The evolution of the eye

The eye of a vertebrate or an octopus looks so complex that it can be difficult to believe it could have evolved by natural selection. In fact, light-sensitive organs (not all so complex) have evolved 40 - 60 times in various invertebrate groups - which suggests either that the Darwinian explanation faces a 40 - 60 fold more difficult problem than the vertebrate eye alone presents, or that it may not be so difficult for the things to evolve after all.

Simulating the evolution of the eye

Nilsson and Pelger simulated a model of the eye to find out how difficult its evolution really is. Their simulation begins with a crude light-sensitive organ: a layer of light-sensitive cells sandwiched between a darkened layer of cells below and a transparent protective layer above. The simulation therefore does not cover the complete evolution of an eye. To begin with it takes light-sensitive cells as given (which is an important but not absurd assumption, because many pigments are influenced by light) and at the other end it ignores the evolution of advanced perceptual skills (which are more a problem in brain, than eye, evolution). It concentrates on the evolution of eye shape and the lens; this is the problem that Darwin’s critics have often pointed to, because they think it requires the simultaneous adjustment of many intricately related parts.

From the initial simple stage, Nilsson and Pelger allowed the shape of the model eye to change at random, in steps of no more than 1% change at a time: 1% is a small change, and fits in with the idea that adaptive evolution proceeds in small gradual stages.

The model eye then evolved in the computer, with each new generation formed from the optically superior eyes in the previous generation; changes that made the optics worse were rejected, as selection would reject them in nature. The particular optical criterion used was visual acuity – the ability to resolve objects in space. The visual acuity of each eye in the simulation was calculated by the methods of optical physics. The eye is particularly well suited to this kind of study because optical qualities can readily be quantified: it is possible to show objectively that one model eye would have better acuity than another. (It is not so easy to imagine how to measure the quality of some other organs, such as a liver or a backbone.)

Results of the simulation

The simulated eye duly improved over time, so that after around 1000 steps the eye had evolved to be rather like a pin-hole camera eye. Then the lens starts to evolve by a local increase in the refractive index of the layer that started out simply as transparent protection. The lens to begin with has poor optical qualities but its focal length improves until it equals the diameter of the eye, at which point it could form a sharply focussed image.

How long did it take?

The complete evolution of an eye like that of a vertebrate or octopus took about 2000 steps. What had looked like an impossibility actually turns out to be possible in a short interval of time. Nilsson and Pelger used estimates of heritability and strength of selection to calculate how long the change might take; their answer was about 400 000 generations. With a generation time of one year, the evolution of an eye from a rudimentary beginning would take less than half a million years. Far from being difficult to evolve, the model shows that it is rather easy. The work illustrates the value of building models to test our intuitions.

The entire process is illustrated in the following animation.

Figure: late stages in the evolution of the eye. (a) the eye is protected by a transparent cover of skin, and part of the cellular fluid has differentiated into a lens. (b) full complex eye, as found in octopus and squid.