Abstract
The arid desert coast of northern Peru has traditionally been viewed either as existing in stasis, or as experiencing punctuated change from sudden flood events, followed by a return to system equilibrium. Despite these environmental extremes, the region was home to agriculture-based societies for millennia, and the success of these farming systems is considered an early example of irrigation technology transforming marginal landscapes. However, a closer examination of the long-term human-environment history of the Chicama Valley, one of the largest valleys in the coastal region, demonstrates that this landscape is the product of protracted interactions across at least three systems: the local environment, El Niño Southern Oscillation (ENSO), and farming. Here, El Niño floods, typically considered high-risk events, are fundamental to local biodiversity and renewal, resulting in a desert ecosystem that is both robust and elastic. The prehispanic farmland known as the Pampa de Mocan (1100BC–AD1460), is presented as a case study to observe the co-evolution of agricultural technology and an ENSO-hyper-arid environment. This ancient farming system developed the capacity to toggle between sudden floodwater inputs and periods of water scarcity. Alongside water and soil conservation practices, prehispanic agriculturalists implemented technologies that were designed to mitigate El Niño flooding and incorporate its byproducts to supplement available resources. The convergence of these interacting systems on the Pampa de Mocan offers new insights into the role of risk in building resilience.
Introduction
Hyper-arid, marginal landscapes, such as the north coast of Peru, are considered fragile ecosystems, where the slightest change in biomass, moisture, or human intervention has the potential to throw the system out of balance (Sullivan, 1996). This view leaves little room for investigating the resilience and dynamism of these landscapes, even less so once natural disasters enter the equation. Despite the apparent risk, scholars document irrigation-based agriculture, here, on the arid north coast of Peru, by 2150 cal BC (Pozorski and Pozorski, 1987; Pozorski and Pozorski, 2005). The north coast is one of the most intensively managed landscapes in the Andes, and archeologists have long credited the development of sophisticated canal systems for the transformation of this desert into an agriculturally productive area, tying the local culture history inextricably to the history of irrigation (Farrington, 1977).
Tracing the development of irrigation technology on the north coast is inherently challenging, in part, because modern farming has erased or obscured most of the prehispanic record. Scholars have turned to case-studies from other arid-zone centers of agriculture, including Mesopotamia, to approximate the patterns of growth and impact of irrigation systems on the desert environment (Billman, 2002; Moseley, 1983; Steward, 1955). These models center the early stages of expansion near riverbanks, where headgates and canals were dug to carry water to rich alluvium on the nearby floodplain. To bring more land under cultivation and avoid some of the negative effects of irrigation, such as salinization, canal branches were extended, drawing water along longer distances, further from the river source (Gibson, 1974; Moseley, 1974, 1983). On the north coast of Peru, the earliest phases of complex irrigation are associated with the Cupisnique archeological culture (loosely dated to 1100–500BC) (Billman, 2002; Jones, 2010; Pozorski and Pozorski, 1979). Over time, alongside a growing population and production needs, the Moche (AD200–900) and Chimu (AD900–1460) societies expanded canal systems to bring more land under cultivation than has ever been achieved before or since (Clement and Moseley, 1991; Koons and Alex, 2014; Moseley, 1983). Settlement patterns, which, due to preservation issues, are dominated by monumental sites rather than small farmsteads, tend to support this river-out expansion model (Billman, 2002; Moseley, 1974).
The harsh conditions of the north coast made irrigation a necessity for agricultural production; at the same time, the technology set new terms for system equilibrium. From this perspective, the sociopolitical organizations of the prehispanic past depended on both the functioning of a river-based canal system and on remaining in balance with the natural limitations of the environment. One factor that threatened these precarious dynamics was the natural hazard known as El Niño events.
El Niño Southern Oscillation (ENSO) describes the see-sawing of both oceanic and atmospheric conditions across the Pacific (from Indonesia to the western coast of South America). These can be combined in three ways: La Niña (strong easterly winds and cooler sea surface temperatures), Neutral (average sea surface temperatures and uncoupled atmospheric patterns), and El Niño phases. During the El Niño phase of the ENSO phenomenon, sea surface temperatures (SSTs) rise and easterly winds (blowing east to west) weaken and occasionally reverse. One of the effects of warm SSTs reaching the coasts of Peru and Ecuador, is sudden rainfall on the otherwise arid region, which can trigger flash floods – also known as El Niño events.
The role of El Niño flood events in the socio-political history of the north coastal plain of Peru has been viewed from the perspective of hazard research (see Bawden and Reycraft, 2000), an approach born out of US policy efforts in first half of the 20th century to re-evaluate the role of government in responding to flood events (Burton et al., 1978; Macdonald et al., 2012; White, 1945). When applied through the lens of anthropology, the approach seeks to identify the vulnerabilities of a population or society – technological, political, social, environmental, or economic – that can contribute to the preconditions for disaster (Hewitt, 1983; Oliver-Smith, 1996; Oliver-Smith and Hoffman, 1999). Scholars have argued that a profound vulnerability of ancient north coast societies was precisely the technology at the center of system equilibrium: canal irrigation.
Headgates, canals, turnouts, sluice gates, and other river-based irrigation components of prehispanic agriculture can be easily damaged or destroyed by the sudden input of floodwaters (van Buren, 2001; Craig and Shimada, 1986; Moseley, 1987; Moseley and Keefer, 2008; Nials et al., 1979). From the “irrigation-civilization” perspective, flood damage on a massive scale, could throw any socio-technological symbiosis that existed into chaos, potentially triggering sociopolitical and economic shifts (Steward, 1955). The hazard research perspective combined with the equilibrium concept is a very attractive explanatory model: if an El Niño event causes major damage to the canal system, then total-system shock and even at times, collapse, can be expected. However, detecting direct evidence of that link in the archeological record has proven challenging.
Billman and Huckleberry (2008) demonstrated that part of this challenge derives from the use of long-term environmental records as proxies for El Niño flood events (102). While ice and lacustrine cores do capture wet and dry periods that include signals from La Niña, Neutral, and El Niño phases of the ENSO phenomenon, wet periods do not necessarily correspond to a flood event on the coast (Schneider et al., 2018; Thompson et al., 1984, 1985). In fact, the precise location, timing, and impact of El Niño floods are difficult to predict with any precision (Rodríguez-Morata et al., 2019). As Billman and Huckleberry (2008) have shown, within the north coast region or even within a single valley, the impact of El Niño floods can vary depending on where rainfall occurs, and the size of the catchment basin (Billman and Huckleberry, 2008; Waylen and Caviedes, 1986). When local El Niño event records are available, tying a particular flood event to material evidence of irrigation damage or its corollary in sociopolitical change is also challenging. In response, and following a hazard research approach, some scholars have centered their investigations on El Niño events that appear to intersect with other sociopolitical and environmental stressors. For example, Moseley (1983, 1999) and others (Craig and Shimada, 1986; Keefer et al., 2003; Satterlee et al., 2000; Shimada, 1994) have argued that extreme prolonged drought, tectonic uplift, and earthquakes around AD600 and AD1100 coincided with El Niño flooding to cause a rupture in both the Moche (AD200–900) and Chimu (AD900–1460) sociopolitical systems (see also Fernandini Parodi, 2018; Mauricio, 2018; Uceda et al., 2021). Importantly, for these authors, the resulting damage to complex irrigation technology was a crucial tipping point in this process (van Buren, 2001).
When scholars make reference to society-changing floods, they are typically referring to somewhat unusual, large-magnitude or “canonical” events. Based on the analysis of flood deposits from the north coast Moche Valley, Nials et al. (1979) identified two “mega-el Niño” events, dated to 500BC and AD1100, which were estimated to have reached 2–4 times the magnitude of the infamous 1925 El Niño (see also Chang Huayanca, 2014; Pozorski, 1987). Such an event would have impacted settlements and entire irrigation systems, possibly causing system collapse. Similar events have been identified on the south coast and point to large-scale repercussions on ancient agricultural systems (Keefer et al., 2003; Moseley and Keefer, 2008; Satterlee et al., 2000). Other scholars point out that the prolonged secondary effects of such floods, such as disease, food, and water shortages, would have also disrupted society and possibly inspired mass human sacrifice (see Kiracofe and Marr, 2008; Prieto et al., 2014; Zhou et al., 2002). While these rare events likely tested the capacities of prehispanic socio-political organizations, they do not represent the most common form of society-El Niño interaction. In fact, El Niño presents in several different “flavors”(Sandweiss et al., 2020), and historical and local flood records suggest that ancient farmers were encountering weak and moderate events on a much more frequent basis.
Billman and Huckleberry (2008) carried out excavations of an 800-year record of floodwater runoff that had built up behind an aqueduct in the north coast Chicama Valley. The record presented evidence of 33 runoff events between AD1175 and 1998. The authors compared this flood record to historical references to floods and concluded that El Niño-related rainfall events, which do not always result in runoff or flooding, can occur as frequently as every 3–4 years (110) (Garcia-Herrera et al., 2008; Waylen and Caviedes, 1986). In other words, El Niño-related weather creates inputs of varying scale and degree on a regular basis in this environment, a revelation that complicates the idea of a fragile ecosystem constantly on the brink of collapse.
The Pampa de Mocan, located in the Chicama Valley, is a preserved farming landscape dating from 1100BC to AD1460 that presents an opportunity to observe the dynamics of arid-zone farming and El Niño events in situ and over time. The ancient irrigation works and farming strategies observed in the Pampa de Mocan demonstrate that managing the impact of flood events was an important focus of the overall farming strategy, however, evidence also points to a far more complex relationship between agriculture and ENSO. Here, both agricultural activities and El Niño effects drew upon and changed the local environment. All three systems – the environment, ENSO, and agriculture – interacted as components of the same agroecosystem.
The environment of coastal Peru
The Andean rain-shadow, the Humboldt current, and the South Pacific anticyclone (southern oscillation) reached their present states by at least the Pleistocene, establishing the hyperarid conditions of coastal of Peru by 8000–6000 BP (Sandweiss, 2003; Sandweiss et al., 1999; Thompson et al., 1995; Wells, 1988). In normal years, mean annual precipitation along the coast ranges from just 5 to 40 mm (Wells and Noller, 1999), with some localities, such as the Pampa de Mocan, receiving an average of 12 mm per year.
The north coast, between 3 and 12°S, consists of 13 entrenched river valleys formed out of the Paleozoic to Cretaceous western Andes (Gilboa, 1971; Hosmer, 1959; Figure 1). Despite being classified as an arid desert, the region is varied in terms of soil types, climate, seasonal water availability, topography, and vegetation communities, even within a single valley. In the Chicama Valley, for example, significant portions of the valley are covered in desert lithosols (rock and gravelly soils of igneous and sedimentary origin) or desertic regosols (soils covering large plains composed of unconsolidated Quaternary marine sediments and eolian deposits); the vast majority of agricultural activity is found on and around alluvial soils (Gilboa, 1969). The valleys themselves are often bounded by igneous intrusions of granite, granodiorite and diorite – coastal cerros or hills. Each valley consists of alluvial fans, piedmont plains and east-west transverse rivers. In the inner valley, along the river floodplain, typical riverine vegetation is present. South of 8°S, lomas plant communities form on the flanks of cerros in winter months; in the northern latitudes, Prosopis forests and mangroves (pantanos) are part of the local ecology. Finally, humedales or wetlands are found all along the coast.

The north coast of Peru, the Pampa de Paijan and the Pampa de Mocan indicated north of the Chicama Valley. Google Earth Pro (December 13, 2015). 17M 692549mE; 9152011mS Landsat/Copernicus.
Just outside of the floodplain, surrounding and separating each river valleys are desert sedimentary plains or plateaus, known as pampas. While today these landscapes are typically devoid of agricultural activity, ancient canals, fields, and prepared surfaces have been recorded especially along the edges of the modern irrigated floodplain (Caramanica et al., 2020; Eling, 1987; Hayashida, 2006; Huckleberry et al., 2018; Kus, 1972; Watson, 1979). Pampas are distinct from river valleys both in geological form and structure (Hosmer, 1959). On the north coast, pampas are located near the fall zone of the western Andes, and the resulting plains are inclined in the direction of the Pacific Ocean. Piedmont sediments bury the rocky flanks, and are typically made up of hard, crystalline rocks and covered in eolic deposits, shales, siltstones, and feldspathic sandstone (Hosmer, 1959: 90). The pampas surrounding the Chicama Valley, including the Pampa de Paiján and the Pampa de Mocan, share this characterization.
The north coast pampas are areas of both new and very old depositional formations, adding to their dynamism: the sediments here are undergoing near constant chemical and physical breakdown, through both gravitational and transport erosion. Gilboa (1969, 1971: 78) describes the northwestern corner of the Chicama as covered in cemented gravels typical of an old delta. Meanwhile, further inland, near the Pampa de Mocan, eolic deposits are covered in gyspic ryzoliths (crystallized and hardened rootlets) and active dunes encroach on Pleistocene-era ridge and swale topography. As a result, El Niño flooding on these landscapes can have varied and spectacular effects, both above and below the surface.
Interactions between El Niño and the local environment
Strong El Niño events have occurred on the Peruvian coast as frequently as every 6–7 years and as infrequently as every 14–20 years (Garcia-Herrera et al., 2008; Ortlieb, 2000; Quinn et al., 1987; Sandweiss et al., 2020), but this pattern has changed through time. By analyzing marine fauna, with a particular focus on molluscan remains, Rollins et al. (1986), Sandweiss (2003), and Sandweiss et al. (1996, 1999, 2001, 2004) have demonstrated that waters along the Peruvian coast cooled around 5800 cal BP reducing the frequency El Niño years between 5800 and 2900 cal BP (see also Carré et al., 2014: 10450; Huckleberry and Billman, 2003; Thompson et al., 1995). Toward the end of that period, El Niño years occurred in much more rapid succession, beginning to reach modern-day patterns by 2900 cal BP (Sandweiss et al., 2020). It is therefore likely, that early agriculturalists on the north coast, including Cupisnique (1100–500BC) irrigation farmers, were reckoning with a changing ENSO regime, and this would have had implications both for the development of irrigation practices and the floodplain itself.
El Niño floodwaters play a central role in floodplain development on the north coast. Manners et al.’s (2007) work in the middle and upper Mocquegua River Valley demonstrates that river channels widened by 30% after an El Niño event in 1946, causing the loss of 19 ha of land and increased entrenchment of the main river channel (235). Manners et al.’s (2007) study is consistent with the response of rivers to discharge surges in other arid environments, where channels are recorded moving up to 3.2 km laterally and up to 6 m in depth through incision after major flood events (Graf, 1983; Huckleberry et al., 2012; Manners et al., 2007). Consequently, terraces of varying height and age surround an increasingly entrenched river (see also Hudson, 2004). However, in the years after the Mocquegua 1946 event, sediments were deposited along the new channel terraces, resulting in the slow recovery of land. The stratigraphy of older terraces along the Mocquegua river show near-continuous deposition of alluvium at a rate of 3 mm/year (245 cm since AD1230) (Magilligan et al., 2008). Consequently, Manners et al. (2007: 243) conclude that 80% of the Mocquegua Valley is younger than 550 14C years.
Flooding along the river, causes channels to widen, leading to the immediate loss of arable land near the narrow valley neck; however, floodwaters transport that sediment and deposit it onto fan terraces and aggrading river terraces. In other words, rather than being “lost” outright, sediment is mobilized, remaining in the system. Large-scale sediment transport occurs in the immediate aftermath, but, in the longer-term, vegetation growth made possible by the influx of water, works to slow down erosion (Tote et al., 2011). This sediment transport cycle catalyzed by El Niño flooding creates inputs for this landscape, including the pampas surrounding the floodplain.
El Niño and local hydrology
Both the western Andean flanks and the coastal mountain range are made up of near-impermeable igneous rock. In those areas where either the coastal mountain range or near-shore coastal uplift slows groundwater flow to the Pacific, subterranean aquifers have formed close to the surface (5–20 m). In the Chicama Valley, Gilboa (1969) identified three important aquifers: a large aquifer in the area of the Cartavio hacienda in the lower valley, a “saline swamp” just to the west of Paiján, and a projected aquifer in the area of the Pampa de Mocan – the piedmont between the western flanks and the coastal mountain range. Today these aquifers make up a significant portion of the Chicama Valley water supply. In the 1960s, Gilboa (1969: 79) reported over 550 wells in the area, which were being exploited at a rate of 175 mcm/year. Over 80% of the Chicama wells are less than 30 m deep (Gilboa, 1969: 79). Together, these data point to an expansive and important aquifer supply in the Chicama Valley, making the question of recharge particularly salient.
Gilboa (1969) argues that the majority of aquifer recharge occurs through river and irrigation seepage. Other scholars suggest that recharge can occur through flooding. Taltasse (1973) suggests that significant differences between river and groundwater temperature and chemistry point to the likelihood of groundwater recharge through geological faults (Taltasse, 1973). Magilligan et al. (2008: 26–27) go a step further and use isotopic composition from groundwater samples to argue that groundwater supplies are ultimately controlled by ENSO patterning. The authors hypothesize that ancient springs were recharged after El Niño flood events in the prehispanic past, making El Niño crucial to the maintenance of groundwater levels. In the Mocquegua Valley, Tiwanaku canals were located near such springs, which were then tapped to carry water to marginal areas on the edges of the floodplain (Magilligan et al., 2008). These Tiwanaku canals date to around AD730 – coincident with a significant El Niño flood event (Magilligan et al., 2008: 28). El Niño serves a vital role in aquifer recharge, and therefore, would have allowed for agricultural expansion beyond the valley-bottom floodplain in the prehispanic past.
Ethnographic accounts speak to the possible use of aquifer recharge for farming purposes on the north coast. In their survey of the Chicama Valley sometime after the 1997/1998 El Niño event, Gálvez Mora and Runcio (2011) interviewed a local farmer, Sr. Julio Flores. Sr. Flores reported using water from an El Niño-recharged spring in the Quebrada Santa Maria, which is one of the quebradas that empties into the Pampa de Mocan system, for 8 years after the flooding event. He constructed a 330 m long canal to carry water to a small reservoir from which he irrigated 3 ha of fields. With two harvests each year, Sr. Flores produced 10,000 kg of maize per hectare per year – approximately doubling the yield of his typical valley-floor harvests (Gálvez Mora and Runcio, 2011: 90–91). While this is an anecdotal example, Julio Flores’ account does provide a point of reference for the scale of production that could be extracted from springs recharged by El Niño events. Gálvez Mora and Runcio (2011: 83) went on to identify five springs in the Quebrada Santa Maria area alone.
El Niño and local biodiversity
Meanwhile, above the surface and beyond the floodplain, the desert pampas are directly affected by the mobilized sediment load. Two variables contribute to this process: (1) sediment transport and (2) plant growth (Tote et al., 2011: 1777). Londoño et al.’s (2012) study of the Ilo dune fields in southern Peru demonstrated that periods of influx of eolian deposition are preceded by “hydrologic excess” in nearby floodplains (Londono et al., 2012). In other words, El Niño floodwaters carry sediments from the eroded land surfaces of the upper and middle river valley to the river delta and coastal beach ridges (Wells, 1988: 176). There, trade winds pick up the loose sediments and transport them to the pampas where they add to the active dune fields and desert pans – also known as “peri-desert loess” or more commonly, “soil dust” (Goudie and Middleton, 2006; Noller, 1993; Pye and Tsoar, 2009). The influx of dust in north coast deserts enriches the soil content, adding higher percentages of both silt and clay than expected for hyperarid sediments (Noller, 1993: 133–157; see also Mächtle and Eitel, 2013). Ultimately, El Niño events contribute to the topography and formation of these desert or pampa areas, and crucially to the enrichment of sediments – adding loess inclusions to the otherwise largely gravel, sand, and sandstone piedmont soils. When combined with the influx of El Niño-related water, these soils support blooming events that have far-reaching consequences for the landscape and the ecosystem.
The arid and semi-arid deserts or pampas undergo spectacular transformations in the aftermath of El Niño flooding. During the 1982/1983 and 1997/1998 El Niño events on the north coast, the annual streamflow was 5.4 times higher than average and rainfall was up to 25 times the norm (Holmgren et al., 2006). This influx of water into desert environments resulted in explosive blooming events with staggered appearances of first, herbaceous ground cover, followed by shrubs, and finally trees (Richter and Ise, 2005). The typical (non-El Niño) vegetative community in Peruvian deserts was recorded in the Sechura Desert in the far north of Peru and includes permanent and sporadic members making up a maximum of 25% plant cover: Galvezia limensis, Encelia canescens, Alternanthera halimifolia, Maytenus octogona, Haageocereus pacalaensis, Prosopis pallida, and Capparis scabrida (Richter and Ise, 2005: 143–144). However, during the El Niño rainfall in March 1997/1998, plant cover changed to include largely herbs, and grasses increased 100% ± 20% in just 3 months (Richter and Ise, 2005: 143–144). Vining et al. (2022) recorded similar effects in the pampas surrounding the Chicama Valley after the 2017 El Niño Costero event.
Arid ecosystems are capable of responding quickly to precipitation input due to seed banks that have built up over time – Holmgren et al. (2001, 2006) argue that annual species survive the long interims between flood events by producing a large number of seeds with long dormant periods (Holmgren et al., 2001). However, the boom in propagation of grasses quickly declines: for both the 1998 and the 2017 El Niño, by October almost all of the fast-growing grasses had perished (Richter and Ise, 2005; Vining et al., 2022: 144). In contrast, tree and shrub growth can persist for much longer and therefore continue to affect the water and sediment supply. Richter and Ise (2005: 144) report an increase in tree cover in the Sechura Desert from 1.75% in 1997 to 7.5% in 2004. Woodland restoration impacts both the drainage system and environment, affecting vegetation cover and the growth of companion species, erosive processes, and groundwater levels, which in turn have significant consequences for agriculture (see also Beresford-Jones et al., 2009).
The north coast desert environment, far from being static, or fragile, undergoes near constant change and transformation of surfaces, soils, vegetation, and water sources, thanks, in part, to El Niño dynamics. In the Pampa de Mocan, these integrated systems converged with ancient agricultural practices, which in turn impacted the local environment and availability of resources. The Pampa de Mocan reflects the result of a package of interacting inputs – anthropogenic and natural – which, over the long-term, evolved to be both productive and resilient in face of extreme aridity and flood events.
Resilient technologies and robust landscapes
The main and permanent water source of the Chicama Valley is the Chicama River. The Pampa de Mocan is located approximately 30 km north of the River, on the southern edge of the Paijan Desert, and itself has no local perennial or active water sources (Figure 1). The Pampa de Mocan, a general term for the areas of the Pampas de Huatunero and El Inca, and the Playa Mocan, totals approximately 5800 ha, however, the present study centers on a 1707 ha polygon in the northeastern extreme of the Valley margin (Figure 2).

Canal system of the Pampa de Mocan as it related to the Chicama River and Ascope Canal System. The Pampa de Mocan is located at approximately: 17M 698449mE; 9159849mS.
Although the Pampa de Paiján (just north of the Pampa de Mocan) was a site of early Holocene occupation (Chauchat et al., 1998), the desert traditionally marks a long-standing boundary between northern and southern political spheres for prehispanic, irrigation-based agricultural societies such as the Moche (AD200–900) and Chimu (AD900–1460) (Castillo Butters and Donnan, 1994; Chauchat et al., 1998; Hoyle, 2001 [1934]; Koons and Alex, 2014; Sandweiss et al., 2001; Wells and Noller, 1999). However, archeologists have recorded preserved canals and fields reaching north from Chicama and south from the neighboring Jequetepeque Valley into the Paijan Desert and nearby pampas.
Using historic aerial photographs, Watson (1979) identified seven large intake canals that terminated in Mocan. According to Watson (1979) the need to provision a growing conquest population drove a Chimu state (AD1100–1460) push to bring the Pampa de Mocan under cultivation. However, Watson and others (Pozorski, 1987) hypothesized that this effort ultimately failed, pointing to the apparently marginal conditions. The Pampa was permanently abandoned shortly before Spanish conquest, leaving a near-pristine record of a prehispanic agricultural landscape, complete with trunk, intake, feeder and drainage canals, turnouts, wells, and field systems.
The first systematic fieldwork on the farming landscape began in 2012 with the Proyecto Arqueo-ambiental del Valley Chicama (PAAVC) and continued through 2017 with the Proyecto Arqueo-ambiental de la Pampa de Mocan (PAAPM). These projects were aimed at collecting the archeological record of irrigation systems adapted to marginal landscapes. Significant among the findings, the evidence pushed the date of agricultural activity back to Cupisnique occupation by 1100BC, much earlier than the previously hypothesized Chimu state activity (AD900–1460). A major contributing factor to the Pampa de Mocan’s viability was El Niño, the resulting floodwaters, and the consequent dynamics that played out in the local environment. Here, landscape technologies such as canal systems, the local social-ecological-system, and regional-scale climate dynamics (ENSO) can be observed in situ as they interact over time as part of one agroecosystem (Gallopín, 1995; Power, 2013; Wezel et al., 2009).
Agricultural building blocks: Soil and water
Today, soils in Pampa de Mocan are classified as regosols, or unconsolidated materials, originating from relatively recent erosional or alluvial formations, and lacking horizons – undeveloped and low in organic matter (see Watson, 1979: 208; ONERN, 1973). However, studies show that these soil conditions can be partly mitigated via the deposition of fine sediments through the application of irrigation water (Sandor et al., 2022; Strawhacker, 2013). For example, natural quebrada soils in the Pampa de Mocan are gravelly and highly permeable, allowing for good drainage. Repeated application of silt suspended in irrigation water would increase the silt and clay content of these soils, resulting in increased water retention capacity and chemical reactivity, which can enhance fertility (see Hesse and Baade, 2009; Sandor et al., 2022).
Salinization is one of the most acute threats to irrigation-dependent agriculture in arid environments. In soils that are poorly-drained, salts dissolved in irrigation water evaporate through capillaries toward the surface. Salinization is devastating to plant growth and has been attributed to agrarian collapse in the Nile Delta, Tigris-Euphrates, and even Huang-Ho Valleys across time (Artzy and Hillel, 1988; Harris, 1960; Jacobsen and Adams, 1958). Moreover, salinization can occur in different areas within a single valley system. In coastal Peru, in the La Leche Valley, Nordt et al. (2004) reported that over 50% of soils downstream of the river suffered high percentages of salinity, while those located in the mid- and upper valleys, with loose gravel content, enjoyed better drainage qualities.
While some evidence of salt accumulation, such as shattered clasts and stone-splitting (salt-wedging), exists even in the mid- and upper valleys (including in the Pampa de Mocan), the salinity of desert soils on the north coast is generally low, even when compared to the south coast (Noller, 1993: 201). This is likely due to the frequency of El Niño events throughout the Quaternary, which may have helped to flush out salts in this region (Noller, 1993: 158). Soil pH is closely related to salt content and, consequently, north coast desert soil pH tends to be mild, ranging from 5.2 to 8.8 with a mean of 7.7 (Noller, 1993: 168).
Another fundamental element for irrigation agriculture is access to a reliable water source. In desert environments, surface water is scarce, or flow can be highly seasonal. While traditional canal systems utilize the river as the main source of water, supply in the arid farmed landscapes of the north coast, in particular, the Pampa de Mocan, was supplemented through a variety of surface-flooding and subsurface-flooding technologies. Surface flooding systems can involve irrigation through channeling river flow, capturing runoff or floodwaters, or “harvesting” water from dammed reservoirs. Once water reaches a prepared field, the system can be further broken down into either border-strip flooding, at times called low terraces, check-flooding, also known on the coast as “posas” (Hatch, 1974), and furrow irrigation. Finally, raised fields are typically found in areas of standing water, seasonal inundation, or high-water table.
Subsurface irrigation has also been identified across the coast, most notably at the site of Chan Chan and at the mouth of the Chilca River, where sunken gardens, also known as mahamaes or wachaques have been studied (Engel, 1966; Knapp, 1982; Moseley and Feldman, 1984; Smith, 1979). Such fields were located in areas of “backmarsh” where the water table could be reached through excavation. No sunken fields were identified in the Pampa de Mocan, although drainage wells, or kochas (Sabogal-Wiesse, 1974; Schreiber and Rojas, 1995), were identified in the lower bajada of the Pampa. These wells were located at points along the slope where drained runoff from upslope fields or canals would naturally seep to the near-surface, indicating the use of the water-table.
Farming in a desert: Technological diversification adapted to arid-land and ENSO dynamism
The irrigation and farming infrastructure of the Pampa de Mocan landscape reflects both the dynamism of the local climate and hydrological conditions, but also deep knowledge of soils and soil erosion. Agricultural activity in the area dates to as early as 1100BC, with Cupisnique sites and fields, and as late as AD1460, with a high density of Chimu and Lambayeque evidence. Eight trunk canals (A-H) and their branches terminate in the 1707 ha study area (Figure 2). While the point of origin of several of these canals has been obscured over time, trunk canals A-C appear to have connected to the river-based Ascope Canal System, best known for the iconic Ascope Aqueduct, at some point during their use-lives. Each of these trunk canals was modified significantly and had multiple use-lives. Due to the ad hoc nature of use of these channels, they were deemed unreliable proxies for establishing a chronology; instead, chronology is tied to ceramic sherd scatter on related field systems rather than on canals themselves. No attempt to date the trunk canals through OSL or radiocarbon was carried out by these projects.
The Pampa de Mocan also presents with remarkably well-preserved fields. Several field types identified by the PAAVC and PAAPM had been previously recorded in the southern edge of the Chicama Valley near the Quebrada del Oso (Kus, 1972). Kus (1972) compiled a detailed record of ancient furrow types, identifying six dominant furrow patterns. He argued that the classic serpentine-shape furrow, iconic of prehispanic coastal agriculture (Kosok, 1965: 107), is closely related to field slope, occurring when slope is 3.1% or higher. The serpentine furrows were likely constructed to slow down water flow and prevent erosion. The six furrow types recorded originally by Kus were also identified in the Pampa de Mocan fields; meanwhile, the PAAPM recorded an additional five types, which were interpreted as closely adapted to environmental and climatic extremes, including both flooding and drought (see Caramanica et al., 2020; Figure 3). These types included raised fields, embankment fields, border-strip fields, check-flood fields, and lithic mulch fields.

Spatial distribution of identified field types in the Pampa de Mocan.
Raised fields
Raised fields are constructed globally to reclaim water-logged areas, low-lying marshlands or swamps, and lake or coastal margins. In the Andean region, raised fields, or campos de camellones (Valdez, 2006), are best known in the highland region of Mojos de Llanos, Bolivia (Whitney et al., 2014), Lake Titicaca Basin (Erickson, 1988, 1993; Kolata, 1993; Stanish, 1994), and a rare coastal example in the Casma Valley (Moore, 1988; Pozorski et al., 1983; Zak, 1984). Just beyond the core Andean region, spectacular examples of fields have been identified in the seasonally flooded Sabana de Bogotá, Colombia (Parsons, 1982; Plazas et al., 1993; Rodríguez Gallo, 2019). However, in an arid landscape, such as the Pampa de Mocan, these types of fields would likely be constructed within an embankment or reservoir built to capture floodwater. Examples of raised fields with associated Cupisnique material (1100–500BC) in the Pampa de Mocan were largely located in floodouts or on the banks of ancient, ephemeral streambeds.
Embankment fields
Embankment fields are prepared, leveled areas, typically bordered by a low mound or wall, which allows runoff flow and debris to pool within the field (essentially, flooding a bordered field). In the Pampa de Mocan, embankment fields captured El Niño floodwater flow and allowed for the controlled distribution of floodwater from the embankment (Caramanica et al., 2020). In several examples, rockpiles were recorded within the embankment fields. Rockpiles have been analyzed in other arid-farming systems in the Sonoran and Negev deserts, where they served as sediment traps (Evenari et al., 1982; Fish, 2000; Fish et al., 1985; Lahav (Lavian) and Steinberger, 2001). Floodwater would have been directed to these rockpiles, which then acted like a net: organic debris and detritus carried by high-energy flow and rich in nutrients could be collected and used to enhance desert soils.
Border-strip irrigation
Watson (1979) described this type of field system as long, wide strips of land bordered by low walls arranged parallel to the direction of slope. These fields are common in areas of low slope (0.1%-1.0%) (Watson, 1979: 143). Border-strip irrigated fields in the Pampa de Mocan were typically located around embankments, indicating that these were fed by trapped, controlled floodwater, likely resulting from El Niño events (Caramanica et al., 2020). These fields helped to mitigate the effects of aeolian soil erosion during arid periods by adding sediment and organic debris to the surface of the fields.
Check-flooding or posas
In these systems, fields are divided into small, cell-units (posas) with mounded borders (bordos). There is only one irrigation ditch at the top of the field and one “posa” or unit is inundated at a time. Hatch (1974) conducted an ethnographic study of modern farmers in the town of Motupe, Lambayeque Valley; there, these fields were used to when the number of water users (usuarios) just met or exceeded the water supply. During dry seasons or when water systems were being pressured upstream, water users would soak each unit for several hours, and this soaking would be sufficient until the next watering cycle, sometimes more than 30 days later. This field system, and possibly similar practices, was well-adapted to conditions of water-scarcity in the Pampa de Mocan.
Lithic mulch fields
Lithic mulch agriculture (LMA) has been described by Lightfoot (1996) as a strategy for arid-land agriculture to increase moisture in a given field while reducing the possibility of wind-erosion. Lithic mulching involves the application of pebbles or small rock fragments to the field and thus leaves a distinct surface signature. According to Lightfoot, LMA is known historically in the Negev, the Mediterranean, Atacama, northwest Argentina, New Mexico, New Zealand, the Canary Islands, and China. Typically, a borrow pit, a pit dug for the extraction of desired rock material, is found in close proximity to the field. The examples of lithic mulching in the Pampa de Mocan were likely particularly important during dry periods, as the surfaces of formerly-flooded or irrigated fields became vulnerable to high winds and erosion (Figure 4).

Example of preparation of lithic mulch fields recorded in the Pampa de Mocan.
The variety of landforms and surfaces across the Pampa de Mocan witnessed multiple geomorphic processes, including fluvial activity and sediment deposition related to El Niño events, and transformations triggered by existing landscape technologies, which also changed how water flowed downslope, seeped into the water table, and encouraged plant growth. The survey and excavation projects studied over 235 ha of ancient fields in detail and projected that approximately 746 ha of the area within the research polygon was under cultivation at some point during this landscape’s history. While the fields have been relatively dated, using ceramic sherd cover as a proxy for cultural periods, precise dating has proven challenging. Future work will attempt to determine if patterns can be observed in floodwater versus river-based irrigation over time, however, it is clear that El Niño was a constant resource throughout the Pampa de Mocan’s occupation. The myriad field systems are scattered across the alluvial platforms of Pampa de Mocan and several dip under the sand dunes encroaching from the northwest into the Pampa de Paiján. Many of them are related to the trunk canals (A-H) described above, however others are apparently vestiges of now-extinct water sources. Ancient farmers utilized different surfaces, water sources, and technologies to transform and take advantage of the ever-transforming elements of this environment.
This evidence suggests that prehispanic irrigation farmers used a variety of practices to mitigate the risks of pampa-agriculture, while taking advantage of some of the opportunities created due to El Niño dynamics: well-drained soils, local biodiversity, and occasional inputs of water and nutritious debris. Beginning with the Cupisnique, farmers began combining river-based and El Niño-farming, and the practice continued through the Moche (AD200–900) and Chimu (AD900–1460) occupations. The density of Moche and Chimu material in the Pampa de Mocan is greater than that pertaining to the Cupisnique; due to the iterative feedback between the three systems – the natural environment, El Niño, and agriculture – the landscape itself became more resilient and, consequently, more attractive to farmers over time. Together, mitigating and adaptive practices contributed to the overall resilience of the Pampa de Mocan.
Conclusions
Discussion of El Niño and hyper-aridity is often centered on risk, catastrophism, or collapse. Warm El Niño currents lead to mass deaths in fisheries, and floodwaters cause erosion, breach irrigation infrastructure, and perhaps most destructive for local smallholders, flood agricultural fields. Secondary effects include diseases, rodents, insects, and food and water shortages. However, ENSO has “positive” effects that add to local biodiversity and create new water sources. Erosion and sediment transport are integral to the formation of soils in the floodplain and flushing out salts; as a result, the sediment load that eventually settles in the sandy pampas around the valleys is rich in loess and clays. El Niño floodwaters likely recharged subterranean aquifers, activating local springs. Settlement patterns suggest that ancient inhabitants understood the dynamics of the floodplain, preferring older, more stable surfaces, such as alluvial fans, old dunes, coastal deposits, or the ancient floodplain (Wells, 1987). Perhaps most spectacularly, El Niño germinates an entire ecosystem in a matter of months in the desert: herbaceous plants emerge first from extensive seed banks, followed by shrubs and woody species. Woodlands can be particularly resilient once established and lead to long-term changes to the water regime. Moreover, the ethnographic record makes clear that local farmers have developed innovative responses to redirect excess flow from El Niño floods (Gálvez Mora and Runcio, 2011). These range from the construction of diversion canals and check dams, to floodwater fields. While the co-evolution of the hyper-arid environment and El Niño resulted in a robust natural system, north coast smallholders incorporated risk management, and risk opportunism in their farming technologies, further enhancing the productivity of this landscape.
The Pampa de Mocan record suggests that the protracted feedback between these systems, built resilience and elasticity into this desert landscape. The dynamics between El Niño, the natural environment, and agricultural systems – both the collective actions of smallholders and possibly the directed practices of states – resulted in near-constant change and a convergent agroecosystem. This does not preclude the possibility of eventual collapse of one or several of these systems, however, the perceived precarity of arid and marginal environments is misplaced. Instead, this study shows the unfolding of these interactions contributed societal resilience in the face of natural disasters.
Footnotes
Acknowledgements
I thank the many individuals and institutions that provided support for this research, including the National Science Foundation, the Graduate School of Arts and Sciences at Harvard University, and the Anthropology Department of Harvard University, Dumbarton Oaks Research Library and Collection, the Universidad Peruana Cayetano Heredia, the University of Arizona, the Pontificia Universidad Católica del Perú, and the Universidad del Pacífico. Special thanks to John Yellen and Anna Kerttula de Echave, Jeffrey Quilter, Luis Jaime Castillo B., Gary Huckleberry, Carlos Wester, Jorge Wester, Luis Alberto Sanchez Saavedra, Solsire Cusicanqui Marsano, Ana Tavera Carito Medina, Enrique Estrada Mariluz, Roxana Tornero, Fiorella Villanueva Rojas, Geraldine Borja, Gabriel Prieto, Noa Corcoran-Tadd, Michele Koons, Yadira Rivera, Marianne Fritz, and Linda Ordogh.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Science Foundation, Long Term Human Ecodynamics in Coastal Peru: A Case Study of Polar-Tropical Teleconnections, Award 1152156, 2011-2016.
