> >

Instructors: Prof. Ofer Bar-Yosef, Dr. Richard H. Meadow


The Phenomenon of El Niño (ENSO)
Its Impact on the Transition to Agriculture in Coastal Peru

Alexandre Tokovinine
Department of Anthropology
Harvard University
January 2004



The coast of Peru offers one of the most fascinating cases of the transition to agriculture. Being clearly a result of adoption rather than invention, this process took a considerable span of time. On the coast, a few cultivated plants make their first appearance at the sites of Paloma (7700-4800 BP) and Chilca 1 (5600-5000 BP) during the so-called Middle Preceramic period commonly dated to about 7950-4450 BP (Piperno and Pearsall 1998: 271-272).1 However, agriculture becomes the primary form of subsistence only at the end of the Late Preceramic period also known as Cotton Preceramic (4450-3750 BP). During this long transitional phase, the inhabitants of the coast develop successful local adaptations based on the intensive exploitation of marine resources and some cultivars. A shift to agriculture-based subsistence schemes during the emergence of complex societies in the Initial Period (3750-2850 BP).

The goal of this paper is to estimate to what extent the beginnings of agriculture in the coastal Peru were determined by a specific environmental setting. I am especially interested in the El Niño phenomenon and its possible impact on human choices of subsistence strategies. The idea is not particularly new. Osborn (1977) was the first to look for the role of El Niño in the emergence of complex societies on the Peruvian coast. Recently, Piperno and Pearsall (1998: 267-272, 279-280) stressed the importance of the chronological relationship between the onset of El Niño and the occurrence of the first cultivars at the coastal sites of Paloma and Chilca 1. However, in my opinion, these suggestions do not amount to well-fleshed models. Of course, I will review some frequently cited and comprehensive theoretical schemes for the Peruvian case. Finally, I will consider the data on the key Preceramic sites available in the literature and evaluate the possible impact of El Niño against other plausible alternatives in explaining the transition to agriculture the alternatives which are probably not mutually exclusive.


The Peruvian coast: a setting

The Peruvian coast comprises three distinct ecological zones deserts, coastal valleys, and lomas (Parsons 1970: 295-296). The Peruvian desert is an almost lunar landscape that extends from 4° south of the equator into the northern Chile and some 15-20 miles inland (Pickersgill and Smith 1981: 89). It is mostly devoid of vegetation except the areas at elevations between 250 and 800 meters where heavy fogs (garruas) support a specific vegetation cover of herbs and shrubs known as lomas. These fogs and the associated lomas growth are seasonal restricted to the period between May/June and October (Pickersgill and Smith 1981: 92). Besides, the amount of moisture in the fog banks varies from year-to-year resulting in different extent of the lomas (Parsons 1970: 260). Although the lomas may provide a rich source of edible roots, seeds, tubers, as well as grazing animals, their productivity is unpredictable and their distribution is discontinuous (Cohen 1977: 151; Pickersgill and Smith 1981: 92). Both factors make the lomas an unlikely candidate for a reliable source of food.

Some 40 river valleys cut across the desert plain. They vary considerably in the annual amount of water flow and in the size of the alluvium (Parsons 1970: 295; Pickersgill and Smith 1981: 93; Wilson 1999: 338). The general trend is that the southern valleys are narrower and drier. The valleys with a seasonal flow (December-March) or the temporally flooded alluvium of broader valleys support loma-like bushes and small trees. Bands of woodland grow along the margins of more permanent rivers. There are trees with edible fruits and/or seeds algarrobo (Prosopis spp.), pacae (Inga feuillei), pepper tree (Schinus molle) and plants with edible seeds, fruits or roots like cattail (Typha), totora (Scirpus), and Cyperus. Moreover, the northernmost river valleys like that of the Zana river (Dillehay et al. 1989) are characterized by moister conditions and support a distinct set of ecological zones with the mangrove and estuarine environment in the lower, the deciduous brush and thorn forest in the middle, and the tropical forest in the upper parts of the valley. Both lomas and the river valleys support a variety of animals including deer and camelids that could be hunted by humans. The floodplains of the river valleys provide the setting for the transition to agriculture.

The extreme aridity of the Peruvian coast (except its northernmost part) is caused by the temperature inversion above the cold offshore waters of the Peru-Chile or Humboldt Current that moves northward along the coast from Valparaiso, Chile, to Cabo Blanco, Peru (Parsons 1970: 292-293; Piperno and Pearsall 1998: 81-82; Marchant et al. 1998: 1164; Wilson 1999: 339). This current is characterized by strong upwelling: the prevailing Southern and Southwestern winds drive the coastal waters seaward so that the cold deep waters replace the warmer surface waters near the coast. The deep waters are extremely rich in nutrients and the upwelling conditions support one of the highest concentrations of zooplankton and phytoplankton, which in turn provide the nutrient base for an extremely productive fishing area. Shellfish, anchovies, larger fish, and guano birds are prolific and, without doubt, have been a major food source for the coastal human populations.


El Niño, its impact, and human responses

The Peru-Chile Current usually extends to a few degrees south of the equator before turning west. The boundary between warm and cold waters shifts seasonally: every year around Christmas, a warm countercurrent sweeps along the coast of southern Ecuador bringing a rainy season in January-April (Marchant et al. 1998: 1165). However, at irregular intervals, this warm countercurrent called El Niño extends much farther south and hits Peru disrupting the marine ecosystem and bringing torrential rains on the coast. This is the El Niño event.

El Niño events are eastern Pacific manifestations of a wider phenomenon known as El Niño Southern Oscillation (ENSO). The warming of the coastal surface waters defined as El Niño proper is accompanied by the atmospheric changes in the tropical Pacific referred to as Southern Oscillation. ENSOs mechanism is quite simple. The southeast trade winds create an accumulation of warm waters and a rise in the sea level in the western Pacific. When these winds become weaker, the accumulated water flows eastward. This movement strengthens the North Equatorial Countercurrent that hits the coast bringing the warm waters and deepening the thermocline (Merle 1998: 465; Marchant et al. 1998: 1165). A typical El Niño event starts in January, peaks in April-June, and ends with a weaker warming in November-December (Piperno and Pearsall 1998: 268). Sometimes, as in the case of the 1997-1998 event (Aceituno 1998), El Niño is followed by a period of strong upwelling conditions known as La Nina.

As the thermocline deepens, the nutrient-rich upwelling waters become impoverished: chlorophyll and diatoms are reduced and replaced by tropical and subtropical diatoms and dinoflagellates (Reycraft 2000: 103). It results in a drastic reduction in the amount of cold-current phytoplankton and zooplankton that in turn affects the rest of the food chain. The populations of anchovies, sardines, and guano birds collapse. Fur seals and sea lions die or migrate to the south. Cold-water shellfish populations are also severely damaged, especially if El Niño is accompanied by a rise in the sea level (Diaz and Ortlieb 1993). The cold-water species are replaced by subtropical and ocean species such as tuna, bonito, and dorado.

The strongest El Niño events like those of 1982-1983 and 1997-1998 are accompanied by the rise of the sea level (up to 30-40 cm above normal in1983), the sea surface temperature anomalies (up to 5-10 C), and torrential rains that produce massive mud flows and flash floods (Reycraft 2000: 103; Teves Rivas 1993: 101, 107-110; Aceituno 1998). The problem is that the duration and the accompanying effects of El Niño events are highly variable. For example, the mega-El Niño of 1997-1998 began and ended very abruptly compared to the prolonged development and ending of the 1982-1983 event (Aceituno 1998: 446-447). As Wilson pointed out (1981: 100), El Niño events do not necessarily bring rainfall. According to the results of Rome-Gaspaldys and Ronchails research on the monthly rainfall data for 1960-1990 from 21 stations in Peru (Rome-Gaspaldy and Ronchail 1998: 684), a statistically significant relationship between rains and known normal El Niño events (i.e. except 1982-1983) was observed only at Piura in the northernmost part of the Peruvian coast (i.e. in the area of the strongest impact). The only prominent feature of El Niño is the increase in the sea surface temperatures and its impact on the ecosystem of the Humboldt Current.

It is also worth mentioning that according to historical data and modern observations, the torrential rains sometimes brought by El Niño do not hit the entire coast, but rather create some hotspots of damage surrounded by relatively untouched areas (Reycraft 103-104; Huertas Vallejos 1993: 360-361). One may also expect that the agriculture-based communities would suffer much more extensively from the floods and mudflows than the coastal fishers. Reycrafts archaeological data on the impact and the aftermath of a major El Niño event in the Ilo valley are especially telling. First, some Chiribaya valley sites like Miraflores were simply washed away and buried by the debris flow that could have been 5-6 m high and likely moved at a speed of 110 km/h (Reycraft 2000: 104-106). Second, once the irrigation-based agriculture system had been completely devastated, a new set of small Chiribaya-derived specialized fishing settlements appeared on the seacoast (the Burro Flaco sites), which maintained an exchange network with a new set of Chiribaya-derived valley sites (Reycraft 2000: 106-111, 115-118).

Whereas the events like that of 1982-1983 leave massive and unambiguous evidence of their impact on the marine food chain (billions of fish, millions of birds, thousands of sea lions and fur seas dying and littering the coast, waters black with hydrogen sulfide produced by decomposition etc.), the average damage of a normal El Niño event is much harder to estimate. For example, Wilson (1981: 101) simply takes the average productivity of the open tropical Pacific waters and assumes it to be representative of the El Niño years. This allows him to estimate the ratio between the productivity of the current during the normal and El Niño years as 6:1. However, the situation is complicated by the fact that the intensity of upwelling varies through the year and one has to distinguish the impact of El Niño from these seasonal fluctuations. Moreover, the upwelling waters are not distributed homogeneously but occur as tongues extending from the coast or as patches with warm waters in-between (Piperno and Pearsall 1998: 82).

One method applied by Nixon and Thomas (2001) is to estimate the fluctuations of the entire upwelling system using satellite imagery of Chlorophyll concentrations in the surface waters of Peru. The resulting picture for the time span of 1997-1999 (El Niño followed by La Niña) is that of high of inter-annual variations in the area of the productive habitat but a much lesser difference in mean annual area between the El Niño and La Nina years up to nine-fold difference between the individual months but only two-fold one between the mean annual values (Nixon and Thomas 2001: 2524). Another approach was suggested by Marchant, Hebbeln, and Wefer (1998). The scholars measured the seasonal changes in the flux of planktic foraminifera during the El Niño and normal conditions (1991/1992 vs. 1993/1994) near Coquimbo, Chile. This research revealed a clear pattern of strong annual fluctuations with changes in the composition of the planktic foraminiferal fauna and a lower-than-expected mean annual difference between the El Niño and normal years (3:4). The authors also cite an earlier observation based on the records of the 1982-1983 event that showed a greater contrast in the mean annual foraminiferal fluxes between the El Niño and normal year (Marchant et al. 1998: 1181), which is similar to the difference in the total area of the upwelling system estimated by Nixon and Thomas.

Therefore, on one hand, Wilsons six-fold estimation seems to be somewhat exaggerated. On the other hand, as Steve LeBlanc pointed to me (personal communication 2003), a certain reduction in the amount of the available food does not indicate that a proportionate amount of those who eat this food would starve, but rather that the entire population would starve. Given the fact, that El Niño affects the base of the wood chain, this initial effect would surely multiply before it hits humans (more than half of the anchovies would starve and die, even larger proportion of predators would starve and die, etc.). The final estimate would depend, as Wilson notes, on the human position in the food chain (Wilson 1999: 354).

One can make an interesting observation comparing the annual pattern of variation in the upwelling conditions suggested by Zuta et al. (cited in Piperno and Pearsall 1998: 82), by Nixon and Thomas (2001: 2525, Fig. 2), and by Marchant et al. (1998: 1177, 1182) with the monthly sea surface temperature fluctuations, which are supposed to reflect the same phenomenon (Quispe Arce 1993: 121-123, Fig. 4a-6a). Zuta et al. associate the strongest upwelling conditions with the months of May and September in the North and June and August in the south. This estimation is similar to the pattern of the sea surface temperatures measured off the north coast of Peru: there is a distinct annual temperature minimum in September October and a temperature maximum around February-March. Same period reappears in the reconstruction suggested by Nixon and Thomas. However, the annual cycle based on the direct measurement of the foraminiferal flux is somewhat different: an upwelling period between August and January with a maximum in January and a stratified period between February and June. I would speculate that this difference arises not of the erroneous measurements, but rather of the fact that the data collected by Marchant et al. reflects a response of the ecosystem to the changes in the environment. Another important observation is that the El Niño events follow the general seasonal pattern in the upwelling conditions characterized by high inter-annual fluctuations even without the impact of El Niño.

Most authors who consider the possible human responses to El Niño event assume that given the unpredictability of this phenomenon, solutions such as storage are unlikely (Moseley 1975: 45-46; Wilson 1981: 100-101; 113). Instead, as Reycraft (2000: 104) suggests, the slow onset speed of major El Niño events would favor short-term adjustments such as re-gearing for catching tropical fish and relocating settlements to safer locations. The coastal population would know about the onset of the El Niño conditions due to some signs of the coming crisis like massive seabird migrations or the presence of huge black clouds against the Andean slopes (Wilson 1981: 100). However, I believe that a distinction should be made in this case between the mega and normal El-Niño events, as well as between their prominent negative impact on the upwelling system and the torrential rains, which are far less common. As for marine resources, as it has been discussed above, the impact of El-Niño events parallels seasonal fluctuations in the availability of fish, which are regular and predictable. Therefore, the solutions to cope with seasonal shortages of marine resources would be applicable to the occurrences of El Niño events. Here, some forms of storage are the most likely candidate.

The pluvial aspect of El Niño is far less regular and it seems unlikely that any long-term solution was in place. Moreover, the torrential rains would destroy all the stored food, which, given the prevailing dry conditions, is usually kept in simple sand pits. Huertas Vallejos cites a 16th century account that describes the disastrous aftermath of the torrential rains in the Lambayeque Valley: se les perdieron todas las comidas y ansi mismo lo que tenian enterrado debajo de la tierra como es su costumbre todo se les perdio (Huertas Vallejos 1993: 379).2 From this point, the mega-El Niño phenomena present a different challenge for the coastal populations. Not only would it put a severe stress on marine resources, the torrential rains would eliminate all supplies. Finally, such El-Niño may be followed by a drought (Huertas Vallejos 1993: 361).

A group under stress may chose to move. Since the worst aftermath of El Niño tends to be restricted to one particular area of the coast, such population movements are not meaningless and there is some historic evidence that they did occur (Huertas Vallejos 1993: 369-370). Nevertheless, migration is always the last resort and, as Wilson notes, implies either conflict with the neighbors or adaptation to some new environmental conditions (Wilson 1981: 113). Alternatively, a group may attempt to diversify its resource base. Indeed, post-El Niño conditions offer some new subsistence opportunities. For example, unusually extensive lomas and desert vegetation may appear after the torrential rains creating a temporal habitat rich in edible plants and grazing animals (Parsons 1970: 301; Ferreyra 1993: 262-263). Moreover, a large area would be temporarily available for cultivation. For example, after the mega-El Niño event in 1925, rich pastures spread as south as Morelos and cotton could be grown in some usually desert areas for the next three years (Parsons 1970: 296). According to the ethnohistoric data collected by Hocquenghem and Ortlieb for the 19th century, medium or even strong (not mega) El Niño events were often welcomed in the Peruvian countryside because they allowed cultivating extra plots of land (temporales) and exploiting larger and richer pastures (Hocquenghem and Ortlieb 1992: 203). Therefore, the capacity of the group to exploit different resources would greatly increase its chances to get through the lean period.


Estimating the antiquity and the frequency of El Niño

Reconstructing the onset of the ENSO phenomenon and the timing of specific El Niño events is crucial for understanding the role of El Niño in the transition to agriculture in coastal Peru. Several kinds of sources allow tracking past El Niño occurrences. Since instrumental records cover only a little over last hundred years, the most important sources are historical accounts and geoarchaeological evidence including ice cores, lake sediments, paleontological deposits, and beach ridges.

Historical data provide the most detailed information on the El Niño events that occurred in 16th-20th centuries, but the 16th-17th century accounts are less consistent and internally verifiable (Hocquenghem and Ortlieb 1992; Huertas Vallejos 1993). Since ENSO is a global phenomenon, some comparative historical information is available in the Old World. For example, Quinn (1993) demonstrated that there is a strong correspondence between large-scale ENSO events and low Nile river flood levels, which are accounted since AD 629. It is still well too recent for the purpose of my study. Nevertheless, these data provide the only way to check and calibrate the chronologies based on other sources.

Andean ice cores are another reliable source on the past ENSO events. A core from the Quelccaya ice cap provided a high-resolution picture of the climate change and variation over the past 1500 years that was successfully compared with the historical reconstructions (Thompson et al. 1984; Thompson 1993). Later research on the two cores from the col of Huascarán in the Cordillera Blanca allowed establishing a general picture of the climate change over the past 15000 years including the evidence of the Younger Dryas (Thompson et al. 1995). According to the pollen records and the oxygen isotopic ratios retrieved from the core, the warmest conditions for the Holocene prevailed between 8400 and 5200 BP, whereas modern conditions (reflected in the vegetation patterns) were established around 3000 BP (Thompson et al. 1995: 49). However, the resolution of this reconstruction is too low for identifying ENSO events.

Lake sediments constitute the source of data that combines high resolution and time depth. A recent study of the sediment core from an alpine lake of Laguna Pallcacocha in Ecuador provided the first detailed history of ENSO phenomena from 15000 BP to the present (Rodbell et al. 1999). Storms caused by individual ENSO events were identified as the layers of light-colored inorganic, clastic laminae (Rodbell et al. 1999: 518). The absolute chronology of the sediments was established with fourteen AMS 14C dates of terrestrial macrofossils and seven tephra layers. The full sediment record revealed a two-stage transition from rare and weak ENSO events between 15000 and 7000 BP (low sedimentation rates, periods over 15 years) to modern periodicity of ENSO (2-8.5 years) around 7000 BP, and to modern rates of sedimentation reflecting strong ENSO events by 2400-1200 BP (Rodbell et al. 1999: 519).

The increase in the strength of El Niño phenomena is documented by the beach ridges, which are formed out of enormous amount of sediments carried down by the rivers during mega-El Niño events. For example, the mass of suspended sediments in the Jequetepeque river increased about twenty times above average in 1982-1983 because of torrential rains brought by mega-El Niño (Teves Rivas 1993: 109). The formation of a new beach ridge in Northern Peru was documented during the same event (Woodman and Mabres 1993). A research on the two beach ridge series formed by the Chira river floods and by the temporal rises of the sea level identified ten or eleven successive major El-Niños or clusters of such events, the earliest dated to about 5090 BP, but seven occurring after 3000 BP (Ortlieb et al. 1993).

The overall picture of climate change and the reconstructions of the shifts in the frequency of ENSO events are also confirmed by archaeological data from the southern Peruvian coast. The site of Quebrada Tacahuay contains several El Niño-related deposits, which were dated with respect to the Early Holocene human occupation episode (Keefer et al. 1998). The deposits below and above the occupation layer suggest dry conditions interrupted by rare El Niño-brought floods. A period of several rare floods (once every 700-800 years) dated to about 12500-8900 cal. BP was followed by only one flood between 8700 and 5300 cal. BP.

Yet another method to track the history of El Niño was applied by Sandweiss, who tracked the changes in the shellfish midden assemblages at the coastal archaeological sites (Sandweiss et al. 1996; Sandweiss 1996b, 2003). Contrary to the present situation, tropical mollusk species were found as far south as 10°S, not only north of 4°S, at the sites dated to 11000-5000 BP. Temperate taxa appeared around 5000 BP indicating a shift to modern conditions (Sandweiss 2003: 30-33). The final transition that Sandweiss dates to around 3000 BP and identifies with the onset of the present El Niño frequency rate is marked by the disappearance of those temperate species which are most sensitive to temperature shifts from the faunal assemblages at the sites located between 9° and 4°S (Sandweiss 2003: 33-34). Sandweiss method is inherently contestable because past food preferences of the coastal inhabitants may have reflected cultural choices, and yet his conclusions are confirmed by recent data on lake sediments (see above). The only adjustment that one can make is that the latest change in mollusk assemblages on the coast was likely related not to higher frequency, but to greater strength of El Niño events.

Therefore, at least three lines of evidence indicate that there were two major climatic changes related to El Niño, which surely affected the coastal populations. The onset of the modern frequency rate of El Niño events during Mid-Holocene is of special importance because it coincides with the appearance of the first cultivars on the coast. One can also extrapolate the present seasonal regime and the upwelling area of the Humboldt Current on the Middle-Late Preceramic times. However, mega-El Niño disasters seemingly did not play any major role in the Middle-Late Preceramic adaptations.


El-Niño and the transition to agriculture: concepts

Despite its obvious impact on the life of Peru today, the El Niño phenomenon did not fare well in the academic discussions on the origins of agriculture on the Peruvian coast. The underestimation of El Niño is especially evident if one compares the two theories that governed the minds of the scholars back in the seventies and still produce some interesting ramifications the population growth and the maritime complexity models.

The population growth model was initially suggested by Cohen (1977). It was based on the theoretical assumption that the population pressure is the prime factor in the evolution of human societies. The data came from the results of the systematic archaeological research in the Chillon valley and the adjoining Ancon region (Central Peru). According to Cohen, sometime after 5600 BP, the population of Ancon-Chillon adopted the first cultivars from the region of Ayacucho in response to diminishing resources in all the three available ecological zones: river valleys, seacoast, and especially lomas (Cohen 1977: 161). It was still the same population of mobile hunters and gatherers who roamed the lomas in search for food (Cohen 1977: 159). In other words, at some point, there was no way to intensify the local hunting and gathering economy. Consequently, people switched to agriculture of which they already knew due to the contacts with the highland population. Cohen argued that the style similarities between the Canario complex in Ancon-Chillon and the Piki complex in Ayacucho indicate some form of interaction between the two areas as early as 7000-6200 BP about the time of the first domesticates in Ayacucho (Cohen 1977: 158).

An important feature of the model was that Cohen interpreted the archaeological evidence as representing a single population that exploited all three ecological zones and progressively failed to meet the demands imposed by its growth. The top demographic limit for the hunter-gatherers economy was reached around 4500 BP and the full-scale transition to farming followed (Cohen 1977: 162). Cohen suggested that the first villages on the coast like the site of Pampa (4500 BP) were specialized fishing settlements and that all the cultivated plants attested at those sites came from the unknown villages in the river valleys, which constituted a single exchange network (Cohen 1977: 162-165).

By contrast, Moseley (1975) offered a model based on the assumption that all costal developments were conditioned by the transition to full-time fishing. The scholar relied on the same Ancon-Chillon data but interpreted it differently. In this case, the coast provided the main source of food for the mobile hunters and gatherers (already equipped with a few cultivars by 5600 BP). Moseley suggested that a resource conflict prevented these early groups from concentrating on one of the three available ecological zones, but around 4500 BP the exploitation of marine resources became the dominant subsistence pattern (Moseley 1975: 42-43). The main idea was that the marine resources of the central Peruvian coast were exceptionally rich, easily exploitable, reliable, and predictable. Moseley discounted the possibility of food shortages caused by past El Niño events arguing that such lean periods would have been short while some extra resources like weakened sea birds would have been temporally available (Moseley 1975: 44-46). According to Moseley, agriculture played a minor and essentially complementary role: although some settlements obviously enjoyed a location that allowed fishing on the coast and farming in the river valley, the greater emphasis on agriculture was simply a result of the transition to sedentary life (Moseley 1975: 47-48). It was only during the Initial period when further population growth and the introduction of irrigation made farming a viable alternative to fishing (Moseley 1975: 57).

Both models earned some well-deserved critique. Raymond (1981) was one the first scholars to sum up a considerable amount of evidence against the affluent fishers. Nevertheless, his critique can be reduced to two arguments. First, the location of the Late Preceramic coastal sites could be better explained by the preference of the valleys with a potential for floodwater farming (Raymond 1981: 817). Second, the actual assemblages of faunal and floral remains could not support the idea of predominantly marine-based subsistence when compared with the population estimates during the Late Preceramic (Raymond 1981: 808-811).

An even more substantial set of arguments was proposed by Wilson (1981) who rejected the idea of bold calculations of bones, shells, and plant remains, but tried to estimate the maximum potential of the coastal environment. First, he tried to calculate the impact of El Niño events on the productivity of the Humboldt Current (Wilson 1981: 100). Second, he estimated the carrying capacity of the marine ecosystem during normal and bottleneck El Niño conditions taking into account different scenarios based on the optimistic and pessimistic reconstructions of the effectiveness of Preceramic fishing technology (Wilson 1981: 104-106). Finally, he compared the results with the estimations of the carrying capacity of the coastal maize agriculture and demonstrated that only agriculture-based subsistence system could have sustained the Preceramic populations (Wilson 1981: 105-106).

The population growth model was recently questioned by Hastorf (1999) who argued that the adoption of agriculture by sedentary foragers was rarely governed exclusively by the subsistence issues. Instead, the regular use of domestic crops in this case seems to be more about cultural symbols and kinship relations than hunger (Hastorf 1999: 37). Hastorf suggested that the discontinuous distribution of the adopted cultivars during the Late Preceramic period indicates that the new plants, non-local tasty food, were used as markers of ethnicity, but not as staples (Hastorf 1999: 48-54).

Nevertheless, the old models are still there. Hastorfs arguments are based on the assumption that the marine resources could have sustained coastal populations well into the Initial period (Hastorf 1999: 52) an assumption that dismisses all research that has been done on the carrying capacity of marine adaptations. In a similar way, Wilson distinguishes between the small coastal settlements of subsistence fishers who effectively controlled the population growth and the villagers in the valleys who for some reason chose to adopt farming, but provides no argument why different choice were made (Wilson 1981: 113-114). Something is missing in either case. In other words, it is important to consider the social and cultural choices, but it is essential to reconstruct the circumstances that made such choices possible. Understanding the evolution of the coastal climate and its impact on human populations provides a potential direction for such research.


First fishers, first farmers, and the onset of El Niño

Understanding cultural change in coastal Peru is not an easy business. Few valleys can boast more or less complete early-to-late Preceramic chronologies and some existing reconstructions have been recently questioned because of the inconsistencies in 14C dates (Pozorski and Pozorski 1999). There are several chronological gaps, notably, the one between Early and Middle Holocene coastal sites (Chauchat 1992: 340-343). If the new dates for Late Preceramic coastal sites are correct, there will also be a hiatus between Mid-Preceramic and Initial Period preceramic sites in several valleys of the north-central coast of Peru. Therefore, I will try to center on the most reliable data and avoid all chronologically sensitive cases. I assume that the well-documented Mid-Preceramic sites in the Chilca and Huarmey valleys provide enough information to consider the impact of the onset of El Niño on the coastal societies. As for the earlier times, I will focus on the cases with good chronological attribution.

The Paijan culture is one of the earliest coastal adaptations in northern Peru. The Paijan hunting complex is represented by several ephemeral sites in the Chicama Valley, which may have been in existence around 10380 - 8260 BP (Chauchat 1992: 340-341, 353). The coastal component of this culture was submerged during postglacial sea rise, but the remaining sites, which likely correspond to seasonal procurement of inland resources, provide some evidence of the sea-based economy of Paijan people. Although the known sites are located some 14-36 km inland and the ancient coastline was some 5-20 km further away, fish remains dominate the faunal assemblages at some of them.

The Paijan tool kit is distinguished by the ubiquity of projectile points designed to spear or to harpoon large fish rather than terrestrial fauna (too large for small mammals, too fragile for any hard-skinned game) an indicator of some coastal estuaries or lagoons, where such fishing style could have been efficient (Chauchat 1992: 336-337, 354). The marine assemblage consists mostly of tropical fish including catfish (Ariidae) and lisa (Mugil spp.), as well as some temperate species, notably anchovies (Chauchat 1992: 356-357). Small cañán lizards were the primary hunting goal of the inland expeditions given their predominance in the overall sample. The cañán and other animal remains generally correspond to the lomas vegetation and ethnographic evidence suggests that Paijan people may have exploited these inland resources seasonally from May to September (Chauchat 1992: 358). No plant remains were recovered at the sites.

The southern coast generally provides better chances to track Terminal Pleistocene Early Holocene coastal sites since sea level change was not that significant because of a narrow sea-shelf (Richardson 1992: 80). One of the earliest sites Quebrada Tacahuay bears evidence of human occupation (butchered sea bird and fish bones, stone tools, hearths) dated to 10770-10530 BP (12700-12500 cal. BP). The local subsistence seems to have been based on sea bird, fish, and mollusks (Keefer et al. 1998). Interestingly, anchovy bones are common indicating that coastal fishers could catch them (with fine-mesh nets off coast!). There are no plant remains at the site. The deposits below and above the occupation layer suggest dry conditions interrupted by rare El Niño-brought floods. A period of rare floods (once every 700-800 years) dated to about 12500-8900 cal. BP was followed by only one flood between 8700 and 5300 cal. BP the warmest part of Holocene according to Peruvian ice cores. Several bifacial and unifacial flakes are similar to the tools found at the Ring site to the north.

The faunal assemblage pertaining to the earliest occupation layers at the Ring site dated to about 10570-7670 BP also provided evidence of a specialized maritime economy (Richardson 1992: 81). No remains of terrestrial animals (except 4 bones of mice) were found at the site despite its proximity to the lomas area, but sea birds, fish, and mussels constituted the bulk of the diet. In contrast to Quebrada Tacahuay, most fish species are larger carnivores that could be fished with hook and line. Indeed, several bone and shellfish hooks were found at the site. The seacoast was originally at some distance from the site and its advance through time, according to Sandweiss (1996b: 134) is reflected in the increase in the amount of shellfish remains: once the coast was nearby, people could eat the mollusks at the site. Notably, almost all (97.5%) fish and mollusks were cold-water species (Sandweiss 2003: 29).

Faunal remains at a roughly contemporaneous site of Quebrada Jaguay (Sandweiss et al. 1998) indicate an even more specialized fishing economy as the assemblage is dominated by the bones of drum fish (Sciaenae spp.) and by the shells of the wedge clam (Mesodesma donacium). The small average size of the drum fish suggests that nets were used. Some knotted cordage that may be parts of fishnets was attested. There is no permanent water source at the site and the occupation was likely seasonal. The site is remarkably poor in terms of plant remains, and yet the presence of prickly pear seeds (Opuntia) indicates that higher elevations were exploited (Sandweiss 2003: 28-29). Even more distant highland connections are suggested by the finds of obsidian flakes from the Chilca source some 130 km away (Sandweiss 1998: 1832). Finally, despite a presumably seasonal occupation, a set of postholes identified as remains of a semicircular structure some 5 m in diameter with a central hearth were found in the latest strata (10500-10800 BP).

In contrast to the three southern sites discussed above, following the sea-level rise but before the onset of modern El Niño conditions around 6000 BP, the ecosystem of the coast of Peru north of 10°S was characterized by the predominance of the tropical marine fauna. The Ostra Base Camp dated to about 6250 and 5450 BP (7150 and ~6250 cal. BP) represents the Mid-Holocene adaptations that postdate the Paijan culture but predate the climate change (Reitz and Sandweiss 2001: 1093-1095; Sandweiss 2003: 31-32). The faunal assemblage from the earlier layers at this coastal site (~6000 BP) indicates an intensive exploitation of a nearby estuary. The inhabitants of the Ostra Base Camp centered on procurement of tropical mollusks and fish, as well some sea turtles and cormorants. The temperate fish species including anchovies constituted only 30% of the total marine vertebrae sample. The poor preservation of non-carbonized plant remains did not allow toreconstruct this aspect of the local economy, but a few carbonized specimen were identified as huarango (Acacia spp.) and algarrobo (Prosopis spp.) tree fruits, which are common for Peruvian coast.

The Ostra Base Camp was abandoned around 5500 BP, as well as other sites to the north including Quebrada Chorrillo, Siches, and Avic, which also subsisted on tropical species (Sandweiss 2003: 31). The assemblages at a new set of site such as Los Morteros in the Chao Valley (4660-4010 BP) that appeared some 500 year after the climate change reflected a subsidence system based on temperate marine species (Cardenas 1999: 147, 151-157). By contrast, there was no discontinuity in the occupation or faunal assemblages at the coastal sites south of 10S. Moreover, their subsistence schemes spread to the north. The above-mentioned site of Los Morteros is contemporaneous or slightly earlier than PV35-6 in the Huarmey Valley and Chilca I in the Chilca Valley, which illustrate the success of the southern adaptation after the onset of modern conditions. Although Cardenass report on Los Morteros is highly fragmentary, one may generalize that the site corresponds to a subsistence system similar to Huarmey and Chilca sites. Besides temperate fish and mollusks, Los Morteros features some gourd remains, pacae, and algarrobo seeds. Grinding stones and net weights are common indicating a new trend towards greater reliance on plant food.

Given its long occupation history (7700-4800 BP) and the amount of archaeological research already done, the site of Paloma provides the best case for the southern Mid-Preceramic adaptations. Paloma is located in the Chilca Valley (some 7.5 km from the river) near the southern margin of the central coast area where most Cotton Preceramic developments would take place. Compared to any earlier site, Paloma is a substantial and maybe even permanent settlement that occupies about 15 he and features 200 graves and some 144 semi-subterranean houses that represent single or multiple occupations with a trend towards clusters of larger houses with patios (Benfer 1999: 224-227, Fig. 5-7). Since, the Chilca River provides only seasonal flow during the summer, agricultural production was likely not the major factor in choosing the settlement location. However, Parsons suggested that there would have been more water in the river when it was not diverged for irrigation with upstream canals (Parsons 1970: 298).

The faunal assemblage at the site is dominated by the remains of fish and mollusks - 30.2% and 69.1% of the probability sample collected by Reitz (1988: 46, Tab. 5). Anchovies, drums, and herrings (Clupeidae spp.) are the most common fish species, whereas mussels (Mytilidae) were the most frequent invertebrates (Reitz 1988: 33, 44-45, Tab. 3). Remains of terrestrial and marine mammals are rare (0.5% of the faunal assemblage). The estimations of the total biomass show that fish, mussels, and mammals were chief contributors to the animal part of the diet 75.6 %, 8.7 % and 8.6 % respectively (Reitz 1988: 46, Tab. 5). Reitz believes that the list of hunted species (sea lions, giant fulmar, guanaco, and deer) suggests year-round occupation (Reitz 1988: 34). The assemblage suggests that the role of sea mammals and anchovies in the diet increased through time (Benson 1990: 306).

The floral remains indicate that Palomans relied on a variety of wild plants, which were collected in the nearby lomas: guayaba (Psidium sp.), mito (Carica candinas), lucuma (Lucuma obovata), cactus fruits (Haageocereus spp.), and some wild tubers including Begonia (Weir et al. 1988: 63, 67). Nevertheless, Paloma is the earliest coastal site where three cultivated plants squash (Cucurbita spp.), beans (Phaseolus and Canavalia spp.), and gourd (Lagenaria siceraria) are attested.

A study of coprolites revealed that the marine and plant foods were prominent in the local diet (Weir et al. 1988: 63). Fish, marine mammals, and crab constituted were complemented by a significant percentage of plant fibers and seeds: seaweed, grasses, solanaceous seeds, squash, and some chenopods. Plant epidermis and fiber were present in 74% of coprolites and a considerable portion of the plant epidermis could have been tubers (Weir et. al. 1988: 66).

The bones and hair of the Palomans also provide some clues to their dietary preferences (Weir et al. 1988: 61-62). Thus, higher levels of magnesium (presumably caused by eating seafood, especially seaweed) distinguish the Paloma sample from Late Preceramic sites. The heavy reliance on marine foods is also suggested by evidence of strong tooth wear, although there is a clear trend of decreasing wear indicating a shift to less abrasive food like tubers (Benson 1990: 301, 304). However, very low percentage of dental caries hints that there was no substantially increasing consumption of carbohydrates. The trace elements in hair and the Harris lines of limb elements indicate that the population experienced regular, probably seasonal stress and shifts in diet, but nothing like a major stress/famine caused by strong El Niño events. The general trend throughout the site occupation history is of decreasing stress and increasing life expectancy (Benfer 1990: 292-293, 304). Later inhabitants were larger, taller, and with smaller teeth. Skeletal remains also indicate that a shift to more activities involving the upper part of the body (for example, fishing) took place, as well as that men spent a lot of time swimming and diving in cold water (Benfer 1990: 305).

The subsequent trend in human subsistence is apparent from the assemblage at the Chillca 1 site dated to somewhat later period (5600-5000 BP). Chillca 1 is located just several kilometers south of Paloma, closer to the river and its mouth with sandy beaches. Demographic studies based on the skeleton sample at Paloma indicate that after a period of rapid growth, its population actually decreased during the last stages of the occupation at the site. Since this decrease happened in the context of increasing life expectancy and decreasing food stress, the only plausible explanation is that the location of Chilca 1 was more attractive and the population slowly shifted to the new site or even migrated outside the valley (Vradenburg et al. 1997). An alternative explanation would be population replacement (Benfer 1990: 299; 1999: 232-233). One may wonder if the period of overlap in the occupation of the two sites reflects their distinct functions a specialized fishing village (Paloma) and a central valley settlement (Chilca 1). That may explain the increasing role of fishing at Paloma.

Anyway, Chilca 1 is quite similar to Paloma, both in architecture, stone tools, and the composition of floral and faunal assemblages (Weir et. al. 1988: 64). The evidence from bones and coprolites suggests no major difference in the diet between the two sites (Weir et. al. 1988: 62, 66). In either case, humans subsisted mostly on marine and wild plant resources and were subject to some seasonal food stress. However, whereas a species of wild Begonia is a major candidate for a tuber component of the diet, Chilca 1 is distinguished by the presence of another root crop achira or Canna edulis (Weir et. al. 1988: 64). Cotton was also attested alongside with the more traditional cordage made of maguey and junco (Weir et. al. 1988: 64).

Further to the north along the Central Coast, the valley of Huarmey River provides a long sequence that comprises several sites (middens and storage pits), the earliest being contemporaneous to Paloma. The Huarmey sites are essential for understanding the changes triggered by the establishment of modern climate conditions. There are some problems with the data, notably very late direct AMS dates of the maize cobs from the main site of Gavilanes (Bonavia 1982: 73-74) and the absence of anything identified as habitation units, but the abundant floral and faunal remains document an increasing reliance on domesticated plants between 6500 and 4000 BP.

The site of PV35-106 located near the extinct estuary to the north of the Huarmey River is a dense accumulation of refuse some 60 cm deep with occupation dated to about 6430 BP or between 7590 and 6897 cal. BP (Bonavia 1982: 256; Bonavia 1996: 172). According to Bonavia, the environment of the lower part of the Huarmey Valley around 6000 BP was more humid than at present and supported estuaries or swamps, as well as a dense vegetation cover near the river (Bonavia 1996: 172). Since the Huarmey Valley is north of 10S, the occupation at PV35-106 is interesting as it predates the proposed climate change. Sandweiss notes that the published data do not allow proving or contesting this hypothesis, but the later faunal assemblages at Gavilanes and at PV35-6 are dominated by temperate-water species (Sandweiss 2003: 32). The stone tool kit probably corresponds to the Mongoncillo industry in the Casma valley that in turn may have been derived from the Paijan culture (Bonavia 1996: 174). There are few botanical remains, but squash and bottle gourd were identified (Bonavia 1996: 175). Some tubers, too small to be used for sustenance, were also present. Animal remains include mollusks, sea urchins, fish, and seals, but no quantitative data are provided.

Bonavia suggests that the PV35-106 was roughly contemporaneous to the earliest occupation at a major site of Gavilanes (PV35-1) (Bonavia 1996: 176). However, the first epoch at Gavilanes was already characterized by a wider list of domesticates such as achira (Canna edulis), peanut (Arachis hypogaea), and pacae (Inga Feuillei). It would probably be safer to assume that PV35-106 is somewhat earlier. The domesticated plants were probably cultivated in the river valley with simple flood farming techniques (Bonavia 1996: 182). Given that the relatively narrow Huarmey Valley has been extensively used by agriculturalists since Late Preceramic times, there is no chance to find any evidence of the earliest farming attempts (Bonavia 1982: 257).

The site of PV35-6 occupied around 4005 BP or 4949-3982 cal. BP corresponds to the transition between the first and the second period at Gavilanes dated to about 4140 (14C) and 4800 (TL) BP or 5204-4240 cal. BP (Bonavia et al. 1993: 413; Bonavia 1996: 176-178). Although Bonavia, who has been digging the Huarmey sites over the past thirty years, strongly argues in favor of local developments, there is a considerable similarity between the subsistence schemes at Chilca I and PV35-6. PV35-6 is located to the south of the river and only a low hill separates it from the beach (Bonavia et al. 1993: 411). Indeed, the subsistence system of the site was largely based on the exploitation fish and mollusks of the temperate waters of the Humboldt Current (Bonavia et al. 1993: 422-423; Sandweiss 2003: 32). Nevertheless, the pollen sample from the coprolites suggests that the inhabitants of the site also spent some time in the river valley (Bonavia et al. 1993: 434-437). Grinding stones appear in the lithic assemblage (Bonavia et al. 1993: 413). The plant remains recovered at the site include squash and bottle gourd, as well as several new domesticates: cotton (Gossypium barbadense), beans (Phaseolus spp), and avocado (Persea americana) (Bonavia et al. 1993: 417-421; Bonavia 1996: 176-177). It is interesting that cotton is attested as plant remains, cloths, and threads, but fishing nets (quite simple if compared with those at Gavilanes) were made of different plant fiber (Bonavia et al. 1993: 437). The only terrestrial mammals in the diet were rodents: their hair and bones were found in the coprolites (Bonavia et al. 1993: 425). Therefore, although there are no estimates of the relative importance of different components of the diet, the trend towards an increasing amount of domesticates is quite apparent.



The data presented above suggest the following reconstruction of events. There were at least two distinct adaptations in the early Holocene. Whereas Paijan people actively exploited the tropical setting of the northern coast, a distinct subsistence scheme emerged in the south and spread as far north as the temperate waters of the Humboldt Current. The climate change around 6000BP resulted in the increase of the upwelling system, but also in dryer conditions that made the southern coast particularly inhospitable. In case of Quebrada Jaguay, a shift to drier climate occurred even earlier: the region was virtually abandoned by 8000 BP (Sandweiss 2003: 29). However, the temperate maritime adaptation persisted on the Central Coast and the site of Paloma clearly represents a success compared to the sites such as the Ostra Base Camp. The most important questions are why the temperate maritime adaptation favored the introduction of domesticates, why this trend became stronger as the adaptation spread further north, and what the onset of El Niño events had to do about it.

As it has been discussed above, the temperate setting is characterized by the abundance of marine resources. The data from Quebrada Tacahuay, Ring site, and Quebrada Jaguay suggest that the Early Holocene fishers could exploit these resources as efficiently, as in Late Holocene times. They also faced the same set of problems, notably seasonal fluctuations in the upwelling conditions. Although the available data on skeletal remains is restricted to the sites of Paloma and Chilca 1, one may predict the same pattern of stress for Early Holocene fishers. These fluctuations would gradually increase to the north. After the onset of modern El Niño frequency, the amplitude of these fluctuations would become even greater. A distinct south-north pattern would remain. Therefore, on one hand, there would a new area between 4S and 10S suitable for intensive fishing, but on the other hand, the seasonal variations in the availability of the marine resources would be higher and the predictability of the year-to-year variation would be lower.

One possible solution of the problem of seasonal shortages is transhumance. The data presented above demonstrate that lomas provided an important contribution to the diet of the coastal inhabitants as early as Terminal Pleistocene - Early Holocene. The climate between 8400 and 5200 BP would even support a somewhat richer vegetation cover. However, the transhumance would not be such an ideal choice as soon as modern conditions were in place. A lean season for fishers that falls between February and June would partially overlap with bad times for loma hunters and gatherers between the months of November and May. Of course, some loma tubers would be available year-round and that may explain the prominence of Begonia at Paloma, but there would be still at least two months of diminished terrestrial and marine resources. Moreover, the most bountiful seasons for marine and terrestrial resources would also partially overlap creating a classic case of resource conflict. Besides, one should consider minor regional variations in seasonal abundance of particular resources. For example, there are specific anchovy runs in the area of Paloma, which were likely exploited by its inhabitants back in Middle Preceramic (Jackson and Stocker 1982: 17).

Another predictable response to seasonal variation is storage. Storage could have played a more significant role than the available data indicate. The most common form of storage was sand pits simple, effective, but virtually invisible archaeologically. The case of storage pits at Gavilanes a form of storage still practiced in that area (Bonavia 1982: 68-71) demonstrates that one has to look specifically for this type of installations in order to find them. The lenses of anchovy remains at Paloma likely represent a similar case. The absence of flies suggests that anchovies were covered with sand immediately after deposition (Jackson and Stocker 1982: 17). Over 500 such simple storage pits were excavated at Paloma, a typical content being headless and salted anchovies (Benfer 1990: 311). Besides, there are some candidates for stone-lined storage pits at the Late Preceramic sites of Asia, Aspero, and El Paraiso (Bonavia 1982: 260-264).

The third solution is to diversify the resource base by adding some cultivated plants to the menu. Given that planting should follow seasonal floods that come around December-January, agricultural products would help with the food stress depending on their period of growth. Therefore, either short-growth or well-storable plants would be preferred. Peanut and squash would seem to be a good short-growth-period solution, while achira could have been easily stored (just left in the ground). The month of March would be critical anyway, and it is reasonable to suggest that cultivated plants and storage pits were part of a single package. Moreover, some rains brought by El Niño would actually favor simple farming in the same time when maritime component of the subsistence system would be under an extra stress. Torrential rains would be uncommon prior to 3000 BP.

Therefore, the onset of modern El Niño conditions and the spread of temperate maritime adaptation north of 10S would exert a relatively constant pressure on the subsistence system in favor of storage and resource diversification. As Raymond (1981: 817) first noticed, most Late Preceramic sites were located in proximity to the coast, but also to self-flooded alluvial plains with a potential for simple floodwater farming.

Another key issue is where the domesticates came from. There are several potential donors: coastal Ecuador, river valleys in northernmost Peru, sierra, and the central highlands. Unfortunately, notoriously little is known about the interaction networks in Preceramic Peru. Recently, Sandweiss suggested that at least two interaction zones comprising Ecuadorian and North-Central Peruvian coasts were in place by Mid-Holocene times (Sandweiss 1996a). One was represented by a plain pebble offering tradition. The earliest offerings containing plain pebbles appear at Las Vegas in Coastal Ecuador around 8000-6600 BP. First Peruvian equivalents occur at Paloma (!) and later at Alto Salavery and Aspero. Another interaction sphere was a Northwest Andean pebble figurine tradition marked by the presence of planed-incised pebbles attested at Valdivia sites in Ecuador and at the tropical-water sites of the Northern coast including Ostra, Siches, and Honda. Given these early contacts, it not surprising that several earliest cultivars (gourd, squash, Canavalia beans, achira) on the coast belong to the lowland complex initially attested in the coastal Ecuador (Pearsall 1992: 194).

The Mid-Holocene interaction spheres likely collapsed with the onset of modern climate conditions, since the expansion of Sechura desert would create a formidable obstacle between the coastal Ecuador and Peru (Sandweiss 1996a: 48). Moreover, the failure of the tropical maritime adaptation in Peru resulted in the abandonment of the sites with strongest Ecuadorian affinities. Nevertheless, there is evidence that some contacts between the coastal sites in Peru and some sites in a tropical setting continued after 5500 BP. Thus, the faunal assemblage at Paloma features femurs of a spider monkey and a mountain lion (Reitz 1988: 34). An even more striking discovery was made at PV35-6, where coca leaves (Erythroxylum sp.) were found (Bonavia et al. 1993: 420, 429-431).

Pearsall suggests the term Andean Mid-Elevation Complex and identifies the common beans, the lima beans and the coca as derived from the mid-elevation area of the western and the eastern slopes of the Andes (Pearsall 1992: 195-196). I would speculate that the Zaña Valley in Northern Peru could have been an origin place of the exotic things at Paloma and PV35-6. The setting of this valley is fairly unique right between the Lowland, Mid-Elevation and High-Elevation complexes identified by Pearsall (1992). First farming communities appeared in the valley between 8000 and 6200 BP (Rossen and Dillehay 1999; Dillehay et al. 1989). Plant remains include domesticates such as squash, peanuts, quinua, ciruela, and some unidentified tubers. Mollusk shells indicate that contacts with the coast were maintained. Interestingly, some shells belong to temperate water-sensitive species (especially C. chorus) an argument for somewhat later chronological placement of this culture (or some part of it) after 6000 BP (but before 3000 BP). There is an indirect evidence of coca consumption (lime powder that is usually added to masticated coca) and the valley is an ideal setting for coca cultivation.

By contrast, the sierra seems to be a less likely candidate. The Guitannero Cave in Callejón de Huaylas is frequently cited as the earliest site near the coast where a whole set of cultivars including oca, common beans, lima beans, pacay, aji, and squash were in use around 10600-7600 BP (Smith 1980). Direct AMS dates for common and lima beans turned out to be much younger 4337 and 3495 BP (Smith 1998: 162). The new dates make the interaction between the coastal and the sierra inhabitants a potential source of the first cultivars on the coast as there is seemingly no more gap in time between the domesticated plants at Guitarrero and the first cultivars on the coast. However, no evidence of the interactions with the coast (except one shell) was found at the Guitarrero Cave and at the contemporary sites of Callejon de Huaylas: its inhabitants did practice seasonal transhumance, but in the sierra and the adjacent mountains (Lynch 1970; 1980: 5, 310-311). Same lack of evidence characterizes potential contacts between the coast and the highlands, although the general agreement is that the beans at Paloma and Chilca 1 should have been adopted from somewhere (Piperno and Pearsall 1998: 271).

Since at least three distinct zones tropical lowlands, mid-elevation areas, and the highlands were responsible for the coastal package of domesticates, it seems that there can be no simple picture of the interactions, which led to the adoption of specific plants. These interactions were in existence before the onset of El Niño conditions and they persisted afterwards, albeit in a changed form. Another important feature is the absence (at least from the point of the available data) of any frontier between the early farmers and fishers. Instead, it seems that the coastal fishers adopted some plants through long-distance networks and the only pressure they experienced was that of local environmental conditions, notably the seasonal fluctuations in the availability of terrestrial and marine resources aggravated through the onset of modern climate.

Of course, one should not overestimate the scale of these changes. In terms of Zvelibils classic model (Zvelebil 1996: 324-325), the onset of El Nino initiated a long substitution phase when agriculture contributed less than 50% of the diet. It is also important to consider to what extent the conditions that triggered the introduction of agriculture favored its further development. Thus, the spread of cold waters of the Humbolt Current further to the north would surely bring drier conditions along the coast, hitting the lomas segment of the subsistence system. However, same conditions would not be particularly favorable for simple farming.

The appearance of the incipient agriculturalists strengthened the trend towards social and economic differentiation that was probably in place for the sedentary and storing society of the coastal fishers. Notably, there are yet no signs of social differentiation at Paloma, but the emergence of larger clusters of houses during its latest occupation hints that some social change was under way. The introduction of cotton was also an important step: the data from Paloma and Huarmey sites indicate that fishing was an unlikely factor in the adoption of this plant. It was initially used to make fancier cloths, not fishing nets. The role of cotton as a trade item was strongly emphasized by some scholars (Quilter 1991a) and this suggestion seems reasonable. The quest for extra surplus could have promoted further emphasis on agriculture or increased reliance on particular cultivars. For example, it would be hard to explain the shifts in the frequency of cultivated plants at the Late Preceramic site of Huaca Prieta (Bird 1988: 9) in purely ecological terms. However, this story has little to do with El Niño events.


Concluding remarks

A common misconception about the maritime setting of the Peruvian coast is that it is just abundant. The truth is that there are considerable fluctuations in the availability of marine resources. The seasonal pattern of these fluctuations is predictable, but the year-to-year magnitude is not. The worst years are characterized by El Niño a warm countercurrent that sweeps away the rich ecosystem of the Humboldt Current and brings torrential rains to the coast.

The climate conditions during Early-Middle Holocene differed from the present. The tropical coastal waters extended down to 10°S and El Niño years occurred much less frequently. Two alternative efficient fishing adaptations existed on the coast during that period. Modern conditions were established around 6000 BP and the temperate fishing adaptation spread northward. The onset of El Niño lowered the predictability of the seasonal variation of temperate species. Besides, the impact of El Niño was stronger in the north. Therefore, the fishers had to diversify the resource base in order to exploit the newly available temperate waters north of 10°S and to cope with greater food shortages south of 10°S. As a result, some cultivated plants were added to the menu.

The economy of the Late Holocene coastal settlements was still based on fishing that would remain the primary form of subsistence well into Cotton Preceramic. However, the problem of the seasonal food stress was solved not only through further elaborations in fishing technology, but also through the introduction of more cultivars. In the two cases when local Mid-to-Late Preceramic sequences are available, there is a clear trend of increasing reliance on wider array of domesticated plants.

The interaction network that connected the Peruvian fishers with the first agriculturalists of the Ecuadorian coast likely played an important role in the introduction of the first cultivars. However, a disruption of this network occurred after the onset of frequent El Niño events and the collapse of tropical fishing adaptations in northern Peru. Subsequently, other areas, notably the valleys of the Western Andean slopes of Northern Peru, were the source of some domesticates. The available data do not allow understanding the nature and the historical development of these Late Holocene networks, but it seems that they were characterized by long-distance communications rather than frontier interactions.



ACEITUNO, Patricio
1998    El Fenómeno El Niño 1997-1998
Bull. Inst. fr. études andine 27(3): 443-448

BIRD, Robert McK.
1988    Preceramic Archaeobotany of Huaca Prieta: Investigations from 1946 to 1986
In: Economic Prehistory of the Central Andes, Elizabeth S. Wing and Jane C. Wheeler, eds. P. 31-94
BAR International Series 427, Oxford

BENFER, Robert A.
1999    Proyecto de Excavaciones en Paloma, Valle de Chilca, Perú
In: El Periodo Archaico en el Perú: Hacia una Definición de los Origenes, Peter Kaulicke ed.
P. 213-237
Bolletín de Arqueología PUCP 3, Fondo Editorial

1990    The Preceramic Period Site of Paloma, Peru: Bioindications of Improving Adaptations to Sedentism
Latin American Antiquity 1(4): 284-318

1996    De la Caza-Recollección a la Agricultura: Una Perspectiva Local
Bull. Inst. fr. études andine 25(2): 169-186

1982     Los Gavilanes: Mar, Desierto y Oasis en la Historia del Hombre
Editorial Ausonia, Lima

BONAVIA, Duccio, Laura W. JOHNSON, Elizabrth J. REITZ, Elizabeth S. WING, Glendon H. WEIR
1993    Un Sitio Precerámico de Huarmey (PV35-6) antes de la Introducción del Maíz
Bull. Inst. fr. études andine 22(2): 409-442

1992    Préhistoire de la Cote Nord du Pérou: Le Paijanien de Cupisnique
CNRS, Centre Régional de Publication de Bordeaux

COHEN, Mark N.
1977    Population Pressure and the Origins of Agriculture: An Archaeological Example from the Coast of Peru
In: Origins of Agriculture, Charles A. Reed, ed. P. 135-177
Mouton Publishers, The Hague

1993    El Fenómeno El Niño y los Moluscos de la Costa Peruana
Bull. Inst. fr. études andine 22(1): 159-177

1989    Middle Preceramic Public and Residential Sites on the Forested Slope of the Western Andes, Northern Peru
American Antiquity 54(4): 733-759

1993    Registros de la Vegetación en la Costa Peruana en Relación con el Fenómeno El Niño
Bull. Inst. fr. études andine 22(1): 259-266

HASTORF, Christine A.
1999    Cultural Implications of Crop Introductions in Andean Prehistory
In: The Prehistory of Food: Appetites for Change, Chris Gosden and Jon Hather, eds. P. 35-58
Routledge, London

1992    Eventos El Niño y Lluvias Anormales en la Costa del Perú: Siglos XVI-XIX
Bull. Inst. fr. études andine 21(1): 197-278

1993    Anomalías Cíclicas de la Naturaleza y Su Impacto en la Sociedad: El Fenómeno El Niño
Bull. Inst. fr. études andine 22(1): 345-393

1982    Perus Preceramic Menu
Field Museum of Natural History Bulletin 53: 12-23
Field Museum of Natural History, Chicago

1998    Early Maritime Economy and El Niño Events at Quebrada Tacahuay, Peru
Science 281: 1833-1835

LYNCH, Thomas F.
1980    Guitarrero Cave: Early Man in the Andes
Academic Press, New York

1970    Preceramic Transhumance in the Callejon de Huaylas, Peru
American Antiquity 36(2): 139-148

MACNEISH, Richard S.
1992    The Beginnings of Agriculture in the New World
Revista de Arqueologia Americana 6: 7-34

MARCHANT, Margarita, Dierk HEBBELN, Gerold WEFER
1998    Seasonal Flux Patterns of Planktic Foraminifera in the Peru-Chile Current
Deep-Sea Research I 45: 1161-1185

MERLE, Jacques
1998    South Pacific Climate Variability and Its Impact on Low-Lying Islands
Bull. Inst. fr. études andine 27(3): 461-473

MOSELEY, Michael E.
1975    The Maritime Foundations of Andean Civilization
Cummings Publishing Company, Menlo Park

NIXON, Scott, Andrew THOMAS
2001    On the Size of the Peru Upwelling Ecosystem
Deep-Sea Research I 48: 2521-2528

1993    Beach-Ridge Series in Northern Peru: Chronology, Correlation, and
Relationship with Major Late Holocene El Niño Events
Bull. Inst. fr. études andine 22(1): 191-212

1977    Strandloopers, Mermaids and Other Fairy Tales: Ecological Determinants of Marine Resource Utilization the Peruvian Case
In: For Theory Building in Archaeology, L. R. Binford, ed. P. 157-205
Academic Press, New York

PARSONS, Mary Hrones
1970    Preceramic Subsistence on the Peruvian Coast
American Antiquity 35(1): 292-304

PEARSALL, Deborah M.
1992    The Origins of Plant Cultivation in South America
In: The Origins of Agriculture: An International Perspective, Wesley Cowan and Patty Jo Watson, eds. P. 173-205
Smithsonian Institution Press, Washington, D.C.

PICKERSGILL, Barbara and Richard T. SMITH
1981    Adaptations to a Desert Coast: Subsistence Changes through Time in Coastal Peru
In: Environmental Aspects of Coasts and Islands, Don Brothwell and Geoffrey Dimbleby, eds.
P. 89-115

BAR International Series 94, Oxford

PIPERNO, Dolores S. and Deborah M. PEARSALL
1998     The Origins of Agriculture in the Lowland Neotropics
Academic Press, San Diego

POZORSKI, Shelia and Thomas POZORSKI
1999    Una Reevaluacion del Desarrollo de la Sociedad Compleja durante el Preceramico Tardío en Base a los Fechados Radiocarbónicos y a las Investigaciones Arqueológicas en el Valle de Casma
In: El Periodo Archaico en el Perú: Hacia una Definición de los Origenes, Peter Kaulicke ed.
P. 171-186

Bolletín de Arqueología PUCP 3, Fondo Editorial

1991    The Impact of Radiocarbon Dates on the Maritime Hypothesis: Response to Quilter
American Anthropologist 93(2): 454-455

1990    Reexamining the Critical Preceramic/Ceramic Transition: New Data from Coastal Peru
American Anthropologist 92(2): 481-491

QUILTER, Jeffrey
1991a    Late Preceramic Peru
Journal of World Prehistory 5(4): 387-438

1991b    Problems with the Late Preceramic Peru
American Anthropologist 93(2): 450-454

QUINN, William H.
1993    The Large-Scale ENSO Events, the El Niño and Other Important Regional Features
Bull. Inst. fr. études andine 22(1): 13-34

QUISPE Arce, Juan
1993    Variaciones de la Temperatura Superficial del Mar en Puerto de Chicamay del Índice de Oscilación del Sur: 1925-1992
Bull. Inst. fr. études andine 22(1): 111-124

REITZ, Elizabeth J.
1988    Preceramic Animal Use on the Central Coast
In: Economic Prehistory of the Central Andes, Elizabeth S. Wing and Jane C. Wheeler, eds. P. 31-55
BAR International Series 427, Oxford

REITZ, Elizabeth J., Daniel H. SANDWEISS
2001    Environmental Change at Ostra Base Camp, a Peruvian Pre-ceramic Site
Journal of Archaeological Science 28: 10851100

REYCRAFT, Richard M.
2000    Long-Term Human Response to El Niño in South Coastal Peru
In: Environmental Disaster and the Archaeology of Human Response, Garth Bawden and Richard M. Reycraft, eds. P. 99-120
Maxwell Museum of Anthropology, Anthropological Papers No. 7

1992    Early Hunters, Fishers, Farmers, and Herders: Diverse Economic Adaptations in Peru to 4500 BP
Revista de Arqueologia Americana 6: 71-90

RODBELL, Donald T., Geoffrey O. SELTZER, David M. ANDERSON, Mark B. ABBOTT, David B. ENFIELD, Jeremy H. NEWMAN
1999    An ~15000-Year Record of El Niño-Driven Alluviation in Southwestern Ecuador
Science 283: 516-520

1998    La Pluométrie au Pérou pendant les Phases ENSO et LNSO
Bull. Inst. fr. études andine 27(3): 675-685

1999    La Colonización y el Asentamiento del Norte del Perú: Innovación, Tecnología y Adaptación en el Valle de Zaña
In: El Periodo Archaico en el Perú: Hacia una Definición de los Origenes, Peter Kaulicke ed. P. 121-139
Bolletín de Arqueología PUCP 3, Fondo Editorial

2003    Terminal Pleistocene through Mid-Holocene Archaeological Sites as Paleoclimatic Archives for the Peruvian Coast
Palaeogeography, Palaeoclimatology, Palaeoecology 194: 23-40

1996a    Mid-Holocene Interaction between the North Coast of Peru and Ecuador
Latin American Antiquity 7(1): 41-50

1996b    Environmental Change and Its Consequences for Human Society on the Central Andean Coast: A Malacological Perspective
In: Case Studies in Environmental Archaeology, Elizabeth J. Reitz, Lee A. Newsom, Sylvia J. Scudder, eds. P. 127-146
Plenum Press, New York

SANDWEISS, Daniel H., Heather MCINNIS, Richard L. BURGER, Asuncion CANO, Bernardino OJEDA, Rolando PAREDES, Maria del Carmen SANDWEISS, Michael D. GLASCOCK
1998    Quebrada Jaguay: Early South American Maritime Adaptations
Science 281: 1830-1832

SANDWEISS, Daniel H., James B. RICHARDSON, Elizabeth J. REITZ, Harold B. ROLLINS, Kirk A MAASCH
1996    Geoarchaeological Evidence from Peru for a 5000 Years BP: Onset of El Niño
Science 273(5281): 1531-1533

SMITH, Bruce D.
1998    The Emergence of Agriculture
Scientific American Library, New York

SMITH, Earle C.
1980    Plant Remains from Guitanero Cave
In: Guitarrero Cave: Early Man in the Andes, Thomas F. Lynch, ed. P. 87-119
Academic Press, New York

TEVES Rivas, Nestor
1993    Erosion and Accretion Processes during El Niño Phenomenon of 1982-1983 and Its Relation to Previous Events
Bull. Inst. fr. études andine 22(1): 99-110

1993    Reconstructing the Paleo Records from Tropical and Subtropical Ice Cores
Bull. Inst. fr. études andine 22(1): 65-83

1984    El Niño-Southern Oscillation Events Recorded in the Stratigraphy of the Tropical Quelccaya Ice Cap, Peru
Science 226(4670): 50-53

1997    Evaluating Archaeological Hypotheses of Population Growth and Decline on the Central Coast of Peru
In: Integrating Archaeological Demography. Multidisciplinary Approaches to Prehistoric Population, Richard R. Paine, ed. P. 150-172
Center for Archaeological Investigations

WEIR, Glendon H., Robert A. BENFER, John JONES
1988    Preceramic Animal Use on the Central Coast
In: Economic Prehistory of the Central Andes, Elizabeth S. Wing and Jane C. Wheeler, eds. P. 56-94
BAR International Series 427, Oxford

WILSON, David J.
1999    Indigenous South Americans of the Past and Present: An Ecological Perspective
Westview Press, Boulder

1981    Of Maize and Men: A Critique of the Maritime Hypothesis of State Origins on the Coast of Peru
American Anthropologist 83(1): 93-120

WOODMAN, Ronald F., Antonio Mabres
1993    Formación de un Cordón Litoral en Máncora, Peru, a Raíz de el Niño de 1983
Bull. Inst. fr. études andine 22(1): 213-226

ZVELEBIL, Marek 1996    The agricultural frontier and the transition to farming in the circum-Baltic region
In: The Origins and Spread of Agriculture and Pastoralism in Eurasia. D.R. Harris, ed. P 323-345
Smithsonian Institute Press, Washington, D.C.


1 There is a confusion with the coastal chronology. One of the problems, as far as I understand, is that some coastal preceramic sites may have been contemporaneous with the Initial Period ceramic centers located in the river valleys (for example, Alto Salaverry on the coast versus Pampa de las Llamas-Moxeke in the Casma Valley or El Paraiso in the lower Chillon Valley versus La Florida in the middle Rimac Valley: see Pozorski and Pozorski 1990, 1991, 1999; Quilter 1991b). I decided to cite the chronological framework suggested by Quilter (1991a: 392). This choice is based on the assumption of lesser overlap between the Preceramic and Initial phases. For example, Piperno and Pearsalls chronology (1998: 274-275) reflects the opposite point of view. Given the growing amount of evidence in favor of later dates, the Cotton Preceramic for the sites of El Paraiso, Salinas de Chao, and Alto Salavery should be redefined.

2 all food was lost as well as what they kept buried beneath the ground as was their custom, they lost everything