Here by “reproduce” I just meant “make more copies of itself” in an immediate sense (so reproductive fitness is just “how fast it replicates right now”). For example, in Lenski’s long-term evolution experiment, some variants were selected not because they increased the bacteria’s daily growth rate, but because they made it easier to acquire further variants that themselves increased the daily growth rate. These “potentiating” variants were initially detrimental (the copy number of these variants decreased in the population), and only after a long long time they took over the population. So, according the definition of reproductive fitness I used, they lead to a lower reproductive fitness – the reason they were eventually selected for is not that they’re good for reproduction, but that they’re good for evolvability. Of course, you can say that eventually they increased in copy number, but that would be defining “reproduction” in a different way, that I find less intuitive.
Now, is that other definition (how gene copy number increases over the long term) what evolution ultimately selects for? I’m not sure. To quote Kokko’s review on the stagnation paradox:
“Trees compete for sunlight and attempt to outshade each other, but when each tree consequently invests in woody growth, the entire forest must spend energy in stem forming and—assuming time or energy trade-offs—will be slower at converting sunlight into seeds than a low mat of vegetation would have been able to. Every individual has to invest in outcompeting others, but the population as a whole is negligibly closer to the light source (the number of photons arriving in the area is still the same). This is why in agriculture, externally imposed group selection to create shorter crops has improved yields.”
She gives other examples. In these cases, the number of individuals tend to decrease over time, even in the long run.
(Sorry I missed your comment)
Here by “reproduce” I just meant “make more copies of itself” in an immediate sense (so reproductive fitness is just “how fast it replicates right now”). For example, in Lenski’s long-term evolution experiment, some variants were selected not because they increased the bacteria’s daily growth rate, but because they made it easier to acquire further variants that themselves increased the daily growth rate. These “potentiating” variants were initially detrimental (the copy number of these variants decreased in the population), and only after a long long time they took over the population. So, according the definition of reproductive fitness I used, they lead to a lower reproductive fitness – the reason they were eventually selected for is not that they’re good for reproduction, but that they’re good for evolvability. Of course, you can say that eventually they increased in copy number, but that would be defining “reproduction” in a different way, that I find less intuitive.
Now, is that other definition (how gene copy number increases over the long term) what evolution ultimately selects for? I’m not sure. To quote Kokko’s review on the stagnation paradox:
“Trees compete for sunlight and attempt to outshade each other, but when each tree consequently invests in woody growth, the entire forest must spend energy in stem forming and—assuming time or energy trade-offs—will be slower at converting sunlight into seeds than a low mat of vegetation would have been able to. Every individual has to invest in outcompeting others, but the population as a whole is negligibly closer to the light source (the number of photons arriving in the area is still the same). This is why in agriculture, externally imposed group selection to create shorter crops has improved yields.”
She gives other examples. In these cases, the number of individuals tend to decrease over time, even in the long run.