Where parks end up

Why do a country's protected wildernesses sit almost entirely on land nobody wanted to farm?

Open a map of almost any country's protected areas and the shading collects in the same places: the high ground, the steep ground, the dry ground, the far ground. The generous reading is that this is where nature survived, so this is where conservation went. The suspicious reading is that nobody was using these places, so protecting them cost nothing. Both readings predict the same map — which means the map alone cannot settle it. What would?

Think about how a parcel of land actually gets its designation, not in theory but in sequence. Long before any conservation agency exists, the good land is taken: flat, watered, fertile, near a market. Then the merely useful land goes — timber, pasture, minerals, whatever repays the cost of reaching it. Protection is not competing for land at the start of this process. It arrives at the end, and it is offered what is left.

That already predicts the pattern, but it predicts something sharper too. If protection is a residual, then the characteristics of protected land should look like the characteristics of unwanted land, and they do. Systematic comparisons across roughly 150 countries found protected areas biased toward higher elevations, steeper slopes, poorer soils, and greater distance from roads and cities — consistently, and not as a scatter of national quirks.

Now ask what follows, because this is where a curiosity becomes a problem.

First, representation. Ecosystems are not equally protectable under this rule, because they are not equally useless. The biomes on good flat soil — temperate grasslands above all, with Mediterranean scrub and lowland tropical forest on fertile ground — have been converted at enormous rates and protected at tiny ones, while rock and ice are handsomely covered. A residual system is biased against exactly the ecosystems most worth defending, since being worth farming and being biologically productive are the same property seen twice.

Second, and less obvious: it corrupts the measurement of whether parks work. Compare deforestation inside parks with deforestation outside and parks look extraordinarily effective. But the inside was chosen for being steep, remote and unprofitable. Compare a protected slope with an unprotected slope of the same steepness and remoteness and the gap shrinks sharply — work on Costa Rica using that kind of matching found protection avoided clearing on roughly a tenth of the protected forest, far below the naive comparison. Parks do work; they work much less than the raw numbers claim, because those numbers mostly report where parks were put.

Third, it makes area targets gameable. If a country must protect 30 per cent of its territory, the cheapest route to the number is more of the same worthless land — so the target can be met while representation gets worse.

So is the pattern purely economic? Not quite, and the second filter is worth naming. The early parks — Yellowstone in 1872, Yosemite before it — were chosen for spectacle: canyons, geysers, waterfalls, peaks. Nobody was counting species. Monumental scenery happens to be strongly correlated with terrain that cannot be farmed, so the aesthetic filter and the economic filter pushed in the same direction and are nearly impossible to separate in the historical record.

And one place where the pattern breaks, which is the best evidence that it is real elsewhere. Where wildlife itself became valuable — the great African savanna parks, sitting on grazing land with genuine agricultural potential — protection did outbid other uses. Opportunity cost was not zero there; it was paid, because tourism and prestige made the land worth more wild than tilled.

The question's own wording needs correcting, and this is the important refinement. "Land nobody wanted to farm" is not the same as "land nobody wanted". A great deal of what became park was inhabited, hunted, herded or burned by people whose uses were not agricultural and were not counted. Yellowstone and Yosemite were cleared of their residents; the pattern repeated across colonial and post-colonial park creation, and the critique of "fortress conservation" is aimed exactly at it. Low opportunity cost, measured in crop yield, was often high cost borne by people the accounting did not see.

Two further qualifications. The residual pattern is a strong central tendency, not a law — systematic conservation planning, which scores candidate sites for representation and complementarity rather than availability, exists precisely to break it. And "protected" covers wildly different things, from a strictly guarded reserve to a category permitting logging and settlement, so two countries with identical percentages may be doing entirely different things.

• How systematic conservation planning scores sites for complementarity and irreplaceability instead of availability.
• Why paying landowners for conservation on valuable land runs into the same selection problem, since the cheapest offers come from those who were never going to clear it.

Every term the collection defines is gathered in the glossary.