This post was written with Claude's help for background research, as well as some structure & phrasing tweaks.

What Has Nature Done For Me Lately?

In this sequence I try to determine the most valuable aspects of nature, after observing that lots of people find it valuable. One of the major candidates is “ecosystem services,” a term describing benefits we get from nature.^[1] I find that the total value of all ecosystem services is high, but the marginal unit wouldn't meaningfully diminish the benefits we receive.

In this post I explore 

• How valuable ecosystem services are (about $45 trillion/year)
• What those ecosystem services are (nutrient cycling, etc)
• How possible it is to get those same benefits from other sources besides nature
• And whether biodiversity is providing the instrumental gain we get from nature. 

A quick clarification: Biodiversity is a measure of how many distinct species exist in nature (although it ambiguously can refer to local density of species, total extant global species, diversity within a species, and/or variety of ecosystems). Biodiversity is not equivalent to ecosystem services - as I demonstrate in this post.

Breakdown of Ecosystem Services

The estimated dollar equivalents of global benefits people obtain from ecosystems are around $45 trillion ($33T, $42T, $44T, $125T, $150T). Many environmentalists treat this as “case closed, nature is as valuable as the entire human economy, therefore modern conservation spending is obviously justified.” But there are several flaws with this argument. 1) The next marginal piece of nature might not be valuable. 2) Ecosystem services might be easily and cheaply replaced. 3) Biodiversity might not be important to continue providing the ecosystem service. Take a look at this table to see for yourself how the situation pans out: 

This changes the picture somewhat, and challenges some assumptions I had. 

• We do not appear to be totally dependent on ecosystem services to survive, as is often stated or implied.
• We have a lot of functional replacements for ecosystem services, especially when there are efficiencies at large scales.
• Dependency on ecosystem services seems to be concentrated in low income countries.
• Only a handful of species are necessary to provide each ecosystem service. One or two services might rely on high biodiversity to provide benefits.
• There are some categories where ecosystem services are still our best option: water regulation, coastal flood control, wetland water purification, and fishing primarily.

According to the 6 global environmental assessments, the most major ecosystem services are water regulation, nutrient cycling, climate stabilization, flood control, and soil formation. Pollination, disease regulation, and pest control are commonly cited as important ecosystem services and linked to biodiversity.^[3] 

The Value Estimates In This Table Are Flawed

First a word of warning: the value estimates in this table are not very good. The methodology used to generate them is a shot in the dark from 1997. There are articles showing these estimates are inaccurate. I am using this set because all six of the other global ecosystem service assessments (see references) determined that it is misleading to provide dollar estimates. Many services interact in non-linear ways and tallying up value doesn’t make much sense. 

Examples:

• A lot of ecosystem services are preventative. They don’t provide value until they prevent a massive disaster, and their value depends on what they are protecting. So it doesn’t make much sense to add up flood control everywhere there are wetlands. Yet where those wetlands do protect people, it's immensely valuable.
• Every bit of wetland lost increases the value of the remaining wetland.
• A lot of ecosystem services are provided, but not used.
• Water regulation and flood control would be double counting the same wetland.
• It's really hard to estimate the aggregate value of rare events, like pests in a drought year. It’s not even clear if pest control is an ecosystem service, or pests are an ecosystem disservice.
• Dollars are substituteable. Nature is non-substitutable. You can’t convert everything to the most valuable ecosystem service to max out your returns.
• When attempts have been made to estimate ecosystem services, the error bars have been larger than the estimates. 

But I think it is useful to have a very very rough sense of the size and ratio between ecosystem services globally, and this highly questionable Costanza 1997 data is what we have to go on. I don’t think it’s too important the exact dollar value or even relative dollar value. I think it’s safe to conclude that they are very valuable, but only some of them are irreplaceable.

Some Other Ideas for Measuring Value

The FAO reports 4.3 million tonnes of lost productivity due to land degradation (primarily nutrient loss and salinization). This seems to roughly verify that ecosystem services provide a lot of value.^[4] 

Here are a couple such examples: 

The planning committee for New York city compared constructing a water filtration plant versus  purchasing land upstream to naturally filter the water. They found it was far cheaper to obtain the same service by maintaining ecosystem services.

Using the ESVD (ecosystem service valuation database) accounting, ecosystem service values are stacked on top of each other. The original paper for this method is de Groot from 2012. I do not endorse these estimates, but include them to represent some common claims regarding the value of ecosystem services. They helpfully show the breakdowns of which ecosystem services provide the most value, and how ecosystem services might outperform land exploitation.^[5]

A more accurate representation comes from Strand 2018 which does spatially explicit accounting. It isn’t complete - it is only looking at 4 out of 17 identified ecosystem services. It shows 65% of the Amazon forest is producing value <$17/ha and 12% is at least $57/ha-$737/ha.

This is all to say that ecosystem services provide a lot of value, and there are some places where we are actively undermining ourselves by converting land that is more valuable when it is providing ecosystem services.

Option Value

Another common counterargument is option value. What about all the uses of nature we haven’t discovered yet? How many species will turn out to be instrumentally valuable to us by the end of time? 

Probably this is indicated by how many are valuable now and extrapolating to hypothetical future uses. Based on the table of ecosystem services above, that would be about 1% of species.^[2] A report on Climate Change and Longtermism gave an arbitrary estimate of 10%. I’ll use 15% as a safe demo estimate. So to estimate future needs for future purposes, maybe 3x as many as we value now, plus room for error. So, using this logic, we would preserve somewhere between 5-60% of species for their option value. I think this shows how option value is not a valid argument for protecting 100% of species, but it does indicate we should protect more than our currently known useful species.(maybe 4x?)

A Case for Biodiversity as a Biotech Resource Library

I can think of an exception, where biodiversity^[6] is instrumentally valuable. Healthcare is huge (~500 billion), industrial biotech is also large (~300 billion), and synthetic biology is going to be an ever-growing field. The question is whether biodiversity as an “evolution-tested library” will be valuable for bioscience development and to me the answer looks to be “Yes.” 

This is a clear case where it is reasonable to suggest that biodiversity itself will be useful for biotechnology development and it will remain so for a long time to come even with information extraction. I think this holds true, even accounting for highly capable AI decrypting and designing synthetic biology on its own.

I invite you to skip ahead to the verdict on ecosystem services in the conclusion, or continue reading if you are interested in my justification for biodiversity as valuable for biotech development.

-

Protein discovery from other species does indeed translate over to useful information for our own species’s healthcare. Furthermore, information from large bodied animals is useful in ways that single-celled organisms can’t contribute, so multicellular biodiversity would be valuable. We still need a LOT more bioscience development, even with protein folding mostly solved.  (eg spatial transcriptomics, protein/enzyme prospecting for conformational stability/protein profiling, purifying via chromatography, epigenetics, structural proteins for industry application, precision targeting of receptors/ion channels and metascience expanding biotech capabilities) It looks like these can all be informed by biodiversity prospecting.

More specifically, being able to feed “genetic large language models”^[7] with lots of reference genomes and proteins seems likely to be very useful. Naturally occurring proteins have been useful for biotech in a variety of cases. Venoms seem to be especially useful, due to their highly specific targeting, bioactive potency, unusual structure, and high stability. Notably poisons are found across the taxonomic tree - reptiles, amphibians, insects, and plants.

Another way biodiversity can be useful is with highly selected phenotypes. These can make biological pathways stand out. Because biological systems are top-down and bottom-up simultaneously, these systems are very complex and we still require a lot more data to understand and manipulate them. There are also a lot of valuable things we might be able to copy. Theoretically we could learn to prevent cataracts by copying hawks, prevent metabolic disease by copying hibernating bears, prevent pain from naked mole rat biology, get antibiotics from crocodiles and vultures, and repair organs by copying axolotls. Most phenotypes are not going to translate to our biology, but I expect some pathways we will be able to recreate for our own treatment. So I do think that 1) biotech has many wide applications, and 2) can benefit from biodiversity prospecting in a variety of ways. 

A few major caveats are that 1) this does not support the preservation of wild biodiversity; only the preservation of small populations of living organisms, 2) is we won’t need to keep biodiversity around after we’ve gathered all their information. My counter point is that there is a lot of bioscience development to go^[8] and I’m pretty confident it’s going to take a long time to gather this information (if only because observing biological interactions seems arduous). I think biodiversity will remain valuable for a long time.

We also don’t need 100% of biodiversity to gain this information, just the subset of the most richly informative species. While we have some good ideas of what species will be fruitful (venomous, mammals, extremophiles, evolutionarily diverse, etc), repeat examples are useful for training models and biotechnology is useful in a wide variety of applications. For example today we can identify antifreeze proteins well, in part because they occur similarly in a variety of species. (Antifreeze has applications in food preservation, cryopreservation, cryosurgery, and anti-icing materials.) So, maybe 50-70% of all species would be useful for training data?

The main takeaway is that I think biodiversity will be useful for biotechnology development and it will remain so for a long time to come. Even accounting for highly capable AI decrypting and designing synthetic biology on its own, I think biodiversity will remain useful for the foreseeable future. 

Or I could be wrong - there might be some basic principles that underlie all species. 

Conclusion 

From the above exploration, it seems that 

1. Biodiversity is mostly not required for ecosystem services.
2. Ecosystem services can be substituted for, in most cases.
3. The exceptions are water regulation, coastal flood control, wetland water purification, and fishmeal.
4. Value from ecosystem services is concentrated in places which cannot afford to replace them.
5. The value ecosystem services provide is real, and large, but the marginal loss of nature doesn’t reduce ecosystem services meaningfully.
6. Biodiversity may be highly valuable as a biotech research library to train medical LLMs on. Or it might not.
7. Instrumental value from nature does not really justify conservation priorities or costs.

However, it seems clear that people would not be happy to replace wilderness with infrastructure providing the same outputs. So, if not ecosystem services, where does all the value that people see in nature come from? I go on to explore x-risk in the next post, but I think biodiversity is mostly valuable for longterm flourishing reasons. Which I will explore in the subsequent post.

This is part of a sequence on where the value of biodiversity comes from:
1) People really care about something they call nature
2) Ecosystem services don't fully explain it 
3) Ecological collapse is not an x-risk 
4) The real value comes from long-term flourishing
5) Conservation in the next century is going to look a lot different than environmentalists think

 

This sequence is being written as part of a project for EcoResilience Initiative, an EA group focused on biodiversity and ecosystems.

Appendix:

Sources for my Claims about Ecosystem Services

Here I go through each of these ecosystem services and give some indications of where the information in the table came from. I start with the most valuable ones and go over their marginal value, replaceability, and relationship to biodiversity. 

Nutrient Cycling

Nutrient cycling is more than 30% (~$13 trillion) of some of these ecosystem service valuations, so let's look at nutrient cycling. The main nutrient cycles^[9] are nitrogen and phosphorus. 

Nitrogen

Famously, <1% of plant species provide nitrogen (~2000 total) for wild habitat. For human purposes, nitrogen is currently produced using the Haber–Bosch process uses natural gas and 1-2% of global energy consumption.^[10] Alternative realistic natural sources of nitrogen are from manure, sewage, and plants (legumes). These sources are a tiny percentage of commercial food production because they are bulkier to transport, contaminated, and don’t release the nutrients as quickly as pure NH₃. Nitrogen provision is a valuable service but a small set of species provides it for the globe, and this is not relevant to most human agriculture as we already have a more efficient non-natural replacement. ‘Biodiversity’ isn’t a provider of nitrogen.

Phosphorus

Phosphorus is mined. 70% of known global reserves are in Morocco. Other sources are bones, fish meal, manure, and sewage. None of these are as cheap and scalable as simple mining. Crystallizing phosphorus from wastewater is about twice as expensive ($750 vs $400) as mining it. (magnesium ammonium phosphate/struvite generated by injecting magnesium into the water treatment process)^[11] Harvesting it from plants fundamentally doesn’t work because they can only gather it from the environment, not generate it. And even producing phosphorus biologically, if we had to, would not be done by preserving biodiversity. It would be produced at scale by one or two highly-bred, top performing species.

Water Supply 

This means freshwater provision like groundwater recharge. It seems that about 20 species generally would suffice, by reducing evaporation, slowing runoff, etc. For reference, ecosystems have somewhere between 50-150 species/ha for a “typical” habitable environment.^[12] Which 20 species will change from location to location. And resilience over longer time frames would mean a few more species. But fundamentally, very few species are necessary for water regulation. 

Pest Control

On massive scales it is cheaper, more reliable, and more efficient to have a tightly controlled, hyper-productive monoculture with pesticides than to rely on natural pest control like bats, birds, “good” insects etc. 

Something these examples have in common, is that biodiversity is more efficient than the manual replacement for small-scale users. But once you massively scale up production, the ecosystem services become less efficient than manual provisioning. 

The required species numbers are pulled from thin air after getting a sense that diversity does matter for biological control. Since I think pesticides replace them at scale, I didn't make time to find a better estimate.

Wild-Caught Fish

Fishing is one ecosystem service that involves hundreds of species, at least in the tropics. However, my best guess is that this is less about desire for a variety of fish and more that these markets sell all edible catch. Therefore I revised the “necessary” species down from 100s to 20-50 with the rationale that if there were fewer species at the same levels of abundance, this would not pose a negative effect on the fishing industry. As it is, farmed fish is starting to supersede wild-caught fish.

For value I'm using FAO data and then adding more for subsistence fishing.

Pollination

While many species of wild bees do pollinate crops, it is more efficient to bring in one species (European honeybees) for a few weeks. It is a much more efficient process than dedicating a large swath of land for bee habitat beside every crop on every farm across the nation. 

Disease Control

Decreased transmission of disease has been linked to higher biodiversity. The proposed mechanism is that higher biodiversity decreases the frequency of encounters with disease carriers. If you have 30 species of prey to eat, you are less likely to eat the same pool of infected prey as often. This is called the dilution effect. It is hotly debated. A competing hypothesis is that the sorts of species that make good disease hosts are the sorts of species that survive and thrive after human-caused biodiversity loss (“strategies that favor growth, reproduction, and dispersal, over defense against parasites”).

Another explanation is human-wildlife contact increases with habitat destruction. This is called the spillover effect. It doesn’t necessarily imply less deforestation is better for human health, but that maintaining distance from deforested areas is healthier.

Flood Control

We build levees, dams, and canals to control floods. We do so because these work as well or better than natural areas. (Yes we lost a lot of benefits of natural flood control, but we gain greater benefits, else we would tear them back down.) We also gain fine-grained control over water levels throughout the year. In natural flood control we typically plan for 5 fast-growing tree species to anchor banks, 5 more shrubby species further back, and perhaps another 5 higher up on the bank to reduce flow speed and erosion. So about 15 species achieve flood control per region. 

Water Purification

The typical tradeoff for water purification is that chemical and physical processing takes more input and produces more toxic byproducts, but is faster, more reliable, and can process more types of waste than nature-based solutions. Typically the mechanical solution is cheaper in the short term, but on long enough time horizons the nature-based solution is cheaper. Nature-based water purification also has co-benefits in wildlife habitat and recreation zones. 

Major Assessments of Current Value from Nature:

Recent Assessments:

IPBES 2026:

The biggest source of measurement is IPBES. They recently released an assessment of the dependency of business on biodiversity. (Feb 9 2026) https://www.ipbes.net/bba-report/media-release, https://www.ipbes.net/business-impact 

The first global assessment was completed in 2019, the second is planned for 2030. 

https://www.ipbes.net/global-assessment

(There are also regional assessments - Africa, Europe & Central Asia, Asia & Pacific, and Americas.)

Global Canopy 2025:
Global Canopy report on the damages by synthesizing 360 reports in 2025:

https://tnfd.global/publication/evidence-financial-effects-of-nature-related-risks/

The Dasgupta Review 2021:

Which attempts to calculate the value of nature: https://www.gov.uk/government/publications/final-report-the-economics-of-biodiversity-the-dasgupta-review

Boston Consulting Group 2021:

More recent synthesis estimates: Around $150 trillion/year is now commonly cited, roughly 1.5x global GDP. This comes from TEEB and BCG in 2021 (Boston Consulting Group)
https://www.bcg.com/publications/2021/biodiversity-loss-business-implications-responses

Swiss RE 2020:

There is also the Swiss RE which values nature at 41.7 trillion according to how reliant the economic sectors are on nature.

https://www.swissre.com/institute/research/topics-and-risk-dialogues/climate-and-natural-catastrophe-risk/expertise-publication-biodiversity-and-ecosystems-services.html#/

https://www.swissre.com/media/press-release/nr-20200923-biodiversity-and-ecosystems-services.html

World Economic Forum 2020

World Economic Forum (WEF) analysis suggests that USD 44 trillion of economic value generation – just under half the GDP of the world

Older Assessments:

Historically there is the one by Robert Costanza: 1997 paper "The value of the world's ecosystem services and natural capital" which estimated $33 trillion per year. He updated in 2014 with $125 trillion per year.

Millennium Ecosystem Assessment 2005:

Longer ago (2005) there was the millennium ecosystem assessment: https://www.millenniumassessment.org/en/index.html But this has been superseded by the IPBES.

The Economics of Ecosystems and Biodiversity:
TEEB has been working in parallel and has produced rough estimates of various ecosystem services 2010, some ongoing 2020: https://www.teebweb.org/seea-natural-capital-accounting-and-valuation-of-ecosystem-services-project/

https://teebweb.org/publications/teeb-for/national-and-international-policymakers/

https://wedocs.unep.org/items/1c5c3435-03c8-4cc5-94d6-8e63d885e9fa

Other Assessments:

UK National Security 2026:

Recent National security report on biodiversity loss for the UK - 

https://www.gov.uk/government/publications/nature-security-assessment-on-global-biodiversity-loss-ecosystem-collapse-and-national-security

Wildlife OU (ongoing):

Wildlife OU is creating a calculator for businesses to measure the impacts of wildlife value. 

https://www.endangeredwild.life/

1. ^{^}

    Technically ecosystem services include ALL forms of value including cultural and aesthetic, but for this post I treat ecosystem services as only instrumental benefits. Several of the patterns I describe apply to both instrumental and cultural benefits. Eg:, most indigenous cultures have about 300 species important to their culture which works out to about 10% of species. 

2. ^{^}

   For reference, ecosystems have somewhere between 50-150 species/ha for a “typical” habitable environment. (This goes down to 25-100 for “hostile” environments like taiga and desert. The lowest biodiversity goes to about 10 species per hectare (salt flats, invaded prairie, hyper arid desert) and the highest are 300-500/ha (tropical rainforest).) To get a rough global estimate of required biodiversity for ecosystem services, I multiply each species required count by the number of global biomes (85-500) and get ~5,000-100,000 (0.01%-1% of total extant species). Total biomes are somewhere between 85 (at the Formation level, which would be minimizing global species) and like 500 (at the Division level which would be using locally thriving species). Totaling that up comes out to somewhere between 7,480 (0.086% out of all extant species) and 81,500 species (0.94%) to support global ecosystem services. Flaws with this estimate: There are unlisted ecosystem services. I’m overcounting habitats since biomes like “ephemeral salt lakes,” “ground water” and “mud plains” would provide more limited sets of ecosystem services. Some species could fulfill multiple ecosystem services. Some species could fulfill multiple biomes. Supporting ecosystem services would be needed. I think this estimate is really low and use 15% of extant species as a generous minimum later on in this article.

3. ^{^}

   For further reading, see What Has Nature Ever Done For Us? by Tony Juniper and this article by Zero Carbon Analytics

^{^}

Another indication that ecosystem services are valuable is estimating the increase in destruction from their degradation. A napkin estimate I came up with in 10 minutes is adding up the creation of malaria-carrying-mosquito breeding habitat (100,000), excess deaths from lack of nutrients due to pollinator decline (86,000), zoonotic spillover (10,000), flooding from mangrove loss (2,000), flooding from forest loss (100,000), and get something like at least 300,000 deaths from lost ecosystem services. 

^{^}

According to the paper, inland wetlands are worth $26,000/ha mainly due to water regulation ($6000/ha), disturbance moderation ($3000/ha), waste treatment ($3000/ha), erosion prevention ($3000/ha), and recreation ($2000/ha). Tropical forest is $5000/ha which mainly comes from climate regulation ($2000/ha), recreation ($1000/ha), and medicinal resources ($2000/ha). Another paper using roughly the same accounting method placed Iran tropical forest at $100,000/ha which mainly comes from water supply ($41,000/ha), raw material ($10,000/ha), recreation ($45,000/ha), and existence value ($3000/ha). In contrast, Iran temperate forests are $5000/ha in this paper, mainly from air quality ($1000/ha), climate regulation ($500/ha) and existence value ($2000/ha) Badamfirooz 2021. These compare with about $1,250/ha for palm oil,# $500/ha for soy, $75/ha for cattle ranching - which are common causes of tropical deforestation. Again, I don’t endorse the ecosystem service valuations written here. I include them to represent the argument that ecosystem services provide non-monetizable value that in aggregate is more valuable than private profit capture from the land.

^{^}

 as in global total extant species, not number of co-occurring species

^{^}

 Not the name I would have picked.

^{^}

 Sequencing their genomes, copying their spatial transcriptomics, profiling their proteins, understanding their epigenetics and whatever other bioinformatics we want to extract…

^{^}

Carbon, oxygen, and water could also be counted, but are usually considered a separate category or are not counted at all. The marginal value of a little more or less is essentially zero. Services like oxygen production are fundamental, but they don’t really make sense to value unless events unfold well beyond realistic scenarios. And carbon cycles are covered far better elsewhere, so I don’t go into it.

^{^}

First nitrogen is harvested from the air by cooling any air to a liquid and then boiling the nitrogen to separate it, while the oxygen remains liquid at -196C. H2 is generated from natural gas by reacting with steam at high temperature over a catalyst, followed by a “water-gas shift reaction” to get more H₂.Then the nitrogen N2 is combined with hydrogen H2 at high temperatures 400-500C and pressures 150-300 atm to create bio-active ammonia. NH3. CO2 is the main waste product, and is responsible for 3% of global carbon emissions.

^{^}

Trying to use hyperaccumulator plants and then fertilizing crops fundamentally doesn’t work because the hyperaccumulator plants would need to be mopping up high concentrations of phosphorous, which is just dispersed mining. The phosphorous source probably wouldn’t last at global scales. (unless perhaps it’s part of a global cycle, which appears to be the case in the Amazon - from Sahara dust and smoke from other parts of Africa) Then transporting and depositing the compost as fertilizer would be bulky and inefficient. It's a remediation tactic, but not a viable industry.

^{^}

and down to 25-100 for “hostile” environments like taiga and desert. The lowest biodiversity goes to about 10 species per hectare (salt flats, invaded prairie, hyper arid desert) and the highest are 300-500/ha (tropical rainforest).

Show all footnotes