What should you plant, and how far apart? A reader asks.

A reader wrote in with the following question:

The short answer is that both people are right about different kinds of land, and neither is telling the whole story about balsa. The longer answer is that the question of three meters versus five meters is the wrong question to start with. Site condition and species choice together determine the spacing. Spacing alone is a downstream parameter, and the specific claim that a dense balsa overstory shades native seedlings into existence is mechanistically wrong on the kind of site most Costa Rican landowners are actually trying to reforest.

A harder question than it looks

Costa Rica has more than thirty years of peer-reviewed research on reforestation with native species. Most of it is concentrated at the Organization for Tropical Studies station at La Selva in the Caribbean lowlands and at Las Cruces in Coto Brus, in the southern premontane. A landowner in the southern Pacific zone is sitting within an hour or two of one of the best-studied tropical restoration laboratories in the world.

Almost none of that published research directly tests three meters against five or six meters in the specific species mix the reader was quoted. The evidence base forces an inferential answer rather than a recipe. And the two recipes she is comparing come from different intellectual lineages. The three meter prescription is borrowed from commercial balsa silviculture in Papua New Guinea and Ecuador, where trees are clearfelled at five to six years for composite wood cores. The five to six meter prescription is closer to the practitioner norms recommended by the Framework Species Method and by the Costa Rican agroforestry literature for species with wide spreading crowns. The two numbers come from different theories of what a planted forest is for.

The choice between those two theories depends on what the site is and what the planting is for. To answer the reader's question, start with what balsa actually is.

Why plan a spacing strategy at all?

One objection is worth meeting head-on before any of the rest of this matters. In mature Costa Rican forest, you can walk past big trees growing a meter apart from each other, and the forest has sorted itself out without anyone measuring spacing. Why plan a strategy at all? If two trunks can stand shoulder to shoulder as adults, why should a planter fuss over three versus five meters on a pasture?

The observation is empirically correct. The answer starts with who survived. Those two large adult trees standing a meter apart are not the blueprint of the forest. They are the survivors of their cohort. The parent stand dropped thousands of seeds into the same few square meters over many years. Most were eaten or rotted. Of the seedlings that germinated, seedling-bank studies at La Selva and similar neotropical sites find roughly sixteen thousand woody seedlings per hectare below the one meter height class, against around four to five hundred adult stems per hectare above ten centimeters of stem diameter. That is a juvenile-to-adult ratio of about thirty to one, and for the largest emergents it climbs closer to a thousand to one. The pair the reader saw shoulder to shoulder are a fraction of a percent of their original cohort that made it through. The rest are compost.

The thinning happens through mortality. Reineke described the general rule in 1933: in any even-aged stand, as the trees grow larger they run out of crown and root space, and the stand must shed stems to stay within a self-thinning envelope. Every tree that dies in that envelope is a tree that somebody either planted and paid for, or that the forest produced out of its own seed rain for free. When the forest is doing the thinning on its own, it costs nothing. When a planter is doing it, every dead seedling is wasted money, wasted seedling stock, and wasted carbon. In a fifteen-year restoration trial at Coto Brus in southern Costa Rica, discussed in more detail below, Karen Holl and Rakan Zahawi planted four species on former cattle pasture: two slow-growing native timber trees and two fast-growing nitrogen-fixing legumes. By year twelve to fourteen the two legumes had dropped to around fifty percent survival, and that mortality was a planned feature of the design, built in to open light for the slower native timbers growing alongside them. In the Slesak 2025 Pacific Northwest trial, tight one to two meter plantings of Douglas-fir lost relative survival steadily across forty years as density-dependent mortality caught up. The forest sorts itself out, but it sorts itself out through death. A planter's job is to pick a starting density and species mix that minimizes the pointless deaths between year zero and canopy closure.

The other half of the answer is that whether the reader needs a planting strategy at all depends on what kind of site she is actually looking at, and the rest of this article is mostly about how to tell. The "big trees a meter apart" observation describes a site that has finished its succession. It is not an argument against planning a restoration strategy for a pasture that has not started one.

What balsa actually is

Ochroma pyramidale is a pioneer species in the mallow family, native to Costa Rica and found in all lowland, disturbed sites on both coasts. Its natural stand density in the wild is roughly two to three trees per hectare. A commercial plantation packs around a thousand trees into the same area, three orders of magnitude denser than anything the species does by itself.

Balsa grows up to 27 meters in ten to fifteen years, flowers by its third year, and rarely lives past thirty or forty. Older trees develop a water-filled core called "water-heart" that disqualifies the wood from commercial use. Its wood density is the lowest of any commercially measured hardwood, around 0.16 grams per cubic centimeter, and its natural durability is rated seven of seven, the lowest class in standard tropical timber scales. Once a balsa tree dies, the wood decomposes rapidly.

The crown is the critical detail. Trees of Costa Rica's Pacific Slope describes balsa architecture in a single load-bearing sentence: "Branches in this species are thick but they are relatively few in number and the crown they produce is often narrow and always thin." Balsa is fast in height but thin and narrow laterally. A three meter balsa stand produces a tall open upper canopy with sparse lateral cover. Density-packing at three meters is the mechanism that produces any canopy shade from balsa at all.

Balsa cannot fix atmospheric nitrogen. It sits in the mallow family, not the legume family, and forms no rhizobial root nodules. That matters because a "nurse tree" is usually expected to do at least two things: throw shade over young seedlings, and supply nitrogen to the soil around them. Balsa cannot do the second one at all. Its litter is carbon-rich, lignin-heavy, and low in nitrogen, so it decomposes slowly and ties up soil nitrogen rather than releasing it.

The canonical silvicultural reference for the genus is the ITTO Balsa Manual, written for Papua New Guinea's national balsa industry. It says the point directly:

Field evidence matches. Vleut and colleagues in 2013 compared ten to fifteen year old balsa-dominated secondary forests in Chiapas with diverse secondary forests at the same stage of succession. Canopy openness under the balsa stands was greater than fifteen percent. Leaf litter accumulated to more than ten centimeters deep. Understory shrub diversity and tree seedling density were both lower in the balsa-dominated stands. The authors' conclusion was that balsa stands "constrained the establishment of understorey tree species, especially late-successional species, decreasing successional development." The one published study that measured what happens under a balsa canopy found the opposite of what the reader was sold.

Where the three meter number actually comes from

The ITTO Balsa Manual has two spacing prescriptions. The standard schedule for unimproved seed stock is 2.5 by 2.5 meters, which works out to 1,600 stems per hectare. The recommended schedule for genetically improved stock is 3 by 3 meters, or 1,110 stems per hectare. Both schedules clearfall the stand at five or six years. The three meter spacing is the upper end of the pulpwood range, not the lower end of a restoration range.

The manual is explicit that initial spacing beyond three meters using unimproved seed stock slows bole length production and increases stem defects. Three by three meters only works as a spacing for improved stock on a short rotation. The rotation ceiling is hard. Red Heart rot "can occur in trees as young as 3 years old but it is usually more prevalent in trees from 5 years onwards. The presence of this degrade at this age has influenced the choice of clear falling commercial plantations at age 6 years." Commercial balsa plantations do not exist beyond ten years because the wood stops being usable.

The three meter prescription comes as part of a package: improved seed stock, five to six year rotation, clearfell, and no intercropping after month four. A Costa Rican project that borrows the three meter number without the rest of the package is importing a pulpwood spacing into a context where it was never silviculturally derived.

What actually works as a nurse crop in Costa Rica

The longest-running relevant trial in Costa Rica is the "Islas" experiment run by Karen Holl at UC Santa Cruz and Rakan Zahawi at the University of Hawaii. It is built on fifteen fifty by fifty meter plots established in 2004 on former cattle pasture and coffee land around Agua Buena and Coto Brus. The experiment exists because a handful of southern Costa Rican smallholders agreed to let the researchers fence off quarter-hectare patches of their land and leave them alone for the next fifteen years. That cooperation, more than any grant or institution, is what has made the trial the closest thing the country has to a direct empirical answer for the reader's question. It has generated more than a dozen peer-reviewed papers.

The four species they planted were Terminalia amazonia, Vochysia guatemalensis, Inga edulis, and Erythrina poeppigiana. The last two are naturalized nitrogen-fixing legumes. Balsa was not used. Holl and colleagues wrote in 2020 that the two legumes were chosen explicitly "to provide rapid shade cover and increase soil nutrients." That is the function the published Costa Rican science actually assigns to a nurse crop.

Nichols and Carpenter in 2006 interplanted Inga edulis with Terminalia amazonia at San Vito de Coto Brus. At four, eight, and eleven years, Terminalia grew significantly better in mixture with Inga than in pure stands, and foliar nitrogen in Terminalia leaves was more than double next to Inga than in controls. The facilitation pathway from a legume to a companion native timber species was measurable in the wood.

Inga edulis fixes roughly 124 to 168 kilograms of nitrogen per hectare per cycle from prunings alone. Erythrina poeppigiana fixes around 100 kilograms of atmospheric nitrogen per hectare per year, enough that its pruned biomass provides the recommended nitrogen application rate for coffee in Costa Rica. Gliricidia sepium can raise soil total nitrogen by 180 kilograms per hectare per year in silvopastoral systems. These are the actual nitrogen numbers behind the words "soil fertility." Balsa contributes zero.

Cole and colleagues in 2011 tested direct seeding of five large-seeded native trees into three site types: bare pasture, young secondary forest, and a three year old plantation of the four Holl and Zahawi species. Year two seedling survival was 75% under the nitrogen-fixer plantation, 44% under secondary forest, and 41% in open pasture. Seedlings under the nitrogen-fixer canopy were 22 to 31 percent taller. Foliar nitrogen was 1.96 percent under the plantation versus 1.19 percent in pasture. Cost per 100 surviving seedlings was $4.55 to $8.55 under the nitrogen-fixer canopy and $116.75 to $182.25 for nursery-grown transplants. The evidence-based Costa Rican nurse-crop prescription is to plant legume nitrogen-fixers first, then direct-seed large-seeded late-successional species under their canopy three years later. That is the two-stage approach, and it does not involve balsa.

Crown architecture is the spacing variable

The real variable behind any spacing prescription is how quickly a tree's crown covers lateral area and how deeply it shades the ground beneath. The Framework Species Method, developed at the Forest Restoration Research Unit of Chiang Mai University and now applied across the tropics, is explicit about this. Elliott and colleagues wrote in 2023 that "ideal spacing among planted trees and natural regenerants depends on tree-growth rates, especially crown expansion, to shade out weeds." Practitioners use closer spacing in drier climates, around one meter, and wider spacing in wetter climates, up to three meters, because tree crowns expand faster under higher rainfall. There is no universal right number.

Vochysia guatemalensis, known in Costa Rica as mayo for the yellow flowers that cover its crown every May, has a round spreading crown. The World Agroforestry profile for the species recommends a four by four meter spacing and is explicit that the distance "should not be less than this, as the tree tops close very quickly." A three meter Vochysia stand is already too tight to let the species do what the species does. Inga edulis forms a broad, flat, moderately dense canopy and will close a four meter alley in standard alley-cropping trials. A single Inga shades out grass over a much larger lateral area than a single balsa, in less time. Balsa, again, has a crown that is narrow and thin. The density-packing at three meters exists to compensate for a crown that does not spread.

Jucker, Bouriaud, and Coomes reported in 2015 that crown plasticity is the strongest driver of canopy packing in diverse mixtures: most tree species expand their crowns when growing next to neighbors of different architectural types. A monoculture of narrow-crowned emergents has no vertical stratification and no crown-plasticity benefit. Mixing a flat-crowned Inga with a round-crowned Vochysia and a narrow-crowned Terminalia packs canopy space more efficiently at wider spacing than any monoculture can at tight spacing. Close crowns are the mechanism, and several species native to Costa Rica get you there at four meters that balsa cannot manage at three.

Site type is the prior variable

The most load-bearing finding in the Costa Rican and global restoration literature is that before species and spacing, there is the site. Robin Chazdon's synthesis in Second Growth and Stephen Elliott's 2023 Framework Species Method review both organize restoration approaches along a ladder keyed to how degraded the starting site is. A minimally degraded site needs no planting. An assisted natural regeneration site needs fire exclusion and maybe some enrichment. Moderately degraded pasture within reach of forest fragments is where the Framework Species Method fits. Severely degraded or arrested-succession sites, where a bracken fern carpet or an Imperata grass thicket has held the land in an alternative stable state for years, need soil amelioration, a nurse crop, and then diverse planting.

Holl and colleagues in 2018 derived a practical diagnostic from the Coto Brus data. Two variables measured a year and a half after cattle were removed, how much of the ground is still dominated by pasture grass and whether natural woody regrowth (young pioneers and shrubs that came up on their own) is already pushing up through the grass and starting to shade it from above, together explained between 47 and 87 percent of the variation in forest recovery at eight and a half years. Surrounding forest cover, prior pasture duration, and soil variables explained very little additional variance once that early signal was known. The practitioner's test is to walk the site eighteen months after the cattle leave. If the grass is still unbroken and nothing woody has come up through it, plant actively. If volunteer pioneers are already climbing above the grass line and closing a layer over it, the site will probably get there on its own.

Jiménez-Rojas and colleagues asked a related question of the same Coto Brus plots in 2022: how far out from a restoration site does the surrounding landscape still shape what grows back? They measured forest cover in circles of increasing radius around each plot and looked for the circle size that best predicted two things. For the arrival of late-successional tree seedlings, the strongest predictor was the amount of forest within a neighborhood of roughly 312 meters. Forest further out than that added almost nothing to the signal: seedling recruitment is a local process, driven by whatever forest sits within a few hundred meters. For the presence of forest-dependent birds on the plot, the relevant neighborhood was larger, around 698 meters. Birds integrate a wider area of habitat around them than seeds do, so their presence on a plot reflects how much forest there is in a wider ring of landscape. The practical takeaway is that closer forest matters more than forest far away, and if there is any fragment of forest within walking distance of the planting, both volunteer seedlings and seed-carrying birds will reach the site and begin filling in the spaces between the trees the landowner planted. If the nearest remnant is a kilometer or more away, volunteer recruitment thins out sharply and every late-successional species has to be put in by hand.

Where balsa actually belongs is the arrested-succession case, and the evidence base for that use is older than any peer-reviewed paper. The Lacandon Maya, who still farm the Selva Lacandona in Chiapas, have sown balsa seeds into bracken-fern-infested fields for generations as part of a long-fallow agricultural cycle. The tree's huge leaves and rapid growth smother the fern, its litter enriches the soil, and within a decade the cleared ground is handed back to the forest. Douterlungne and colleagues in 2010 measured what Lacandon farmers had been observing for a long time: balsa transplanted at two by two meter spacing into bracken clearings reached 92 percent first-year survival. Levy-Tacher and colleagues followed the same system for four years and documented complete bracken rhizome eradication in plots with balsa plus biweekly or monthly weeding. The authors emphasized that the long-term outcome depended on what natural regeneration came in under the balsa, which is to say the Lacandon practice only works when the dying balsa is replaced by the surrounding forest's own seed rain. The balsa-as-weed-killer prescription is not in dispute. It is the evidence base that the carbon credits company is implicitly invoking when it pitches eighty percent balsa at three meters, whether or not the company has ever heard of the Lacandon. If the reader's land near Uvita is moderately degraded African-grass pasture within reach of a forest fragment, that evidence base does not apply to it.

What the reader can check on her own land

Three checks can be done on foot, before any contract is signed. The first is the distance to the nearest forest fragment. If there is continuous or near-continuous tree cover within a few hundred meters, the Zahawi seed-rain data suggest that the restoration will benefit strongly from natural seed input during the first decade, and a wider spacing with a smaller planted mix becomes defensible. If the site sits in the middle of an unbroken pasture with no remnant tree within walking distance, spacing will not save it. The reader will have to plant densely and use every disperser-attracting trick available, including perching snags and fruiting trees positioned to draw in birds and bats from whatever forest is reachable.

The second check is the grass. Most of Costa Rica's post-pasture landscape is dominated by introduced African tussock grasses in the Hyparrhenia, Pennisetum, Melinis, and Brachiaria genera. These form dense swards up to three meters tall and are aggressive, but they are not in the same category as a bracken fern carpet or an Imperata thicket. If the land is African-grass-dominated pasture, it falls into the moderately degraded category. If it has been invaded by a woody fern carpet or a sugar-cane monoculture that the cattle never ate, it falls into the severely degraded category. A half day of walking will tell her which.

The third check is the horizon of her goal. The three meter balsa prescription assumes clearfelling at year five or six. If she wants a forest at year forty, the balsa cohort will be dying back well before then, and whatever grew up underneath it will have to carry the canopy forward. The five to six meter mixed prescription is designed to answer the year-forty question directly. The two recipes answer different questions about what the land is for. She should pick her question before she picks a planter.

One honest caveat about the second planter's recipe. The peer-reviewed literature in Costa Rica does not report a direct canopy-closure timeline for a uniform five to six meter grid in a diverse native planting. The wider recipe has practitioner tradition behind it, but no published spacing trial that the reader can look up. The closest empirical analog is the Holl and Zahawi applied nucleation treatment at Coto Brus, which planted eighty-six trees on a fifty by fifty meter plot (around 344 trees per hectare, the same stem count as a 5.4 meter grid) but clustered them into small islands rather than spreading them evenly. Inside those islands, canopy cover reached above ninety percent by year fifteen. The ground between islands was still filling in at year seven to nine, with drone measurements showing about forty-five percent matrix cover above two meters.

The wider-spacing recipe operates on a ten to fifteen year closure timeline, longer than the three to five year timeline of a dense planting, and it depends on the site being close enough to a forest fragment for volunteer seedlings to bridge the gaps between the trees the landowner put in. That is a coherent choice as long as the site and the patience are both there.

There is a social detail worth holding in mind, and it comes from the farmers themselves. Holl and colleagues recorded that the Coto Brus smallholders whose land hosted the experiment often described the sparsely-planted, volunteer-dependent plots as "messy" or "unused" compared with the neat rows of the full plantation treatment. The slower, cheaper, more natural recipe carries a social cost that a landowner has to be willing to pay. A pasture that looks like a pasture with a few saplings scattered across it, even when it is on its way to becoming a forest, does not always look like work getting done. Neighbors walk by and see abandonment where the research sees a functioning restoration design. That perception is part of the real price of the wider recipe, and a landowner who picks it should weigh it in advance rather than discover it later.

The carbon math problem

Balsa wood stores roughly 75 kilograms of carbon per cubic meter of stemwood. Teak stores about 259. Plantation eucalyptus about 273. For the same volume of wood, balsa holds roughly a third as much carbon as a dense-wooded tropical timber species. Faster growth does not automatically translate into faster carbon sequestration once wood density is accounted for.

Brienen and colleagues in 2020 analyzed 110 species from tropics to boreal forests. A 50 percent increase in early growth rate corresponded on average to a 23 percent reduction in maximum lifespan. Seventy-four of 82 species showed negative growth-lifespan correlations. Fast pioneers store less carbon in the long run because they die sooner. Büntgen and colleagues in 2019 reported the same effect at stand level: faster-growing tropical Borneo sites contained less aboveground carbon density than slower-growing sites. Carbon residence time matters more than annual productivity.

Lewis, Wheeler, Mitchard, and Koch calculated in 2019 three scenarios for the Bonn Challenge target of 350 million hectares under restoration. Natural forest regeneration across that area would store roughly 42 petagrams of carbon by 2100. Commercial monoculture plantation would store about one petagram. Agroforestry would store around seven. The ratio between natural forest and commercial plantation on the same land is about forty to one.

Drought physics lean the same way. Hutchison and colleagues in 2018 reported that during the 2013 to 2015 El Niño dry sequence at the Sardinilla experiment in Panama, monocultures had significantly higher mortality than mixtures. Five-species plots showed the smallest growth decline and no early-warning spectral signals of potential die-off. Powers and colleagues in 2020 measured 2015 drought mortality in Guanacaste dry-tropics species and found that hydraulic safety margin, not wood density alone, predicted which species died. Species with large xylem vessels, the end of the spectrum where balsa sits, are hydraulically vulnerable. Costa Rica's Pacific coast is drying. The 2015 drought dropped rainfall at Santa Rosa to 39 percent of its long-term mean. Any species list chosen now for the southern Pacific has to hold up through thirty more years of that trend.

A Costa Rican coda

Two observations tie the general story back to Costa Rican specifics. The first is that the FONAFIFO Pago por Servicios Ambientales program sets 816 trees per hectare as the planting density for its reforestation payments, which works out to roughly 3.5 by 3.5 meter spacing. The number has been treated as a received standard for three decades. Its provenance traces to Maderas Preciosas Costa Rica, which adopted it in 1995 for commercial teak and pochote plantations in Guanacaste, derived from Miller's 1969 teak growth table. It is industrial teak silviculture applied across the board to natives, exotics, and endangered species alike, without published ecological derivation for any of them. Ochroma pyramidale does not appear on any FONAFIFO species list. Voluntary carbon credit projects operate outside the Pago por Servicios Ambientales framework and are not required to follow its rules, species lists, or densities.

The second is the balsa stand around the Sirena ranger station in Corcovado National Park. A popular account holds that the trees are descendants of a World War II balsa plantation started by the Allies and then abandoned when aluminum replaced wood in combat aircraft. The primary documentation for that story has not been located. Paul Allen's 1956 Rain Forests of Golfo Dulce, the canonical Osa botanical survey written eleven years after the war, does not mention any wartime plantation, Corcovado, or Sirena. Ecuador supplied the Allies with balsa for the de Havilland Mosquito. The United States wartime strategic-plant program in Costa Rica was abaca, a fiber substitute for Manila hemp, not balsa. Sirena was Don Félix Avellán's private cattle ranch until 1975, when Decree 5343-A created the park. The simplest explanation for the balsa around Sirena is that it is post-abandonment natural regeneration from an old pasture edge, which is exactly what balsa does across the Pacific lowlands of Costa Rica. The contemporary balsa work on the Osa, at Piro and Cerro Osa, is Osa Conservation's Framework-Species-Method planting with balsa as one of forty-plus native species, which is a restoration design, not a monoculture.

Plantation or forest: what the law actually says

One more consideration is specific to Costa Rica, and it is legal. Ley Forestal 7575 treats a plantation and a forest as different legal objects, and the difference controls whether a landowner can later change her mind.

Article 3 of the law defines bosque as a native ecosystem of at least two hectares, with canopy cover above seventy percent and at least sixty trees per hectare of fifteen centimeters of diameter or more. A plantación forestal is land of at least one hectare cultivated with forest species whose principal objective is timber production. Article 28 then says that plantations, agroforestry systems, and individually planted trees do not require a cutting permit, a transport permit, or an industrialization or export permit. Article 19's prohibition on cambio de uso, the rule that otherwise stops a landowner from clearing forest, applies only to terrenos cubiertos de bosque. Because a plantation is not bosque under Article 3, Article 19 does not reach it.

For a carbon-credit balsa plantation on private pasture, that is the legal machinery underneath the pitch. At year five or six the stand is registered as plantación, Article 28 exempts it from the permit regime, and the owner can clearfell on the planned rotation without ministerial authorization beyond a regente-issued certificate of origin. The plantation is legally cheap to reverse, and the carbon developer is counting on that.

The interesting part starts around year fifteen. The Procuraduría General de la República adopted in its Opinion OJ-050-2011 a fifteen-year benchmark for when an abandoned stand becomes a verdadero bosque, a "true forest." The Sala Constitucional had earlier held, in Voto 3923-2007, that a plantation left to natural regeneration can become forest through the inercia of its owners, and it declared Article 28 unconstitutional by omission for failing to require an inspection before cutting. The Sala ordered the Asamblea Legislativa to enact a precautionary inspection regime. Eighteen years later, the legislature has not complied, and Costa Rica faces a petition before the Inter-American Commission on Human Rights (P-746-17) over the non-compliance. The doctrine stands even though the statutory fix has not arrived.

The practical effect is that the difference between a five-year balsa rotation and a fifteen-year diverse native planting is not only silvicultural. The balsa rotation is timed to harvest the stand before it can legally become a forest. A diverse native planting designed to still be standing at year twenty is betting on the flip. Once the stand crosses the Article 3 thresholds of canopy cover, stem density, and native-ecosystem character, a regente can no longer honestly certify it as a plantation, and Article 19 locks in against clearfell. The landowner has traded a cheap-to-reverse plantation for an expensive-to-reverse forest. For a reader whose goal is a forest at year forty, that is the right trade. For a reader whose goal is a repeating five-year carbon rotation, it is the wrong one.

Closing

To the reader's three questions. Both of the people she spoke with are right about different kinds of land. The three meter balsa recipe answers a different question than the five to six meter diverse recipe, and the carbon credits company has borrowed a short-rotation pulpwood spacing for a purpose it was not silviculturally designed for. She can tell which recipe fits her land by walking it, by looking at the grass, the distance to the nearest forest fragment, and the trajectory of any canopy that has already begun to form. The year-and-a-half canopy-closure test is a practitioner diagnostic she can apply herself.

A research-aligned planting for moderately degraded pasture in the southern Pacific of Costa Rica, within reach of forest fragments, looks like a mix of Vochysia guatemalensis, Terminalia amazonia, Inga edulis, and Erythrina poeppigiana, planted at roughly 2.8 to 4 meter spacing depending on rainfall, with direct seeding of large-seeded late-successional species under the canopy after year three. That is the Holl and Zahawi Coto Brus prescription, and it has fifteen-plus years of published Costa Rican evidence behind it. She is sitting in one of the best-studied tropical restoration landscapes in the world. The answer is already in the literature.