Nature makes order from randomness.
The photo above shows one of the pods from my Yellow Sugarsnap Project with peas segregating for seed colour. The pod is from one of my F2 hybrid plants (the second generation after the original cross) so the peas inside are F3. As immaculate as this alternating pattern is, it's entirely random.
I've just spent three days typing up descriptions of all my little packets of F3 seed from the Yellow Sugarsnap Project into a nice tidy table, and even as I handled each of my sixty-two seed packets (each plant's seeds carefully saved separately) and stared at them hour after hour I didn't notice the pattern. I noticed that some of the packets of seed are very uniform while others show a bit of variability. I thought that factor might be significant, so for each one I wrote down how variable the seeds were, and which traits they varied for. Sometimes it was size or colour, but more often it was a case of a few wrinkly seeds showing up in a batch of smooth ones. I dutifully jotted all this down but I still didn't notice the pattern. D'uh!
And then I was asked to do a little recorded talk about Mendel and his peas for a University of Bath podcast, just a very brief grounding in the history of genetics for psychology undergraduates. Not trusting myself to not screw it up, I did some refresher-research on Mendel. And in doing so I thought very hard about his experiments, and how he'd been the first person to notice the recurrence of 3:1 ratios in inherited traits. And it was only then that I twigged that there was a pattern in the seeds I'd collected from my pea project. So I raked them all out of the box and sorted them into different groups, and ker-ching! There it was. A beautiful and very obvious ratio.
As like as two peas in a pod? These F3 seeds from my Yellow Sugarsnap Project vary from smooth to wrinkled in the same pod, as well as varying for colour.
As a romantic idle speculation, I wonder whether Mendel found the same thing in his peas and got the initial idea for dominant/recessive segregation from it. Peas have this wonderful advantage over pretty much all other vegetables, that certain traits show up visibly in the seeds. If Mendel had been experimenting with tomatoes or brassicas this wouldn't happen because the seeds all look very similar no matter how different their genes are. He would have to actually grow the plants to see the differences between them. But with peas being the way they are, he must have seen a pattern very similar to what I have here.
The pattern is this: a number of my seed packets from the F2 plants have perfectly uniform round peas, with no wrinkles. A similar number have all wrinkled peas, with not a single round one among 'em. But a larger number have got variability for wrinkliness. And in every one of these cases they have, roughly speaking, a quarter wrinkled and three-quarters round. There are no other ratios. None of the packets have mostly wrinkled with just a few round, or even half and half. They all have an approximate 3:1 ratio in favour of round peas. A Mendelian ratio in other words. In fact there are two Mendelian ratios at the same time. The packets of round or predominantly round seed outnumber the packets of wrinkled seed by about 3:1, while the ratio of round to wrinkled within each of the variable seed packets is also 3:1.
I sorted the seed packets into types. On the left are all the seeds which are completely round with no wrinklies. On the right are the ones with all wrinklies and no roundies. In the middle are the packets which show a mixture of types. There are roughly twice as many in this middle group, as you can see.
Wrinkliness is one of the traits Mendel experimented with, and he found it to be recessive to roundness. This is now known as the R locus. The round-seeded allele is R and its wrinkle-seeded alternative is r. My original cross was between Golden Sweet (RR) and Sugar Ann (rr), so the resulting F1 hybrid must have had a genotype of Rr. Recombining those Rr genotypes in the F2 generation can go any of four ways, with visible effects in the seeds, like this:
Genotypes in the F2 plants can clearly be assigned to their four respective groups.
Why does seed wrinkliness matter? Well, it's a very useful trait for pea breeders to look out for because it's a rule-of-thumb indicator of sweetness. Sugars shrink more than starches do within pea seeds, so the sweeter ones tend to end up more wrinkly. A high sugar content doesn't guarantee a good flavour (as I found in my taste tests with these) but it helps.
It's obviously very useful to be able to identify the seeds which are likely to produce plants with sweet-tasting peas before you've sown them. If I want to breed a sweet-tasting variety I can just pick out and sow the wrinkly seeds and not the round ones, which will greatly increase my chance of getting what I want. This is a really unusual situation, and only works because the desirable trait shows up in the seed itself in an obvious way, when most other traits don't – you have to grow the plants to find out what their genetic make-up is, and even then you can't always tell. It's only because wrinkliness is recessive that I can be confident it will breed true.
Let me explain from a practical point of view. Dominant traits are a pain in the backside for plant breeders to work with. Say I wanted to breed a new pea with purple flowers, based on a cross between a purple-flowered and a white-flowered variety. Purple flowers show straightforward dominance in peas, so I would get ALL purples in the F1 generation followed by an F2 generation which was three-quarters purple and a quarter white. So I would obviously proceed by saving seed from all the purple-flowered F2s and removing the whites. When I sow the seeds from the purple-flowered plants, will they simply produce more purple-flowered plants? No, only a third of them will be true-breeding for purple. The rest will still have the recessive white-flower allele lurking in their DNA, hidden by its dominant purple twin. Although they look like true purples on the outside, those plants will again produce a 3:1 ratio of purples to whites. Unfortunately there's no way to tell which are true-breeding and which aren't, other than by growing them and removing all the whites in each generation until they eventually stop showing up.
Recessive traits, by contrast, are a joy. They show up in smaller proportions of course, but once you have a plant with the requisite pair of recessive alleles it should breed true from then on without any further mucking about.
That's why the sweet-wrinkly seeds showing up in a Mendelian ratio is such a godsend. Laying all these peas out on my desk in their individual packets, I can see their exact genotype for the R gene at a glance. The round seeded ones are RR and will breed true for roundness. The wrinkled ones are rr and will breed true for wrinkliness. The ones that are mostly round with a few wrinklies are Rr or rR (which amount to the same thing) and will continue to show variability in their offspring.
This is incredibly handy. Not only can I identify the sweet ones without having to grow them all and taste them, I can see which of them are true-breeding for sweetness/wrinkliness. If I want to be sure of getting a full complement of wrinkliness in my plants for ever after, I can instantly pick out the ones with the fully recessive genotype and Bob will be my uncle.
The reason this is possible is because this segregation for seed type is showing up within different peas on the same plant. Compare that to the situation with flowers. If some of the plants were obliging enough to produce a load of purple flowers and a smattering of whites all on the same plant, that would be great. I would know those were not true-breeding for purple. But they don't. They produce all purple flowers and keep the whites hidden in their genome to pass on to their offspring unseen.
OK, so we've established that the plants which produce only smooth, rounded seeds must be RR, and because they have a matching pair of alleles their offspring will also be RR. The technical name for this is homozygous. Exactly the same is true of the plants which produced only wrinkled seed. They are also homozygous, because their genotype must be rr and so all their offspring will be rr too.
The plants which produced a mixture of round and wrinkled types have to be heterozygous. Instead of a matched pair of alleles they have one of each type. That means that when they make seeds they will randomly pass on the four possible combinations to their offspring: RR or rr (which are both homozygous and will breed true) or Rr or rR (which are heterozygous and won't). The heterozygous seeds will express their dominant allele and hide their recessive one, so they will look the same as the RR seeds, and so once again there will appear to be a ratio of 3 rounded to 1 wrinkly.
Note that it's the plants which produced these seeds which are heterozygous, not the seeds themselves. Half the seeds in the heterozygous batch will actually be homozygous, but the other half remain heterozygous and will produce variable offspring which are half homozygous and half heterozygous, and so on ...
These seed packets are all siblings from the Yellow Sugarsnap project ... I still can't get over the amazing diversity made by this one simple cross!
With the two quarters of homozygous seeds separating out like this, you can see that in each generation half the heterozygosity is lost. If continued for a few generations it will all but disappear. That's how new varieties are stabilised.
In practical terms, what does that mean for these packets of variable seeds from the heterozygous F2 plants? Well, I know that I have all four classes mixed up here in approximately equal amounts, and I can see which seeds are homozygous (true-breeding) for wrinkliness, because they're wrinkled. Unfortunately I can't see which ones are homozygous for round seeds, because they look exactly the same as the heterozygous ones. Hence this 3:1 ratio of round to wrinkled. If I were to sow all these seeds, I would find the same 3:1 ratio in the next generation too, and onward.
Finally, a little reminder that all I'm looking at here is the R locus, the gene controlling wrinkliness. That's just one of many thousands of genes in every pea. Segregation is taking place at every other locus at the same time! If I select identical-looking wrinkled peas, I can assume they will be true-breeding for wrinkliness but they may differ enormously in other traits.
Wow, my head feels weird now.