THE ANTHURIUM INFLORESCENCE

Every other genus in this cluster works the same broad way. A chamber, a smell, an insect that goes in and cannot immediately get out, and a transaction that can be described. Anthurium does none of that. There is no trap anywhere in the genus — the spadix stands in the open, the spathe is a backing sheet rather than a room, and nothing is ever held. Three separate studies say so independently, each having looked for a chamber and found none.

What replaces it is an argument. One body of work holds that the flowers give nothing at all — that Anthurium is nectarless and its visitors are collecting perfume, or being fooled. Another holds that the plant secretes sugar from two different organs, at two different times, and that the second one has simply been overlooked. Both positions are on this page, in the part where they belong, and neither is presented as settled.

The order, and why

Parts I to III are what the flower is and how it runs — the structure, the size of the genus, and the sequence a single spadix goes through. Read these once; they make the rest legible.

Parts IV to VII are the argument and the evidence — what the plant may or may not be offering, what it smells like, and every animal anyone has caught doing the job, graded by how good the evidence actually is. Part VII is the one to read if you only read one.

Parts VIII to XII are the operation — reading your own plant, collecting and keeping pollen, making the cross, what happens to the fruit, and what is known about which species will cross with which. That last one is shorter than it should be.

Then the sources, with what kind of evidence each one is.

Every flower is bisexual, and the whole spadix is fertile

In the other three genera the spadix is divided into zones: female at the bottom, sterile in the middle, male above, and often a sterile appendix on top doing the smelling. An Anthurium spadix has none of those divisions. It is covered, base to tip, in small bisexual flowers — each with its own tepals, its own stamens and its own pistil.

The consequence matters more than the anatomy. There is no sterile zone to feed an insect with, no appendix to run a furnace in, and no chamber to hold anything inside. Everything those genera do with architecture, this one has to do some other way.

No trap, and this is not a small point

Three independent studies — on weevils, on oil bees and on gall midges — each looked for a trapping mechanism in the species they were working on, and each reported that there is none. Nothing is enclosed, nothing is held, and no visitor is ever prevented from leaving.

Which means the plant has no way of making a visitor stay long enough to be useful. Whatever brings an animal to an Anthurium and keeps it there has to be worth its time — or has to look as though it is. That is the question Part IV is about.

The spathe is not doing what you think

The colored spathe that sells these plants is a backing sheet, not a room. It does not close, it does not constrict around the spadix, and in most species it reflexes away from it entirely. In the genus's own literature it is treated as a display and orientation surface — a landing board and a background — rather than as part of any mechanism.

The size, and why the number keeps moving

This site counts 1,460 accepted species, from Plants of the World Online. Recent literature gives a lower figure for formally described species — a 2026 paper gives about 1,300 — while estimates of the total can be higher. These are not corrections of one another. Each is a different authority on a different date, and the gap between them is mostly a disagreement about what counts as a species.

The important line in that table is the second one. Engler's 1905 treatment is the last time anybody revised the whole genus. Everything since has been sectional, regional or partial. A genus of well over a thousand species has been running for a hundred and twenty years on a framework built when fewer than five hundred were known.

The sections are mostly not natural

Those sections are how the genus is navigated — how a species gets placed, and how a breeder guesses what might cross with what. Molecular work has not been kind to them.

Even the 1983 sectional treatment said as much about one of its own groups, dropping it from the key with the note that it is “a seemingly unnatural group” for which no defining characters could be found. The problem is old and openly acknowledged.

And the plants in the trade are close relatives

Set against a wild genus of over a thousand species, the ornamental material is startlingly uniform.

Read with the sections above, the practical shape of the genus is this. Enormous wild diversity, an unfinished classification that cannot reliably tell you what is related to what, and a cultivated pool whose breadth is not yet settled. Any cross you make is being made without a map.

Do not average these. They are three different species on three continents' worth of forest, measured by three groups to three different standards. Taken together the only safe statement is that an Anthurium inflorescence is in business for somewhere between one and three weeks, and that the male phase is usually the longer half.

Where it has been watched closely, the phases do not overlap

In the study that watched most closely, “the two sexual phases never overlapped.” The stigmas finish before the first anther opens.

That is the reason most of this genus needs two plants. By the time such an inflorescence has pollen, its own stigmas are done — and unlike the other genera on this site, there is no chamber holding an insect through the changeover to carry pollen from one to the other.

Not every species is on that clock

That is the usual pattern, not a rule. A revision that treats 122 species of Mexico and Central America gives the genus “flowers perfect, usually protogynous” — usually. In one of its species, Anthurium pluricostatum, the stigmatic droplets persist “until all anthers are open,” which the revision calls “only weakly protogynous and might best be considered homogamous” and reads as “a clear possibility for self fertilization.” A second species, Anthurium cotobrusii, is described in the same work with its droplets still there as the first stamens open, and no conclusion drawn from it.

Weigh those two for what they are: single observations by one botanist, most of them made on cultivated plants, with no count of inflorescences behind them and no pollination trial in the same work to test whether the overlap actually yields seed.

Can a plant rescue itself? The two studies disagree

One route is closed for certain. A large plant cannot pollinate across its own inflorescences, because “each individual… produces several inflorescences during the flowering period; however, only one inflorescence is open at one time” — measured across seven species.

Whether a single inflorescence can pollinate itself is a different question, and here the literature splits.

In the first, bagging at bud stage across twenty species left nine producing seed, which the authors recorded as a capacity to self-pollinate. In the second, ten bagged inflorescences of each of seven species at each of two sites gave “none… produced infructescence, which indicates that seed production depends on pollinators”.

What to do with that: assume your plant cannot self, and work as though it needs a partner — collect pollen from one plant in male phase, keep it, and put it on a different plant in female phase. Parts VIII and IX are about doing that well.

But do not read fruit as proof the cross took, because on this evidence it might not be. Part XII.

One detail worth watching for

Usually the flowers open from the base of the spadix upwards, in what the same revision calls “a slow or rapid progression” — but its description of the genus also allows a spadix “sometimes beginning from the middle and developing rapidly throughout the spadix.” The direction has been tied to the shape: long, tapered spadices starting at the base and working toward the tip, cylindroid or clavate ones scattered, or starting at the apex or the middle. Anthurium caperatum is described with “stamens emerging scattered throughout the spadix” though its own spadix is a tapered one.

Whichever way it runs, a spadix is not in one state, it is in a gradient: on a long spadix the flowers at one end can be finished while the others have not started. When Part VII talks about a receptive stigma, it means a receptive band — and on a spadix that opens scattered, a patchwork rather than a band.

The strong claim: aroids pay nobody

A 2002 study of scent across the genus states it for the whole family, and states it flatly:

If that is right, then an Anthurium visitor is either collecting something that is not food, or being deceived. It is a family-wide assertion, and it is the most contradicted sentence in this whole literature.

Against it: somebody measured the sugar

Two species have had their secretions actually tested, and both came back sweet. In Anthurium seibertii the stigmatic fluid ran at 8% sugar — sucrose, glucose and fructose. For comparison, the phloem sap of a related aroid sits at about the same concentration.

The second is older and stranger, and it is the reason this part exists.

So there is a second reward window, opening as the first one closes, from an organ almost nobody inspects. If it is real and at all general, the genus is not nectarless — it is paying twice, and the second payment is timed to the moment the plant has pollen to move.

How common is any of this

The only broad survey is a look at 30 selected species during the day, weighted toward plants that flower frequently under greenhouse conditions and have relatively long stigmatic phases:

So every species looked at was secreting something, and about half were producing drops visible to the naked eye. But that is a survey of appearance, not of sugar. Beyond the two species above, the chemistry has been done on one more, in the structural study below.

Five answers, and which to trust

And then somebody looked properly

The question above — asked in print in 2017, and left at “further efforts are needed” — was answered in 2021, by a structural study that cut sections instead of arguing from chemistry. It found the thing everyone had been looking for in the wrong place.

And Daumann was right in 1930. He said the sweet secretion was on the tepals rather than the stigmas, and that it ran on into the male phase. Ninety-one years later, sections found the glands exactly where he said the fluid was.

Where that leaves the argument

The nectarless claim is dead, and the stigmatic-sugar claim is wounded. What survives is stranger and more useful than either: this genus has two different secretions, from two different organs, doing two different jobs — and the sweet one is the one almost nobody was watching.

What it means at the bench: when your plant is in female phase and the spadix looks wet, that fluid is the receptive signal — Part VII — whatever else it turns out to be. And keep looking at the tepals after the stigmas dry. If Daumann was right, something is still happening there — and the structural study that found the tepal nectaries worked on a single species, Anthurium andraeanum, so on most others the question is still open.

The scent work found that odor type tracks pollinator type closely enough to be judged by nose. Sweet and perfumed points one way; sour, fermenting or fruity points another; nothing at all points a third.

The finding that should change how you read any scent claim

In the gall-midge species the scent was analyzed, the dominant compound identified, and then — unusually — tested on the insects themselves. The result was not the obvious one.

Timing is part of the signal

Scent is not emitted continuously, and when it is released tells you when to expect the visitor. The perfume species scent in the morning, and that is exactly when the bees arrive — a tight window, roughly 08:30 to 12:00. The cucumber-scented species is strictly nocturnal, in both scent and visitation. Anthurium caperatum smells at night, in both sexual phases.

That last detail is worth pausing on. A scent that continues into the male phase is not recruiting for pollination of that inflorescence — its stigmas are finished. It is either moving pollen out, or advertising something the plant is still supplying. Which returns to Part IV.

What it takes to actually show pollination

Three things, in order of difficulty. That the animal carries the pollen. That it touches a receptive stigma. And that it moves between inflorescences, because an animal that does the first two without the third is just rearranging one plant's pollen.

Only two studies in this genus have all three.

The two strong cases

Male oil bees. Across two seasons they were effectively the sole visitor — carrying heavy pollen loads, contacting stigmas, moving between spadices, and marked individuals came back. This is the best pollination evidence anywhere in the genus.

What makes it strange is the payment. There is no oil, no nectar, no oil-secreting tissue and no trap, and the bees ignored the pollen entirely — a stingless bee collected it while the oil bees would not. The authors' proposal is that the bees come for the scent itself, mopping it up with an oil-filmed brush on the abdomen. Their own title ends in a question mark, and this page keeps it there.

Female gall midges. Named as “the only effective pollinators of the studied plant population” — pollen loads photographed, stigmas contacted, both phases visited, an attractant identified by field bioassay, and fruit set in three of five inflorescences. Here the reward is not mysterious: the midges were watched feeding on stigmatic exudate in the female phase and on pollen in the male phase. Two meals, two visits, one insect.

And notably, no brood site — the midges were never seen laying eggs, which is what gall midges usually do in flowers.

The birds, and why they may not be helping

At 2,700 m in eastern Ecuador, five bird species were watched picking up pollen from a male-phase inflorescence — four hummingbirds and a flowerpiercer — arriving with clean bills and leaving loaded. They were feeding on the fluid.

Two details worth keeping. The visits were structured by a dominance hierarchy — one hummingbird held the inflorescence as territory and won every contest — and the birds were feeding on fluid secretions, which returns to Part IV. Whatever the plant is producing, something with a high energy demand was drinking it.

Everything else is weaker, and some of it is very weak

Ants have turned up often enough that they are no longer noise. On Anthurium upalaense in Caribbean Costa Rica, Ectatomma ruidum patrolled every one of six female-phase spadices, feeding on a secretion (the text names the tepals, the figure caption the stigmas), and fed on pollen on three of four male-phase ones. The authors call it a possible pollinator and say plainly that they did not confirm pollen transfer. A genus-wide review of the interaction literature adds further ant visits from lowland forests in Costa Rica, Peru and Brazil — though those are counts compiled partly from herbarium labels and photographs, floors rather than totals, not a survey.

A butterfly on one species carried a pollen load, and the same butterfly species had visited the same plant in female phase two weeks earlier — suggestive, but the two observations are not linked. Moths on another species are identified only to family, and the authors call them visitors twice, in their own words.

Male euglossine bees were reported in numbers on Anthurium huixtlense by a grower in the tropics: many dozens of male Eulaema bees, possibly Eulaema cingulata (velvet-black, bumble-bee-sized, with shiny gold abdomens), massed on four or five receptive spadices across three mature garden plants. The same grower's Anthurium armeniense and Anthurium chiapasense, also in flower, did not seem to draw these bees but a Trigona species instead.³⁸ The report describes visitation and does not say whether any pollen moved.

Then the weakest link in the chain, which is also the most cited:

One species that will not behave

The sharpest illustration that this genus resists generalization: two populations of a single species, in two parts of Colombia, with two different smells and two different visitor guilds. One is foul-smelling and fly-visited; the other is scentless and bee-visited.

The authors considered splitting them and deliberately did not, judging the plants “so overwhelmingly similar in appearance” that it would be unwise “without further, more detailed information on their pollination biology”. So one name currently covers two pollination systems — and that is the honest position, not an oversight.

The cue: the spadix goes wet

Two independent accounts describe the same thing. Up close, “the stigma will exude a drop of fluid which will sit at the apex of each individual flower on the spadix”. From further back, the whole spadix takes on a “velvety and shiny structure… due to wetness”.

When the droplets are there, the flower is ready. That is the entire signal, and it is more legible than anything the other four genera on this site offer.

When to look

The droplets appear in the early morning. Some species present them on every flower at once; others work up the spadix in succession, which is the base-to-apex progression from Part III seen close up.

Practically: look first thing, and look again the next morning. A spadix that showed nothing at noon may have been wet at seven.

Two more things worth watching

The stamens come out in a usual order, and usual is as far as it goes. Across the genus, “usually the lateral pair of stamens emerg[es] first, followed by the anterior then the posterior stamen” — and the filaments are described as “promptly shrinking and withdrawing the anther to the surface of the tepals” once they have done their work. The generic description in that Mexican and Central American revision is looser again, and leaves the last two unranked: the lateral pair usually first, one at a time, followed by the anterior and posterior ones.

And it is departed from. In Anthurium caperatum the stamens emerge “scattered throughout the spadix, sometimes with the alternate stamens preceding the lateral ones” — the lateral-first sequence running backwards, on a long tapered spadix, which is the shape that is otherwise said to open from the base upward. That progression is keyed to the shape rather than fixed: long tapered spadices tend to begin at the base and work toward the apex, while cylindroid or clavate ones are less orderly, scattered, or beginning at the middle or the apex. Both notes carry the same caveat as the intervals above — the plants behind them were mostly in cultivation.

So a flower part-way through its male phase has some stamens out and some still in, and a spadix in mid-changeover is showing you its position in the sequence, not a malfunction. In one species the order was tracked up the spadix as well: the apex had its lateral stamens exserted while the bottom third had not yet finished emerging.

The withdrawal cuts the other way too. Because the filaments pull the spent anthers back down to the tepals, a spadix that has finished its male phase can look as though it never started one. No stamens showing is not evidence that pollen is still to come — check the stigmas: wet means the male phase is still ahead, browned and dry means it may already be behind you.

What you will almost never see

Fluid and pollen on the same spadix at the same time. The field study that watched most closely reported that the phases never overlapped; the grower literature is slightly softer — “rarely will the same plant have stigmatic fluid and pollen at the same time, although it does happen in some species on occasion”.

If you do see both at once, you have something worth recording — but overlap is on record, with at least one species described that way. In Anthurium pluricostatum the two phases run together: it is only weakly protogynous, near enough homogamous, and its stigmatic droplets persist while the stamens emerge. That is a trait of the species, not of the genus, and it comes from a description written largely on cultivated plants — so whether your own plant overlaps is a question about your species.

The gap, and what to do in it

Between the stigmas drying and the anthers opening on the same spadix is roughly a week. That gap is not dead time — it is when you line up the other half of the cross. Either a second plant is coming into female phase, or you are about to have pollen and need somewhere to put it. Part VIII is about keeping it until then.

When it is ready

“Pollen is ready to be collected and/or used when it is visible on the surface of the spadix. The timing of this event is always after the last of the stigmatic fluid is dried up and no longer visible.”

So the sequence on any one spadix is fixed and you cannot rush it: wet, then dry, then later pollen. If you can see pollen, that spadix's own stigmas are already past.

Collecting it

Straightforward, and no equipment worth the name: take it off with a finger or a brush, or tap it into a paper envelope, and keep it cool and dry.

Where to keep it — genuinely unsettled

The grower literature does not agree with itself, and the honest presentation is to say so.

The refrigerator is the disputed variable. One account says pollen kept cold “often becomes moldy”; the primer reports both camps and tells the reader outright to “experiment with various methods to find out which fits their species requirements”.

That is unusual and worth respecting. A practical literature that declines to give one answer is normally telling you the answer depends on something nobody has isolated. Try both, on the same pollen, and write down what happened.

The better workaround is the one the same source gives next: grow more than one clone of anything you intend to breed, so that a wet spadix and a shedding spadix exist at the same time somewhere in the collection.

The four things that actually matter

1. Wet spadix. Not a colored spathe, not a plant that looks ready — visible droplets, or that shiny wet look across the whole spadix. Part VII.

2. Cover the whole spadix, not a spot. The instruction is to smear up and down and around. The flowers are in a gradient, not a state — some bands are receptive and others are not, and you cannot tell which by looking.

3. Repeat, on consecutive mornings. Every practical source says it, and it is also the right hedge against the possibility that any individual flower is receptive only briefly. One application is a gamble; four mornings is a method.

A commercial breeding bulletin gives the rule a reason rooted in the spadix itself: “To achieve maximum seed production, an Anthurium spadix needs to be pollinated more than once.” Its contrast is what makes the point — one pollination will fertilize an entire Spathiphyllum spadix, because every flower on one of those matures at the same moment, while an Anthurium spadix comes into phase in a wave two weeks long. You are not repeating because the first attempt may have failed. You are repeating because most of the spadix was not open yet. No number of repeats is given, no interval, and no seed counts.

4. Tag it immediately. Fruit can take anywhere from one month to two years to ripen (Part X). Nothing you remember will survive that. Parents, date, direction of the cross.

Then wait, and expect to wait badly

There is no early confirmation. No color change, no swelling you can trust in the first weeks. The spadix either begins to develop berries or it does not, and in the slowest species you will not know for the better part of a year.

And there is nothing to do to it in the meantime. The same commercial bulletin is explicit: “To ensure seed set for Anthurium or Spathiphyllum, no special environmental manipulation — such as controlling relative humidity — is needed after pollen transfer.” That is worth saying only because the same three authors demand 100% relative humidity and a wet-toweling wrap for Dieffenbachia the following year — so a reader arriving from aroid breeding advice generally has a real reason to think humidity matters here. For this genus their answer is that it does not, stated flat, with nothing measured on either side of it.

Which is the practical argument for making several crosses in a season rather than one careful one, and for keeping the records that let you learn something from the failures. Part XII is about what the failures mean.

How long the fruit takes — and this varies more than anything else here

More than an order of magnitude, within one genus — the slowest of these takes twenty-four times as long as the fastest. There is a tendency inside it, small scandent species fast and large species slow, and a tendency is not a schedule. Do not average these into a rule — find out what your species does, and until you know, assume the long end. It is the reason Part IX insists on a tag.

Knowing when it is ripe

The cue is mechanical rather than visual, and it is unusually definite: berries are ready when they “literally ‘pop out’ of the spadix”. Some species release the lot at once; others let go over several days as they ripen in succession.

Ripe-looking and viable are not the same moment, and for a collector the difference runs the useful way. The seeds inside are “usually… capable of germination somewhat before the berry becomes fleshy, loose, and obviously mature”, and berries “are also capable of maturing to some extent even on severed inflorescences” — so an infructescence cut early is not necessarily a loss. Both statements come from one monographer's experience of growing the section, with no germination trial, no interval in days and no count behind them. Treat them as a reason to try a berry that looks unready, not as a schedule for picking one.

Most berries hold two seeds, one per locule. But some species regularly carry four, and rarely a species runs to as many as ten, at two to five seeds per locule. The account that records that range names no species for the four- and ten-seeded cases. There is a lookup one level up, though. The revision of the Mexican and Central American species puts more than one seed per locule in sections Tetraspermium, Gymnopodium and Porphyrochitonium, and usually one per locule in the remaining sections. Find your plant's section and you have a prediction — open a berry and you have the answer.

Sowing

Seed sent by mail does not always arrive dried out. Thomas B. Croat of the Missouri Botanical Garden reported that seed of most aroid species may survive a month or more in the mail when it is prepared with care. For Anthurium, the seed is washed free of the sugary pericarp and mesocarp, which invite fungal attack, then folded into a square of dry newspaper (his example was about 20 fresh seed to a square 15 cm, or about 6 in., wide) and sealed in a small plastic bag with padding against crushing, the packet kept moist but not wet. He had used the method for 25 years with great success.³⁹ In a later post he described reasonably successful mailings as far as Australia, with shipments of several weeks not seeming to affect the seed.⁴⁰ Other growers were firmer that Anthurium seed is short-lived: Dewey E. Fisk wrote that seed of most Araceae, Anthurium especially, usually stays viable no more than three or four days,⁴¹ and Don Burns wrote that Anthurium seed allowed to dry germinates poorly.⁴²

Germination takes one to three weeks in most cases, longer in some species. Use a well-drained soilless mix and watch for fungal and bacterial problems, which are the usual cause of loss at this stage. The commercial equivalent is sowing into cells under 75–80% shade, with plantlets moved on at 4–6 months.

What the seed is actually like

Two details from the genus description that no other source here mentions. The berry is two-celled with usually one seed per cell, and the flesh around the seed is “usually translucent, sweet” — which is presumably the point, for whatever eats it.

A first-hand taste report qualifies that. Asked on the Aroid-L mailing list whether Anthurium berries are poisonous, Thomas B. Croat answered that the seeds are not, if the plant is an Anthurium, because he ate the mesocarp all the time. He found it sweet at first but soon acrid, enough to make him spit it out, and suspected that it might do the same to birds.⁴³ It is one informal remark, not a toxicity test. For Anthurium scandens, Julius Boos, replying to a grower who had tasted one greenhouse berry and found it very sweet, called the fruits edible but a poor reward: little beyond flavor, with much skin and many seeds.⁴⁴

And the seeds are “usually with a sticky appendage on at least one end” — usually, and the exception has a geography on it. The revision of the Mexican and Central American species records that “many of the Mexican species of Anthurium lack a sticky appendage on the seeds”. No count and no list goes with that, so it is a warning rather than a lookup: if your plant is Mexican and the seed is not sticky, that is on record, and not something you did to the berry.

That is read as a dispersal structure — and read is the word. The revision that supplies the sectional lookup above says the sticky appendages “seem ideally suited” to bird beaks, and behind that careful phrasing sits no bird watched, no gut passage traced and no seed recovered. The stickiness is described; the bird, in that account, is inferred from it. A separate field note on Anthurium upalaense in Caribbean Costa Rica photographed a manakin (Manacus candei) and a euphonia (Euphonia gouldi) taking berries from four infructescences; no gut passage was traced and no seed recovered there either, so dispersal remains inferred. It explains something practical either way: if the seed sticks to your fingers, that is the plant working as designed, not pulp you failed to clean off.

Then the long part

From sowing to first flower, in commercial Anthurium andraeanum: fifteen months at the earliest, thirty to thirty-six months typically — call it three years from seed to bloom.

Add the fruiting time and a cross made today may not be assessable until 2029. That is the real reason to keep records, make several crosses at once, and label everything twice.

Start here: some of these plants make seed without you

This is also why Part VIII warns against using a plant's own earlier pollen. A spadix that pollinated itself and a spadix that took your pollen look identical.

What the compatibility record holds

Anthurium clarinervium will not cross with its own section. Attempts with other members of section Cardiolonchium failed so consistently that the species' sectional placement itself is now doubted — it is thought instead to be closer to Anthurium berriozabalense, and its section is described as unknown.

That is worth reading twice. A breeding failure was strong enough to overturn a taxonomic placement — which tells you both how weak the sections are (Part II) and how much information a failed cross carries.

Anthurium radicans hybrids are a dead end. The species has produced several attractive foliage hybrids, and “these hybrids all seem to be sterile and will not produce a successive generation”. You can make the first cross; you cannot build a line on it.

And one result that goes both ways at once

One much-quoted compatibility result comes from breeding work reported second-hand in a taxonomic paper. Anthurium amnicola was crossed against two different sections, with opposite outcomes:

Note what it does to the sections argument. Here, section membership predicted the outcome correctly — the crosses worked within Calomystrium and failed outside it. In the Anthurium clarinervium case it predicted exactly wrong.

So the sections are neither useless nor reliable. They are a hypothesis that is sometimes right, and the only way to find out which kind of case you have is to make the cross.

A longer list, and a first-hand one

The other set is longer, and unlike the one above it is first hand. A grower-naturalist growing the Anthurium dressleri complex in Guatemala City published what he had crossed it with. Eight took: Anthurium crystallinum, Anthurium forgetii, Anthurium magnificum, Anthurium papillilaminum, Anthurium regale, Anthurium luxurians, Anthurium warocqueanum, and Anthurium radicans, which sits in a different section altogether. Two did not: Anthurium balaoanum, tried repeatedly, and Anthurium lappoanum.

It is the longest list of partners on this page, and it is one grower's plants — no counts, no protocol, no controls. Read it as a place to start, not as a compatibility table.

A shorter first-hand record is a single pair. Neil Carroll pollinated Anthurium veitchii with pollen of Anthurium tilaranense. By February 28th, 1999 berries were forming, though he could not yet say whether the cross had taken;⁴⁵ by May 8th, 2000 he reported very good seed set and that the seed had germinated.⁴⁶ It has the shape of the records above, a named pair and an outcome without counts, and it stops at germinated seed, short of seedlings that could confirm the cross.

A program that did put numbers to this

The records above share a shape: a named pair, an outcome, and no counts. None of them states how many attempts were made, and most are reported at second hand. Two published programs are built the other way, and the older of the two is the larger. At the University of Hawaii, 1,592 hand cross- and self-pollinations were made among 56 species, and the results were published in 1976. The species were sorted into six morphological groups on Englerian characters — ovules per locule, berry color and shape, inflorescence shape, leaf shape and texture — and then crossed within and between those groups.

Those two rows are the comparison the authors drew, and they called the difference significant at the 1% level. The whole program was larger than the rows above: 1,592 pollinations in total, counting 21 selfs and 52 crosses within a species, for 280 fruiting spadices and 181 flowering progenies. One caution on the smaller number: the printed total of nine between-group progenies does not match its own column, which lists 1 and 7 and otherwise zeroes — a sum of eight. Nine is what the paper prints, and only two between-group progenies are ever named.

That is a map, and a coarse one. Crossing inside a morphological group produced flowering seedlings often enough to be worth the labor; crossing between groups almost never did — 9 flowering progenies out of 886 attempts. Those six groups are not the sections. They were built for this purpose, out of characters a breeder can see on the plant.

The six groups, and how each one behaved

The map is worth printing, because the groups are not equally willing and the differences are large.

Which group is your plant in

The groups are not the sections, and they were not meant to be. They are a working key built from characters a grower can see, and the first cut is the berry: three or more seeds in a berry puts a plant in Group I, whatever else it looks like. Everything with two seeds or fewer is then split by leaf shape, and finally by leaf texture — velvety into Group V, leathery into Group VI. Open a group to see who is in it.

The section printed beside each name is the one that program used, out of Engler's 1905 arrangement, and at least one has moved since: the 1991 revision of section Pachyneurium treats Anthurium hacumense as a member of section Porphyrochitonium instead. Read the sections here to follow the program, not to place your own plant.

Group V is the most willing — nearly a third of its within-group pollinations ended in a flowering hybrid. Group VI took the most work for the least return: 267 pollinations, the largest effort in the program, for the same 38 hybrids Group V got from half as many attempts. The authors put that down to Group VI being the most morphologically diverse of the six, and Group I's low yield down to its chromosome range — 2n = 24, 28, 30, 48 and 84 inside one group. They also suggest Group V would have scored higher still but for B chromosomes affecting viability. Both of those are readings, not tests — no comparison of B-carrying against B-free plants is presented.

Only two pairings ever crossed the barrier

Fifteen group-pairs were attempted between groups. Thirteen produced nothing that flowered. The two that did:

That last one is the most useful single result in the program, because it was tried both ways. One direction gave a vigorous but sterile hybrid; the reciprocal gave weak seedlings that died early. Which parent carries the seed is not a detail. The authors read the cross as suggesting Anthurium triangulum may be misplaced in Group VI, since it made no flowering hybrid with any other species there — their suggestion, offered as a possibility.

Where germination is concerned the barrier is starker than the hybrid counts make it look. Between Groups I and II, 116 pollinations gave six fruiting spadices and a mean germination of 0.3%. Four other pairings germinated nothing at all.

And a second program, counted by section

The other one was published in 1991, inside a revision of section Pachyneurium, and it asks a different question: not whether morphological groups predict a cross, but whether a section does. The crosses were made by R. D. Sheffer at Indiana University Northwest and by Thomas Croat at the Missouri Botanical Garden. No growing conditions are stated for any of it.

Between sections, almost nothing came through. The five sections crossed against were Belolonchium, Cardiolonchium, Dactylophyllium, Porphyrochitonium and Urospadix, and nearly all of the seed that formed was abortive and failed to germinate. Two crosses in the whole between-section record broke through: Anthurium standleyi × Anthurium lilacinum, and Anthurium hacumense × Anthurium glaucospadix.

Inside the section there is a second barrier, and it is geographic. Most of the pollinations that gave viable seedlings were Central American with Central American, or South American with South American, and not between the two. Three pairings are named as the exceptions: Anthurium barclayanum with Anthurium cotobrusii, Anthurium crassinervium with Anthurium bradeanum, and Anthurium halmoorei with Anthurium fendleri, which gave two viable crosses. So if a within-section cross will not take, look at where the two plants come from before you conclude anything about the section.

That paper prints 21 successful crosses by name, out of 226 attempts summarized across 58 taxa, of which 27 produced viable plants. Counted a second way in the same paper, those 27 are 36% of the 74 species that were tested at all.

What the sections can and cannot tell you

The obvious strategy — cross within a section, because those species should be relatives — is weaker than it sounds. As Part II sets out, most sections are not monophyletic, and the diagnostic characters of at least one large section turned out to have evolved more than once independently.

So a shared section means the two plants look alike in ways somebody thought were important in 1905. It is a reasonable place to start and a poor thing to rely on. The Anthurium clarinervium case is exactly this failure caught in the act.

And a breeding record that is Victorian

A French glasshouse manual from 1898 lists roughly twenty-five named hybrids with the cross direction, the breeder and the year, running from 1881 to 1896 — including the first hybrid ever raised from Anthurium andraeanum, made by Bergman in 1881.

Two other things from the same source, both practitioners' rules rather than experiments. Outcrossing beat selfing: pistillate flowers “set seed better with pollen from another plant of the same species”, and to get any seed at all you needed several identical plants in the house. And bigeneric crosses were doubted even as interspecific ones were called easy.

The first of those two is contested. In the 1976 program, 21 self-pollinations gave 17 fruiting spadices (81.0%) and 14 flowering progenies (66.7%), against 52 crosses within a species giving 34 fruiting spadices (65.4%) and 30 flowering progenies (57.7%). The authors read that as “no significant differences in the production of offspring between self- and cross-pollination within a species”.

Do not read the selfs' higher percentages as a win for selfing either. With 21 selfs against 52 crosses the authors declined to call the gap a difference — and no species is named against a self. The 21 are pooled across the six groups, so the figures describe one collection of plants and not any species in it.

Chromosomes

Anthurium has been counted more thoroughly than most aroid genera, and what those counts show — the common number, the polyploid series, and the extra chromosomes carried in one section — sits on the page that deals with the whole family: Why Some Crosses Fail. It is not repeated here, deliberately, so the two pages cannot drift apart.

The short version: the counts exist, and they have been tied to particular crosses only a few times — unique chromosome counts offered as reasons Anthurium scandens and Anthurium trinerve were not expected to hybridize with other Anthurium species, and fifteen regular chromosome pairs recorded in the hybrids of Anthurium amnicola, though their pollen stained at only 3.3–27.8% with aceto-carmine (fourteen hybrid plants, one spadix each), against 87.5% in Anthurium amnicola itself (two spadices). They tell you the genus has room for ploidy barriers. They locate very few of them.

1. The spadix was not receptive

Far and away the commonest. A colored spathe is not a signal — the spathe can be open and perfect for weeks while the spadix is doing nothing. Wet, glistening, droplets at the flower tips. Nothing else counts (Part VII).

And check in the early morning. A spadix inspected at midday may have been receptive at dawn and dry by the time you looked.

2. You only did it once

Every practical source says the same thing: several mornings in a row. Flowers come into phase in sequence up the spadix, and there is a claim — unsourced, but consistent with the rest — that any individual flower is receptive for about a day. A single application catches whichever band happened to be ready.

3. The pollen was dead

The primer's figure is a week or two, with no temperature attached, and the storage advice openly contradicts itself (Part VIII). A 1981 note on Anthurium araliifolium adds one data point: pollen kept five days at 10 °C and brushed onto a later spadix of the same plant set six fruits, and nine of their eleven seeds germinated. Fresh pollen removes the variable entirely — which is the real argument for growing several clones.

Or there was none to collect. Denis of Silver Krome Gardens in Florida reported in 2011 that Anthurium andraeanum varieties produce no pollen until nights drop consistently into the low 60s °F (about 16 to 17 °C) while the flowers are in early development.⁴⁷ In 2013 he limited this to the large-flowered, high-Andean type of Anthurium andraeanum and gave the figure as nights of 55 to 60 °F (about 13 to 16 °C), when pollen starts in the tiny flower buds; in Florida, he wrote, that confines breeding of these plants to mid to late winter. He noted that other members of section Calomystrium may be male-fertile even in the warm summer heat.⁴⁸ The two figures differ, which marks this as a grower's rule of thumb and not a measured threshold.

4. It worked, and you are early

One month to two years from pollination to ripe berry, depending on species (Part X). Before you conclude anything, check what your species does. More than one grower has thrown out a working cross on a schedule borrowed from a faster plant.

5. The cross genuinely will not go

Last, because it is the least likely and the hardest to establish. Clearly documented cases are scarce (Part XI) — and one of them was strong enough to move a species out of its section.

Before concluding incompatibility, rule out one to four. And if you do conclude it, write it down with the numbers: parents, dates, how many attempts, what the spadix looked like each time. That record would be a genuine addition to what is known, which is a sentence this site can write about very few things.

Peer-reviewed literature

Grower and breeder accounts

Chromosome counts

What is missing

1. Whether the secretions are nectar. For one species it has been answered: in Anthurium andraeanum a 2021 study found sugars in the tepal secretion and none in the stigmatic fluid. Since 1930 the only other published sugar figure is a second-hand report for the stigmatic fluid of Anthurium seibertii, and for the rest of the genus the question is open.

2. What the tepal secretion is. It was described in 1930, it is sweeter than the stigmatic fluid, it continues into the male phase, and since then it has been sectioned and sugar-tested in one species, Anthurium andraeanum, where it comes from nectaries on the tepals; elsewhere it has been recorded as droplets in species descriptions, but not analyzed.

3. A germination percentage you can attach to a species. The 1976 crossing program sowed 25 seeds per cross on filter paper in a petri dish and published an average germination percentage for each of its group pairings, across the 56 species it worked with — but the unit of report is the morphological group, so no figure in it belongs to a named plant. No viability curve goes with them.

4. A pollen viability test a grower can run. Staining tests have been published for a few species, in chromosome studies, and they need a microscope. The “week or two” figure has no temperature attached and no sample size.

5. Whether any Anthurium is apomictic at all. Four species are recorded as self-pollinating and one is described setting a full spadix unpollinated. Apomixis has been proposed — a 1983 sectional treatment calls Anthurium scandens and Anthurium trinerve “apparently self-pollinating or apomictic”, and the grower primer calls Anthurium bakeri apomictic — but not tested: no published study has bagged an emasculated spadix and counted what came. Until somebody does, it is unknown whether fruit set without a visible pollinator means selfing or something else.

6. A compatibility matrix at species level. The 1976 program published one for its six morphological groups — pollinations, fruiting spadices, average germination and flowering progenies in every pairing, over 1,592 pollinations among 56 species — and the 1991 program published counts for one section against five others, with 21 successful crosses named. Below the group and the section, what is on record is a scatter of separate crossing lists rather than a table you can look a pair up in.