Terraced Agriculture — How Mountainous Civilizations Fed Themselves

The analysis of ancient terracing systems reveals a consistent pattern: they were built as integrated water management and food production systems, and they fail — both agronomically and structurally — when the water management component is neglected. This pattern has been demonstrated repeatedly as terrace systems are abandoned, partially restored, or improperly repaired without understanding the hydraulic logic of the original design.

Andean andenes: scale, engineering, and function. The Inca terrace system was the largest agricultural infrastructure project in the pre-Columbian Americas, and arguably in the ancient world. The total area of Andean terracing is estimated at 500,000 to 1,000,000 hectares, though much of this is now abandoned or in disrepair. Active terracing still covers perhaps 150,000 to 200,000 hectares in Peru and Bolivia.

The engineering sophistication of Inca andenes has been studied extensively by Clark Erickson, John Earls, and others. Key features include: 1) Subterranean drainage layers beneath each terrace platform, typically consisting of coarse gravel and stones, which prevent waterlogging while retaining moisture in the overlying soil. 2) Irrigation channels, often constructed from shaped stone, that delivered water from upstream sources to the highest terraces, with gravity distribution downslope. 3) Soil composition design — Inca agricultural workers transported specific soil types from lower elevations to construct optimal growing media on high-altitude terraces. 4) Microclimate engineering — the dark stone walls of andenes absorb solar radiation during the day and release it at night, creating a temperature buffer that extends the growing season at high altitudes and reduces frost risk. Night frost is the primary agricultural constraint above 3,500 meters in the Andes; the thermal mass of stone retaining walls was a deliberate response to this constraint.

Experiments conducted by the Peruvian Ministry of Agriculture in restored andenes near Moray and Andahuaylillas in the 1980s and 1990s documented that restored terrace systems produced yields of potato, quinoa, and maize comparable to those achievable on prime lowland agricultural land, with no synthetic inputs and far less erosion loss. The thermal mass effect was confirmed experimentally: air temperature within terrace systems was measured at 2 to 3 degrees Celsius warmer at night than on adjacent unterraced slopes, sufficient to shift frost occurrence probabilities significantly.

The abandonment question. Large fractions of Andean terrace systems were abandoned after the Spanish conquest, primarily because the population collapse from epidemic disease eliminated the labor force required to maintain them. Terrace maintenance is not optional in the way that maintenance of some agricultural systems is — without regular repair of retaining walls and drainage systems, terraces fail by a specific sequence: vegetation establishes on retaining walls, root pressure destabilizes the wall, a section collapses, the terrace platform is undermined, and the fill material — sometimes centuries of accumulated soil — washes downslope in a single erosion event.

The labor requirements for terrace maintenance are estimated at 3 to 5 person-days per hectare per year for routine wall maintenance, increasing to 20 to 50 person-days per hectare for major repair after collapse. Terrace systems abandoned for decades or centuries can require substantial reconstruction investment before they become productive again. The Center for Traditional Andean Agriculture (PRATEC) and various NGOs have documented both the costs and the returns of terrace restoration in Peru — typical payback periods of three to seven years from the point of restored productivity, with subsequent yields that justify the initial investment.

Bali's Subak: water temples and rice terrace management. The Bali irrigation system — the Subak — is among the most studied traditional water governance systems in the world, in part because of the work of ecological anthropologist J. Stephen Lansing, whose 1991 book "Priests and Programmers" documented how the water temple network coordinated irrigation and pest management across a complex landscape of terraced rice paddies.

The Subak system allocates water from shared irrigation sources to individual farms through a hierarchy of associations — the subak, which manages a single water source, and the supra-subak organizations that coordinate between watersheds. Water allocation decisions are made in water temple ceremonies that integrate ritual with practical water management: the timing of fallow periods, which temples coordinate through ritual calendar, is simultaneously a spiritual practice and a pest management strategy. When multiple subaks fallow their fields simultaneously — a requirement of some ceremonies — the landscape-level fallow synchronizes pest populations' disruption and prevents the buildup of the brown planthopper and other rice pests that devastate asynchronous systems.

The Green Revolution disrupted this system deliberately. Agricultural extension programs promoted continuous high-yielding variety cultivation and synthetic pesticide application, eliminating the need for fallow synchronization. Pest outbreaks followed. In areas that adopted Green Revolution practices, planthopper outbreaks caused crop losses of 20 to 40 percent within a few years. In areas that maintained Subak-coordinated cultivation, outbreaks were minimal. Lansing and Kremer documented this contrast using computer modeling of the Bali watershed, demonstrating that the traditional system's pest control performance was not coincidental — it was the product of an evolved, self-organized management system that optimized across the entire watershed simultaneously.

UNESCO recognized the Subak system as a World Heritage Cultural Landscape in 2012 — a designation that acknowledges it as a living system of human-environment interaction, not merely a historical site. The designation was a response to active advocacy by Balinese farmers and the Balinese government, who sought international protection for a governance system that was being eroded by tourism development, land conversion, and continuing Green Revolution promotion.

Ifugao terraces: muyong and communal governance. The Ifugao terracing system of Luzon, Philippines, operates through a land tenure arrangement that is unusual in combining private forest ownership with communal water governance. Each farming family maintains a muyong — a private woodlot typically occupying higher slopes above their terrace fields. The muyong is maintained as forest, which captures rainfall, slows runoff, and delivers a steady water supply through small springs and seeps to the terrace irrigation channels below.

The incentive structure is direct: the family that neglects its muyong — allows it to be cleared or degraded — reduces its own water supply and consequently its rice yield. This private incentive is reinforced by communal social pressure, since muyong management affects downstream water availability for other terrace farmers. The system requires no state enforcement mechanism and no external monitoring — the incentives are built into the productive relationship between forest and field.

UNESCO listed the Ifugao rice terraces as endangered in 2001, citing abandonment of terrace farming by younger generations seeking urban employment, neglect of muyong management, and infestation by golden apple snail (an invasive species introduced by industrial aquaculture programs). Restoration programs have focused on supporting traditional farming as an economically viable livelihood — a problem that is as much about commodity price structures and rural-urban wage differentials as it is about ecology.

Yemen's highland terraces. The terraces of the Yemeni highlands represent a third independent tradition of mountain terrace agriculture, adapted to an arid environment rather than a humid one. Yemeni terraces are designed primarily to capture and retain the scarce and irregular rainfall of the highland zone — each terrace platform acts as a catchment that concentrates rainfall on a limited cultivated area. Stone retaining walls, carefully fitted without mortar, are built with slight inward lean to maximize structural stability and to reduce the temperature of the soil surface through shade.

The crops grown on Yemeni terraces — sorghum, millet, qat, coffee, fruit trees — are adapted to the terrace microclimate, which is measurably cooler and more humid than adjacent unterraced slopes. The coffee of the Haraz and Haima highlands, among the world's oldest cultivated coffee varieties, grows on terraces that have been maintained for at least 1,000 years and possibly much longer.

The ongoing conflict in Yemen has devastated terrace maintenance. Without the labor of local communities, retaining walls have collapsed, and the erosion loss of accumulated soil on abandoned terraces represents centuries of productive capacity being washed away. The reconstruction of Yemeni terrace agriculture — when political stability eventually permits — will require not only physical reconstruction but the reassembly of the knowledge and community structures that operated the system.

The engineering lesson. Terrace agriculture demonstrates a principle that is transferable to contemporary land management: radical landscape modification for long-term productivity is achievable with human labor and local materials over multi-generational time scales, and produces infrastructure that improves in function over time rather than degrading. The contrast with industrial agricultural infrastructure — which requires continuous external inputs to maintain performance and degrades when those inputs are withdrawn — is direct.

The terrace systems of the Andes, Bali, the Philippines, and Yemen were all built without external capital, without synthetic materials, and without the internal combustion engine. They were built by communities with intimate knowledge of their specific landscape, motivated by the direct relationship between their labor and their food supply. This is the model of infrastructure investment that produces durable systems: local materials, local knowledge, local incentives, and time scales measured in generations rather than quarterly returns.