Abstract
Coastal environments of the early mid-Holocene provided challenges and opportunities for agriculturalists living in the Santa Elena Peninsula, Ecuador (Santa Elena Province, formerly southwestern Guayas Province). Cores extracted from swamps in three river outflows, namely, the Río Verde/Río Zapotal drainage (Chanduy estuary), the Río Grande (Punta Carnero locality), and the Río Valdivia, provided pollen, phytolith, sedimentary, and elemental sequences relevant to documenting vegetation and agriculture. The Chanduy record documented maize and other cultigens from 3200 to 500 cal. BC, providing evidence for intensive cultivation of alluvial lands. The Punta Carnero core provided the first evidence for occupation of the peninsula during the ‘hiatus’ between the Vegas and Valdivia periods, as maize was present in a stratum dating to 4857 cal. BC. Records documented mid-Holocene sea-level stabilization, development of low-energy depositional environments, and variation in rainfall attributable to El Niño-Southern Oscillation (ENSO) by 5000 cal. BC. There was no evidence that the region was either markedly wetter or drier in the early mid-Holocene, suggesting that climate controls similar to those of today were in place.
Keywords
Introduction
Santa Elena and SW Guayas Provinces, Ecuador, have been proposed as a center of domestication of cotton (Gossypium barbadense), lima bean (Phaseolus lunatus), squash (Cucurbita ecuadorensis), and jackbean (Canavalia plagiosperma) and are part of a macro-region in which other Neotropical crops originated (Piperno and Pearsall, 1998). This proposition is based on the distribution of related wild species and primitive domesticates and on the appearance of domesticated plants in the archaeological record of the Santa Elena Peninsula during the Preceramic (9300–5300 BC) and Early Formative (4400–1450 BC) periods (Chandler-Ezell et al., 2006; Pearsall, 2003; Pearsall et al., 2004; Piperno and Stothert, 2003; Stothert et al., 2003; Zarillo et al., 2008; Zeidler, 2003; all BC/AD dates are calibrated). Maize (Zea mays) was introduced during the later part of the Preceramic and formed part of the complex of domesticated plants at the Early Formative Real Alto and Loma Alta sites. Archaeological data suggest significant transformations of societies over a period of 8000–9000 years.
The goal of this research was to explore the interplay of environmental and social factors for the development of agricultural societies of Santa Elena and SW Guayas by studying people and plants on the landscape. The approach was to extract sedimentary cores from estuaries of key drainages in the region and to study the microfossil records of past vegetation and climate. Cores capture regional patterning in vegetation – the composite picture of human and natural processes on the landscape. The earliest indications of prehistoric agriculture in forested environments are often environmental disturbances identified in paleoenvironmental records (Piperno, 1993, 2006). Cores were taken from the Chanduy estuary and the Río Valdivia in order to compare two Valdivia period populations, Real Alto and Valdivia. The Punta Carnero locality was cored to document the environmental context of the Preceramic Vegas occupation. While we were unable to explore how agricultural practices may have differed between Valdivia populations, our results documented the importance of maize and floodplain agriculture in the Chanduy region, provided the first evidence for occupation of the Santa Elena Peninsula during the ‘hiatus’ between Vegas and Valdivia, and characterized early mid-Holocene environments.
Study region
Environment
Ecuador straddles the equator on the west coast of South America (Figure 1). This is a collision-edge coast; uplift is greatest along the Peruvian coast and considerably less in the Colombian-Ecuadorian tectonic group (Araya-Vergara, 2007). The shoreline of the Santa Elena Peninsula, as broadly defined (west of Estero Salado, south of Chongón-Colonche mountains), is one of emergence (Hydrotechnics, 1974). Quaternary tablazo (uplifted Pleistocene deposit of sands, sandstone with shells, and limestone) and recent deposits are separated from the sea by sand barriers.
Map of the study region showing vegetation zones and coring localities. The primary localities are the Valdivia Valley, Punta Carnero, and Chanduy Valley.
The peninsula is composed of three uplands, a large interior lowland, and a narrow coastal plain (Hydrotechnics, 1974). The interior lowland is 100 m or less in elevation, composed of tablazo. The coastal plain lies at about 10 m elevation and is characterized by low-gradient stream channels and coastal sand bars. Large, flat salitrals develop behind sand bars when storms breach them.
The coring localities lie in the outflows of drainage basins that differ in size and water capacity. Rainfall in Ecuador decreases from north to south and from the Andes coastward (Neill and Jørgensen, 1999). The seasonal shift of the ITCZ brings precipitation into coastal Ecuador in the austral summer (Carré et al., 2012; Garreaud et al., 2009). Year-to-year and longer fluctuations in rainfall result from El Niño-Southern Oscillation (ENSO). The coasts of southern Ecuador and northern Peru experience high rainfall and flooding at the peak of ENSO (December–March). On the peninsula, rainfall is highest in the Chongón-Colonche mountains (400–700 mm/yr) and decreases to the west/southwest (100–200 mm/yr; Hydrotechnics, 1974). The Valdivia river originates in these mountains and can have surface-water flow throughout the year in its upper reaches. It has a meandering course from Loma Alta to the mouth and lies within an unconformity in this stretch.
The Río Verde and Río Zapotal join at the Chanduy estuary; this large drainage basin falls entirely within the interior lowland. The lower Río Verde runs along the Ballenita-Chanduy fault. The Río Zapotal is characterized by broad meanders within a wide alluvial plain. No drainage diversions or abandoned drainages are evident. The drainage area of the Río Las Vegas, which emerges at Punta Carnero, also falls within the interior lowland. Rivers that do not rise in the mountains are characterized by less reliable surface-water flow. The Río Las Vegas was channeled into the Velasco Ibarra reservoir in 1949, but run-off was so low that the reservoir was dry from 1958 to 1971 (Hydrotechnics, 1974).
Vegetation on the peninsula is determined by the rainfall patterns just summarized and heavily affected by land use. Neill (1999), following Harling (1979), delimits three vegetation types: coastal desert and semi-desert, savanna and deciduous forest, and semi-deciduous forest (Figure 1). In addition, mangrove swamps, largely destroyed by shrimp pond construction and disruptions of freshwater flow, once fringed lagoons. Today, the peninsula appears desertified because of deforestation, but desert and semi-desert vegetation (xerophytic woodland, thorn woodland, and arid scrub) only characterize the far west (Neill, 1999). The lower reaches of the rivers of the peninsula fall into this vegetation type. Farther inland and northward, semi-desert gives way to deciduous forest and savanna.
Prior to research reported here and that of Lim et al. (2014), paleoenvironmental records for Ecuador were limited to cores in lakes in the Andes (e.g. Brunschön and Behling, 2009; Jantz and Behling, 2012; Moy et al., 2002; Niemann and Behling, 2009; Niemann et al., 2009, 2013; Rodbell et al., 1999), Galápagos (e.g. Conroy et al., 2008; Riedinger et al., 2002), and Amazon (Piperno, 1990). In the absence of coastal data, studies suggest conflicting scenarios for early to mid-Holocene conditions in Santa Elena/SW Guayas. Any difference in the mean position of the ITCZ and cold and warm ocean currents would have a profound impact on rainy season length, abundance of rain, and vegetation in this transitional zone. It has been argued by Sandweiss and colleagues (Reitz, 2001; Reitz and Sandweiss, 2001; Rollins et al., 1986; Sandweiss et al., 1983, 1996, 2001), on the basis of remains of warm-water marine faunas at sites in coastal Peru, that the southern boundary of the warm ocean province was far to the south of its current location between 11,000 and 5400 BP, and that ENSO events were absent or significantly different from modern frequency. Such southward displacement would have brought increased precipitation to Santa Elena/SW Guayas during the late Pleistocene and early to mid-Holocene and an expansion of tropical forest southward and westward. There is some evidence from the region to support this reconstruction: remains of avifauna requiring more watered conditions at the La Carolina site (Campbell, 1982).
There are alternative explanations for these anomalous faunal assemblages. Warm-water marine faunas could have occupied localized settings with warmer water in otherwise cold water settings (DeVries and Wells, 1990) or resulted from major ENSO events. Some models propose increased rain in the Andes during the late Pleistocene: higher run-off in coastal rivers would have supported habitats favorable to faunas observed at La Carolina and Talara in northwestern Peru (Lemon and Churcher, 1961; Simpson, 1975). A lake record from the Galápagos (Colinvaux, 1972; Colinvaux and Schofield, 1976a, 1976b), which are subject to the same climatic controls as coastal Ecuador, documented aridity until 10,000 BP. Similarly, an oceanic core documented cold water conditions, correlating with aridity, until 7000 BP (Hough, 1953). A synthesis of proxy records for western South America and the Galápagos by Carré et al. (2012) concluded that during the mid-Holocene, the mean position of the ITCZ was farther north, and the South American monsoon weaker, than at present. New proxy data for coastal ocean temperatures (mollusk shell stable isotopes) also support cooler sea-surface temperatures than today during the mid-Holocene, associated with aridity along the Pacific coast from central Chile to the Galápagos.
One can argue from available data for either drier or wetter conditions in Santa Elena/SW Guayas during the early to mid-Holocene and for either improving or deteriorating conditions for human populations of the region. The only coastal record, a core from mangrove near the La Tolita site in northern Ecuador, documented continuously moist climate since 5000 BP (Lim et al., 2014). Regional pollen and phytolith sequences are needed to document vegetation and the setting for the transition to agriculture.
Cultural sequence
Occupation began in the Santa Elena Peninsula during the early Holocene with Preceramic Las Vegas (Piperno and Stothert, 2003; Stothert, 1985, 1988; Stothert et al., 2003) with traces of Pleistocene human presence extending back to 11,800 BC. Domesticated squash phytoliths were recovered from extracts directly dated to 10,000–8000 BC (Piperno and Stothert, 2003). Maize appeared by 5985 BC (Stothert et al., 2003) along with Calathea allouia (leren) and gourd (Lagenaria siceraria). Plant domestication and acceptance of maize occurred in the context of a small-scale society with broad-based subsistence.
Valdivia, the earliest ceramic-producing (Early Formative) culture of Ecuador (4400–1450 BC; all dates from Zeidler, 2003, 2008; see also Marcos and Michczynski, 1996), follows Vegas. Valdivia is divided into eight phases (Hill, 1972–1974). Machalilla (Middle Formative, 1430–830 BC) developed from Valdivia outside the study region. The Late Formative Chorrera, or Engoroy, tradition developed from Machalilla and dates from 1300 to 300 BC.
Valdivia is best known from Chanduy (Real Alto site) and the Valdivia valley (Valdivia, Loma Alta sites; Raymond, 2003; also see Damp, 1984; Lathrap et al., 1977; Marcos, 1978; Pearsall, 1979; Raymond, 1988; Zeidler, 1984). The Valdivia and Real Alto sites are near both marine and riverine habitats, but five of the eight major early Valdivia sites are located 10–20 km inland (Raymond, 2003). Site locations, and the presence of maize, arrowroot, achira, squash, cotton, and jackbean in Early Valdivia contexts (Damp et al., 1981; Damp and Pearsall, 1994; Pearsall, 2003; Zarillo et al., 2008), suggest a focus on fertile alluvial soils; site spacing suggests politically and economically independent communities (Raymond, 2003).
Middle (phase 3) and Late (phases 4 and after) Valdivia saw changes in settlement patterns in Chanduy (Raymond, 2003). Real Alto grew from a village to a town of 12.4 ha (by phase 3, 2800–2400 BC) of 1800 people and a ceremonial precinct with two mounds and then shrank to a smaller ceremonial center with a dispersed population (Clark et al., 2010; Marcos, 1978, 2003; Stothert, 2003; Zeidler, 1984). Population dispersal began during Late Valdivia (2400–1800 BC), with daughter communities founded along the Río Verde and Río Zapotal (Damp, 1984; Zeidler, 1984, 1986). Survey by Kreid and Garcia (1985; discussed in Schwarz and Raymond, 1996; Zeidler, 2008) identified Late and Terminal Valdivia sites (phase 8, 1800–1450 BC, after abandonment of Real Alto) along tributary streams and interfluvial locations. The domesticated plants described earlier continued to be used (Pearsall, 2003). Further settlement changes occurred during the Machalilla, Chorrera/Engoroy, and Guangala periods (Kreid, 1986; Zeidler, 1986).
In the Valdivia valley, after an initial growth during Early Valdivia, Loma Alta shrank to a hamlet-sized settlement and lost its concentric pattern (Raymond, 2003). The Valdivia-type site remained village-sized, never growing larger than 2–3 ha. As in Chanduy, small settlements appeared along the Valdivia river during the later part of the sequence (Schwarz and Raymond, 1996).
Archaeological research documents significant transformations of local societies of the Santa Elena region over a period of 8000–9000 years. Changes in settlement locations in terms of natural resources, shifts in settlement sizes and internal configurations, appearance (and disappearance) of ritual spaces, and expansions and contractions of population are all indications that dynamic relationships existed among the populations of the region and between populations and their resource base.
Methods
Paleoenvironmental data were obtained by coring sediment accumulations on the edges of lagoons. Lagoons were fringed by mangroves until recently, creating low-energy depositional environments suitable for coring (e.g. Neff et al., 2006b). Cores extracted from naturally accumulating sediments contain pollen that fell in the coring locality (regional pollen rain) and fluvially transported phytoliths (Pearsall, 2015; Piperno, 2006). River outflows were selected for the presence of Preceramic or Formative period sites in the watersheds. Coring was done by vibracorer, removing intact sedimentary sequences. In addition, samples were recovered by hammer-coring in La Tarea, a dry albarrada (earthen water-catchment feature) located east of Chanduy.
Each core was sampled at 2–5 cm intervals for pollen and phytoliths, charred materials were removed for accelerator mass spectrometry (AMS) dating, and sediments were described and sampled. Dates were run at the University of Arizona AMS lab. Sediments were characterized by Freidel using standard particle size analysis procedures (Day, 1965). Pollen was analyzed by Jones at the Palynology lab of Washington State University following standard methods. Phytolith analysis of core samples was carried out by Duncan at the University of Missouri Paleoethnobotany lab; Veintimilla analyzed albarrada sediments at the Archaeobotanical lab of CEAA/ESPOL. Standard procedures were followed (Pearsall, 2015). Elemental analysis was carried out by Neff. See Supplemental information, methods, for methods descriptions (available online).
Results and discussion
Dating
A total of 43 AMS dates were run on charcoal from cores (Supplemental Table 1, available online). One AMS date was run on a phytolith extract from LT-041. Age versus depth graphs were composed for each core (Figure 2; Supplemental SFigure 1, available online). Minor date reversals occurred during a period of rapid sedimentation around 5000 BC; these did not affect data interpretation. Two other reversals were clarified by further dating, which led to rejection of two dates as anomalies. In the case of Ch-045, a sample from 211 cm was rejected as too old in comparison with dated strata above and below it in correct stratigraphic order. The date on phytolith extract from LT-041, 230–235 cm, was also rejected as too old on this basis.
Depth versus age for the primary core sequences: Valdivia, VA-042; Punta Carnero, PC-051; and Chanduy, Ch-044 and Ch-045.
Valdivia locality
Sediments
VA-042 was extracted in three tubes and extended to 464 cm. Tube 3 (464–425 cm) comprised fine loamy sands that fell out of the tube. Sediments were characterized beginning 379–396 cm and were predominantly sandy through 235 cm (Figure 3). Sandy peat formed (238–235 cm), followed by silty–clayey sediments and peat (235–176 cm). Based on the closely spaced dates and appearance of peat, this was a low-energy environment with rapid sediment buildup. There was a spike in sand at 186–176 cm (4317 BC). Organic matter was increasingly abundant from 248 to 154 cm. After this, a high-energy environment was reestablished (154–127 cm, after 4317 BC until before AD 1766). The uppermost 30 cm of the core was dominated by clays and silts.
VA-042 summary results: pollen, phytoliths, and sediment and LOI data. Dot (·) indicates a rare type, <2% of sum. See Supplement SFigures 2 and 3 (available online) for complete pollen and phytolith diagrams.
Elemental analysis
Samples from 421 to 101 cm were analyzed for organic matter, calcium carbonate, and 46 elements. Principal component analysis of elemental data showed two primary sources of elements. Those contributing to PC1 (titanium and related elements) were attributable to inputs from weathered terrigenous sediments; those contributing to PC2 (calcium and related) were associated with in situ depositional processes. Larger contributions of terrigenous sediments (higher PC1) are associated with increased run-off and higher rainfall, and larger contributions from in situ processes (higher PC2) are associated with increased evaporation and lower rainfall (Neff et al., 2006a). Inputs from these sources varied over time in the coring locality (Figure 4; Supplemental SFigures 4 and 5 (available online) show details of the elemental graph; Supplemental SFigure 6 (available online), organics and carbonates). From 7000 to 5100 BC, terrigenous inputs (PC1) were low and in situ depositional processes (PC2) high, indicating relatively dry conditions. PC1 was increasingly variable after 6000 BC. Variability increased and terrigenous inputs trended higher from 5100 to 4400 BC, indicating wetter conditions. Percentage organic matter was high. Amplitude of variability decreased after 4800 BC. There were few data points from 4400 BC to post-AD 1000, but all indicated drier conditions. From ca. AD 1360 to 1560, inputs indicated wetter conditions, similar to earlier in the sequence. ENSO was the most likely source of variability in PC1 in the sequence during both wetter intervals.
PC1 and PC2 for VA-042 elemental data plotted against date. See Supplement SFigures 4 and 5 (available online) for details of the most ancient and most recent parts of the graph.
Sediment and elemental data documented changes in the depositional environment on the north side of the Valdivia valley, close to the present river outflow. A quiet water environment was established around 5000 BC, the result of slowing sea-level rise and stabilization. Decades of research along the tectonically stable Brazilian coast date establishment of modern sea level at 5700–4900 BC (Angulo et al., 2006). This environment continued until 4300 BC, when a high-energy environment was established. Further sea-level rise (mid-Holocene marine transgression) could have led to landward migration of the beach ridge that formed the early lagoon, resulting in sand deposition in the coring locality (the core is ca. 4 m elevation, just above a sea-level high stand of 2–3.5 m). After sea level fell again, a lagoon was not reestablished in this location because of another change in hydrology, such as uplift. Alternatively, a strong ENSO flood event could have breached the beach ridge and changed the river outflow to perpetuate a high-energy environment. An earlier sand spike (186–176 cm) may record a smaller flood event. Whatever the cause(s), another change in land–sea relationship occurred in the historic period, which reestablished quiet water conditions in the coring locality.
Pollen and phytoliths
Pollen was present from 420 to 180 cm (7188–4317 BC) but not in later, sandy deposits (Figure 3; Supplemental SFigure 2, available online). Strata below 5404 BC had extremely low pollen concentrations; pollen became more common as a quiet water environment was established, but concentrations were never high. Mangroves were present throughout the sequence, with highest pollen concentrations at 4317 BC. Sedge, grass, and Canna-type pollen, among other herbaceous taxa, occurred, as did Moraceae, Fabaceae, and some other arboreal taxa. Potential economics included Canna-type, Coccoloba, and Sapotaceae. Phytoliths were present from 420 to 180 cm, but sparse prior to 230 cm (Figure 3; Supplemental SFigure 3, available online). Mangroves do not produce diagnostic phytoliths but contributed to small rugulose sphere, schlerid, and cystolith counts. Abundant sponge spicules and diatoms documented a wet habitat. Forest trees were present, including the useful economics palm, Annonaceae, Chrysobalanaceae, and Ebenaceae. Marantaceae phytoliths were identified. Phytoliths were recovered from sandier strata dating to 4317 BC–AD 1018; assemblages were similar.
Mangroves grew in the Valdivia river outflow prior to and during establishment of an early lagoon. This was likely a small, river mouth swamp, rather than an elongated lagoon like today (Roy et al., 1994). Today, the Valdivia river watershed falls within the deciduous forest zone (Figure 1). As indicated by the elemental study, from 5000 to 4300 BC, rainfall was similar to historic levels and would have supported this formation. However, this is not reflected in the pollen record. Pollen of animal-pollinated taxa is often found in low quantities in tropical forest cores (Colinvaux et al., 1999). Since such pollen grains are not carried by air, they are likely deposited by water flow within the local catchment (e.g. from dropped flowers). The VA-042 pollen record documented few taxa not represented in the xerophytic vegetation zone, but rather assemblages were characterized by xerophytic taxa (Supplemental Table 2, available online). Compared with Valdivia, the Punta Carnero sequence, discussed below, appears moister: Cecropia, Celtis, Machaerium-type, and Morella are all present in Punta Carnero and not in Valdivia, for instance. Both these cores have the same richness of arboreal taxa, but pollen concentrations are much lower in VA-042. Given this, we suspect the Valdivia pollen record is not representative of regional vegetation. This interpretation is supported by phytoliths, which documented deciduous forest taxa (e.g. Annonaceae, Arecaceae, Bombacaceae, Ebenaceae, and Zingiberales). Nonetheless, it is apparent that xerophytic forest extended northward into the lower Valdivia river valley. With the exception of seasonal grasses, phytolith production is not high among taxa of this formation. More phytolith producers occur in deciduous forest, including palms, but potential phytolith influx is still low (MU lab documents and comparative soil samples). VA-042 sediments were acidic (mean pH: 3.1, range: 2.2–4.5); low phytolith counts are not attributable to dissolution.
Useful plants were identified in sediments of the early lagoon, but no clear economic signal: palms, Canna, and Marantaceae are not documented in the xerophytic zone but occur in moister areas in Guayas (Supplemental Table 2, available online). Note, however, that pollen of Amaranthaceae, grasses, and low-spine Asteraceae can indicate human disturbance. The habitat would have been attractive to people, and an older Valdivia site, Loma Alta, is located 15 km inland.
In summary, a high-energy environment existed in the VA-042 locality at 7000 BC, and rainfall was low but sufficient to support mangrove and coastal forest (Table 1). A lagoon formed and disappeared prior to the first documented appearance of Valdivia peoples in the area at 4000 BC. In comparison with earlier in time, conditions were wetter from 5000 to 4300 BC. This is consistent with the trend at 5000 BC for the southern low-latitude tropics: wetter than at 7000 BC (Bush and Gosling, 2012). In the Valdivia valley, moisture levels were not markedly different than in the historic period. ENSO events (moisture or sand spikes) occurred, but less frequently than historically. Any settlements located within the lower valley and oriented to this lagoon would likely have been lost when it failed. The Valdivia site was founded on a low spur on the southern edge of the valley, where it overlooked riverine, lagoon, and marine habitats and would not have been prone to flooding. Unfortunately, our core did not capture human–landscape interactions during the Valdivia period in the lower valley.
Summary of environmental indicators for the Valdivia locality.
xero.: xerophytic forest; decid.: deciduous forest; organ.: organic matter; moist.: moisture.
Punta Carnero locality
Sediments
PC-051 had four sediment zones (Figure 5). Sedimentation was dominated by sand from 555 to 418 cm, documenting rapid water flow from 12,684 to the interval dated 4613–4857 BC. Sedimentation changed between 402 and 387 cm, with sand dropping to 60% and then to 36–38%. These quiet water conditions fell within the 4613–4857 BC interval. This was followed by high sand from 275 to 201 cm and then a return to fine sediments from 201 to 66 cm. Quiet water conditions at the top of the core dated to 3733 BC and after. Loss-on-ignition (LOI) data paralleled the sedimentary record, with more abundant organic matter in quiet water settings, especially after 3733 BC. Sediments from another core, PC-041, not analyzed in detail, also showed an early quiet water interval overlaid by black sand with crushed shell.
PC-051 summary results: pollen, phytoliths, and sediment and LOI data. Dot (·) indicates a rare type, <2% of sum. See Supplement SFigures 7 and 8 (available online) for complete pollen and phytolith diagrams.
High-energy conditions during the earliest section of PC-051 documented lower sea level; the coring locality was located upstream in relationship to the coast. The first episode of quiet water represented a slowdown in sea-level rise, with increased sedimentation and lagoon formation. This depositional environment existed at 4618–4857 BC, during the occupation ‘hiatus’ between Las Vegas (9300–5300 BC) and Valdivia (4400–1450 BC). Subsequent sea/land changes would have obliterated human settlements around this lagoon. The interval was terminated by sand and crushed shell deposition that is not dated. This represented a subsequent high stand of sea-level and beach ridge development or a major El Niño. Conditions from 3733 BC represented the low-energy depositional situation seen today. Valdivia 1 sites, dated 3400–3125 BC, are known from the outer peninsula region (Zeidler, 2003).
Pollen and phytoliths
Pollen was absent in the lowest meter of PC-051, representing the interval of rapid water flow (Figure 5; Supplemental SFigure 7, available online). There were two subsequent pollen zones: 440–260 cm and 260–100 cm. These differed in pollen concentration; pollen all but disappeared at 260 cm and was sparse in the upper zone. In both zones, mangrove dominated but varied relative to pollen of herbaceous taxa, such as Amaranthaceae, grass, Asteraceae, and Cyperaceae. Non-mangrove arboreal taxa occurred throughout the sequence, as did Canna-type. The later high-energy interval (275–201 cm, after 4613–4857 BC) fell across both pollen zones.
Phytolith deposition began around 400 cm (Figure 5; Supplemental SFigure 8, available online). Phytoliths were sparse throughout, but like pollen were more abundant in strata corresponding to the lower pollen zone. Diatoms and sponge spicules were limited to the lower zone, where sedge (Scirpus) also occurred. Fewer taxa were represented in phytolith assemblages later in time. Wild squash, palm, and arrowroot family were present. Three samples in the lower zone with relatively abundant phytoliths documented bamboos and other grasses, in addition to wild squash, palm, and arrowroot family; maize occurred in one stratum, 350 cm (4857 BC). Phytoliths were also examined from another core, PC-041; results were similar (Supplemental SFigure 9, available online). Arrowroot family, palm, and bean (cf. Phaseolus) were documented.
The Punta Carnero watershed falls today within the xerophytic zone (Figure 1). Considering the arboreal pollen assemblage as a whole, there is strong evidence that deciduous (moister) forest extended westward into the watershed during the early mid-Holocene and after (Supplemental Table 2, available online). Animal-pollinated taxa not documented in the xerophytic zone occurred (Machaerium-type, Sapotaceae, and Spondias) as did pollen of xerophytic zone trees (Bursera, Cassia, Coccoloba, Erythrina, Zanthoxylum-type). This mixed suite of local arboreal indicators documented that deciduous forest extended farther westward than today but did not replace xerophytic forests. Higher LOI data during this interval also indicated moister conditions. Taxa represented by phytoliths are consistent with deciduous forest, but overall phytolith influx was low, with counts characteristic of xerophytic forest (MU lab documents). Sediments were acidic (mean pH: 3.8, range: 3–7.4).
Canna, palm, and arrowroot family do not occur in the xerophytic zone but are documented in Guayas Province (i.e. in moister forests; Jørgensen and León-Yánez, 1999; Svenson, 1946a, 1946b). Given the evidence for deciduous forest, it is possible that these plants grew wild. However, the presence of maize at 4857 BC raises the possibility that they were maintained under cultivation. Other potential economic plants identified in the Punta Carnero record included Coccoloba, cf. Phaseolus, Spondias, Moraceae, and Sapotaceae. Pollen of Amaranthaceae and Asteraceae also suggested human disturbance consistent with cultivation. The lack of maize pollen indicated fields were located some distance inland from the mangrove. The Punta Carnero record provides the first evidence for occupation of the peninsula during the ‘hiatus’ between Las Vegas and Valdivia.
In summary, the Punta Carnero sequences extended into the early Holocene, when the locality was a high-energy environment (Table 2). Sea-level stabilization led to the devel-opment of a mangrove-fringed lagoon; rainfall was sufficient to support deciduous forest and plant cultivation, but xerophytic forest also occurred. Subsequent sea-level rise or ENSO flooding after 4600/4800 BC led to a poorer environment for pollen preservation and phytolith deposition, but mangrove and forests still grew close by. Re-establishment of a low-energy environment after 3700 BC did not result in pollen and phytolith concentrations as high as those seen earlier, and indicators of freshwater swamp disappeared. Organic matter was abundant. In combination, these data suggest that conditions were drier, but not extremely so, when Valdivia populations are documented in the region.
Summary of environmental indicators for the Punta Carnero locality.
xero.: xerophytic forest; decid.: deciduous forest; organ.: organic matter; moist.: moisture.
Chanduy locality
Sediments
Three cores were taken below the confluence of the Río Verde and Río Zapotal, on a terrace above the tidal zone (Figure 1). Ch-044 began at 300 cm and extended to 512 cm. It was taken in two tubes; due to expansion of sediments, there was 37 cm overlap between tubes, but sediments were continuous. Deposits dated to 6063–4115 BC and were characterized by increasing proportions of silts over time and highly variable sand contributions (Figure 6). Sediments were gley, with the exception of a stratum of black silt loam and peat at 461–466 cm. A period of rapid sedimentation occurred at 5000 BC. LOI indicated two strata with abundant organic matter, both in this section of the core, and relatively higher levels of organics above 417 cm.
Sediment and LOI data (percentage) from Ch-044 and Ch-051.
Ch-045 extended from 84 to 450 cm. Sediments were not analyzed. Field descriptions indicated three depositional zones. From 450 to 209 cm, sediments were gleyed (clay loam, silty clay loam, and silt loam). No descriptions were available from 354 to 287 cm (after 1540 BC to just before 502 BC). Rapid sedimentation occurred at 1567–1540 BC, depositing 31 cm of clay to silty clay loam. The second depositional zone extended from 209 to 153 cm and was characterized by medium sand followed by fine and medium sand and fine sandy loam. The third zone extended from 153 to 84 cm and consisted of gleyed loam. The top of the first depositional zone and the entire second and third zones dated to after 502 BC (date from 278 to 284 cm), documenting deposition of 2 m of sediment in 226 years.
Ch-051 extended from 45 to 685 cm, dating from 4416 to 3712 BC to after 2747 BC. Sand dominated through 300 cm, with the exception of a stratum of organic-rich silty sand at 660–685 cm (Figure 6). Sediments were less sandy from 300 to 130 cm, with variable contributions of silts and clays. This interval dated to 2700 BC and represented a rapid depositional situation, associated with abundant organic matter from 181 to 130 cm. The upper section of the core, 130–45 cm, was characterized by alternating sandy and silty sediments and lower LOI.
Chanduy cores documented four episodes in which fine-grained sediments built up in a low-energy setting: 5000 BC (Ch-044), 2700 BC (Ch-051), 1500 BC (Ch-045), and 500 BC (Ch-045). The earliest episode, representing slowing of sea-level rise and stabilization, was punctuated by periodic spikes in sand resulting from ENSO flooding. The later depositional episodes are discussed below.
Pollen and phytoliths
Ch-044 was processed only for phytoliths but was unproductive; Ch-051 was not studied. In core Ch-045, pollen was well preserved and present in relatively high concentrations from the base to 275 cm (3236–502 BC; Figure 7). While some mangrove pollen was found, the locality was a freshwater habitat, such as a slough inland from the mangrove, during this interval. The pollen record was dominated by herbs and cultigens, including maize, which was abundant in the basal level of the core (6% by count) and present throughout the sequence in quantities suggesting that fields were located nearby. In addition to maize, Canna-type, Manihot (manioc genus), and Malvaceae (cf. Gossypium, cotton genus) occurred. High percentages of weeds, including Amaranthaceae, Polygonaceae, grasses, and low-spine Asteraceae (ragweeds), indicated intensive human activity in the immediate vicinity of the locality. Arboreal taxa were present in low numbers (<20% by count, excluding mangroves), including potential economics that rarely travel any distance (Annona, Coccoloba, Malpighiaceae, Moraceae, Spondias, and Sapotaceae). Contribution of arboreal pollen varied over time but, in general, declined relative to herbs and cultigens. Species richness also declined, with just over half as many tree taxa present in later strata compared with earlier.
Ch-045 percentage pollen diagram, based on 200 counts. Dot (•) indicates a rare type, <2% of sum. Very rare types are omitted from the graph: Borreria, Canna-type, Manihot, Acacia, Cordia, and Unknown B.
Phytolith concentrations were low from the base of Ch-045 to 379 cm (1567 BC; Figure 8). Phytolith influx then ceased until after 502 BC, when phytoliths were present in variable quantities. The early record documented weedy taxa, including Asteraceae and grasses (Chloridoid and Panicoid), trees, and economics, including palm, Marantaceae (arrowroot family), and Canna. No maize phytoliths were recovered. The presence of sponge spicules, diatoms, and sedges confirmed a wet environment. The same mix of taxa occurred in the upper section of the core (500–276 BC), with the addition of Heliconia (bird-of-paradise).
Ch-045 count phytolith diagram.
Pollen indicated that the coring locality – on the landward side, but within the historically documented Chanduy mangrove – was not mangrove but freshwater swamp from 3236 to 502 BC. Mangrove was likely located nearer the outflow (Figure 1) and did not extend inland to the confluence and the Río Zapotal as in recent times, indicating reduced brackish water incursion. This could have resulted from higher or more regular freshwater influx, although LOI did not indicate continuously wetter conditions. Alternatively, subsidence of the confluence area may have resulted in later mangrove expansion. The lower part of the Chanduy drainage falls today within the xerophytic zone, while the upper is deciduous forest (Figure 1). Pollen documented a mixed suite of local arboreal indicators from both zones (Supplemental Table 2, available online). Phytoliths also documented moister forest taxa. Deciduous forest likely extended closer to the coast in the past. The absence of phytoliths from 1567 to 500 BC may indicate reduced run-off and drier conditions, but influx was never high. Sediments were alkaline (mean pH: 8.6, range: 8.1–9); phytolith dissolution may have contributed to low counts, especially in older strata and those with pH values close to 9, when opal silica solubility begins to increase rapidly (Karkanas, 2010). Over time, fewer kinds of trees were represented in the pollen record. Richness declined equally among taxa from xerophytic and deciduous forest, indicating a more generally open landscape.
The Valdivia period Real Alto site is located 4 km up the Río Verde from Chanduy on a low ridge above the river (Figure 1). Pearsall (1979) argued that river alluvium within 5 km of Real Alto would have been prime agricultural land for the community. A 5-km catchment includes the Chanduy lagoon, Río Verde and Río Zapotal confluence, extensive alluvial deposits up the Río Zapotal, and pockets of alluvium along the Río Verde. Support for agriculture oriented toward naturally watered alluvial terraces came from study of fuel wood charcoal from Real Alto (Pearsall, 1979, 1983). The study documented a focus on legume trees of the xerophytic zone throughout the occupation of Real Alto, with no indication that inland deciduous forests were cut. From this was drawn the inference that alluvial terraces, not uplands, were the focus of agriculture. Ch-045 provides the first direct evidence that Valdivia peoples grew crops in alluvial settings, very near watercourses, by at least 3236 BC. At that time, Real Alto was still a small community, but farmers were already planting maize and other crops near the Chanduy swamp. Real Alto is the only site of this time period in the area.
Ch-045 also documented the stability of alluvium-based agriculture in this transitional environment: the Chanduy area was farmed continuously from 3236 to 502 BC, through most of the Formative (4400–300 BC). This is a minimum time estimate, as conditions in the locality did not preserve pollen after 500 BC, and the core did not extend earlier. Alluvial lands would have existed in the area from at least 5000 BC, when sea level stabilized. Furthermore, documentation of maize in the core until 500 BC demonstrated that Machalilla and Chorrera populations continued to farm in the confluence area. While there are fewer settlement pattern data for these periods, sites of both traditions are known, mostly in riverine locations.
Water management also contributed to successful agriculture on the Santa Elena Peninsula, as indicated by numerous earthen water-catchment features or albarradas (Marcos, 2004; Stothert, 1995). Albarradas are U-shaped embankments oriented to catch rainy season run-off. Excavations date albarrada construction as early as Terminal Valdivia (albarrada San Pablo; Marcos and Tobar, 2004) and Chorrera (albarrada Achallán; Stothert, 1995), with continued use and new construction throughout later prehistory, Colonial, and modern eras (Marcos and Tobar, 2004). Several albarradas are located east of Chanduy; La Tarea (LT-041) was selected for hammer-coring and phytolith analysis (Supplemental SFigure 10, available online). Veintimilla (2004), who analyzed phytoliths from albarradas studied by Marcos and colleagues, identified maize phytoliths in the upper 2 m of LT-041. Unfortunately, all dates were stratigraphically lower and pre-dated known occupation of the region. We estimate that sediments in the pre-maize section of LT-041 accumulated somewhat before and during the early mid-Holocene (Supplemental information, La Tarea dating, available online). Phytoliths counts were low, the overall assemblages characteristic of xerophytic forest. While we could not date the maize remains, the LT-041 phytolith assemblage further documented the strong association of this crop and albarradas: maize occurred in all seven albarradas excavated by Marcos and colleagues, regardless of time period (Veintimilla, 2004). Canna, squash, and bean phytoliths occurred less often.
Sustained agriculture in the Chanduy region raises the issue of anthropogenic influences in the coring records. As mentioned earlier, cores documented four episodes in which fine-grained sediments built up rapidly in the locality: 5000 BC (sea-level stabilization), 2700 BC, 1500 BC, and 500 BC (Table 3). The 2700 BC episode fell near the beginning of Valdivia phase 3 (2800–2400 BC), when Real Alto reached its largest size and population. Maize, which requires fertile, well-watered growing conditions, was commonly used (Pearsall et al., 2004). Well-watered lands within the town’s catchment area would have been intensely used to support this population concentration, as the wood study indicated that uplands were not the focus of farming. Intensive use of alluvial terraces could have resulted in increased erosion and sedimentation. Higher rainfall also contributes to erosion, and LOI (higher organic matter) indicated relatively moist conditions at this time. Pollen and phytoliths were somewhat more abundant but otherwise gave little indication that conditions were wetter at 2700 BC than earlier. This is consistent with the overall climatic trend at 3000 BC for the southern low-latitude tropics: moist, unchanged from 5000 BC (Bush and Gosling, 2012).
Summary of environmental indicators for the Chanduy locality.
xero.: xerophytic forest; decid.: deciduous forest; organ.: organic matter; concent.: concentration.
No microfossil evidence from Ch-044 and Ch-051.
Sediments from 3200 BC to sometime after 500 BC are gleyed, indicating a low-energy environment. Sedimentation was relatively rapid at 1500 BC and after 500 BC. Sediment sizes were more variable after 500 BC.
Rapid sedimentation at 1500 BC occurred just before the transition from Valdivia to Machalilla. Settlement pattern data indicate significant changes: Late and Terminal Valdivia and Machalilla populations both lived in small, dispersed settlements, but while Valdivia populations used both interfluvial and riverine habitats, interfluvial settlement declined during Machalilla and the subsequent Chorrera period (Kreid, 1986). If founding of settlements in interfluvial areas was accompanied by cultivation along tributary streams and uplands, this represented agricultural expansion during Late and Terminal Valdivia. Such expansion could have contributed to increased erosion and sedimentation. Pollen concentration peaked at 1500 BC, indicating consistently moist conditions.
The subsequent contraction back to riverine-focused settlement between 1430 and 300 BC (Machalilla and Chorrera periods) suggests that upland cultivation became unsupportable or unnecessary. The Ch-045 pollen record indicated increasingly open conditions, with reduced richness of forest taxa. This, with falling pollen concentrations, is consistent with drier conditions but could have an anthropogenic component. Cessation of phytolith influx supports reduced moisture, but high pH may have affected preservation. The overall climatic trend at 1000 BC for the southern low-latitude tropics is drier to unchanged (Bush and Gosling, 2012).
Rapid sedimentation at 500 BC fell before the transition between Chorrera and the Regional Developmental culture, Guangala. Guangala settlements were established throughout the interfluvial area, as well as along upper stream courses (Kreid, 1986), suggesting upland cultivation was supportable or necessary and resulted in increased erosion. Pollen continued to document an open, anthropogenic landscape. Following this episode, phytolith deposition resumed, albeit at low levels, suggesting somewhat moister conditions.
In summary, episodes of buildup of fine-grained sediments in Chanduy did not have a single cause. Our data suggest that an interplay of environmental influences – sea level, rainfall – and social/political factors – intensification, expansion of agriculture – led to changes in sedimentation. In most cases, this resulted in enhanced pollen preservation. Phytolith representation was disappointing. The forests of Santa Elena are dominated by poor phytolith producers; there were few phytoliths to become deposited from run-off from xerophytic forest and mangrove, somewhat more from deciduous forest. Production does not explain low phytolith numbers from seasonal grasses or absence of maize phytoliths in the Chanduy sequence. While high pH may have impacted phytolith preservation there (the single sample with a 200 count had the lowest pH (8.1) and was relatively young (post-500 BC)), this does not explain low grass counts at Punta Carnero and Valdivia, where sediments were acidic. We do not believe that processing was at fault (Supplemental information, phytolith processing experiments, available online). Mangrove cores from Pacific coastal Guatemala, where seasonal rainfall today averages 2000–3000 mm, consistently produced good phytolith counts (Neff et al., 2006b). Our working hypothesis is that seasonal rainfall in the range characteristic of the Santa Elena Peninsula today (100–700 mm) results in low phytolith influx into coastal swamps.
Conclusion
Interaction of geology, tectonic setting, topography, climate, sea-level rise, and sediment supply, among other factors, control evolutionary processes in lagoons (Cooper, 1994). Establishment of quiet water settings in which mangroves or freshwater swamps developed in the early mid-Holocene on the Santa Elena Peninsula created the potential for deposition of pollen and phytoliths from local and regional vegetation, but understanding the processes that created these settings and altered microfossil assemblages through time is challenging.
Cores penetrated early Holocene sediments at Punta Carnero, but pollen and phytoliths were not deposited/preserved in what was then a rapid water setting. Our best data on early Holocene environments remain archaeological: continuous settlement by Las Vegas peoples, who farmed, fished, hunted, and gathered on the peninsula from 9300 to 5300 BC (Piperno and Stothert, 2003; Stothert, 1985, 1988; Stothert et al., 2003). Growing squash, gourd, leren, and maize required seasonal rainfall at least as abundant and regular as today. Just to the north, in the Valdivia valley, our research documented that mangrove swamp and xerophytic and deciduous forests existed prior to establishment of modern sea level (back to 7188 BC), another indication that early Holocene climate was seasonally moist. Las Vegas agriculture developed under these favorable conditions.
By the early mid-Holocene, coring localities had become quiet water environments, preserving a record of local and regional vegetation and human activities. The Punta Carnero core documented maize at 4857 BC, establishing occupation of the peninsula during the ‘hiatus’ between Vegas and Valdivia. Other potential economic plants identified included Coccoloba, Spondias, Moraceae, Sapotaceae, palm (arboreal), Canna, squash, arrowroot family, and bean (cf. Phaseolus). Settlements oriented to this early lagoon were likely destroyed when it failed. This perhaps also happened in the Valdivia valley, where a lagoon was established and destroyed before founding of the Valdivia site at 4000 BC. A similar suite of useful plants was documented in VA-042: palm, Coccoloba, Sapotaceae, Annonaceae, Chrysobalanaceae, Ebenaceae (arboreal), Canna, and arrowroot family.
Pollen and phytoliths from all sequences identified a mixture of trees from xerophytic and deciduous forests, establishing that moister forest extended farther westward than today in the early mid-Holocene but did not replace the drier formation. Elemental data from VA-042 indicated moisture levels similar to the historic period. Taken together, results showed the Santa Elena region to be neither markedly wetter nor drier in the early mid-Holocene, suggesting that climate controls similar to today were in place by at least 5000 BC. This extends by two millennia the evidence for climatic stability documented for northern Ecuador by Lim et al. (2014) and argues against southward displacement of the ITCZ until 3400 BC, as argued by Sandweiss and colleagues (Reitz, 2001; Reitz and Sandweiss, 2001; Rollins et al., 1986; Sandweiss et al., 1983, 1996, 2001). Variability in moisture and evidence for flood events (sand spikes) attributable to ENSO are documented from 5000 BC. Our data are not high resolution, but ENSO events appear to be less frequent 5000–4000 BC than during the historic period.
Maize was ubiquitous at Real Alto by Valdivia 3 (2800–2400 BC; Pearsall et al., 2004). Coring data from Chanduy have now confirmed a Formative period agricultural system of maize cultivation oriented to alluvial lands, possibly in association with water-catchment features that extended the cropping season (i.e. through multi-cropping of annuals, like maize, or cultivation of root crops with long growing seasons, like leren; Pearsall, 2014). Phytolith analysis of albarrada La Tarea provided further support for the relationship between these water-catchment features and maize cultivation previously documented by Veintimilla (2004). Core results have added to the growing body of evidence for the mixed, tropical forest-nature of prehistoric agriculture in coastal Ecuador, documenting root crops (Canna, Manihot, and arrowroot family), arboreal resources (Annona, Coccoloba, Malpighiaceae, Moraceae, Spondias, Sapotaceae, and palm), maize, and other useful plants like cotton (cf. Gossypium) and bird-of-paradise. The results of this study documented intensive agriculture in alluvial lands in the vicinity of Real Alto, provided evidence for human occupation of the Santa Elena during the Vegas/Valdivia ‘hiatus’, and provided evidence that early mid-Holocene climate controls operated similarly to today’s.
Footnotes
Supplemental information,methods
Sediments were characterized by Freidel using standard particle size analysis procedures (Day, 1965). Air-dried samples were pretreated with dilute sodium hypochlorite to remove organic grain coatings and content, dispersed with sodium hexametaphosphate, then ultrasonified for one hour in preparation for hydrometer analysis. After the hydrometer procedure sand fractions were separated from fines by wet-sieving, then oven-dried and weighed. Loss-on-ignition was measured on each sample (Davies, 1974). Samples were oven-dried and weighed, then heated at 420 C for 5 hours. Samples were then cooled and weighed, the percent weight loss representing the amount of organic matter.
Pollen was analyzed by Jones at the Palynology lab of Washington State University following standard methods. Samples were treated with dilute HCl to remove carbonates and sieved and swirled to remove larger inorganic particles. Small silicates were removed using concentrated HF, then samples were washed in 1% KOH to remove humates. Acetolysis treatment (Erdtman, 1960) followed, to remove unwanted organic materials. Heavy liquid flotation was used to recover pollen. Identifications were made using a Nikon compound stereomicroscope at 400 x magnification; where possible 200-grain counts were made for each sample (Barkley, 1934; Bryant and Hall, 1993). Results were displayed using Tiliagraph (Grimm, 1988). Zonation was calculated by constrained sum of squares analysis.
Phytolith analysis of core samples was carried out by Duncan at the University of Missouri Paleoethnobotany lab; Veintimilla analyzed albarrada sediments at the Archaeobotanical lab of CEAA/ESPOL. Standard procedures were followed (Pearsall, 2015). Dried samples were first treated with dilute HCl followed by mixed strong acid (concentrated HCl and HNO3) to remove carbonates and certain oxides. Organic matter was removed using bleach and H2O2 (35%). Samples were dispersed using 0.1% Na2H2EDTA, followed by clay removal by gravity sedimentation. Phytoliths were floated from samples by heavy liquid. Dried extracts were mounted in Canada balsam and slides examined until a 200-count was reached, or the entire slide was scanned. Identifications were made using the MU phytolith comparative collection (http://phytolith.missouri.edu). Results were displayed in Tiliagraph.
Elemental analysis was carried out by Neff. Organic carbon and calcium carbonate percentages were determined by weight loss-on-ignition (LOI). Elemental concentrations were measured by laser ablation-ICP-MS of samples after LOI.
Citations: see article reference list.
Acknowledgements
Publication figures were composed by Howard Wilson from drafts provided by the authors. Meghann O’Brien and Michael DeLoughery assisted with fieldwork. We thank Vaughn Bryant, Jr for helpful comments on the manuscript and assistance in composing Supplement Table 2 (available online).
Funding
This research was supported by grants to Pearsall from the National Science Foundation (People, Plants, and Landscapes in Prehistoric Ecuador: A Look at the Causes and Consequences of Agriculture, NSF 0407742, 0509775).