As I say, in all twelve tribes the average fitness increased as the thousands of generations went by. All twelve lines got better at surviving in these glucose-limited conditions. The fitness increase could be attributed to several changes. Populations grew faster in successive flasks, and the average body size of the bacteria grew, in all twelve lines. The top graph opposite plots the average bacterial body size for one of the tribes, which was typical. The blobs represent real data points. The curve drawn is a mathematical approximation. It gives the best fit to the observed data for this particular kind of curve, which is called a hyperbola.* It is always possible that a more complicated mathematical function than a hyperbola would give an even closer fit to the data, but this hyperbola is pretty good, so it hardly seems worth bothering to try. Biologists often fit mathematical curves to observed data, but, unlike physicists, biologists are not accustomed to seeing such a close fit. Usually our data are too messy. In biology, as opposed to physical sciences, we only expect to get smooth curves when we have a very large quantity of data gathered under scrupulously controlled conditions. Lenski’s research is a class act.
Lenski experiment: bacterial body size in one tribe
You can see that most of the increase in body size occurred in the first 2,000 or so generations. The next interesting question is this. Given that all twelve tribes increased in body size over evolutionary time, did they all increase in the same way, by the same genetic route? No, they didn’t, and that’s the second interesting result. The graph at the top of page 123 is for one of the twelve tribes. Now look at the equivalent hyperbolic best fits for all twelve (graph at the foot of page 123). Look how spread out they are. They all seem to be approaching a plateau, but the highest of the twelve plateaus is almost twice as high as the lowest. And the curves have different shapes: the curve that reaches the highest value by generation 10,000 starts by growing more slowly than some of the others, and then overtakes them before generation 7,000. Don’t confuse these plateaus, by the way, with the daily plateaus of population size within each flask. We are now looking at curves in evolutionary time, measured in flask generations, not individual bacterial time, measured in hours within one flask.
Lenski experiment: bacterial body size in twelve tribes
What this evolutionary change suggests is that becoming larger is, for some reason, a good idea when you are struggling to survive in this alternating glucose-rich/glucose-poor environment. I won’t speculate on why increasing body size might be an advantage – there are many possibilities – but it looks as though it must have been so, because all twelve tribes did it. But there are lots of different ways to become larger – different sets of mutations – and it looks as though different ways have been discovered by different evolutionary lineages in this experiment. That’s pretty interesting. But perhaps even more interesting is that sometimes a pair of tribes seem to have independently discovered the same way of getting bigger. Lenski and a different set of colleagues investigated this phenomenon by taking two of the tribes, called Ara+1 and Ara−1, which seemed, over 20,000 generations, to have followed the same evolutionary trajectory, and looking at their DNA. The astonishing result they found was that 59 genes had changed their levels of expression in both tribes, and all 59 had changed in the same direction. Were it not for natural selection, such independent parallelism, in 59 genes independently, would completely beggar belief. The odds against its happening by chance are stupefyingly large. This is exactly the kind of thing creationists say cannot happen, because they think it is too improbable to have happened by chance. Yet it actually happened. And the explanation, of course, is that it did not happen by chance, but because gradual, step-by-step, cumulative natural selection favoured the same – literally the same – beneficial changes in both lines independently.
Lenski experiment: increase in fitness
The smooth curve in the graph of increasing cell size as the generations go by gives support to the idea that the improvement is gradual. But perhaps it is too gradual? Wouldn’t you expect to see actual steps, as the population ‘waits’ for the next improving mutation to turn up? Not necessarily. It depends on factors such as the number of mutations involved, the magnitude of each mutation’s effect, the variation in cell size that is caused by influences other than genes, and how often the bacteria were sampled. And interestingly, if we look at the graph of the increase in fitness, as opposed to cell size, we do see what could at least be interpreted as a more overtly stepped picture (above). You remember, when I introduced the hyperbola, I said it might be possible to find a more complicated mathematical function that would fit the data better. Mathematicians call it a ‘model’. You could fit a hyperbolic model to these points, as in the previous graph, but you get an even better fit with a ‘step model’, as used in this picture. It is not such a close fit as the cell size graph’s fit to a hyperbola. In neither case can it be proved that the data exactly fit the model, nor can that ever be done. But the data are at least compatible with the idea that the evolutionary change that we observe represents the stepwise accumulation of mutations.*
We have so far seen a beautiful demonstration of evolution in action: evolution before our very eyes, documented by comparing twelve independent lines, and also by comparing each line with ‘living fossils’, which literally, instead of only metaphorically, come from the past.
Now we are ready to move on to an even more interesting result. So far, I’ve implied that all twelve tribes evolved their improved fitness in the same general kind of way, differing only in detail – some being a bit faster, some a bit slower than others. However, the long-term experiment threw up one dramatic exception. Shortly after generation 33,000 something utterly remarkable happened. One out of the twelve lineages, called Ara−3, suddenly went berserk. Look at the graph opposite. The vertical axis, labelled OD, which stands for optical density or ‘cloudiness’, is a measure of population size in the flask. The liquid becomes cloudy because of the sheer numbers of bacteria; the thickness of the cloud can be measured as a number, and that number is our index of population density. You can see that up to about generation 33,000, the average population density of Tribe Ara−3 was coasting along at an OD of about 0.04, which was not very different from all the other tribes. Then, just after generation 33,100, the OD score of Tribe Ara−3 (and of that tribe alone among the twelve) went into vertical take-off. It shot up sixfold, to an OD value of about 0.25. The populations of successive flasks of this tribe soared. After only a few days the typical plateau at which flasks of this tribe stabilized had an OD number about six times greater than it had been, and than the other tribes were still showing. This higher plateau was then reached in all subsequent generations, in this tribe but no other. It was as though a large dose of extra glucose had been injected into every flask of Tribe Ara−3, but given to no other tribe. But that didn’t happen. The same glucose ration was scrupulously administered to all the flasks equally.
Lenski experiment: population density
What was going on? What was it that suddenly happened to Tribe Ara−3? Lenski and two colleagues investigated further, and worked it out. It is a fascinating story. You remember I said that glucose was the limiting resource, and any mutant that ‘discovered’ how to deal more efficiently with glucose would have an advantage. That indeed is what happened in the evolution of all twelve tribes. But I also told you that glucose was not the only nutrient in the broth. Another one was citrate (related to the substance that makes lemons sour). The broth contained plenty of citrate, but E. coli normally can’t use it, at least not where there is oxygen in the water, as there was in Lenski’s flasks. But if only a mutant could ‘discover’ how to deal with citrate, a bonanza would open up for it. This is exactly what happened with Ara−3. This tribe, and this tribe alone, suddenly acquired the ability to eat citrate as well as glucose, rather than only glucose. The amount of available food in each successive flask in the lineage therefore shot up. And so did the plateau at which the popula
tion in each successive flask daily stabilized.
Having discovered what was special about the Ara−3 tribe, Lenski and his colleagues went on to ask an even more interesting question. Was this sudden improvement in ability to draw nourishment all due to a single dramatic mutation, one so rare that only one of the twelve lineages was fortunate enough to undergo it? Was it, in other words, just another mutational step, like the ones that seemed to be demonstrated in the small steps of the fitness graph on page 125? This seemed to Lenski unlikely, for an interesting reason. Knowing the average mutation rate of each gene in the genome of these bacteria, he calculated that 30,000 generations was long enough for every gene to have mutated at least once in each of the twelve lines. So it seemed unlikely that it was the rarity of the mutation that singled Ara−3 out. It should have been ‘discovered’ by several other tribes.
There was another theoretical possibility, and an extremely tantalizing one. This is where the story starts to get quite complicated so, if it is late at night, it might be an idea to resume reading tomorrow . . .
What if the necessary biochemical wizardry to feed on citrate requires not just one mutation but two (or three)? We are not now talking about two mutations that build on each other in a simple additive way. If we were, it would be enough to get the two mutations in any order. Either one, on its own, would take you halfway (say) to the goal; and either one on its own would confer an ability to get some nourishment from citrate, but not as much as both mutations together would. That would be on a par with the mutations we have already discussed for increasing body size. But such a circumstance would not be rare enough to account for the dramatic uniqueness of Tribe Ara−3. No, the rarity of citrate metabolism suggests that we are looking for something more like the ‘irreducible complexity’ of creationist propaganda. This might be a biochemical pathway in which the product of one chemical reaction feeds into a second chemical reaction, and neither can make any inroads at all without the other. This would require two mutations, call them A and B, to catalyse the two reactions. On this hypothesis, you really would need both mutations before there is any improvement whatsoever, and that really would be improbable enough to account for the observed result that only one out of the twelve tribes achieved the feat.
That’s all hypothetical. Could the Lenski group find out by experiment what was actually going on? Well, they could take great strides in that direction, making brilliant use of the frozen ‘fossils’, which are such a continual boon in this research. The hypothesis, to repeat, is that, at some time unknown, Tribe Ara−3 chanced to undergo a mutation, mutation A. This had no detectable effect because the other necessary mutation, B, was still lacking. Mutation B is equally likely to crop up in any one of the twelve tribes. Indeed, it probably did. But B is no use – has absolutely no beneficial effect at all – unless the tribe happens to be primed by the previous occurrence of mutation A. And only tribe Ara−3, as it happened, was so primed.
Lenski could even have phrased his hypothesis in the form of a testable prediction – and it is interesting to put it like this because it really is a prediction even though, in a sense, it is about the past. Here’s how I would have put the prediction, if I had been Lenski:
I shall thaw out fossils from Tribe Ara-3, dating from various points, strategically chosen, going back in time. Each of these ‘Lazarus clones’ will then be allowed to evolve further, on a similar regimen to the main evolution experiment, from which, of course, they will be completely isolated. And now, here’s my prediction. Some of these Lazarus clones will ‘discover’ how to deal with citrate, but only if they were thawed out of the fossil record after a particular, critical generation in the original evolution experiment. We don’t know – yet – when that magic generation was but we shall identify it, with hindsight, as the moment when, according to our hypothesis, mutation A entered the tribe.
You will be delighted to hear that this is exactly what Lenski’s student Zachary Blount found, when he ran a gruelling set of experiments involving some forty trillion – 40,000,000,000,000 – E. coli cells from across the generations. The magic moment turned out to be approximately generation 20,000. Thawed-out clones of Ara−3 that dated from after generation 20,000 in the ‘fossil record’ showed increased probability of subsequently evolving citrate capability. No clones that dated from before generation 20,000 did. According to the hypothesis, after generation 20,000 the clones were now ‘primed’ to take advantage of mutation B whenever it came along. And there was no subsequent change in likelihood, in either direction, once the fossils’ ‘resurrection day’ was later than the magic date of generation 20,000: whichever generation after 20,000 Blount sampled, the increased likelihood of those thawed fossils subsequently acquiring citrate capability remained the same. But thawed fossils from before generation 20,000 had no increased likelihood of developing citrate capability at all. Tribe Ara−3, before generation 20,000, was just like all the other tribes. Although its members belonged to Tribe Ara−3, they did not possess mutation A. But after generation 20,000, Tribe Ara−3 were ‘primed’. Only they were able to take advantage of ‘mutation B’ when it turned up – as it probably did in several of the other tribes, but to no good effect. There are moments of great joy in scientific research, and this must surely have been one of them.
Lenski’s research shows, in microcosm and in the lab, massively speeded up so that it happened before our very eyes, many of the essential components of evolution by natural selection: random mutation followed by non-random natural selection; adaptation to the same environment by separate routes independently; the way successive mutations build on their predecessors to produce evolutionary change; the way some genes rely, for their effects, on the presence of other genes. Yet it all happened in a tiny fraction of the time evolution normally takes.
There is a comic sequel to this triumphant tale of scientific endeavour. Creationists hate it. Not only does it show evolution in action; not only does it show new information entering genomes without the intervention of a designer, which is something they have all been told to deny is possible (‘told to’ because most of them don’t understand what ‘information’ means); not only does it demonstrate the power of natural selection to put together combinations of genes that, by the naïve calculations so beloved of creationists, should be tantamount to impossible; it also undermines their central dogma of ‘irreducible complexity’. So it is no wonder they are disconcerted by the Lenski research, and eager to find fault with it.
Andrew Schlafly, creationist editor of ‘Conservapedia’, the notoriously misleading imitation of Wikipedia, wrote to Dr Lenski demanding access to his original data, presumably implying that there was some doubt as to their veracity. Lenski had absolutely no obligation even to reply to this impertinent suggestion but, in a very gentlemanly way, he did so, mildly suggesting that Schlafly might make the effort to read his paper before criticizing it. Lenski went on to make the telling point that his best data are stored in the form of frozen bacterial cultures, which anybody could, in principle, examine to verify his conclusions. He would be happy to send samples to any bacteriologist qualified to handle them, pointing out that in unqualified hands they might be quite dangerous. Lenski listed these qualifications in merciless detail, and one can almost hear the relish with which he did so, knowing full well that Schlafly – a lawyer, if you please, not a scientist at all – would hardly be able to spell his way through the words, let alone qualify as a bacteriologist competent to carry out advanced and safe laboratory procedures, followed by statistical analysis of the results. The whole matter was trenchantly summed up by the celebrated scientific blogwit PZ Myers, in a passage beginning, ‘Once again, Richard Lenski has replied to the goons and fools at Conservapedia, and boy, does he ever outclass them.’
Lenski’s experiments, especially with the ingenious ‘fossilization’ technique, show the power of natural selection to wreak evolutionary change on a timescale that we can appreciate in a human lifetime, before our
very eyes. But bacteria provide other impressive, if less clearly worked-out, examples. Many bacterial strains have evolved resistance to antibiotics in spectacularly short periods. After all, the first antibiotic, penicillin, was developed, heroically, by Florey and Chain as recently as the Second World War. New antibiotics have been coming out at frequent intervals since then, and bacteria have evolved resistance to just about every one of them. Nowadays, the most ominous example is MRSA (methycillin-resistant Staphylococcus aureus), which has succeeded in making many hospitals positively dangerous places to visit. Another menace is ‘C. diff.’ (Clostridium difficile). Here again, we have natural selection favouring strains that are resistant to antibiotics; but the effect is overlain by another one. Prolonged use of antibiotics tends to kill ‘good’ bacteria in the gut, along with the bad ones. C. diff., being resistant to most antibiotics, is greatly helped by the absence of other species of bacteria with which it would normally compete. It is the principle of ‘my enemy’s enemy is my friend’.