On the civilizational importance of dirt
And an unremarked EA opportunity
David R Montgomery’s Dirt was either one of the more eye opening books I’ve read, or it was on the order of Paul Ehrlich, and I actually didn’t know which was the case when I first finished it.
But don’t you just love the variance in those outcomes?
Any book that has such a high variance potential has to be worth reading and thinking about, because if you’ve reached that point, you have a large hole in your world model, and it’s an opportunity to fill it.
So what is the essential thesis?
Basically, we are losing fertile soil to erosion and lazy industrial farming practices at a noticeably higher rate than soil is created.
“In a broad sense, the history of many civilizations follows a common story line. Initially, agriculture in fertile valley bottoms allowed populations to grow to the point where they came to rely on farming sloping land. Geologically rapid erosion of hillslope soils followed when vegetation clearing and sustained tilling exposed bare soil to rainfall and runoff. During subsequent centuries, nutrient depletion or soil loss from increasingly intensive farming stressed local populations as crop yields declined and new land was unavailable. Eventually, soil degradation translated into inadequate agricultural capacity to support a burgeoning population, predisposing whole civilizations to failure. That a similar script appears to apply to small, isolated island societies and extensive, transregional empires suggests a phenomenon of fundamental importance. Soil erosion that outpaced soil formation limited the longevity of civilizations that failed to safeguard the foundation of their prosperity-their soil.
“Ah hah!” you say, “the Romans and Aztecs and whoever didn’t have powerful modern fertilizers, which we create out of fossil fuels, so soil mattered to them for agricultural output, but not to us.”
After all, hasn’t agricultural output been increasing for as long as we’ve been tracking it?
In fact, our land use has been staying flat or even getting more efficient in many cases!
You can ‘t tell me that’s bad for soil - using the same or smaller amounts of land to get 10-20x as much crops out of it pretty much has to be a good thing.
But Montgomery would say:
Modern society fosters the notion that technology will provide solutions to just about any problem. But no matter how fervently we believe in its power to improve our lives, technology simply cannot solve the problem of consuming a resource faster than we generate it: someday we will run out of it.”
Broadly, even if we use our soil 20x more productively we are still going through it much faster than it’s regenerated, and this will still ultimately screw us.
“Whereas the effects of soil erosion can be temporarily offset with fertilizers and in some cases irrigation, the long-term productivity of the land cannot be maintained in the face of reduced soil organic matter, depleted soil biota, and thinning soil that so far have characterized industrial agriculture.”
It seems we’re eroding and losing soil through poor agricultural practices at about 10-100x the rate it’s regenerating.
That’s according to a 2007 paper Montgomery wrote.1 And what are the consequences of that? Over the long run, reduced output, Malthusian dynamics, and in the limit, civilizational collapse, at least for the Romans, Mayans, Aztecs, and most other studied populations.
Montgomery gets a lot of his data from sediment cores in lake beds, which you can inspect volumes and pollen distributions from to know how fast surrounding erosion was, and what the surrounding plant life looked like at various times.
“Erosion from agriculture also caused abandonment of parts of southern Central America. Pollen from a long core pulled from the bottom of La Yeguada's small lake in central Panama records that the rainforest was cleared for slash-and-burn agriculture between 7,000 and 4,000 years ago. Archaeological records from this period indicate considerable population growth as intensified agriculture stripped the forest from the lake's watershed. By the time of Christ, accelerated erosion in the foothills and uplands led to agricultural abandonment of the watershed. Slow forest regeneration suggests depleted soils, and later agricultural settlements were concentrated along previously unoccupied floodplains and coastal valleys.”
“Soils at a site in the Gila National Forest, typical of prehistoric agricultural sites in the American Southwest, were cultivated between AD 1100 and 1150, at the peak of Pueblo culture, and subsequently abandoned. Soils of sites cultivated by the Pueblo culture are lighter colored, with a third to a half of the carbon, nitrogen, and phosphorus content of neighboring uncultivated soils. In addition, cultivated plots had gullies-some more than three feet deep-that began during cultivation. Even today, little grass grows on the ancient farm plots. Native vegetation cannot recolonize the degraded soil”
This is how he builds up his overall pictures of societies burgeoning, expanding to more marginal lands and hillsides, depleting the soil, then collapsing.
“This pattern suggests a fundamental model of agricultural development in which prosperity increases the capacity of the land to support people, allowing the population to expand to use the available land. Then, having eroded soils from marginal land, the population contracts rapidly before soil rebuilds in a period of low population density.”
“This roller-coaster cycle characterizes the relation between population and food production in many cultures and contexts because the agricultural potential of the land is not a constant-both technology and the state of the soil influence food production. Improved agricultural practices can support more people with fewer farmers, but soil health eventually determines how many people the land can support.”
He then points to it as a significant factor in civilizational collapse.
“Time and again, social and political conflicts undermined societies once there were more people to feed than the land could support. The history of dirt suggests that how people treat their soil can impose a life span on civilizations.”
“Yet why would so many unrelated civilizations like the Greeks, Romans, and Mayans all last about a thousand years? Clearly, the reasons behind the development and decline of any particular civilization are complex. While environmental degradation alone did not trigger the outright collapse of these civilizations, the history of their dirt set the stage upon which economics, climate extremes, and war influenced their fate. Rome didn't so much collapse as it crumbled, wearing away as erosion sapped the productivity of its homeland.”
In the Mayans:
“Sediment cores from lakes in the Mayan heartland suggest that agricultural intensification increased soil erosion. The rate that sedimentation piled up on lakebeds increased substantially from 250 BC through the ninth century AD. While not necessarily responsible for the collapse of Mayan society, soil erosion peaked shortly before Mayan civilization unraveled about AD 600 when the food surpluses that sustained the social hierarchy disappeared. Some Mayan cities were abandoned with buildings half finished.”
In the Romans:
“Two centuries later Pertinax offered central Italy's abandoned farmland to anyone willing to work it for two years. Few took advantage of his offer. Another century later Diocletian bound free farmers and slaves to the land they cultivated. A generation after that, Constantine made it a crime for the son of a farmer to leave the farm where he was raised. By then central Italy's farmers could barely feed themselves, let alone the urban population. By AD 395 the abandoned fields of Campagna were estimated to cover enough land to have held more than 75,000 farms in the early republic. The countryside around Rome had fed the growing metropolis until late in the third century BC. By the time of Christ, grain from the surrounding land could no longer feed the city.”
In the Aztecs:
“The soils in Mexico's "cradle of maize" in the Tehuacan Valley about a hundred miles southeast of Mexico City also bear witness to extensive pre-Columbian soil erosion. Remnants of subsoil exposed at the ground surface documented soil erosion from fields, leaving a modern soil that consists of little more than a thin mantle of broken rock. In contrast, areas with little evidence of past cultivation contained a foot and a half of well-developed soil above weathered rock. Abrupt transitions in soil depth between long-cultivated and uncultivated areas suggest that a foot and a half of soil was missing from the farms.”
And the long term effects of soil erosion and degredation are indeed, long term. Medieval farmers in Europe used to own little patches and stripes of different fields, located all over a given area.
Scholars argue that farmers would only put up with “an inconvenient arrangement in which a farmer had no say in the rotation or type of tillage used on his fields - which could be quite distant from each other” for good reasons. And those reasons are that a lone farmer can’t keep enough livestock to fertilize a degraded plot of soil enough to generate adequate harvests, but the collectively pooled livestock of the whole village would be able to generate enough fertilizer to do this.
One thing that makes me feel a little better about all this - all these declines in output and Malthusian bounces played out over many hundreds of years, and some took roughly a thousand years.
But how comforting IS that, really? Hasn’t Europe been populated and farmed for thousands of years? China? India? Have we been running some of these “thousand year soil clocks” out?
And aren’t there notable failure cases where the land *didn’t* bounce back, too? Indeed, he points many out! What about Iraq, aka Mesopotamia, one of the cradles of civilization? Doesn’t look like a breadbasket or Fertile Crescent any more, does it?
Another one - pretty much all of North Africa. These used to be breadbaskets (or at least wine and olive baskets) in the Roman empire. Part of the reason they wanted Carthage was for its land and agricultural capacity! But only Egypt still produces significant crops, due to being a floodplain of the Nile. All the others had their soil eroded to the point it was all swept away, and there’s very little agriculture there now outside of Egypt.
Wait a minute, we keep talking about soil eroding faster than it’s replaced - how is soil replaced? What generates soil?
In a word, worms.
Dirt is tiny rocks that you get from erosion of underlying minerals.
Soil is a teeming ecosystem made of dirt, decomposing organic matter, bacteria, fungi, and microbes. Soil is rich in nutrients, specifically the nutrients like phosphorus, potassium, and nitrogen needed for plants and crops.
This was a topic of study of Darwin himself - the last book he published before his death was on worms.
The way that we get most agricultural soil is worms - worms break down minerals into dirt (when dissected, worm gizzards contain small rocks and grains of sand), and they mix that dirt in with leaves and organic matter that they tear into small pieces as they eat. Worms make soil by slowly “plowing, breaking up, reworking, and mixing dirt derived from fresh rocks with recycled organic matter.”
“When we behold a wide, turf-covered expanse, we should remember that its smoothness, on which so much of its beauty depends, is mainly due to all the inequalities having been slowly leveled by worms.”
Worms plow soil, but over centuries.
Why did all these otherwise quite capable empires let their soil degradation happen?
One of the dynamics Montgomery points to is misaligned incentives, classic “get a return now” at the expense of long term sustainability. He terms this “mining soil fertility.”
“Over the course of history, economics and absentee ownership have encouraged soil degradation - on ancient Rome's estates, nineteenth-century southern plantations, and twentieth-century industrialized farms. In all three cases, politics and economics shaped land-use patterns that favored mining soil fertility and the soil itself. The overexploitation of both renewable and nonrenewable resources is at once well known and almost impossible to address in a system that rewards individuals for maximizing the instantaneous rate”
And at the timescales we’re talking about, they’d be taking a lower return over their entire lifetimes, to improve the situation for somebody 5-20 generations down the line using that same plot of land. And that’s so many generations it’s almost certainly not going to their descendants!
Definitely a classic coordination problem. Get a higher return “this year and every year of my life,” even though it depletes a resource 20 generations down the line? No surprise it’s an endemic problem, with those terms.
He has multiple cites of Roman erosion being pretty signifcant - erosion rates go from an inch every thousand years to an inch every hundred years.
But 1/100 of an inch every year - that’s impossible to notice! Particularly if large Roman landowers aren’t working their own fields. But erosion over time was strong enough that Ostia, once a port, stands miles from the coast now, and Ravenna likewise lost access to the sea.
So where are we?
Soil is important, it declines over centuries or thousands of years, but we don’t really know where we’re at today, because most places have been farmed for thousands of years by now.
Sounds about right. This is where I was when I finished the book. Believing it was probably a credible problem, but not knowing where this really stood in the hierarchy of problems, and on what timescales we may need to worry about it.
But that won’t do! Maybe something has happened to improve our state of knowledge in the last 17 years since the book was published (2007)?
And indeed, it has!
The go-to study on this seems to be DL Evans, et al - Soil lifespans and how they can be extended by land use and management change (2020).
In it, they aggregated 4285 measured soil erosion rates, from 240 studies covering 255 unique locations across 38 countries.
Overall, they tell us that some 90% of soils are thinning overall.
So soils are eroding 10-100x faster than they’re being replaced, in 90% of the world.
That seems pretty bad.
But on the other hand, the rate of absolute loss is quite slow. IF you assume an average soil depth of 30cm (about 11 inches, sounds conservative to me), the most we’ll lose over the next 100 years is 16% of the worst off soils, and it sounds like the remaining 83% will last for thousands of years. So, not my problem? Let some other hundreds-away generation worry about that?
I mean, if we can’t even coordinate on global warming over the last 20 years, you can’t expect us to worry about 83% of soil running out 10k+ years from now, can you?
Also, if the Roman Empire at it’s peak couldn’t do anything about this, you think WE have any chance??
This is the DL Evans money chart:
Broadly, Montgomery and DL Evans and people making this argument would like us to move from the red and blue lines to the green line, and the way to do that is with things like:
Using higher productivity crops and agricultural techniques to minimize agricultural land use, because forested land has healthy, net-increasing soil.
Plant cover crops when fields are fallow - this prevents erosion from rain and wind. “fallowing increased soil erosion rates by more than an order of magnitude compared to those during the cropping period.”
Plowing less and using minimal or “zero tillage” practices - although this increases herbicide use significantly, so not a clear net win.
Contour cultivation or terracing - farming on a slope causes much greater erosion - about 37% of soils on farmed slopes or hills were in the “under 100 years” category - and contouring or terracing greatly reduces that.
All those seem like fine ideas. But what I’m most interested in is breeding super worms.
So why has nobody thought to tackle the other end of this, the soil regeneration end?
Forget 10x developers, why can’t we breed 10-100x “soil fertility” worms?
They can either eat 10x more, or have 10-100x more descendants, or have a 10-100x faster eating + pooping life cycle, or have a certain super effective microbiome, or you can just directly measure soil volume and fertility impacts and breed for that, but there should be some way to increase the “soil regeneration and fertility” end of things.
Breeding worms should be easy:
Worms have a short reproductive cycle (populations can double in 60-90 days), and are ideal for breeding even via simple Mendelian methods.
When breeding for desirable traits, it takes only 8 generations to achieve significant results in marquee animals like cows and dogs.2
When Byalev deliberately set out to breed domesticated foxes, it only took 3 generations until they had no aggression response, within 4 generations some kits would approach humans wagging, like puppies. At the sixth generation, they essentially had the full dog behavioral package.3
Any given trait that is at least 15% heritable is a good candidate for selection via breeding - and things like metabolism and size are at least 15% heritable.4
This doesn’t even get into the possibility of breeding a bunch of worms, observing their spread in desirable traits, then GWAS-ing all of them to isolate the relevant genes and straight CRISPR-ing the desired genes into future generations, then breeding more generations - but maybe that’s overkill, when we know simple Mendelian breeding will work fine.
This is an EA style opportunity
The reason nobody has tackled this is almost certainly some variant of “there’s no market for it.” People aren’t going to pay you for 10-100x worms.
But any “worm breeding” program is going to be trivial in cost, in the low millions. Most companies and charitable organizations in the world could support it, you don’t even need to bring it up to the “government” level, but if you do that, every government can afford it too. It’s just a matter of getting in front of the right decision maker.
More, this is something that one talented individual could do *as a hobby* if they were inclined. When Darwin was studying worms, his house was famously full of jars of worms, earth, and plant matter, as he studied them to understand their behavior and life cycle and “casting” generation, or how much earth they were moving and generating.
According to DL Evans’ paper, there are immediate hot spots of “poor soil” in almost every country that could be positively impacted by 10-100x worms. Moreover, marginal soils are most likely in places where people are more likely to be subsistence farmers, who are closer to the edge and need adequate soil fertility and harvests to feed themselves and their families.
Given the significant potential impact (more and better soils will almost certainly in the aggregate feed many hungry people and save many lives), and the extremely low cost and barriers to entry, this is a tailor-made EA-style opportunity.
Montgomery D R 2007 Soil erosion and agricultural sustainability, Proc. Natl Acad. Sci.
This is the money graph, compare the white circles to the filled circles: https://imgur.com/a/Kp02LWe
Between 2000 and 2016, US dairy cattle breeders, by applying selection pressure to increase the productive life, achieved an increase of about 10 months.
After eight generations of selection, the percentage of dogs with an excellent hip quality score (as assessed by an extended view hip score) increased from 34 to 93% in German Shepherd Dogs and from 43 to 94% in Labrador retrievers.
In dogs, it's generally thought that it takes ~7-8 generations to get a new measurably distinct *breed* entirely.
Then the “domesticated elite” within the foxes went from 18% of the population at 10 generations to 80% in 35 generations, very fast fixation when selected for.
Generally on the order of 20-80% h^2 in mice and humans