Diversity as Infrastructure of Farms and Forests in the Tropics

Working on climate mitigation and adaptation resembles herding chickens. The birds remain together while the enclosure holds. A single gap releases all of them at the same moment, and no amount of subsequent effort gathers them back into a group. Tropical land policy kept two birds in the same enclosure for three decades–production and protection–, and holding those two consumed most of the political attention available. A third arrived with the restoration agenda, which a country can satisfy by planting one exotic species in rows and still report as fulfillment. A fourth arrived with the energy transition, which is necessary and welcome and which lands on a country where mining claims already overlap 370,000 km² of intact forest, the largest overlap of its kind anywhere in the world (Grantham et al. 2021). Four claims now move through a single enclosure, and holding them together requires a way to see all four in a single frame at the resolution where they collide.

The oldest pair of claims is agriculture and forest protection, treated as competing demands on the same land. In the field, the competition is material, because every hectare converted to soybean or pasture is a hectare withdrawn from native vegetation, and every hectare of legal reserve is an area withheld from immediate revenue. In political speech, the opposition hardens into identity, with rural caucuses and environmental movements each reading the other’s advance as its own retreat. Both versions rest on the premise that the two systems are separable. Three decades of measurement in the Brazilian Amazon and Cerrado have shown that they are joined: the water that sustains grain yields in the Mato Grosso state is produced by forest standing elsewhere. Recognizing ecological diversity as productive infrastructure is therefore the condition on which reconciliation of the two agendas depends. That recognition acquires force only once the infrastructure is measured at the scale on which it operates.

THE COMFORT OF THE SHORTEST PATH

Reconciliation has not happened because it is costly for almost everyone who currently profits. If the task were straightforward, it would already be complete, and the international debate would have moved to the deeper work of mitigation and adaptation to a warming driven primarily by greenhouse gas emissions from fossil fuel combustion. Every sector enters the transition intending to win, and to do so through the business model it already operates. Unruh (2000) described the mechanism as carbon lock-in, in which technological systems and the institutions built around them co-evolve into a configuration that resists alternatives, even when those alternatives perform better on the metrics the system claims to value. Lock-in operates on intellectual authority as well as physical assets. Planck (1950) argued that a new scientific truth establishes itself once the generation trained in the previous one has died. Azoulay, Fons-Rosen and Graff Zivin (2019) turned that claim into a testable design, using the premature deaths of 452 eminent life scientists as a natural experiment on the subfields each had dominated, and found that entry by researchers who had never collaborated with them rose markedly, drew on a different body of literature and was disproportionately likely to become highly cited. Their analysis of the mechanism revises the aphorism that motivated it. Almost none of these scientists sat on funding panels, the deterrent operated through the anticipated cost of contradicting a prominent figure, and entry concentrated in subfields whose citation base already reached beyond itself. Permeability to outside literatures is therefore the variable that governs whether new arguments arrive, and permeability is a property that institutions can build deliberately. Delay in accepting that standing forest is an input to agricultural productivity accordingly has a component lodged in how far the relevant fields reach past their own reading lists. Resistance of this kind is directed at the entry of an unfamiliar variety into a settled arrangement, and the case for admitting that variety begins with what a system loses when it holds only one kind of response.

DIVERSITY CARRIES THE LOAD

Diversity is the mechanism through which systems retain function under disturbance. Ashby (1956) established the general form of the principle in cybernetics, showing that a regulator can absorb only as much disturbance as its own internal variety allows. Ecology reached the same conclusion by measurement, through the insurance effect, whereby communities holding species that respond differently to the same perturbation sustain function when any single response fails, a result that Oliver, Heard and Isaac (2015) report as well supported both empirically and theoretically. Response diversity supplies two properties–the feedback channels through which a system registers that something has gone wrong and the redundancy that keeps it operating while the correction is made. A landscape of continuous monoculture holds neither. An institution whose decision makers share one training, one income bracket, one career stage and one geography holds neither. Classifying the maintenance of variety as generosity, therefore, misidentifies the load-bearing element of the structure. Treating variety as infrastructure raises an immediate demand for evidence, since a policymaker asked to fund infrastructure will ask what it produces and how much.

THE EVIDENCE IS ALREADY SUFFICIENT

Evidence that forest cover sustains agricultural production in the tropics is quantitative and spatially explicit. Leite-Filho, Soares-Filho and Davis (2021) measured the relationship between historical deforestation and rainfall across the Southern Brazilian Amazon and found that forest loss reduces rainfall precipitously once it passes a threshold whose value depends on the size of the window examined. The authors characterize the result as a hydrological and economic negative-sum game because reduced rainfall and productivity at larger scales exceed the gains captured locally, and they estimate that avoiding deforestation prevents agricultural losses of up to US$ 1 billion per year in that region alone. Analysis of the climatic envelope of Brazilian agriculture points in the same direction, showing that warming and drying, compounded by forest loss, contract the area climatically suitable for the double-cropping systems on which the country’s grain expansion depends (Rattis, Brando & Macedo 2021). Integrated economic modeling estimates that movement towards an Amazonian tipping point would cost the region US$ 256.6 billion in cumulative gross domestic product through 2050, while the policy package capable of averting it would generate approximately US$ 339.3 billion in additional wealth (Banerjee, Cicowiez & Rattis 2021). The productive value of standing vegetation has been quantified at the scale on which farms operate and priced at the scale on which finance ministries operate. What the rules built on this evidence still lack is a definition of the vegetation that produces the value.

The function at stake belongs to native vegetation with structural and species diversity. A restoration target expressed in hectares can be met by planting a single fast-growing exotic species at commercial density, and countries facing high implementation costs have an incentive to satisfy commitments through exactly such low-cost options, which permits a claim of fulfillment while carbon and biodiversity expectations go unmet (Fagan, Reid & Holland 2020). A homogeneous stand occurs on one phenological schedule, roots at one depth, and responds to drought in one way, thereby removing the response diversity that allows evapotranspiration to continue when conditions shift (Oliver, Heard & Isaac 2015). Protected native vegetation and ecologically diverse restoration deliver the hydrological service agriculture depends on. A single-species commercial stand delivers that service on a single schedule and within a single range of conditions. The definition of a restored hectare therefore governs what the next decade delivers, and that definition is currently drafted in three separate places.

THE ENERGY TRANSITION ARRIVES AT THE FOREST

Decarbonization generates its own demand for land. Sonter, Dade and Watson (2020) mapped global mining areas and found that they potentially influence 50 million km² of the terrestrial surface, that 82% of these areas target materials needed for renewable energy production, and that they coincide with protected areas over 8% of their extent, with key biodiversity areas over 7% and with remaining wilderness over 16%, showing higher mine density in those overlaps than in areas targeting other materials. Bioenergy adds a second claim that splits in two: crop-based biofuel competes for arable land with food production, and wood biomass competes directly with standing forest. Brazil holds substantial reserves of transition minerals and the largest agricultural frontier in the tropics, placing all three claims in the same territory under the same climate justification. Minerals, grain and forest now compete under a single banner, and choosing among them requires a common footing at the resolution where the choice is physically made.

THREE CONVENTIONS, ONE LANDSCAPE

The Rio conventions divided a single landscape among three treaties. Climate, biological diversity and desertification each acquired a secretariat, a negotiation calendar, a national focal point and a reporting obligation–and a producer in Mato Grosso sits within all three mandates at once while coherently reporting to none of them. The desertification convention has moved furthest towards a shared unit through Land Degradation Neutrality, whose indicators parties report at the national level as aggregates. Restoration commitments show what the separation costs, since national pledges are counted in areas with monitoring and management plans largely absent, and 25% of committing countries lost more forest between 2000 and 2015 than they promised to restore over the following fifteen years (Fagan, Reid & Holland 2020). Ostrom’s work (2010) on polycentric governance indicates that institutional fragmentation admits a remedy, because systems with multiple overlapping decision centers can outperform a single central authority when those centers share information and hold one another to account. Evaluation of hybrid public and corporate governance of soil carbon on California rangelands found that alignment between corporate sustainability targets and producer priorities held where complementary public instruments carried the risk of the transition period (Buckley Biggs, Hafner & Mashiri 2021). Polycentric arrangements consequently deliver on the condition that participating centers can verify one another against a common unit, which must resolve well below the national aggregate produced by each current reporting line.

THE OPERATIVE UNIT IS THE NEIGHBOURHOOD

The climate a farm experiences extends across a radius, and the radius has been measured. Cohn et al. (2019) compared undisturbed forest sites across the Amazon and Cerrado against forest loss in concentric rings at 1 to 2, 2 to 4, 4 to 10 and 10 to 50 km, and found that a uniform decline of 25% in forest cover within 50 km raises maximum air temperature by 0.50 °C to 0.95 °C, a magnitude comparable to the warming produced by clearing at the site itself, with the strongest signal between 1 km and 10 km. Rainfall timing varies over a similar range, with each additional percentage point of deforestation delaying the onset of the rainy season by 0.12 to 0.17 days, as measured at 112 rain gauges in southern Amazonia between 1974 and 2012 (Leite-Filho, Pontes & Costa 2019). The rainfall threshold introduced earlier tightens as the window widens, standing at 55% to 60% forest loss within 28 km cells, at 45% to 50% within 56 km cells, and at 25% to 30% within 112 km cells (Leite-Filho, Soares-Filho & Davis 2021). Regional records already capture the compounded signal, with the eastern Amazonia-Cerrado transition showing the largest warming and drying trends in tropical South America between 1981 and 2020, including rising vapor pressure deficit and dry-day frequency, alongside falling precipitation and evaporation (Marengo et al. 2022). A hoe and a twenty-two-row planter therefore both operate within a climate produced by land-cover decisions made kilometers away by people the operator will never meet, and depressed yields push planted area outward onto marginal land, which increases degradation and depresses yields again. Agricultural outcomes in the tropics are governed by a neighborhood spanning tens of kilometers, and Brazilian land policy assigns scores, payments and penalties either at the property or at the State level. Closing that distance requires an instrument that reads the neighborhood and pays the property.

AN INDEX ASSEMBLED FROM THOSE CONSTRAINTS

The Tropical Regenerative Agriculture Index under development at Instituto de Pesquisa Ambiental da Amazônia (IPAM), Woodwell Climate Research Center, Grupo Associado de Agricultura Sustentável (GAAS) and Fundação Dom Cabral (FDC) is an attempt to build that instrument, and each of its design decisions answers one of the findings set out above. Its twenty-five indicators are grouped into biophysical quality, biophysical potential, benefits in the landscape and permanence, which brings response diversity inside the score instead of leaving it as context around the score. Aggregation is by geometric mean, a family of functions that limits the extent to which strength in one dimension can substitute for weakness in another, and the choice of that family is the operational form of the claim that a landscape requires several functions working at once (Gómez-Limón, Arriaza & Guerrero-Baena 2020). A collective landscape bonus activates only when a minimum share of neighboring properties within a hexagonal unit of roughly five square kilometers also performs, a resolution taken from documented pollinator foraging distances and a direct answer to the finding that the operative unit exceeds the property boundary. A randomized trial in Mexico supports conditioning payments on the entire holding, since contracts requiring landowners to enroll all of their forest reduced deforestation by 41% relative to contracts allowing them to choose parcels, quadrupling the program’s cost-effectiveness (Izquierdo-Tort, Jayachandran & Saavedra 2024). The quality of land tenure records caps every score, because a property whose boundaries cannot be verified cannot support a durable claim on public or financial resources. The output offered to financial instruments is the distance between current and attainable performance over three, five and ten years, which gives credit lines, payments for ecosystem services and climate insurance a quantity they can price. An instrument built this way converts each constraint identified in this essay into a coefficient open to challenge, and the challenge it invites is one its authors must deliberately open.

WHAT REMAINS UNRESOLVED

Composite indices can compress what they measure into something easier to game than to achieve. Giller et al. (2021) argue that regenerative agriculture lacks agreed definitions and outcome criteria, a critique that belongs inside the design of any such instrument. Three exposures follow from it and deserve to be stated by the people building the index rather than by its critics. The geometric mean constrains substitution between dimensions without eliminating it, leaving the weights carrying as much of the result as the aggregation rule does. Evidence that deforestation reduces rainfall is considerably stronger than evidence that restoration returns rainfall at a comparable rate, which places the hydrological return on a restored hectare among the quantities still under active revision. An index built jointly by a research institute, a climate science center and a business school will be read by some audiences as a route to cheap credit for producers who have changed very little. The defense against all three readings is to publish the equations, the weights, the uncertainty and the calibration data, so disagreement attaches to specific coefficients and can be settled by evidence. We are currently working on that, even though I do not yet know where the agricultural, forest and energy pathways converge, and I am not persuaded by any account that claims to know while offering no measurement to test the claim. The climate emergency is forcing a pace of decision-making that exceeds our physical and intellectual preparation, leaving the honest position of building the instrument while its destination remains under debate. The immediate step is to apply the index to real properties across the Amazon and the Cerrado and to publish what fails. Criticism aimed at our coefficients is the contribution I am asking for, and the longer task we ask readers to join is to hold four claims within a single science-policy enclosure and keep each fed.

References

Ashby, W. Ross. 1956. An Introduction to Cybernetics. London: Chapman & Hall.

Azoulay, Pierre, Christian Fons-Rosen & Joshua S. Graff Zivin. 2019. “Does Science Advance One Funeral at a Time?” American Economic Review 109 (8): 2889-2920. https://doi.org/10.1257/aer.20161574.

Banerjee, Onil, Martín Cicowiez & Ludmila Rattis. 2021. “An Amazon Tipping Point: The Economic and Environmental Fallout.” Inter-American Development Bank Working Paper 01259. https://doi.org/10.18235/0003385.

Buckley Biggs, Nicole, Jayce Hafner & Fadzayi E. Mashiri. 2021. “Payments for Ecosystem Services within the Hybrid Governance Model: Evaluating Policy Alignment and Complementarity on California Rangelands.” Ecology and Society 26 (1). https://doi.org/10.5751/ES-12254-260119.

Cohn, Avery S., Nishan Bhattarai, Jessica Campolo, Octavia Crompton, David Dralle, John Duncan & Sally Thompson. 2019. “Forest Loss in Brazil Increases Maximum Temperatures within 50 km.” Environmental Research Letters 14 (8): 084047. https://doi.org/10.1088/1748-9326/ab31fb.

Fagan, Matthew E., J. Leighton Reid & Margaret B. Holland. 2020. “How Feasible are Global Forest Restoration Commitments?” Conservation Letters 13 (3). https://doi.org/10.1111/conl.12700.

Giller, Ken E., Renske Hijbeek, Jens A. Andersson & James Sumberg. 2021. “Regenerative Agriculture: An Agronomic Perspective.” Outlook on Agriculture 50 (1): 13-25. https://doi.org/10.1177/0030727021998063.

Gómez-Limón, José A., Manuel Arriaza & M. Dolores Guerrero-Baena. 2020. “Building a Composite Indicator to Measure Environmental Sustainability Using Alternative Weighting Methods.” Sustainability 12 (11): 4398. https://doi.org/10.3390/su12114398.

Grantham, Hedley S., Paolo Tibaldeschi, Pablo Izquierdo et al. 2021. “The Emerging Threat of Extractives Sector to Intact Forest Landscapes.” Frontiers in Forests and Global Change 4. https://doi.org/10.3389/ffgc.2021.692338.

Izquierdo-Tort, Santiago, Seema Jayachandran & Santiago Saavedra. 2024. “Redesigning Payments for Ecosystem Services to Increase Cost-effectiveness.” Nature Communications 15. https://doi.org/10.1038/s41467-024-53643-1.

Leite-Filho, Argemiro Teixeira, Vanessa Y. de Sousa Pontes & Marcos H. Costa. 2019. “Effects of Deforestation on the Onset of the Rainy Season and the Duration of Dry Spells in Southern Amazonia.” Journal of Geophysical Research: Atmospheres 124 (10): 5268-5281. https://doi.org/10.1029/2018JD029537.

Leite-Filho, Argemiro Teixeira, Britaldo Soares-Filho & Juliana Leroy Davis. 2021. “Deforestation Reduces Rainfall and Agricultural Revenues in the Brazilian Amazon.” Nature Communications 12. https://doi.org/10.1038/s41467-021-22840-7.

Marengo, José A., Juan C. Jimenez, Jhan-Carlo Espinoza, Ana Paula Cunha & Luiz E. O. Aragão. 2022. “Increased Climate Pressure on the Agricultural Frontier in the Eastern Amazonia–Cerrado Transition Zone.” Scientific Reports 12: 457. https://doi.org/10.1038/s41598-021-04241-4.

Oliver, Tom H., Matthew S. Heard & Nick J. B. Isaac. 2015. “Biodiversity and Resilience of Ecosystem Functions.” Trends in Ecology & Evolution 30 (11): 673-684. https://doi.org/10.1016/j.tree.2015.08.009.

Ostrom, Elinor. 2010. “Beyond Markets and States: Polycentric Governance of Complex Economic Systems.” American Economic Review 100 (3): 641-672. https://www.aeaweb.org/articles?id=10.1257/aer.100.3.641.

Planck, Max. 1950. Scientific Autobiography and Other Papers. Londres: Williams & Norgate.

Rattis, Ludmila, Paulo M. Brando & Márcia N. Macedo. 2021. “Climatic Limit for Agriculture in Brazil.” Nature Climate Change 11 (12): 1098-1104. https://doi.org/10.1038/s41558-021-01214-3.

Sonter, Laura J., Marie C. Dade & James E. M. Watson. 2020. “Renewable Energy Production Will Exacerbate Mining Threats to Biodiversity.” Nature Communications 11. https://doi.org/10.1038/s41467-020-17928-5.

Unruh, Gregory C. 2000. “Understanding Carbon Lock-in.” Energy Policy 28 (12): 817-830. https://doi.org/10.1016/S0301-4215(00)00070-7.

Submitted: August 10, 2026

Accepted for publication: September 8, 2026

Copyright © 2026 CEBRI-Journal. This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original article is properly cited.