The Pilot Donation Recommendation

After considering many different interventions, we have determined that the Good Food Institute (GFI) is the organization which most effectively reduces global biodiversity loss (in terms of the number of evolutionarily distinct species saved from extinction per dollar).

Donate to the Good Food Institute

How does GFI curb extinction?

1. The key driver of extinctions globally is habitat loss,

2. which is primarily driven by agricultural expansion

3. in order to satisfy the growing market for meat.

4. GFI’s work reduces the demand for meat permanently.

GFI is dedicated to improving the global food system for the planet, people and animals. They strategically tackle neglected gaps for alternative proteins by cultivating scientific developments, increasing adoption of alternative proteins, and encouraging private sector growth. The part of their work that we’re most excited about is precision fermentation: a process which allows specific proteins to be produced at-scale.

If we could replace just 20% of ruminant meat protein with precision-fermented protein by 2050, we could halve annual deforestation; pound-for-pound, precision fermentation requires 79-86% less land area than beef. We penciled an estimate that every $6,800,000 given to the Good Food Institute will result in one more species being saved from extinction in the next 200 years.

Precision fermentation is a good target because it is currently underfunded, highly scalable, at a stage with high leverage, and has the potential for massive global impact.

It has the potential to permanently reduce deforestation and biodiversity loss within 5-10 years, and stands out above other methods to reduce future biodiversity loss.

Our research process

On our mission finding the world’s most effective biodiversity interventions, we have collected a longlist of 65 biodiversity interventions, narrowed them down to have a closer look at a shortlist of ~15 of the most promising ones, identified the top 3 and then roughly quantified their impact. From there, we searched for organizations actively working on our top intervention, compared their cost-effectiveness and capacity to absorb donations, and singled out the Good Food Institute.

Read more about our process here.

Other praise for GFI

• Founders Pledge recommends GFI for their positive impacts on animal welfare.

• Giving Green recommends GFI for their positive biodiversity impacts.

Have we convinced you? Let us know!

If our work has informed your decision to donate to GFI, please do so via the buttons on this page and send us an email so that we can track our impact.

We would love to hear from you!

Couldn’t use the buttons or donated already? No worries - we can still count your donation as long as you email us to let us know.

Donate to the Good Food Institute

Support our work directly:

Like what we do? With your help we could do so much more.

Donate to EcoResilience Initiative

The Case for Fertilizer as a Biodiversity Lever

Theory of Change

The principal cause of recent rapid biodiversity loss has been habitat loss, degradation, and fragmentation. Though partly due to increasing land area allocated to urban development, agricultural expansion has served as the primary catalyst. The global human population has doubled over the past 50 years, and per capita caloric intake has increased by 20%. To meet this need, natural land areas have been modified leading to a decrease in biodiversity across critical ecosystems.

However, fertilizers can serve as levers to mitigate this loss in biodiversity. By increasing crop yield per hectare, fertilizers reduce the total amount of land area needed to accommodate a human population. Thus, fewer parcels of “natural” land area need to be reallocated to agriculture, thereby decreasing habitat loss, and the correspondent loss in biodiversity. This case is interrelated with the previously published shallow dive on Improved Crops, which is worth a read to contextualize the following.

Geographical Relevance

Since Haber and Bosch’s introduction of synthetic fertilizers in 1909, the technology has proliferated widely. However, fertilizer application rates vary widely throughout the world. The rate of fertilizer application is significantly higher in wealthy countries than in the developing world. In 2024, the rate of fertilizer application in China was 394 kg / ha of arable land, while in the US it was 127.8 kg/ha. Conversely, in Sub-Saharan Africa (SSA) it was 18.2 kg/ha. Introducing fertilizer could have an outsized impact in areas where it is underapplied, namely SSA.

The varying relationships between agricultural land utilization and cereal yields provide valuable insights into the scale of the benefits of fertilizer use. In SSA, cereal yield and cereal land use track quite closely over the past 50 years. Conversely, in China where fertilizer use is excessive, the growth in cereal yields has far outpaced the growth in cereal land use. There, enriching the soil has improved the yield of existing farmlands. Following China’s example, fertilizer use could dramatically augment cereal yields in SSA, and allow them to be the principal driver of cereal production growth, rather than expanding land use. SSA is forecasted to have disproportionately high human population growth over the next few decades, resulting in an increased pressure to alter “natural” land area into agricultural land to feed this growing population. Given that SSA is a biodiversity hotspot, thwarting the existing trend is necessary for preserving critical habitat.

Risks and Countervailing Forces

Challenges to Land Sparing Logic

While the logic above seems to have held in the natural experiment of regions with high and low fertilizer usage over the past 50 years, it is not immediately obvious that it would hold as an intervention in SSA now. Jevon’s Paradox describes the economic phenomenon wherein decreasing the cost of per unit production incentivizes further production. Jevon noticed this with the increasing efficiency of coal engines in the 1800s leading to an increase in rates of coal consumption. Applied here, it would suggest that increased crop yield from fertilizers could incentivize further agricultural land expansion, rather than sparing more habitat.

Jevon’s Paradox is contrasted with the hypothesis of the Green Revolution pioneer Norman Borlaug, who introduced many yield-increasing programs to the Global South. The Borlaug hypothesis holds that any increase in yield will disincentivize habitat degradation as it would be more economical to more intensively farm existing agricultural land than convert ‘natural’ land. Existing literature points to the truth being somewhere between Jevon and Borlaug’s frameworks: yield increasing technologies do not seem to have as severe of a backfire effect as Jevon posits, but are not nearly as effective at preventing habitat loss as Borlaug asserted. It stands to reason, then, that fertilizers could contribute to diminishing the expansion of agricultural land as SSA’s population continues to grow, but the exact degree remains unclear.

Second Order Effects of Fertilizer: Nutrient Pollution and Eutrophication

Along with fertilizer application comes the risk of nutrient pollution, wherein nutrients from fertilizer that are not absorbed by crops significantly alter surrounding and downstream ecosystems. This leaching of nutrients harms these areas, most notably through soil acidification and eutrophication. As a result, biodiversity often dwindles. Nitrogen Use Efficiency (NUE) quantifies this impact by comparing the percentage of nitrogen in a yielded crop and the amount applied to the soil as fertilizer. Global NUE is roughly 40-50%, indicating that about half of applied nitrogen is not absorbed by intended plants, and instead leached into the environment. In countries with extremely high rates of fertilizer application, NUE is much lower. In China, it is 0.25, indicating that 3 out of every 4 kg of fertilizer applied is leached into the surrounding environment, providing even more detriment to biodiversity and intactness of downstream ecosystems.

While high rates of fertilizer application bring significant risk of negative secondary effects, these may not be as prevalent in SSA. At present, the NUE in SSA is over 100%, indicating that soils are being actively stripped of nutrients rather than receiving an excess. As such, there is likely leeway for increased fertilizer application before nutrient pollution becomes a significant concern. However, existing studies have found that smallholder farmers in the developing world are disproportionately likely to overapply fertilizers due to their lack of access to other inputs such as high quality seed and machinery. Given the relatively low governmental regulation capacity, mitigation/intervention may prove difficult. It thus appears that even under modest increases to fertilizer access in SSA, downstream nutrient pollution remains a risk. It’s not immediately clear what the optimal fertilizer application in SSA would be, but it certainly seems to be greater than it currently is from both a welfare and a biodiversity conservation perspective.

Takeaways and Comparative Efficacy

Increased fertilizer usage in SSA is a promising biodiversity intervention with ample positive spillovers for human welfare and soil sustainability. The present NUE over 100% implies that increased fertilizer application would initially bring minimal risk of nutrient leaching. Moreover, this concern could be further mitigated through programs that incentivize optimal application. Regulatory and technological support seem to augment impact potential.

Finally, contextualizing this efficacy alongside other biodiversity interventions can provide insight into paths forward. Existing literature supports varying conclusions about the most effective approach, with some finding that shifting diets in developed countries away from livestock more effectively alleviates the pressure to expand agricultural land use, while others support closing yield gaps to be of superior impact. As it stands, increased fertilizer usage in SSA seems to have a high potential to mitigate biodiversity loss in a hotspot, and high counterfactual impact given the current trends in land use intensification and population growth therein. This solution requires scaffolding to achieve its optimal impact, but can prove to be a valuable piece of an effective biodiversity intervention.

Terra Bioindustries is doing exactly the right work, but they’re not set up for donations at this time. If you are looking for a place to donate right now, you can give to the Good Food Institute here, which supports the broader field.

How we Chose: Current Bottlenecks

There are several important blocks holding back the development of precision fermentation into a dominant food source. Here is a well-composed list of the current unsolved obstacles. (Briefly, they are: sugar feedstock, microbial strains, bioreactors, fermentation processes, valorization, and downstream processing.) Sugar feedstock may be the biggest one - half the price of operation for precision fermentation is from sugar input. And sugar demand is going to rise to more than double current global production according to Synthesis Capital.

Particularly for biodiversity, removing the need for sugar crops would reduce conversion of biodiverse areas to cropland due to reduced cattle and cattle feed production. Precision fermentation is a net positive for global extinction reduction, even with sugarcane feedstock. This holds true across all levels of market penetration - it causes positive land use change for biodiverse continents. Still, sugarcane is currently mostly grown in India, Brazil and Thailand - and if demand increases, it risks expanding to other tropical regions.1 This is a biodiversity threat we can foresee and avoid, while also increasing the speed that precision fermentation expands.

To make massive amounts of sugar, it is important to use a plentiful and cheap resource, and a similarly low-cost process. Beating current sugar crops is a tall order, but it will need to happen to compete with other protein sources. Here is Good Food Institute’s deep dive on feedstock production. Agricultural waste is the prime target here, primarily because of its size.

The Contenders

A few companies are working on sugar yields - Terra Bioindustries in particular is doing exactly what we are looking for - they are isolating sugar from ag-waste (corn stover and brewers spent grains). Other important actors are: Hyfé, which is producing low-cost sugars from water waste. And Fibenol is producing sugars from wood-waste.2 Pow.Bio is another interesting company, with their “continuous fermentation” technology halving costs.3

Terra Bioindustries’ direction is most exciting to us. Their goal is mass production, not expensive, specialized, small batch proteins. They are producing both sugar from ag-waste, and mycoprotein. They estimate that their protein flour will cost more than most soy protein but less than most pea protein. Their mycoprotein is “incomplete” - this means it is missing one of the nine essential amino acids (specifically: lysine). Their more important project from our perspective is refining sugar from corn stover and brewers spent barley grains. Corn stover is especially good, because it is the highest volume ag-waste. Both protein and sugar are valuable, but for advancing precision fermentation, it is the work on sugar that is most exciting to us at ERI.

There are some drawbacks to Terra Bioindustries sugar production process. Their sugar, “recyclose” contains both glucose and xylose. Pure glucose (C6 sugars) is desirable for fermentation. Xylose (C5 sugars) is much less accessible, and including xylose can slow down the process of fermenting. To be clear, producing a mixture of C5/C6 is typical for ag-waste feedstocks; it is an unsolved issue. One proposed solution is engineering microbes to use C5. This would help precision fermentation be more sugar-efficient, but optimizing microbe strains to produce protein already has many constraints. Using microbe strains which are already optimized for other traits is crucial to bringing costs down and improving production efficiency. Terra Bioindustries says that their sugar “recyclose” is indistinguishable from glucose for some precision fermenters. But they may be testing with microbe strains that are capable of digesting C5 xylose sugars.

Despite these drawbacks, we believe Terra Bioindustries presents a transformative opportunity. Their goals closely match our strategic priorities for advancement in precision fermentation biotech. There are few other companies or labs working on this specific problem. At their current stage, funding would have outsized impact - they are scaling production to demonstrate commercial viability by 2027 and are looking for $5 million to reach this milestone. Last year they came in 6th out of 1000 entries to The Livability Challenge. Early-stage investment now could determine whether they accelerate to commercialization or are forced to end their development trajectory altogether. Their performance has a chance to kick-start ag-waste feedstocks as a viable option, bringing the costs down for precision fermentation and preventing tropical forest loss from happening.

Please note, that this is only an outside view: We haven’t checked with Terra about their situation, eg. if they already have secured enough funding, or how exactly are their expectations for the future. To really make a reliable donation recommendation, we should also check that part and form a well-informed opinion on their team and plan. But that takes more time and work. This is how far we got for now.

Another limiting factor is that we aren’t a fund, so there’s not much we can promise to them in exchange for their time and potentially secret data.

Recent Breakthroughs

A few researchers are also working on sugar yield science. NREL has a division with several top scientists working on the problem.

In particular Xiaowen Chen, now of the University of Minnesota, recently participated in a breakthrough where their process using corn stover might produce sugar at 28 cents/lb (low-end estimate). For comparison global sugar price varies from 10-25 cents/lb, and is higher (35 cents/lb) in protected markets. This is a cutting edge development, published July 2025, 4 months ago at the time of this writing.

Hongzhang Chen of the Chinese Academy of Sciences is working on some low-cost methods to break down lignocellulose (plant matter) to free up the sugars more easily using “steam explosion.” Steam explosion is probably more cost effective than specialized enzyme mixes.

Jake Lindstrom at the FDA also studied sugar yield from fast pyrolysis.

Donate Now?

Unfortunately we don’t know of a way to easily donate directly to the above impactful initiatives. We at ERI want to make it easy to support this groundbreaking work, and will do what we can to provide access to interested donors as soon as we are able. Any updates will be posted here. If you are a big donor and want to take initiative, you can contact Terra Bioindustries by reaching out to Rebecca Palmer directly or using their contact form. Alternatively, you can reach Xiaowen Chen here, Hongzhang Chen here, and Jake Lindstrom here.

However, if you just want to donate today, an easy way to do so is to give to the Good Food Institute. The main reason to do so is that they are trying to funnel money, talent, and information to where it will bring alternative proteins to mass adoption.4 We penciled an estimate that every $6,800,000 given to the Good Food Institute will result in one more species being saved from extinction in the next 200 years. (If you donate, we request that you add a comment mentioning ERI or tell us you donated, so we can track our effect. And thank you for improving our future!)

The main caveats are that the Good Food Institute is not directly concerned with biodiversity consequences, and is not prioritizing transitioning away from sugar-crop feedstocks. They are not directly supporting Terra Bioindustries nor the other scientists I have mentioned here.

Good Food Institute does support ag-waste processing including Hyfé in a general way, although they prioritize other areas. We were able to rapidly gain in-depth understanding of the industry, key bottlenecks and leverage points using their documentation. They connect scientists, companies, funders, and provide comprehensive and accurate reports which inform decision-making. They are pushing alt-proteins forward and shortening timelines. Without their work, the field would be less informed, connected, funded, and much more neglected. They also have a grant program which channels money where they believe it will most effectively bring alternative proteins to mass adoption. For further evidence see Giving Green’s 2025 analysis (recently released!), Animal Charity Evaluator’s 2023 analysis, and Charity Entrepreneurship’s 2025 recommendation.

Good Food Institute has a donate button here. (Thank you for making a difference!)

Please consider sharing this analysis with others interested in effective giving, or who want to learn about high-impact conservation interventions.

Precision Fermentation addresses one of the biggest causes of biodiversity loss - pressure to convert natural land to agriculture and livestock. The industry is in a nascent state, when funding can have a high impact on the course of history. This intervention has the potential to permanently reduce biodiversity loss. This makes it our strongest recommendation.

Why we are confident it will succeed:

Alt proteins have become more accepted and more widespread in recent years. (see the rising number of flexitarians, market acceptance of impossible burger, and the expected tripling of market growth of alt proteins in 6 years.) The main barriers to adoption are taste, convenience, nutrition, and price. These are diminishing as market research matures, prices fall rapidly, and accessibility increases.

Why this is so impactful for biodiversity:

Habitat loss is by far the number one cause of extinction risk. Most land conversion is due to livestock and livestock feed. This trend is predicted to worsen as global wealth increases, particularly in biodiverse regions. Precision fermentation dramatically reduces land use: achieving 20% market penetration could halve deforestation, while 80% adoption could halt agricultural land conversion altogether.

How funding will be put to use:

According to Ambitious Impact scale-up-funding is very neglected. Precision fermentation suffers from one-time start up costs (mostly infrastructure) that make it difficult to compete with the established competition.

While you might assume commodity protein production would attract sufficient investment, several factors prevent adequate funding: The eventual market will operate on low profit margins and growth requires specialized research and scaling equipment. Therefore, precision fermentation does not receive adequate attention and funding from investors.1

This is a critical time to fund research, particularly for biodiversity conservation. If “second generation” precision fermentation processes can swap input feedstock away from tropical high-sugar crops to plentiful agricultural waste, it would not only eliminate feedstock land use but also significantly lower operational costs. Early adoption is important, because production chains will be locked in by existing factories. This will make it increasingly arduous for companies to switch and for ag-waste methods to compete.

Our Runner-Up Recommendation for Biodiversity Intervention: BioBanking

The possibility of solving extinction entirely makes biobanking a crucial path to pursue with long timelines. Traditional conservation has largely overlooked this intervention, despite the rock-bottom price per sample and multiple conservation benefits. Right now there are very few biodiversity biobanks. The main remaining uncertainty is whether adequate gestation technology is possible, but we expect this challenge is most likely surmountable. A potentially disqualifying drawback to this intervention is that it does not contribute directly to habitat conservation. We would like to note that biobanks help prevent extinction through supplementing genetic diversity and allowing the piecemeal rebuilding of ecosystems. Read more on this newly released shallow dive!

Other highly effective biodiversity interventions: Introducing Keystone Species

Introducing keystone species includes some very high-impact outcomes, particularly through wetland creation. Keystone species benefit multiple specialized species, are self-perpetuating, and represent a relatively inexpensive intervention for their impact size. Some habitats cannot exist at all without these species. However, a drawback is that implementation varies from location to location, requiring localized approaches that create unavoidable scaling overhead. Before we can recommend specific keystone species introduction programs, we need to conduct additional research.2

What Makes our Top Choices Powerful

As we investigated 15 different contender interventions, we looked for interventions that had strong performance across several key criteria:.

• Evolutionary Distinctiveness Impact

• Scaling Well

• Long Lasting/Self Perpetuating Effects

• Simple Implementation

• Addresses Land Use

• Leverages Tipping Points

• Positive Ecosystem Complexity

• Neglected

• High Potential Upside

• Avoids Development Issues

• Addresses Climate Change

• High Funding Need

• Benefits Many Species

Precision fermentation caught our eye because it satisfied many of these criteria at once: It would be self perpetuating if it becomes an affordable desirable food source. It addresses habitat loss, which is one of the biggest causes of extinction (and it is expected to continue to worsen, particularly in biodiverse tropical countries). Precision fermentation changes incentives and reduces pressure to convert land to agriculture globally. It’s underfunded. It could have a massive total impact if it scaled up to be a significant proportion of our diet. After we investigated further, it held up as a plausible funding target: even with sugar cane feedstock and the first 20% of market penetration, it would cause positive land use change for biodiverse regions.

Biobanking has similar characteristics. It is not self-perpetuating, but it could address the threat of extinction in a scalable, cost effective way for the most evolutionarily distinct threatened species. It remains underfunded by conservation organizations despite its potential for global impact if strongly supported.

Keystone species possess many desirable intervention characteristics.Though less scalable than our other top recommendations, they are self-perpetuating, leverage ecological tipping points, and substantially improve ecosystems. Scaling may be possible to improve, if generalist keystone species could have positive impacts in a variety of locations. Introducing keystone species may be neglected due to the popular resistance against taking an active role in conservation.

Other interventions required ongoing funding, did not scale well, or weren’t as neglected. For a combination of reasons, the other interventions we reviewed did not emerge as top contenders for this first year of our biodiversity intervention assessment. We would like to highlight some of the other interventions which performed well in this comparison, and we hope to share more details on our comparisons eventually.

Expected Value Analysis

Most academic papers estimating cost seem unrealistically high.3 We found examples ranging from $5,200,000, $1,300,000, $682,396, $635,000, to $380,000 annually per species. Cost-effectiveness estimates are rare in conservation literature, likely because it is not advantageous to produce such estimates for a number of reasons. Ongoing maintenance is often required, unpredictable variance is common, and it’s difficult to accurately estimate reduction in extinction risk. The lack of cost/benefit awareness is a known systematic problem. We were unable to find more readily available project-based estimates of cost per extinction reduction rate. Manual estimation from available data would help improve these comparisons.

Precision Fermentation

Precision Fermentation is difficult to estimate because it reroutes economic drivers and lowers future deforestation rates. It is unclear which locations or species populations will be impacted by international market forces. At 20% of market share for microbial proteins, 50% of future deforestation would be prevented. That is 78Mha of forests saved every 30 years, particularly in sub-Saharan Africa. We estimate this would save somewhere between 345 - 4810 species after 200 years of deforestation averted in sub-Saharan Africa. If we spend 3x as much as we have for the last 10 years (~20% of $5.5 billion) in technological development to reach 20% of market capture, then the cost for research would be $3-$15.5 billion, and the per-species saved rate would be between $45,000,000 on the high end and $600,000 on the low end, with a midrange of $2-10 million. This is not including all the non-forested areas prevented from conversion to grazing/agricultural land. Even so, this is 7.6x-130x or about 20x less per species extinction than academic estimates for 200 years of perpetual funding.

The full estimated cost to reach 20% of market share is $100 billion-$200 billion from build out and supporting technology development. Investors have funded alternative proteins to the current multi-billion market size. If the current pace of growth continues (tripling in 6 years), it will take 18 years to reach 225 billion and the market reach we hope to achieve.4

This estimation involves substantial uncertainty and should be carefully revised before it is used for further comparisons. These academic cost estimates are likely unreasonably high. Adoption of precision fermentation would “solve” demand for habitat loss more permanently than most traditional conservation actions. The further we project, the better precision fermentation’s relative performance. But conclusions beyond 200 years seem unwise. For example, other land use threats like energy generation may make such agricultural land use savings necessary but insufficient.

Keystone Species

Keystone species were much simpler to estimate. Introducing a keystone species in the Dhansiri River in Aassam, India, costs approximately $2,470,000/species for 200 years of operation. This included floating mats of vegetation for water quality remediation, and mussel propagation for 8 years. For comparison, the actual sewage treatment plant budget for the Dhansiri River is $680,000. This budget includes awareness building, river monitoring, pollution tracking, reforestation, straining, counteraction for violations, etc. for 25 years. This sewage treatment benefits many highly evolutionarily distinct native species. Assuming reconstruction or expansion of the plant every 50 years, the cost would come out to $2,720,000 for 200 years of operation. These two interventions are roughly equivalent in cost and impact. But we caution that there is enormous uncertainty in these estimations.

Biobanking

Biobanking is even more cost effective at 127k per species for 200 years of preservation. It provides essentially unlimited time to wait for resurrection technology to become highly cost effective. Cloning currently costs somewhere between 6k-1,500k suggesting an eventual cost somewhere in the middle. Our rough conclusion is 133k-172k per species. The upper limit is set by Colossal’s current price at $188 million per species,5 and the concern that artificial wombs might be quite expensive.

NB: This is an ongoing process and much of what we present here is imperfect - think of it as a “poor man’s methodology” - but we believe that there is value in being transparent about where we are and how we got here.

Core Values

Our priorities are extinction prevention and thriving ecosystems, with a focus on long-term timescales. Read more about us on our website and on the EA forum.

We are extremely resource-limited and had difficulty assessing thriving ecosystems, so we decided to focus almost exclusively on extinction prevention for our first assessment. We hope to correct this in later iterations.

We follow a cut-down version of Charity Entrepreneurship’s research process.

Big List of Interventions

We started with a big funnel - listing every intervention we are aware of, and checking for more. We sorted them by the biggest extinction risks, level of funding, thriving ecosystems, impact on extinction prevention, and duration of their impact. We collected a list of important things to consider such as scaling, feasibility, complexity, track record, and problem growth/shrinkage, which would come in handy for our next phase.

Prioritization Dashboard

We then created a “prioritization dashboard” where we specifically tracked Tree of Life, Thriving Ecosystems, Longtermism for every intervention. This dashboard was designed to figure out “where to start researching.” It is an internal thinking tool, with guesses and informal comments. Please note that it does not reflect our final conclusions, nor the rigor of our deliberation. In the dashboard we started to list our certainty, the area impacted, counterfactuals, funding neglect, and other relevant notes. We began marking which interventions merit further investigation. This dashboard went through two or three re-formulations as we rethought our prioritization approach.

A key part of this phase was posing our biggest crux questions for each intervention that would determine whether they would be good candidates for further research, or if they had fundamental issues. For example: “What is the best ecosystem engineer, and how much impact do they create?” “Where do canals exist in the world, and are these the same places that need more wetlands?” After answering questions like these, we narrowed down to about 20 from the original 65 (and many more maybes).

It is during this phase that we stopped tracking “thriving ecosystems” as a component of our current impact assessment. Ecosystem health is not a well documented part of conservation, trying to evaluate it was too complex and it was simply too time consuming to pursue.

Key Evaluation Factors

We realised that we needed to investigate which factors would have the biggest impact on the outcomes of these interventions. With support from the Sustainability Incubator Program we hosted an “adaptive action” workshop on our five key factors for evaluating intervention effectiveness: Species Affected, Extinction Risk, Future Projection, Operational Requirements, and Community Support. We reached out to dozens of multidisciplinary professors, and in the workshop we explored the merits of a range of indicator metrics for each key factor.

A Turning Point

Our efforts were refocussed following an enlightening conversation with another researcher (also working on biodiversity prioritisation), where we discussed their process and how it differs from our own. This revealed how we can take advantage of our unique research strengths in this space: EcoResilience Initiative - as a small team of agile researchers outside of traditional conservation organizations - can more easily research interventions with “left field” approaches and high-risk high-reward profiles, and are more free to openly assess controversial methods. We began writing the shallow dives immediately afterwards with renewed confidence.

Shallow Dives

We started by listing the characteristics of a strong intervention. Next we wrote a “first impressions” paragraph on every intervention; stating what makes it likely/unlikely to be the top intervention. Each month we selected 4 more to research in-depth, writing a 1-5 page overview, and published our findings. In the next 3 months we released 13 shallow dives. These painted a picture of each intervention’s track record, potential for impact, and biggest bottlenecks. These preliminary research summaries allowed us to rank interventions or discard them if we found intractable problems. At the end we could clearly select the top 3-7 interventions, and began comparing them in a more rigorous way. We chose to leave 5 other interventions from our top 20 for later investigation.

Top Interventions

We will compare the top 3-4 interventions side by side and will compare them with benchmark “average” conservation interventions. We will produce a final intervention recommendation in the next month. We are also beginning to list the organizations that are working on those interventions in preparation for our November donation recommendation.

What’s next?

In our upcoming October newsletter we will announce our initial results, along with a quantified comparison with a benchmark “average” conservation intervention. Don’t forget to subscribe and look out for updates!

Subscribe now

Others working in this space: GivingGreen’s Biodiversity Recommendation!

How can I help?

Donate - Connect us with funders - Connect us with collaborators - Volunteer

Share EcoResilience Initiative

Subscribe now

Biobanking is the idea of preserving (usually by freezing) tissue samples (ideally embryos) from species at risk of extinction. The hope is that at some point, almost certainly within the next 200 years, we will be able to use these samples to re-create the organism and effectively “de-extinct” a species.

Why aren’t we Biobanking?

We aren’t doing it en-masse right now for several reasons. Primarily because it’s a new technology. St. Judes in 1976 was perhaps the first frozen biobank. We weren’t able to even read DNA 60 years ago, let alone edit it until 12 years ago, we didn’t anticipate the usefulness of intact tissue samples until recently.

Although de-extinction is beyond our scientific capabilities today, we naturally expect that biotechnology will improve in the future, perhaps eventually making this a viable, scalable conservation strategy. Other environmental-conservation groups, perhaps because of a tendency to see technology in a negative light, underrate this avenue.

Given that it seems inevitable that many species will go extinct over the next decades/centuries, dismissing de-extinction approaches for essentially public-relations reasons (the belief that talking loudly about de-extinction will lessen public support for other kinds of conservation) seems ill-advised.1

Consequently, only a few places are biobanking genetic material from wild species, and perhaps only Revive and Restore is doing so with the explicit objective of aiding hypothetical future de-extinction projects as we’ve outlined here. In that sense, this work is extremely neglected.

Cost

On top of that, the cost is minimal. Something like $2000 for 100 years for one sample.

Some example biobank costs:

$200 for freezing one sample, pluripotent cells.

$1 per vial per year storage

$1-5 per year storage

Collection and other one time costs would add more to the total.

$15,000 for a biological collection expedition is reasonable. The collectors might gather 30 per expedition or $500 per sample. Scaling up and targeting easily collected species could greatly decrease cost per sample. But let’s estimate $1500 in one time costs for travel, record keeping, operations, equipment, sample preparation etc and $500 for 100 years storage again due to storage, operations, record keeping. $2000 for one sample seems like a reasonable estimate.

Multiple Samples

We will want to do many more than one sample per species. To properly de-extinct a species you will need a thriving population of diverse individuals. The specific number varies2 but the common rough estimate is 500 breeding individuals. And then we will want some redundancy, so maybe three times that, so there are samples all over the world and we can lose 30% of them to unforeseen disasters and slow degradation. We would target something like 1500 samples for each species we are preserving. That is about $3,000,000 for 100 years per species, and we can save them, effectively forever. For comparison, one study estimates it costs about $1,300,000 per year to keep critically endangered species surviving in the wild with insurance populations in zoos (which the authors consider a low cost).

But there are other problems and concerns that should be mentioned.

Sample Degradation

How much samples degrade over time is unknown because biobanks have only existed for less than 50 years. If samples degrade too far before they are used for de-extinction, they may not save a species from being lost. So far we know they can be viable for 50 years. (27 years in humans) We see indications that embryo vitrification is basically indefinite in storage length and thawing is successful 80-95% of the time.

Perhaps de-extinction can only be achieved with embryos. That would raise the cost and difficulty of acquiring the tissue samples significantly, maybe as much as 100x as much.

Required De-Extinction Technology

It is possible that we cannot bring species back without a functional womb. Perhaps surrogate wombs from related species can work. The major obstacles are: rejection of a foreign body, gestation time mismatch, and uterine and placental differences. These seem difficult to achieve, especially if we’ve lost the ability to study what the embryos need. Artificial wombs may be required. But these do not seem like they are impossible obstacles. Given that we couldn’t read DNA 60 years ago, merely another 60 years might be enough. Foreign body rejection can be suppressed with current immunosuppressents in humans. Gestation time and uterine/placental differences seem eventually solvable. We have just begun to apply modern breakthroughs to biotechnology and AI is likely to continue to accelerate our expansion of capabilities. It seems highly likely that it will be achieved at some point within the next 200 years.

Parental Upbringing

Perhaps the species needs an upbringing, or certain initial conditions to survive.

Upbringing is a concern for social animals which receive some kind of care from their parents (altricial). It may be very difficult to replicate many aspects of rearing, such as the transference of the gut microbiome and the teaching of social cues. Some amount of parental care is nearly 100% in mammals and birds, but rare in most species overall.

It is important to document upbringing in endangered altricial species and try to recreate it when performing de-extinction, but fortunately it will not pose an obstacle for the majority of species (Insect parental care is understudied, but appears to be similar to the above percentages).

Re-introduction to the Wild

Perhaps their habitat is lost and cannot be re-created.

It is important to preserve habitats for re-introduction. Practicing de-extinction is one of the ways habitat can be re-created. The hope is that we could de-extinct species one at a time, rebuilding until the habitats are sufficiently advanced to support the other species’ de-extinction. Fortunately plants tend to survive mass extinctions, are relatively easy to keep surviving populations, easy to preserve, and easy to de-extinct relative to other species.

There are several cases of totally dependent ecological relationships. Obligate mutualism such as all lichen species (20,000), corals (800, most parasites (200,000), and ant-plant (250) relationships. Pollinators are mostly generalists. Perhaps 4,000 species amongst the orchid specialists, fig wasps, specialized bees, birds, and fly pollinators. In total this represents ~300,000 known obligate species (0.003%) that it may not be possible to practice de-extinction. We should take special care to help these species survive.

Current state of de-extinction technology:

• We are able to clone a deceased individual from a 33 year old biobanked sample. (Black footed ferret)

• We are able to implant an embryo into closely related species and bring it to term. (Guar and Bongo)

• Colossal Biosciences is able to insert bits of extinct species DNA to create hybrid individuals. (Dire wolves)

• Immunosuppressants increase birthrates from embryo transfers (Several)

• We are able to trigger stem cell/pluripotent cell generation (2006)

• We are learning how to recreate lost habitat (Tallgrass Prairie Preserve, Pleistocene Park)

• We can teach species some of their lost survival skills (Whooping crane, Condor, Orangutan )

• Protein folding has been solved (AlphaFold2) and AI holds promise for unlocking other biomedical abilities

• There are some large uncertainties, which is one reason why this is not our top recommendation for biodiversity preservation.

Remaining bottlenecks:

• The importance of embryos vs tissue samples

• Working immunosuppressents, implanting embryos, and pluripotent cell generation across diverse species types

• Successful birth from distantly related wombs or artificial wombs

• Teaching altricial species all their lost survival skills

• Recreating lost habitat

• Successfully re-introducing extinct species to the wild

• Other unknown problems

Even if We Fail

Even if we never fully de-extinct any species we will have preserved the evolutionary history, cell structure, and biochemistry of these species. Their information can help inform the development of biotechnology, the study of our evolutionary history, genetic rescue, the creation of artificial biodiversity, or ecological monitoring. Conversely, if these species never go extinct the samples will still be useful for genetic diversity.

An Opportunity for Extinct Species

There is the possibility of biobanking species that are already extinct. Namely the specimens thawing out of the permafrost right now. There are very few people collecting these samples3 on shoestring budgets, as thousands more priceless specimens rot away every year. The researchers in Siberia and Canada are mostly collecting large mammals like mammoths, wolves, and sabertooths. Small mammals, insects, plants, and reptiles are rarely collected despite the importance of these extinct species. This is an ongoing tragedy where a very small targeted investment could make a massive difference for our long-term future.

Final Notes:

The possibility of solving extinction entirely makes this a crucial path to pursue. There are few biodiversity biobanks currently. It is not being pursued by traditional conservation. The cost is unbeatably low. The main remaining uncertainty is the possibility of providing adequate gestation.

Effective Altruism has been excited about alternative proteins for a long time (in 2018 and 2023 on the 80k hours podcast). If you haven’t heard about it yet, it’s difficult to believe it’s real.1 But it is very real, and turns out to be one of the best methods of biodiversity conservation.

Instead of growing the whole cow, chicken, pig, etc, what if we just grew the protein? Precision fermentation is duplicating the molecules we like to eat. It sounds futuristic except we’ve been doing it for decades for specific ingredients like citric acid (lemon flavor), acetic acid (vinegar) and vitamins.

The process is:

1. use microbes to generate your molecule

2. nurture them so they’ll reproduce really fast

3. come back and harvest the molecules they mass produced for you

It’s a lot more complicated in practice, but that’s the basic idea. Much like fermenting to make beer, getting the ratios just right is very difficult.

Precision fermentation is exciting because it’s potentially much more efficient than raising cattle2 or growing crops. You are growing just the molecules you want to eat, instead of the whole organism. It’s great for animal rights. It’s great for land use (79-86% less area than beef). It can’t spread infectious diseases to wild animals or livestock. According to a comparative model in 2023, it is also the most nutritional source of protein with the lowest overall environmental impacts (such as water eutrophication, carbon emissions,3 nitrogen/soil acidification, and more).

This is a big deal because per capita meat consumption is increasing, especially in developing countries. It has doubled before and it could double again - from 25kg in developing countries (mostly in tropical megadiverse countries) to 100kg per year as the levels are in industrialized countries. Meat is far less efficient and takes far more resources and land area to produce, compounding the impacts. Producing cattle meat takes about 24 input calories per output calorie. Precision fermentation uses 3-5 input calories per output calorie. That is 5x less already, and this is expected to improve.

Unlike the “improving crops” intervention, this intervention would unambiguously reduce total agricultural land. If it were cost competitive, it would drive down land use demand. If precision fermentation expands to take up 20% of the market, we would see a 50% drop in future deforestation (78Mha particularly in sub-Saharan Africa) even with increased population, according to this well-constructed scenario. Cropland would still expand because of increasing meat demand and sugar requirements for fermentation. At 80% adoption, almost no natural land is converted (0.6 Mha/year), especially in the Congo, Amazon, and Central America.

It is not cost competitive yet, mostly because of startup costs in infrastructure, but some projections say it will be cheaper than chicken in 2040. Prices are already looking like they will drop from $4-5 per kilogram to $2.5 per kilogram in the imminent future. It’s a new emerging technology that might revolutionize agriculture and sustenance. (4 billion dollar industry and is projected to 10x in 5 years). It’s not just about meat substitutes either. Adding protein to flours could improve diets while decreasing the need for land.

Precision fermentation is fed by sugars. These sugars are mostly from corn, sugar beet, sugarcane, and wheat. Anything that is high in sugar, grown locally, and already mass produced. This would imply that as the world shifted to precision fermentation proteins over livestock proteins, agriculture would shift from livestock feed crops (like soy, barley, sorghum) to surgery crops (like sugar beet and sugarcane). Other crops work for both (like corn, wheat, and cassava). That means places that can only grow starchy livestock crops and not sugary fermentation crops, like parts of Brazil, Canada, Australia, would potentially deconvert from agriculture to wild lands. Other areas that previously could not grow starchy crops, might convert land from natural areas to sugar crops. Like in Ukraine and Russia for sugarbeet, and increased pressure in India, Brazil, Thailand, and China for sugarcane and cassava. This is not a desirable transition to high sugar crops, as they tend to grow best in tropics. Sugar beet and wheat have better prospects for land conversion.

However, looking at the scenarios in this analysis replacing beef with microbial protein: even using sugar cane, even in biodiverse countries, this is still a win! The authors project more cropland in Latin America and India (negative for biodiversity). However, this is more than traded for by 1) gaining forest in Latin America and India, 2) losing (massively) less forest in Sub-Saharan Africa and (some) southeast Asia, 3) deconverting large amounts of pasture in Sub-Saharn Africa, Latin America and China, and 4) less crop gained in China. This is an overall win for the biodiverse countries as well across all scenarios 20% to 80% microbial protein replacing beef, even using sugar cane feedstock.

And a way to circumvent this problem is underway. The precision fermentation industry is looking into using ag-waste products or wood pulp as feedstock as well. This is a very real possibility. The drawback is that processing these feedstocks is very energy intensive (explosive steam) to convert it to usable feedstock. Local energy production makeup then becomes an important factor for land use and carbon emissions. The first factories are being built now, so in a few more years (2030) we should see increasing build out. It may be possible in the upcoming future to reach places like South Africa, use yam waste, and create protein rich cassava replacement at cost.

Giving Green recommends it as an environmental intervention. Effektiv Spenden wrote a great article about it here. And for the first time Ambitious Impact recommended it out of their research program. Their summary is: scale-up-funding is very neglected and precision fermentation is great for both the environment and for animals. I generally see the Good Food Institute being recommended.

One of the remaining obstacles is the visibility, perception, and availability of these products. Future use is determined by price, but it is also determined by how well these products integrate and how pervasively they are used. So it might make a big difference to be an early adopter. Using them in recipes, increasing their visibility, and thereby increasing their availability and future use by others. Specifically, it is important that these proteins are not seen as an impractical novelty food, but as a normal ingredient. Other ways to increase their expansion would be to directly donate or invest in precision fermentation companies.

Common Asset Trusts or CATs are fundamentally about capturing and redistributing resource rents - they’re mechanisms to ensure extraction/degradation fees flow to citizens rather than being captured by elites or corporations. Only Alaska has created one, although several others are proposed. Their structure is: They are a trust, separating ownership from the users. They generate revenue from fees on resource degradation. They distribute benefits to all citizens, not just users. They operate at multiple scales including national, using formal legal mechanisms with courts for enforcement. CATs would be ideal for countries with extractive resources leading to rentier economies. But due to Dutch Disease, they often can’t even collect basic royalties, let alone manage complex trust structures. CATs need strong legal systems, banking structures, and political stability. This is best at ensuring sustainable yield or managing revenue generating resources. Biodiversity can be one of many targets for their resource flows.

Compensatory Mitigation Banks are financial mechanisms where developers or companies pay into a pooled fund to compensate for unavoidable environmental damage their projects cause.

They typically work by 1) a development project will harm biodiversity (destroy habitat, impact endangered species, etc.), and that harm can’t be avoided or minimized on-site, 2) the developer pays into a fund. 3) The money is then used to protect, restore, or enhance biodiversity elsewhere.

Critics argue they can become “licenses to trash” - allowing destruction of irreplaceable ecosystems in exchange for money that may not create equivalent ecological value elsewhere. This doesn’t have to be the case. They could theoretically trade less valuable biodiversity loss for greater biodiversity gain in the most important areas and for the most tractable threats. Their usefulness depends on how the funds can be used, and how strict the validation of how successful the restoration projects were.

This does not seem like a promising intervention because it relies on industries voluntarily or (more likely) being forced to give up their profits and governments choosing to spend it on biodiversity (in distant, potentially foreign sites) over other priorities. It can only work in places with strong regulatory institutions. I don’t see how this has enough momentum to be implemented consistently, nor that it will fund lasting conservation work in the places that need biodiversity conservation the most.

The deep ocean contains valuable minerals (⏬) which can be mined in areas that have extremely low biodiversity and very common habitat types. There are rocks (called “nodules”) lying exposed on the sea floor, containing high concentrations of Manganese (27%), Nickel (1.4%), Copper (1.3%), Cobalt (0.25%), and rare earth elements (0.05%). The nodules can be picked up by jets of water and returned to the surface for processing. These mineral-dense areas (such as the Clarion-Clipperton Zone) are vast, and there is a lot of material available -- over 200 years’ worth of global supply of Manganese and Cobalt, and about 75 years’ worth of Nickel.

It is theoretically a viable and lucrative mining source. These minerals are valuable for many kinds of technology and production that are all increasing in the coming decades. They are especially beneficial for solar power, electric motors, electricity transmission, wind turbines, and batteries. Currently 50% of nickel comes from Indonesia, a region high in evolutionary distinctive biodiversity. Net-zero scenarios require large expansions of mineral extraction from somewhere.

Deep ocean mining is not very disruptive to biodiversity for some basic reasons:

• The deep ocean is 60% of the earth’s surface, so the seafloor is by far the most common type of habitat worldwide. The Clarion Clipperton Zone (CCZ) is in Abyssal Clay sediment (31% of the deep ocean floor), as opposed to biogenic sediment like Calcareous (37%) and Sileaceous (23%) ooze which are mostly shallower areas.

• Habitat types are defined by variation in depth, sediment, topology, water masses, and marine snow. All of these vary extremely little across the ocean floor compared with terrestrial variation. Variation comes from the nodules themselves, as they are a hard substrate.

• The seafloor cannot become as densely populated as the surface due to the energy and nutrient limitations, making it more akin to desert or tundra than to something like a forest (or the coral reefs and kelp forests that thrive in shallower waters).

This does not necessarily mean that abyssal life does not vary much genetically. Life down there may specialize for small differences in habitat and have evolved for long periods of time. Especially because this habitat is extremely stable, which fosters survival of evolutionary distinctive species. It is still unknown if the Clarion Clipperton Zone has unique life, found nowhere else in the ocean.1

Biodiversity is very low in the CCZ, and what species are present are mostly worms and organisms smaller than 1 cm. Even if the life in the CCZ is 1) unique, 2) highly evolutionarily distinct, and 3) will be wiped out forever by mining activity, there is an opening for moral judgements on the value of small, million year old, metabolically efficient wormlike life versus high-energy tropical terrestrial large vertebrates. Relying on evolutionary distinctiveness is a countermeasure against our human biases, but it is not the only value in conservation choices.2

The hope is that the deep ocean mining would displace3 highly damaging mining that is taking place elsewhere in the world, particularly in biodiverse jungle. This looks possible in New Caledonia, the Philippines, Brazil and Gabon.

The mining that is most likely to be reduced by ocean competition are nickel and manganese mines. Copper is too readily available and too large of a market to shift with ocean mining. Similarly cobalt isn’t likely to be impacted much, because it is mostly produced as a byproduct in copper and nickel mines. Unfortunately the cobalt mines in the DRC are unlikely to be closed by price reduction, as these “artisanal” (not my wording) mines are more a symptom of poverty than market demand.

Nickel:

Nickel is a small enough market that the deep ocean deposits will change the market significantly. Unfortunately the recently expanded4 nickel mines in Indonesia are relatively low cost and are unlikely to be the first to close.5 Still, nodule production may limit the growth of mining in Indonesia. There is more hope for closing nickel mining in places like New Caledonia and the Philippines, as they are higher cost and could be supplanted by nodule production. The economic disruption would likely hit marginal, environmentally damaging operations first, such as mines facing depletion. This is relevant to biodiversity in the area, as mines near depletion tend to worsen their impact by trying to scrape the last bits of valuable minerals out.

Manganese:

Manganese is available in large amounts in the nodules, so much that it will impact the manganese trade despite it being a large market. Some of the largest manganese mines are located in tropical rainforests - in Brazil and in Gabon - others are located in semi-arid regions like South Africa. The more expensive mines are the ones that are most likely to close first. Another factor is that oceanic nodules can often be processed with much less energy and less CO2 footprint than terrestrial alternatives.

Most places are more biodiverse than the deep ocean, so if deep ocean mining reduces or prevents mines almost anywhere, it would be less damaging to the environment than the deep ocean mines.

Right now deep ocean mining is being held back by environmental groups (including WWF and Greenpeace), government lethargy, and potentially the mining companies themselves that would not want to have their established businesses be interrupted by new competition. It is also held back by technology - these specialized mining rigs are not mass produced yet, in part because the supply is not accessible. Recently President Trump from the USA has opened up deep ocean mining, and the International Seabed Authority has not given a response. However, the United States is not part of the ISA and so USA companies may be exempt from ISA approval.

The technology seems accessible. Nautilus Minerals deployed successful collectors, but failed at scaling up. The Metals Company has deployed successful collectors for weeks and riser pipes at 4km deep but is waiting on NOAA and has yet to scale up to commercial production. Other mining operations have been able to run for months at a time. The biggest technological challenge seems to be the risers (elevators transporting the nodules to the surface). But this part of the process has multiple potential routes forward, so it seems likely that it is only a matter of time until it is solved economically. Once the technological difficulties have been overcome, scaling up will be dependent on manufacturing and investment rather than the richness of the mineral vein (unlike traditional terrestrial mining). This industry seems to me to be very close to opening up. Some deep ocean mining companies are planning to start operations in 2027 and 2028.

It is possible that this backfires if deep ocean mining makes nickel, manganese or rare earth metals cheap enough to encourage complimentary mines, say of lithium, to open in biodiverse hotspots. This depends a bit on how valuable the green energy transition (and other technological developments) is for biodiversity. I leave that to other climate research orgs, like Giving Green, to evaluate.

To read more about the deep ocean, metals, and deep ocean mining, see the “Critical Ocean Minerals Research Center” website. For counterpoints see this scientific article and this scientific american article.

Final conclusion: What if opening the deep ocean to mining leads to extensive permanent damage to the sea floor? Will I regret this damage, looking back after hundreds of years? Maybe. I think our conservation should be based on overall extinctions and loss of the tree of life. With this in mind, I think the positive effects on biodiversity outweighs the losses from deep ocean mining. The numbers of species are just too small compared to the known damage of terrestrial mines.

One of the biggest drivers of biodiversity loss is habitat conversion to agriculture. This is especially happening where livestock is profitable and human populations are growing. So one intervention might be making agriculture more efficient to reduce land use. Perhaps by genetically engineering food crops for better yield, improved use of fertilizers, laws encouraging larger more efficient operations, more tractors, or other ways of improving land efficiency.

(This is going to be a deliberation ladder, aka a twisty ride, so get ready!)

Better crops will help biodiversity:

We can anticipate much more forest clearing in Africa, where population growth is projected to expand most. As detailed in the book Not the End of the World, Africa has a lot of unrealized potential crop improvements that could mean greater production of food using less area. The idea is this would benefit wildlife by reducing habitat conversion to crops.

But wait, better crops will actually harm biodiversity!

But making crops more profitable runs the risk of increasing land conversion, via Jevon’s Paradox. Any time production becomes cheaper, supply and demand dictates that more land will be converted as it becomes more profitable to produce food. Even in poor, food-insecure societies, farmers are not so completely disconnected from markets that growing more food would be a profitless activity. Even remote jungle farmers trade extra produce. When crop productivity rises, agricultural pressure on natural habitat will rise. Therefore, improving crop productivity is not a good way to benefit biodiversity – although to be clear, it is still a benefit to human civilization.

But wait, better crops CAN help biodiversity (with regulations)?

There are cases where increases in agricultural output occurred alongside decreases in land conversion. This has happened when strong government regulation prevented additional land use (or converted land back to nature). It is a very unpopular position for governments to take, but it does happen:

Costa Rica: Intensifying cattle ranching and crops on existing lands combined with payments for ecosystem services allowed significant reforestation.

Vietnam: Increasing rice yields on existing paddies reduced pressure on upland forests when combined with strong forest protection laws. As a side note, Vietnam is now facing ecological troubles as a result of their water management for rice yields.

India: The Green Revolution (improvements in wheat/rice yields) decreased expansion into forested areas. Several forests were legally protected. Without these constraints, crop productivity gains would have been a force for accelerated expansion.

Malawi: This research found a decrease in deforestation as a result of increased agricultural productivity without external government pressure.

But even without regulations, better crops will help global biodiversity even as they harm locally??

Linus Lomqvist of The Breakthrough Institute argues: Once you take into account the global context, improved crops do decrease land use. This is because human civilization gets diminishing marginal utility from growing more food, so global food demand is relatively “inelastic” in economic terms.1 If you grow more food in one place, prices will decline and other places will grow less. Therefore, local productivity improvements increase local agricultural land conversion, but decrease land conversion globally.

But wait, this effect will actually backfire in Africa, harming biodiversity???

Unfortunately, in Africa, we will likely see increased global deforestation.2 Today, Africa is a very small portion of the global agricultural market; it also has extremely poor agricultural productivity. Increasing agricultural productivity there would greatly increase Africa’s market share - their (improved) yields remain well below the global average. So, improved African agricultural productivity would ironically cause global average agricultural productivity to drop precipitously, ultimately increasing global cropland.

But wait, better crops will ultimately help biodiversity even in Africa’s case:

Eventually,3 African agricultural productivity will rise to exceed today’s global average. The paper projects that African crop improvements will increase global land use for around 20 years before finally turning a corner and creating a net global decrease in loss of natural land.

Whew, that was a wild ride!
So, what conclusions can we take from all this?

For human flourishing, agricultural productivity improvements (especially in the world’s poorest countries) are always great. But for biodiversity purposes, it seems important to stay mindful of how local improvements will impact global average agricultural productivity.

Compared to helping African farming become just barely productive enough to become competitive on the global market (which would trigger a significant wave of habitat destruction), it seems important to aim to continue to improve productivity so they start being a positive influence on net global natural land (ending and then reversing the wave of habitat destruction) as quickly as possible.4

Similarly, one could seek to improve productivity in the nations that are already the most advanced:

• The Breakthrough Institute also found that boosting US crop technology by doubling R&D funding would spare land at an area:area rate of 1:10 (1.8 million more acres used in the United States would spare a whopping 18 million acres around the globe). Because it would be so much more efficient than agriculture everywhere else.

• Therefore funding the biggest successes to further improve might be a good idea. The actual situation on the ground is not straightforward though. A lot of US production is, for example, ethanol biofuel.

Introducing agricultural techniques is already a top priority around the world. It is a major focus of the Gates Fund, World Bank, etc. However, those programs are likely prioritizing crop improvements where there is the greatest human need, rather than where there is the greatest biodiversity threat. So there is potential for the deployment locations of crop improvements from a biodiversity perspective, despite the strong charitable emphasis.

Unfortunately, even with crop improvements, I am not sure that reducing agricultural land use is guaranteed. It’s possible people’s diets may change to prefer more meat. With animal meat, a much greater land area is required for production of each calorie. If instead the world ate less meat per person, it would greatly alleviate the habitat loss problems threatening biodiversity. (See the section on precision fermentation!)

Finally, here are some further considerations on this topic, which add yet more twists!

Global land use might not tell the whole story

We’ve actually already hit peak overall agricultural land, thanks to declining pasture land.
The land conversion threat to biodiversity is now decreasing globally, sorta. Decreases in world population growth, and increase in agricultural productivity have resulted in “peak agricultural land.” The productivity of pastureland (from improved breeds and grazing management) has outpaced the demand for livestock products. However this is pastureland, not cropland. Pastureland is in regions that are less productive, drier, and less biodiverse than cropland.5 Cropland on the other hand continues to expand, particularly in the tropics. So even though total agricultural area is now decreasing, habitat loss from agricultural land may still be worsening. This gets into land sharing versus land sparing, which is complicated.

Even if total land is decreasing, it may still be depleting soils, requiring fertilizer inputs, and moving to new locations. Therefore an unchanging total area of operational agricultural land may still threaten biodiversity. (see soil and pesticide shallow investigations!)

The question remains, where will land be spared as a result of crop improvements and local crop expansion? It stands to reason that it would be the least efficient crop land - which as far as I can tell are in desert, tundra, rocky, steep, and other marginally productive landscapes. These areas are unlikely to be high in biodiverse areas as they are not productive for biomass. On the plus side, they may be locations which support rare species, and are more likely to be adjacent to wilderness.

This seems like a good intervention with potential.

A lot of interventions can only work on one species at a time. In the recent set of interventions we talk about several single-focal-species interventions and explore how we would pick a species that has a big positive effect on many other species. This would amplify the effect and may bring it back to being a powerful intervention again. For example, rats are everywhere and impact a ton of species. If we had a “rat treatment” it would maybe have a big positive conservation effect even though it’s just for a single species. The impact distribution for invasive species is a pareto curve: About 100 species cause 80% of invasive-caused habitat degradation globally, so this intervention could scale well. In the links below, we discuss a few interventions regarding invasive species which have outside negative effects and ecological engineers which have outside positive effects on biodiversity.

Example Thought Experiment

Wild Water Buffalo (Bubalus arnee): This species creates aquatic habitat and wallows, as well as disperses seeds. Water buffalo tend to increase biomass, heterogeneity, and nutrient cycling. But it is endangered,* losing habitat, and located in Assam India which does not have local conservation support on the level of first world countries. A complicating issue is disease transmission to and from livestock.

*There are only about 4,000 wild water buffalo left, but there are 200 million domesticated water buffalo. These were domesticated 6,000 years ago, and thus would not be significantly phylogenetically different. Therefore I am removing it from the recommended keystone species as they are “already thriving.”

Ancestral wild water buffalo (4000 subspecies population) are most threatened by wetland habitat loss due to increased erosion and invasive species.

• Mimosa, Peruvian water primrose/ludwigia, and water hyacinth are actively converting land from marshland to shrubland. Biological control may be an option for each.
  

  • Weevils are highly effective against water hyacinth

  • Peruvian water primose/ludwigia. (see the Biological control section!)

  • but they are best combined with herbicides like 2,4-D or water hyacinth extract for mimosa using integrated pest management (see our pesticide intervention review coming soon!).

  • Barriers to water hyacinth control are the facilities to breed the biological control, their 90 day generation time and 2-4 year lag until peak effectiveness.

  • The periodic flooding of Assam may also make weevils ineffective in Assam.

It sounds like the increased erosion is coming from decreased water input which may be related to climate change or agriculture.

However, rather than try to keep them in place, there are locations where feral previously-domesticated water buffalo are thriving such as northern australia and the amazon floodplain. It stands to reason that endangered wild water buffalo could thrive there without needing caretaking, if we chose to use assisted migration (see assisted migration section!) to establish them there. Northern Australia would not be a good choice because the rare species there are destroyed by water buffalo. Historically there were only Diprotodons (giant wombats) which do not have the same kind of grazing pattern as water buffalo. In the Amazon historically there were Toxodons (rhino analogues) which is a closer match, and other larger grazers. None of these are equivalent to water buffalo, but they did likely create more grazing disturbances than the current equilibrium.

Ecological Engineer species can be a huge leverage point, where just a few animals can transform a landscape into a more bountiful, unique habitat. Some of the best cases are:

• beavers (and others that create unique aquatic terrain)

• oysters (and others that filter water)

• prairie dogs (and others that create burrows for other species to use)

• buffalo (and others that convert habitat through periodic grazing)

• mountain lion (and others that change herbivory patterns).

The reason ecological engineers are so powerful is that they create many microhabitats or shifting patterns of use. This results in unique combinations that support specialized species. One of the most impactful roles of keystone species is building persistent physical structures that other species obligately depend on, such as coral reefs. Dozens of species depend on and could not exist without these structures. For example: burrowing owls don’t actually dig their own burrows! Rather, they repurpose prairie dog warrens.

Ecological engineers often create disturbance regimes of particular frequency, intensity, and scale. Ideally we want engineers that push the landscape toward intermediate disturbance at multiple nested scales. Not too connected, not too fragmented. Not too stable, not too chaotic. It’s context-dependent, which is why ecosystem management is so difficult.

Creating aquatic-terrestrial interfaces (channels, dams, wallows etc) generally tends to boost biodiversity, because the water-land edge is inherently rich and many species require both habitat types in their life cycle. Beavers are the most famous example, but many other species have related effects. However, many of them are not good targets for additional conservation funding due to either being deadly (hippopotamuses, crocodiles), high profile (elephants, atlantic salmon, coral), or already thriving (american alligators, wild boar, nutria). Here are some species that may be good targets for conservation as important and disappearing ecological engineers:

• West African Manatee: Deepens river systems by digging out river bottoms and narrow channels, creating habitat for fish and passage for fishermen. They impose a lot of grazing pressure (including on invasive water hyacinth).

• Prairie Crayfish: Their unfortunate fate of being a crustacean means they do not get conservation funding despite their importance for nutrient circulation in wetlands, creating microhabitats, increasing water retention in soil, and creating burrows for other species. They are located in the United States so they may not be neglected.

• Lechwe: Semi-aquatic antelope that follow the flood cycles in massive herds.

• Mussels (many species): renowned as important ecosystem engineers via their filtration effects. But as an uncharismatic animal they are understudied and underfunded, especially in Asia and Africa.

West African Manatees also need improved water quality, protection from hunting, and assured access between freshwater and ocean habitats. Here is an action plan for their threats and an organization working to combat hunting which is their greatest threat across their range.

Mussels typically decline from pollution and sediment. Buffer zone regulations 20-100 meters for erosion control and pollution runoff are the typical prescriptions for water quality. Mussels can be re-introduced from areas they were driven out of.

Other keystone species will have unique conservation plans for their particular situation and needs.

Another key example are disturbance-dependent communities. Large grazers in grasslands, elephants in savannas, etc, cause disturbances like heavy grazing and migration trampling. Without these periodic disturbances, many ecosystems converge on one kind of forest or shrubland rather than a mosaic of different ages and types. We already covered the potential for this in the “Fencing Improvements and Passage for Wildlife” last month. There are several places for re-establishment of large grazers -- areas that are expansive enough to support true migration, not threatened by encroaching development, where conservation is supported by the government and locals:

• Buffalo are slowly gaining ground against the many hurdles from ranchers and land use disputes, see the American Prairie Reserve.

• Guanaco in Patagonia National Park is another example.

• Kazakhstan’s central steppe is another prime location, recently being rewilded with Przewalski’s horses and kulan (Wild Ass).

Overall, this intervention feels promising to me. It uses conservation concepts which are well established, but which I suspect are under-applied because the environmental movement is still averse to active interventions. The work on introducing keystone species is still focused on one or two species (wolves, bison, and beavers). There seem to be opportunities here that are tractable and highly effective, managing to aid many species at once with some small changes with more or less permanent improvements.

Control of invasive species is a common focus of conservation efforts. But first: is eliminating invasive species really a top priority? Invasive species seem unlikely to be major drivers of extinction compared to land conversion and resource exploitation. It is difficult to tell how much invasive species contribute to the extinction of another species, because almost every threatened species faces multiple threats (on average 2.5 threats).

As this Society for Conservation Biology paper noted in 2022, invasive species were listed as a primary cause in 6.8% of red list species on the IUCN. This is much lower than the #1 threat, habitat loss which was the primary cause in 71%, and lower than direct harvesting: 7.4%. Pollution and climate change were lower at 5% and 2%. Among species that have actually gone extinct, pressure from invasive species has been cited as a cause in less than 2%, of cases.

Oftentimes, invasives are not so bad, allowing many other species to continue to thrive alongside them. Sometimes they even fill ecosystem functions and ecosystems adapt to their presence. In this scenario, scientists tend to call them “non-native species” rather than “invasives”.

So, overall, invasive species do not cause large amounts of damage relative to other extinction threats. This makes it difficult to see biological control or fertility control methods ever becoming among the most important biodiversity interventions.

However, the impact distribution from invasive species is on a pareto curve: the worst 100 invasive species cause 80% of all invasive-caused habitat degradation worldwide. Therefore, individual interventions can be effective by focusing on the worst invasives.

Some attempts to control invasive species involve labor-intensive methods like trapping/hunting individual animals or spraying individual plants. But these don’t seem very scalable outside of individual niche applications (like ridding a single small island of goats or etc), which makes them unsuited for dealing with non-localized invasions. So, we’ll focus on biological control (leveraging natural predators / parasites to constrain an invasive species) and fertility control.

Even then, fertility and biological control as general strategies still have a scalability problem, since solutions must be carefully designed and are unique to individual species and particular locations.

Fertility Control means using long-lasting contraceptives to control the reproduction of species whose populations would otherwise balloon to ecologically unhealthy levels. Often these are species whose natural predators have been hunted away by humans, or invasive species that lack the predation, competition, or disease pressure they experienced in their original environment.

Fertility control is currently used on horses and deer in some places.

• Wild horses in the United States are typically shot with darts containing a fertility-control vaccine. Unfortunately, many horses have learned to stay out of shooting distance. And the fundamental strategy of shooting darts, while more humane than shooting to kill, is not much less labor-intensive. (Assateague Island’s population-control program counts every horse on the island six times a year!)

• Hastings on Hudson in the United States has trialled deer bait that contains contraception.

The Botstiber Institute (International Conference on Fertility Control for Wildlife), Faunalytics, and the Hastings on Hudson Deer Immunocontraception Project are all sources of information on this intervention.

Long-lasting contraception seems preferable to culling from an animal-welfare standpoint, and is also potentially less labor-intensive if cheap, scalable delivery methods can be found.

However, overall, fertility control interventions do not seem promising, because:

1. 2-3 years of immunocontraception is the longest duration achievable by most methods, due to biological mechanisms which break down doses over time. (Although there is 2019 paper showing effectiveness of one treatment for 5-10 years by using a slow release mechanism, so future progress might be possible here.)

2. Some species are not viable candidates for contraceptives:

   1. Feral cats are perhaps too beloved to be managed.

   2. Commercial livestock might eat contraceptive bait.

   3. Cane Toads and rabbits produce so many offspring that fertility reduction would need extremely high success rates to have an impact.

   4. Contraceptive control is not an option for dealing with invasive insects, fish, and amphibians. This is because of their reproductive rate, brood size, lack of hormone controlled fertility, as well as delivery methods not being able to reach a high percentage of the population.

   5. Plants and fungi also can’t be controlled via current sterilization techniques because they produce hundreds to millions of seeds/spores, and most sterilization is very coarse grain methods like “heat” and “chemical burning” that would damage surrounding species equally. Viral control methods are one of the few targeted methods we have.

3. In most instances, culling and poison are far cheaper and far more effective.

Gene drives have many similarities to fertility control via contraception, but have some important differences – they would be self-spreading, as permanent as we design them to be, and have very high rates of effectiveness. This addresses many of the current drawbacks of fertility control via contraception. Gene drives work by being a genetic snippet that has a virtually 100% chance of being passed on. See our previous coverage:

“Gene drive technologies show promise for eradicating invasive species, particularly mice on islands. The main coalition is called GBiRD for Genetic Biocontrol of Invasive Rodents. New Zealand and Australia have active research programs in this area. All of these gene drive efforts are still in research phases. The current technique is to have a gene that ensures all future mice are born sterile until the population crashes. If the gene escapes into other populations out in the world (a real concern with mice) the gene is preset to self-eliminate after a set number of generations, or could be tied to a mutation that only occurs in the target population. This would safeguard against causing lasting harm in mice elsewhere in the world.”

Overall, fertility control feels doubtful to be a scalable intervention. I really want it to be highly effective, but the current applications are really lacking. As far as I can tell baits are not currently an option and each species will need its own contraceptive program plus delivery mechanism. That said, I don’t understand why there is not more of a push for this, because it would be a far better replacement for the current methods. Gene drives seem uncertain to arrive anytime soon, (5-10 years?) although they would be more widely applicable once they do arrive.

Update Aug/2025: Conservation X Labs is now working on this!

Biological Control

Biological control is the introduction of natural predators usually from the native range of the invasive species. Control of invasive species using biological organisms. Using biological control gets a bad rap based on a history of very embarrassing and destructive failures in the 1800s. But modern projects are much more thought-through and have a good track record.

• Hawaii was the site of many infamous early failures of biocontrol, such as the disastrous introduction of mongooses (intended to prey on invasive rats, they also decimated many native bird populations on the islands). However, Hawaii today has one of the world’s most sophisticated biocontrol programmes, with a decades-long record of many successful introductions of new species to solve ecological problems caused by invasives on the island.

• In Canada and the USA, the Purple Loosestrife flower was successfully controlled by introduced leaf-eating beetles. Loosestrife is now controlled by 90% established leaf-eating beetles, and native species are recovering.

• On the Galapagos islands, the Cottony Cushion scale insect was controlled by the introduction of Vedalia beetles. Vedalia is pla

  ying a role in rescuing at least 17 endangered species in the Galapagos that were being harmed by the scale insect, and is used to control Cottony Cushion scale in many other places in the world.

The big upsides of biological control are that it can scale exponentially to match the size of the problem, and afterwards ideally becomes a permanent, self-maintaining feature of the environment, and dynamically responsive to the ongoing need for control.

Biological control tends to work best for plants and insects: these organisms often have very specialized predators / parasites / diseases that can be used to precisely target them. Conversely, introducing large carnivores to control animals rarely works as intended because large carnivores will eat many different species.

Biological control approaches might generally be a little easier than fertility control. Rather than creating a contraceptive and delivery mechanism, it involves finding a suitable natural predator. On the other hand, the predators need to survive in the new habitat and not become invasive themselves by preying on off-target organisms. Therefore, biological control methods require multi-year field trials.

Lasting “forever” makes this very worthwhile. It’s not horrible to keep using pesticides or hoping the habitats adapt to the invasive species, but even a very lengthy, expensive process to find a permanent self-maintaining solution is worth it in the long run.

Overall, this intervention feels effective and underapplied because it has a negative reputation despite more recent techniques being far more successful. Solving the problem more permanently than the other methods of invasive species control seems like a good way to redirect funds to more effective plans. It can’t work in all situations, but in many instances it seems like a good tactic.

• Species blocked by roads and fences are opportunities where manually moving species could unlock the species potential to take care of itself after the barrier is crossed.

• Another target are species that can’t move quickly like plants and small bodied animals

It may be a good idea to start moving species to distant locations they would never reach naturally if it benefits the survival of those species. See

• Gilbert’s Potoroo. It was abundant for an extended range along the southwestern coast of Australia but has become the smallest population size of any mammal in Australia. A population was artificially established on Bald Island, Middle Island, and in Waychinicup National Park. The native habitat was reduced to 15 individuals and a wildfire reduced that to 7 survivors. On the artificial transplanted island locations there are 100 as of 2018.

• The western swamp turtle, also in Australia has had 4 populations (83 turtles total) transplanted 300km to more suitable habitat. As of 2020 growth rates were normal for their natural habitat, although it will take more time to see if survivorship and reproductive rates indicate long-term success.

• Torreyey Pine, an evolutionarily distinct species has plans to be moved from Florida to North Carolina, a distance of at least 300 miles.

It seems like we can successfully do this for highly endangered species and/or species without futures, if we can put aside the “naturalness” (and fairness) of taking these conservation actions. Unpopulated islands are a limited resource, but the western swamp turtle is an example of simply finding suitable habitat, wherever it may be, and moving the endangered species to it.

Generally, addressing one species at a time is prohibitively resource intensive. So, assisted migration for keystone species would be the logical target.

Biodiverse mediterranean habitat can be especially drought threatened and especially evolutionarily distinct from gondwana relics. They would thus be good targets for multiple instances of assisted migration. For example, Proteas (flower group) of South Africa fynbos lives in nutrient poor soil that favors other gondwana relics and co-evolved gondwana invertebrates. The fynbos species are highly threatened including Proteas. In Australia there are Banksia species in a similar situation. Porteas and Banksia have important effects on their soil and environment which would be useful for other species to adapt as well. Banksia are important food sources for birds and insects. They can be grown at scale in facilities, but the facilities do need specialized conditions. They can be planted with 50%-75% success rates within the first 2 years.

A group that is working on this task is the Big Scrub Rainforest Conservancy in Australia. They are working with a Godwana rainforest remnant in eastern Australia. (A region with high evolutionary distinctiveness.) They are carefully monitoring the isolated remnant genetic diversity of endangered and keystone rainforest trees, and working on a long term plan to revegetate private land with these species. “...long-term sustainability for discrete, sustainable translocation recovery plantings of the genetically optimal threatened species.”

Assisted Migration has a few cases of win-wins. Coastal wetland can be migrated inland to preserve habitat that is often important nurseries for fish and bird migrations while also protecting people from storm surges. Coastal wetlands are extremely good for biodiversity as massive sources of nutrients and life stages. Techniques for doing this include adding sediment from dredging, planting vegetation to trap sediment, diverting inland streams to deposit silt, and simply removing tidal barriers to allow inland flooding.

For further reading, see this 2023 global evidence map of assisted migration.

Overall, this intervention feels likely to be effective, as long as it targets keystone species. I would not advocate for assisted migration as a generalized tactic.* It would be primarily keeping a habitat and supporting the many species that use it. Trees and other habitat modifying plants would be the obvious target. They are easy to move as a sprout and are important underlying fabric for future generations. It seems like something we should be willing to do more often than we currently are, once again because environmental groups are averse to active modification. This would be addressing both endangered species, endangered habitats, and having an outsized effect by using keystone species while also relying on already practiced techniques.

*unless I learn of a more efficient method than carefully moving individuals a few at a time, as seems to be the current practice.

Assisted Evolution is the act of improving the genetics of a species. Ideally we won’t be required to do deep research projects on each species’ genomes, one at a time. There are some routes to avoid that scenario:

1. We can work on keystone species that will help several others by their flourishing. There are projects already working on this technique with warm water corals!

2. We can work on systems that are shared across many species and are quite fundamental. For example, heat shock proteins, metabolic efficiency, and dna repair mechanisms. These are ancient and identical in many species, but better versions could be found (in other species or by us) and then inserted into many species with replicable and scalable methods. We are exploring this with rubisco in crop plants. We would be using this knowledge to prevent extinctions and increase flourishing in nature.

3. The third way is to improve the microbiome - gut bacteria, soil bacteria, etc. Doing so can improve nutrient availability, disease resistance, and environmental health. This method is less invasive to individual species’ genomes and more easily deployed by distributing different microorganisms into the environment. It also avoids some favoritism. We are already doing this for coral known as “BMCs” or beneficial microbes for corals without modifying them beforehand.

This write up from Yale explores assisted evolution from teaching bilbies to avoid predators, to the manipulation of American Chestnut to resist chestnut blight.

All of these technologies are far away from deployment. Some are in research phases, others have identified the genes, very few have edited species, and none have been released into the wild. Blight resistant american chestnuts have been a program since 1980s. Gene editing began in 2006, in 2012 they found a gene that worked. By 2019 they had a desirable tree line tested for safety and in 2024 is their latest application for deregulation. ~ 10 years since the gene was identified and years to identify the gene. The cost of the program is at least 6.8 million since 2012.

Warm water resistant corals have at least $4million of funding in 2016 at the Oppen lab. 8 years later they are now testing their corals in the wild on the Great Barrier Reef.
15 years of work at CSIRO has produced corals with higher temperature resilience after a 4 year of lab testing. There was $6 million committed to CSIRO’s coral heat tolerance research in 2018. Australia has overall committed $2billion to “reefs” over the next decade, a smal part of which includes CSIRO’s work. Next steps are outlined in this 2024 paper, and mainly involve deploying and testing in the wild for several years tracking uptake of coral symbionts, scaling up production (while keeping genetic integrity and affordability), and eventually global deployment.

Overall, this intervention feels like it is not an immediate solution, but a long term necessity. Biotechnology of this calibre is still a ways off, and then applying it to wild species is yet further away. Fundamentally changing basic cellular processes seems unlikely to be easily improved over time-tested versions by evolution, but I would estimate that there are some opportunities. Then deciding which species to apply this to is a moral dilemma. Overall this feels really important to address as the caretakers of the earth, but it’s difficult to see any immediate or clear paths forward in this area. Long term, this needs to be pursued.

Following up on our Big List of Environmental Interventions, let’s take a closer look at the top dozen or so. These are shallow dives, trying to get a picture of each intervention’s track record, potential, and current bottlenecks. These preliminary research summaries allow us to rank them for further research or discard them if we find intractable critical problems.

Characteristics of a good intervention

A good intervention should score highly on an importance, tractability, neglectedness framework.

Importance:

• The best interventions should have large positive effects on global biodiversity. Interventions should help many species of great evolutionary distinctiveness and increase ecosystem complexity (rather than a simplified system).

• Frequently this will mean addressing the biggest causes of biodiversity loss: land use and climate change.

Tractability:

• The best interventions should be highly scalable -- absorb a lot of funding and benefit from economies of scale as it’s deployed in more places, rather than being bottlenecked by factors like training, bureaucracy approval, or manufacturing capacity.

• High leverage is valuable. Influencing major ecological tipping points to convert a small amount of effort into a huge positive effect might be one of the best tactics.

• Ideally we should see improvement in a few years rather than needing decades before benefits appear. (But it is most important that the intervention is eventually self perpetuating, or otherwise long-lasting, rather than requiring constant effort to maintain the benefits.)

Neglected:

• The best interventions to direct our marginal effort should be neglected - diamonds in the rough with undiscovered potential, not projects that already have lots of funding and attention.

• Some of the best interventions will probably come from a “hits-based” philosophy, where the rewards are uncertain but huge upside is possible.

No single intervention will fit all the criteria, but we want to get as close as possible. Scaling up and across different parts of the world, different ecosystems, and different cultures in different countries is particularly difficult with conservation. It is also difficult for anything to persist for a long time in nature, which by its very nature is in a perpetual state of flux.

Funding and Increasing Enforcement of Protected Lands:

Conservation efforts often create “paper parks”—protected lands that exist in name only due to insufficient funding for enforcement. What if we just increased funding so that parks could combat poaching in their borders?

Unfortunately, increasing enforcement is not as simple as “more funding,” or “more people/guns/cameras.” The fundamental challenge is that enforcement agencies can never match the local community’s intimate knowledge of the land when residents are determined to circumvent protection measures.

Poaching, driven primarily by subsistence needs rather than trophy hunting or medicinal purposes, remains the greatest direct killing of biodiversity. Since subsistence hunting stems from basic survival needs, the most effective solution lies in addressing the underlying issue: human development. When local communities have alternative livelihoods and food security, the pressure on protected wildlife often releases. Unfortunately, “solving poverty” is often even harder than protecting wildlife. It is important to understand the context of enforcing protected areas, and to pursue human development as a way to improve biodiversity protection.

Preventing land conversion to agriculture is another one of the most important drivers of biodiversity loss. Modern satellite imagery gives early warning of forest clearing after just a few days. Enforcement is important for park boundaries, stopping early habitat development before it becomes an industry. Still, enforcement usually needs community support more than it needs funding. See the next section on using microphones to detect and prevent illicit logging, which also faces this issue.

Microphone Coverage to Detect Illegal Logging

One of the greatest causes of biodiversity loss is the logging of tropical forest, particularly the Amazon. It is a problem in many tropical forest parks. Surprisingly, logging is primarily driven by soybean profits, not profits from lumber. Individuals log the protected forest undetected, sell the wood, and grow soybeans. The soybeans deplete the soil, and the farmers then sell the land to ranchers who raise beef cattle on the logged land. Ultimately the land is thus taken up mostly by ranchers -- but the mechanism driving the process is soybean farming and lack of enforcement. Patrolling the entire park is out of the question, and covert logging often takes place before the park managers notice. After the forest is logged, the damage is done.

Logging makes a lot of noise, so one approach is to monitor the forest by using small continuously-recording microphones. The microphones would ideally send real-time alerts when logging is detected; authorities can then investigate and stop the logging before it gets completely underway. Microphones have become much cheaper, and advances in AI allow them to become more accurate at detecting the sound of logging. Finding lots of microphones in the jungle is nearly impossible, which makes destroying or jamming them nonviable. These devices would greatly increase park enforcement power and greatly discourage logging.

The current limits to these microphones are: battery life, connectivity, and environmental damage. Current generation field microphones need to operate longer and more reliably to be practical. They also need a way to broadcast alerts in real-time.

1. Battery Life: Current field microphones can operate for 6-12 months on their batteries. Reaching these remote locations to replace the batteries severely limits their usefulness. Most batteries aren’t produced for the long-term, low-power applications required for remote forest deployment. The few potential advancements on the horizon are not yet commercially available at practical prices. Solar panels aren’t a good option, either, as the forest canopy blocks the vast majority of sunlight. Software improvements could make the alerts more accurate, improving “wake up” and transmissions by reducing false positives, effectively making batteries last 15-25% longer.

2. Failure Rates: Microphones also often fail. They get knocked down, water damage, overheat, or animals interfere by chewing them, knocking them over, blocking up their ports, or otherwise impairing them. Many field microphones fail for one of these reasons.

3. Connectivity: Today’s field microphones only record data, to be physically retrieved when workers visit the devices months later. This is fine for studying patterns of illegal logging, but obviously no good for active prevention and law enforcement -- real-time data transmission is needed. But there are no cell towers in much of these remote wilderness regions, and trying to relay information via satellites would take too much energy for these extremely low-power, battery-operated devices (especially given that the forest canopy blocks a lot of reception). The best options would be radio (via technologies like LoRa) or using cell tower coverage in the few places where it is available. Satellite connection is expensive and connections to satellites require large powerful broadcasting equipment. Rather than stationary microphone complete 24/7 coverage, a roving drone flying program with 2-3 drones extending the range of connectivity/monitoring would perhaps be a great and cost effective improvement on the current patrol systems. These are being deployed in Brazil by the WWF and the Rainforest Foundation.