IDENTIFYING AMORPHOPHALLUS

Most genera get harder to identify because the plants look alike. Amorphophallus is harder for a different reason: its modern classification is largely invisible. The genus was sorted into four subgenera by a molecular analysis, and for two of those four the paper that named them does not list a single character you could check on a plant.

That is not a complaint about the paper. It is how botanical naming works: a new name has to be published with a diagnosis, and an old name or an automatic one does not. Two of these four subgenera were new in 2017 and carry full descriptions. The other two already existed, so they arrived with a distribution and nothing else¹.

The practical result is that half the top-level structure of this genus cannot be keyed. You can place a plant in it, but only by knowing where the plant sits on a tree built from DNA.

And the character that works best is not a shape

There is one striking exception, and it is what this guide is really about. The single character that lines up most convincingly with one of these molecular groups is not something you look at. It is a question about what the plant did over a whole season — and it was already sitting in a key published twenty-one years before the DNA work that vindicated it.

Part I sets out the four groups. Part II is about the two that have no description. Part III is that character. Part IV is the short list of things you genuinely can look at, and Part V is what remains unsettled.

Before 2017 this genus had exactly one subgenus to its name. Everything else was a set of sections — ten of the eleven Engler had accepted in 1911, unrevised since — and molecular studies, sampling about three in ten of its species, had already found one of them, section Rhaphiophallus, not to be a natural group¹. Then a team sequenced 157 species — about seven in ten of the genus as it was then counted — and recovered the four major groups an earlier 69-species study had found¹.

What was actually done

One nuclear marker and two plastid ones: ITS1, rbcL and part of matK, analyzed by both Bayesian inference and maximum likelihood. The genus itself came out monophyletic at the highest support either method offers — 1.0 posterior probability and 100% bootstrap. Within it, four clades recurred in both analyses, and the authors formally raised all four to subgenus. Two of the names already existed. Two were new here, and one of those, Scutandrium, absorbed the whole of what used to be the separate genus Pseudodracontium¹.

And the other three do not work that way

They all occur in mainland Southeast Asia and they overlap there heavily. A plant from Thailand or Laos or Vietnam could belong to any of the three, and its locality tells you nothing about which¹. For those, you need characters — which poses the problem.

Open the taxonomic treatment in that paper and read the four entries in order. Two of them contain a heading that reads Diagnosis, followed by tubers, spathes, staminodes, styles and berries. Two of them contain a sentence explaining the name and then a distribution¹.

Which two, and why

Scutandrium and Afrophallus were new names in 2017. A new name has to be published with a diagnosis, so both carry one. Metandrium had existed since 1924, and subgenus Amorphophallus is an autonym — it exists automatically because the genus type falls inside that clade. Neither needed a description, and neither was given one¹.

What that means when you have a plant

If your plant is African, Part I already settled it. If it is Asian, you are choosing between three subgenera, and only one of the three — Scutandrium — has a character list you can hold it up against. For the other two, membership is a position on a tree. It was established by sequencing, and in the general case it is confirmed the same way.

That is worth stating plainly rather than hiding, because the alternative is a reader concluding they have failed at something that cannot currently be done. You cannot key an Amorphophallus to subgenus Metandrium from a description, because no description has been published.

What you can do instead is Part IV. But first, the one character that turned out to carry more weight than anybody expected.

In 1996, two decades before the DNA study that sorted it into four subgenera, Hetterscheid and Ittenbach published a key to it. Couplet 41 of that key does not ask what any part of the plant looks like. It asks what season things happened in.

The same paper puts it in prose as well. African species “all develop a leaf in the same season as flowering, with or without pollination and fruit set”; the Asian non-evergreen species, in the wild, refuse to make a leaf after flowering and effective pollination, so “fruiting plants are never found with leaves”².

Then the DNA arrived, and agreed

Twenty-one years later the molecular analysis erected subgenus Afrophallus for the African clade — and when it came to say what distinguishes that subgenus, the character it reached for was the growth cycle. Each year, both flowering and leafing occur in mature tubers; that cycle, the paper says, supports the molecular monophyly of the subgenus, and it is its most prominent apomorphy¹.

It is not the only one

Scutandrium's diagnosis runs on time too. Its inflorescence is described as appearing every second year, alternating with a solitary leaf in the next — with a named exception, Amorphophallus rhizomatosus, which carries leaf and inflorescence together¹. That is a two-year observation written into a subgenus description.

The mechanics of that cycle — why the tuber can fund only one structure at a time, and what the two calendars cost the plant — are Part II of Amorphophallus Morphology. Here it matters only as evidence.

And it assumes a plant that rests

An evergreen species complicates the calendar. On December 29th, 2006, Wilbert Hetterscheid described Amorphophallus coaetaneus as an evergreen whose leaves may stay on for three to five years¹⁰. In Mobile, Alabama (USDA zone 8b), Russell Coker’s two plants of that species had their foliage cut back by frost in the fall of 2009; the next spring the first came up with a flower, and leaves on both followed almost two months later¹¹. That is a flower ahead of the leaf, the reverse of the order couplet 41b gives for an Asian species that does both in one season. It comes from two plants after one hard winter, and the report does not say whether the order was the species’ own or a response to that winter.

Two of the four subgenera do have published character lists. Here they are, side by side, as close to the original wording as prose allows. This is the whole of what the literature offers for keying an Amorphophallus above the level of species¹.

So the short version

Of six characters, one does real work: the staminodes. Shield-like and broad, or not there at all. Two more help — a strongly constricted spathe and a flattened, disc-shaped tuber both point African. One is a calendar. One is a pair of named exceptions. One is noise.

The four groups are solid. How they relate to each other was not, and the 2017 paper is candid about it.

The backbone

The four clades held up under both methods. The branching order between them did not. In the authors' words, the relationships between the clades “remain poorly supported and would require further molecular investigations”, and the arrangement was “not at all resolved in previous studies” and still unclear¹.

That gap was later addressed with far more data — a phylogenomic study using plastid genome sequences, aimed squarely at resolving the subgeneric backbone³. It is worth being precise about what that means: it filled a hole the 2017 paper had explicitly left open. It did not overturn a result, because on this question there was no supported result to overturn.

What to do with all this

Three things, in order of how much they will actually help.

Record the season. Note the date the inflorescence opened, and note whether a leaf came up that same year, before, after, or not at all. That single habit answers couplet 41, which is the character that carries the most weight in this genus and the one that no photograph can recover later. Note too whether old leaves were still on the plant when it flowered. On February 22nd, 2010, Wilbert Hetterscheid wrote that Amorphophallus polyanthus has a strong tendency to flower immediately after leaf senescence sets in, sometimes while old leaves are still more or less present¹².

Photograph the inflorescence, not the crown. Every character in both published diagnoses is on the inflorescence or the tuber. When one opens, get the inside of the spathe base, the spadix, and — if the plant has them — the staminodes. Those few days are the only time the evidence exists. The leaf is a poor substitute. Michael Pascall of Whyanbeel Arboretum, in Far North Queensland, wrote on August 16th, 2001 that seed of a spotted-petiole form of Amorphophallus galbra, collected at Mount Amos near Cooktown, gave him only one seedling with a spotted petiole and all the rest green, and that the only true way to identify these variable plants is by the flower¹³.

Note the provenance. For an African or Madagascan plant the subgenus follows from the locality alone. For an Asian one the locality will not settle it, but it still narrows the species. Wilbert Hetterscheid reasoned this way on August 7th, 1997. In the vegetative state, he wrote, Amorphophallus bulbifer is difficult to separate from Amorphophallus muelleri. He had no reports of Amorphophallus bulbifer from Burma, where Amorphophallus muelleri is known, so for a plant from there Amorphophallus muelleri was a likely candidate¹⁴. The tuber might help: smooth in Amorphophallus bulbifer, and covered all over with distinct, raised, annulate root scars in Amorphophallus muelleri^14, 15.

Fifteen sources. Six are papers. One molecular analysis supplies the classification and both of the published diagnoses; one key from twenty-one years earlier supplies the character that turned out to matter most; one phylogenomic study addresses the question the first one left open; and two more carry the figures — a 1911 monograph plate and a 2023 set of regional keys. The sixth, a 2003 revision of the African species, supplies the character table discussed in Part IV. The other nine, numbered 7 to 15, are messages to the Aroid-L mailing list: statements made in correspondence, not published determinations.