What Evolution Can and Cannot Explain About Complexity

the challenge

The challenge is not merely to imagine a sequence from light-sensitive cells to a camera eye. It is to show that each small change improves performance enough for natural selection to retain it, while embryonic development can still make the structure. Critics often point to the vertebrate retina, where nerves pass in front of photoreceptors, creating a blind spot, as evidence that the eye is too awkward to have been engineered stepwise. That point cuts the other way. Historical compromise is exactly what evolution predicts.

what the data show

A classic paper by Dan-Eric Nilsson and Susanne Pelger in 1994 modelled a plausible route from a flat patch of photoreceptive cells to a camera-type eye. Allowing only modest changes in curvature, aperture and refractive properties, each improving spatial resolution by about 1 per cent, they estimated that a functional eye could evolve in fewer than 400,000 generations. The number is rough and dependent on assumptions, but the point was never the exact total. It showed that no absurd timescale is required.

Developmental genetics strengthened the case. The Pax6 gene, studied in fruit flies, mice and other animals, helps organise eye development across very distant lineages. That does not mean all eyes evolved once in their current form. It means old regulatory circuits were repeatedly redeployed. Comparative anatomy tells the same story of tinkering. The mammalian middle ear, often taught for its own sake, is also a textbook case of exaptation: bones that once formed part of the jaw joint in synapsid ancestors were co-opted into hearing. Complex organs are often assembled by moving old parts into new jobs. Eyes are no exception, drawing on pre-existing light-sensitive molecules, epithelial tissues and neural circuits.

the challenge

The challenge here is historical reconstruction. Bacteria do not leave a generous fossil record of soft molecular machines, and the proteins involved evolve quickly. That makes it harder than with bones or shells to trace exact lineages. Sceptics then leap from “hard to reconstruct in detail” to “cannot have evolved”. Those are not the same claim.

what the data show

Comparative genomics and structural biology have steadily eroded the claim of irreducibility. Several core flagellar proteins are homologous to components of the type III secretion system, a needle-like apparatus used by some bacteria to inject proteins into host cells. The evolutionary relationship is complicated, and researchers disagree about which system came first in particular lineages, but the broader point stands: these are members of a family of secretion and motility structures built from related parts. There is no isolated miracle machine sitting outside biology.

Flagella themselves are not uniform. Some bacteria use simpler motility systems than the textbook model from Salmonella. Different lineages have added, lost and rearranged components. That modular variation matters because it shows that the system is evolvable. Parts can be co-opted, duplicated or dispensed with while preserving function. Selection can then refine performance. This is the same general logic seen again and again in evolution: scaffolding first, optimisation later.

the challenge

Here the challenge is to explain both complexity and robustness. Clot too little and an animal bleeds to death. Clot too much and vessels block. Critics argue that such balance requires all the parts to be present from the start. That argument confuses present optimisation with ancestral necessity. Early systems need only be better than uncontrolled bleeding, not as fine-tuned as a human trauma ward would prefer.

what the data show

Comparative genomics shows that much of the vertebrate cascade expanded through gene duplication and divergence. Related serine proteases were copied, modified and recruited into a chain of activations. Ancestral protein reconstruction has made that process unusually concrete. Joe Thornton and colleagues, working on steroid hormone receptors and other systems, reconstructed ancient proteins to show how new molecular specificities evolve through a combination of permissive and function-switching mutations. The clotting field has drawn on the same logic: present-day specificity can be traced back to more promiscuous ancestors that later specialised after duplication.

Jawless vertebrates such as lampreys and hagfish are especially instructive because they lack several clotting factors familiar from jawed vertebrates, yet they still haemostase. That observation alone punctures the claim that the mammalian cascade is irreducibly complex. Evolution can elaborate a simpler system into a more layered one when ecological demands change, body size increases or injuries become more severe.

The same historical pattern that reshaped the mammalian middle ear is visible here at the molecular level. Old components acquire new jobs. Exaptation is not a rare trick but a routine mechanism. A protein used in digestion, immunity or extracellular remodelling can be copied and shifted into haemostasis if the chemistry is close enough for selection to begin refining it.

the challenge

The challenge is twofold. First, to show that new coding or regulatory elements really arise rather than merely being discovered after the fact. Second, to explain how they avoid being neutral junk long enough to matter. De novo genes, in particular, once seemed implausible because a random stretch of DNA was assumed unlikely to become a useful protein.

what the data show

Gene duplication is the most established route. A copied gene can preserve the original function in one copy while the other accumulates changes, sometimes partitioning tasks, sometimes acquiring a new role entirely. Whole-genome duplications in the ancestry of vertebrates probably supplied abundant raw material for innovation, although the exact number and timing of those events remain debated.

De novo gene birth is now documented in several lineages, from yeast to flies to mammals. Most such genes are short and many will be transient, but some gain expression, function and selective value. New genes also arise by exon shuffling, retroposition and the domestication of transposable elements. The placenta offers a famous example: syncytin genes in mammals were derived from ancient viral envelope genes and recruited for cell fusion. There is nothing mystical about this. Genomes are opportunists.

Lenski’s E. coli lines again matter because they show novelty emerging under observation. The citrate-using trait was not a brand-new gene, but a regulatory innovation that created a new ecological opportunity. That is often how complexity starts: not with a miracle protein, but with a changed context in which existing parts interact differently. Over longer periods, selection and drift can then stabilise and elaborate those interactions into pathways that later look inevitable.