from current data alone. Evolution’s patterns are real enough to be useful, but not yet reliable enough to serve as a crystal ball.
Pattern and Chance Together
The most accurate answer to whether evolution follows a pattern is that it does both. General outcomes, like the existence of predators, fast swimmers, or organisms that build reefs, appear to be highly repeatable because physics, ecology, and developmental biology funnel life toward a limited set of workable designs. The specific details of which lineages produce those designs, through what genetic changes, and on what timeline, are deeply contingent on history, luck, and context. Evolution is patterned enough to study, model, and sometimes predict, but contingent enough that the tape of life, replayed from the beginning, would almost certainly produce a recognizably different cast of characters.