5 ms·

I wonder - once we start replacing more extensive segments that represent complex functions, not only does it become a less probable natural solution but to ste

by tomxor 8y ago

I wonder - once we start replacing more extensive segments that represent complex functions, not only does it become a less probable natural solution but to steal a term from ML - perhaps there is a danger of temporal "over-fitting"... The natural version maybe sub-optimal in the current climate, but that historical baggage could have evolved to that state because it allowed the continuation of that species through a more diverse range of climates, something that could be useful in the future. There is the potential that forcing a "more efficient" solution for _now_, could be also creating a more fragile and endemic species? Perhaps not even for the distant future, if details in the original in-fact make it better able to cope with subtle climatic variations more efficiently.

sampo 8y ago
Crop plants and livestock animals are already pretty much unable to survive in the wild, outside of farms. So the "overfitting" is as old as farming.
- Gravityloss 8y ago
  The dangers of monoculture as well.
- tomxor 8y ago
  True, but this is direct, cultivation is by proxy, and does not necessarily remove unexpressed genes.
  - n4r9 8y ago
    That's a confusing thing to say... surely GMO doesn't necessarily remove unexpressed genes either? To provide a counterargument, with GMO we additionally have a decent understanding of the underlying mechanics and which genes have changed. This puts us in a better position to correct things if they do turn out to have unintended consequences.
    - tomxor 8y ago
      > That's a confusing thing to say... surely GMO doesn't necessarily remove unexpressed genes either? You may be correct, this was speculation based on imagination on my part - I do not know if unexpressed genes were removed in this experiment, I only have a suspicion they may have been part of the "inefficient" original sequence that was replaced.
- JohnJamesRambo 8y ago
  Livestock animals live in the wild pretty well. Google “feral hogs.”
  - GlenTheMachine 8y ago
    Not quite true. Some livestock animals do ok in the wild: hogs and, in some warmer climates, cattle. There are feral goats in the Caribbean. But the climate has to be right. You don't see feral goats in Maine. Or even Ohio. Domesticated chickens, OTOH, are toast anywhere there are predators. There are feral chickens in Hawaii. Nowhere else that I know of. There are also few feral sheep, as they need to be sheared. Domesticated horses also tend to not do well. They've been selectively bred for traits that make them somewhat fragile without specific care. As for crops: most domesticated food crops don't tend to become invasive. Every once in a while you'll see volunteer corn left over from last years' crop. and volunteer tomatoes are practically ubiquitous in backyard gardens. But more and more, the seeds from food crops are not fertile, they're hybrids. And even in backyard gardens of "heritage" crops, without human intervention the weeds almost always win out in the end. The only vegetable I've ever seen that persisted long after the gardener was gone was, oddly, rhubarb.
    - JohnJamesRambo 8y ago
      That's a far cry from OP's comment that livestock animals are "pretty much unable to survive in the wild." I agree with you that crop plants are much more fragile and unable to compete without watering and fertilization. Tomatoes bred to make big 1 lb bags of water for their seeds will have a rough time compared to the tiny tomatoes made by the wild ancestor of tomatoes. https://www.smithsonianmag.com/travel/why-wild-tiny-pimp-tomato-so-important-180955911/ https://www.smithsonianmag.com/travel/why-wild-tiny-pimp-tom...
      - GlenTheMachine 8y ago
        I'd argue that this is partly a philosophical argument: what do we mean when we say “the wild”? None of the feral animals listed, with the possible exception of feral pigs, could survive if humans hadn't killed off the major predator species. There are wild cattle where I grew up, cows that escaped one of the local farms. But there are no wolves or mountain lions, which in a true natural state would, I suspect, would take them down in short order. So to be generous to the OP, he might have a point.
- candle_sticks 8y ago
  this is over... simplifying
btown 8y ago
That baggage may be more useful insofar as it allows species to adapt more quickly over generations to new climates. In contrast, the techniques we develop based on our intelligence can be put into effect much more quickly, effectively making us more agile in the face of change. That said, monocultures increase our vulnerability to a class of threats even faster-evolving than our technology: diseases. As with all things, we should approach this with balance.
adrianN 8y ago
If you're concerned about this you should financially support some seed banks. They effectively keep backups.
- tomxor 8y ago
  Indeed, they are worthy cause. However I am not concerned, just interested. I've also visited kew and donated to their seed bank. I think it's important to try to reason about the evolution of the things we attempt to modify but is not necessarily a cause for alarm so long as one modified species is not so aggressive as to risk completely supplanting other variations.
dekhn 8y ago
I think the right way to think about it is that modern enzymes are at least partly optimized to handle a wide range of conditions (more or less what you said), but also lock in a fair amount of legacy details that would be hard to evolve out due to the amount of dependencies on them (everything in biology is tightly related, even modular systems leak). I think of it in an RL context- evolution has acted as a subtle force on the fitness, in which natural selection evaluates mutation- but the resulting actions and policies are extremely subtle and integrate a wide range of conditions, and weave a large number of players together. Then scientists come along an re-optimize for a single variable and claim they've "fixed a bug". No, it wasn't a bug, you just don't understand the full aspects of the biology. (note: I work in ML and biology full time, but what I say above is effectively non-scientific, impossible to prove, and mainly for enjoyable thought-discussions).
- tomxor 8y ago
  > but also lock in a fair amount of legacy details that would be hard to evolve out due to the amount of dependencies on them (everything in biology is tightly related, even modular systems leak). I remember visiting a clone of the Encephalartos_woodii (AKA "the loneliest plant in the world") which is thought to have evolved into a genetic dead end due to it's massive genome which is said to be 20 times the size of the human genome... I wondered if this was due to some kind of dependence, which sounds a lot like what you are saying?
  - dekhn 8y ago
    it's not clear what the plants that have super-large genomes are doing. Some of them are perfectly capable and show little to no negative issues associated with their genomes. I think the case you're describing, there was just a large full-genome duplication that was tolerated (somehow the gene regulation still works and the appropriate amount of protein products are expressed, etc). What I meant more is that most enzymes work in regulated webs and you can't easily tweak one part of the web without having knock-on effects on all the other genes. Nature seems to handle this via gene duplication- once a gene is duplicated, it seems like one can carry on the original function for its dependents, while another is freer to evolve new functions. Sorry I don't have a better explanation for this- to really talk about this you kind of need to have a huge background and know a ton of biology and have read all the debates from all the players about fitness and evolution and even then, nobody really has good answers for this.
bsder 8y ago
Maximally efficient is minimally robust.
candle_sticks 8y ago
there's a lot of baggage with photosynthesis because plants have been around for a long, long time. in fact, they have been around and have existed in so many climates and microclimates that there are actually different types of photosynthesis: CAM, C2, C3, and C4. some came about because the gases in atmosphere changed over time, others because of heat and humidity. if you have a half dozen or so houseplants, most likely you have plants that create food via 2 different types of photosynthesis. for example cacti use CAM photosynthesis to prevent water loss during the day when it's really hot out. a number of succulents also use this and it's such an important part of survival in specific climates that it's actually an example of convergent evolution where something like the euphorbia plants in Africa developed while never sharing a close genetic ancestor with the cacti and succulents of North and Central America.
randomuser0 8y ago
From a past life as a biologist looking at photosynthesis, IIRC, at least the photosynthetic systems of algae were optimized for (1) "dynamic range" of sunlight and (2) molecular stability. (1) Intermittent flecks of light - not guaranteed continuous sunlight in the wild. (2) every photon captured destabilizes the molecular assembly and some captured energy may be better lost than build up oxidative stress. Of course, this is from a ten year old memory and may be completely wrong.

by tomxor 8y ago

sampo 8y ago

Crop plants and livestock animals are already pretty much unable to survive in the wild, outside of farms. So the "overfitting" is as old as farming.

Gravityloss 8y ago
The dangers of monoculture as well.
tomxor 8y ago
True, but this is direct, cultivation is by proxy, and does not necessarily remove unexpressed genes.
- n4r9 8y ago
  That's a confusing thing to say... surely GMO doesn't necessarily remove unexpressed genes either? To provide a counterargument, with GMO we additionally have a decent understanding of the underlying mechanics and which genes have changed. This puts us in a better position to correct things if they do turn out to have unintended consequences.
  - tomxor 8y ago
    > That's a confusing thing to say... surely GMO doesn't necessarily remove unexpressed genes either? You may be correct, this was speculation based on imagination on my part - I do not know if unexpressed genes were removed in this experiment, I only have a suspicion they may have been part of the "inefficient" original sequence that was replaced.
JohnJamesRambo 8y ago
Livestock animals live in the wild pretty well. Google “feral hogs.”
- GlenTheMachine 8y ago
  Not quite true. Some livestock animals do ok in the wild: hogs and, in some warmer climates, cattle. There are feral goats in the Caribbean. But the climate has to be right. You don't see feral goats in Maine. Or even Ohio. Domesticated chickens, OTOH, are toast anywhere there are predators. There are feral chickens in Hawaii. Nowhere else that I know of. There are also few feral sheep, as they need to be sheared. Domesticated horses also tend to not do well. They've been selectively bred for traits that make them somewhat fragile without specific care. As for crops: most domesticated food crops don't tend to become invasive. Every once in a while you'll see volunteer corn left over from last years' crop. and volunteer tomatoes are practically ubiquitous in backyard gardens. But more and more, the seeds from food crops are not fertile, they're hybrids. And even in backyard gardens of "heritage" crops, without human intervention the weeds almost always win out in the end. The only vegetable I've ever seen that persisted long after the gardener was gone was, oddly, rhubarb.
  - JohnJamesRambo 8y ago
    That's a far cry from OP's comment that livestock animals are "pretty much unable to survive in the wild." I agree with you that crop plants are much more fragile and unable to compete without watering and fertilization. Tomatoes bred to make big 1 lb bags of water for their seeds will have a rough time compared to the tiny tomatoes made by the wild ancestor of tomatoes. https://www.smithsonianmag.com/travel/why-wild-tiny-pimp-tomato-so-important-180955911/ https://www.smithsonianmag.com/travel/why-wild-tiny-pimp-tom...
    - GlenTheMachine 8y ago
      I'd argue that this is partly a philosophical argument: what do we mean when we say “the wild”? None of the feral animals listed, with the possible exception of feral pigs, could survive if humans hadn't killed off the major predator species. There are wild cattle where I grew up, cows that escaped one of the local farms. But there are no wolves or mountain lions, which in a true natural state would, I suspect, would take them down in short order. So to be generous to the OP, he might have a point.
candle_sticks 8y ago
this is over... simplifying

btown 8y ago

That baggage may be more useful insofar as it allows species to adapt more quickly over generations to new climates. In contrast, the techniques we develop based on our intelligence can be put into effect much more quickly, effectively making us more agile in the face of change. That said, monocultures increase our vulnerability to a class of threats even faster-evolving than our technology: diseases. As with all things, we should approach this with balance.

adrianN 8y ago

If you're concerned about this you should financially support some seed banks. They effectively keep backups.

tomxor 8y ago
Indeed, they are worthy cause. However I am not concerned, just interested. I've also visited kew and donated to their seed bank. I think it's important to try to reason about the evolution of the things we attempt to modify but is not necessarily a cause for alarm so long as one modified species is not so aggressive as to risk completely supplanting other variations.

dekhn 8y ago

I think the right way to think about it is that modern enzymes are at least partly optimized to handle a wide range of conditions (more or less what you said), but also lock in a fair amount of legacy details that would be hard to evolve out due to the amount of dependencies on them (everything in biology is tightly related, even modular systems leak). I think of it in an RL context- evolution has acted as a subtle force on the fitness, in which natural selection evaluates mutation- but the resulting actions and policies are extremely subtle and integrate a wide range of conditions, and weave a large number of players together. Then scientists come along an re-optimize for a single variable and claim they've "fixed a bug". No, it wasn't a bug, you just don't understand the full aspects of the biology. (note: I work in ML and biology full time, but what I say above is effectively non-scientific, impossible to prove, and mainly for enjoyable thought-discussions).

tomxor 8y ago
> but also lock in a fair amount of legacy details that would be hard to evolve out due to the amount of dependencies on them (everything in biology is tightly related, even modular systems leak). I remember visiting a clone of the Encephalartos_woodii (AKA "the loneliest plant in the world") which is thought to have evolved into a genetic dead end due to it's massive genome which is said to be 20 times the size of the human genome... I wondered if this was due to some kind of dependence, which sounds a lot like what you are saying?
- dekhn 8y ago
  it's not clear what the plants that have super-large genomes are doing. Some of them are perfectly capable and show little to no negative issues associated with their genomes. I think the case you're describing, there was just a large full-genome duplication that was tolerated (somehow the gene regulation still works and the appropriate amount of protein products are expressed, etc). What I meant more is that most enzymes work in regulated webs and you can't easily tweak one part of the web without having knock-on effects on all the other genes. Nature seems to handle this via gene duplication- once a gene is duplicated, it seems like one can carry on the original function for its dependents, while another is freer to evolve new functions. Sorry I don't have a better explanation for this- to really talk about this you kind of need to have a huge background and know a ton of biology and have read all the debates from all the players about fitness and evolution and even then, nobody really has good answers for this.

bsder 8y ago

Maximally efficient is minimally robust.

candle_sticks 8y ago

there's a lot of baggage with photosynthesis because plants have been around for a long, long time. in fact, they have been around and have existed in so many climates and microclimates that there are actually different types of photosynthesis: CAM, C2, C3, and C4. some came about because the gases in atmosphere changed over time, others because of heat and humidity. if you have a half dozen or so houseplants, most likely you have plants that create food via 2 different types of photosynthesis. for example cacti use CAM photosynthesis to prevent water loss during the day when it's really hot out. a number of succulents also use this and it's such an important part of survival in specific climates that it's actually an example of convergent evolution where something like the euphorbia plants in Africa developed while never sharing a close genetic ancestor with the cacti and succulents of North and Central America.

randomuser0 8y ago

From a past life as a biologist looking at photosynthesis, IIRC, at least the photosynthetic systems of algae were optimized for (1) "dynamic range" of sunlight and (2) molecular stability. (1) Intermittent flecks of light - not guaranteed continuous sunlight in the wild. (2) every photon captured destabilizes the molecular assembly and some captured energy may be better lost than build up oxidative stress. Of course, this is from a ten year old memory and may be completely wrong.