Farming: The Domestication Centres and the Great Systems

Status: Researched draft — cultural review pending

How to read this entry: Documented fact = published archaeology, archaeobotany, genetics, or ethnohistory cited in Sources. Tradition — the people's own account = Indigenous oral history or a people's own published teaching, labelled as such. Debated / Claim = a point where published scholars disagree, or a figure given only as a sourced range. Scope note: a history of farming in the Americas, system by system, each named for its own people and place.

Farming Began More Than Once

• No single origin. Documented fact: Farming in the Americas did not spread from one centre. Syntheses recognise several independent centres where local peoples domesticated local plants: Mesoamerica, the Andes–Amazonia region, and eastern North America, among others (Smith, 2006, PNAS; Piperno, 2011, Current Anthropology). Each domesticated different species, in different environments, on its own chronology.
• A long transition. Documented fact: Dependence on domesticates took thousands of years — some 6,000 years in Mesoamerica on Smith's Guilá Naquitz argument (Smith, 1997). Later systems combined crops from more than one centre; that movement is history, not a single origin.

1. The Independent Centres

Mesoamerica — Squash First in the Record, Then Maize

• Squash at Guilá Naquitz, Oaxaca. Documented fact: Seeds, peduncles and rind of domesticated squash (Cucurbita pepo) from Guilá Naquitz cave, Valley of Oaxaca, Mexico, were directly dated by accelerator mass spectrometry (AMS). Nine AMS dates place cultivated C. pepo there between about 10,000 and 8,000 calendar years ago, identified as domesticated by increased seed length and peduncle diameter and changes in fruit shape and colour (Smith, 1997, Science). At publication this predated directly dated maize and beans by more than 4,000 years.
• Maize from teosinte. Documented fact: Maize (Zea mays) was domesticated from Balsas teosinte, Zea mays subsp. parviglumis, native to the Balsas River basin of southwestern Mexico (Matsuoka et al., 2002, PNAS; Doebley, 2004, Annual Review of Genetics). Much of the stark morphological difference maps to a few major loci, including tb1 and gt1 (Doebley et al., 1990; Wang et al., 2015). Mangelsdorf and Reeves's 1938 Tripsacum hybridisation hypothesis was refuted by genetics.
• Maize dating — disagreement printed. Debated / Claim: Genetic divergence estimates place a single domestication about 9,000 years ago in the central Balsas valley (Matsuoka et al., 2002). The oldest directly dated cobs are younger: maize from Guilá Naquitz at about 6,250 calendar years ago (Benz, 2001, PNAS; Piperno & Flannery, 2001, PNAS). Maize phytoliths and starch on stone tools from Xihuatoxtla rock shelter, Guerrero, were reported from layers dated to about 8,700 calendar years ago (Piperno et al., 2009, PNAS). These measure different things and are not interchangeable.
• Beans. Documented fact: Common bean was domesticated in Mesoamerica (and separately in South America); its earliest directly dated Mexican beans are about 2,300 years ago, Coxcatlán Cave (Smith, 2006).

Andes–Amazonia — Potato in the Highlands, Manioc in the Lowlands

• Potato — single highland origin. Documented fact: Genetic analysis supports a single domestication of cultivated potato (Solanum tuberosum) from the Solanum brevicaule complex in the highlands of southern Peru, in the wider Lake Titicaca region (Spooner et al., 2005, PNAS). Summaries place it about 8,000–10,000 years ago; farmers selected against bitter glycoalkaloids and for larger, clustered tubers.
• Potato diversity. Documented fact: By the Inca period (15th–early 16th century CE) Andean farmers cultivated thousands of named varieties roughly between 3,000 and 4,500 metres, a diversity still conserved in part in the International Potato Center (CIP) genebank in Lima (CIP, n.d.). Debated / Claim: Precise historic totals ("more than 3,000," "more than 4,000") depend on whether folk varieties, landraces or accessions are counted; this entry prints "thousands" only.
• Manioc in southern Amazonia. Documented fact: Manioc (Manihot esculenta) was domesticated in the southern Amazon from wild subsp. flabellifolia (Olsen & Schaal, 1999, PNAS; Piperno & Pearsall, 1998). Starch shows it at Áspero and Caral, coastal Peru, in 3000–1800 BCE by exchange (Scientific Reports, 2023) — movement, not coastal domestication.

Eastern North America — The Eastern Agricultural Complex

• An independent centre, defended in print. Documented fact: Bruce D. Smith's PNAS synthesis, "Eastern North America as an independent center of plant domestication" (2006), reviewed archaeological and genetic evidence for marshelder (Iva annua), chenopod (Chenopodium berlandieri), squash (Cucurbita pepo) and sunflower (Helianthus annuus), and concluded the evidence strongly supports independent domestication of all four in the eastern United States.
• Squash, twice. Documented fact: Cucurbita pepo was domesticated independently in Mexico and in eastern North America, as subsp. ovifera (Sanjur et al., 2002; Smith, 2006). Eastern squash was grown initially more for edible seeds and for rind containers than for thick flesh (Fritz, 2019).
• The full complex. Documented fact: The Eastern Agricultural Complex (EAC) includes domesticated chenopod (subsp. jonesianum), marshelder (var. macrocarpa), sunflower, squash and erect knotweed (Polygonum erectum subsp. watsoniae), plus cultivated maygrass and little barley (Fritz, 2019; Mueller, 2017, 2018; Smith, 1987, 2007, 2011; synthesised in American Antiquity, 2024). It was important west of the Appalachians for over 4,000 years.
• Dates and habitat. Documented fact: Syntheses place incipient cultivation at about 5300 BCE, with several species cultivated by about 1800 BCE, including at the Riverton Site, Illinois (Smith, 2006; Fritz, 2019). Chenopod, marshelder and squash are floodplain weeds of disturbed soils after floods. Species sequences remain Debated / Claim, repeatedly revised by AMS dates.
• Displacement. Documented fact: Maize from Mexico was present in small amounts by about 200 BCE in some syntheses and became central much later, especially after about 800–1000 CE. Cultivation of most local domesticates other than squash and sunflower then declined and was abandoned, and the domesticated forms of chenopod, marshelder and erect knotweed were lost (Smith, 2006; Fritz, 2019).

2. The Three Sisters and the Spread North

• The system. Documented fact: Three Sisters planting intercrops maize, beans and squash in hills: maize supports climbing beans, beans fix nitrogen through root bacteria, and squash covers the ground, shading soil and retaining moisture (Lewandowski, 1987, Agriculture and Human Values; U.S. National Agricultural Library [NAL], n.d.). Versions were practised by Haudenosaunee, Cherokee, Mandan, Hidatsa, Hopi and Zuni, among others, with local names (NAL, n.d.).
• Assembly in time. Documented fact: Squash was present for millennia; maize arrived over centuries; beans latest, the full trio widespread in the Northeast roughly 1000–1300 CE (Hart & Lovis, 2013; Fritz, 2019).
• Haudenosaunee fields — women's agriculture. Documented fact: Among the Haudenosaunee (Mohawk, Oneida, Onondaga, Cayuga, Seneca, Tuscarora), the Three Sisters — Diohe'ko, "these sustain us" in Seneca usage recorded by Arthur C. Parker (1910) — were the foundation of subsistence, planted in hills and cultivated and harvested by work parties of women (Lewandowski, 1987; Parker, 1910, in NAL, n.d.). Women farmed assisted by children, most maize dried for winter (New York State Museum, n.d.).
• Tradition — Haudenosaunee account. Corn, beans and squash sprouted from the body of Sky Woman's daughter — corn from her breasts, beans from her hands, squash from her navel (Carnegie Museum of Natural History, 2018; LaFrance, in Words That Come Before All Else).
• Mandan, Hidatsa and Arikara gardens. Documented fact: On the upper Missouri in present-day North Dakota, Mandan, Hidatsa and later Arikara women farmed fertile river bottomlands in family gardens typically about 3 to 5 acres, growing maize, beans, squash and sunflowers (Wilson, 1917; State Historical Society of North Dakota [SHSND], North Dakota Studies, n.d.). Varieties were kept in separated patches, preserving early-maturing northern varieties (SHSND, n.d.).
• Buffalo Bird Woman's record. Tradition — the people's own account, as recorded: Buffalo Bird Woman's Garden (1917, as Agriculture of the Hidatsa Indians), compiled by Gilbert L. Wilson from Waheenee — Maxidiwiac, Buffalo Bird Woman (born about 1839, Hidatsa) — who learned from her mothers (her mother and her mother's sisters) from about age 12 (Wilson, 1917; SHSND, n.d.). She describes bird-watching from a platform, digging-stick preparation, bison-scapula hoe and antler rake, and maize in hills about four feet apart. Wilson is recorder; the knowledge is Hidatsa.

3. Hohokam Irrigation — Canals in the Sonoran Desert

• People and place. Documented fact: The Hohokam — Huhugam, "those who have gone," in Tohono O'odham usage; the Akimel O'odham and Tohono O'odham are their descendants in published tribal and archaeological accounts — farmed the Salt and Gila valleys of southern Arizona from roughly the early centuries CE to about 1450 CE, growing maize, beans, squash, cotton, tobacco and agave (National Park Service [NPS], 2017; Haury, 1976).
• Snaketown. Documented fact: Snaketown, on the Gila River in the present Gila River Indian Community, was about 250 acres at maximum extent, with population estimated as a range at about 1,000 to 3,000, a plaza and ballcourts, and surrounding farmland (Haury, 1976, The Hohokam: Desert Farmers and Craftsmen; NPS, 2017). It was backfilled after excavation and is closed to the public by decision of the Gila River Indian Community (NPS, 2017).
• Canal extent — ranged. Debated / Claim: Totals differ by what is counted (mains only or branches too; one valley or Salt and Gila together; one period or all summed). Print figures include about 200 to 250 miles in older summaries, about 500 miles in the Phoenix Basin alone in others, and individual mains at 7 to 10 miles, and in basin accounts 20 to 30 miles (NPS, 2017, reporting canals up to 10 miles, averaging about 10 feet deep and 30 feet wide; Phoenix Basin syntheses). Acreage is likewise ranged: about 25,000 acres for the Phoenix Basin in one synthesis, up to 70,000 acres in accounts centred on the Gila / Snaketown system. No single total is asserted; the network was North America's largest documented pre-Columbian irrigation system, built without draft animals or wheels.
• Working and decline. Documented fact: Woven-mat dams channelled flow into mains and distribution canals (NPS, 2017). Debated / Claim: Major centres were largely depopulated by about 1450 CE after decline from the 1300s. Published explanations combine flooding that damaged headworks, tree-ring drought, waterlogging and soil salinisation from long irrigation, and social reorganisation (Bayman, 2001; Fish, 2007, in later syntheses). No single cause is established.

4. Mexica Chinampas — Farming the Lakes of Tenochtitlan

• Not floating gardens. Documented fact: A chinampa is a raised field anchored in shallow lake water: stakes in the lakebed, a woven enclosure, layers of mud, vegetation and soil built to just above the surface, canals between fields, and willows at the edges whose roots stabilise the plot (Armillas, 1971, Science, "Gardens on Swamps"; Coe, 1964). "Floating gardens" is a misnomer.
• Maintenance. Documented fact: Fertility was kept by dredging canal mud onto fields, and by seedbeds and transplanting (Armillas, 1971; Sanders et al., 1979).
• Place and time. Documented fact: The great districts lay in fresh-water Chalco and Xochimilco, south of brackish Texcoco, protected by dikes expanded in the 15th century after the Mexica subdued the southern lake city-states by about 1465 (Armillas, 1971; Parsons, 1976). Chinampas predate the empire; the gridded expansion to feed Tenochtitlan (founded 1325) was Mexica-period.
• Extent — Armillas and his critics. Debated / Claim: Pedro Armillas (1971), from aerial photographs, estimated about 12,000 hectares reclaimed in Chalco–Xochimilco, reduced by him to about 9,000 viable hectares after allowing for canals — the figure used by Parsons (1976) and Sanders et al. (1979). Later re-measurement (Crossley, 1999; Frederick, 1999) argued the photogrammetry exaggerated field area, with field-to-canal ratios of roughly 3:2 to 1:1. This entry gives about 9,000–12,000 hectares reclaimed, viable field area debated downward from 9,000, and no more precise total.
• Yields — ranged. Debated / Claim: Published maize yields include about 3 tonnes per hectare in accounts of chinampas supplying Tenochtitlan and about 4 tonnes per hectare cited from Sanders (1957), with two to three harvests per year of different crops on the most intensive plots (Armillas, 1971; Sanders et al., 1979). On assumptions of about 160 kg of maize per person per year, Armillas and Sanders modelled support for about 117,000 to 200,000 people. Those are model outputs: Tenochtitlan also drew food by tribute and trade, and chinampas grew many crops besides maize.

5. Maya Systems — Milpa, Terraces, Wetland Fields, Forest Gardens

• A repertoire, not one system. Documented fact: Maya peoples farmed lowlands and highlands with a varying combination of gardens, milpa, terraces, wetland fields and managed forest (Turner & Harrison, 1983; Ford & Nigh, 2015). "Slash-and-burn" alone misstates it.
• Milpa as a managed cycle. Documented fact: For Ford and Nigh, from El Pilar archaeology and living Maya farmers, milpa is a 12-to-24-year cycle, not abandonment: forest cut and burned for annuals, then managed orchard-garden, then closed-canopy useful trees — every phase tended (Ford & Nigh, 2015).
• Forest gardens. Documented fact / Debated in degree: Ford found dominant El Pilar forest trees are disproportionately useful species, arguing the forest is largely a relic forest garden (Ford & Nigh, 2015). How far it explains the whole forest remains Debated / Claim.
• Terraces. Documented fact: On slopes Maya farmers built stone terraces — limestone retaining walls on bedrock with soil behind them — checking erosion, retaining moisture and levelling planting surfaces. Systems around Caracol, Belize, are linked to supporting that city's large Classic population (Chase & Chase, 1998; Foster, 2015).
• Pulltrouser Swamp raised fields. Documented fact: In wet lowlands some Maya built raised fields by digging canals and heaping the material, renewing canal muck on top. At Pulltrouser Swamp, northern Belize, B. L. Turner II and Peter D. Harrison's project excavated three sites and concluded from soils, pollen, settlement and artefacts that the features are Maya-made raised and channelised fields, with settlements on terraces 2–3 metres above the swamp (Turner & Harrison, eds., 1983, Pulltrouser Swamp).
• Extent — disagreement. Debated / Claim: Early aerial surveys claimed raised fields across much of the southern lowlands, including almost 3,200 hectares for northern Belize (Siemens & Puleston, 1972; Turner & Harrison, 1981). Ground-truthing found some features were natural, and a major review concluded definitive evidence is confined to a few places — Pulltrouser and Albion Island, Bajo Morocoy, Río Azul, and south of La Milpa, with only modest constructions south of Tikal (Johnston, 2003, Journal of Anthropological Archaeology). Pulltrouser itself is not in doubt; its regional scale is.

6. Andean Systems — Terraces, Vertical Control, Raised Beds, Chuño

• Terracing. Documented fact: Andean terraces (andenes) — stone-faced slope platforms — levelled soil, reduced erosion and controlled irrigation. They long predate the Inca, who expanded earlier systems (Denevan, 2001).
• Vertical archipelago — at published level. Documented fact (model): John V. Murra, from 16th-century Spanish visitas, proposed "vertical control of a maximum of ecological floors": a community (ayllu) holding dispersed islands at different altitudes — maize valleys, higher potato and quinoa plots, higher still camelid pasture, and lower eastern-slope coca plots — farmed by colonists retaining rights in the core (Murra, 1972; 1975, Formaciones económicas y políticas del mundo andino). Debated / Claim: Later scholars (Van Buren, 1996) argue it was applied too uniformly and trade also moved goods between floors.
• Suka kollus at Lake Titicaca. Documented fact: Around Lake Titicaca (about 3,800 metres, Peru and Bolivia), farmers built raised beds in Aymara suka kollus (also suka qullu; waru waru in Peru, camellones in Spanish): platforms several metres wide separated by water-filled canals (Erickson, 1985, 1988; Kolata, 1996, Valley of the Spirits). Major expansion is associated with Tiwanaku (roughly 500–1000 CE), with earlier and later use in the basin; construction in Puno is reported as beginning as early as about 1000 BCE (Erickson, 1987).
• Frost, and the experimental record. Documented fact: Clark Erickson's Huatta project (1981–1986, with agronomist Ignacio Garaycochea and local farmers) excavated ancient beds and rebuilt them on ancient models. Canal water absorbs heat by day and releases it at night, blanketing beds in warmer air and reducing frost damage — the central hazard at Titicaca — while canals drain waterlogged soil, supply moisture and yield muck dredged onto beds (Erickson, 1985, 1988). Debated / Claim: Basin extent estimates range from tens of thousands of hectares to over 80,000–100,000 hectares (Díaz & Velásquez, 1991, reporting about 102,440 hectares of remains in Puno). Experimental yields exceeded local controls, but precise tonnages vary; no single figure is asserted.
• Diversity in practice. Documented fact: Households planted mixtures of varieties across altitudes, spreading risk, and maintained diversity through seed exchange (Brush, 2004; CIP, n.d.).
• Chuño — documented practice. Documented fact: Chuño (ch'uñu) is potato freeze-dried by the altiplano climate. In dry winter (June–July), small, often bitter potatoes are spread on flat ground to freeze in sub-zero nights (around −5 °C or colder above 3,800 metres), thaw in sun, and over several nights are trampled underfoot by family groups to press out water and loosen skins. Dried in sun they become black chuño; soaked in running water for weeks and peeled before drying, white chuño or tunta (Acosta, 1590; modern ethnographic accounts). Dried chuño keeps for years and provisioned Inca armies via state storehouses (qollqas).
• Chuño — antiquity, carefully stated. Documented fact / Debated in depth: Starch evidence identified as chuño is reported at the Middle Horizon site of Quilcapampa La Antigua, Peru (about 600–1000 CE), and chuño remains at Tiwanaku sites — at least several centuries before the 15th-century Inca Empire, whose chroniclers described it as established (Quilcapampa starch study, 2022; Tiwanaku finds in chuño literature). Deeper claims rest on inference, not directly dated chuño.

• Continuity. Documented fact: These systems continue as living practice — Three Sisters, Xochimilco chinampas, Maya milpa, and Andean terraces, chuño and rebuilt suka kollus (Ford & Nigh, 2015; Erickson, 1985, 1988).

Sources

Acosta (1590); Armillas (1971) Science 174; Benz (2001) PNAS 98; Brush (2004) Farmers' Bounty; Carnegie Museum of Natural History (2018); Chase & Chase (1998), in Foster (2015); Denevan (2001); Doebley (2004) Annual Review of Genetics 38; Erickson (1985, 1987, 1988); Ford & Nigh (2015) The Maya Forest Garden; Fritz (2019); Haury (1976) The Hohokam; Johnston (2003) Journal of Anthropological Archaeology 22; Kolata (1996) Valley of the Spirits; Lewandowski (1987) Agriculture and Human Values 4; Matsuoka et al. (2002) PNAS 99; Mueller (2017, 2018); Murra (1972, 1975); National Park Service (2017); New York State Museum (n.d.); Olsen & Schaal (1999) PNAS 96; Parker (1910), in U.S. National Agricultural Library (n.d.); Parsons (1976); Piperno (2011) Current Anthropology 52; Piperno & Flannery (2001) PNAS 98; Piperno & Pearsall (1998); Piperno et al. (2009) PNAS 106; Sanders, Parsons & Santley (1979) The Basin of Mexico; Scientific Reports (2023) 13:14143 (Áspero); Smith (1997) Science 276; Smith (2006) PNAS 103; Spooner et al. (2005) PNAS 102; State Historical Society of North Dakota (n.d.); Turner & Harrison, eds. (1983) Pulltrouser Swamp; Van Buren (1996) American Anthropologist 98; Wilson (1917) Agriculture of the Hidatsa Indians; International Potato Center (n.d.); Wang et al. (2015) Genetics 200.

Flagged claims and weakest-sourced points: (1) Maize chronology — genetic (~9,000 BP), Xihuatoxtla (~8,700 cal BP, indirect) and cobs (~6,250 cal BP) measure different things. (2) Hohokam totals (200–250 to ~500 miles; 25,000–70,000 acres) vary by scope counted. (3) Chinampa area and yields are model outputs, Armillas's figures argued downward. (4) Maya wetland-field regional scale — Pulltrouser secure, wider claims not ground-truthed. (5) Suka kollus extent and tonnages vary by report. (6) Chuño antiquity beyond c. 600–1000 CE evidence, and precise potato variety counts. (7) EAC synthesis dates still revised by AMS work.