In Nicolas Poussin’s 1638 painting The Arcadian Shepherds, or Et in Arcadia Ego, unsettled shepherds gather around a starkly cut stone tomb, while one of them traces with his finger the inscription; “and [yet] in Arcadia, [here too] I am.” They are accompanied by an imposing female figure, stylistically based on the Juno Cesi statue currently in the Capitoline Museum. With her diadem, blue and yellow clothing, and hand on the back of the shepherd pointing at the letter “R” on the inscription, she most likely represents the personification of Reason, offering some form of comfort or interpretation of their disruptive discovery. A laurel tree rises behind the tomb, and the shepherds are wearing laurel wreaths. The line is adapted from Virgil’s fifth Eclogue, where two shepherds lament the death of Daphnis, son of Hermes, lover of Pan, and follower of Artemis, joined in mourning by the whole of the natural world.

Daphnis was a mythical hero of pastoral culture. Raised by Sicilian shepherds on the slopes of Mount Etna, he was said to have sung the first pastoral poems. He lived in prototypical pastoral fashion: tending herds, composing songs, and dancing with Pan and Artemis. His death was brought about after he fell in love with Aphrodite herself, wasting away with the withering vegetation in the heat of summer and entrusting his pipes to Pan as he was hauled off to Hades by Eros—the general fate of any mortal lover of the goddess.

While Daphnis was Sicilian, the essential concept of the lost pastoral idyll that he represented became associated with the province of Arcadia in the central Peloponnese of Greece, hence the location of his tomb. Many of the specific features of the Arcadian ideal—rolling hills, misty sun-dappled valleys, scattered groves of oak and poplar trees—derive from actual landscapes. As a landlocked and mountainous interior region unsuitable for large-scale crop agriculture, much of Arcadia was indeed sparsely inhabited by pastoralists, said by classical urbanites to have their own distinctive mythology and to preserve ways of life long lost in teeming cities like Athens and Alexandria.

Before there was Arcadia, there was Eden and any number of myths of lost harmony with nature and the origin of alienation. Human cultural traditions seem to universally produce these stories, and our modern civilization has proved no different—from the Industrial-era English preoccupation with preserving the beatific countryside to today’s environmentalist assumptions, both popular and academic. But Arcadia contains cadavers rather than world trees. The image of a pastoral landscape as homeostatic is only possible due to a severe lack of temporal perspective. Rather than a timeless refuge, pastoral systems are one of the first results of the transformations of nature by human beings. Historically, they were its leading edge.

The foundational image of ancient Arcadia was “the hills in which herdsmen summer their flocks—in the classical world, the border between tilled land and wilderness.” These borderlands could be located along the expansion front of an agrarian society or form pockets “left behind” for reasons of climate or topography. Evan Eisenberg goes on to note that “[this] setting is especially apt in that pastoralism is in the short run the easiest and pleasantest, in the long run the most destructive of ancient ways of life.”

Ancient Arcadia itself was not spared from widespread erosion and deforestation due to relentless goat browsing pressure, which also has some responsibility for maintaining the parkland-like landscape associated with the concept. The currently somewhat improved ecological state of the region is due to ongoing state-imposed goat population reduction. This started with the Metaxas regime’s 1937 declaration of “war against the goat,” featuring compulsory slaughter of goats that entered designated forest areas and forcing shepherds to move to cities or switch to field agriculture.

Ancient authors were well aware of the damage caused by overgrazing and of the instability of pastoral systems. Throughout classical discourse, a dialectic developed between the positions that the observed deterioration of Mediterranean ecology under human exploitation was an inherent result of the general aging of the earth, as posited most notably by Seneca, or that it was caused by human mismanagement and excess. Pliny laments that “[we] blame the barrenness of the earth on its age, as if it were growing old and exhausted. But it is our own faults—our excessive greed for pasture, the trampling of flocks, and the constant burden of agriculture—that have drained the earth of its strength”.

The corollary that deterioration caused by human mismanagement should be modifiable by improved management was generally not drawn at societal scale. Classical agricultural manuals certainly give descriptions of practices which would preserve soil structure and fertility, but these are couched in terms of individual ethics to be followed by the virtuous farmer, and they address crop agriculture more than pastoralism. By the late 4th century, classical society was suffering from structural problems of sufficient severity to render the question of an organized response moot and Ambrose rails that, “The flocks consume all in their path, leaving the earth as if it were plundered. How will the earth yield her fruits when she is stripped bare by the appetite of beasts and the folly of men?” (Hexaemeron, Book 3, Chapter 16).

Given that disharmony and destruction were apparent and that ancient Mediterranean society did not generally see itself as capable of competent ecological management—because it could not conceive of itself as a system able to offer such a coordinated homeostatic response—there had to be a reason for this obvious relational rupture. Something had to have died. And here we find the body of Daphnis of the forest, the representation of this loss and alienation, who could feed his “fair flocks” in harmony with Pan and Artemis. Daphnis was tricked by Aphrodite into falling for a city-dwelling princess as part of the arc toward his demise. This is no accident.

But, as his inscription implies in its switch to direct address in the penultimate line, perhaps Daphnis is alive and can be found from the forest to the stars, if only we can see past his tomb. Though our understanding of ecology and biology today is far deeper and more granular than the classical world’s, we are strikingly similar to them in one key respect: we too fail to notice our full ability not just to negatively affect but also to heal and to steward the natural world. Since the Industrial Revolution, our release of long-trapped carbon into the atmosphere has made us, whether we like it or not, geoengineers. We should not try to ameliorate this outcome with half-measures, but rise to meet our newfound responsibility with the full breadth of our technological tools and in cooperation with nature’s own gatekeepers of carbon: plants.

The Amazon Was Created By a Lost Civilization of Gardeners

There is no intrinsically correct form of the concepts a society uses to structure its relationship between matters seen as internal to its processes and those seen as forming the external frame of the world within which it is contextualized. Concepts such as Daphnis dead in Arcadia, or the formation and separation of ideas of the “human” and “natural” worlds, come about to explain observed patterns, events, and causes.

The entirety of any human societal processes is as fully dependent energetically on plant photosynthetic production as all other parts of the heterotrophic biosphere. As such, accounts of separate and uncontrollable natural forces—and of changes in the complexity of the internal processes and productions of a human society—are part of the formation of a societal self-concept, rather than accurate descriptions of energy and influence. Much can be learned by examining which elements of the world a society decides to define as outside of its domain of responsibility and the ways that this patterns its engagement with reality.

In 1998, Betty Meggers, one of the leading American archaeologists studying Amazonia, wrote that terra preta, the dark, fertile soil that is curiously found in the Amazon, “is unlikely to have been cultivated indigenously because habitation sites are also burial grounds.” Meggers wrote within an intellectual tradition that was incapable of understanding the evidence before its eyes of large-scale human engagement with the biosphere outside of the familiar frames of field agriculture, pastoralism, and urbanism.

In Amazonia, humans created and lived within a managed forest over thousands of years and miles, in a manner invisible to Western lenses. Additionally, Meggers was predominantly responsible for propagating the standard view held by Western anthropologists and archeologists in the mid-twentieth century: that indigenous Amazonians were, at best, agriculturalists of opportunity, scraping a meager harvest from nutrient-depleted tropical soils, which are famous for their lack of fertility. Her 1971 book Amazonia: Man and Culture in a Counterfeit Paradise, introduces the trope that seemingly paradisical Amazonia was a place where nature imposed harsh and intrinsic limits on human habitation.

This view became untenable after the demonstration in the last decades that legends of lost garden cities of Amazonia were in large part historically accurate. This was a result of many avenues of evidence, most compellingly in the last few years with the large-scale use of lidar to uncover remnant structures beneath the forest canopy. A 2022 study in southwestern Amazonia found that “large settlement sites are surrounded by ranked concentric polygonal banks and represent central nodes that are connected to lower-ranked sites by straight, raised causeways that stretch over several kilometres. Massive water-management infrastructure, composed of canals and reservoirs, complete the settlement system in an anthropogenically modified landscape.”

Amazonia was the domestication site for many of our currently dominant cultivated plants, including cacao, peppers, pineapple, sweet potato, and tobacco. Out of the roughly 16,000 known Amazonian tree species, 50% of the total forest cover is composed of only 227 species, just 1.4% of the total diversity—yet these species are overwhelmingly those useful to humans or species that modify their local ecology in a direction favorable to humans. The scale of human management suggests a pre-Columbian population of around 10 million. Previous estimates under the Meggers hypothesis ranging as low as 500,000 are now discredited. But the fundamental issue remains that freshly cleared rainforest soil is extremely infertile. In a high-energy, high-diversity system such as a tropical forest, the vast majority of available nutrients spend almost all their time rapidly exchanging between the bodies of living organisms, which are specialized to quickly absorb decomposition products, leaving little to accumulate as a soil reservoir. So how were all these people fed?

In much of Amazonia, the lower populations in recent centuries allowed a form of shifting cultivation that is often termed “slash and burn” or swidden agriculture. A plot is cleared of forest, and the felled vegetation is burned on the site, forming a surface layer of soil enriched with nutrients in which crops can be grown for a few years before fertility declines. After this, humans move on to a new plot, allowing the first to recover through natural regenerative succession. Assuming no banking of nutrients, this method does impose limits on population density according to the recovery time, and is only sustainable at low density.

In the Brazilian Amazon, freshly-cleared forest soil in an area without a history of human cultivation is known as terra comum, common soil. Repeated cycles of swidden agriculture result in an accumulation of carbon and other combustion products in the soil, somewhat improving its fertility; this soil is then known as terra mulata, brown earth. The most fertile soil, terra preta, is enriched in carbon to a greater degree than terra mulata, also containing a mixture of compost products, animal bones and feces, pottery shards, and other evidence of human modification. This is the soil that Meggers mistook as residue from burial grounds and middens. But these soils cover between three to ten percent of the total forest area—far more than could be accounted for through funerary practices or as the unintentional remnant of a low-density population. In Bolivia, the majority of the forested area surrounding watercourses in the llanos is likely anthropogenic, as nearly all of it is underlain by highly human-modified terra preta.

Carbon as a soil additive has several unique properties; most importantly, it “adsords” or adheres charged particles onto its charged surface, and is porous with a vast surface area when the feedstocks producing it are burned in the right way. These advantages accrue to the current use of biochar in soil, in addition to it being a fixed and extremely stable form of carbon with negligible atmospheric return.

Amazonia, then, is an example of a highly human-modified system—in no sense an untouched wilderness or counterfeit paradise. This was invisible to Western archeologists because the boundaries they expected to see were not present, and the cultivated species were intermixed with the wild ones. Eduardo Neves, a Brazilian archeologist at the forefront of the reevaluation of the history of Amazonia, notes that “[the] Indigenous worldview does not differentiate between the domain of culture and the domain of nature. The diversity of the Amazon, the presence of many large nut trees and fruit-bearing palm trees, is a result of Indigenous practices.”

These indigenous practices also increased the amount of carbon held in the land as charcoal many times over what it would have been without human management. What appeared as wilderness is in reality a garden forest shaped to human design, and certainly not an Arcadian edge. Deforestation for monocultures of soy or cattle pasture is destroying an interwoven artifact of human action as part of the biosphere—not a pristine system alienated from human influence—and replacing it with a simpler, energetically less favorable one.

Once we can see the relational system connecting the structure of the forest system to the human activities that modified it, conceiving of a system like Amazonia as separate from its human inhabitants becomes erroneous. Here, cultural blindness and essentialization lead to a failure to recognize the evidence before our eyes of long-term, intentional human management. Similarly, our understanding of phenomena like anthropogenic climate destabilization is bounded by the extent of our contextual knowledge, since the Earth system as a whole is also not energetically separable from its human inhabitants.

Arcadian Thinking Limits Our Responses to Climate Change

Just as the human influence on Amazonia flew under the radar of Western agriculturalists and anthropologists, so too often does the energetic reality of the climate crisis escape adequate understanding when viewed through eyes determined to separate the cultivated from the wild. All of the geoengineering methods that have the potential to enable homeostatic regulation of the climate system lie across this boundary because they all involve targeted interventions to change the parameters of planetary-scale processes. In this way, a small energetic input can have an exponentially greater effect or, in other words, an intended consequence.

Ethics that place these methods a priori out of bounds will hobble our ability to respond as a species to the problem our own success has presented to us, because systemic changes are appraised indistinguishably from highly negative unintended consequences. Often, in a specific instance, the problem space is comparatively limited. For example, Populus tremula x alba 717, or “Deer” for short, is a well-known variety of hybrid aspen that exhibits a more rapid carbon assimilation and, therefore, growth rate by a novel and evolutionarily innovative mechanism.

Yes, this would result in forests in which this was the dominant species of Populus showing a faster overall growth rate. Therefore, the forests would not exactly match the properties of a preindustrial forest. Still, we must be free to question why improving the carbon removal capacity of the terrestrial biosphere should be equated with returning it to whatever form it was in at whatever time slice our society derives its Arcadian ideal from—or to some notion of a wild uncontaminated by human influence.

Every open system drawdown approach suffers from its own version of this image problem. In the case of oceanic iron fertilization, an equation with “dumping” and assumption of the inevitability of toxic algal blooms and some combination of zero effect and systemic breakdown has limited experimentation for decades. Too often, the methods of carbon removal companies are dictated by the idiosyncratic preferences of carbon markets built to reward comparatively minuscule scale removals by closed system approaches such as direct air capture. This method is widely known not to be scalable in a relevant time frame to remove carbon at anywhere near the rate we need, but it has the advantage of producing a product (compressed carbon dioxide gas) usable by the fossil fuel industry at rates sufficient to offset the emissions of software companies.

Concepts are imported which make sense in a business context but which are entirely inadequate to the task at hand, that is, altering the control parameters of a continuous planetary flux process. A focus on permanence and traceability of specific removals to specific customers only makes sense if equal weight is given to scalability and the percentage of energy input under direct human management. Even some open system approaches have become elaborately baroque in their attempts to satisfy the demands of myopic and largely image-driven carbon credit issuance registries.

Behind all of this lies the simultaneous anxiety and reassurance of the availability of solar radiation management as a comparatively low-cost solution, of a sort, should the situation continue to deteriorate without sufficient near-term carbon removal. But solar radiation management solves the temperature increase problem in the near term while making other problems worse. It will not prevent ocean acidification and it will reduce incident solar radiation in photosynthetic wavelengths, while also removing the temperature rise control signal for carbon accumulation. At high atmospheric carbon dioxide levels, the therapeutic index of solar radiation management is narrow—that is, the required annual aerosol dose falls between narrow bounds. In particular, the consequence of underdosing in a given year would be wide swings in average global surface temperature, since the underlying climatic forcing is counterbalanced by the dimming sulfur dioxide.

Humans have the potential to be the first species in the history of the biosphere to enable homeostatic regulation of the carbon dioxide concentration in the atmosphere—if we abandon our desire to alienate nature from human influence. This requires recognizing that the alienation is a consequence of our countervailing desire to act without consideration for the presence of living ecological systems throughout even the most human-dominated environments, and to ignore our dominant influence on planetary energy flux.

The only long-term solution to our spiraling negative impacts on the climate system and the functional connectivity of the global ecosystem is to work with the rest of the biosphere to create ecologies which assume human influence as a given factor. This is how we—the only species yet produced by the Earth system to understand our own origins—can open our eyes to the threads connecting us to the whole. In doing so, we can awaken from our Arcadian dreams to create new forests and regenerate waterways in partnership with the species surrounding us.

The only ethic that should apply here is that of systemic functional improvement in retention of energy and communication between living systems. This requires that we work with boldness in cooperation and restraint from alienation and a priori categorization. These attitudes have been in eclipse recently in many different contexts, but the rewards are great, no less than bringing Gaia to life through discovering and applying control inputs to planetary carbon metabolism. We would be functioning as the eyes of the biosphere seeing itself for the first time.

Yes, we must decarbonize our energy production, but this will not serve to remove what has accumulated over the last several centuries or to reduce our global ecological dominance as energy-intensive heterotrophs. We need to go talk with the autotrophs—the plants—and empower them to remove the carbon we have emitted, for both our sakes, but especially for our own. We must ask them to increase their rate of drawdown, and reduce the rate of return of the carbon they fix.

They will not do this without being asked. Increased atmospheric carbon dioxide deselects plants for efficiency of carbon removal; while it does stimulate growth rate, it makes metabolically expensive carbon capture optimizations less likely to evolve and spread. Increased temperature accelerates the rate of respiration-mediated decay of biomass and liberation of fixed carbon into the atmosphere so, likewise, the autotrophs will not pull this lever for us. Thankfully, most of the words are already written, and we can cut and paste most of our messages together. The human discipline devoted to facilitating genetic communication between organisms is called biotechnology, and if harmony is found in communication, it is here where we will discover Daphnis alive.

Conceptually Confused Regulation is Blocking Biotechnology Progress

The African iroko tree is oxalogenic: the tree secretes oxalic acid from its roots. Over time, this oxalic exudate accumulates and decomposes to calcium carbonate, or limestone, with the accompanying liberation of half its carbon as carbon dioxide. Therefore, the soils surrounding the roots of an oxalogenic tree become enriched in fixed carbon at a ratio of half the carbon in the oxalate secreted. Oxalogenic trees are rare. Iroko is one of the only well-known examples, and it has a restricted distribution and a slow growth rate; it takes over a century of growth for the stem of an iroko tree to approach a meter in diameter.

Genetic biotechnology could enable the elucidation of the oxalate secretion pathway in iroko and the introduction of this pathway into faster-growing tree species with broader climatic tolerance bands. The introduction of these trees into areas where they effectively integrate into local ecologies—likely those that match the ancestral range of the recipient species chosen—would enable an increase in the rate of carbon removal in the forests of the regions in which they were planted. Because these trees would still perform their original role, including reproducing on their own, as well as removing more carbon from the atmosphere, this would enable a significant shift in global carbon flux toward drawdown. No method like this will be sufficient alone, but integrating genetic biotechnology with open system drawdown holds significant potential to meaningfully contribute to carbon removal.

But unfortunately, under our current regulatory framework, such methods are almost impossible to test, even on a small scale. Creation of a fast-growing, oxalogenic tree could proceed within the walled garden of academia, but widespread deployment even of a method with demonstrable promise is almost impossible. The Salk Institute, an organization named after the creator of the first polio vaccine, and so—one would think—committed to significantly impacting global human problems, has spent close to two decades researching a genetic biotechnology innovation that increases soil carbon storage by upregulating suberin (cork) production by the roots of grasses.

Field trial after field trial has been run, success has been demonstrated, but no deployment has occurred. The institute has a division dedicated to this work, the Harnessing Plants Initiative, who narrate that they have “made significant progress in the lab” and have registered the trademark Ideal Plants. All this seeming progress obscures the fact that there is almost no viable regulatory route to widespread adoption. The Salk Ideal Plants only differ in specific respects from other varieties; in the case of the Salk plants, the root system is larger, and the cork wall of each root is thicker. These differences, while specific to unusual anatomical regions, are no larger in scale than those between many horticultural varieties. They are certainly less than those between, say, domesticated maize and its wild progenitor teosinte. However, the status of biotechnology regulation is such that how a thing is done carries far more weight than what it is.

It is no accident that most genetically modified organisms with legal approval for unrestricted use are bulk foods. Only agricultural-chemical conglomerates have historically had the financial and legal resources to shepherd an innovation through the arduous approval process, which can easily take over a decade, while organizations such as the Salk Institute provide demonstrations but do not assume the responsibility for deployment. Meanwhile, genetic modification of bacteria and fungi to produce drugs and foods proceeds apace and arouses little protest, and the idea that the leakage of these organisms into the environment is an existential risk is hardly acknowledged.

Beyond the specific regulatory issue is market-driven adoption. Concern for mitigating anthropogenic carbon emissions has resulted in the issuance of carbon credits and the development of systems of carbon accounting, mostly on a voluntary basis but with increasing governmental support globally. These credits are associated with determinants of quality which stipulate conditions beyond removal of carbon. Because participation in carbon markets and the purchase of carbon credits are generally voluntary acts associated with the environmental movement, the determinants of stipulative quality of a given carbon removal tonnage include many conditions beyond verification of the removal itself. The two dominant carbon credit registries in the United States both include a stipulation that the use of genetically modified organisms (GMOs) for removal be prohibited for credits issued.

In the baseline public consciousness, despite a little recent progress, “GMOs” are still often considered a simple bad, and their avoidance is an obvious necessity for the ecologically literate, especially GMOs placed in natural environments such as forests. So biotechnological methods, such as that of the Salk Institute, that have shown efficacy but would only ever work if a significant portion of photosynthetic carbon, either in agricultural or wildland systems, were fixed by their plants, are unlikely to be viable without changes in public understanding. The idea of GMOs as a special danger, and the intrinsic stigma it generates, compounds on top of the general difficulty of open system approaches in gaining acceptance.

The idea that organisms produced by genetic biotechnology represent the possibility of catastrophic contamination of a previously pure biosphere has caused a strange Arcadian fixation, seen both in perception and regulation, on ranking the techniques used in their production on a scale of how much impurity they introduce. The rest of the biosphere, however, continues to communicate for its own purposes, neglecting to respect any such rules.

There are currently three dominant methods of introducing novel genetic script to a eukaryotic organism’s genome: particle bombardment (adhering DNA sequences to physical carriers such as gold nanoparticles to provide mass to shoot directly into a cell), transformation using a living mediator (using the pre-existing ability of organisms to integrate sequences into a host genome), and a class of enzyme-mediated methods such as CRISPR often collectively termed “gene editing.” These have all been developed since the 1980s.

The division of the U.S. Department of Agriculture (USDA) responsible for biotechnology regulation is known as Animal and Plant Health Inspection Service (APHIS). In 1987, APHIS established regulations for “the safe movement and importation of genetically engineered organisms,” simultaneously with its publication of regulations aimed to control species that spread disease. These regulations specified that Agrobacterium, the major organism used in plant genetic transformation, was a plant pest and regulated any genetic sequence identifiable as deriving from Agrobacterium as being a “plant pest sequence.” APHIS went on to establish severe restrictions on the import, export, trade, and movement of all organisms containing “plant pest sequences.” Because genetically-transformed organisms can be identified by the presence of flanking sequences derived from the Agrobacterium vector used in their transformation, any product of Agrobacterium transformation was regulated in this way.

This introduced a fixation on “genetic purity” into the legal structure dealing with the entire nascent field of plant biotechnology. Other methods of transformation, such as particle bombardment, were not regulated in any way by the USDA, as long as they too did not involve the introduction of “plant pest sequences.” Although the intent was to regulate the products of biotechnology in themselves, the letter of the law portrayed the introduction of minute genetic sequences from disease-causing organisms used in transformation as somehow capable of introducing the contaminating essence of the pest into the organism hosting them. The regulatory superstructure which developed, therefore, could be imported wholesale from existing rule systems serving to reduce the risk of the spread of infectious diseases, but with a whole class of organisms transformed via non-vector mediated methods being excluded.

There are very few examples of ecological problems caused by genetically modified plants. Those that have happened underline the importance of separating our evaluations of technique and application. To much of society, genetically modified plants as a class are synonymous with the specific example of glyphosate-resistant crops, with the added conceptual confusion of the actual use—enabling spraying of glyphosate to kill weeds in the crop without harming it—being conflated with nebulous ideas including the production of glyphosate by the plants themselves, or simply an unknown and therefore uncanny enhancement that could escape and dominate.

The problems caused by these plants are part of the general problem inherent in managing large areas of land in a manner that progressively depletes the soil biome, reduces mineral nutrients, and prevents carbon sequestration. That is, they are the problems of industrial agriculture itself. Nonetheless, the historical domination of the regulatory system by the concerns of the purveyors of these plants has led to patterns of decisions by the USDA illustrating the incoherence of its legislative approach.

In 2003, a creeping bentgrass variety (Agrostis stolonifera), transformed by Monsanto to be resistant to glyphosate—for intended use on golf courses—escaped from a field trial in Oregon. The grass has since spread widely through multiple habitats across the state, and will likely continue to expand. In 2016, APHIS approved glyphosate (brand name Roundup) resistant creeping bentgrass, claiming that it did not have enhanced weediness potential, long after it had escaped from cultivation. Creeping bentgrass is a Eurasian species brought to North America during European colonization and had already shown itself capable of establishing itself in place of native grasses in disturbed areas, with a history of “noxious weed” classification.

The intended use case for the grass was to improve the ease of weed control with Roundup on golf courses. APHIS noted that “[a] primary herbicide used to eliminate [creeping bentgrass] where it is not desired is glyphosate,” but that alternate herbicides were available. Glyphosate is indeed the dominant herbicide used by ecological restorationists seeking to remove invasive plants such as bentgrass, so it seems bizarre that the USDA could claim that a bentgrass variety specifically engineered to be resistant to this herbicide did not have increased weediness potential.

Noting the incoherence, the fact remains that Roundup-resistant Agrostis remains identical to conventional Agrostis in all but one troublesome aspect, and a minor threat at best. The voluminous field trial data and long timeline required for the approval stem from the intensive investigation of properties of the plant, triggered by the presence of “plant pest sequences” in the construct, although the grass was transformed via biolistics. These included “agronomic, compositional, disease and insect evaluations” for “65 field releases in 20 states and 40 counties performed between 1999 and 2002.” Unsurprisingly, none of these properties differed from conventional bentgrass. Glyphosate-resistant creeping bentgrass spreading from highly managed and intensively irrigated golf courses is hardly a poster child for integrative biotechnology, but is more annoyance than danger.

The vast expense poured into gathering evidence to support the approval of such a marginal candidate, which was already causing problems in the wild when the decision was made, while organizations like the Salk Institute write letters to congress petitioning for funding to run the same exhaustive field trials required for approval of their plants, makes clear some of the processes retarding the balanced development of plant biotechnology. The Scotts-Miracle Gro company had already abandoned commercialization of the grass when approval was granted, and the only mention on their website now is a brief historical account and a phone number to report the grass for eradication.

After over thirty years, APHIS updated its regulations in 2020 in a piece of legislation termed the SECURE (Sustainable, Ecological, Consistent, Uniform, Responsible, Efficient) rule, replacing the determinant of regulability from “plant pest sequences” to whether a given transformation “could have happened in nature.” The implication now was that “unnatural transformations” were more dangerous and thus more regulatable than those that could have occurred without human intervention.

Because living organisms exist in a context where high energy particles occasionally collide with the nucleotides in their genomes, and because the replication of these chains of nucleotides is not absolutely perfect every time, changes in individual nucleotides, including deletions, additions, and replacement of one of the four by another, are common events. These are known as point mutations. Also common are transpositions of sections of the chains, changing the order of the sequence within an organism.

The framers of the SECURE rule classified these as natural transformations and applied minimal regulation to human methods that mimic them, known collectively as gene editing. Regulation increases as more nucleotides are changed, the reasoning being that every additional nucleotide transformed takes us further from the state of nature. Regulation is most stringent in instances where sequences from one species are introduced into another, being that this is seen as the least likely scenario to occur in nature. The organisms created by these methods are termed transgenic, as opposed to the cisgenic products of gene editing.

The recent advent of CRISPR and other methods allowing targeted single nucleotide changes at specific points in the genome has made anthropogenic point mutation relatively simple, and a gene can often be rendered inactive, more active, or have the properties of its protein product modified, by one or only a few nucleotide changes. The text of the regulation allows the stacking of repeated single nucleotide changes so long as each stage of the project is given individual approval. But gene editing is not capable of introducing entire new metabolic pathways to an organism. These need genetic constructs enabling the production of novel proteins, either in code written by humans or taken from a donor organism, creating a transgenic synthesis.

Gene editing is not, as usually portrayed, simply a technological advance in bioengineering that facilitates easier and safer application; it is rather a specific method that allows targeted genetic changes to small numbers of nucleotides. CRISPR will therefore never replace transgenic transformation, but rather is a complementary method. Even in instances where metabolic transformations could be achieved by specific patterns of gene editing, this would be slower and clumsier than simply introducing a construct.

This is not to imply that the results of gene editing are generally less major in their effect on the organism, only that it is incomplete as a transformation toolkit. Taking a step back, we can see that what APHIS did with the SECURE ruling was to replace one criterion of purity with another. The plant biotechnology industry reoriented itself around the new measure and immediately began producing literature that emphasized the safety and technological advancement of CRISPR, positioning it as a replacement for the supposedly clumsier and more dangerous older transgenic methods. Unfortunately for the framers, there is very little that does not happen in nature, and naturally transgenic organisms are relatively common. Like other methods before, gene editing is now often used for specifically regulatory reasons, not because it is better and safer, but because it does not produce a GMO as defined by APHIS.

In December 2024, a Northern California district court vacated the SECURE rule in its entirety, granting that both the gene editing exemption and a failure to incorporate noxious weed authority were “arbitrary and capricious” decisions by APHIS. The judge argued that there was no basis to assume all conventionally-bred plants were risk-free, and therefore that it did not make sense to use them as a baseline to determine the scope of regulatory oversight. With this decision, APHIS has currently reverted to the pre-SECURE rule regulatory structure, although this situation is widely considered unsatisfactory, and further revision is expected. Given that it took fifteen years to update the original regulation, significant delay is likely.

The products of plant genetic engineering do not present a unique risk. Instances of them becoming noxious weeds are possible but rare, and if this happens, these weeds can and should be managed. But a regulatory mindset deriving from ideals of purity and contamination has greatly inhibited the development of a field of knowledge and practice that represents unprecedented biospheric integration and is one of our best hopes for near-term carbon drawdown. Many find this a scary proposal, but it already happens in nature all the time without our knowledge or involvement.

Nature Recognizes No Difference Between Its and Our Interventions

The black cottonwood (Populus trichocarpa) is a deciduous poplar tree common to the Pacific Northwest. A variant of it contains the composite gene BSTR, composed of three sections pasted together: the first codes for a portion of the enzyme glycosyltransferase from the bacterium Streptomyces, which lives symbiotically within poplar trees as an endophyte; the second, for part of a DNA-binding protein from an ant, Trachymyrmex septentrionalis. This specific ant is a farmer, cultivating a fungus in subterranean galleries, which it feeds with poplar wood and leaf litter.

The third portion of BSTR is a significant chunk of the coding sequence for RuBisCo, the carbon-fixing protein in photosynthesis and the most abundant enzyme on Earth, deriving from the cottonwood’s own chloroplast. These three sections, when joined together in this specific pattern, form a gene whose protein product enables the photosynthetic apparatus of the tree to adapt more rapidly to changes in light intensity, leading to significant gains in its carbon fixation rate.

BSTR black cottonwood is a compelling example of the deep contingency and synthetic power of nature. That two enzyme fragments, from a prokaryote and an animal, can combine with a portion of chloroplastic RuBisCo to form a composite gene integrated into the plant’s nuclear genome is already bizarre. That such a gene could be transcribed is improbable. That its transcript could be translated to a functional protein stretches credulity. That the function of this protein should be to improve the efficiency of photosynthesis itself is outside of any interpretation in probabilistic terms. With this being said, the organism is also obviously highly transgenic and born full-leafed from nature’s face. What can be interpreted probabilistically is the time frame when the event occurred.

The first wild tree discovered is almost certainly not the mother of the whole line. Black cottonwood, like most poplars, is proficient in asexual reproduction. Adventitious roots sprout easily from stem cuttings, the root system is laden with dormant buds, and regrowth from stumps and branches is rapid. Given this ability, the frequency of the tree in Pacific Northwest forests, and the comparatively recent development of human survey and sequencing techniques, the transformation event is likely to have occurred several millennia ago, assuming that the enhanced growth rate increases the likelihood of encountering a BSTR cottonwood over time. We also do not know whether the tree represents an ongoing replacement of unboosted cottonwoods, or if there are physiological costs we are as yet unaware of which select against BSTR in some environments, limiting its spread.

The question of how the event occurred must, at this point, be answered speculatively, but there are clues. A scenario involving shed cells in the salivary glands of ants harvesting living poplar stems or roots hosting Streptomyces comes to mind. Then the release of fragmented sequences during concurrent lysis of bacterial and ant cells, mixed with damaged chloroplasts, and all this happening while the genome of the tree was being actively read, perhaps during an immune response to these insect predators. Most improbable, but at least possible if repeated billions of times over millions of years, which the ongoing use of cottonwood as an ant agricultural feedstock certainly provides.

Human transformation techniques such as particle bombardment, with no method to ensure genomic integration other than physically shooting the construct into the nucleus, similarly rely on this “law of large numbers,” but even biolistics is likely far more efficient than whatever happened several thousand years ago somewhere near what is now Seattle.

There is no essential difference between the creation of a genetically engineered plant and a naturally occurring event such as BSTR cottonwood. While acting as agents of change should correctly lead us to a sense of responsibility for the consequences of such change, there should be no wholesale abandonment of what, from my perspective, is our duty. As the bearers of intentionality and concentration of energy flux in service of this intentionality, we must integrate living systems in directions that move the system as a whole toward homeostasis.

While wholesale abandonment may seem a strong term, it accurately names the central human tendency to avoid direct perception of sources of shame. If we consider shame to be an essentializing abandonment of our self-perception, avoiding perceptions that cause this becomes a survival strategy. Thinking about a constructed human world abutting a given natural world, without considering their inevitable intermingling, allows us to cognitively and emotionally withdraw from our contextual reality. This framing also makes it easier to externalize costs, and neglect to register energetic dependencies, while causing action paralysis even in cases of clear necessity.

This shame also bears significant responsibility for a lacuna in ecological science regarding the study of directed human improvement of ecosystemic function. Practices such as permaculture and agroforestry give hints of such a science of applied ecology, but most ecological theory still categorizes human activity as a disturbance outside the system, rather than as an inevitable component of any modern ecology that could be actively beneficial to both us and the system as a whole when managed correctly.

While properties such as biodiversity are correctly seen as important to protect and restore, there is a surprising lack of theoretical undergirding to give quantitative descriptions of their benefits as general properties or with predictive power regarding the effects of changes. We need general measures of ecological functionality and responses to degradation that restore functionality. However, in order to derive these, we must release the common assumption that restoration of function is synonymous with the restoration of some prior Arcadian state of affairs.

In the world as we find it, the boundaries of entire climatic zones are shifting, and the effects of humans are ubiquitous. Human assistance could enable many organisms to persist through these centuries, leveraging our intentionality in service of accelerating the evolutionary process. For example, increasing the thickness of the bark of forest trees would increase carbon storage while reducing mortality from wildfires. Enhancing the downy layer on the underside of many leaves would reduce the transpiration rate and improve drought tolerance.

As a sketch of what such a theoretical and regulatory environment would look like for synthetic biology, imagine a world in which we had derived stable metrics of ecological functionality using standardized data collection and measures of energetic flux. In this world, genetically modified plants would be regulated on the level of practices rather than around speculative catastrophic contamination risks. Regulatory approval could be necessary for increments of area planted, or for commercial use or use above certain production thresholds—use that would have an energetic impact. Noxious weed regulation could remain intact, and apply regardless of the construct inserted, at the level of the species worked with.

Unlike many pressing issues of our time, this is one where the financial and regulatory obstacles are identifiable and clear, and scientific experimentation and iteration can be conducted immediately and rapidly. In fact, I have been doing so for the last five years: my public benefit corporation successfully bred and planted hybrid poplar trees that assimilated carbon from the atmosphere thirty to fifty percent faster than normal, work for which we were profiled in The New York Times. I am continuing to work with these trees, answering basic research questions about their behavior and carbon removal rate in various contexts, as one branch of the nonprofit organization Carbocene Industries. For those who wish to help rebalance planetary carbon flux and become technological gardeners of our Earth, I invite you to join me. The whole Earth is already our garden; it is high time we recognize this and treat it so.

Assuming “business as usual,” CO₂ concentration is projected to be 600 to 800 ppm by the end of the century. Optimistic projections not involving sustained negative emissions starting at mid-century give an average of ~2°C warming by 2100. Even that is considered by authorities to be a problem, but it’s less than the paleoclimatic analog.

It’s usually assumed that if we reduce emissions, then CO₂ will steadily decline on its own due to natural sequestration processes. This is based on the notion of a set residence time in the atmosphere—essentially a guaranteed life cycle which these kinds of predictions can be based on—as if the pre-industrial climate was some kind of strong equilibrium to which excursions will quickly return.

But that’s not really how it works. In fact, the assumption that our pre-industrial climate was a robust equilibrium is an existentially dangerous one. If there’s more carbon around, the biosphere will just operate with more carbon in the cycle, including in the air, until some process causes permanent sequestration. To understand this, as well as the unusual and metastable nature of our current climate, we have to go back to the geological history of climate change.

The Current Climate Is An Icy Anomaly

Prior to 20 million years ago, both carbon dioxide concentration and temperature were significantly above current levels, and had remained there for most of earth’s history.

Even during the much of the cooling trend of the last 20 million years, the Arctic Ocean was ice free in the summer, and the northern margins of the surrounding continents, where today there is barren tundra, were clothed in temperate forest. The Antarctic Ice Sheet covered only a small area around the pole, the rest of the continent supporting mixed forests of conifers and southern beeches. The coastlines of the continents were many miles further inland than today.

In contrast to the usual apocalyptic narratives, the tropics were not scorched wastelands. They were covered with luxuriant rainforests containing up to 100 species of apes, including our own ancestors. Temperatures were up to 10 ℃ higher than today at the poles, but the tropics were not much hotter. Recent observations are showing that warming is concentrated in high latitudes, causing mostly reduced seasonal variations. 

Prior to the last 20 million years, the climate was mostly warm and wet, with much reduced seasonality.

The last time the climate was anything like today—in other words, the last time there was an ice age, the current one being the Quaternary glaciation—was the late Carboniferous period. That was 300 million years ago, before the dinosaurs, with an ice age lasting approximately 60 million years. Prior to that, the only other ice age since the advent of multicellular life was the Ordovician glaciation, about 450 million years ago, at which time the atmosphere had not yet reached its modern composition.

The key point to understand is that the recent low carbon, low temperature condition is a significant anomaly from the norm. We are in an interglacial period—with reduced but still significant glaciation—of an ice age which has lasted for the past 2.6 million years. Ice ages are rare: the Earth has had significant polar ice caps for at most 20% of the last 540 million years.

The remaining 80%, including the entire Mesozoic era, saw what are often described as hothouse or greenhouse conditions. The greenhouse state is the default equilibrium. Glacial or icehouse periods, on the other hand, are historically not stable. They are comparatively short term events occurring because of particular contingent conditions.

Most of Earth’s history has been warm, wet, and high-carbon. Ice ages like the current one are rare, few, and short. The recent climate has not been a long-term equilibrium, but a short-term metastable condition. This is key background for any discussion of global warming, and any possible response.

Prehistoric Volcanic Warming

Though there are other factors, historical large changes in the climate equilibrium are closely correlated with addition and removal of CO₂ from the atmosphere. The addition is usually due to massive and sustained volcanic activity on a scale unseen for tens of millions of years. For example, large areas of Siberia and Northern India are covered with the remains of lava flows extending hundreds of miles, from eruptions which continued for tens of millenia, known as flood basalts or traps.

The Siberian Traps were formed simultaneously with the Permian-Triassic mass extinction, the worst in history, where 95% of marine and 70% of terrestrial species were wiped out. Coming close on the heels of this extinction event, the Wrangellia eruptions in the Carnian stage of the early Triassic caused a further mass injection of CO₂ into the atmosphere, and another episode of sudden warming. This increased evaporation from the oceans to such an extent that extreme precipitation caused almost continuous flooding across the entire planet. Current evidence suggests that average global rainfall reached the levels now seen in temperate rainforests on the Pacific Northwest, for a period of 2 million years. Known as the Carnian Pluvial Event, this caused another large scale turnover in the biosphere, instrumental in the evolution of the dinosaurs.

The Indian flows, or Deccan Traps, happened at the same time as the extinction of the dinosaurs, and were possibly caused or intensified by the asteroid impact that caused the extinction. This caused a warming period leading to the Paleocene-Eocene Thermal Maximum, an event much studied as a possible analogue of current climate change.

The volcanic hotspot underlying Yellowstone National Park is a possible source of a future flood basalt event.

Historically, these eruptions seem to cause an initial global cooling, caused by the release of sun-blocking aerosols like SO₂ into the upper atmosphere, and then a large degree of warming caused by CO₂ as the Sulfur dissipates. The associated mass extinctions are likely exacerbated by this climate whiplash. They also prolong the effects: the initial event and dimmed cooling period reduces biological productivity, reducing the ability of the biosphere to sequester carbon. This exacerbates ocean acidification as CO₂ dissolves into the water. Ocean acidification, combined with the sudden reheating, in turn keeps biological productivity much reduced and slows recovery. It can then take millions of years for the biosphere to return the CO₂ level to an equilibrium.

In contrast, mass extraction of CO₂ from the atmosphere is associated with events where unusual factors enable the removal of large quantities of plant material from circulation before the biomass either decomposes or is consumed by animals (which would release the CO₂ absorbed by plant life back into the atmosphere). One example of this would be mass burial. Once icehouse conditions are established—that is, ice age conditions with relatively lower temperatures, such as the present day—they are maintained until the sequestration mechanism ceases to operate.

In the early history of the Earth, this happened multiple times, and these ice ages are termed Snowball Earth events. The earliest one was likely caused by cyanobacteria, the first widely successful photosynthetic organism. The advent of photosynthesis both removed CO₂ directly, and removed methane, another potent greenhouse gas, by oxidation. The transition to an oxidizing atmospheric chemistry caused a mass extinction, as other life wasn’t adapted to a harsh oxygen atmosphere. As a result, the cyanobacteria could draw down CO₂ until nearly the entire surface of the planet was covered in ice. These events may have ended by the injection of CO₂ through eruption of flood basalts through the ice cover.

The important point to take away from this is that massive CO₂ addition to the atmosphere has historically been caused by volcanic eruptions, and slow carbon drawdown has occurred through the response of the biosphere to this addition, returning conditions to a hothouse equilibrium. Large-scale CO₂ removal is caused by distinctive and contingent biological events, rather than by any kind of ongoing and stable geophysical process, and results in a metastable icehouse state.

Our current addition of CO₂ differs from this pattern by being caused by industrial activity, and occurring on a far shorter timescale and higher rate than is typical, although with a lower potential total addition.

Contextualizing Anthropogenic Warming

For comparison to our own activity, the most intense phase of the Deccan Traps eruptions, 66 million years ago, released several tens of thousands of gigatons of CO₂ into the atmosphere, over a period of about 30,000 years. Since 1750, we have emitted about 1500 gigatons, half of this in the last 40 years. We have roughly 3000 gigatons of potential emissions from easily accessible fossil fuels remaining in the ground, although to release this we would have to extract and burn all the world’s coal, which is unlikely. So, even assuming these figures to be underestimates, and factoring in CO₂ equivalents in methane and other warming gases, total realized and potential anthropogenic emissions are unlikely to exceed 4000 gigatons.

This is dwarfed by the effect of the largest Deccan eruptions. Realistically, we will only ever emit about a tenth of the CO₂ released in a flood basalt volcanic event; we will probably not use all available coal reserves as the carbon intensity of economic growth is either flat or on a slowly declining trend. The major difference is that our emissions are occurring over a period of centuries, rather than tens of millennia. Even the most extreme natural events would release only on the order of 1 gigaton per year, whereas in 2017, global CO₂ emissions were about 41 gigatons.

So our own maximum total impact is well within the limits of the largest natural events that have occurred in prehistory, but is happening much faster, and is therefore significant. Judging by these prehistorical analogues, by default we can expect a very large impact: a mass extinction, significant loss of coastal land, and large changes to global weather patterns.

Most importantly, the unprecedented rate of our emissions may cause significant instability in the climate, which is already precariously balanced in an unusual ice age. Such instability has been seen recently in the transition from the last glacial maximum to the current interglacial period, though this was not directly caused by CO₂ emissions.

This transition was not gradual or stable. Temperatures rose and fell multiple degrees within decades. In the Younger Dryas event, only 13,000 years ago, global temperatures declined 2-6℃ from values similar to the present, within about 30 years, after having risen on a similar timescale. The colder climate lasted 1000 years, causing glaciers to advance once more across Eurasia and North America, and then ended as abruptly as it started. There is evidence that the sudden increase in aridity caused during this time disrupted and postponed early development of agriculture in the Levant.

Our economic systems and patterns of settlement are not equipped to deal well with this degree of instability nor its possible result of a return to the hothouse equilibrium.

Analogs For Intervention

It’s obvious that something will need to be done about this, either to adapt to changing global climate patterns, or ideally to control them. Fortunately, the geological record does not only forecast harsh change, but also contains example mechanisms of mass carbon removal from the atmosphere, which could help us to understand our options.

The two most relevant massive drawdown events occur during the Carboniferous period, and the Eocene epoch:

In the Carboniferous, around 300 million years ago over a period of tens of millions of years, CO₂ concentration declined tenfold from over 2000 ppm to about 200 ppm, similar to levels at the most recent glacial maximum, and the only other time they have fallen this low in the last 500 million years. Unsurprisingly, this also resulted in the previously mentioned ice age, a particularly long and deep one. The carbon drawdown was caused by the evolution of land plants before terrestrial herbivores and fungi became adept at digesting wood. Huge forests proliferated across the continents and dead trees simply laid where they fell without decomposing efficiently, until they were eventually buried.

Millions of years of this process sequestered enough carbon to raise the oxygen concentration of the atmosphere to 35% (today it is 20%). This allowed the evolution of giant insects, whose respiratory system depended on this elevated oxygen, and terrifyingly destructive firestorms, where even wet wood would burn. When fungi, termites, and herbivorous vertebrates evolved and proliferated enough to end the sequestration process, and start returning carbon trapped in dead vegetation to the atmosphere, the ice age ended. Glaciation did not occur on Earth for another 270 million years. Much of the coal we burn now represents carbon removed from the atmosphere during this ice age.

The second example, the Azolla event, was only discovered in the last 20 years, and was extremely specific in its mechanism. In the early Eocene, global surface temperatures were 8 ℃ higher than today. Tropical forests extended almost to the poles, and atmospheric CO₂ concentration was 3500 ppm, in contrast to today’s 400 ppm. Around 49 million years ago, the global climate suddenly cooled by 6 ℃, with a simultaneous drop of atmospheric CO₂ to 650 ppm. The cause of this drop in temperature and CO₂ was unknown until a convincing hypothesis was proposed in 2004:

The Arctic Coring Expedition took advantage of reduced ice cover to take core samples from the Lomonosov Ridge, in the Northern Arctic Ocean. To their surprise, they discovered huge deposits of the freshwater aquatic fern Azolla locked under the seabed. These deposits, up to 20 meters deep, were laid down 49 million years ago, sustained over 800,000 years, and exactly coinciding with the observed drop in temperature and CO₂. Noting this correlation, the researchers proposed a causal link, now widely accepted, between the strange growth of Azolla and the massive drawdown in atmospheric carbon dioxide, accounting for the sudden global cooling in the Eocene.

Azolla is incapable of growth in salt water, and so was only able to grow in huge volumes in the Arctic Ocean due to a set of unusual factors involving a freshwater surface layer. These factors also ensured that the carbon absorbed by the fern through photosynthesis was sequestered on the seabed rather than being returned to the atmosphere when the plants decomposed.

During the Eocene, the Arctic Ocean had limited water exchange with the rest of the world oceans, and received the flow of a number of large rivers. The lower density freshwater from these rivers pooled on the surface, causing a low salinity layer to form. This layer prevented mixing of the water column, and created a seabed with extremely low oxygen levels, while also enabling Azolla to proliferate across the surface. As a floating plant with leaves above the water, Azolla had direct access to atmospheric CO₂. Additionally, and unusually among ferns, Azolla is able to fix nitrogen directly from the atmosphere using a bacterial symbiosis, removing a further limitation to growth. Most decomposing organisms require oxygen for their metabolism, and this was severely depleted on the seabed. Therefore, when the ferns died and sank, they were preserved intact until buried by sediment, and the carbon they had absorbed in life was locked into the sedimentary rock under the sea. Ironically, much of the oil and natural gas resources currently under active exploration in the Arctic are composed of these Azolla deposits.

A common and vital lesson can be drawn from these historic events: previous episodes of warming have only ended when specific events in the biosphere, like the Carboniferous drawdown and the Azolla event, caused mass drawdown of CO₂. The cessation of emissions alone, and the background rate of carbon extraction from the atmosphere, have never been sufficient to reduce atmospheric CO₂ either to a level or on a timescale relevant to human civilization.

Possible Geoengineering Methods

Even currently locked-in emissions and warming may cause dramatic climate instability, or a phase transition to a stable hothouse equilibrium, where the factors maintaining the equilibrium may be powerful enough to resist any attempt to return. Therefore, active intervention in the climate system, not merely changing structure of economic activity, seems necessary to avoid the potential of an uncontrolled climate catastrophe.

Such intervention could be no different in principle to the examples we’ve just looked at. The Azolla event is particularly informative, as it shows how specific conditions obtaining in the physical environment can cause a self-sustaining biological process with a favorable ratio of carbon extraction to energy investment.

In any case, some kind of geoengineering or controlled carbon drawdown is necessary to ensure optimal and stable conditions for human civilization.

While previous examples of biological carbon drawdown have mostly occurred over timescales of at minimum several millennia, any human effort would need to be effective on a much shorter timescale.

There are four realistic candidate geoengineering methods that could be used to solve this problem. These are direct air capture (DAC), bioenergy with carbon capture and storage (BECCS), injection of aerosols into the high atmosphere, and oceanic iron fertilization:

Direct Air Capture is simple, but is currently too expensive at approximately $100 per ton of CO₂. DAC startups don’t even plan to sequester, but rather to produce carbon-neutral fuels for profit, which isn’t helpful for geoengineering. Government funding could provide incentive for sequestration, but the expense would be too high, and the technology won’t be mature enough to scale any time soon; Canadian startup Carbon Engineering’s pilot plant only extracts one ton per day.

BECCS consists of biomass being cultivated at large scale and used as feedstock to create biofuels such as ethanol, replacing fossil fuels in the same roles, or directly burned to generate electrical power. The CO₂ released in the process is captured by scrubber systems and injected deep underground, usually into porous rock formations. This enables a portion of the CO₂ collected by the plants in photosynthesis to be sequestered, while the rest is used to create carbon neutral biofuel and electricity. In theory, this enables a BECCS system to achieve net negative emissions while remaining economically viable from the sale of biofuels and electrical power.

BECCS is an elaboration of simple carbon capture and storage, where scrubbers are used to capture exhaust gases from fossil fuel burning, or biomass is pyrolyzed into biochar to be buried or used as soil amendment. These methods are already in operation where carbon credits act as an economic incentive.

Because of its close integration with existing technologies and emissions sources, BECCS is currently the favored method for achieving net negative emissions, but like DAC suffers from problems of scalability. A BECCS project would need to economically support both its fuel and power production, and its method of carbon sequestration, using the sale of the commodities produced. This is unlikely to be viable in the near term without either significant carbon taxes or increases in fossil fuel prices, which the advent of fracking has made unlikely for now.

Using current methods, removing 10 gigatons of CO₂ annually using BECCS would require a land area the size of India to be brought under cultivation for this purpose alone, which would entail an expansion of agricultural land without compare in recent history and be ecologically disastrous in itself. If BECCS projects replaced existing agriculture, this would displace the cultivation of food on arable land. BECCS will almost certainly be a component of future strategies to effect carbon drawdown, but cannot be rolled out on the scale required within a short time frame, and would be too destructive as a sole solution.

Atmospheric Aerosol Injection is the next method. It is possible to provide a direct cooling effect by injecting aerosols—most models using SO₂—into the stratosphere to partially block out the sun. This method is not included in mainstream roadmaps, but would be extremely economical and deceptively efficient. By directly reducing incoming solar radiation and causing global dimming, SO₂ injection mimics the short term effects of a large volcanic eruption, and enables the prevention of warming without addressing CO₂ emissions at all. By this mechanism, aerosol pollution from industry likely masked some of the impact of increasing CO₂ levels during the 20th century.

While SO₂ causes acid rain if released at low altitude, if injected into the upper atmosphere it would have little to no effect at ground level, and would provide effective cooling if continuously replenished. SO₂ could be injected through large hoses attached to tethered balloons, from pressurized canisters aboard specialized or commercial aircraft, or through airburst artillery. Addition of a few million tons of SO₂ yearly would be sufficient to offset the warming effect of a doubled CO₂ concentration. This is a comparatively small quantity, and could be distributed by a fleet of modified 747 airliners for a cost of 1 or 2 billion dollars yearly. This would provide effective mitigation of climate change within months. With costs comparable to those of operating a small airline, this method could be implemented by most states, or even by wealthy individuals.

The ease of implementation of this method makes it almost inevitable that it will at some point be employed unilaterally. For example, if a strong heat wave occurs in a city in Southern Asia during a period of high humidity, as will likely happen at some point in the next few decades, and wet bulb temperatures rise above 35℃, this will cause mass casualties directly attributable to climate change. Currently, the impacts of heatwaves are seen as an increase in the background death rate, and mainly affect the very old, very young, and people with existing health problems. But in this case, previously healthy people will start dying in the streets without access to air conditioning.

This level of obvious impact will create an imperative for the government of the country concerned to do something, anything, to mitigate the situation. Given the low cost of aerosol injection, and its immediate effects, there is a strong possibility that injecting SO₂ into the stratosphere will be their response, with easy justification using the argument that the tragedy was the result of centuries of emissions by large wealthy nations. This could have complex geopolitical consequences, given that aerosol injection will modify the global climate quickly and obviously, causing easily observable whitening of the sky.

At first glance, aerosol injection seems to be the ultimate “get out of jail free card” for climate change, and it does have a place in preventing immediate catastrophic consequences of runaway warming. I would not be at all surprised if it were employed at some point. It does not, however, provide a long term solution.

First of all, shading the planet will not affect the rate of ocean acidification, which creates severe problems in its own right. Dissolution of CO₂ into the ocean lowers the pH of seawater, disrupting the oceanic biosphere. We can see this happening already: since 1750, the average pH of the world ocean has decreased from 8.25 to 8.14.

Second, any aerosol injection system requires continuous management and calibration to give the right degree of cooling. Successful short term management of climate using this method will take the pressure off and thus reduce the employment of more long term solutions that address CO₂ directly. If CO₂ concentrations continue to increase during a period of effective solar radiation management, an unstable situation could result where any interruption or miscalibration in aerosol injection will result in sudden extreme warming. This is the same mechanism that caused previous mass extinctions after flood basalt events, as we noted above.

Oceanic Iron Fertilization. The ideal geoengineering method for tackling the carbon problem would mimic a naturally occurring process which has been shown to mitigate past episodes of global warming, would not be prohibitively expensive or environmentally destructive, would have beneficial effects beyond those of reducing climate change, would not create dependency on continuous careful management, and would have self-limiting negative consequences if misapplied. Such a method exists, has been demonstrated experimentally for decades, and does not require the creation of any massive infrastructure or technological innovation. This method is oceanic iron fertilization.

Iron fertilization takes advantage of the fact that the growth of marine phytoplankton is primarily limited by micronutrient availability, specifically that of iron, and that repletion of available iron in the photic zone of the oceans is able to induce massive algal blooms, which then sink into the abyss, sequestering vast quantities of carbon. Iron fertilization has been experimentally demonstrated twelve times since 1993, most notably in 2007 in the South Atlantic by an Indo-German consortium (project LOHAFEX, 2007), and in the North Pacific in 2012, by the Haida Salmon Restoration Corporation (HSRC). In both cases the addition of ferrous sulfate in quantities between 10 and 100 tons resulted in massive plankton blooms visible from space, with an estimated production of up to 100,000 tons of algae per ton of added iron. Even assuming that much of the carbon contained in this algae is eventually returned to the atmosphere, this is a most favorable ratio. Oceanographer John Martin was aware of this as early as 1988, when he stated “give me half a tanker of iron, and I will give you an ice age.”

Several iron fertilization experiments have been of doubtful legal status, and societal reactions to them have been polarized both in the popular press and academic contexts. The HSRC release was organized by the HSRC in collaboration with Russ George, founder of the San Francisco based company Planktos Inc.. They were careful to release the iron in technically international waters, which gives an idea of the ambiguous legal situation. George was subsequently described in much of the media as a “rogue geoengineer” and “Pacific Ocean hacker,” and accused of manipulating the Haida into supporting his scheme. In reality the project was collaborative and the Haida were equal partners, standing to benefit economically from the expected increase in salmon catch.

As intended, there was a correlated massive increase of the 2013 year class of sockeye salmon by up to 400% due to increased food availability. The causality is still debatable, but a similar increase in salmon population occurred in the years following the 2008 eruption of the Kasatochi volcano in Alaska, which introduced a large quantity of iron-rich dust into the atmosphere, providing strong evidence that iron fertilization in general increases fish stocks.

Currently, the majority of iron input to the deep oceans originates from desert dust storms over the Sahara and Central Australia, which also contribute to the fertility of the Amazon rainforest, due to their global transport by wind currents in the high atmosphere. There is evidence that the increased input of windblown iron rich dust into the oceans due to aridity in the last glacial maximum was partly responsible for maintaining low CO₂ and cold temperatures. The quantity of this windblown dust has been decreasing in recent decades, and this is correlated with a decline in ocean fertility.

Another major source of iron and other nutrients is whale feces. Whales distribute nutrients throughout the world ocean by feeding on krill and small fish in productive regions of upwelling, and spreading nutrients throughout their migrations by defecating in surface waters. Krill contain so much iron that the feces of whales become red, containing 10 million times the concentration of iron in ocean surface waters. But the global population of large whales is still only at most 20-30% of that before industrial whaling, despite a decades old ban. Studies of average water transparency suggest that there has been an annual decline of 1% in ocean phytoplankton populations throughout the 20th century. Artificial ocean fertilization would help reverse this trend as a side effect.

While one ton of iron can theoretically stimulate algae growth to fix 80,000 tons of CO₂, the actual sequestration efficiency is reduced by several factors. As noted above, the HSDC project was associated with a large increase in salmon population, which represents the transfer of carbon into the bodies of these fish. Some were caught and eaten, and others returned to the rivers of their birth to spawn. None of this carbon was sequestered, but the increased productivity of the salmon fishery and transfer of nutrients to Pacific coast forests are beneficial in themselves. Sequestration is dependent on downwelling currents separating phytoplankton from surface nutrient cycling. Because of this, choosing the right location to fertilize is of paramount importance. The Antarctic circumpolar current and areas of the Central Pacific with sustained downwelling are currently the leading candidates. Estimates of sequestration from recent iron fertilization experiments average about 20%, with the Southern Ocean giving the higher reported values.

While much work remains in sorting out the details, a quick calculation is possible at this point. Assuming a 20% near term sequestration efficiency in optimal sites, and accounting for inevitable inefficiencies in the distribution of iron, a sequestration of 10,000 tons of CO₂ per ton of iron is reasonable. Global anthropogenic carbon emissions in 2017 were 41 gigatons. Were we to attempt to entirely offset these emissions with iron fertilization, 4.1 megatons of iron would be required. Global crude steel production in 2017 was about 1.7 gigatons, so this represents the diversion of roughly 0.2% of annual iron production, although using scrap and recycled iron would be preferable. Unfortunately, the algae would become limited by other nutrient deficiencies long before this point, there may not be enough downwelling areas in the oceans to remove such massive quantities of carbon from circulation, and global marine ecosystems would be heavily disrupted by such a massive ongoing fertilization campaign, though these challenges could no doubt be mitigated to some degree.

In a real world scenario including ongoing emissions reductions and limitations to fertilization efforts, 20 percent at most of this quantity of iron would be used—about 800,000 tons. This is equivalent to the capacity of 2 large oil tankers. If we assume the use of an easily soluble iron source, such as iron sulfate, this brings us up to 2.2 megatons, about 6 tankers per year. While this is economically non-trivial, it could be done for less than 10 billion dollars per year, comparing favorably with BECCS and DAC, which require human intervention at all stages of the sequestration process, and would cost at minimum several hundred billion in order to offset a significant fraction of current emissions.

Oceanic iron fertilization actually offers a very attractive and practical method to help solve the problem of global warming, assuming it is deployed alongside other negative emissions technologies and at least some degree of emissions reduction. There is no need to use only one method to the exclusion of all others.

Humans Now Control The Carbon Cycle

Examining the prehistorical natural analogues is vital to predicting the future consequences of different geoengineering methods. For example, the closest analogue to BECCS is the mass burial of terrestrial vegetation in the Carboniferous. Aerosol induced global dimming mimics the effects of large volcanic eruptions without the associated CO₂ injection. DAC is a form of accelerated weathering. Iron fertilization has analogues in both the Azolla event and more directly in the addition of iron rich wind borne dust to the oceans during recent glacial periods.

While we don’t yet know everything, and we will continue to learn more about the costs, efficacy, and other effects of the different methods, oceanic iron fertilization appears very attractive. It is more economically efficient compared to BECCS and DAC, has none of the Faustian instability of aerosol injection, and has the significant benefit of increasing fisheries and oceanic ecosystem productivity.

What is clear is that a focus on emissions reductions alone is not going to work. Projecting climate impacts in terms of arbitrarily defined residence times of CO₂ in the atmosphere ignores the fact that ocean dissolution is not an acceptable long term means of removing CO₂, that natural sequestration is not a process occurring independently from human activity at a constant rate, and that even optimistic emissions reduction scenarios may throw the global climate into a period of severe instability.

Whatever our intentions, and independently of our fossil fuel emissions, humans now exert a dominating influence on the carbon cycle. We appropriate roughly 30% of the primary production of the terrestrial biosphere to grow food and produce raw materials such as wood, paper, and textiles. We cannot expect the biosphere to continue to function as though 7.7 billion humans are not present, simply because we reduce or eliminate fossil fuel use.

Unlike the many previous organisms which have caused global disruptions in climate simply by doing too well, we have the ability to intervene deliberately to mitigate the worst environmental consequences of our sudden dominance, and even stabilize and optimize the global climate system and carbon cycle to our liking.

This entails taking an active geoengineering approach where we stop acting as if the climate system is both beyond human control and too sacred to affect, and instead admit ourselves direct control, responsibility, and husbandry over the Earth. This transition from treating the Earth’s climate and ecological systems as wild and beyond the human order, to treating them as gardened integral parts of the human order, would massively impact how we see ourselves. New areas of opportunity would open up with the rise of such technologies.

For example, oceanic iron fertilization opens the door to fertilization and productivity management of ecosystems in general. What if we could turn deserts into productive pastures, turn barren areas into rainforests, boost the productivity and diversity of existing ecosystems, and have enough sustainable wild fish to feed everybody? What if we could choose to manage the earth to massively increase gross biological productivity using fertilization and careful control of the carbon cycle towards a massively more fertile and prosperous planetary ecosystem? These things are not immediately possible, and there will no doubt be pitfalls along the way; however, once we take the plunge into geoengineering, such possibilities start to seem almost inevitable.

On the other hand, we could also continue to stumble blindly toward extremely unpleasant climate disasters while desperately trying to minimize our impact, and descending into accusatory but practically useless political infighting. There is a clear choice here, and while geoengineering will require unprecedented global coordination and unity of purpose, it’s not even all that expensive. In the long term it is our only tenable option, and the opportunities are endless.