‘Mayacene’ floodplain and wetland formation in the Rio Bravo Watershed of northwestern Belize

Abstract

This is the first article to characterize the soil and fluvial geomorphology of the Rio Bravo’s fluviokarst watershed in the Rio Bravo Conservation and Management Area, northwestern Belize. Although the watershed has had little-altered tropical forest cover since c. 1000 BP, humans inhabited it for millennia, especially during the Maya Preclassic and Classic, c. 3000–1000 BP. We studied soils and floodplain formation in four excavation transects in the Rio Bravo to understand long-term human impacts on this watershed. Archaic to Preclassic (c. 3000–1700 BP) sedimentation rates ranged from 0.82 mm yr⁻¹ at Chawak But’o’ob to 1.5 mm yr⁻¹ on the Gran Cacao floodplain. The late Preclassic through Classic (c. 2300–1000 BP) rates rose 0.98–2.03 mm yr⁻¹, and the Classic (c. 1700–1000 BP) rates ranged from 1 mm yr⁻¹ to as high as 9.12 and 16.27 mm yr⁻¹ at ancient Maya wetland field sites. Post Classic rates dropped back in the one dated profile, and the well-developed topsoils indicate long-term surface stability. Older soils at the edges and higher islands of the valley had more vertic features and full Vertisols, whereas Vertic Mollisols and Aquerts have formed in younger sediments. We also present new evidence for late Classic Maya wetland field formation at Chawak But’o’ob, which shows field raising with canalization in this wetland of low ionic water. All the soil profiles with dating and stable carbon isotope evidence exhibited increased δ¹³C in the profiles through the Classic period sediments, although some were equivocal. The two wetland field δ¹³C profiles through the Classic period sediments increase by c. 6‰ at Chawak But’o’ob and 3‰ at the Birds of Paradise (BOP) Field center, although earlier BOP profiles increased by as much as 7‰. Hence, this watershed exhibits three large diachronic shifts: from lower to higher and again to lower deposition over pre-Maya, Maya, and post-Maya times. These changes along with earlier evidence for ancient intensive agriculture from 3000 and 1000 BP lie sandwiched between the ancient and contemporary little-altered tropical forest.

Keywords

aggradation ancient Maya wetland agriculture floodplains geoarchaeology Maya Lowlands paleosols Vertisols

Introduction

The Rio Bravo is a little studied fluvial karst watershed in northwestern Belize (Figure 1). Today, tropical forest and some savanna cover the watershed, but evidence for human impacts date to the Archaic period in the middle Holocene, at least 5000 yr BP (Beach et al., in press; Luzzadder-Beach and Beach, 2009). Beach et al. (2003) argued that erosion and sedimentation started early here, in the Preclassic, and other studies showed this occurred elsewhere in the Maya Lowlands (Anselmetti et al., 2007; Beach et al., 2009, in press). This paper continues this line of research focusing on fluvial change in the watershed from floodplain aggradation and wetland field formation based on a series of excavations through its main valley. While geomorphological research is de rigueur in many places, it is still in its infancy in this region and much of the tropics (Luzzadder-Beach and Beach, 2009).

Figure 1.

Map of the Rio Bravo watershed above a graph of the river’s longitudinal profile and water content of dissolved sulfate and calcium in ppm.

Other papers have synthesized the large-scale changes of the Maya period subsumed under the concept of the early Anthropocene or, for the Maya world, the ‘Mayacene’, characterizing ancient human impacts from c. 3000 to 1000 BP (Beach et al., in press; Kennett and Beach, 2013). In this paper, we look for the Longue Dureé structures and markers of landscape change (Braudel, 2011) within the Rio Bravo watershed. Here, we make the first attempt to outline sediment movement over time along four excavation transects through a footslope, floodplains, and terraces and to compare these with earlier soil studies in three other parts of the watershed. We also make the first assessment of a new zone of wetland field formation at Chawak But’o’ob (heretofore Chawak). Karl Butzer has shown over decades of scholarship on four continents that geomorphic and geoarchaeological systems are complex puzzles and their triggers multifaceted (Butzer, 1959, 2005; Butzer et al., 2008). For the Rio Bravo, we start to solve a complex puzzle of floodplain deposition, soil formation, and ancient wetland field use within the valley’s broader geomorphic evolution.

Environments

This northwestern Belize watershed lies almost entirely within the Rio Bravo Conservation and Management Area (RBCMA) (Figure 1), managed by the Programme for Belize, a Belizean non-profit, conservation organization (Nations, 2006: 243). The vegetation is largely old growth, subtropical moist forest (Hartshorn et al., 1984), with a small tract of wetland savanna (Bridgewater et al., 2002) along its Cacao Creek tributary and a small tract of pasture. The forest types include upland, mesic, riparian, bajo scrub, and transitional, with the riparian forest having a canopy of up to 30 m and a tree diversity of up to 59 tree species (⩾ 10 cm dbh) ha⁻¹ (Brokaw et al., 1993).

The Rio Bravo is a transboundary basin that rises from the eastern edge of Guatemala’s Petén and drains over northwestern Belize through aquifers and valleys structured by normal faults. The limestone bedrock is covered by sascab, a saprolitic and marly limestone. The highest point of the watershed is c. 300 m above sea level (m a.s.l.), and its base level confluence with the Booth’s River is at c. 15 m a.s.l. The graben is a tilted normal fault with its deepest section adjacent to the Rio Bravo escarpment, and the graben gradually rises eastward away from the Rio Bravo channel up to the Booth’s River escarpment (Figure 1). Upstream from the Rio Bravo graben, the tributaries collect from another structural valley and the La Lucha escarpment.

The underlying geological structure is the Yucatan Platform, composed of Cretaceous and Tertiary limestone and evaporite (Hartshorn et al., 1988; Marshall, 2007; Perry et al., 2009). Some of these layers are terminal Cretaceous in age and formed from Chicxulub bolide impact ejecta (King et al., 2004), which is one source of the region’s sulfate-rich groundwater. Perry et al. (2009) contended the large polje sinks here mark the zone of maximal sulfur-rich ejecta, since sulfur minerals like gypsum and celestite dissolve preferentially in groundwater flows. The platform descends along a series of Pliocene-induced normal faults (30–80 m high) from the central Petén at about 400 m a.s.l. to the barrier reef of Belize. The escarpments are composed of denser limestone with higher quantities of Mg (Brennan et al., 2013) weathered into steep karst hills and sinkholes, the largest of which are large, internally drained sinks regionally called bajos (Dunning et al., 2002). These depressions have intermittent water, but the Rio Bravo is perennial albeit with widely varying flows. Dissolution, infiltration, and slow mass wasting are dominant geomorphic processes. Under typical forested conditions, most runoff infiltrates into soils and bedrock, but with deforestation, runoff rises, infiltration falls, and erosion and deposition accelerate (Beach, 1998; Beach et al., 2008).

The Maya region has few fluvial studies (Beach et al., 2008). One recent study by Solís-Castillo et al. (2013) considered soils and human interactions on the Pleistocene and Holocene terraces of the lower Usumacinta River, Mexico’s largest. This research studied a chronosequence across the terraces, pointing out that Vertisols, and soils with Vertic features, were the oldest and often connected to Preclassic era artifacts.

There are also few soil studies in this region, but the small-scale soil surveys (King et al., 1992: 221) map the Yaxa soil suite and Yalbac subsuite in the region, based on limited field sites and laboratory analyses. These are mainly Lithic and Vertic Rendolls on the uplands and are fertile, calcareous, black, clay soils with smectitic or vermiculitic minerals largely formed in situ from impurities in the limestone such as chert and from volcanic and Saharan sources (Bautista et al., 2011; Cabadas et al., 2010; Tankersley et al., 2011). Stable geomorphic surfaces have development of expansive clays and produce Vertisols (King et al., 1992: 223). These Rendolls and Vertisols merge with Mollisols, Histosols, and thicker, depression Vertisols in the study area.

The tropical wet and dry climate receives about 1500 mm of precipitation and averages 26.4°C annually, ranging from an average of 303 mm of precipitation in October to 39 mm in March and from 29°C in May to 21°C in December (based on a 22-yr record from Gallon Jug Farm in the watershed). We still lack a paleoclimate study for this area, but based on regional speleothem and lake core studies, we recognize droughts in the late Preclassic and late to Terminal Classic, the Post Classic, and the ‘Little Ice Age’ (Dunning et al., 2012; Hodell et al., 1995, 2005; Kennett et al., 2012; Medina-Elizalde et al., 2010). Such climate changes could have influenced floodplain and wetland formation, and the regional abandonment of wetland fields coincides with the Terminal Classic drought for as yet unknown reasons (Luzzadder-Beach et al., 2012).

Historical overview

Although humans and their tools were here from before the Holocene (Valdez and Aylesworth, 2005), only by about 5500 years ago do we observe traces of agriculture and deforestation (Beach et al., in press; Jones, 1994; Pohl et al., 1996). We also see tools modified as Archaic adaptations to new uses since these middle–late Archaic tools include features such as constricted adzes used in horticulture. Widespread agricultural diffusion occurred between 4000 and 3000 yr BP, and Maya Civilization spanned from the Preclassic (3200–1700 BP) to Classic (1700–1100 BP) and to Postclassic (after 1100 BP). The small rural population of the middle Preclassic (2900–2400 BP) relied on slash-and-burn horticulture, and populations and land uses expanded several times in the late Preclassic (2400–1700 BP) (Adams et al., 2004: 329). Regionally, sporadic growth and decline occurred in the early Classic (1700–1400 BP) as populations expanded into rural areas (Sullivan and Valdez, 2006), although the site of Blue Creek continued to grow (Adams et al., 2004; Guderjan, 2007). As settlement spread in the Classic, agricultural terracing and major water management expanded (1700–1100 BP; Beach et al., 2002; Scarborough et al., 1995). Most evidence suggests the late Classic period (1400–1100) had the highest populations and land use intensities. The Maya abandoned many regional sites around 1200–1100 BP (Guderjan, 2007; Guderjan et al., 2009; Valdez and Scarborough, 2014), until approximately 900 BP, when minor re-occupation occurred in the Birds of Paradise (BOP) Fields and Akab Muclil in this study area. There was less Postclassic environmental impact, and forests returned and lake sedimentation dropped in the early Postclassic (Mueller et al., 2010), although some sites like Lamanai, 20 km away, continued to have Maya occupation to European conquest (Graham et al., 1989).

Methods

The data in this article come from excavations on the footslopes and floodplains within the RBCMA. We have detailed soil, water, and geomorphology methods in previous publications. Methodologies include water chemistry (see Luzzadder-Beach and Beach, 2008, 2009 for field and laboratory water chemistry methods) and field coring, soil and sediment stratigraphy, excavations, magnetic susceptibility, extensive laboratory analyses of pollen and other microfossils, soil chemistry, and radiocarbon dating (see Beach et al., 2006, 2009, 2011, 2013, 2015 for methods). Here, we use excavations to bedrock, where possible, to expose sediment and soil sequences. Groundwater chemistry provides a uniformitarian window on potential past water quality conditions, since groundwater reflects the mineral characteristics of the aquifers in which it resides, and the geology has remained constant. Most of the dating relies on new accelerator mass spectrometry (AMS) dates (Table 1) with some ceramic evidence and synthesis of previous dates (Beach et al., 2003). This article defines sediment and soil sequences in broadly recognized terms of color (by Munsell), texture (by hydrometer), structure, carbonate (by HCl reaction and LOI), soil organic matter content (by LOI), and other soil terms as outlined by the Soil Survey Manual (Beach et al., 2008, 2009, 2011; USDA, 1993). For the buried soils in our study, we use the ‘soil memory’ concept of Targulian and Goryachkin (2004), which refers to the lines of soil information that generally remain stable after burial (Solís-Castillo et al., 2013). This article also presents sediment profiles of carbon isotopes for new soil sequences to understand the inputs of C₃ for C₄ vegetation on soil humin (Figure 5) (see methods in Awiti et al., 2008; Beach et al., 2011, 2015; Webb et al., 2004). We used only AMS radiocarbon determinations on organic sediments, terrestrial organics, or charcoal. We used a GF Instruments Magnetic Susceptibility Meter SM-20 based on 10⁻³ SI units to assess magnetic susceptibility in the field on soil columns at 50 mm intervals.

Table 1.

Radiocarbon dates and sedimentation rates.

Chawak Fan							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	AD 1000 mm yr⁻¹
NOSAMS-116652	C	60	1300	−25.89	AD 663–767	1183–1287	0–60	1300	0.46	2
NOSAMS-116653	Ab1	80	1600	−24.08	AD 406–536	1414–1544	60–80	300	0.67
NOSAMS-116653	Ab2	108	2360	−23.9	508–387 BC	2337–2458	80–108	760	0.37
Chawak Wetland Field 1							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	AD 1000 mm yr⁻¹
AA96018	C	55	1955	−22.5	38 BC–AD 125	1825–1988	0–106
AA96020	Ab	106	1283	−12.6	AD 657–859	1091–1293	106–188	1283	0.83	3.75
AA96019	2C	188	2119	−22.6	349–43 BC	1993–2299	188–212	836	0.98
AA96021	2C	212	2478	−23	772–429 BC	2379–2722	212255	359	0.67
AA96022	2C	255	2936	−22.3	1385–925 BC	2875–3335		458	0.94
Chawak Canal Field Edge
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP
AA92895		98–103	269	−26.3	AD 1494–1950	0–456
AA92892		135	1678	−22.7	AD 252–427	1523–1698
Chawak Wetland Field 2							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	AD 1000 mm yr⁻¹
AA92894	Ab	208	1228	−24.9	AD 688–885	1065–1262	0–208	1228	1.69	9.12
AA96023	2C	160	2507	−22.4	792–512 BC	2462–2742	0–130			4.6
AA96024	2C	175	2436	−24.5	705–405 BC	2355–2700	0–175	2597	0.67	1.1
Chawak Levee 3							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	1000 AD mm yr⁻¹
AA96027	Ab	85	1821	−22.9	AD 93–318	1632–1857
AA96028	2Cs	187	8513	−20
AA96029	2Cs	220	3698	−21.2
AA96030	2Cs	328	3648	−20.4	2134–1929 BC	3879–4084	0–328	3698	0.89
Dos Hombres (Beach et al., 2003)							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	1000 AD mm yr⁻¹
RB 2 op 10a	Ab	100	2000	−25	205 BC–AD 240	2155–1710	0–100	2000	0.5	1
RB 2 op 21a	Ab	94	1890	−19.8	195 BC–AD 430	2145–1520	0–140	1890	0.74	1.57
Gran Cacao 50 m test unit							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	1000 AD mm yr⁻¹
AA92883	Ab1	150	2119	−23.9	359 BC–AD 2	1948–2308
AA92884	Ab1	200	1889	−22.2	AD 31–229	1722–1919	0–200	2004	1	2
AA92885	2Ab	410	3344	−21	1738–1526 BC	3476–3688	200–410	1340	1.57
Gran Cacao 250 m test unit							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	1000 AD mm yr⁻¹
AA92886	Cg	70	853	−25.5	AD 1048–1262	688–902	0–70	853	0.82
AA92887	Cgss	330	2210	−25.9	386–173 BC	2123–2336	70–330	1357	1.92
AA92888	Ab	440	3380	−24.7	1768–1534 BC	3484–3718	330–440	1170	0.94
BOP Unit 2							Deposition rates
Sample number	Horizon	Depth (cm)	Age	D13C	Calib 7.0 2 σ ADBC	Dates BP	cm	Years	mm yr⁻¹	1000 AD mm yr⁻¹
AA92889	Ab	103	1263	−24.2	AD 666–867	1083–1284	0–103	1263	0.82
AA92890	2Ab	225	1338	−26	AD 641–767	1183–1309	103–225	75	16.27
AA92891	3Ab	330	1856	−28.6	AD 66–245	1705–1884	225–330	518	2.03

BOP: Birds of Paradise.

Findings

Water quality and quantity

The key elements in this region’s water chemistry are calcium (Ca²⁺) and sulfur (S), usually measured as sulfate ( $S O_{4}^{2 -}$ ) (Luzzadder-Beach and Beach, 2009) because both undergo dramatic increases over the course of the river (Figure 1). Over most of the river, beginning at elevations of about 140 m a.s.l., the water is low in both elements, but below c. 30 m a.s.l., Ca²⁺ and $S O_{4}^{2 -}$ increase through a series of springs that presumably drain water exposed for a much longer residence time in the sulfate-rich aquifer. The stream ranges from mostly overland flow with 48–338 ppm of Ca²⁺, and 0–11 ppm of $S O_{4}^{2 -}$ in the uplands, to extremely solute-rich water with 708–1402 ppm of Ca²⁺ and 567–1275 ppm of $S O_{4}^{2 -}$ on the coastal plain (Figure 1). Upper watershed dissolved loads thus have little influence on coastal plain aggradation, but floodplains below 30 m a.s.l. are subject to rising groundwater tables that will precipitate gypsum in the soil because of evaporative pumping driving the water to saturation. Gypsum saturation is defined as $S O_{4}^{2 -}$ concentrations of about 1600 ppm and Ca²⁺ concentrations of about 636 ppm, varying by temperature, by accompanying sodium chloride (NaCl) load, and by other chemicals present (Hounslow, 1995; Luzzadder-Beach and Beach, 2009). Surface water below elevations of 30 m a.s.l. in the coastal plain often is a mixture of high ionic groundwater and low ionic wet season runoff, which often dilutes calcium and sulfate far below saturation and may lead to dissolution and mobilization of gypsum in sediments. Thus, seasonal cycles may precipitate and dissolve gypsum in lowland or coastal plain floodplains. Second, this high-solute concentration in the water chemistry restricts the growth of certain types of vegetation and crops such as maize (Luzzadder-Beach and Beach, 2008, 2009), partly dependent on oxygen at the root level (Kramer and Boyer, 1995).

The Rio Bravo runs year round from near Chan Chich downstream to its confluence with the Rio Hondo, and during the wet season, it expands across its floodplain. Other tributaries in the region dry up during the dry season, and currently, Blue Creek/Rio Azul, which forms the border between modern NW Belize and Mexico, runs dry in the summer because of modern abstraction from its tributaries. Groundwater is available year round in the upland and lowland zones and represents a significant dry season resource for the ancient Maya as well as modern inhabitants. The Rio Bravo base flow is supplemented by spring flow as it enters the coastal plain. The lowland wetlands of the Rio Bravo watershed have near-surface groundwater during the dry season and regularly flood during the wet season.

Soil and geomorphic transects

Chawak footslope

Throughout this region, we have studied soil profiles on slope catenas to understand soil formation and erosion in the Holocene. We found paleosols in footslope soil profiles at La Milpa (Beach et al., 2003, 2006) and Blue Creek (Beach et al., 2008) and in agricultural terraces at Mahogany Ridge and La Milpa (Beach et al., 2003, 2011). The Blue Creek and La Milpa fans had paleosol sequences dated to Preclassic by AMS dates, and the terraces at Mahogany Ridge and La Milpa had ¹³C enrichment through the Classic period sediments indicative of organic matter enriched in C₄ plant residues. Each of these indicates landscape instability associated with the period of Maya land use, as well as late Classic soil and water conservation (Beach et al., 2002, 2008, in press).

To confirm earlier findings on footslopes and to trace sediment from slopes to floodplains, we excavated a footslope unit (Figures 2a and 2b) at the mouth of an arroyo just above the floodplain. This sequence provided 140 cm exposure to bedrock. The sequence had some mixed, non-diagnostic artifacts (ceramics and lithics), but the main lines of evidence for soil formation and environmental change are the well-developed paleosol formed between 108 and 140 cm and the sediment accumulation above this zone to the surface, where a cumulic topsoil has formed in the upper 30 cm. The top of the paleosol’s Ab3 horizon yielded an AMS date of 2360 BP at 108 cm (Table 1), which coincided with the lowest ceramics and the Maya late Preclassic. At 80 cm in the Ab1 an AMS date on charcoal was 1600 BP, and another in C horizon at 60 cm was 1300 BP. These samples date the Ab1 and C horizon deposit to the Maya Classic, and the well-developed topsoil in the upper 30 cm indicates relative landscape stability since.

Figure 2.

(a) Chawak But’o’ob and Dos Hombres transects, (b) Chawak But’o’ob soil sequence transect, and (c) Rio Bravo graben ecosystem model.

Evidence for instability through the Maya Classic zone also included an increase in high-energy sediments through the Maya Classic sediments. The paleosol’s Ab3 and C horizons have c. 40% clay, the profile’s highest. The topsoil is composed of 28% clay and the profile’s lowest coarse fraction, indicating little, recent high-energy inputs. The Classic period sediments, however, had gravel and 39–53% sand between the paleosol (25% sand) and the lower topsoil (23% sand) and lower melanization, consistent with initially high organic and clay-rich sedimentation in the Ab1 that diminishes upward to below the current topsoil. The decrease in fines and increase in sand may imply source reduction in clay-rich topsoils in the watershed with Maya Classic period erosion. We estimate sedimentation rates using the AMS dates and intervening sediment thicknesses (Table 1) to show low deposition in the late Preclassic to Classic (Ab2 to Ab1) of 0.37 mm yr⁻¹, faster rates in the late Classic of 0.67 mm yr⁻¹, which declines back to 0.46 mm yr⁻¹ after 1300 BP. Given the clay texture and well-developed A horizon, sedimentation was probably even lower since the Terminal Classic, and the late Classic rate was probably closer to 2 mm yr⁻¹ before Postclassic reforestation. Elevated magnetic susceptibility is also sandwiched between the Classic period sediments and the lowest paleosol, rising to 0.351 10⁻³ SI units between 0.136 at the surface and 0.064 10⁻³ SI units at the Cg horizon. For comparison, we studied a nearby footslope soil below a slope that unlike the Chawak footslope had sparse ancient mounds, terraces, and walls. This soil was 50 cm thick and had no paleosol, only the cumulic A horizons expected on footslopes. Hence, the more anthropogenic footslope soil shows c. 100 cm of deposition above the Preclassic paleosol, similar to other examples at La Milpa and Blue Creek. Also, Wright et al. (1959) noted that footslopes often have larger trees because of plentiful moisture and increased nutrient flow from bedrock parent material (Wright et al., 1959) and soil deposition and moisture retention (Becker et al., 1988). We noticed high biodiversity and very large tree specimens around this site, which may be related to Maya erosion and aggradation.

Chawak floodplain

On the graben floor below Chawak, we analyzed four units (Figure 2c). Here, the floodplain has strong elements of ancient Maya use, with the nearby site and its hydrologically engineered tributaries and wetland fields around its spring-fed lagoons. This riparian forest is the most mature and diverse (Brokaw et al., 1993), but it transitions eastward into the less diverse and shorter bajo scrub forest where the graben rises slightly to the east (Figure 2c). The Rio Bravo at this point runs through the structural valley with two branches, one at an elevation of c. 37–39 m a.s.l. on the escarpment edge that gains water from the stream and groundwater seeps and the main stem that courses some 700 m east across the large, flat graben at c. 40–41 m a.s.l. The graben extends eastward another c. 400 m before the tilted block gradually ascends eastward. The Rio Bravo enters this graben about 4 km south from a horst about 20 m higher (Figure 1). Thus, excavations at Chawak bisect the floodplain near the southern fringe of the graben.

The first field excavation was in a suspected ancient Maya wetland field (Figure 2b, profile 1). It displays a topsoil built on clays that bury a paleosol to a depth of c. 106 cm. The upper 160 cm of soil is about 60% clay, except for an activity area of decomposed gravel, cobble, and clay from 100 to 50 cm that buries the paleosol. A radiocarbon age from charcoal at 55 cm in the C Horizon is 1955 ± 38 BP (Table 1), which is late Preclassic and earlier than most Maya occupation evidence (Walling, 2005). The top of the paleosol has a ridged surface, accentuated by krotovinas and argilloturbation (Figures 2a and 2b). A date near the top of this buried soil at 106 cm is 1283 ± 36 BP and one below the bottom of an adjacent Maya canal near this site is 1678 ± 37 BP (Table 1). These are both Classic Period dates, although the second date is well below the canal and earlier than most cultural evidence. The charcoal date at 106 cm is late Classic and contemporary with the Maya site. The paleosol consists of an Ab horizon from c. 106 to 160 cm with few unidentifiable lithic and ceramic fragments and gravel clasts. Below this is a series of laminated sediments from 160 to 260 cm, which indicate cyclical reduction and oxidization with dark gray and yellow laminations caused by periods of water saturation and drainage (Figures 2a and 2b). These sediments date from 2119 ± 37 BP at 188 cm to 2936 ± 79 BP at 255 cm (Table 1), a time of both early Maya occupation of the watershed and climatic instability (Beach et al., in press).

Using the AMS dates and the sedimentation thicknesses, we estimated Preclassic sedimentation rates (Table 1) of 0.82 mm yr⁻¹ with no soil formation for c. 800 years. The upper 106 cm formed at the same rate of 0.83 mm yr⁻¹, while the anthropogenic paleosol formed at the comparable rate of c. 0.98 mm yr⁻¹. Since the topsoil is nearly pure clay and the upper activity area is about 50 cm deep, most of the sedimentation likely occurred by the end of the Terminal Classic site, and thus, the late to Terminal Classic deposition rate is close to 3.75 mm yr⁻¹ (c. 106 cm over 283 years). In comparison, the footslope also has a prominent paleosol buried at 108 cm and also dates to the Classic Period. In both cases, in the Classic period a soil had formed with greater melanization, from Munsell color N5/ and N6/ to N4/. We hypothesize the date at 55 cm of 1955 ± 38 BP is sediment dug from below by the Maya in excavating the canal and building up the field above the floodplain and thus managing water and planting surfaces (Table 1). These dates are in accord with other wetland fields in the watershed (Beach et al., 2009, 2013, 2015, in press).

The other lines of evidence for this sequence include δ¹³C, microscopic charcoal, and pollen. Pollen was poorly preserved in several layers; hence, we discuss only presence data. The δ¹³C profile produced a prominent increase from the low δ¹³C value of −27.54‰ at the surface to the high −21.53‰ near the top of the paleosol in the late Classic before decreasing back to −24.05‰ in the Preclassic surface (Figure 5). The δ¹³C value of the AMS sample at 106 cm was −12.6‰ and thus similar to a C₄ plant in this region of few such plants (Beach et al., 2011). Although pollen preservation was variable, grass pollen rises to 50–60% of total pollen through the Classic Zone of δ¹³C increase. The lowest level of the δ¹³C profile in the middle Preclassic (between the date of BP 2478 and 2936) had a δ¹³C decrease to −25.58‰ in a dark zone dominated by Typha pollen, which Das et al. (2010) measured at −26.1‰ in a cattail species (Typha domingensis) from the United States. This dark layer unexpectedly had the highest fine and larger particulate charcoal and may indeed be anthropogenic given the fire-inhibiting nature of this wetland (Figure 2b). Although Chawak dates to the late Classic, the nearby site of Dos Hombres has cultural and paleosol evidence from the middle Preclassic when charcoal deposition occurred.

A second excavation on a wetland field provided a similar sequence: a dark topsoil built on a Cg Horizon composed of reduced clays and a conspicuous gravel and cobble layer at 50–60 cm that has allowed oxidation of the layer to orange hues (Figure 2b, Chawak Wetland Field 2). This layer shows up in Unit 1, where it is less distinct and more bioturbated and falls in a zone that is elevated in δ¹³C. Paralleling the first unit, the zone from 100 to 150 cm is an Ab horizon with melanization of Munsell color N4/ mottled with yellow especially around rare gravel deposits. Coring around the unit site showed the Ab sequence was as deep as 220 cm in some place, and a date from 208 cm in one of these places was 1228 ± 36 BP. This shows differential fill to engineer the wetland. This also shows a localized much faster deposition rate of 1.69 mm yr⁻¹ to the present or 9.12 mm yr⁻¹ to 1000 BP (Table 1). The Ab, as in wetland field 1, lies above a laminated sequence of dark and reduced and oxidized layers that have higher amounts of sand. Two dates from depths of 160 and 175 cm, where the Ab and above is 150 cm deep, are 2507 ± 38 and 2436 ± 32 BP. These laminated layers as in Field 1 date to the middle Preclassic.

The third excavation (Figure 2b, Chawak Canal) was at a canal at the base of the escarpment near these wetland fields. The canal had c. 100 cm of dark colored, organic fill above a deep sequence of gleyed clay (Figure 2b). One AMS date below the canal at 135 cm was 1678 ± 37 BP and date in the mixed canal sediments was 269 ± 35 BP. The 135 cm date is 35 cm below the canal, and the upper date is Postclassic fill, accumulated since Terminal Classic abandonment.

The floodplain between the first three units and the levee unit transitions southward into a gilgai environment with scrub forest and sedges and bamboo. The levee topsoil is thick with 30 cm of A horizon uninterrupted by disarticulated B or C horizons with strong slickensides and tilted and buried A horizons that extend down to 160 cm. The whole Vertisol extends down c. 2 m, where it lies above laminated sediments with oxidized sands and reduced clays down to 330 cm. An AMS sample from 220 cm yielded a date of 3698 ± 28 BP and one from 328 cm produced a date of 3648 ± 32 BP. An AMS date on organic matter in calcareous sand at 187 cm produced a date of 8513 ± 51 BP, which likely reflects allochtonous organics. The δ¹³C profile produced a less pronounced increase from −26.75‰ to −23.62‰ just above the early Classic to late Preclassic AMS date (Figure 5). This may reflect an increase in δ¹³C associated with Maya land use and even maize production, but the zone is clearly highly argillo- and bioturbated in the Vertisol. We interpret the dating to indicate a varying depositional environment before 2000 BP in the Preclassic to Archaic c. 3700 BP, which shifts into the Vertic soil conditions in the upper 2 m, dating from the Archaic to Classic period. The upper zone has a series of disarticulated buried A horizon sections caused by the contraction and expansion of 2:1 clays and a major change of this landscape from the Preclassic to the Classic (Beach et al., 2003).

Each of the units and cores has produced a sequence from the Preclassic and Archaic of horizontally bedded floodplain sediments of varying energies and oxidation states to the late Preclassic and Classic soil horizons with anthropogenic impacts. The levee unit shows the most extreme transition to a Vertisol from the Archaic laminations. The Vertisol only formed in 2 m of sediments aggraded over up to 3700 years, which implies a substantial environmental change. We suggest several possible hypotheses to explain this changeover, but all must account for the top 2 m of sediment in which the current soils formed.

First, climatic change induced aridity could have tipped the scale of soil moisture to deficit, thus causing contraction and topsoil toppling into open cracks. As noted above, the evidence for recurring aridity is high in the late Preclassic, late Classic, and Post Classic. After the deposition associated with Maya late Preclassic and Classic periods aggraded the floodplain by 2 m, drought could have tipped the seasonal drying toward Vertisol formation.

Second, human alteration by deforestation would have produced more extreme wet season runoff and dry season soil moisture deficits. Another aspect of this is that upper soils are largely clay, whereas the lower section of horizontal beds varies between clay and coarser textures. Furthermore, this shift of diverse textures to mainly clays may be related to the ‘Maya Clays’ deposited from watershed deforestation and soil erosion, where 2:1 clays dominate the region’s slope soils. This all would have started in the Preclassic since the charcoal amounts from these levels indicate widespread burning both regionally and locally. Together, the anthropogenic explanation is thus a complex, diachronic model of land use, soil moisture, erosion and deposition, added mass for diapir formation (Beach et al., 2003), and clay expansion. The anthropogenic model also may have a climatic change component, with the recurring aridity events in the Classic and later periods.

Third, the Rio Bravo at this section has dissected into the floodplain by c. 2 m, which along with channel migration could have decreased flooding, causing floodplain deposition of clays while more coarse sediments remained in the channel, and lowered the water table and soil moisture stability.

Dos Hombres transect synthesis

We studied multiple excavations in the Rio Bravo from 1995 to 1998, which lie c. 1.5 km downstream from and c. 1 m lower in elevation than Chawak (Beach et al., 2003). These excavations ranged from areas highly influenced to little influenced by river flooding, including deep alluvial sequences near the Rio Bravo, a floodplain Mollisol with a buried Vertisol paleosol, Vertisols with evidence of aggradation, an oxbow core, and a Vertisol developed on a former point bar (Figures 3a and 3b). These excavations had a similar valley sequence of alluvial soils near the escarpment that transition into gilgai and more Vertisol formation through the higher central graben on progressively older soils away from the main active channel.

Figure 3.

(a) Dos Hombres survey transect and (b) Dos Hombres soil sequence transect.

A soil near the western base of the Rio Bravo graben (RB S3100 E 10) had no evidence for wetland field canals but similar morphology (without the Ab) to the wetland field sites at Chawak (Figure 3b). This site had a thin A horizon of 10–20 cm overlying a BW horizon that extended down to 37 cm and a thick Cgss horizon with many cracks and slickensides and prismatic structure. This soil had reduced, N5/, N6/ Munsell colored horizons. This Aquert did have well-developed slickensides and tilted, wedge-shaped structural aggregates. Below 65 cm, the gray clay was mottled with 2.5Y5/8 yellow clay and kankars or decomposing limestone cobbles, which lie atop sand and clay laminations from 125 to 200 cm. The thin A horizon may reflect rapid deposition that does not allow a mature topsoil to form. We have no dates here, but if they replicate the Chawak wetland strata, the upper clays represent Maya Classic clay aggradation over the Preclassic sand and clay laminations.

The soil (Op 10a) in the graben, just 100 m east of the Rio Bravo escarpment, shows 100 cm of sediment burial over a well-developed Ab horizon with a date of 2155–1710 BP (Figures 3a and 3b). The clay sediments above the paleosol had several Classic or non-diagnostic ceramics and some larger gravel clasts. The modern topsoil was moderately developed compared with the thick and dark, Preclassic Ab horizon, which was similar to other regional modern and buried soil sequences (Luzzadder-Beach and Beach, 2009). Both the surface soil (0–100 cm) and paleosol (100–150 cm) have slickensides, mottled gray and yellow colors (indicating soil moisture variability), and prismatic structure, but only the paleosol has diagonal Ab horizons plunging into the Css horizon (Beach et al., 2003).

About 100 m east on a slight rise just west of the Rio Bravo, we studied a well-developed Vertisol (RB2 Op 16). Aggradation has occurred here, but the abundant cracking, slickensides, and a disarticulated and distorted Ab horizon ranged from 70 to 150 cm in depth. Argilloturbation has partially obliterated the paleosol, integrating the Ab horizon into subjacent clay and limestone saprolite and superadjacent clayey sediments (Figure 3b).

The number of oxbow lakes in the valley testify to significantly different past meander patterns. At the eastern edge of the Rio Bravo Graben, a pollen core from 60 to 85 cm from an oxbow (Figure 3a, near Op 12) dates to 2010–1860 BP and 2860–2740 BP (Dunning et al., 1999). The lower date came from the underlying, floodplain clay and the upper date came from lake sediment and is a more reliable date for the thin 25-cm organic sediment because the lower date could represent an unconformity from channel erosion. An excavation (Op 12) on an adjacent low ridge revealed a Vertisol developed in clay soils overlying point bar sands (below 124 cm). The key aspects here are Vertisol formation sometime after the adjacent early Preclassic date (2860–2740 BP), evidence for river mobility in the broad Preclassic period (1860–2860 BP), and the abundant pollen of Zea mays and disturbance pollen in the late Preclassic period (Dunning et al., 2003; Lohse, 2006).

Vertisol features, scrub forest of Palo de Tinto (Haematoxylum campechianum) and sawgrass (Cladium mariscus subsp. Jamaicense), and gilgai increase eastward with distance and elevation from the river. The Vertisols at Ops 9A-D still receive flows from a small, broad, yazoo-like tributary of the Rio Bravo. The low gradient, seasonal stream flows in small, anastomosing channels that follow gilgai cracks, accentuating the wallows and vegetation-anchored hummocks. On the far eastern edge of the graben is a zone of mature Vertisols in a depression subject to little or no river deposition but from erosion of the surrounding slopes, which had intensive habitation in the late Classic (Lohse, 2006). This soil also has a highly distorted and disarticulated Ab horizon with strong yellow and gray mottling (Beach et al., 2003). The radiometric date from the Ab at 100 cm of 2145–1520 BP provides broadly similar late Preclassic to early Classic chronology as the Vertisols at Chawak and Op 10A.

Gran Cacao floodplain transect

The next transect (Figure 4a) was at 20 m a.s.l. across the Rio Bravo floodplain after the channel passes the confines of the Rio Bravo canyon, some 18 m lower than at Dos Hombres. Here, the river’s tighter meanders of the graben give way to a wider meander belt of the coastal plain. This broader lowland zone is also subject to hydraulic damming of the nearby confluence of the Booth’s River and Rio Azul. Thus, although the floodplain is wider, it is still subject to flooding. The water chemistry also changes abruptly upstream of this site at 25–30 m a.s.l., where the solute load of calcium and sulfate rises because of the mixing of groundwater from a regional aquifer (Figure 1). Hence, sediment load is both higher from the rivers and the nearby water table, which can add mass to soils through calcium carbonate (CaCO₃) and gypsum (CaSO₄ + H₂O) precipitation from water that becomes saturated through evaporation (Beach et al., 2009; Luzzadder-Beach and Beach, 2009).

Figure 4.

(a) Gran Cacao and BOP transect and (b) Gran Cacao and BOP soil sequence transect.

We excavated three deep trenches under high, riparian forest canopy on this floodplain (Figures 4a and 4b). The trenches were at 50, 250, and 500 m away from the Rio Bravo’s low season floodplain and levee. The floodplain through this zone ranged from a relatively flat surface from frequent inundation and deposition toward the hummocky gilgai surface of Vertisol formation, where seasonal drying and wetting allow contraction and expansion on the terrace at 500 m.

The 50-m south unit (Figure 4b) provided an exposure of 450 cm of dominantly clay sediments through its 3500 year sequence. The topsoil A horizon was 25 cm here, overlying a Cg horizon from 25 to 150 cm with an intervening Ab2 horizon from 100 to 120 cm. The main upper Paleosol at this site from 150 to 200 cm had AMS dates of 2119 ± 64 BP at 150 cm and 1889 ± 37 BP at 200 cm, reversed but not unexpected in a bio- and argilloturbating topsoil. A lower date on charcoal in the Ab3 horizon at 405–415 cm was 3344 ± 39 BP and shows relatively rapid deposition in this lower floodplain at the edge of the coastal plain. Although there is no close site, Maya lithics and ceramics are still evident at 100–200 cm. The stable carbon isotope data provide evidence of C₄ plant inputs into soil humin in the paleosol and also in the whole sequence. The topsoil is −22.65‰ and rises to −20.92‰ in the late Preclassic, dropping to −24‰ in the lower Archaic depth (Figure 5).

Figure 5.

δ¹³C isotope profiles of soil sequences in the Rio Bravo floodplain.

Although there is a faintly developed cumulic Ab2 horizon that dates to the late Preclassic with few artifacts (ceramics and lithics), the change in sedimentation and thus instability is too subtle to differentiate. From c. 3350 to 2000 BP (410–200 cm), the deposition rate was 1.5 mm yr⁻¹, and from c. 2000BP to the present in the upper 200 cm, the rate drops to c. 1 mm yr⁻¹. Based on limited dating, the combined Preclassic deposition is 1.5 times faster than the combined Classic to present deposition.

The 250-m south unit was more distant in the floodplain but equally aggraded as the 50-m site, again with a 450-cm sequence dating from c. 3400 years to the present. Here, the topsoil was cumulic, 60 cm thick, and an AMS date from charcoal was 853 ± 35 BP in the silty clay Cy horizon at 70 cm. Two other AMS dates were 2210 ± 45 BP in an Ab2 horizon at 330 cm and 3380 ± 39 BP in an Ab3 horizon 440 cm. The rate of deposition in the Archaic to late Preclassic zone is 0.9 mm yr⁻¹, and the late Preclassic to Postclassic zone from 330 to 70 cm increased to 1.9 mm yr⁻¹, with the Postclassic to present zone dropping to 0.8 mm yr⁻¹. This indicates an approximate two-fold increase in deposition through the main Maya occupation in this region. With better dating, the 50-m unit may also reflect this chronology. The stable carbon isotope data provide equivocal evidence of C₄ plant inputs into soil humin in the whole sequence because the topsoil is −23.6‰ and decreases to −26.3‰ at 25 cm, before rising to c. −24‰ from the Post Classic to the Archaic depth (Figure 5).

The site at 500 m south of the river was 5 m higher and less frequently flooded by the river given its closeness to the weathered bedrock limestone at 160 cm. This soil is a Vertisol similar to the Dos Hombres floodplain at Op 10a because of the gilgai surface and gray and yellow mottled Cgss horizons and prismatic structures. It did have an Ab horizon from 110 to 150 cm with increase in melanization but no datable material. The site had no disarticulated, vertical, or diagonal A horizons in the subsoil, as at Unit 3 in the Chawak transect and the Ops 9, 16, and 21 on the Dos Hombres transect. The stable carbon isotope data provide evidence of increased C₄ plant inputs in soil humin in the paleosol, rising from the topsoil at −27.3‰ to −22‰ just above the buried paleosol (Figure 5). If this parallels other buried soils, then it dates to the Preclassic, and the sediments that bury it date to the late Preclassic and Classic period.

BOP transect

We excavated a series of units to understand wetland formation 1 km east and down river, both by natural and anthropogenic factors (Beach et al., 2009, 2011; Luzzadder-Beach et al., 2012) (Figures 4a and 4b). Here, the floodplain lies at 15–17 m a.s.l. at the confluence of Cacao Creek and the Rio Bravo. This is the ancient Maya wetland field site called the BOP Fields, where we excavated many trenches across canals and field edges to study ancient Maya wetland fields and environmental change. Wetland fields occupy at least 1 km² on this savanna floodplain and more in surrounding forest. The Maya wetland fields at this site are mainly drained fields with 1–2 m wide canals running east–west and north–south at interval of c. 10 m; hence, burial from the canal excavated sediments should not cover the majority of the intervening fields to much depth. For the current floodplain formation study, we excavated a series of units in the middle of wetland field on a north–south axis away from Cacao Creek (Figures 4a and 4b).

Our deepest sequence, at BOP 2, extended through sediments and paleosols to 350 cm, and a date at 330 cm was c. 1856 ± 40 BP. The AMS date at 225 cm was 1338 ± 36 BP and another in a zone of soil disturbance at 103 cm was 1263 ± 36 BP (Table 1). The zone between 225 and 100 cm has many signs of disturbance, and the two dates separated by 120 cm, from 225 to 103 cm, are virtually identical (Figure 4b). These dates parallel those from many wetland field and canal sequences with canal sediments dating to the late to Post Classic and the fields predating the canals in the late Classic (Luzzadder-Beach et al., 2012). The paleosol or activity area spans a zone from 90 to 200 cm that has zones of oxidized soil with a brown color, melanization, artifacts (ceramics and lithics), and a prominent increase in δ¹³C from −26‰ at 75 cm to −23‰ at 145 cm before decreasing back to −25.91‰ at 250 cm (Figure 5). This increase in C₄ species soil humin parallels findings from the wetland field edges and canals, which also found the bulge in δ¹³C correlated with increased grass pollen, Z. mays, charcoal, and Maya late Classic ceramics (Beach et al., 2009). It is clear that this region is highly anthropogenically altered, and if we ignore the possible sedimentation caused directly by the Maya, the deposition rate is c. 2 mm yr⁻¹ in the early to late Classic, c. 19.27 mm yr⁻¹ in the active paleosol zone, and 0.8 mm yr⁻¹ in the late Classic to present. This rate parallels the rate of deposition for the other fields reported earlier (Beach et al., 2009) of 1.47–2.44 mm yr⁻¹, with the highest rates occurring through the active field building zone in the late Classic. We posit three possible explanations for the rapid rate of deposition through the active paleosol zone: accelerated flooding near this triple confluence, active wetland field raising through this zone, and gypsum aggradation. Here, there is a marked increase in gypsum crystals, which certainly contributes to the mass increase in the sediments, but the aggradation rate here is much higher than two other wetland field sites at Chan Cahal and Sayap Ha where gypsum plays a larger role.

Synthesis floodplain and depression formation at Chan Cahal and BOP

The Chan Cahal area (Figure 1) is a special case of wetland formation because it receives runoff from a localized slope and lies in an area of perched water table fed by the high Ca²⁺ and $S O_{4}^{2 -}$ springs. This region of intense ancient Maya wetland agriculture was aggraded by four main contributing factors: slope erosion from fluvial and mass wasting, human manipulation from canal digging and field raising, and gypsum precipitation, but likely not river flooding (Luzzadder-Beach and Beach, 2009). Gypsum precipitation is a major component of aggradation at this site since groundwater with high dissolved load is its main source with less mixing of overland runoff during storms. Based on research from 2000 to 2008 at Chan Cahal, wetland aggradation experienced a slow rate of 0.65 mm yr⁻¹ in the fields and canals. Both are about half the rate of the previous BOP fields and much lower than in the present study.

Sayap Ha is a wetland field on the floodplain downhill from the Chan Cahal site (Beach et al., 2013, 2015) and provides a well-dated sequence of aggraded sediments. The raw deposition rate in the late Classic here from 180 cm to the surface is c. 1.21 mm yr⁻¹ and is 1 mm yr⁻¹ in the Post Classic. The canal at this site produced an overall aggradation rate of 1.4 mm yr⁻¹. The upper canal and field deposition rates were both very close to 1 mm yr⁻¹ (1.06 and 1 mm yr⁻¹, respectively), but the lower rates were 1.37 mm yr⁻¹ in the field between 180 and 65 cm and 2.3 mm yr⁻¹ in the lower 58 cm of the canal. Except for the high late Classic aggradation for BOP above, these were comparable rates to the fills at the earlier BOP fills (Beach et al., 2009) and higher rates than at Chan Cahal. All the wetland fields thus far show aggradation in the late Classic, canal excavation in the late Classic, and rapid canal fill near the onset of the Post Classic. The highest rates of aggradation are in the lower floodplain at Sayap Ha and BOP and at Chawak during active human wetland field construction on active floodplains.

Conclusion

This paper has provided the first assessments of floodplain formation and evidence for a new ancient Maya wetland field system in the Rio Bravo, a little studied tropical river. To study these, we examined water chemistry and four soil transects through the Rio Bravo valley, spanning the alluvial zone from channels to the lowest terraces or bedrock islands, where the floodplain changes form alluvial sequences with young soils to infrequently flooded, more developed Vertisols. Where soils are older and deeper, more vertic features and other time-dependent features like Fe and Mn nodules have formed. Thus, one implication of this work is that Preclassic and Archaic archaeological surveys will have to focus on Vertisols or need to use deeper excavation and geophysical prospection.

The system experiences two extreme variations: in Dry Season to Wet Season flows and seasonal and longitudinal water chemistry. Flows are low from January to June through the Dry Season but much higher and spasmodic in the Wet Season. There is an equally great variation in water chemistry through the system from the low-solute upper watershed to the point below 30 m a.s.l., where high-solute groundwater mixes with the stream’s base flow. This mixing of groundwater with high dissolved load of $S O_{4}^{2 -}$ and Ca²⁺ and surface water with a low dissolved load creates great seasonal variation in water chemistry. These water chemistry attributes affect both human use and long-term floodplain formation.

All the sites have well-developed topsoils with evidence of minimal aggradation of mostly clay over the last millennium. Before and during the outset of Maya Civilization, floodplain deposition rates ranged from 0.82 mm yr⁻¹ with no soil formation for c. 800 years at Chawak in the sequence through laminated sediments to 1.5 mm yr⁻¹ at GC 50 and 0.9 mm yr⁻¹ at GC 250 in an Archaic to late Preclassic zone. Thus, Archaic to late Preclassic rates were c. 0.8 to 1.5 mm yr⁻¹ in most of the floodplain and much lower on the footslopes. Early deposition rates were even low where gypsum precipitation contributes to aggradation at Chan Cahal, Sayap Ha, and the Gran Cacao units. The deposition rises through the late Preclassic to Classic layers at all sites, and the Chawak and Dos Hombres areas experienced a large changeover from layered sand and clay deposits to mostly clays with occasional gravel and cobbles in the late Preclassic through Classic period. We interpret the early units as variable, slowly depositional fluvial and lacustrine deposits and the latter as overbank floodplain and colluvial mixes during the intensive Maya land use periods.

Several sites have activity layers from the Classic and paleosols from the Preclassic Maya periods. The sites of BOP, Chawak, and Sayap Ha have activity levels and increased deposition that date to the late Classic period (1400–1100 BP), as does the footslope at Chawak. At these sites, some combination of Maya field building and increased floodplain sedimentation increased aggradation rates by 16.27 mm yr⁻¹ at BOP to 1.6 mm yr⁻¹ in the Gran Cacao trenches. The later rate increases through the Maya periods by two times, but the rates are much higher in the Maya wetlands fields at BOP. Both rates were higher than deposition rates up the canyon that ranged from 0.83 to 0.96 mm yr⁻¹ at Chawak, which had clear evidence for, if lower rates of, erosion and deposition. The higher rates of deposition at BOP probably reflect the triple confluence of the main rivers, plus the confluence of Cacao Creek, and the Maya field building in this large-scale ancient wetland management system. Most Postclassic deposition rates (Chawak, Gran Cacao, Chan Cahal) during this time of abandonment recede back to the Archaic to early Preclassic levels.

The delta C isotopes in soils show variations driven by changes in the species that use the C₄, C₃, and CAM photosynthetic pathways. Wetland fields at Chawak and the BOP floodplains exhibit δ¹³C enrichment in the late Classic sequences, as does the footslope at Chawak. Since we have overlapping evidence for Z. mays and grass increases at BOP and Chawak, we reason that maize must have been an important component of C₄ species input into soil humin. We see some increase in the δ¹³C enrichment in the Preclassic sequences, but it is possible that like near the Usumacinta River (Solís-Castillo et al., 2013) the drift toward lower ¹³C in the Preclassic is related to a dryer climate and its impacts on vegetation. We do have growing evidence both in disturbance and Z. mays pollen at Laguna Verde and Chan Cahal (Beach et al., 2009) and Laguna Juan Pioja (Dunning et al., 2003) that the drift toward lower ¹³C in the Preclassic and Classic is related to human-induced vegetation change. The levels of δ¹³C enrichment are similar to levels in Beach et al. (2011), suggesting that C₄ plants made up only about 25% of the late Classic vegetation where C₄ species represent a small part today’s ecosystems. Since other disturbance and crop species are C₃, the 25% C₄ estimate probably indicates more than 25% deforestation. But even around the large city of Tikal, Lentz et al. (2015) estimate that 40% of forest remained through the late Classic.

Finally, two key research goals were to investigate whether the Chawak fields with their prominent canals were indeed wetland fields and to study whether wetland fields occurred in the zone of low-solute water of the Rio Bravo. We conclude that these are indeed Classic Maya canals and field systems in this zone of low dissolved load based on three lines of evidence. First, the field and canal building with activity areas in the wetlands dates to late classic, the same activity period as fringing site of Chawak. Second, the δ¹³C proxy of C₄ plant organic matter increases through late Classic–dated activity areas both in the footslope and wetland soil. Third, these chronological and morphological characteristics parallel findings at nearby Chan Cahal and BOP, where we have a host of additional proxy lines of evidence for ancient wetland agriculture (Beach et al., 2009, 2015). This evidence continues to expand the areal extent of this ancient, intensive, wetland agroecosystem.

Footnotes

Acknowledgements

In Belize, we worked within the Maya Research Program with Dr T Guderjan, Director, the Programme for Belize Archaeological Project, Dr F Valdez Jr, Director, and with the gracious cooperation of the Department of Archaeology, Institute of Archaeology, National Institute of Culture and History, the Programme for Belize, and the communities of Blue Creek and San Felipe. We thank Nicholas Brokaw and Dr Michael Brennan for reviews and field help, as well as P Magana, K Cox, Esq., D, B, and M Dyck, anonymous reviewers, and many graduate and undergraduate students of our institutions. We also thank Karl Butzer for his half-century influence on geoarchaeology, always looking for the Longue Dureé of landscape change with a skeptical eye toward convenient explanations. The findings and interpretations here are the responsibility of the authors, not of the supporting agencies, institutions, and reviewers.

Funding

We thank the following organizations for supporting this research: Georgetown University’s School of Foreign Service, the Cinco Hermanos Chair in Environment and International Affairs; The University of Texas at Austin, The C.B. Smith Centennial Chair; grants from the National Geographic Society (CRE-7506-03, CRE-7861-05; T. Beach and S. Luzzadder-Beach PIs), the Guggenheim Foundation, Dumbarton Oaks, and the National Science Foundation (Nos BCS-0924510, T. Beach, PI; BCS-0924501, S. Luzzadder-Beach, PI; BCS-0241757; GEO-CNH-1114947, Brokaw, Ward, Cortes-Rincon, Luzzadder-Beach, Walling; SBR 963-1024, V. Scarborough and N. Dunning); George Mason University’s Center for Global Studies and Provost’s Office.

References

Adams

REW

Robichaux

Valdez

Jr . (2004) Transformations, periodicity, and urban development in the Three Rivers region. In: Demarest

Rice

(eds) The Terminal Classic in the Maya Lowlands: Collapse, Transition and Transformation. Boulder, CO: University Press of Colorado, pp. 302–323.

Anselmetti

Hodell

Ariztegui

. (2007) Quantification of soil erosion rates related to ancient Maya deforestation. Geology 35: 915–918.

Awiti

Walsh

Shepherd

. (2008) Soil condition classification using infrared spectroscopy: A proposition for assessment of soil condition along a tropical forest-cropland chronosequence. Geoderma 143: 73–84.

Bautista

Palacio-Aponte

Quintana

. (2011) Spatial distribution and development of soils in tropical karst areas from the Peninsula of Yucatan, Mexico. Geomorphology 135: 308–321.

Beach

(1998) Soil catenas, tropical deforestation, and ancient and contemporary soil erosion in the Petén, Guatemala. Physical Geography 19: 378–404.

Beach

Luzzadder-Beach

Lohse

(2013) Landscape formation and agriculture in the wetlands of northwestern Belize. In: Lohse

(ed.) Classic Maya Political Ecology. Los Angeles, CA: UCLA Cotsen Institute of Archaeology Press, pp. 43–67.

Beach

Dunning

Luzzadder-Beach

. (2006) Ancient Maya impacts on soils and soil erosion. Catena 66: 166–178.

Beach

Dunning

Luzzadder-Beach

. (2008) Human and natural impacts on fluvial and karst depressions of the Maya Lowlands. Geomorphology 10: 301–331.

Beach

Luzzadder-Beach

Cook

. (in press) Ancient Maya impacts on the Earth’s surface: an early anthropocene analog? Quaternary Science Reviews.

10.

Beach

Luzzadder-Beach

Dunning

. (2002) Upland agriculture in the Maya lowlands: ancient conservation in northwestern Belize. Geographical Review 92(3): 372–397.

11.

Beach

Luzzadder-Beach

Dunning

. (2003) Depression soils in the lowland tropics of northwestern Belize. In: Gómez-Pompa

Allen

Fedick

. (eds) The Lowland Maya Area: Three Millennia at the Human–Wildland Interface. Binghamton, NY: Haworth Press, pp. 139–173.

12.

Beach

Luzzadder-Beach

Dunning

. (2009) A review of human and natural changes in Maya Lowlands Wetlands over the Holocene. Quaternary Science Reviews 28: 1710–1724.

13.

Beach

Luzzadder-Beach

Dunning

. (2011) Carbon isotopic ratios of wetland and terrace soil sequences in the Maya Lowlands of Belize and Guatemala. Catena 85: 109–118.

14.

Beach

Luzzadder-Beach

Guderjan

. (2015) The floating gardens of Chan Cahal: Soils, water, and human interactions. Catena 132: 151–164.

15.

Becker

Rabenold

Idol

. (1988) Water potential gradient for gaps and slopes in a Panamanian tropical moist forest’s dry season. Journal of Tropical Ecology 4: 173–184.

16.

Braudel

(2011) Memory and the Mediterranean. New York: Random House.

17.

Brennan

King

Shaw

. (2013) Preliminary geochemical assessment of limestone resources and stone use at Maya sites in the Three Rivers Region, Belize. Journal of Archaeological Science 40: 3178–3192.

18.

Bridgewater

SGM

Ibanez

Ratter

. (2002) Vegetation classification and floristics of the savannas and associated wetlands of the Rio Bravo Conservation and Management Area, Belize. Edinburgh Journal of Botany 59: 421–442.

19.

Brokaw

NVL

Mallory

Schulze

. (1993) Vegetation of the Rio Bravo Conservation and Management Area, Belize. Manomet, MA: Manomet Bird Observatory.

20.

Butzer

(1959) Contributions to the Pleistocene geology of the Nile Valley. Erdkunde 13: 46–67.

21.

Butzer

(2005) Environmental history in the Mediterranean world: Cross-disciplinary investigation of cause and effect for degradation and soil erosion. Journal of Archaeological Science 32: 1773–1800.

22.

Butzer

Abbott

Frederick

. (2008) Soil-geomorphology and ‘wet’ cycles in the Holocene record of North-Central Mexico. Geomorphology 101(1–2): 237–277.

23.

Cabadas

Solleiro

Sedov

. (2010) The complex genesis of red soils in peninsula de Yucatán, Mexico: Mineralogical, micromorphological and geochemical proxies. Eurasian Soil Science 43(13): 1439–1457.

24.

Das

Wang

Hsieh

(2010) Chemical and carbon isotopic characteristics of ash and smoke derived from burning of C₃ and C₄ grasses. Organic Geochemistry 41: 263–269.

25.

Dunning

Beach

Luzzadder-Beach

(2012) Kax and Kol: Collapse and resilience in lowland Maya civilization. Proceedings of the National Academy of Sciences 109: 3652–3657.

26.

Dunning

Jones

Beach

. (2003) Physiography, habitats, and landscapes of the Three Rivers Region. In: Scarborough

Valdez

Jr Dunning

(eds) Heterarchy, Political Economy and the Ancient Maya: The Three Rivers Region of the East-Central Yucatan Peninsula. Tempe, AZ: University of Arizona Press, pp. 14–24.

27.

Dunning

Luzzadder-Beach

Beach

. (2002) Arising from the bajos: The evolution of a neotropical landscape and the rise of Maya civilization. Annals of the Association of American Geographers 92: 267–283.

28.

Dunning

Scarborough

Valdez

Jr . (1999) Temple mountains and sacred lakes, and fertile fields: Ancient Maya landscapes in northwestern Belize. Antiquity 73: 650–660.

29.

Graham

Pendergast

Jones

(1989) On the fringes of conquest: Maya-Spanish contact in early Belize. Science 246: 1254–1259.

30.

Guderjan

(2007) The Nature of an Ancient Maya City: Resources, Interaction and Power at Blue Creek, Belize. Tuscaloosa, AL: University of Alabama Press.

31.

Guderjan

Beach

Luzzadder-Beach

. (2009) Understanding the causes of abandonment in the Maya Lowlands. Archaeological Review of Cambridge 24: 99–122.

32.

Hartshorn

(1988) Tropical and subtropical vegetation of Meso-America. In: Barbour

Billings

(eds) North American Terrestrial Vegetation. Cambridge: Cambridge University Press, pp. 365–390.

33.

Hartshorn

Nicolait

Hartshorn

. (1984) Belize Country Environmental Profile: A Field Study. Belize City: Robert Nicolait and Associates, Ltd.

34.

Hodell

Brenner

Curtis

(2005) Terminal Classic drought in the northern Maya Lowlands inferred from multiple sediment cores in Lake Chichancanab (Mexico). Quaternary Science Reviews 24(12–13): 1413–1427.

35.

Hodell

Curtis

Brenner

(1995) Possible role of climate in the collapse of Classic Maya civilization. Nature 375: 391–394.

36.

Hounslow

(1995) Water Quality Data: Analysis and Interpretation. Boca Raton, FL: CRC.

37.

Jones

(1994) Pollen evidence for early settlement and agriculture in northern Belize. Palynology 18: 205–211.

38.

Kennett

Beach

(2013) Archaeological and environmental lessons for the Anthropocene from the Classic Maya Collapse. The Anthropocene 4(13): 88–100.

39.

Kennett

Breitenbach

Aquino

. (2012) Development and disintegration of Maya political systems in response to climate change. Science 338(6108): 788–791.

40.

King

Jr Pope

Petruny

(2004) Stratigraphy of Belize, north of the 17th parallel. Gulf Coast Association of Geological Societies Transactions 54: 289–303.

41.

King

Baillie

Abell

TMB

. (1992) Land Resource Assessment of Northern Belize (Bulletin 43, vol. 1). Southampton: Natural Resources Institute, Overseas Development Administration.

42.

Kramer

Boyer

(1995) Water Relations of Plants and Soils. San Diego, CA: Academic Press.

43.

Lentz

Magee

Weaver

. (2015) Agroforestry and agricultural practices of the ancient Maya of Tikal: Resilience and management of an essential resource. In: Lentz

Dunning

Scarborough

(eds) Tikal: Paleoecology of an Ancient Maya City. New York: Cambridge University Press, pp. 152–185.

44.

Lohse

Awe

Griffith

. (2006) Preceramic occupations in Belize: Updating the Paleoindian and Archaic record. Latin American Antiquity 17: 209–226.

45.

Luzzadder-Beach

Beach

(2008) Water chemistry constraints and possibilities for the ancient and contemporary Maya Lowlands. The Journal of Ethnobiology 28: 211–230.

46.

Luzzadder-Beach

Beach

(2009) Arising from the wetlands, mechanisms and chronology of landscape aggradation in the northern coastal plain of Belize. Annals of the Association of American Geographers 99: 1–26.

47.

Luzzadder-Beach

Beach

Dunning

(2012) Wetland fields as mirrors of drought and the Maya abandonment. Proceedings of the National Academy of Sciences 109: 3646–3651.

48.

Marshall

(2007) Geomorphology and physiographic provinces of Central America. In: Bundschuh

Alvarado

(eds) Central America, Geology: Resources and Hazards. London: Taylor & Francis Group, pp. 75–122.

49.

Medina-Elizalde

Burns

Lea

. (2010) High resolution stalagmite climate record from 395 the Yucatan Peninsula spanning the Maya terminal classic period. Earth and Planetary Science Letters 298(1–2): 255–262.

50.

Mueller

Islebe

Anselmetti

. (2010) Recovery of the forest ecosystem in the tropical lowlands of northern Guatemala after disintegration of Classic Maya polities. Geology 38: 523–526.

51.

Nations

(2006) The Maya Tropical Forest: People, Parks, and Ancient Cities. Austin, TX: University of Texas Press.

52.

Perry

Payton

Pederson

. (2009) Groundwater geochemistry of the Yucatan Peninsula, Mexico, constraints on stratigraphy and hydrogeology. Journal of Hydrology 367: 27–40.

53.

Pohl

Pope

Jones

. (1996) Early agriculture in the Maya Lowlands. Latin American Antiquity 74: 355–372.

54.

Scarborough

Becher

Baker

. (1995) Water and land at the ancient Maya community of La Milpa. Latin American Antiquity 6: 98–119.

55.

Solís-Castillo

Solleiro-Rebolledo

Sedov

. (2013) Paleoenvironment and human occupation in the Maya Lowlands of the Usumacinta River, Southern Mexico. Geoarchaeology 28: 268–288.

56.

Sullivan

Valdez

Jr (2006) The Late Preclassic to Early Classic Transition in the Three Rivers Region. Research Reports in Belizean Archaeology, vol. 3, pp. 74–83. Belize: The Institute of Archaeology, NICH.

57.

Tankersley

Scarborough

Dunning

. (2011) Evidence for volcanic ash fall in the Maya Lowlands from a reservoir at Tikal, Guatemala. Journal of Archaeological Science 38: 2925–2938.

58.

Targulian

Goryachkin

(2004) Soil memory: Types of records, carriers, hierarchy and diversity. Revista Mexicana de Ciencias Geologicas 21: 1–8.

59.

USDA Soil Survey Staff (1993) Soil Survey Manual. US Department of Agriculture Handbook, Vol. 18. Washington, DC: US Government Printing Office.

60.

Valdez

Jr Aylesworth

(2005) A fluted Paleoindian point and other chipped stone tools from August Pine Ridge, Belize. Mono y Conejo: Journal of the Mesoamerican Archaeological Research Laboratory 3: 35–39.

61.

Valdez

Jr Scarborough

(2014) The prehistoric Maya of Northern Belize: Issues of drought and cultural transformations. In: Iannone

(ed.) The Great Maya Droughts in Cultural Context: Case Studies in Resilience and Vulnerability. Boulder, CO: University of Colorado Press, pp. 255–269.

62.

Walling

(2005) Archaeological investigation of Prehispanic Maya residential terraces, commoner housing and hydrology at Chawak But’o’ob, Belize. Antiquity 79(304). Available at: http://antiquity.ac.uk/projgall/walling304/ (accessed 24 May 2015).

63.

Webb

Schwarcz

Healy

(2004) Carbon isotope evidence for ancient maize agriculture in the Maya Lowlands. Journal of Archaeological Science 31: 1039–1052.

64.

Wright

ACS

Romney

Arbuckle

. (1959) Land in British Honduras. London: Her Majesty’s Stationery Office (Colonial Research Publication 24).