Improving on nature?


I have two invited reviews due this summer, building on the theme from my book, that past natural selection improved trees (and the wild ancestors of our crops) much more than it has improved the overall organization of forests (and other natural ecosystems):

In Global Food Security, Andy McGuire and I will ask, "What can agriculture learn from nature?" If natural selection or some other process had consistently improved the overall organization of natural ecosystems, then agriculture might benefit from copying that organization. If every natural ecosystem had some process that adjusted the relative abundance of species to maximize ecosystem-level productivity and/or stability, then we could (for example) try to match the ratio of grasses and legumes in our pastures to those in nearby grazed meadows. I expect to argue, however, that nothing has consistently improved natural-ecosystem organization, so mindless mimicry of natural ecosystems is unlikely to improve agriculture. The wild ancestors of key crops grew naturally as monocultures, but that doesn't necessarily mean polyculture wouldn't be better. It's still worth studying how natural-ecosystem organization affects productivity and stability, and thinking about which features of natural ecosystems might be worth copying.

In "Evolutionary tradeoffs as crop-improvement opportunities", intended for Field Crops Research , I will argue that past natural selection has been improving individually-beneficial plant traits like drought tolerance for millions of years, leaving few simple, tradeoff-free options for further improvement. Accepting tradeoffs rejected by past natural selection has been key to past crop improvement and that is probably still true.

For a preview, see my discussions with farmer/blogger Chris Smaje and soybean-breeder Clem Weidenbenner in the comments for this post on Small Farm Future.

Chris argues that rotating annual crops with pasture is copying nature. I don't see any close analogs to such rotations in nature, so disagree. The pasture phase might benefit from copying some aspects of natural grazing systems, though.

Clem has various examples of plant breeding improving crops in ways that natural selection hasn't. I agree, but would any of those changes have improved individual-plant fitness in nature? If not, what are the prospects for improving traits like stress tolerance, which would (if tradeoff-free) have improved individual fitness?

Increasing or decreasing oil content beyond its natural range would presumably decrease fitness, even though it may be useful to us. Clem mentions range expansion of crops, which could show that humans can improve traits like cold tolerance in ways that past natural selection on the crop's wild ancestors didn't. I need to read more about this, but I find it interesting that high-altitude maize picked up cold-tolerance genes from teosinte, not the other way around.


Wondered whether you might bring this conversation home (err.. here).

As for increasing or decreasing grain protein and/or oil content beyond the range that evolution has established WITHOUT the hand of man might be a more apt way to put this. So - as this is useful to us, when we deploy such materials to wide acreage aren't we by default increasing these domesticate's fitness?

** Yes, but this doesn't affect my argument that improving traits that have already been improved by millions of years of natural selection for individual fitness (such as stress tolerance) is difficult, whereas improving traits that only enhance fitness in modern agriculture (better taste, as evaluated by humans, or adaptation to high soil fertility) is much easier. -- Ford **

On the matter of cold tolerance genes moving from teosinte to maize instead of the other way round - Hufford et al actually discuss toward the end of their paper that this should be expected... as maize in this particular instance is the invader, or 'new kid on the block'. Teosinte, having colonized the higher altitudes long before maize came to be would have already solved the cold tolerance problem.

** Similarly, I would argue the wild ancestors of our crops had already solved the drought tolerance problem, except to the extent that there are tradeoffs rejected by natural selection that are acceptable to us.**

No need to reinvent the wheel (if you bother to maintain some sort of cross compatibility with your progenitor - which is actually the part I find more interesting). Maize/teosinte hybrids then, following on would likely find it possible to acclimate to altitude (thus cold) using the teosinte genes. Hufford et al also discuss the point that this result makes sense from a population density aspect (the invading maize having fewer gametes in the air to potentially pollinate the teosinte [I think we should be calling this mexicana to be more precise].

Now lets change from high altitude Central America to high latitude North America... say Ontario and the Dakotas (or even Minnesota :) These are ecosystems where no maize ancestors had spread to before Homo sapiens brought them. The cold tolerance mechanism(s) in the modern cultigen may or may not be related to the mechanism(s) in the Mexican germplasm -

** Do you know a good review about those mechanisms? It would be interested to compare them with those that evolved naturally.**

but even if they are they would not be appropriately packaged with other genes for high latitude adaptation. And again, I'm not arguing that evolution could not have 'solved' this situation... just that it didn't (unless you allow that as natural beings we humans are merely helping evolution do what it does). Context.

There are several papers I've run across since seeing your original comment at Chris' blog. These center on sunflower (another native to America... this one having a much wider latitudinal adaptation and sympatric overlaps of wild and cultivated with some remaining cross compatibility). Many of these papers have Loren Rieseberg as an author. One perhaps most relevant to the abiotic tolerance is Whitney et al 2010 New Phytologist 187:230-239. Others can be found through Google Scholar - Rieseberg has a profile there.

** That paper shows adaptive introgression from one wild sunflower species to another. If we're comparing improvement by humans to improvement by natural selection over millennia, we need to look at introgression from cultivated species to wild relatives, and ask whether the traits that "stick" are broadly useful traits like drought tolerance (which would convince me of the superiority of human selection for such traits) or narrow adaptations to recently-evolved pests or to herbicides. I agree that human-mediated evolution can be faster than natural selection -- we breed for resistance to new diseases, rather than waiting for natural selection -- but natural selection has been working on drought for so much longer....**

Ford, with respect I think you’re misrepresenting my position a little.

** Sorry (my responses labeled ** -- Ford)

This part of my argument arose in the context of current efforts to breed perennial grain crops, where I think I’m in agreement with you that this is ‘misguided mimicry’ inasmuch as there’s a tradeoff between perenniality and seed allocation which can’t readily be optimised. What farmers have often done instead historically is split their farms between an annual grain part and a perennial grass part, each of which has its pros and cons but which delivers net benefits across the whole system. I agree with you that the rotational aspect of this is not found, or only barely found, in nature, but I don’t think that this is the most important thing to focus on. The ‘mimicry’ is in appreciating what annual grains and perennial grasses respectively can give you, and the ‘improvement’ is in devising a rotation between the two that gives you (sort of) the best of both worlds.

** I agree.**

From this perspective, I think emphasising the contrast between mimicry and improvement is not especially enlightening.

** I meant to emphasize the contrast between mimicry of natural ecosystems (not consistently improved by natural processes) versus mimicry of the adaptations of wild species, which have been improved (at least in terms of individual-plant fitness) by natural selection. **

If you adopt a very strong definition of ‘mimicry’, ie. that mimicry must involve pretty much the exact copying of natural ecosystems, then agriculture is never mimicry, always improvement.

** I agree that exact copying is rarely possible and so would not use such a narrow definition. But even imperfect copies of some wild-species adaptations may be better than anything we could invent from scratch.

But then we probably have to get into a semantic argument about what ‘mimicry’ really means. I don’t think I’m in fundamental disagreement with you or Andy’s basic arguments, but I feel Andy’s posts overstate the implications of his stances on natural ‘balance’, agricultural ‘improvement’ etc – particularly when you put it into the context of social history and the fairly disastrous career of ‘agricultural improvement’. I suspect there are also some tradeoffs in terms of labour/energy input, ecosystem services and biodiversity if you focus too single-mindedly on improvement.

** We need to use multiple criteria for improvement, but I don't expect everyone to agree on those criteria.**

So I agree, mindless mimicry is out, but maybe there’s scope for mindful mimicry – and it’s not always entirely obvious what we need to be mindful about before the fact.

Thanks, Ford.

** I meant to emphasize the contrast between mimicry of natural ecosystems (not consistently improved by natural processes) versus mimicry of the adaptations of wild species, which have been improved (at least in terms of individual-plant fitness) by natural selection. **

Yes, that's a good point which I elided a bit. As per the post on my blog, there are still some issues about emergent properties, labour/energy issues and ecosystem services that make me a bit nervous about embracing 'improvement' quite as wholeheartedly as Andy does - however, your point about the non-improvement of ecosystems is surely a powerful one that needs to be borne in mind. In fact, I received a grant to plant a forest garden, spent a year procrastinating because I wasn't sure why I was doing it, bought your book, and then gave the grant money back shortly afterwards. So I'm financially the poorer for encountering your work, but intellectually the richer I hope...

This seems awkward - replying to myself... but here goes:

** That paper shows adaptive introgression from one wild sunflower species to another. If we're comparing improvement by humans to improvement by natural selection over millennia, we need to look at introgression from cultivated species to wild relatives, and ask whether the traits that "stick" are broadly useful traits like drought tolerance (which would convince me of the superiority of human selection for such traits) or narrow adaptations to recently-evolved pests or to herbicides. [Ford's inserted comment]

I'm having a hard time with your premise that somehow we need to have genes from a cultigen being passed to a wild relative (and then 'stick') in order to demonstrate the value of human breeding. I think this view assumes that genes can flow evenly back and forth between the two, and I'm not sure this assumption has been vetted. Do you know whether it has?

And if one does demonstrate that a gene from a cultigen is passed to the wild progenitor - what sort of test will be needed to answer whether the resulting hybrid has received the necessary benefit?

A relatively simple demonstration of the value of selection for adaptation to a new environment can be made with a common garden plot of a cultigen grown along side a wild relative. But this too assumes far too much (but it does create images that capture the imagination).

Ford said:
Plant breeders who focus on traits we value, but which were not favored by past natural selection, will therefore make faster progress than those who try to beat natural selection at its own game.

Nice summary and I agree with only a minor quibble (which may be more due to a personal perspective than any real philosophical difference). Breeding from a commercial perspective is necessarily a bet that one can achieve some sort of change in the germplasm that will yield a downstream benefit of sufficient value to repay the investment made and provide some profit for the effort. Under this rubric then, I don't need to beat natural selection at its game... I need to beat the other plant breeder(s). Its like the story of the two fly fishermen at the trout stream when a bear comes down to the opposite bank. Our first fisherman beats it for the bank and starts getting out of his waders and into his boots. Our second fisherman scoffs: "You'll never outrun that bear"; to which our first replies: "Its not the bear I'm trying to outrun".

So, trying to beat natural selection is not the only play to consider. But even when one ignores the commercial end and considers just the breast beating "I can do anything better than you" human arrogance there can still be cases where we find ourselves positioned today in a footrace with natural selection where we are not necessarily thousands of years too late to the party.

Consider all the extant germplasm that exists today - regardless of how long nature has been testing it and imposing selective pressures on it. All these genes and gene combinations have strengths and weaknesses yet. As we come to the field and bring our own particular values (values not necessarily in step with simple fitness) AND with our values our ability to 'coddle' or modify a specific ecosystem to favor a certain outcome - then it seems to me we change the nature of the race. But lets be more specific.

Drought tolerance. In our conceptual foot race each side gets to start with the same resources - the extant germplasm [conceptually this may seem unfair, we didn't develop all the extant germplasm - but thems the breaks]. On Nature's side she goes about her business, rainfall or lack thereof occurring stochastically. On our side - rain out shelters, irrigation if necessary to prevent untimely death and total loss of germplasm (Nature gets a 'go back to start' card in this latter event). We get to double haploids at will. We get to generate and leverage enormous numbers of new genetic combinations. And we get to grow our materials in multiple nurseries per calendar year. Nature has to rely on chance cross pollinations, occasional mutations, and hit or miss screening nurseries. Poor gal - its as though we have her bound, gagged, and in chains.

But to what avail are all these modern tools if she has already solved the problem? I take it this is your central point? Well, even small (read really tiny) improvements can take on enormous value when deployed across such a vast landscape as the human agricultural ecosystem. If one can develop a new cultivar with only a 0.1% increase in yield (coming say from a miniscule change in water use efficiency)- and this cultivar is adapted to about 10% of North America's corn belt - which for the moment comprises about 91 million acres - then a little farming math continues as follows: 10% of 91 million acres is 9.1 million. Typical US average corn yield of 148 bu per acre should give us 1.3 billion bushels... for which a one tenth of one per cent increase is 1.3 million bushels. Today on the CBOT corn is trading at $4.42 per bushel. Nature may not give a fig what corn is trading for, but more than a handful of her humans do.

So after all the sophistication of modern breeding technology is brought to bear - we may only push the needle a teensy tiny bit in our favor... but that little bit can be leveraged to great advantage.

Yeah, a 50% increase... I'm not that optimistic either. And I agree with the wisdom of seeing various differences in terms of the trades being made.

Sounds like we're in agreement then... time for a beer.


I’m not wholly convinced that forest fires or wetland emissions can necessarily be regarded unambiguously as disservices, but leaving that aside I’m just musing on your arguments in relation to my example of a semi-natural permanent pasture or rangeland. So, this ecosystem isn’t optimised by natural selection at the level of the ecosystem, but it comprises various plants which are variously well adapted to various different circumstances – cold, heat, flood, drought, fire, pests/diseases, nutrient pulses etc. The way we’ve generally ‘improved’ such pastures, at least where I live, is by selecting for species like perennial ryegrass which produce a lot of biomass in response to high levels of fertilisation, leading to more meat per hectare (albeit also with the addition of concentrates from outside the system). The mimetic strategy might accept the loss of some productivity in return for retaining that multiply adaptive potential to various eventualities and system stresses in the mixed ‘non-improved’ grass, and look to some different kinds of ‘improvement’ in the herbivore-ruminant-grass interactions. Or to frame it in the terms you used in your comment on my blog, how realistic is the assumption that species interactions in the semi-natural polycultures will really be ‘random’ with respect to human interests? I can see that they would be more random with respect to one or a few human interests than species chosen specifically by humans for their capacity to furnish those interests. However, since they’re quite well adapted in aggregate evolutionarily to cope with the range of conditions that our grassland will experience, with fairly minimal intervention from us, and since we have a human interest in the persistence of the grassland, might it be that this distribution is not so random and we’re accepting a poor tradeoff whereby the additional productivity does not ultimately compensate for the additional work and the additional risks involved in achieving it? Maybe that’s not inevitably the case, but how realistic is the opposite view of complete randomness with respect to human interests?

Thanks for that clarification - makes sense to me. I'll let you get back to your lab now, while I go and precision plant some seeds...

Wondering if you've seen Henry & Nevo Pl Biotech J. 2014 Exploring natural selection to guide breeding for agriculture.

Haven't read it through yet, but thus far it has touched on SO much of what we've talked about here.

Unfortunately it doesn't look like you've been cited. Ausies...

Leave a comment