Eco-augmentation can be defined as humans’ deliberate and strategic interactions with nature, intended to create more resilient, robust and sustainable ecological systems. This includes activities that lead to increases in the diversity and functions of ecosystems. It is now possible in principle to modify organisms to make them more resilient to environmental shifts such as climate change; to introduce genetic changes that will propagate through a population of pathogens or parasites, and render them harmless; and to create organisms that can restore biological functions that have been lost from an ecosystem, or enhance those that exist. However, these technical advances can also pose severe risks to ecosystems and humans, such as runaway invasive species or even ecological collapse.

On a political and ethical level, there is little consensus about what would constitute augmentation of an ecosystem. Ecosystems are always changing, with species or functions being variously lost or gained, and it is not always obvious whether these changes are “good” or “bad”. Arguments can be made that “augmentation” means:

• restoring an ecosystem to its previous state (pre-human or at some point in time)

• reshaping it to suit a “new normal” (such as the effects of a changed climate)

• increasing some measure of productivity (especially in agricultural ecosystems)

• adding new functions (such as processing an additional nutrient)

• boosting dominant species in the hope they will support co-existing rare species

• increasing the diversity of micro- or macro-organisms

• adding functional redundancy to make the ecosystem more robust

• limiting invasive species, pests, pathogens and other undesirable species

Multiple scientific fields are integral to eco-augmentation. Ecology is one: to augment an ecosystem, we must first understand it and have some idea of the ways in which it might be “improved”. This includes understanding the microbiology of the ecosystem, including any extremophiles, and the underlying genetics. It also includes understanding the connections amongst and between species and communities of all size-classes and their roles in ecosystem function. Likewise, modelling and forecasting are key to predicting the potential consequences of any intervention.

Another key field is synthetic biology, which allows precise edits to be made to a species’ genome, or even the creation of entirely new forms of life. These applications are in their infancy but developing fast, so deliberate or accidental release of synthetic organisms into ecosystems is a real and imminent possibility. Unfortunately, despite considerable progress in the characterisation of ecosystems it remains extremely challenging to predict the impacts of such an introduction. Unintended effects are possible or even likely, potentially including low-probability, high-impact events such as a runaway invasive species or ecological collapse. Improving our ability to predict how ecosystems will respond to an intervention is thus essential.

Eco-augmentation also poses significant ethical and philosophical challenges. Some argue that eco-augmentation should benefit humans, while others believe ecosystems have intrinsic value regardless of their utility and should therefore be enhanced for their own sake. But defining what constitutes an improvement to an ecosystem, or the restoration of a degraded one, is non-trivial, depending on humans’ relationship to ecosystems and the ways we conceive of them. These questions rest as much on cultural and political values as on technical questions. These values will derive significantly from the communities who will be most directly affected by any attempt at augmentation — who must therefore be consulted and whose cooperation must be secured, because they will have no way of opting out of the consequences.

Key opportunities

Illuminating the biodiversity darkspots

Despite our growing knowledge of the world’s ecosystems, there remain many ecological “darkspots” where little or no data is available. These include the deep seas — the largest biome on earth — the McMurdo Dry Valleys of Antarctica, some particularly biodiverse terrestrial regions, and the microbial realm — particularly those microbes we cannot yet cultivate. Our lack of knowledge still hampers our ability to model and understand these ecosystems, and the world-spanning interactions between ecosystems. Filling in these darkspots is critical to our ability to develop key models to understand the effects of change, both passive and human-directed, on our global ecosystem.

Forecasting ecosystem changes

The same factors — habitat loss, invasive species, climate change and so on — pose similar threats to many ecosystems. That such threats may be as predictable as the weather opens up the possibility of forecasting and early-warning systems. These would enable rapid intervention, mitigation or enhancement when key events are detected or predicted. This requires three advances. First, frequent surveys using multiple tools and collecting multiple forms of data to maintain baseline measurements for an ecosystem. Second, rapid and easy access to this data online, ideally through open-access databases. Finally, the ability to model ecosystem changes and their evolution through advances in computational capacities, artificial intelligence and complex systems science.

Creating more robust ecosystems

In general, the most robust ecosystems seem to be those which have a high degree of functional diversity. For instance, nutrients can be processed in differing ways, creating a complex web of interactions among species and remaining functional across a wide range of conditions. The tools of modern biology and ecology, coupled with more information and understanding of metabolic function, will give us an opportunity to build this robustness into future ecosystems. One possibility would be to identify preserve those functions that enable ecosystem robustness, reducing the risk of collapses and other harmful events. However, diversity that may seem non-functional at a given time may turn out to be necessary at another, and is thus worth preserving for its own sake.

Hacking co-evolution

Changing a human society changes the associated ecosystems, and vice versa. Synthetic biology and de-extinction technologies have advanced to the point where we will soon be able to design and synthesise new organisms, or even entire ecosystems, rapidly and cheaply, and release them into the wild. The question then is: what do we want to do with these technologies? Deepening our understanding of human-ecosystem co-evolution helps us to identify ways to bring about positive change for humans and ecosystems, and to understand and avoid the associated risks.

Introduction to eco-augmentation

It is now possible, in principle, to deliberately modify ecosystems and enhance their functions: “eco-augmentation”. This might include releasing new genes or entire new organisms, with the aim of restoring a lost ecosystem function; making a keystone species more resilient to climate change; or boosting an ecosystem service deemed to be desirable, such as carbon storage. In theory, such eco-augmentation has the potential to undo some of the damage humans have done to ecosystems, to stabilise or restore ecosystem services, and to reduce the risk of catastrophic collapses of ecosystems and of the societies that depend on them.¹

Existing approaches to eco-augmentation include the introduction of new species, of modified species that can use new resources or survive in particular conditions, or of nutrients in new environments where they are limiting factors (such as iron in the sea). Advances in synthetic biology are now enabling the creation of highly modified and entirely new forms of life, particularly bacteria. Tools like CRISPR allow genomes to be edited with unprecedented precision and thus allow rational design of modified organisms. While this form of biotechnology is still incipient, it is developing extremely rapidly, so synthetic organisms could soon be released — deliberately or otherwise — into the wild.

However, the impacts of eco-augmentation are barely understood. Ecologists can make predictions about the likely impacts of a given intervention in an ecosystem, but these predictions often come with enormous error bars. In particular, it is not possible to state with confidence that extremely harmful impacts, such as a new species becoming invasive and dominating the ecosystem, will not occur. The extreme complexity of ecosystems means that we may not be able to predict such low-probability, high-impact events with significant confidence for many decades, possibly indefinitely.

There is already considerable potential for engineered organisms to escape into the wild or to be deliberately released. Some synthetic biologists are working on safety techniques, often focused on making it impossible for these organisms to survive without human intervention: for instance, making them dependent on chemicals that are not found in nature.² However, the rapidity with which organisms — especially micro-organisms — can evolve means that such measures may rapidly become ineffective and so are not guaranteed to prevent escape, as happened in the widespread and rapid dissemination of antibiotic resistance among bacteria.

It is profoundly difficult to identify the forms that eco-augmentation should take and its possible environmental consequences. What would constitute a beneficial impact on an ecosystem and what are the acceptable trade-offs? Should ecosystems be restored to a previous state, or radically transformed for a new post-climate-change normal? These questions are about sociocultural values as much as they are about the technicalities of science; they raise profound questions about the relationship of humans to ecosystems, and to the Earth as a whole. A human-centric viewpoint implies that ecosystems should be managed for the benefit of human society, but it is also possible to view the ecosystems as intrinsically valuable, potentially being recognised as having rights of their own — legalistic protections like those recently granted to such geographical features as mountains and rivers. Our legal and political systems have not yet come to grips with these questions.

Reading and writing ecosystems

Our capacity to read an ecosystem — to identify its constituent organisms, components and processes — is increasing rapidly. Technologies such as efficient environmental DNA analysis, faster and cheaper DNA sequencing, protein and other metabolites analysis capabilities, remote sensing and machine learning have dramatically improved both the volume of data available and the insights that can be gleaned from it. Complex interactions between the behaviours of individual species and the functioning of whole ecosystems are being mapped.

This includes inter-species communication, which is highly prevalent and important for the functioning of ecosystems, and takes numerous, often surprising forms. For instance, bumblebees can cause plants to speed up flower production during times of pollen scarcity by damaging their leaves.³ Chemical agents, including volatile chemicals that can travel significant distances, also exert strong influences over organisms and ecosystems.⁴

Meanwhile, models of ecosystems are growing in sophistication. It is now possible to simulate the key species and processes in an ecosystem, and to make predictions about shifts such as changing distributions under climate change,⁵ but such models can have much wider applications. However, these models will need to be improved in order to analyse not just the supposed key species but the majority of those in an ecosystem, including the rare ones, since they may also prove critical for the function and resilience of an ecosystem.

The growing range and depth of ecological data opens the possibility of creating “digital twins” of ecosystems, with information on species composition and interaction derived from a range of survey data. However, for the majority of ecosystems, the baseline data needed to feed the models are scarce. Closing the baseline knowledge gap would allow these digital twins to be used to better understand the processes underpinning the ecosystem and to identify stressors.⁶

The range of ecosystems under scrutiny has also expanded. While early ecological studies often focused on large animals and plants living on land, there is growing understanding of remote marine areas, including previously under-surveyed areas like the deep sea,⁷ and the microbial ecosystems which form the basis of all food webs.⁸ But even on land, there are still many darkspots where little information has been gathered.⁹ One survey of vascular plants identified 32 darkspots, almost all found within biodiversity hotspots such as Borneo and Colombia.¹⁰

In other cases, data exists but is not readily available in any systematic way. Such “greyspot” ecological data remains locked away in individual papers, often behind paywalls. However, initiatives are under way to increase accessibility and to create large, open-access databases of basic information and analysis, along the lines of the very successful NCBI GenBank model. These initiatives have the potential to greatly increase our understanding of ecosystems. For instance, the BioTIME database hosts over 8 million abundance records for over 40,000 species, drawn from over 500,000 locations.¹¹

Alongside this, there is growing potential for “writing” ecosystems. Molecular biology and synthetic biology are providing a growing range of tools which can be used for this purpose: for example, CRISPR-Cas9, which allows precision editing of an organism’s DNA. Meanwhile, synthesis of DNA is becoming faster, cheaper and more accurate. Taken together, these allow us to modify existing organisms or design entirely new ones.

The potential is exemplified by the Synthetic Yeast Genome Project (Sc2.0), the aim of which is to produce the first eukaryotic organism with a wholly synthetic genome.¹² Sc2.0 researchers aim to replace all the existing “natural” chromosomes in a yeast cell with synthetic ones. The consortium has synthesised all the chromosomes individually and is now focusing on their integration.¹³ Other researchers have engineered micro-organisms that produce biochemicals not found in nature, such as “non-canonical” amino acids.¹⁴

While most such research has been lab-based, the same technologies can be applied in ecosystems. The Mice Against Ticks project aims to reduce the numbers of people infected with Lyme disease on the islands of Nantucket and Martha's Vineyard. The disease is spread to humans by ticks, which are first infected by white-footed mice. The researchers aim to use CRISPR to create mice with heritable immunity against Lyme disease — meaning that their offspring will also be immune, thus permanently cutting the chain of infection if such immune mice interbreed with wild populations.¹⁵

Mice Against Ticks has been designed in consultation with community members and is intended to be geographically contained. Other proposals envisage organisms which depend on particular environmental or nutritional conditions, thus restricting their potential spread. But not all approaches to eco-augmentation can be readily constrained, meaning that they may affect people or ecosystems who were not initially considered. For example, gene drives — suites of genes edited to “override” natural selection and spread rapidly through a population — could spread widely once released.¹⁶

There is precedent for how to handle such open-ended possibilities. The 1975 Asilomar Conference on Recombinant DNA established guidelines for managing the risks associated with DNA editing, including the prohibition of certain types of experiments.¹⁷ The widespread adoption of these guidelines built trust among policy-makers and the public, enabling the growth of the biotechnology industry. Analogous guidelines could similarly build trust in synthetic biology and eco-augmentation.

Unfortunately, there is no tradition of communication between the fields of synthetic biology and ecology. In particular, there have been few well-controlled experiments in which synthetic micro-organisms are released into realistic laboratory ecosystems to monitor their effects, and even fewer field experiments. Synthetic biologists have instead focused on designing organisms that depend on highly specific conditions and are thus unlikely to survive in a wild ecosystem.¹⁸ This yields a different, more technological, way of providing assurance that experiments or deployments are unlikely to have dramatic unforeseen consequences. However, since organisms have evolved many ways to escape imposed constraints, there may not be a safe way to provide containment at the desired level of confidence.

5-year horizon

• Advances in the design of warning systems and of conservation genomics for threatened species and systems, based on multiple monitoring tools and sensors, perhaps using AI to inform and accelerate decision-making.

• DNA sequencing and synthesis that is many times cheaper and more accurate than today’s. Further improvements to cost, accuracy and accessibility of other “reading” technologies, such as single-cell sequencing and spatial omics.

• Design of multiple, widely applicable safeguards to help prevent synthetic organisms from surviving in the wild if they escape.

• Creation of a single-celled eukaryote (yeast) with wholly artificial chromosomes.

• Development of large collections of viable bacteria, seeds, embryos and so on to preserve their genomic resources prior to the release at large scale of genetically engineered organisms.

10-year horizon

• Implementation of early-warning systems for threatened species and systems, based on multiple monitoring tools and sensors, perhaps using AI to inform and accelerate decision-making.

• DNA synthesis primarily done using enzymatic techniques, replacing existing chemical-based techniques. Efficient, cheap benchtop synthesisers for long DNA.

• Implementation of multiple, widely applicable safeguards to help prevent synthetic organisms from surviving in the wild if they escape.

• Designer chromosomes for eukaryotes.

• Automated construction of synthetic genomes and insertion into host organisms.

• Widespread use of DNA molecules to store massive amounts of information.

25-year horizon

• Portable DNA synthesisers

• Systematic methods for culturing “microbial dark matter” — the vast majority of micro-organisms, whose favoured growth conditions are not well understood today

• Writing of other -omics such as proteomics and 3D cellular structures.

• Customised organisms leading to new biodiversity resources.

• Transitioning ecosystems

Ecosystems are constantly changing. New species evolve, others become extinct, others enter from outside and environmental conditions alter, with knock-on effects on all the constituents. Ecosystems are always in transition. But these processes used to take thousand or even millions of years to occur. Today, due to human activity, they are taking place at a much faster rate, exceeding the rate at which the constituents can adapt .

Remote and extreme ecosystems are among the most prone to fast transitions because they exist in precarious states. For instance, the Antarctic McMurdo Dry Valleys are the coldest and driest habitat on Earth, and the closest analogue we have to Martian conditions.¹⁹ Yet they are inhabited by a considerable diversity of micro-organisms.²⁰ While the productivity of these ecosystems is low, they have a high level of functional redundancy and have survived for at least 100,000 years even as Antarctica has undergone major climatic shifts. It is essential to study these ecosystems: they are likely to undergo transitions in the near future whose potential effects remain unclear.

There is also a need to better understand why some organisms and ecosystems are particularly prone to dramatic transitions while others are highly resilient. Corals are a notable example: even a small rise in average temperatures can trigger bleaching, in which the corals expel their symbiotic algae, leading to mass death. This means corals are extremely vulnerable to climate change. Global coral-reef cover fell by 50% from 1957 to 2007, and only 10% of coral reefs are expected to survive past 2050.²¹ However, corals in the Red Sea are more temperature-resistant than those elsewhere, so might be preserved if non-temperature stressors are minimised.^{22 23}

On the other hand, the extremely diverse oasis of Cuatro Ciénegas Basin in Northen Mexico — an analogue of early Earth due to the diversity and abundance of stromatolites — is deeply endangered. The deep aquifer that feeds the ponds is under extreme pressure due to the bad management of alfalfa agriculture in the desert. This is particularly poignant, since this site has survived isolated for hundreds of millions of years, allowing the evolution of many new lineages in the tree of life.^{24 25}

The BioTIME database has already acted as a valuable sense-check: it indicates that, contrary to popular belief, species richness is not declining globally, but that there is considerable regional variation, with some ecosystems declining in species richness while others are increasing.²⁶ BioTIME also shows that many species and communities are shifting to smaller body sizes.²⁷ There is an urgent need for more such open-access ecological databases.

This kind of knowledge is enabling ecologists to identify warning signs of imminent changes or collapses in ecosystems. Just as weather forecasts provide advance notice of extreme events, ecosystem monitoring could be operationalised to give early warning of dramatic shifts. This would enable interventions to protect or strengthen threatened ecosystems, such as swapping in a new organism to perform an ecological function that is being lost.²⁸

Such models and related simulations should be used to help identify harmful unintended consequences of eco-augmentation interventions. It is distinctly possible that a synthetic organism or gene could achieve exponential growth in a particular ecosystem, essentially becoming invasive. While some ecosystems might be resilient against such invasions, we do not know which ones. The precautionary principle suggests that we should consider all ecosystems to be vulnerable.

5-year horizon

• Massive open-access online repository for diversity and ecosystem data, analogous to GenBank for genomes.

• Remote sensing using, for example, hyperspectral imaging — imaging areas at many points on the electromagnetic spectrum simultaneously — to identify ancient grassy ecosystems at large scale.

• Screening systems based on volatile chemical cues to detect disease outbreaks and ecological shifts.

• Detection of ecosystem tipping points, aided by AI, through monitoring of variables such as spatial patterns of vegetation or fire spread.

• Study of the stratosphere in order to understand the dynamics of aerosols, as well as the ozone hole and the effect of nitrogen oxides and rocket fuel at such high altitude.

10-year horizon

• Widespread automation of biodiversity monitoring, reducing global blind spots.

• Development of “digital twins” for ecosystems around the globe.

• Advances in identification and monitoring of small-organism and underground biodiversity, for example, using passive acoustic monitoring.

• Rapid and quantifiable marine ecosystem monitoring via deep-learning techniques to interpret audio and video.

• Significantly expanded mapping of deep-sea and sea-bed ecosystems.

• Biobanks for ecosystems.

25-year horizon

• Reliable predictions of ecosystems’ ability to withstand multiple stressors, integrating global climatic change and atmospheric and geochemical process.

• Ability to fully reconstruct or repair complex ecosystems, with a focus not just on lists of species but on functional interactions such as nutrient flows.

• Hacking co-evolution

Ecosystems — including the physico-chemical environment, microbes, plants and animals — have been co-evolving for millions of years. And in many cases, these ecosystems and human biology as well as culture continue to co-evolve. Human activity is changing these co-evolutionary processes, often inadvertently and sometimes to the detriment of our own interests. In a dramatic example, humanity’s widespread use of antibiotics and pesticides has created selection pressure which has caused many organisms to evolve resistance, so they are once again threatening our healthcare and food systems.²⁹

By using multiple techniques, ranging from traditional surveys to sequencing of environmental DNA, ecologists are now able to monitor such co-evolutionary processes over large areas and many years. In theory, it should be possible to intervene intentionally and rationally in order to obtain desirable outcomes — such as ecosystem restoration or improved human health — to “hack” co-evolution. For instance, urban communities often have elevated rates of allergies and other immune-mediated conditions, but Finnish children in kindergartens with higher levels of biodiversity showed enhanced immunoregulatory pathways.³⁰

Such “beneficial” interventions could be considered as eco-augmentation focused on the human ecosystem. A key challenge for such interventions is that evolution is known for its capacity to produce surprises even in the simplest ecosystems. For instance, simply growing two bacterial species at 32ºC instead of 22ºC was enough to upend their ecological relationship: the predatory species became the prey.³¹ Such complex and unpredictable outcomes are not the rule, but ecologists do not yet have a means of predicting when they are more or less likely to occur. Interventions may also be more or less effective depending on the organisms involved, the kind of intervention and the complexity of both the environment and of the intervention.

Complex-systems theory offers some guidance, because it indicates that the underlying architectures of ecosystems are often relatively simple, but even in these systems the results can be complex and even chaotic. Ecologists have also identified many emergent rules that apply across ecosystems: for instance, the majority of species in any ecosystem are rare. However, even the best models struggle to identify “tipping elements” in ecosystems, where a small push can cause a big change — yet it is precisely these elements that offer both the greatest opportunities and the biggest risks.

Further challenges arise around the ethics of such interventions. There is little consensus on whether it is acceptable to alter ecosystems for the benefit of humans, what the criteria for deciding would be, or who of the different human (or animal or plant) groups in a given ecosystem deserves to benefit. While concepts like ecosystem services can be interpreted as identifying the ways ecosystems benefit humans, ecosystems have their own intrinsic value independent of humanity in some philosophies such as “deep ecology”.

Consent may prove to be a critical issue, but may be different to reach them if different human groups have very different values, and also because the lack of data. The history of conservation suggests strong engagement with affected human communities is essential to any eco-augmentation project during both design and implementation. Restoring an ecosystem is not simply a technical task, but is also about the people who do the work, whose relationship with nature is often profoundly changed by the process.³²

5-year horizon

• Systematic understanding of which ecosystem and human-health outcomes are supported by biodiversity and relative effect sizes.

• Capacity to fully “read” and understand any accessible ecosystem: complete description of genomic composition, as well as other "omics" (transcriptomes, epigenomes, proteomes and so on) plus data on its single species (population biology), community and ecosystem ecology, physiology and interaction with the environment, in particular in face of global change and so on.

• Advances in detailed understanding of biochemical and metabolic processes like bacterial metabolism to ecosystem-scale effects

10-year horizon

• Experimental interventions in controlled environments to promote human health by promoting and augmenting biodiversity.

• Large-scale meta-analyses of co-evolution, and experiments involving interactions between several species, offering estimates of how predictable evolution is and the distribution of rare evolutionary events.

• Detailed computer simulation of ecosystems becomes a general practice, using detailed descriptions of candidate gene pathways and responses to major drivers, and of the results of the interaction between species and their populations, community and ecosystems dynamics, augmented by more powerful computing systems and AI.

• Detailed, well-controlled and replicated experimental management of particular ecosystems, for instance to analyse how to maintain their robustness or to increase a chosen ecosystem service.

• First “de-extincted” animal (a modern animal with some traits of an extinct species, rather than a revived but truly extinct animal).

25-year horizon

• Large-scale experimental interventions to promote human health by promoting and augmenting biodiversity.

• Biodiversity experimental interventions for intergenerational human-health outcomes, such as changing the environment of a parent in order to reduce asthma risk in their descendants.

• Active management of ecosystems, for instance to maintain their robustness or increase a chosen ecosystem service.

• Automated and rapid (perhaps seasonal) assessments of ecosystem species diversity, interactions, genomics and health.

• “Meta-models” linking detailed biochemical processes like bacterial metabolism to ecosystem-scale effects