Most people who eat poi today have eaten one variety of taro. They may have eaten it their whole lives and never known there was a choice.
That wasn’t always true. Poi used to taste of somewhere. Different islands, different valleys, different water produced recognisably different poi, and people had opinions about it the way people elsewhere have opinions about apples or olive oil. What happened to that variety — and where Hawaiʻi’s taro is grown now — is a story about water, land prices, and one shipment of huli — cuttings, not seedlings — that left the islands.
Hundreds of kalo, not one
Hawaiians named and maintained an extraordinary catalogue of taro. Traditional accounts put it at as many as 300 named varieties, distinguished by corm colour, leaf shape, petiole markings, growing habit, cooking behaviour, and taste. Names like Lauloa ʻEleʻele ʻUlaʻula, ʻUlaʻula Poni, Manini Kea, and Nihopuʻu were ordinary vocabulary in Hawaiian communities in the way that variety names are ordinary vocabulary to any farming culture.
These were not interchangeable. Some were prized for poi and some for table taro; some for lūʻau leaf; some suited flooded ground and others upland slopes; some cooked sweet and moist, others dry and starchy. A valley might be known for a particular kalo the way a French village is known for a cheese.
Among them, one name became the standard for poi. Lehua maoli — maoli meaning native, common, or ordinary, though the connotation is closer to the true one — is described as a favourite variety of Hawaiians, grown throughout the islands, producing red poi of excellent quality; the widely advertised “Lehua red poi” that often commanded a premium came from this variety. The Bishop Museum notes it grows in both dryland and wetland cultivation, matures in eight to twelve months, and is said to be the original taro brought to Hawaiʻi.
That last claim is tradition rather than settled fact, but it tells you the standing the variety held.
Wet and dry
Here’s a thing that surprises people: most taro in the world is not grown in water.
Taro is grown across the tropics — Asia, Africa, the Caribbean, the Pacific — and the majority of it is upland or dry cultivation, rain-fed or irrigated but not flooded. Hawaiʻi practised this too. Upland taro, kalo maloʻo, was grown here for hundreds of years, and dryland systems fed large populations on islands and slopes where diverting a stream wasn’t possible. Poi could also be made from breadfruit — poi ʻulu — though it was considered inferior.
What Hawaiʻi developed to an unusual degree was the flooded alternative: loʻi kalo, terraced pondfields fed by ʻauwai, hand-built irrigation channels that tapped a stream above and returned the water below. It is one of the most sophisticated agricultural systems built anywhere in the Pacific, and it produced yields per acre that dryland cultivation could not match.
The water has to move
The critical detail — and the one that decides where taro can economically be grown — is that a loʻi is not a pond. It is a section of a river.
Water in a loʻi must flow continuously. If it stands still it warms, and warm standing water is a death sentence for kalo. Taro is highly sensitive to heat; if the water in a loʻi becomes stagnant or too warm, the crop can develop root rot. A functioning ʻauwai system maintains constant vigorous flow of cool mountain water, which oxygenates the soil and keeps the corms cool, letting the plant grow slowly and develop a dense, nutty flavour and creamy texture.
The mechanism is straightforward once stated. Warm water holds less dissolved oxygen than cool water, and the pathogens that attack taro roots — Pythium species, water moulds rather than true fungi — thrive precisely in waterlogged, low-oxygen conditions, and survive between crops as resistant oospores that can lie dormant in soil for years. Flow solves both halves at once: it carries heat away and it keeps oxygen up. Stop the flow and you have built an incubator.
This is why the old growers cared so much about which water. Cold, fast, high-elevation stream water made better taro and healthier plants than sluggish lowland water, and the flavour difference was real enough that people could taste which valley their poi came from.
It also means that taro land is not really land. It is land plus a reliable, legally secured flow of cool fresh water — which in Hawaiʻi is among the scarcest and most contested assets there is.
Waikīkī was a taro field
The clearest illustration of what’s been lost sits under the most photographed square mile in the Pacific.
Before the hotels, Waikīkī was a vast estuary fed by three streams — Makiki, Mānoa, and Pālolo — descending from the Koʻolau range. Its coastal marshes, loʻi kalo, and loko iʻa (fishponds) sustained Native Hawaiians for centuries. Nearly 85% of what is now Waikīkī was in wetland agriculture or aquaculture. Kamehameha chose to live there after taking Oʻahu. It was, in the most literal sense, the breadbasket of the island.
Then the crop changed. In the 1860s and 1870s, former Asian sugar plantation workers — Chinese and Japanese — replaced the taro and farmed more than 500 acres of those wetlands as rice, also raising fish and ducks in the ponds. By 1892, Waikīkī had roughly 550 acres in rice, nearly 12% of all rice planted on Oʻahu.
This transition makes complete sense once you understand the loʻi. Wetland rice and wetland taro want nearly the same infrastructure: terraced, bunded, level pondfields with controlled water in and water out. Chinese farmers arriving from rice country found a landscape already engineered for the work they knew. The labour is comparably brutal in both — flooded fields, hand transplanting, constant bunding and channel maintenance, everything done bent over in mud and water.
And then it ended. The Ala Wai Drainage Canal, begun in 1921 and completed in 1928, drained and filled the remaining ponds and irrigated fields. The canal also destroyed wetlands that had functioned as a sponge against flooding and as a natural filtration system capturing sediment and nutrients before they reached the ocean. Everything that made Waikīkī agricultural was engineered away in seven years, and the result is the state’s economic engine.
The narrowing
Meanwhile the varieties themselves were collapsing into a handful.
The cause was commercial milling. Once poi became a manufactured product rather than a household one, poi mills — like any manufacturing operation — needed to deliver the predictable, consistent product their customers expected. Consistency means one input. Hundreds of cultivars with distinct flavours and cooking behaviours are an asset to a family pounding its own and a problem to a factory.
So the catalogue narrowed to what milled well and yielded heavily. Lehua maoli became the commercial standard — and was then displaced in turn. Maui Lehua, a cross between Moi and Lehua maoli that emerged in the 1970s, is now the most commercially grown taro in Hawaiʻi; most bags of supermarket poi and most poi served at restaurants and commercial lūʻau contain this variety. It won on agronomics: fast to mature, a hearty yielder, monocrops well, consistently large. Today’s crop is dominated by just a few cultivars — Maui Lehua and Moi for poi, and the Chinese cultivar Bun Long for lūʻau leaf and chips.
Two centuries ago you could taste the difference between islands. Now most poi is one hybrid.
The economics that push taro offshore
The acreage tells the story bluntly. At its peak, kalo farming is estimated to have covered around 35,000 acres across the islands. Current figures put it at a few hundred — roughly 557 acres harvested across almost 270 farms in 2022, which was just seven acres more than in 1991. Thirty years of effort has produced essentially no net growth.
The demand is there. Hawaiʻi consumes something like 6.5 million pounds of taro a year, local farms supply around three-quarters of it, and the state imports roughly 2 million pounds annually. Kalo was worth about $1.24 per pound in 2021 — a good price for a vegetable. So why doesn’t production expand to meet it?
Because the inputs are the most expensive versions of themselves anywhere in the United States.
Land with water rights. Not just agricultural land — agricultural land with a secured, flowing supply of cool fresh water, in a state where water allocation is litigated for decades and where much agricultural land is held by large landowners not using it for farming. Despite 47% of Hawaiʻi’s land being zoned agricultural, satellite and field survey work found only about 8% actually growing crops.
Labour. Loʻi work resists mechanisation almost completely — it is hand work in mud, and it always has been. Hawaiʻi layers onto that some of the highest labour costs in the country, including the state’s Prepaid Health Care Act, which since 1974 has required employers to provide health coverage for employees working over twenty hours a week, plus mandatory temporary disability insurance. These are good laws. They are also real costs, and they fall hardest on labour-intensive agriculture with thin margins.
Put those together and you get a crop that is culturally central, in genuine demand, priced decently — and still barely expanding, because nobody can assemble the land, water, and labour to grow more of it at a price the market will bear.
Which brings us to Honduras
If the constraint is land, water, and labour rather than the plant itself, the obvious question is what happens when you take the plant somewhere those things are cheap.
So we tried it. In July 1999 we sourced 250 huli — the cut tops of harvested kalo, the traditional planting stock, which regrow into new plants — of Lehua maoli, and flew them to Honduras. Honduras’s national agricultural research foundation, FHIA, took on the cultivation at San Pedro Sula. Eighteen months later, in January 2001, they air-freighted back two boxes of hand-peeled taro — 145 kilos, San Pedro Sula to Miami to Dallas to Honolulu.
Several things about that are worth sitting with.
The variety travelled well, which shouldn’t be surprising — Lehua maoli grows in both wetland and dryland conditions, and Central America is well within taro’s climatic range. Central America is in fact an established taro-growing region already, largely for export to North American markets.
It travelled better than that. In October 2000 Dr Ramón de la Peña of the University of Hawaiʻi — the taro breeder who advised both sides — inspected the Honduran plots and reported that the plants there were larger than Hawaiian plants of the same growth age, and that the growth cycle looked likely to be shorter.
Hand-peeled is the phrase that does the most work in that sentence, and it deserves its own explanation.
The peeling problem
Raw taro fights back. The corm carries needle-shaped calcium oxalate crystals that irritate on contact — they irritate the throat and mouth, which is why taro must be cooked thoroughly before eating — and prolonged handling of raw corms makes your hands sting and itch. It is genuinely unpleasant work.
So most operations don’t do it. They cook the taro with the skin on, and the skin then slips off easily from the cooked corm. It’s faster, cheaper, and kinder to the crew.
It also costs you flavour. Cook the skin on and some of what’s on the skin ends up in the pot — a faint muddy, earthy note that carries through into the finished poi. Anyone who eats poi seriously can taste it. It is the difference between poi that is fine and poi that is clean.
We peeled raw instead: a Hobart abrasive potato peeler for the bulk of it, then finished by hand, because a drum peeler cannot follow the irregular contours of a taro corm the way it can a potato. Slower, more labour, more discomfort. Distinctly better poi.
That trade-off is a fair miniature of the entire industry. Nearly every shortcut available in taro processing costs you a little flavour, and the market rarely pays enough to justify refusing them. Peeling raw was the single most expensive processing decision we made, and it happened to be the one that made the biggest difference in the bowl. It is also, not coincidentally, the step that is prohibitively expensive in Honolulu and viable elsewhere. The whole offshore economic case sat in that hyphenated adjective.
And it worked, in the narrow sense. The taro came back. It was good.
You can export a cultivar, not a watershed
But return to the water for a moment.
If the flavour of poi genuinely depends on cool, fast stream water — if Waiāhole tastes like Waiāhole because of the temperature and oxygen of what runs through its ʻauwai — then a variety grown somewhere else is not the same product, however faithfully the genetics were preserved. You can export a cultivar. You cannot export a watershed. Whatever came back from Honduras was Lehua maoli by name and lineage, but it was not Lehua maoli grown in Hawaiian water, and by the logic the old growers applied, that is a different taro.
There is a second question, less technical and harder. Kalo is not merely a crop here. In the Hawaiian genealogical account, Hāloa — the taro plant — is the elder sibling of the Hawaiian people, and the relationship of care between them is the model for the relationship between people and land. Sending huli of a heritage Hawaiian variety abroad to be grown by a foreign government, so it can be peeled by cheaper hands and flown back, is a transaction that some will read as pragmatic problem-solving and others will read as something closer to extraction.
We found out which. The offshore taro solved the cost problem completely and the market problem not at all. People did not want poi made from taro grown somewhere else. They said so.
Not because they could taste it. A caption written at the time put it at decent, but not wonderful, and that is fair for the taro on its own. The Honduran taro made perfectly good poi — a little different from poi made with taro grown here, which is exactly what you would expect from different water and different soil. Blended with local taro, the results were indistinguishable. We know that because we did it.
And they still wanted affordable poi.
That is the whole bind in two sentences, and it isn’t hypocrisy — it’s two entirely reasonable preferences that happen to be incompatible at Hawaiʻi’s cost of land and labour. Poi is not priced like a commodity because it isn’t experienced as one. A customer buying poi is buying a relationship to a place, and the provenance is part of what’s in the bag. Take the provenance out to make it cheaper and you have removed some of what they were paying for. No amount of blind-tasting data changes that, because they aren’t wrong about what they want.
The June problem
Then there was timing, which nearly nobody outside the trade thinks about and which may be the hardest constraint of all.
Poi demand in Hawaiʻi is violently seasonal. It peaks in June — graduations, alumni reunions, the lūʻau season, the month when everybody is feeding a hundred people at once. If you sell poi, June is when you make your year.
June is also peak growing season. Taro matures over eight to twelve months, and in the months before harvest the corm is still bulking — putting on the weight the farmer will eventually be paid for. Because taro is bought by the pound, harvesting in June means harvesting a crop that hasn’t finished getting heavy. The farmer is being asked to leave money in the mud.
So farmers don’t want to harvest in June. Entirely rational. And the processor doesn’t want the alternative either, which is mountains of taro arriving in September when demand has collapsed and there is nowhere to put it.
Fresh poi cannot bridge that gap. It has a shelf life measured in days before it sours and a couple of weeks refrigerated. You cannot buy in September and sell in June.
Which is the real argument for preservation, and it has nothing to do with mainland nostalgia or convenience. A shelf-stable poi decouples the harvest calendar from the demand calendar. Buy the crop when the farmer wants to sell it and it’s at its heaviest, process it then, and hold it until the month people actually want it. The farmer gets a better price at a better time. The processor smooths a demand curve that would otherwise be unmanageable. And the June customer gets poi that isn’t scarce.
Frozen achieves part of this and brings a cold chain and freight bill with it. Freeze-dried achieves all of it — a product that sits on a shelf for years and comes back as poi on the morning of the graduation. Seen that way, freeze-drying wasn’t a novelty product at all. It was an inventory solution to an agricultural timing problem, and the fact that it also travels well to the mainland was a bonus rather than the point.
What you can move and what you can’t
Set all of it side by side and a pattern emerges. Almost every constraint on poi is a constraint of place or of time, and the two behave very differently.
Time turned out to be solvable. The June-versus-September problem is an engineering problem, and preservation technology answers it. You can hold a harvest across a calendar.
Place did not. Water temperature isn’t portable, and neither is provenance. Growing the taro elsewhere fixed the arithmetic and broke the thing people were buying — and they were not being irrational about that, because with poi the place is part of the food.
Which leaves the fact underneath all of it, and it is a fact about Hawaiʻi rather than about Honduras: a place that grows the best taro in the world on a few hundred acres, imports two million pounds a year, and cannot afford to grow more of its own staple food on its own land. The offshore experiment was a symptom. The land, the water, and the labour are the disease.
[SIDEBAR — first-hand note]
The huli went out in July 1999 — 250 Lehua maoli tops, bought from a local grower for $291.45 and invoiced on to FHIA at a quarter each as commercial samples. Moving live planting material was the slow part. It took a Honduran import permit from the Ministerio de Recursos Naturales, a US phytosanitary certificate, and the best part of a year of USDA quarantine paperwork. Taro was not on the list of crops approved to enter Hawaiʻi, which is why what came back came back peeled.
It grew, and it grew well. Two boxes — 145 kilos of hand-peeled taro — flew back in January 2001 for a processing trial. The freight came to $353.39.
The poi tasted fine. A little different from poi made with taro grown here, which is what you would expect from different water and different soil. Blended with local taro, the results were indistinguishable.
If I had the whole thing to do again I would reverse the order. Honduras was willing to spend money on this as an agricultural project. I would have concentrated there first and got ten acres or more under cultivation, so there was a steady stream of taro coming back to Honolulu. Once that supply was assured, then I would have started the Poi Company.
Reporting current as of August 2026. Historical acreage figures are estimates and vary between sources.
Also in this series