V1

Suppose that one type of eye is better at detecting information about the world than another. Then it will allow its owner to respond more usefully to the visual world, whether it’s finding food, avoiding predators and harsh conditions, monitoring its own movement, or whatever. As always, natural selection works at a local level, local both in space and time. So it doesn’t matter whether this eye is good, only whether it is better than the competition around it. Half an eye is better than no eye. In fact half an eye is better than 49% of an eye…

Darwin admitted that the evolution of something as intricate as the human eye, by nothing more than natural selection, seems “absurd”. I love science when it’s so counter-intuitive! I know that everything in the universe is made out of atoms and quarks and things, but it doesn’t look like it, intuitively. Stars at night don’t look like they’re billions of miles away. The Earth doesn’t look round. Anyway, back to the eye…

Back in 1994, Nilsson and Pelger presented a model of the possible evolution of eyes, consisting of a sequence of steps where each step is a) small enough to happen in one generation; and b) leads to an improvement in spatial acuity. They start with a simple, flat light-sensitive circle and see where it leads. Basically, the sequence starts with an increasing central depression, which means that light from either side gets cut off by the surrounding “ridge”. Thus the light falling on any one part of the patch is coming from a smaller and smaller region of the world, as the depression gets deeper. This “specialisation” defines an increase in spatial acuity. Then later, the “ridge” surrounding this central depression constricts, ultimately forming a “pupil”, and further limiting the amount of “world” whose light reaches each part of the light-sensitive “retina”.

Unfortunately, this process inevitable cuts down the total amount of light reaching the retina, so you end up with an eye that has really high spatial acuity, but that only works in really bright conditions… But if you add a clear protective covering over the initial light-sensitive patch, and if this covering gradually thickens just so, then it will start bending the light in such a way that acuity improves further without having to make the pupil any smaller. This “lens” allows a trade-off between acuity and sensitivity, so you can see clearly now, even in lower light levels.

Through various plausible, even pessimistic, assumptions, Nilsson and Pelger argue that you could get from the initial flat patch of light-sensitive cells to a fully-functioning vertebrate eye in just 1829 steps! Then with some more (plausible) assumptions about evolution and inheritance, they estimate this could occur in a little over 350,000 generations, something like 1500 times faster than the evolution actually managed.

The upshot is a nice straightforward explanation of the development of many different sorts of eyes through natural selection. Whether the historical evolution of humans involved this sequence is not clear, but that hardly matters: here is one way it could have happened, an existence-proof of the possible evolution of the eye.

In the paper, they also point out it could, theoretically, have happened a lot faster, if for example different changes happened in parallel rather than sequentially. Given ideal conditions, a truly benevolent ecology, I wonder just how quickly such an eye could evolve? It’s a big universe out there: what’s the record fastest evolution of an eye, from a standing start, I wonder…?

Nilsson, D., Pelger, S. (1994). A Pessimistic Estimate of the Time Required for an Eye to Evolve. Proceedings of the Royal Society B: Biological Sciences, 256(1345), 53-58. DOI: 10.1098/rspb.1994.0048