The Karwinskii Sub-Variety Complex

Overview

This chapter is a working snapshot of the single most active unresolved area in agave taxonomy: the Agave karwinskii sub-variety complex. The species is one of the few trunked agaves in commercial mezcal use, and it carries the longest list of named sub-varieties of any Oaxacan mezcal plant. Maestros mezcaleros recognize ten or more distinct producer-named forms (madrecuixe, cuixe, bicuixe, barril, tobaziche, cirial, san martinero, largo, tripón, sierrudo, and a handful of regional one-offs), each one consistently associated with a particular trunk-and-piña morphology, a particular maturation window, and a particular distillate character. The botanical literature has carried A. karwinskii as a single broadly variable species since Howard Scott Gentry's 1982 monograph and has been slowly catching up to producer-recognized infraspecific structure ever since.

Two recent developments make 2024-2026 the inflection point. The first is a 275-individual, 19-descriptor multivariate phenotypic study from Aragón-Parada and colleagues that, almost incidentally, documented A. karwinskii as the most internally heterogeneous of every Oaxacan mezcal species tested. The second is a complete genome assembly for the species published by Peña-Ramírez and colleagues in 2024, which gives any future population-genomic team the foundational reference sequence it needs to ask the structured question directly. The discriminating molecular study has not yet been published in print. Watch this space: a peer-reviewed paper resolving (or formally rejecting) the producer-named sub-varieties as discrete genetic clades is almost certainly coming within the next 24 months.

The editorial posture this chapter recommends, until that paper lands: maestros mezcaleros are very likely correct that the named sub-varieties track real biological differences. The botanical literature has not yet formally codified that recognition because the discriminating study has not been published. Treat sub-variety names as locally meaningful and operationally useful for production, not as formal Linnaean varietas or subspecies designations.

The species

Agave karwinskii Zucc. was described by Joseph Gerhard Zuccarini in 1833 from material collected by the German botanist Wilhelm Friedrich Karwinski von Karwin (the species name commemorates Karwinski). The original publication is in volume 16(2) of Nova Acta Physico-Medica Academiae Caesareae Leopoldino-Carolinae Naturae Curiosorum, page 677. Three nineteenth-century redescriptions (Agave laxa Karw. ex Salm-Dyck 1834, Agave corderoyi Baker 1877, Agave bakeri Ross 1896) were eventually folded back into the Zuccarini name as heterotypic synonyms in the modern Kew Plants of the World Online (POWO) treatment. The species sits in family Asparagaceae, subfamily Agavoideae, with its native range recorded as the Mexican states of Oaxaca and Puebla.

Geographically, the species occurs across the Mixteca Alta and Valles Centrales of Oaxaca and through the Tehuacán-Cuicatlán corridor in Puebla, at elevations of roughly 1,500 to 1,900 metres, on dry hillsides and rocky scrub. Higher-resolution distribution mapping is sparse outside producer knowledge; the actual range almost certainly extends a bit beyond what POWO records. Maturation runs 10 to 18 years, with the longer end common in undisturbed wild populations.

The trunked-agave anomaly

Most agaves used at scale for distillation present as low rosettes pushed up from a near-ground stem: the leaves arch out from what looks, at a distance, like ground level. Agave karwinskii breaks that pattern. Mature karwinskii plants develop a woody trunk that can exceed four metres, topped by a relatively narrow rosette of lance-shaped leaves. The whole plant reads more like a tree-form yucca than like an espadín or a tobalá. The piña reflects the architecture: instead of the spherical pineapple shape that gives the heart of the plant its Spanish nickname, a karwinskii piña is sausage- or baseball-bat-shaped, elongated, and considerably less massive than its rosette cousins. Piñas typically weigh between 8 and 25 kilograms, sometimes reaching 35 kilograms in older trunked specimens.

The trunked habit matters for the sub-variety question because trunk morphology turns out to be one of the discriminating phenotypes maestros track. Cirial (named for its straight candlestick trunk), largo (named for its long trunk), and tobaziche (lean and narrow-trunked) are all named on trunk characters. Barril is named for piña shape, a feature that depends on the trunk-to-piña ratio. Madrecuixe is the form with the longest trunk and the largest piña. The morphology the producer reads off the plant in the field is the same morphology a formal taxonomist would have to use to separate populations.

The species also reproduces unusually for the genus. Agave karwinskii produces basal offsets via creeping rhizomes in addition to flowering on a 2 to 6 metre quiote (the central flowering stalk). This dual reproductive strategy is uncommon among mezcal agaves and matters for sub-variety dynamics: rhizome-based clonal patches can hold a single phenotype in place over decades, while bat-mediated cross-pollination at flowering moves genetics across distance. The species is therefore simultaneously clonal and outcrossing, which is exactly the recipe for genuine infraspecific structure to develop and persist.

The sub-variety question

The question that has hung over Agave karwinskii since at least the 1980s is straightforward to state: what are the producer-named sub-varieties, botanically? Three live hypotheses sit in the literature.

Hypothesis A: one highly variable species. The standard view inherited from Gentry's 1982 monograph and carried through most twentieth-century herbarium work. A. karwinskii is a single species with very broad morphological plasticity; the named sub-varieties represent ecotypes, age cohorts, or microhabitat responses, not formally distinct taxa.

Hypothesis B: a polymorphic species with real infraspecific structure. The view increasingly supported by recent phenotypic work. There is a species, but populations within it are deeply structured, and the producer-recognized names track real population-genetic clades that may eventually warrant varietal or subspecies rank.

Hypothesis C: a species complex requiring splitting. The most ambitious view, held by some Oaxacan producer-researchers and a minority of academic botanists. At least some named varieties are not A. karwinskii at all but are misattributed; sierrudo, in particular, is often placed within the A. americana group instead.

The empirical evidence so far points strongest to Hypothesis B, with Hypothesis C live for individual cases like sierrudo. The cleanest way to read the current state of the field is: the producer naming system has been operationally tracking real biological differences for generations, and the formal taxonomy is now beginning to acquire the tools to confirm or refute that tracking.

The 2024 phenotypic and genomic evidence

Two 2024 publications shift the empirical ground.

The first is Aragón-Parada and colleagues, Agro Productividad (2024), "Phenotypic analysis of mezcal agaves from the Central Valleys of Oaxaca." The team assessed 275 individuals across seven Agave species (espadín, tobalá, karwinskii, tepeztate, americana, and others) using 19 morphological descriptors and standard multivariate techniques: principal-component analysis, hierarchical clustering, multivariate analysis of variance. The headline finding for this chapter: A. karwinskii and A. potatorum showed the greatest intra-population phenotypic variability of any species in the study, with statistically significant within-species differences. In plain English, the populations the field crew sampled as "karwinskii" did not behave like a tight morphological cluster. They formed structured sub-groups before the analysis was told anything about producer-recognized sub-varieties. Medium confidenceMedium confidence: most claims are backed by reputable secondary sources, but some details rely on inference or have not yet been verified against primary sources.The structure emerged from descriptor-based clustering, and the paper was not designed to test the sub-variety hypothesis directly; medium confidence that the sub-groups in the multivariate space map one-to-one onto specific producer-named varieties, high confidence that infraspecific structure is real.

The second is Peña-Ramírez and colleagues, Biodiversity Genomes (2024), a complete genome assembly for A. karwinskii produced from a single wild-collected individual using paired-end Illumina sequencing. The assembly was published as part of the Yucatán-Peninsula plant genomes project, with the sequence deposited at GenBank under accession JAXCLU000000000. The genome itself does not answer the sub-variety question; it gives the next research team the reference sequence they need to design a population-genomic study sampling multiple named sub-varieties across multiple sites. That study has not yet been published. The cost of doing it has dropped sharply now that the assembly exists. The bottleneck is now sampling logistics and grant funding, not sequencing.

Population-genetic precedent in adjacent species sets expectations. Figueredo-Urbina, Casas, and Torres-García (PLOS ONE 2017) used 10 nuclear microsatellite loci to show that morphologically distinct populations of Agave inaequidens, A. cupreata, and the domesticated A. hookeri form genetically distinguishable clusters; that paper is the clean methodological template for what a future karwinskii study will look like. Work on bat-mediated gene flow (Trejo-Salazar and colleagues, 2016 and later contributions) has shown that nectar-feeding bat pollinators move pollen across substantial geographic distances within and between agave populations, which means karwinskii sub-varieties under wild conditions should be cross-pollinating extensively, exactly the scenario producers describe when they report that madrecuixe seed yields offspring spanning several "named" types.

The ten named sub-varieties (with what producers say)

What follows is a synthesis of producer knowledge from the Oaxacan artisanal mezcal community (Mezcal Vago, Real Minero, El Jolgorio, Rey Campero, Lalocura, Mezcaloteca, Neta Spirits, Pensador, Mal de Amor) with the limited academic literature. Each entry walks the morphology, the maturation window, and the sensory profile across the full tasting arc (aroma, first sip, midpalate, finish, mouthfeel) so the entries can be read side by side. Confidence is flagged on each.

Madrecuixe (madrecuishe, madre-cuixe). The "mother" of the karwinskii complex. Producers in Miahuatlán and especially around San Luis Amatlán report that the seeds of madrecuixe, when allowed to germinate, yield offspring across most of the other named karwinskii types. The plant is the longest-trunked and largest-piña form in the group, with the most aromatic distillate of the sub-set.
Aroma: the broadest of any karwinskii; layered roasted-agave sweetness underneath a distinctly mineral lift (the "wet-stone" character of slate and flint, the same quality found in a Chablis or a dry German Riesling) and a faint cocoa-nib note that is almost unique to this sub-variety.
First sip: rounded and sweet on entry, with the cooked-piña sugars and the mineral spine arriving together.
Midpalate: savory and broad, with toasted-almond and dried-fig undertones from the long maturation.
Finish: long, cocoa-tinged, slightly smoky in cooler-roasted expressions and almost wine-like in warmer ones.
Mouthfeel: the heaviest and most coating of the karwinskii sub-set, a function of the largest piña and densest sugar load in the cluster.
Maturation 12 to 18 years. Botanical status: A. karwinskii; confidence high as the most consistently documented sub-variety in producer accounts.

Cuixe (cuishe). Shorter trunk than madrecuixe; narrower leaves; smaller piña. Producers use "cuixe" both as a standalone variety and as a generic for the whole complex outside Oaxaca, a major source of confusion when reading labels.
Aroma: lean and direct; dry herb (especially oregano and bay leaf), white-pepper top notes, less of madrecuixe's cocoa and almost none of barril's honey.
First sip: comparatively dry and lifted, with the cooked-agave sugar reading more as cane husk than as honey.
Midpalate: the "mineral spine" of the karwinskii flavor space, with citrus pith (think grapefruit zest pith rather than pulp) and a touch of green herb.
Finish: clean and pepper-edged, shorter than madrecuixe, with little of the slow-decay sweetness barril carries.
Mouthfeel: medium-light, drier than madrecuixe, more lifted on the palate.
Maturation 10 to 15 years. Botanical status: A. karwinskii; confidence high as a producer category, medium that "cuixe" in San Luis Amatlán is the same population as "cuixe" in Santiago Matatlán (probably is; not formally tested).

Bicuixe (bi-cuixe, vicuxe). Sometimes a regional synonym of cuixe, especially in the Zona Refrescadera around Santiago Matatlán; sometimes a distinct producer-recognized variety with a shorter trunk and a slightly thicker piña than cuixe.

Aroma: the lean herbal-pepper register of cuixe softened by a touch of barril's honey, often described as "cuixe with a little more body."
First sip: drier than barril and rounder than cuixe; the entry has a slightly richer texture without the explicit fruitiness barril carries.
Midpalate: mineral with a touch more body than cuixe, sometimes with stone-fruit pit (apricot kernel) showing up where cuixe gives only citrus pith.
Finish: medium-length, with the pepper edge of cuixe and an undertone of warm cooked-agave that points toward barril.
Mouthfeel: medium, between cuixe's lift and barril's coating richness.
Botanical status: A. karwinskii; confidence medium (often a regional synonym; sometimes a distinct phenotype). Reference bottles: 5 Sentidos Bicuixe, Tr3s Tiempos Bicuixe.

Barril ("barrel"). Named for the barrel-shape of the piña: fatter and more rounded than cuixe or madrecuixe, with a shorter trunk relative to piña diameter. The shape matters because a wider piña stores sugars more like a rosette agave than a trunked one, which is why barril reads sweeter and rounder than its siblings.
Aroma: the most overtly honeyed of the karwinskii sub-set, with orange-blossom and a faint candied-citrus lift; some expressions add a baking-spice note (cinnamon stick, allspice) from the cook.
First sip: notably sweet on entry, with cooked-agave sugar arriving without any of cuixe's herbal dryness.
Midpalate: broad and fruit-leaning; stewed tropical fruit (pineapple, ripe mango) under the agave sweetness, with the mineral character of the karwinskii complex pulled back into the background.
Finish: lingering and slightly oxidative (almost sherry-like in older bottlings), with the honey character extending through.
Mouthfeel: rounder and slightly fuller than cuixe; less coating than madrecuixe but more pillowy than tobaziche. The karwinskii closest to a "fruity mezcal" profile and the easiest entry point for drinkers coming from tequila or añejo styles.
Maturation 10 to 14 years. Botanical status: A. karwinskii; confidence high (one of the most consistently recognized sub-varieties across producer accounts; the morphology is distinctive enough that misidentification is rare).

Tobaziche (tobasiche). Leaner trunked form; narrow leaves; piña relatively small and elongated. The narrow piña stores less sugar than the barrel-form of barril, which is why tobaziche reads dry where barril reads honeyed.
Aroma: the most austere of the karwinskii sub-set; cut green herb (rosemary stem, agave-leaf juice), a flinty mineral edge, almost no fruit, almost no smoke.
First sip: sharp and lifted, with very little entry sweetness; the cooked-agave sugar reads more as savoury cooked-vegetable than as honey.
Midpalate: herbal-green and mineral; some expressions show celery leaf or cilantro stem. The character is closer to a dry biodynamic white wine than to a sweet mezcal.
Finish: dry and pepper-edged, with a mineral aftertaste that lingers without sweetness.
Mouthfeel: the lightest of the karwinskii sub-set, almost crystalline; aerates well in a copita.
In Santa Catarina Minas, tobaziche is frequently called largo, a regional synonym worth flagging. Maturation 10 to 15 years. Botanical status: A. karwinskii; confidence high. Reference bottles: Lalocura Tobaziche, Mezcal Vago Tobaziche, El Jolgorio Tobasiche.

Cirial ("candle"). Named for the straight, candlestick-like trunk and the unusually tall, straight flowering quiote. The candlestick form means the plant invests in vertical-trunk growth at the expense of piña diameter, which produces a smaller, denser piña and a comparatively dry, savory distillate.

Aroma: quiet on entry; dry mineral character, faint cooked-agave, with a subtle dried-tobacco note in some bottlings.
First sip: lean and savory, similar to tobaziche but less herbal and slightly more rounded; the entry sweetness is restrained.
Midpalate: mineral and umami-tinged, with notes some tasters call "savory broth" (mushroom-stock or kombu) and others read as a faint resinous pine.
Finish: longer than cuixe, with a savory tail rather than a sweet or fruity one.
Mouthfeel: medium-light, similar in body to cuixe but with less of the green-herb lift.
Botanical status: A. karwinskii; confidence medium. Reference bottles: Mezcal Nacional Cirial, occasional Mezcaloteca releases.

San martinero (san martín). Producer-recognized variety from San Martín Lachilá and adjacent communities. In the Zona Refrescadera, "san martinero" is often used as a synonym for barril, which means most san martinero bottlings drink very close to barril rather than as a distinct experience.

Aroma: honeyed in the barril direction, with orange-blossom and warm cooked-agave; some San Martín Lachilá expressions add a slightly drier, almost saline edge that the Miahuatlán barril bottlings do not carry.
First sip: rounded and sweet on entry, indistinguishable from barril for most drinkers.
Midpalate: broad and fruit-leaning, with the stewed tropical fruit barril is known for; sometimes a touch of dried apricot.
Finish: lingering and sweet, similar to barril; in the few san martinero bottlings that drink distinctly, the finish carries a faint saline mineral note where barril is purely sweet.
Mouthfeel: rounded and pillowy, matching barril.
Botanical status: A. karwinskii; confidence medium (often a regional alias for barril; sometimes a distinct phenotype, especially in San Martín Lachilá releases).

Largo ("long"). In Santa Catarina Minas, "largo" refers to tobaziche, and the bottle drinks like a tobaziche should: dry, mineral, herbal. In other producer accounts (notably Real Minero's catalog), "largo" denotes a separate long-trunked form distinct from both madrecuixe and tobaziche, with its own sensory profile.

Aroma: clay-influenced (Real Minero's expression is distilled in fired-clay pots rather than copper) with a mineral, faintly damp-stone character on top; dry herb in the cuixe register but with more savory depth.
First sip: lean and savory, with the clay-pot character pulling the entry into a quieter, less fruit-forward space than a copper-distilled karwinskii would occupy.
Midpalate: mineral and herbal-green; touches of cooked-vegetable umami and a faint dairy-cream undertone that drinkers familiar with clay-pot ancestrals will recognize.
Finish: clean and mineral, longer than cuixe but with the same dry-finish discipline.
Mouthfeel: medium-light; the clay-pot distillation produces a quieter, less coating texture than copper-distilled karwinskii expressions.
Botanical status: A. karwinskii; confidence low-medium (genuinely ambiguous between tobaziche-by-another-name and distinct phenotype; depends on producer). Reference bottle: Real Minero Largo.

Tripón. Producer-recognized variety grown almost exclusively in Santa Catarina Minas. Less widely documented than madrecuixe or barril, it appears in Comunidad-series releases and some Real Minero work. The piña is unusually dense for the karwinskii sub-set, which is what producers mean when they call tripón "thick."

Aroma: dense and savory; less aromatic lift than madrecuixe, less herbal than tobaziche, with cooked-agave heaviness and a faint resinous note (think pine pitch or charred mesquite resin) in some expressions.
First sip: broad and weighted on entry; the cooked-piña sugars arrive without the bright mineral lift of cuixe or the candied character of barril.
Midpalate: the "thickest" of the karwinskii sub-set in producer language; the spirit coats the palate in a way few of its siblings do, with savory depth and a slight smoke undertone (a function of the piña density holding heat longer through the cook).
Finish: long and savory; less sweet than madrecuixe, less peppery than cuixe, with the weight extending well past the swallow.
Mouthfeel: the heaviest and most coating of the karwinskii sub-set after madrecuixe, sometimes heavier in lower-proof bottlings where the weight has nowhere to go.
Botanical status: A. karwinskii; confidence medium-low (limited geographic range; minimal academic documentation; the local-population character may or may not survive a future taxonomic revision). Reference bottle: Comunidad #12 Wild Tripon.

Pata de borrego ("sheep's foot"). Producer-recognized variety from Oaxaca's Sierra Sur, named for the trunk and basal-rosette shape that producers compare to an ovine hoof.

Aroma: likely within the karwinskii mineral-and-herbal register; producer notes suggest a savory-leaning profile closer to tobaziche than to barril, with less aromatic lift than madrecuixe.
First sip and midpalate: the morphology and the maturation window predict a moderate body and a savory-mineral character, but the lack of bottling samples means specific notes (citrus, herb, smoke) cannot be asserted.
Finish and mouthfeel: not enough source material to describe with confidence; expect medium body and a dry-leaning finish if the karwinskii family pattern holds.
Botanical status: provisionally A. karwinskii; confidence low.

Sierrudo. The most contested name in the complex. Treated separately in the next section, including its likely A. americana attribution and the distinct sensory profile that follows from being the wrong species for this list.

The sierrudo placement controversy

Sierrudo is the karwinskii sub-variety that is probably not a karwinskii sub-variety at all. Some producer sources and several European importers list sierrudo as a karwinskii form. Other sources, including the South Embassy import label and Sacapalabras Mezcal, place it explicitly in Agave americana with 18 to 25 year maturation and a much larger piña than typical karwinskii.

Three pieces of evidence stand against placing sierrudo squarely in A. karwinskii. Maturation time: reliable sierrudo bottlings consistently report 18 to 25 years; karwinskii typically matures in 10 to 18 years. Piña size: sierrudo piñas are routinely 50 to 80 kilograms in producer notes, comfortably in A. americana territory and well above the karwinskii typical 8 to 25 kilograms. Growth form: some sierrudo populations are described as classic A. americana large-rosette plants without the karwinskii woody trunk.

Counter-evidence in favor of karwinskii placement is mostly nomenclatural; the name appears alongside karwinskii sub-varieties in some producer catalogs rather than being supported by morphological or genetic data.

The broader point sierrudo illustrates is that vernacular names in Oaxacan mezcal can attach to plants of more than one species, especially when the producer-recognized morphology and the formal botanical species don't align. The sierrudo question will not be closed by a single paper; it will be closed by per-bottle producer attribution and, eventually, by population-genomic sampling that tells us which "sierrudo" plants share ancestry with karwinskii populations and which share ancestry with A. americana populations.

Producer-traditional vs. botanical-scientific taxonomy

The cleanest way to understand the tension is to notice what the two communities use the names for. A maestro mezcalero uses a name like barril or cirial to predict roast yield, fermentation behaviour, and distillate character: practical production-side outcomes the maestro has tracked across many seasons. The name is operationally useful regardless of whether it corresponds to a published subspecies. Two plants the maestro calls madrecuixe will, in his experience, behave more like each other than either will behave like a plant he calls barril.

A botanical taxonomist uses a name to denote a population that meets criteria of monophyly, morphological distinctness, reproductive cohesion, and geographic coherence: criteria that derive from a different operational frame (taxonomic stability, herbarium logic, peer-reviewable distinction).

These two frames can agree, and sometimes do. A. cupreata and A. inaequidens are good cases, where producer-recognized "papalote" and "lechuguilla" map cleanly onto distinguishable species. Sometimes they don't. "Lechuguilla" as a Spanish name attaches to four or five unrelated agaves and one Dasylirion, depending on which side of which sierra one is standing on.

For A. karwinskii, three things follow.

Editorial consequence. Until formal varietal or subspecies names are published, the safest practice is "A. karwinskii, locally called madrecuixe" rather than "A. karwinskii var. madrecuixe." The latter is a botanical claim that has not yet been substantiated. This site holds that line consistently across every spirit and producer page.

Regulatory consequence. NOM-070-SCFI-2016 (the mezcal NOM, currently under review) requires species disclosure for varietal mezcal but does not regulate sub-variety naming. Producers can label madrecuixe or barril with no formal botanical backing; the label is a marketing claim, not a regulatory category. If a future NOM revision codifies sub-variety names, it will do so on producer-knowledge grounds, in advance of (and possibly in collision with) the formal taxonomy. Worth watching.

Sustainability consequence. Sub-varieties that turn out to be genetically distinct populations may have different sustainability profiles. Madrecuixe, the "mother" with the broadest cross-pollination capacity, may be ecologically more valuable to preserve as standing wild seed parents than the same-genus offspring varieties. Conservation planning that treats all A. karwinskii as a single management unit may mask important within-species pressure on specific sub-populations.

What comes next

The literature is moving. With a complete genome assembly in hand and a phenotypic structure paper already published, the discriminating population-genomic study is methodologically tractable. The question is who will publish it first and what they will find. Three outcomes are all live.

The "Hypothesis B confirmed" outcome. Sampling across producer-named sub-varieties at multiple sites recovers genetically structured populations that correspond closely to the producer names. The literature catches up; the next NOM revision codifies sub-variety labelling; the bottle label discipline tightens around what is now informally implied.

The "partial confirmation" outcome. Some producer names (madrecuixe, barril, tobaziche) recover as discrete clades; others (san martinero, largo) collapse into existing types or split further. The producer naming system gets re-mapped onto the formal taxonomy with a handful of clean wins and a handful of necessary corrections.

The "Hypothesis A holds for most names" outcome. Most producer-named sub-varieties turn out to be ecotypes, not genetic clades. The botanical recognition fails to materialize, but the operational producer practice continues unchanged because the morphologies and the distillate behaviors are still real even if they do not map onto distinct populations.

Whatever the eventual outcome, the next 24 to 48 months should produce the discriminating study. Until it lands, the editorial discipline this site applies is the only honest one: treat sub-variety names as locally meaningful, operationally useful, and provisionally botanical. Cite the producer attribution. Flag the species attribution as disputed where it genuinely is. And keep watching for the paper.

The chapter you have just read is itself a snapshot, dated 2026-05-28. When the discriminating publication lands, this page is one of the first this site will revise.