Copy number

The standard picture of adaptation is slow and incremental.  Mutations turn up at random, most are useless, occasionally one helps, and over enough generations the helpful ones become common.  It’s a good picture.  It’s also, for anything living in water that is warming on a timescale of decades, uncomfortably leisurely.

So a paper from October is interesting: it argues that some diatom populations get there faster by a much blunter route.  They double the entire genome.

Polyploidy, if you haven’t met it, is having more than the usual two copies of every chromosome.  It’s unremarkable in plants - bread wheat carries six sets, cultivated strawberries eight - and it tends to be treated as a rare, dramatic event that happens once and then defines a lineage for millions of years.  The claim here is that it’s also something that happens on the timescale of a warming experiment, and that populations which do it adapt faster.

Why extra copies would help isn’t obvious, and I don’t think it’s settled.  There are three standard arguments.  Redundancy: with two copies of a gene doing the job, a third and fourth are free to drift into doing something slightly different, which is a way of getting novelty without the risk of breaking what works.  Dosage: more copies can simply mean more product, and if the limiting factor under heat is how much of some protein you can make, that’s immediately useful.  And buffering: with several copies, damage to one matters less.  They aren’t mutually exclusive and the paper doesn’t need to choose between them to make its case.

What I keep getting stuck on is one level below that, and it’s the sort of thing that decides whether the result means what it appears to mean.

Two stories produce this observation.

In the first, heat causes the doubling.  Something about high temperature disrupts the machinery of cell division - and division is exactly where you’d expect heat to cause trouble, since it involves a lot of protein assemblies doing precise mechanical work - and cells come out of it with double the chromosomes.  Under this story, doubling is a response.  It’s available to any population that runs into heat, whether or not any of its members had the trait beforehand.

In the second, the doubling was already there.  In any large population a small fraction of cells will have doubled by accident.  Heat arrives, kills most of the ordinary cells, and the doubled ones happen to do better and take over.  Under this story, doubling is not a response to anything.  It’s a lottery ticket, and you have to have bought it in advance.

Same endpoint.  Same measurement before and after.  Opposite mechanisms, and quite different implications - the first says this is a card any population can play, the second says it depends entirely on what was already sitting in the population when the temperature moved, which in turn depends on population size, history and luck.

The measurement that separates them isn’t exotic.  You measure ploidy repeatedly through the temperature ramp rather than only at the start and the end.  If doubled cells appear during the stress in numbers that can’t be accounted for by growth of the ones already present, that’s induction.  If the doubled fraction simply expands from where it started while everything else declines, that’s selection.  Flow cytometry does this and it isn’t expensive; it’s just several more sampling points, and someone has to be there for them.

I should say clearly that I’m working from what’s reported rather than from a close reading of the full analysis, and it’s entirely possible the paper addresses this and I haven’t got to it yet.  If it does, good - that’s the version of the result I want.  But it’s the question I’d ask first, and the fact that it’s the question I’d ask first tells you something about how easy it is to publish an endpoint comparison and let the reader supply the mechanism.

The general shape of this problem escapes biology immediately.  Any time you’re told that an event changed a population, there are two things it can mean: the event changed the individuals, or the event changed which individuals were left.  A cohort that came through something difficult may have been altered by it, or may simply be the subset that was always going to come through.  Before-and-after measurements can’t tell those apart, and they will produce a confident number either way.

None of this is a doom story, and I don’t want it read as one.  I don’t think warming oceans are a foregone catastrophe for diatoms, and I don’t think they’re fine either - the honest position is that these are ancient, wildly varied organisms and we don’t yet know enough to say.  What a paper like this does is add a mechanism to the list of things that might be happening.  That’s genuinely useful.  It’s just that a mechanism on the list and a mechanism demonstrated to be operating are different objects, and the distance between them is usually one experiment nobody has funded.