C-FIN is Canopy's country-level view of conservation finance: for every African nation, an estimated annual cost to manage its protected estate, the money Canopy can verify reaching it, and the shortfall between the two. This note sets out every calculation and its source.
In plain terms: for each country we multiply its protected land area by a published per-square-kilometre management-cost benchmark to get an annual need, then subtract the recurring, annualised conservation slice of the deals Canopy has independently verified to get a gap. Headline deal size is not the conservation figure, and a one-off payment is shown as realised capital rather than treated as recurring. Every number is arithmetic on public figures. Nothing on the page is generated by a model, and where an input is missing we say so rather than show a zero.
C-FIN answers one question for a whole country: is the money that reaches its protected areas anywhere near what running them costs. It sits alongside two neighbours and does not duplicate them. Funders profiles the institutions that supply capital. The Nature Finance tracker records individual transactions. C-FIN is the demand-and-gap view: the national need, and the shortfall once verified funding is counted.
It is not investment advice, and it is not a complete census of conservation finance. It is a deterministic, reproducible estimate built to be argued with, not a precise budget.
For each country the need is a single multiplication:
Two rates, and each country says which one it uses. Since 1 September 2026, 20 countries are priced at their own fitted rate from Correa et al. (2024) Table 2, the same study behind the shared reference. The remaining 32 are priced at the shared reference of the published rate/km²/year. Every country carries a rate_source saying which, and a pooled-priced country carries the reason it is pooled. The fitted rates range from $148/km²/yr (Namibia) to $5,494 (Chad), which is wider than the cited literature range and is the variation the single rate was averaging away.
South Africa is not priced at its own rate either, for a different reason. Correa et al. do not cap South Africa, but their Table 2 puts its current spend at 17.0 times its own necessary spend, more than three times the next highest row, and their current figure is 2.8 times the recurrent funding C-FIN can independently source for the country. Both columns of that row disagree with the evidence, so the fitted value is not used as a cost floor and South Africa prices at the shared reference. This exclusion is asymmetric on purpose: a country spending far below its fitted requirement is the funding gap C-FIN exists to measure, while a country apparently spending 17 times its requirement means the requirement estimate is wrong.
Four countries are deliberately not priced at their own rate. Correa et al.'s Table 2 footnote caps the Central African Republic, DR Congo, South Sudan and Nigeria at $10,000/km²/year in 2015 dollars, because the authors judged the raw fitted values implausible; CAR's was $84,788/km²/year. C-FIN does not publish a rate its own source says it does not believe, so those four are priced at the shared reference. Using the capped values instead would have raised total need by 148% rather than 19%, with almost the entire increase falling on those four countries.
C-FIN's planning reference is the published rate/km²/year in the stated dollar vintage. It is based on the $1,034/km²/year in 2015 US dollars result published by Correa et al. (2024), converted using US CPI-U. The study linked annual recurrent protected-area spending to elephant population change in 80 protected areas across 25 sub-Saharan African countries, including forest and savannah systems. It is a common planning reference, not an audited national budget or a cost to reach 30% protection.
Why this is the headline. C-FIN uses a stated selection rule, rather than taking an average of unlike studies: select the broadest Africa-specific estimate that is outcome-linked, explicitly covers recurrent management costs, states its price year, and spans more than one major terrestrial ecosystem. Correa et al. meets those criteria. It is still a planning reference, not an audited national budget or a claimed cost for every individual protected area.
How the number is calculated. C-FIN carries forward the transparently rounded 2018-dollar conversion of the study result, then converts it to the live dollar vintage using U.S. CPI-U. The published inputs and rounding are documented in the evidence dossier. US consumer inflation is a transparent common-dollar convention; it does not claim that field costs moved identically in every African country.
The headline is not an average of the following figures. They answer different questions and are kept separate:
| the published rate / km² / yr | Headline planning reference in the stated dollar vintage: Correa et al.'s $1,034 2015-USD median to stabilise elephants, transparently converted. |
| $1,347 / km² / yr | High-intensity lion sensitivity: Lindsey et al.'s $1,271 2015-USD threshold, transparently rebased. It is not the headline because it is a lion-landscape threshold. |
| $282 / km² / yr | Older Congo Basin forest reference: Blom's $212 2004-USD effective-management estimate, transparently rebased. It is not the headline because it is regional and older. |
| $254 / km² / yr | Median observed spending in Correa et al.'s 2015-dollar study sample. Observed spend may be underfunding, so C-FIN does not use it as the need. |
June 2026 sensitivity band: / / per km²/year.
One rate across every biome, with an explicit limit. The selected study is broader than a savannah-only source, but it does not make the published rate a local budget for desert, forest, wetland or high-tourism estates. Country, biome, governance, threat, access, scale and tourism conditions can move a site's true budget materially above or below it. C-FIN does not adjust any country until it has a separately cited, deterministic input that can be applied consistently.
All CPI figures are the U.S. Bureau of Labor Statistics CPI-U, all items, U.S. city average. The conversion is a transparent common-dollar convention. It does not claim that prices for rangers, fuel, equipment or community programmes moved in lockstep with U.S. consumer prices in every African country.
The source selection comes before the inflation conversion. C-FIN did not start with Lindsey et al.'s lion figure and inflate it. It reviewed the available Africa-focused management-cost studies and selected Correa et al. as the all-estate reference because it is the broadest recent study that links recurrent protected-area management spend to a measured biological outcome across both forest and savannah elephant systems.
| Correa et al. (2024) | Selected headline source. The authors compiled annual site-level spending and repeated elephant population surveys for 102 protected areas. Eighty sites in 25 sub-Saharan African countries had both the spending and population-change data needed for the final analysis, covering more than 40% of the combined forest and savannah elephant population. The headline is the modelled median spend associated with zero annual elephant population change: $1,034/km²/year in 2015 USD. |
| Lindsey et al. (2018) | Retained as a high-intensity lion comparison, not selected as the headline. Its $1,271/km²/year in 2015 USD is a threshold model for 115 lion protected areas, designed to predict lion populations at or above 50% of carrying capacity. It is a valuable, deliberately demanding savannah lion-landscape reference, but it is not an all-biome average and does not include forest elephant systems. |
| Blom (2004) | Retained as a forest context reference, not selected as the headline. It estimates an effective protected-area network for the Niger Delta-Congo Basin forest region. Its $212/km²/year 2004-USD estimate is geographically specific and older. It is not used to label all land outside elephant range as cheap. |
What Correa et al. actually estimates. The study uses repeated population counts to estimate each site's annual elephant population change. Its model tests the relationship between that change and annual spend per unit area while accounting for protected-area size, government effectiveness, human population density and GDP per person. The selected result is the spend level that the model associates with a zero annual population trend, holding the other variables at their stated values. It is an empirical, outcome-linked planning reference. It is not a claim that every African protected area has elephants, that every protected area should have elephants, or that the published rate is a locally costed budget.
What the number covers. Correa et al. describes annual recurrent management spend, including personnel, law enforcement, infrastructure and road maintenance, habitat management, and engagement with adjacent communities. It does not turn capital establishment, land purchase, protected-area expansion, ecological restoration, compensation, debt service or every country-level administrative cost into an annual management need. C-FIN keeps those categories outside this benchmark unless a future source explicitly and comparably includes them.
Why the observed-spend figure is not used. The study's median observed annual spend was $254/km²/year in 2015 USD, far below its modelled median spend needed to stabilise elephant populations. Using actual reported spend as the benchmark would confuse present expenditure with a threshold associated with the stated conservation outcome.
No biomass measure is used in the published-rate calculation. Nor does C-FIN currently use a protected-area polygon overlay, lion range polygon, elephant range polygon or forest-biome polygon to calculate a country's published need. The country formula uses a single World Bank reported terrestrial protected-area extent multiplied by the published reference.
Why no spatial tier is yet published. C-FIN has run an internal exploratory overlay of terrestrial protected-area polygons with known-extant lion range, known-extant savannah and forest elephant range, and forest biome polygons. It uses a conservative species rule, excluding ranges marked possibly extant, uncertain or extinct. That exercise is useful for testing a future tool, but it does not yet establish a cost for all remaining desert, wetland, mountain, grassland or multi-use protected land. Applying the lower Congo Basin forest figure to every non-elephant polygon would be an unsupported assumption, so the overlay currently changes no country rate, total or gap.
What would be needed before a spatially differentiated rate is defensible. C-FIN would need a fixed protected-area data vintage; licensable and cited current range and biome layers; a reconciled denominator with the published country coverage series; an evidence-backed recurrent-cost tier for every mutually exclusive land class; and sensitivity tests that show how much a country result changes when the spatial inputs move. Until then, the published number stays a single continent-wide planning reference with explicit limits, rather than a false-precision country model.
The per-ecosystem rate gate, live. Rather than leave this to prose, C-FIN runs a standing gate: each ecosystem earns its own cited rate the moment the evidence clears the bar, shown here with its status and source. Today one ecosystem carries a provisional cited rate and the rest fall back to the pooled reference. No ecosystem yet changes a country's need.
| Ecosystem | Status | Rate /km²/yr | Basis |
|---|
Illustrative partial pooling, for honesty about the gaps. The gate above abstains for every ecosystem without a cited rate. As a transparency check, C-FIN also runs a hierarchical (partial-pooling) model that instead gives every ecosystem a number and an explicit uncertainty band, shrinking data-poor ecosystems toward the pooled reference. It is deliberately shown as a sensitivity, not a result: with a single cited ecosystem rate today, the data-poor ecosystems all collapse to the pooled reference and their bands are set by the model's prior, not by evidence. Even the one evidenced ecosystem is pulled off its raw cited rate by that prior. It changes no country rate, total, or need.
| Ecosystem | Basis | Pooled median /km²/yr | 95% band |
|---|
Direct country figures, shown without changing the portfolio benchmark. Correa et al. also publishes a fitted necessary-spend figure for 25 of C-FIN's 54 countries. The table below rebases those direct 2015-dollar figures through the same rounded CPI chain as the pooled reference. The other 29 countries abstain to the pooled rate; no proxy, interpolation or default country adjustment is invented.
Country briefs offer a display-only toggle between the uniform pooled reference (the default on every page load) and the published country-specific figure where one exists. Selecting the country-specific view recomputes that brief's rate, annual need and gap together; uncovered countries visibly retain the pooled reference. The toggle never changes the authoritative portfolio figures on C-FIN or any stored data.
| Country | Current pooled rate | Study country rate | Basis |
|---|
Central African Republic, DR Congo, South Sudan and Nigeria are shown at the source paper's own $10,000/km²/year ceiling in 2015 dollars, before CPI rebasing. Their raw uncapped figures remain in the registry for provenance.
Existing protected estate only. C-FIN multiplies the rate by land already reported as protected. It does not estimate the cost of expanding a country to 30% protection, buying land, restoring ecosystems, or funding new capital works. Target 3 is contextual coverage information, not a country budget formula.
Terrestrial only. Marine protected areas are out of scope for this terrestrial cost benchmark, so marine extent does not enter the need.
Latin America is a useful external comparator because it includes large tropical forests, high-biodiversity protected landscapes, community and Indigenous land governance, tourism economies and strong variation in human pressure. It does not supply a substitute Africa-wide rate. Different wage structures, land tenure, institutions, threat patterns, protected-area categories and management objectives mean that importing a Latin American dollar-per-kilometre figure would create false precision.
| Lessmann et al. (2024) | Outcome evidence, not a cost benchmark. The study tested protected-area effectiveness across 17 Latin American national systems and examined 27 individual protected areas in Ecuador. In Ecuador, larger funding deficits were associated with lower effectiveness in avoiding deforestation; the relationship was particularly important in places facing high agricultural pressure. This supports C-FIN's view that funding need should be read alongside pressure and governance, not as a uniform area-only rule. |
| Waldron et al. / IDB (2026) | Method design comparator, not a current-estate benchmark. The IDB's 30x30 scenarios for 26 Latin American and Caribbean countries model future terrestrial management costs using protected-system scale, local human pressure, agricultural rent and site revenue. It shows why the cost of defending land varies materially by context. It is not applied to C-FIN because it estimates a future 30x30 system, not Africa's currently reported protected estate, and it is a scenario model rather than an Africa-specific outcome study. |
| UNDP and TNC (2010) | Finance-definition warning. The 20-country LAC exercise distinguishes basic from optimal management needs but states that country methodologies and definitions varied, making regional results difficult to compare. That is a direct warning against averaging heterogeneous country budgets into one supposedly universal rate. |
How Latin America will be used. C-FIN will use the LAC evidence to test whether a future “other terrestrial” tier needs to vary with pressure, scale, remoteness and management type. It will not lower the rate for non-elephant land simply because it is outside an elephant or lion polygon. Any future tier must have an Africa-relevant recurrent-cost source, a precise scope, a stated dollar year, and a reproducible spatial or country-level rule. Until then, Correa et al. remains the single published Africa-wide planning reference.
What this means for the next benchmark decision. The next decision is not whether to replace the published reference with a Latin American average. It is whether the combined Africa and LAC evidence supports a transparent scenario tool alongside the common reference: for example, a wildlife-security case, a forest case and an evidence-backed other-terrestrial case. The tool will only be published if every class has a defensible cost basis and the output is presented as a sensitivity range, not as an audited national budget.
Funding is the money C-FIN can stand behind. C-FIN keeps two channels distinct: the conservation slice of verified market-finance transactions, drawn from Canopy's deals record, and a growing band of cited recurrent funding from protected-area agencies and NGO operating accounts. The rules are deliberately conservative, so the gap is never flattered by soft or mislabelled numbers.
Recurring finance is what the gap nets against. For each deal we take the documented conservation portion, terrestrial only, and annualise it over a documented tenor. That recurring, annual-equivalent figure is what is subtracted from the annual need. A deal with no documented tenor (a one-off results payment, a forward carbon purchase) is not annualised, so it does not enter the recurring figure; no tenor is invented to manufacture one.
Conservation capital is shown alongside, never hidden. The cumulative documented conservation dollars a country has received, one-off and recurring, are reported next to the recurring figure, so a genuine one-off payment shows as realised capital rather than being erased. Capital is context; it is not netted against the annual need.
The conservation slice, not the headline. A deal's headline size is the instrument, not the conservation spend. A $750M sovereign green bond whose proceeds fund clean transport and water contributes $0 conservation; a $150M wildlife bond of which $10M reaches rhino conservation contributes $10M. Where a deal's conservation portion is not documented, it is left null, not guessed.
Marine is context, out of scope. The need is terrestrial, so conservation dollars in marine instruments (blue bonds, marine debt-for-nature swaps) are shown as context and are never netted against a terrestrial need.
Multi-country deals are unallocated, never split. A deal naming two or more countries with no documented per-country split is credited to no single country: it is marked unallocated and shown on a portfolio line, never split across countries and never duplicated into each. (Earlier such a deal's full value was credited to one arbitrary country; that is fixed.)
Carbon and funders are context, not funding. Carbon-credit issuance is not verified annual management revenue, and a funder naming a country in its posture is a stated intent, not a placed dollar. Both are shown on each country brief for context and are never added to counted funding.
No double counting. Only deals flagged to count toward the total are summed, and each counted deal's conservation slice is credited exactly once, so a deal and a funder commitment describing the same money cannot both land in the number.
Recurrent funding is a separate, sourced channel. C-FIN includes a public protected-area agency budget only where the published document identifies the relevant agency or protected-area programme. It includes NGO spend only where an annual report or audited account identifies country or protected-area-specific operating expenditure. It can include a dated country-specific project disbursement, or a project amount divided by its published term, only when the protected-area scope and calculation are explicit. If a direct record is from another year, overlaps a project envelope, or cannot safely be combined with the current source, it appears as “Other dated direct funding evidence” and is not summed. A global NGO budget, a broad environment-ministry vote, a one-off grant, construction and equipment are not silently allocated or annualised.
Time is shown, not concealed. The planning reference is expressed in June 2026 US dollars. Funding evidence retains its reported source year or stated project term, because the current dataset is not yet a harmonised FY2026 budget series. The main comparison therefore uses the label Latest annual evidence: it is an evidence cue against the common reference, not a claim of a country’s 2026 funding balance.
No double counting across channels. Funding is counted at its final operating destination once. An NGO grant already captured in a tracked deal or inside an agency budget is not added again. Multi-year operating support is included only when its annual amount or documented term makes annual treatment reproducible.
The deal gap isolates the market-finance channel. The identified-funding comparison adds cited public-agency and NGO operating funding, but is computed only for countries with a sourced recurrent figure. Because source years differ, it is not a country’s literal 2026 funding balance; the main page labels the relationship as evidence against the benchmark. It is null, with a reason, everywhere else: missing research is never recoded as $0 funding. Both calculated values cannot go below zero. The aggregate identities are asserted in a regression test rather than left to be eyeballed.
C-FIN published this percentage per country and portfolio-wide until 2 September 2026. It has been withdrawn rather than caveated, because the problem is not one of interpretation. Need is area × rate. Protected area varies 2,660-fold across the 54 countries (100 to 266,007 km²) while the applied rates vary 34-fold, so the denominator is dominated by the size of the estate and the ratio inherits that. Measured across the 51 countries carrying both terms, adequacy correlates −0.617 with protected area and +0.109 with the funding it claims to measure.
The clearest demonstration is two countries priced at the identical pooled rate. Mauritius holds 100 km² and $2.75m of sourced funding, and read 1,889%. Niger holds 266,007 km² and $1.72m, and read 0.4%. Their funding differs by a factor of 1.6; their land differs by a factor of 2,660. Essentially the whole gap between those two figures was land area.
Two fixes were tested and ruled out before withdrawing it. Per-country fitted rates shipped and the relationship did not improve (−0.700 to −0.617). Splitting the cohort on whether a national agency budget exists shows adequacy has no relationship to funding even where the numerator is complete (−0.023 for the 30 countries with an agency budget). The defect is structural: adequacy = funding ÷ (area × rate), and area varies more than anything else in that expression.
What is published instead. Both inputs, side by side and unrounded: the annual planning benchmark and the sourced recurrent funding, each with its own confidence band and source years. The division is what manufactured a performance reading out of a size difference, so the page states the two numbers and leaves them un-divided. A replacement measure, built from what is at stake, how likely it is to be lost and how far management falls short, is scoped in the tracker rather than promised here.
The field remains in the data. funding_adequacy_pct is still computed and still present in funding_gap.json, so the audit trail and every historical figure quoted below stay reproducible. It is no longer rendered on any surface.
The confidence band reports coverage as well as recency. Two independent things undermine a funding figure, and the band shows the weaker of them. Recency is how much of the sourced funding is recent and dated. Coverage is whether a national protected-area agency budget is among the sources at all: it is the only source that spans the whole estate, so a figure built from project grants and NGO operating spend alone omits the largest component and understates national spend by an unknown amount. A country without a sourced agency budget is never banded strong however recent its records are, and its recurrent figure should be read as a floor because the true total is higher. 23 of the 54 countries are in that position today, and they carry $3.19B of the $7.73B annual need. Sourcing their agency budgets is the single largest improvement available to this model.
Sourced funding above the benchmark is a scope cue, not a surplus. Where a country's sourced funding exceeds its annual need, the usual cause is that the sourced national conservation figure is broader in scope than the terrestrial protected estate the benchmark prices, or that the protected estate is very small. Read it as a prompt to check the source's scope, never as an over-funded verdict.
Withdrawing funding adequacy left the question it was badly answering still open. Need answers how much, and it does that honestly: protected-area extent times a planning rate. It cannot answer where, because area varies 2,660-fold across the estate and dominates the expression. So the two questions are now answered separately. This one emits a ranking rather than dollars, and the cost sits beside it, labelled a cost. It is published on the C-FIN country page and in data/conservation_priority.json.
Live reading.
Three terms, and they multiply. Stake is what would be lost. Threat is the rise in human pressure, cropland and built-up land, inside the 10 km ring outside each park between 2019 and 2024. Capacity gap is how far the PACE security, governance, leadership, budget and fundraising reads sit from the top of their scale. A sum would let one strong term carry a country that fails the other two; a product does not. A precious place under no threat does not need money this year, a threatened place with little at stake is not worth it, and a threatened, precious place whose managers are already coping does not need more. The cube root is the geometric mean, which keeps that property and returns the result to the same 0-1 scale as the terms.
Every roll-up is a mean over a country's parks, never a sum. This is the easiest way to rebuild the artefact the index exists to remove. The number of parks a country has correlates +0.569 with its protected area, so a sum, or an area-weighted mean, reimports area through the back door. The park mean of PACE correlates −0.024 with area, and the mean is what is used.
Fixed anchors, never scaled against the estate. PACE runs 1 to 5, the EPI indicators run 0 to 100, and ring encroachment is anchored on a fixed band of −1.0 to +5.0 percentage points. Nothing is min-max scaled across the 54 countries, so a country's reading cannot move because another country moved. That is precisely the property funding adequacy did not have.
The stake term is provisional, and it is gated rather than chosen. The risk in any index of this shape is trading one single-variable proxy, area, for another, species count. So candidate sources are screened: any source whose country values correlate with protected area at |ρ| ≥ 0.40 abstains, whatever its provenance. Today two sources qualify, both from the 2024 Yale Environmental Performance Index: the Red List Index and the Species Habitat Index. Both are read at the country rather than at the park, and both measure how imperilled a country's biodiversity is rather than how irreplaceable it is. The Species Protection Index and the Biodiversity and Habitat composite abstain because they grade protection performance rather than stake. IBAT's STAR metric and Key Biodiversity Area irreplaceability abstain because neither is licensed; either would be what promotes this term from provisional, and would also move the whole index from the country to the park.
Both readings are published, so the provisional term can be judged. Every country carries the ranking with the stake term and the ranking without it, and the rank correlation between them is recomputed on each build. It is not small. A ranking that depends on a provisional term is a hypothesis, and saying so is cheaper than being right about it.
Threat is the ring, not the site-hazard index, and that was a correction. The index was first built on the C-RISK site-hazard layer, which reads five satellite feeds in the current period. 102 of its 199 scored parks sit at exactly zero, so inside a product it decided the ranking on its own: the first build correlated +0.887 with that one term, and five countries collapsed to a priority of exactly zero on a quiet week in a single-park cohort. Ring encroachment is the measurement that answers how likely a place is to be lost. Acute hazard is still published, beside the index rather than inside it, on the same reasoning the conservation-condition axis already uses for the same pair of layers: one week cannot overturn six years.
The boundary result is deliberately not in the capacity gap. It would be natural to add it, since the ring comparison reads whether a boundary holds. It cannot be added: that comparison is the change inside the park minus the change in the ring, and the ring is already the threat term. Using both would wire one quantity into two terms of a product with opposite signs and manufacture a relationship between them that no data supports. The boundary result is carried as context on each country instead.
A placebo runs on every build. Each term is rotated across the country list by a different fixed offset, so every term keeps its own spread and lands on the wrong countries. If that rotated version still ranked like the real one, the ranking would be coming from the shape of the formula rather than from stake, threat and gap occurring together in the same place. The rotated correlation is published beside the real one.
What it does not do. It does not emit dollars: converting a ranking back into a budget is a separate problem and is not attempted here. It does not feed any other Canopy score. C-RISK readiness already reads C-FIN through its traction slice and country-mean PACE through its asset slice, and the capacity-gap term here is PACE, so letting this index replace the annual need would make readiness consume PACE twice. Nothing reads the artefact except the pages that display it, and a guard walks the repository on every change to keep it that way.
Terrestrial only, and it abstains rather than guessing. Marine protected areas are out of scope for all three terms, so a marine-dominated country's priority is a statement about its land estate alone. A country missing any term is not ranked and says which term is missing; it is never scored on the terms it does have, and never filled in from a regional average. The threat term is the binding constraint on coverage, because the impact layer reaches 138 of 200 parks. Each ranked country also carries a confidence set by the parks behind its thinnest term, not its park count.
The coverage figure is the share of a country's land inside protected areas, shown against the 30% by 2030 target of the Kunming-Montreal Global Biodiversity Framework (Target 3). It comes from World Bank Open Data, which is WDPA-derived and open without a token: terrestrial protected share (ER.LND.PTLD.ZS), marine protected share (ER.MRN.PTMR.ZS), and land area (AG.LND.TOTL.K2).
A floor, not a ceiling. Protected Planet / WDPA is the authoritative 30x30 tracker; these World Bank series are WDPA-derived. OECM coverage is under-reported globally, so reported figures are a floor, not a ceiling.
Landlocked is not zero. Landlocked countries carry marine coverage as not applicable, never as 0.
The figures at the top of C-FIN are sums across the countries with a computable value: total annual need, total recurring deal finance, the deal-finance gap, sourced recurrent funding and the identified-funding gap. The latter is summed only across countries with a sourced recurrent figure. Shown beside them: total verified conservation capital (the cumulative documented conservation dollars, split into the portion attributable to a country and the multi-country portion held unallocated), the marine total held out of scope, and the headline tracked deal value, which is the instrument size and not the conservation figure. A country with no computable need is left out of the need and gap totals rather than counted as zero, so the aggregates never quietly absorb a missing input.
The headline need carries its range. The total annual need is published with the span produced by the cited low and high African rates, each applied flat to the whole terrestrial estate. That span is the distance between published rates, not a statistical confidence interval, and neither bound is biome-resolved. It sits on the headline rather than inside a footnote because the choice of rate, not the arithmetic, is what moves the total.
Marine protected areas are out of scope. Every need figure prices the terrestrial estate only, because the management-cost reference is terrestrial. A country whose protection is largely marine therefore carries a small need beside sourced funding that covers both, and the two are not comparable. Seychelles is the clearest case: 313 km² of terrestrial protected land against national conservation funding that is overwhelmingly marine.
Nine fitted rates fall outside the cited band. The literature the model cites supports $375 to $1,791/km²/yr. Nine countries price at a fitted rate outside it: Chad ($5,494), Tanzania ($5,094), Zimbabwe ($3,175), Cameroon ($2,236) and Kenya ($1,797) above it, and Namibia ($148), Botswana ($160), Ethiopia ($174) and Burkina Faso ($307) below. Their need figures are extrapolations rather than rates the sources bound, and each is marked as such on its country row.
A priority ranking is published beside the annual need, answering the question withdrawing funding adequacy left open. priority = (stake × threat × capacity gap)1/3, each term on 0 to 1, rolled up as a mean over a country's parks and never a sum, on fixed anchors rather than scaled against the estate. It correlates −0.280 with protected area where the annual need correlates +0.926, no single term stands in for it, and a placebo that rotates the terms apart correlates −0.146 with the real order. The stake term is provisional: it is gated, any source correlating with protected area at |ρ| ≥ 0.40 abstains whatever its provenance, and today only the 2024 EPI Red List and Species Habitat indices qualify. Both the ranking and the ranking without that term are published, and they agree at +0.610. 31 of 54 countries are ranked; the rest abstain with their missing term named. It emits no dollars, changes no need, funding figure or gap, and feeds no other Canopy score.
Funding adequacy is withdrawn from every surface. It measured protected area, not funding: across the 51 countries carrying both terms it correlates −0.617 with area and +0.109 with funding, because need is area × rate and area varies 2,660-fold while rates vary 34-fold. Mauritius read 1,889% and Niger 0.4% on the same rate, on funding that differs by a factor of 1.6. Two candidate fixes had already been measured and ruled out: per-country fitted rates, and completing the funding numerator. The percentage is gone from the portfolio headline, the country table and the country brief; the need and the sourced funding are both still published, un-divided. Two disclosures that existed only in the data now reach the page: the nine fitted rates outside the cited $375-$1,791/km²/yr band are marked on their country rows, and the terrestrial-only scope is stated wherever a need is shown. No need, funding figure or gap changed, and funding_adequacy_pct remains in the data for the audit trail.
The funding-confidence band now reports coverage as well as recency, and shows the weaker of the two. Previously it measured only how recent and well dated the sourced funding was, so a country whose entire figure was a single recent project grant could be banded strong while its funding numerator omitted the national protected-area agency budget entirely. That is the reading the band exists to prevent, since adequacy is a floor and the band is what separates underfunded from under-sourced. 16 countries moved from strong to moderate, adequacy is now shown with a ≥ for the 21 affected countries, and each says in its reason that the agency budget is missing. No need, funding figure, adequacy percentage or gap changed.
South Africa no longer prices at its own fitted rate. Correa et al. do not cap it, but Table 2 puts its current spend at 17.0 times its own necessary spend, more than three times the next highest row, and that current figure is 2.8 times the recurrent funding C-FIN can independently source. Both columns of the row disagree with the evidence, so the fitted value is not used as a cost floor and South Africa prices at the shared reference. Its annual need moves from $33.3m to $169.6m and its funding adequacy from 609.5% to 119.7%; total need moves from $7.60bn to $7.73bn and portfolio adequacy from 12.1% to 11.9%. The exclusion is asymmetric by design and is recorded per country in the rate registry rather than applied as a threshold.
Country-specific planning rates went live. 21 countries are now priced at their own fitted rate from Correa et al. (2024) Table 2, and 31 remain on the shared pooled reference; each country states which rate prices it and why. Total annual need moves from $6.38bn to $7.60bn (+19.1%) and portfolio funding adequacy from 14.4% to 12.1%, with no change to any funding evidence. The four countries whose fitted estimate that paper's own footnote flags as implausible are priced at the shared reference, not at the capped value; publishing the capped values would have raised need by 148%. The sensitivity band remains a counterfactual on the rate choice and is not rescaled around a fitted rate, so a country's need may now fall outside it and every such country declares that explicitly.
Added funding adequacy, sourced funding as a share of the annual need, computed in the pipeline and published per country and portfolio-wide. It separates the severity of a shortfall from the size of the estate, which the absolute dollar gap alone cannot do. It is reported only where both a need and a sourced recurrent figure exist, is never capped above 100%, and abstains rather than reading 0%. Promoted the cited-rate need range from the collapsed benchmark panel onto the portfolio headline, labelled as the span between published rates rather than a confidence interval. No need, funding figure or gap changed.
Published the per-ecosystem rate gate: a live table of all 14 RESOLVE biomes showing which carry a cited management-cost rate and which abstain to the pooled reference, each with its status and source. Display only, reading the machine-readable gate output. No country need, funding figure or gap changed; every rate still falls back to the single pooled benchmark until an ecosystem's rate is firm and a spatial layer exists.
Added the cited June 2026 low, central and high benchmark sensitivity rates and their portfolio range. Each is applied as a single flat rate to the whole terrestrial estate; this is not a biome-resolved estimate, and issue #386 remains the open path to a spatial rate.
Relabelled the funding comparison as “Latest annual evidence” and made the time distinction explicit: the planning reference is in June 2026 US dollars, while funding sources retain their stated source years. The comparison is an evidence cue, not a harmonised 2026 national funding balance.
Replaced Mozambique’s older BIOFUND observation with BIOFUND’s US$12.23m 2025 disbursement to 23 conservation areas. The source mixes terrestrial, coastal and marine areas, so no terrestrial share is inferred.
Added African Parks’ disclosed first carbon-funding cycle for W National Park in Benin, annualised across its stated 2024–2025 operating period and kept separate from the park’s commercial income.
Added FPRCI’s signed 2025 Côte d’Ivoire grant for six named parks and reserves. The grant is kept separate from OIPR’s public-enterprise budget because that budget does not identify it as a resource.
Added the European Union’s €9m, three-year Sierra Leone biodiversity and community-development programme as related landscape funding. Its broad community, resilience and public-service scope means it is not added to the direct protected-area annual comparison.
Added the EU’s named €3m Kidepo Game Reserve project in South Sudan, annualised across its stated three-year term. The unsplit Boma–Gambella cross-border programme remains unallocated rather than being divided between countries.
Updated Seychelles to SPGA’s 2026 full-expenditure forecast, replacing the prior 2023 audited amount. The separately scheduled capital projects are not stacked on top of this operating forecast.
Updated Mauritius to the 2026/27 National Parks and Conservation Service recurrent allocation. The new fiscal-year estimate replaces the prior 2025/26 amount; capital expenditure remains excluded.
Updated Ghana’s Wildlife Resources programme to the published 2026 Forestry Commission budget line, converted using the Bank of Ghana’s reported January–June 2026 monthly average exchange rates. This current budget replaces the prior-year programme estimate.
Refreshed park-specific earned revenue from African Parks’ 2025 annual report: Iona, Liwonde, Nkhotakota, Kafue and Matusadona. The newer annual amounts replace the prior-year figures rather than stacking with them.
Added Zimbabwe’s approved 2026 current grant to ZimParks. The separate capital allocation and wider environment-sector support are excluded; the published ZiG amount is converted at the official exchange rate stated in the same budget cycle.
Added the US$1.19m protected-area-management, new-PA and OECM component of Ethiopia’s new Rift Valley GBFF project, annualised across 48 months. Its distinct restoration, livelihoods, knowledge and M&E components are excluded.
Added the US$874,465.80 site-linked human–wildlife-conflict component of Gabon’s active Target 3 project, annualised across 48 months. The separate PFP design, knowledge and M&E components remain excluded, as do all prospective transition and endowment figures.
Added only the US$412,883 protected-area-and-OECM-restoration line in the newly approved Central Togo project, annualised across its 60-month term. The broader US$6.6m project total is not counted; its agriculture, livelihoods, policy, monitoring and management amounts remain excluded. The project identifies its coordination with the pre-existing Togo landscape project to avoid duplication.
Added a source-specific US$9m, five-year Canadian development-assistance grant for Ghana’s Wechiau Community Hippo Sanctuary and Avu Lagoon Community Protected Area, annualised as US$1.8m. It is labelled as a time-bound, multi-purpose site allocation and is separate from the national Wildlife Resources programme budget.
Replaced Côte d’Ivoire’s third-party copy of the 2025 OIPR budget with the identical official Directorate-General-of-Budget annex. The amount is unchanged; the country page now links directly to the state budget record and states its state-transfer, own-resource and expenditure detail.
Updated Rwanda’s Akagera National Park earned-revenue source from the published 2024 lower bound of US$4.7m to the newer 2025 lower bound of US$5m. This supersedes rather than adds the prior year, and remains a site-specific revenue source separate from Nyungwe and the Dian Fossey Gorilla Fund programme.
Added the documented US$500,000 protected-area-management-and-connectivity share of the active Basse-Lobaye project in the Central African Republic, annualised across its 72-month term. The separate OECM, restoration, livelihoods, policy, knowledge and M&E allocations are not counted.
Added only the explicitly protected-area-management components of two active GEF projects: Maiombe National Park in Angola and protected-area management and connectivity in Cameroon’s Dja landscape. Each component is annualised over its documented project term; the broader policy, livelihoods, landscape, knowledge and M&E components remain excluded.
Replaced Uganda Wildlife Authority’s older budget with the Auditor General’s FY2024/25 board-approved UGX 181.3bn amount (including the published recurrent and capital split). Removed the separately listed protected-areas project receipt because the authority-wide audit already reports donor income, avoiding a potential overlap.
Added two additional, source-specific 2024 park revenues from African Parks: US$516,000 at Zakouma in Chad and US$20,800 gross commercial income for W National Park operations in Benin. Both remain site-specific and separate from the broader sources already shown for their countries.
Added Rwanda’s separately reported Nyungwe National Park 2024 tourism revenue (at least US$2.333m). It is kept separate from Akagera’s reported park revenue and from the Dian Fossey Gorilla Fund’s Rwanda programme; no national protected-area total is inferred.
Added separately reported 2024 park revenue at Liwonde in Malawi and at Bangweulu, Kafue and Liuwa in Zambia from African Parks’ annual report. Each is a site-specific earned-revenue source, not an inferred park budget or a national total; none is merged with the countries’ agency allocations or other geographically distinct project sources.
Added a conservative Republic of the Congo park-revenue source: African Parks reports that Odzala-Kokoua tourism generated more than US$300,000 in 2024 and covered core costs. C-FIN records US$300,000 only, without inferring a total budget from the percentage, and keeps it separate from Nouabalé-Ndoki’s USAID project disbursements.
Corrected Comoros’s pre-existing IATI figure: its public record says it is cumulative, so it is replaced by the protected-area-network project’s US$4.024m grant annualised across its 60-month term. Added the direct protected-area component of the launched Blue & Green Economy project (US$5.204m over 60 months); its terrestrial-and-marine, associated-city scope is stated and is not attributed solely to the terrestrial benchmark.
Added Mauritania’s separate 2026 MRU47m allocation for Awleigatt National Park infrastructure and equipment. This direct, park-specific investment-budget line is kept distinct from IMPADRA’s arid-region PA project. Banc d’Arguin’s mixed marine/coastal support remains outside the terrestrial comparison.
Added Angola’s current protected-area-system project only for its two direct PA-management components (US$1.669m over 60 months). Livelihoods, financial-leverage, knowledge and M&E components are excluded. The new national-system project remains separate from Angola’s earlier Luengue-Luiana/Iona project envelope.
Added four current GEF Wildlife Conservation for Development allocations at the documented protected-area component level: Ethiopia’s direct PA-management component, Kenya’s PA-management/HWC component, Eswatini’s Big Five Nature Reserve component and Zambia’s Lower Kafue protected-habitats component. Each is annualised over its published term. Policy, livelihoods, disease, community-landscape, communications, knowledge and M&E components remain excluded; the additions are kept separate from each country’s public-agency or existing project source.
Added two current Mozambique GEF allocations only at their published terrestrial protected-area component level: Gilé National Park (US$1.785m over 60 months) and TRANSFORM’s Conservation Area Networks component (US$4.703m over 84 months). Mixed-project marine, community-landscape, policy, knowledge and M&E components are excluded; the sources remain separate from BIOFUND’s earlier multi-site support and Gorongosa’s reported project budget.
Added Liberia’s separate US$3.104m, 60-month GEF Northwest Liberia Landscape project, annualised from its published grant and duration. Its primary document identifies Gola National Park, proposed protected areas and adjacent lands, and implementation was launched in 2025. It remains distinct from the FDA’s whole-authority allocation and from the earlier UN programme expenditure.
Added Sierra Leone’s US$5.550m, 60-month GEF Gola Forest Landscape project, annualised from its documented grant and duration. It is under implementation and includes the National Protected Area Agency, but covers a forest reserve and associated community lands; it is therefore labelled as a broader, time-bound project allocation and not conflated with the NPAA’s audited recurrent expenditure.
Added Guinea’s separate Nimba–Bossou/Ziama protected-area project: US$5.202m of GEF grant financing divided across the documented 60-month term. The Ministry’s implementation material confirms the project has been launched. It is kept separate from the geographically distinct Bafing–Falémé project and labelled as time-bound project funding, not a national agency outturn.
Replaced Mali’s narrower 2025 Tienfala project line with the Ministry’s published 2026 XOF500m Programme 3.004 forecast for wildlife development and protection. The programme expressly covers protected-area creation, management, equipment, operating structures and infrastructure, but also supports wider wildlife functions; it is therefore labelled as a broader programme forecast, not audited protected-area expenditure. The two lines are not added together.
Added Cabo Verde’s explicitly named protected-areas maintenance component in the Council of Ministers’ 2025–27 Climate and Environmental Action Programme: CVE40.833m divided across its published three-year term. No water, infrastructure, forestry, fisheries or awareness component is included. The source does not split terrestrial, coastal and marine sites, so that limit is shown on the country source ledger.
Added Liberia’s approved FY2025 US$4.184m Forestry Development Authority allocation. The budget identifies it as Government of Liberia recurrent funding with no donor component. It is labelled as a broader forestry-and-protected-areas authority allocation rather than a protected-area-only outturn, and remains distinct from the separately reported UN protected-area programme expenditure.
Added Equatorial Guinea’s 2025 Finance-Law support to INDEFOR-AP: the XAF540m authority transfer and a separately listed XAF100m domestic project to strengthen the institute. The authority figure is transparently whole-agency and the second is a one-year project line, not audited protected-area expenditure. Both remain distinct from the separately dated GEF park-project disbursement.
Replaced Ghana’s lower Wildlife Division budget with the current published GHS24.54m allocation for the Ministry’s dedicated wildlife-resources protection and sustainable-use programme. It is clearly labelled as a budget estimate, including goods, services and assets but no separately published staff line. It is not an audited protected-area outturn, and it is not stacked with the earlier overlapping figure.
Updated Gambia’s broader biodiversity programme proxy to its published 2026 budget estimate, replacing rather than adding the lower 2025 allocation. Added the CEO-endorsed US$7.142m RECOSERV grant divided across its 60-month term. Both are explicitly scoped: the former is not parks-only, and the latter is a time-bound landscape project rather than an agency outturn.
Added Mauritania’s 2026 public-investment allocation to the named IMPADRA protected-areas project: MRU41m, comprising the published domestic and UNEP grant components. This is a current allocation for one terrestrial arid-region project, not a national system budget; coastal and marine funding remains separately excluded from the terrestrial comparison.
Added the Zimbabwe Government’s stated US$2.5m European Union support expected in 2026 for wildlife protection, human–wildlife conflict and community resilience. It is visibly labelled as wider environment-sector support rather than a ZimParks or protected-area-only budget, and remains separate from Hwange and Matusadona sources.
Added FAPBM’s separately reported 2024 support to 29 New Protected Areas in Madagascar. It is summed with the already-recorded, distinct Madagascar National Parks portfolio, yielding the fund’s reported US$5.612m for those two portfolios; neither figure is presented as a national-system outturn.
Added Egypt’s official 2025 natural-reserve self-financing revenue to the annual-source comparison: EGP600m, converted at the Ministry of Finance/CBE 2025 annual-average rate. This is labelled as a reserve-specific own-revenue flow used for reserve development, not audited operating expenditure or a wider EEAA budget. The annual-source rule now states this narrow, source-confirmed revenue treatment explicitly.
Corrected Guinea’s Bafing–Falémé annualisation against the primary GEF CEO-endorsed document: US$7.060m over 72 months, rather than the inconsistent secondary-project amount. The entry remains explicitly limited to a time-bound protected-area landscape allocation, not a national operating budget.
Added the CEO-endorsed GEF/UNDP Day and Mabla protected-areas project in Djibouti, divided by its published 72-month duration. It is explicitly distinguished from the country’s marine funding and from any national-agency budget.
Added Guinea’s six-year Bafing–Falémé protected-area landscape project and Chad’s published 2023 minimum Greater Zakouma expenditure invested in the local economy. Guinea’s overlapping cumulative transaction observations remain contextual rather than additive; Chad’s figure is labelled as a minimum for one co-managed ecosystem, not a national protected-area total.
Added Lesotho’s parliamentary-reported Environment and Forestry Ministry recurrent outturn, explicitly labelled as a whole-authority figure with wider functions, and Morocco’s five-year Ifrane National Park biodiversity project annualisation. Neither is described as a protected-area-system total or a park-only audited outturn.
Added two active, protected-area-linked GEF projects: Burkina Faso’s PÔ-Nazinga-Sissili landscape programme and Togo’s Fazao-Malfakassa landscape programme. Each is divided by its source-published project term and labelled as time-bound landscape funding, not as a national agency budget, annual outturn or full protected-area-system total.
Replaced Nigeria's hearing-reported personnel-and-overhead subtotal with the official 2025 Treasury/GIFMIS budget amounts for National Park Service headquarters and each named national-park unit. The result is labelled as a full-year allocation that includes capital and may reflect revisions, not as audited expenditure; the incomplete January–October payment total is not annualised.
Added Guinea-Bissau's approved GEF/IUCN Cantanhez National Park project as a clearly labelled five-year grant annualisation. It is park-and-landscape project funding, not an audited IBAP outturn or a national protected-area-system total.
Added the approved GEF/IUCN Obô Natural Parks project to São Tomé and Príncipe as a clearly labelled five-year grant annualisation. It replaces no agency outturn and is not represented as audited annual expenditure or the parks' full operating budget.
Added the documented 2024 share of Pendjari National Park's carbon-credit revenue allocated to park-management costs. Community and government shares of that revenue remain excluded, and the park-specific funding remains distinct from Benin's CENAGREF public subsidy.
Added Mali's published 2025 domestic investment allocation for the Tienfala parks-development project. It is shown as a one-complex, time-bound project allocation, not as a national protected-area budget or operating outturn.
Added Burundi's 2024/25 approved OBPE allocation alongside the existing Kibira National Park project line. OBPE manages protected areas but also has wider environmental and forestry functions, so the country brief labels this as a whole-authority allocation rather than a protected-area-only outturn.
Replaced Sierra Leone's 2025 National Protected Area Authority estimate with the government's audited 2024 recurrent outturn: separately reported personnel and goods-and-services spending. The source has no authority-specific capital actual, so no capital figure was inferred or added.
Added the Central Bank of Libya's 2025 National Parks Administration expenditure and Niger's published 2025 protected-area project allocations. Libya is labelled as agency-wide spending that includes non-terrestrial activity; Niger's two entries are explicitly time-bound project allocations, not a national-agency total.
Replaced Namibia's earlier current-spending-only estimate with the 2025/26 approved Wildlife Management, Protected Areas and Conservation programme allocation, which also covers human-wildlife conflict and community-based natural-resource management. Added Uganda Wildlife Authority's published 2024/25 receipt for a protected-areas project. Both entries are labelled with their programme or project scope and are not presented as audited whole-agency operating outturns.
Added a published 2025 Gorongosa protected-area project budget to Mozambique and replaced South Africa's government-transfer-only figure with SANParks' audited 2023/24 full expenditure. The latter remains explicitly marked as a whole-agency total that includes marine-park activity as well as terrestrial national parks.
Expanded the research supplement with published 2024–25 park revenue and operating-funding lines for Rwanda, Malawi, the Central African Republic and Zimbabwe; replaced Kenya's earlier recurrent-only KWS figure with its audited 2023/24 total expenditure; and replaced Cameroon's narrow public-investment line with the 2025 dedicated wildlife-and-protected-areas programme allocation. Country entries state whether the amount is a full agency total, a park-specific revenue source, or a restricted operational share; excluded shares remain excluded.
Added a machine-readable research supplement for new annual, country-specific funding evidence and merged it deterministically into the published country records. Its first entries add Virunga National Park’s published 2025 operating budget to the Democratic Republic of the Congo and replace Malawi’s partial Parks and Wildlife cost-centre sum with the wider 2025/26 national Wildlife Management and Conservation programme allocation. Both retain their source and scope limits.
Replaced Zambia's 2023 Wildlife Conservation and Management allocation with the newer 2025 Appropriation Act figure: ZMW 388.23m. Its USD equivalent uses the arithmetic mean of the twelve Bank of Zambia monthly 2025 ZMW/USD mid-rates, published in the Bank's 2025 annual report. Zambia's annual sourced total changed accordingly.
Updated country-brief labels and source text to state the June 2026 US-dollar planning reference consistently. No calculation changed.
Added Zambia's 2023 Wildlife Conservation and Management programme allocation from the Ministry of Tourism's official budget statement: ZMW 300.2m, converted at the World Bank's 2023 annual-average rate. It is labelled as an approved national programme allocation, not an audited protected-area operating outturn, and is shown separately from Conservation Lower Zambezi's audited operating costs. Zambia's sourced annual total changed accordingly.
Changed the published and applied planning reference from $1,095/km²/year in fixed 2018 dollars to the published rate/km²/year in the stated dollar vintage. This carries forward the prior transparent CPI conversion of Correa et al.'s $1,034 2015-dollar median, then expresses it in the live dollar vintage. All country needs and portfolio totals were rebuilt; the funding-source and abstention rules did not change.
Removed the redundant coverage stat strip from country briefs. The coverage chart now carries the terrestrial and marine values once, alongside the target-distance explanation. Simplified the C-FIN opening sentence to describe identified public-source funding rather than imply a complete measure of funds reaching every country. No calculation changed.
Removed repeated summary statistics from the C-FIN landing page and the top of every country brief. Each measure now appears only in the relevant table column or thematic band, where its scope and source are explained. Removed the current-dollar equivalent from the country funding strip. No calculation changed.
Removed the aggregate annual-need headline card from C-FIN. The total remains available as a reproducible portfolio result, but is no longer presented as a top-page headline. No calculation changed.
Simplified the benchmark panel on C-FIN by removing the five numeric comparison cards. The full rate, CPI conversion, study median, observed-spend figure and lion sensitivity remain documented here. No country need, funding figure or gap calculation changed.
Expanded the country evidence ledger with source-level GEF/IATI records for Guinea and Guinea-Bissau. The donor activities explicitly describe their reported disbursement series as cumulative, so C-FIN labels them as dated cumulative project-funding observations, not annual disbursements or national operating budgets. They are visible on the country pages but do not change the annual comparison.
Added a Latin America comparator section. It explains how outcome evidence from Ecuador, 30x30 scenario modelling and LAC finance assessments inform a future stress-tested scenario tool without being averaged into, or substituted for, C-FIN's Africa-wide benchmark. No country need, funding figure or gap calculation changed.
Moved version history to the end of this methodology. Added the full evidence chain for the June 2026 dollar equivalent, the separate roles of the elephant, lion and Congo Basin studies, and explicit non-use rules for biomass, biomes and species or protected-area polygons. No country need, funding figure or gap calculation changed.
Replaced the all-estate headline benchmark with $1,095/km²/year in 2018 US dollars. This is a transparent CPI rebase of the $1,034 2015-USD median in Correa et al. (2024), an outcome-linked study of 80 protected areas across 25 sub-Saharan African countries, including forest and savannah elephant systems. The prior $1,271 figure was found to be a 2015-dollar lion-landscape threshold, not a 2018-dollar all-biome figure. It is now retained as a separately labelled high-intensity sensitivity at $1,347 in 2018 dollars. Added a public machine-readable benchmark evidence dossier with the selection rule, source values, conversions and limits. Every country need and portfolio total was rebuilt.
Added a separately labelled “Other dated direct funding evidence” section to country ledgers. It shows direct protected-area disbursements for Angola, Benin and the Democratic Republic of the Congo where combining different years, project envelopes or potentially overlapping sources would create a false annual total. None of these records changes the annual comparison.
Withheld Somalia and Eritrea from the C-FIN comparison table. The World Bank coverage series returns 0.0 where terrestrial coverage is unreported, so neither country has a defensible planning benchmark. Their underlying records remain available for future sourcing; no funding calculation changed.
Added separately labelled marine-protected-area funding for Mauritania and a qualified parks-entry revenue context for Lesotho. Neither is added to the terrestrial annual comparison; both retain native amounts, conversion method and source limits.
Clarified the benchmark panel on C-FIN with plain-English help controls for the published threshold, observed mean, median actual spending and current-dollar sensitivity. No calculation changed.
First logged release. Added source-level annual funding evidence on country pages, including public agency budgets, NGO operating support, dated country-specific project disbursements and explicitly labelled annualised project allocations. Added separate non-counted categories for own revenue, marine funding, related ecosystem funding and multi-country funding without a country share. The annual comparison remains terrestrial only and never divides a regional total between countries.
C-FIN is a directory of country briefs; open any country from C-FIN for its full six-band view. ← Back to C-FIN