Why is the bottleneck idea widely accepted? It seems to my non-academic brain that each bottleneck would make an ongoing population less likely, as it is an opportunity for the population to die out. So it seems like us having this chat today is evidence that there weren’t too many bottleneck events. The more we would have had, the less likely it would be that humans would have survived until today. 0 bottleneck events seem most likely though that lens, 1 - less likely, 2 - even less, and so on.
As far as I understand it, the bottleneck idea is based on a back-calculation of "effective population size" in the past from existing genetic diversity in the present. (This is definitely the case in the article, which specifies "the population of breeding individuals".)
There are a couple of things that tend to be miscommunicated:
1. An effective population size of N means an actual population size that is significantly larger; children too young to reproduce, adults too old to reproduce, and nerds too awkward to reproduce are all not counted in effective population size.
2. If a group gets demographically wiped out at time T, their ancestors at time T-1000 vanish from the effective population at time T-1000, even though those same ancestors were part of the effective population at time T-1000 if you did the same calculation at time T-200.
(This is true to the extent that the vanishing ancestors don't also have other descendants in groups that survive. But the effect is quite significant - the effective size of the pre-Cherokee population in the year 1000 was much larger in the year 1450 than the same quantity, the effective size of the pre-1450-Cherokee population in the year 1000, is today. Was there a bottleneck? Sure; Amerinds in the region of the United States got wiped out. How long did that process take? When did it happen? We know a lot about the shape of the population over the relevant time period - is it something we're comfortable calling a "bottleneck"?)
So we might estimate our group of 1300 effective individuals as reflecting maybe 430 men of reproductive age, 870 women of reproductive age, plus 220 more women of reproductive age who fail to reproduce, 650 more men of reproductive age who fail to reproduce, and a few thousand children and elders not of reproductive age. And if this group later loses a war, their population could be a lot higher than that.
At minimum you need a breeding pair to keep the population going. That's a whole lot less than 1200, so from that macro perspective, I'd say it's fairly hard to wipe most things out, especially the things that have made it this far, just like it's hard for me to completely get rid of the duckweed that keeps growing in my fish tank. Even if I pluck out all the visible duckweed, there's always one or two in the filter that re-seed the population and within a few days the whole surface is covered again. Most life is a little bit like this, to the point where I don't think it's very improbable for a species to have bottlenecks. For a species to make it, surviving bottlenecks has to be a defining feature, and for the most part we are exposed to species that make it.
If you have a fair coin, and then you toss it ten times and it lands ten heads in a row, that result is no longer unlikely -- it has a probability (in whatever sense that word is now meaningful) of 100 %.
You can only judge hypotheses more or less likely in relation to each other. When you have multiple hypotheses that all explain the current situation, one may be more likely to have happened in the past than another. So if all you know is that on the first 20 tosses half were heads, you might say it's less likely the sequence was ten heads followed by ten tails, than some mix of heads and tails.
But of course, the difficult part is judging whether two hypotheses explain the same outcome. That's often somewhat subjective when it comes to historical analyses like these.
Once you've won the lottery, the odds of you having won the lottery is 1 no matter how small they were before the draw.
So, no, in other words, you can't draw that conclusion. Consider that in any instance where bottlenecks led to extinction we couldn't be here to discuss it, and so the only possibilities is that we won the lottery, whatever the odds were.
That’s an interesting point about the dependence of likelihood on the outcome.
However, a counter-argument is that if you win a lottery 250 times, it would be reasonable to conclude that the lottery is biased in your favor. If you lost a bet on a coin toss picking tails 250 times in a row, you’d assume the coin is biased.
In that case, every toss with the same outcome informs you of the bias.
So does an outcome really say nothing of the chance for that outcome?
In a similar way, our survival might say that a chance for extinction was sufficiently low each year/month/day/period. The longer we survive, the more confidently we can say that the possibility of survival was high and extinction — low.
If you knew the number of attempts, the outcome would say something of the probability.
So to your last paragraph, yes, we can say something about the odds of survival any given period, because we see many periods.
But we see only one instance of our survival, and we inherently need to be here to see it, so our survival in general for the whole period tells us nothing of how likely (few events reducing our chances) or unlikely (many such events) our survival was in the past, because no matter how unlikely we wouldn't be here to discuss the probability if we didn't survive, and unlikely things do happen.
Yes, if you look at survival as one instance, it does become difficult to say anything about the likelihood based on the outcome.
But I think it's too easy to dismiss this as survivorship bias. There are also other selection biases, which might explain the higher chances of survival of a species if an alive species is selected.
There's also the multiverse theory, or more locally, something we could call "multi-environment". It does make sense that if we were here at all, we would most likely be in an environment where we were most likely to survive. Imagine 20 planets, or 20 timelines, where some make our survival nearly impossible, some - very difficult, and some - relatively easy and likely. We'd probably find ourselves alive on a planet or in a timeline where it was relatively easy and likely.
I just read a bit more on this and it looks like the debate for whether human survival implies the low risk of extinction in the past is ongoing with good arguments on each sides. So I think I'll leave it up to more academic people than me to debate this, but thanks for your thoughts.
Whether or not there are other instances of long term survival, without having access to them, they can't form a basis for our probability calculations, so the point remains that our single survival is insufficient to use as a basis for the probability of events in our past.
Right?