1. Introduction
Bayesian chronological modeling has significantly advanced our understanding of the dynamics of settlement development, growth, and decline across the Maya World (Arroyo et al. Reference Arroyo, Inomata, Ajú, Estrada, Nasu and Aoyama2020; Culleton et al. Reference Culleton, Prufer and Kennett2012; Ebert et al., Reference Ebert, Culleton, Awe and Kennett2016; Hanna et al. Reference Hanna, Graham, Pendergast, Hoggarth, Lentz and Kennett2016; Hoggarth et al. Reference Hoggarth, Culleton, Awe, Helmke, Lonaker, Davis and Kennett2021; Inomata et al. Reference Inomata, Triadan, MacLellan, Burham, Aoyama, Palomo, Yonenobu, Pinzón and Nasu2017; Prufer et al. Reference Prufer, Thompson, Meredith, Culleton, Jordan, Ebert, Winterhalder and Kennett2017; Tsukamoto et al. Reference Tsukamoto, Tokanai, Moriya and Nasu2020; Vadala and Walker Reference Vadala and Walker2020). Since 2014, the Stann Creek Regional Archaeology Project (SCRAP) has investigated such processes at Alabama, an Ancestral Maya townsite in East-Central BelizeFootnote 1 . Unlike many settlements that developed slowly over centuries or millennia, Alabama is a rare example of a “boomtown,” a rapid-growth community settled during the 8th and 9th century CE or the Late to Terminal Classic periods (Peuramaki-Brown Reference Peuramaki-Brown2017). Alabama’s sudden growth appears to coincide with the early decline and disintegration of many other Classic Maya polities, a phenomenon commonly referred to as the “Maya collapse” (Aimers Reference Aimers2007; Culbert Reference Culbert1973; Demarest et al. Reference Demarest, Rice and Rice2004; Dunning et al. Reference Dunning, Beach and Luzzadder-Beach2012; Iannone et al. Reference Iannone, Houk and Schwake2016; Lucero Reference Lucero2002; Webster Reference Webster2002). While the factors contributing to Alabama’s boom may never be fully understood, developing a more detailed understanding of where and when people settled at the townsite enables exploration of the relationship between patterns of settlement growth at local and regional scales and the broader sociopolitical, economic, ideological, and demographic shifts that characterized this period of Ancestral Maya history.
In this article, we present 62 new AMS 14C dates from three residential groups in Alabama’s outlying settlement zone. Using a Bayesian statistical framework, we model these radiocarbon data in stratigraphic sequence to generate detailed developmental and occupational histories for the townsite. This approach addresses challenges in chronology building at Alabama and comparable sites in East-Central Belize, including poor pottery preservation, obscured stratigraphic profiles due to earthen-core construction techniques, and difficulties constraining radiocarbon dates from the Late to Terminal Classic period (ca. 600–900 CE), many of which lie on a plateau/reversal in the 14C calibration curve. We describe our excavations of platform architecture at each residential group, highlighting distinct stratigraphic contexts and their associated AMS 14C dates. The results offer the first directly dated sequence models for East-Central Belize. They provide crucial chronological reference points for understanding the spatiotemporal dynamics of this rapid-growth community and its surrounding region. Insights from this study also offer methodological guidance for integrating excavations of earthen-core architecture with AMS 14C dating, informing future reconstructions of site development and occupational histories along the eastern frontier of the Maya lowlands and beyond.
2. Background
2.1. Archaeological investigations at Alabama, East-Central Belize
Alabama is a major inland Ancestral Maya center of East-Central Belize, a unique material culture sub-region of the eastern frontier of the Maya lowlands (Figure 1; Graham Reference Graham, Evans and Webster2001; Peuramaki-Brown Reference Peuramaki-Brown2017). The sub-region features diverse environmental zones accessible within a day’s walk or paddle, including high-canopy broadleaf forests along rivers and creeks, pine ridges and savannahs, coastal plains, lagoons, and coastal beaches and mangroves. Within this landscape, Alabama occupies the uppermost terrace of an alluvial pocket within the first range of the foothills along the eastern slopes of the Maya Mountains, approximately 20 km inland from the Caribbean coast. This economically and politically strategic position prompted both Graham (Reference Graham1994, 132) and MacKinnon (Reference MacKinnon, McKillop and PF1989a) to propose the site may have served as a “gateway community” (Burghardt Reference Burghardt1971). Indeed, the region has long captivated the attention of scholars interested in economic processes such as resource acquisition and use and the movement of resources and goods within the region and beyond (Dunham Reference Dunham and Fedick1996; Graham Reference Graham1987, Reference Graham1994; Jordan et al. Reference Jordan, Peuramaki-Brown, Chiac, Saqui and Tzib2021; MacKinnon Reference MacKinnon, McKillop and PF1989a, Reference MacKinnon1989b; Mackinnon and Kepecs Reference MacKinnon and Kepecs1989, Reference MacKinnon and Kepecs1991; MacKinnon and May Reference MacKinnon and May1990; Shipley Reference Shipley1978; Stomper et al. Reference Stomper, Brown and Pope2004; Tibbits et al., Reference Tibbits, Peuramaki-Brown, Brouwer Burg, Tibbits and Harrison-Buck2023). Alabama was strategically located not only in terms of its position at a nexus of resource zones and proximity to inland and coastal trade routes but also because of its vicinity to the inland site of Pearce, located ca. 10 km north along an upper tributary of the South Stann Creek and accessible via a gap in the foothills of the Maya Mountains (Peuramaki-Brown and Morton Reference Peuramaki-Brown and Morton2019a). The relationship between Alabama and Pearce is an anticipated focus of future SCRAP research.
Figure 1. Map of central Belize, showing the location of Alabama within the material culture subregion of East-Central Belize and other select archaeological sites.
For the past decade, SCRAP has worked to understand the tone and tempo of Alabama’s development and growth. Building on research by Graham (Reference Graham1994) and MacKinnon (Reference MacKinnon1987, Reference MacKinnon1988a, Reference MacKinnon1988b, Reference MacKinnon, McKillop and PF1989a, Reference MacKinnon1989b; MacKinnon et al. Reference MacKinnon, Olson and May1993), archaeological investigations in the monumental core and at various loci in the outlying settlement zone suggest rapid development at the townsite during the late facet of the Late Classic to Terminal Classic periods (ca. 700–900 CE). Although home to only around 1,000 people at its peak, the townsite exhibited many urban hallmarks typical of larger and longer-occupied settlements (Peuramaki-Brown and Morton Reference Peuramaki-Brown and Morton2019b). Located outside the limestone plains and karst landscapes that characterize much of the Maya lowlands, builders relied on locally available granite, slate, phyllite, quartzite, and various clays and alluvial sediments for constructing platforms and various surfaces (Graham Reference Graham, Evans and Webster2001, 684). Although investigations indicate human presence to some degree in the monumental core during the Early Classic period (ca. 250–600 CE), limited architectural evidence suggests that this settlement was small and relatively short-lived. Today, Alabama’s monumental core lies covered by tropical broadleaf forest, while the settlement zone sits amid an active, industrial-scale orange orchard.
Building chronologies at Alabama has presented several challenges (Peuramaki-Brown et al. Reference Peuramaki-Brown, Morton, Longstaffe and Jordan2023). First, the region’s clay-rich and highly acidic soils, which include alluvium from the valley bottom and hill-wash sediments and soils derived from granites and related rocks of the Cockscomb Granitic Complex (Kesler et al. Reference Kesler, Kienle and Batesom1974; Martens et al. Reference Martens, Weber and Valencia2010; Shipley Reference Shipley1978) coupled with above-average rainfall (Wright et al. Reference Wright, Romney, Arbuckle and Vial1959, 148), result in zero preservation of bone or shell, and exceptionally poor preserved and highly fragmentary pottery assemblages. Pottery sherds recovered from construction core contexts typically lack diagnostic attributes such as well-defined forms, rims and bases, and preserved slip or other surface treatment, limiting our ability to assign types or varieties within established lowland Maya ceramic typologies (e.g., Adams Reference Adams1971; Culbert Reference Culbert1993; Gifford Reference Gifford1976; Sabloff Reference Sabloff1975). While distinctive types, varieties, or, most often, wares are occasionally identified (Jordan et al. Reference Jordan, Peuramaki-Brown, Chiac, Saqui and Tzib2021, 5), these represent exceptions rather than the norm. Consequently, analyses of ceramic assemblages have focused instead on more durable modal attributes such as paste and fabric (Jordan et al. Reference Jordan, Peuramaki-Brown, Chiac, Saqui and Tzib2021).
Stratigraphy plays a crucial role in establishing chronological control over archaeological contexts (Harris Reference Harris2014), and proper interpretation is vital for constructing Bayesian sequences (Bayliss Reference Bayliss2015; Bronk Ramsey Reference Bronk Ramsey2009a; Dye and Buck Reference Dye and Buck2015). However, documenting and understanding stratigraphy at Alabama is challenging due to site formation processes related to the clay-rich earthen materials used in platform construction cores. Builders sourced platform construction materials from nearby borrow pits or local alluvial deposits (Jordan Reference Jordan, Peuramaki-Brown and Morton2019:226-232). Large borrow pits encircle the monumental core (Peuramaki-Brown and Morton Reference Peuramaki-Brown and Morton2019b, 5, Figure 3), and one sits adjacent to the largest outlying settlement site in the orchard, ALA-002. Post-depositional alterations, such as erosion, swelling, heaving, and bioturbation by burrowing animals and tree roots, can complicate distinguishing between stratigraphic layers without extended profile exposures. Variations in soil/sediment colour and texture, along with relative artifact frequencies, aid in distinguishing between construction and occupation contexts, though these boundaries often remain ambiguous. Determining whether changes in soil composition indicate discrete construction events or result from basket-load construction by different task units (Loten and Pendergast Reference Loten and Pendergast1984) proves challenging, except in rare cases where builders buried facing stones during architectural remodeling. In some instances, changes in material selection appear intentional, contributing to building stability or carrying cultural significance (Inomata et al. Reference Inomata, Triadan, Vázquez López, Fernandez-Diaz, Omori, Méndez Bauer, García Hernández, Beach, Cagnato, Aoyama and Nasu2020; Sherwood and Kidder Reference Sherwood and Kidder2011).
Charcoal samples from construction core contexts occasionally produce 14C determinations that are wildly improbable, sometimes dating thousands to tens of thousands of years too old (see Nolan Reference Nolan2012; Schilling Reference Schilling2013; Wallis et al. Reference Wallis, McFadden and Singleton2015 for similar challenges in the American Southeast). These extreme outlier examples almost certainly originate from deeply buried strata containing non-anthropogenic charcoal formed via natural fires. In addition to quarry sites, borrow pits may have also functioned as refuse disposal areas. Therefore, we presume that most cultural materials in construction cores, including pottery and charcoal, are secondary materials redeposited during platform construction (Schiffer Reference Schiffer1987). Radiocarbon data from these contexts remain valuable as they provide maximal ages, or termini post quos, to constrain 14C dates in overlying contexts in the sequence (Bayliss et al. Reference Bayliss, van der Plicht, Bronk Ramsey, McCormac, Healy, Whittle, Whittle, Healy and Bayliss2011; Manning and Birch Reference Manning and Birch2022). Furthermore, when coupled with other lines of evidence, information derived from outlier samples can offer insights into the timing of earlier settlement activity.
Even when calibrated radiocarbon dates align reasonably with known contextual information, they may indicate two or more discrete calendar periods due to wiggles, reversals, or plateaus in the 14C calibration curve (Culleton et al. Reference Culleton, Prufer and Kennett2012, 1585; Manning Reference Manning2024; Manning and Birch Reference Manning and Birch2022; Manning et al. Reference Manning, Birch, Conger and Sanft2020; Meadows et al. Reference Meadows, Rinne, Immel, Fuchs, Krause-Kyora and Drummer2020; Taylor et al. Reference Taylor, Stuiver and Reimer1996, 661-662). This issue is prevalent at Alabama, where the principal period of residential construction and occupation coincides with a minor reversal and plateau. This often results in ambiguous bi-modal Late Classic calendar probabilities or wide date ranges encompassing both the Late Classic and the Terminal Classic periods. These two cultural-historical periods exhibit markedly different dynamics: the former represents the apogee of the Classic period Maya city-states, while the latter signifies their disintegration and the accompanying political, social, economic, ideological, and demographic transformations.
3. Methods
3.1. Sample selection
We selected all samples for radiocarbon dating from excavations at three residential groups (ALA-002, ALA-045, and ALA-047) on the alluvial bottomlands of Alabama’s settlement zone, approximately 1 km to the southwest of the monumental core (Figures 2 and 3). These excavations occurred in 2016, 2018, 2019, 2022, and 2023. We chose samples from various stratigraphic contexts related to platform construction and occupation to estimate the timeframe of cultural activities associated with these settlement sites. We prioritized samples from contexts with clear natural or architectural stratigraphy wherever possible. We collected all samples in situ, recording three-dimensional information (depth, distances from excavation unit walls) along with lot numbers and assigned cultural contexts. We reviewed and, when necessary, adjusted context designations post hoc using our best judgment, informed by stratigraphy, associated material deposits, and other radiocarbon data.
Figure 2. GPS map of settlement zone of Alabama, showing location of settlement units discussed in the text.
Figure 3. Topographic maps of settlement units discussed in the text, indicating locations of excavation units: (top) ALA-002; (middle) ALA-045; and (bottom) ALA-047.
3.2. AMS 14 C dating
Radiocarbon samples were processed at Beta Analytic (USA) and the A.E. Lalonde AMS (Canada) laboratories, following the sample pretreatment and analytical procedures of each laboratory (see https://www.radiocarbon.com/beta-lab.htm and Crann et al. Reference Crann, Murseli, St-Jean, Zhao, Clark and Kieser2017, respectively). Samples analyzed by Beta Analytic are prefixed with “Beta,” while those by the AEL-AMS Laboratory are “UOC.” Most samples selected for dating were single pieces of charcoal, with priority given to small, and thus likely short-lived, pieces of wood (i.e., twigs) to help mitigate the potential for inbuilt age (Dee and Bronk Ramsey Reference Dee and Bronk Ramsey2014) and the “old wood” effect (Schiffer Reference Schiffer1986). Two samples (UOC-16316 and 16317) exhibited < 1 % carbon content, indicating they were not pure charcoal and were consequently analyzed as bulk sediment and may have averaging built into their dates (Ashmore Reference Ashmore1999).
We report all dates in Table 1, with radiocarbon ages calculated as -8033ln (F14C) and reported in 14C yr BP (BP=AD 1950; Stuiver and Polach Reference Stuiver and Polach1977; Reimer et al. Reference Reimer, Brown and Reimer2004). We corrected dates to calendar years (cal yr BP) using OxCal v.4.4 (Bronk Ramsey Reference Bronk Ramsey2009a) and calibrated them using the IntCal 20 calibration curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards and Friedrich2020). Discontinuous ranges are reported in Table 1 and are simplified in the text. Calibrated and modeled dates are 2-sigma calibrated ranges (95.4% probability) unless stated otherwise. Calibrated dates are expressed as “cal AD” or “cal BC,” while calendar and seriational dates (e.g., culture-history periods or ceramic periods) employ “CE” or “BCE.”
Table 1. AMS 14C dates from Alabama’s settlement zone
* Samples identified as outliers within their respective Sequence models were removed from analyses. See supplemental materials.
3.3. Bayesian modeling
Bayesian statistical analyses provide a powerful tool for constraining the probability distributions of calibrated radiocarbon dates and testing the extent to which dates fit interpretive expectations (Bayliss Reference Bayliss2009; Buck et al. Reference Buck, Cavanagh and Litton1996; Hamilton and Krus Reference Hamilton and Krus2018). We utilized Bayesian statistical tools in OxCal 4.4 (Bronk Ramsey Reference Bronk Ramsey2009a) to model AMS 14C dates across six ordered stratigraphic sequence models. These tools allow analysts to incorporate “prior” information, such as stratigraphy, ceramic sequences, and culture-historical frameworks, to constrain sample date ranges. Our Bayesian models make prior information explicit by formalizing assumptions and utilize these constraints to generate narrower probabilistic ranges (Bayliss Reference Bayliss2015, 681; Buck and Meson Reference Buck and Meson2015, 571; Lulewicz Reference Lulewicz2018, 59–60). We use the default uniform prior distribution in OxCal, which assumes all dates within a specified range have equal probability. This reflects the assumption that, without additional constraints, an event (such as platform construction or deposition of refuse) could have occurred at any point within the range. OxCal codes used in the analysis are provided in Supplementary Material 1.
The graphical output of OxCal displays the effects of modeling in the generated histograms, where the original, unmodeled dates are greyed in the background, and constrained (posterior) ranges are darkened in the foreground. The models’ posterior estimates should be considered interpretive rather than absolute and may change as new excavations are carried out and dates from additional contexts are added to the models. Individual runs of models will have very slightly different results. The modeled output includes each date’s refined posterior probability distribution and indicators regarding their congruency with the model, enabling the investigation of problematic dates as potential outliers. Agreement indices (A) are generated for both the model (A-model) and the posterior distributions of each radiocarbon date, with a scale of approximately 0–120%, with a cutoff at 60% (A’c = 60) indicating a poor fit (analogous to 0.05 significance level in a χ2 test; Bronk Ramsey Reference Bronk Ramsey2009b). It is important to note that an agreement index above A’c does not necessarily confirm the model’s “correctness;” instead, it signifies a good fit between the model and the data, and it is up to the analyst to determine whether the agreement is significant (Culleton et al. Reference Culleton, Prufer and Kennett2012, 1577). Agreement indices aid in identifying outliers, which the analyst can manually eliminate based on expert knowledge of the depositional processes relevant to the archaeological context.
Formal outlier analysis methods are more sophisticated and entail assigning a prior probability to the likelihood that a given measurement, event timing, or sequence placement is incorrect. These models allow the analyst to determine appropriate revisions after considering all available information. Following Bronk Ramsey (Reference Bronk Ramsey2009b, 1024–1028), first, we manually rejected the most apparent outliers. To identify additional outliers using statistical methods, we applied a “General” temporal outlier model (Bronk Ramsey Reference Bronk Ramsey2009b, 1028; Dee and Bronk Ramsey Reference Dee and Bronk Ramsey2014). This model uses a long-tail Student’s t distribution with 5 degrees of freedom to detect outliers by assigning a probability that a radiocarbon date is anomalous due to various issues such as contamination, redeposition, or incorrect contextual association. When the outlier model is applied, radiocarbon determinations identified as outliers—based on their consistency with all other available information, including other radiocarbon dates and model parameters—are down-weighted, reducing the influence of these dates on the overall chronology. If outliers are retained in the model, the results essentially represent a sort of “averaging” between a model that accepts the date and one that rejects it (Bronk Ramsey Reference Bronk Ramsey2009b, 1024). We removed the identified outliers to avoid such averaging and reran the model. This was an iterative process; in some cases, removing one date from the sequence introduced new outliers. Once satisfied, we removed the outlier analysis code to generate the stratigraphic sequence models presented in this article. Results of each step of the outlier analysis are provided in Supplementary Material 2. Unsurprisingly, most outlier dates are associated with construction core contexts containing redeposited secondary materials or other contexts affected by site formation processes that moved samples out of their original context. While we assume that most, if not all, 14C determinations from construction core contexts stem from secondary materials and thus should be interpreted as termini post quos for platform construction, outliers were those dates that had the most detrimental effect on the overall model agreement.
Culleton and colleagues (Reference Culleton, Prufer and Kennett2012, 1577, their table 3) summarize the basic commands of OxCal and detail the relevant stratigraphic contexts to which these commands apply (see also Bronk Ramsey Reference Bronk Ramsey2009a; Hamilton and Krus Reference Hamilton and Krus2018). We used Sequences to establish the basic ordered stratigraphic structure for each model. We modeled strata separating directly dated contexts as Boundaries, estimating the probabilistic start and end times of events that are not directly dated. Our placement of boundaries relied on stratigraphic information, including changes in soil matrices, artifact density and composition, or clear examples of architectural modifications or additions. We grouped all radiocarbon dates belonging to the same stratigraphic deposit (e.g., construction core, occupation horizon, refuse deposit) within a Phase, providing a container for otherwise unordered dates within the ordered Sequence and giving each date equal weighting. OxCal allows users to choose among different types of relationships between groups coded as phases (contiguous, overlapping, or sequential), each of which treats the data differently (Bronk Ramsey Reference Bronk Ramsey2009a, 348). Contiguous phases refer to groups that follow one after another with a transition event and use the same boundary for the end of one phase and the start of the next. Sequential phases refer to cases where there is a hiatus between groups, but with the placement of an additional boundary to indicate the beginning of the second phase comes after the end of the first. Overlapping phases treat groups of dates within the sequence independently and allow for the modeling of contexts where one phase might start before the previous one has ended. In some instances, we applied the After command to define terminus post quos within models to incorporate prior ceramic data. We justify each instance below.
4. Results
4.1. Settlement Unit ALA-002
ALA-002 is one of Alabama’s largest outlying settlement units, featuring three platforms arranged around a 2,250 m2 raised boulder and cobble core plaza (Figure 3a). The two largest platforms, ALA-002A (2.5 m tall) and ALA-002B (2.75 m tall), are relatively well preserved, while orchard activities have partially disturbed ALA-002C, a 0.5 m tall L-shaped platform. We conducted horizontal and stratigraphic excavations at all three platforms and placed a 1 m × 1 m test unit in the plaza (Peuramaki-Brown et al. Reference Peuramaki-Brown, Blaine, Chiac, Peuramaki-Brown and Morton2022, forthcoming). Additionally, we carried out 258 shovel test excavations (0.5 m × 0.5 m) on a 5 m × 5 m grid across the plaza and to the sides and rears of the platforms (Longstaffe Reference Longstaffe, Peuramaki-Brown and Morton2022). These investigations identified three distinct occupation components: a buried occupation that dates to the Early Classic, a Late Classic to Early Postclassic occupation represented by the visible platform architecture, and a reoccupation of two platforms during the Late Postclassic period.
4.1.1. Early Classic occupation
Small quantities of Early Classic ceramics and outlier charcoal AMS 14C dates in the group’s Late Classic period platform construction cores suggest an earlier occupation near ALA-002. Although we have not identified Early Classic architecture in Alabama’s settlement zone, we documented cultural deposits that we posit date to this period in stratigraphic profile windows at each platform and below the plaza core, buried beneath 25–30 cm of sterile sediment. Compared to permanent survey monuments, the absolute depths of these spatially separated contexts show only a few centimetres of difference relative to the current ground level, indicating they likely represent a single, coherent, stratigraphic level. One deposit, a ca. 10 cm thick artifact-bearing layer of olive-yellow to yellowish-brown sediment behind ALA-002A, included an Early Classic basal flange with preserved red and black paint on an orange/cream slip. A similarly buried layer at ALA-002C contained artifacts, though it lacked diagnostic ceramics. Additionally, we documented eight blackened sand features buried beneath the plaza and Step 1 of ALA-002B, with the uppermost containing unidentifiable ceramics. No carbon was found in these features or floated sediment samples. Assuming the outlier charcoal samples in the construction cores resulted from human activity coeval with these deposits rather than natural processes, they may provide valuable insights into the timing of this earlier occupation. We modeled seven outlier dates as an unordered phase with boundaries marking the start and end of activity (Table 2). Using OxCal’s Date function, we estimate this activity between cal AD 345–545. An older age determination (UOC-23599: 1950 ± 20 BP, cal AD 15–125) from charcoal in terrace construction core at the back of ALA-002B was identified as an outlier at 79% probability and excluded from this analysis.
Table 2. Bayesian model of Early Classic radiocarbon ages and calibrated date ranges
Amodel = 110.2; Aindex = 111.9
4.1.2. ALA-002A
Excavations at ALA-002A revealed contexts associated with platform construction and ritual or ceremonial use of the structure (Table 3, Figure 4). The platform’s function is inferred from its location on the plaza’s east side, square base, a tall granite slab–potentially an uncarved stela–toppled at mid-stair level, and the relative scarcity of residential artifacts (Chase and Chase Reference Chase, AF and SD1998). Activity around the platform began between cal AD 530–765, but likely in the Late Classic, between cal AD 620–750 (1-sigma). Before constructing the platform, builders levelled an occupation horizon to create a stable surface, represented by a ca. 5 cm-thick layer of tamped olive-brown loamy sand, extending beneath the platform, as revealed by a 1 m-wide profile exposure (Figure 4b). A 3 cm-thick deposit of carbonized materials, possibly from a ritual burning event, was observed directly beneath the platform atop the occupation horizon and dated to cal AD 650–760 (UOC-12578). The ancient builders piled earthen material over this deposit to form the platform’s construction core. In the same profile, fallen stones and colluvium were removed, exposing the lowest, heavily slumped four-course high terrace of the platform’s rear façade.
Table 3. Modeled results for the ALA-002A stratigraphic sequence
Amodel = 77.9; Aindex = 76.7
Figure 4. (a) Bayesian model for ALA-002A. Letters correspond to radiocarbon samples plotted on profiles. Blue shading shows commands used for Bayesian modeling in Oxcal; (b) profile of east (rear) side of ALA-002A, showing locations of radiocarbon samples; (c) profile of west (front) side of ALA-002A, showing locations of radiocarbon samples. For both (b) and (c) outliers mentioned in the text are in red and labelled. Colored stones are in place or slightly slumped.
Three AMS 14C dates from charcoal in the construction core indicate platform construction during the Late Classic. The stratigraphically highest sample, from just below a surface of small alluvial cobbles–possibly the base of a terrace surface at the level of the superior course of the lowest terrace of the rear platform face–dates to cal AD 680–770 (UOC-16274). The second sample, recovered behind the basal course of the same terrace, dates to cal AD 675–775 (UOC-16252). The third sample is from Feature 2, a dense deposit of carbon and fragmented ceramic sherds associated with three squared granite slabs within the construction core toward the platform’s top, dating to cal AD 680–770 (UOC-23596). This sample likely did not form in situ, as neither the ceramics nor the surrounding soil show signs of burning. The slabs associated with this feature are aligned north to south, perpendicular to the building’s central axis, and may serve as capstones for a yet-to-be-excavated primary deposit. A smashed Platon Punctated-incised tripod dish with rattle feet and preserved red slip was found on an adjacent granite slab, likely an offering still in its primary context. These standardized British Honduras Ash Ware vessels–simple outflaring or outcurving dishes with incised lines near their rims or bases–were produced in the Belize River Valley throughout the Late and Terminal Classic periods (Chase and Chase Reference Chase and Chase2015; Gifford Reference Gifford1976, 257; Jordan et al. Reference Jordan, Davenport, Goodwin, MacDonald, Ebert, Hoggarth and Awe2022; LeCount et al. Reference LeCount, Yaeger, Leventhal and Ashmore2002). At centers in the Toledo District to the south, British Honduras Ash Wares did not become common until after 780 CE, making them an important diacritical marker of the onset of the Terminal Classic (Braswell Reference Braswell and Braswell2022a, 88). As such, their presence in Late Classic contexts at Alabama suggests that residents may have had stronger or earlier trade ties to the Belize Valley than did contemporaneous settlements farther south.
Above Feature 2, excavators documented a pile of cobbles, designated as Feature 1, within the earthen construction core. The purpose of this feature remains to be determined; it could be a cluster of alluvial stone inclusions or a marker for the location of Feature 2, among other possibilities. Early facet Late Classic AMS 14C dates from this feature (UOC-23595: 1360 ± 20 BP, 93% probability of cal AD 645–680) and from the construction core just above the carbon layer atop the occupation horizon (UOC-16253: 1420 ± 30 BP, cal AD 590–660) were excluded from the model as outliers due to low agreement indices. Despite their exclusion, these outlier dates highlight the possibility of earlier activities elsewhere in the settlement unit during the Late Classic, prior to ALA-002A’s construction. We suspect that ALA-002B was the first platform constructed at the settlement site (see results below) and that ALA-002A was a later addition to the group.
A 1-m-wide profile exposure revealed that the platform’s lowest masonry courses were buried beneath the boulder and cobble core of the plaza pavement, indicating that the platform predates the formal plaza (Figure 4c). Initially, the natural ground surface between the group’s platforms functioned as the plaza. This is evidenced by an artifact-bearing, yellowish-brown sandy clay stratum just below the terminal plaza construction core. Pockets of habitation debris and several features were identified at various loci across this surface. A separate refuse deposit, directly associated with ALA-002A, formed a 35–40 cm thick stratified layer atop the occupation horizon, abutting the face of the lowest rear platform terrace. Notable differences were observed between the lower and upper sections of the matrix, with upper layers containing numerous carbon flecks, friable ceramics, and heat-exposed chert (pink/red with potlid scars), indicating a significant burning event toward the end of the structure’s use. Additional refuse was found in isolated deposits intermingled with fallen and slumped granite stones along the back and front of the platform.
Without a way to stratigraphically link these spatially separated contexts–specifically, the refuse dispersed across the penultimate plaza and the refuse associated with the platform–they were modeled as two independent overlapping phases. This approach allowed for independent estimates of the timing of refuse accumulation at ALA-002A relative to the refuse accumulation and features associated with the original plaza surface, as well as the subsequent construction of the cobble and boulder plaza, without making assumptions about the temporal sequence or stratigraphic relationship between these events.
The first of these phases incorporated AMS 14C dates associated with ALA-002A platform refuse, including a charcoal sample from between fallen architectural blocks on the front of the structure, dated to cal AD 710–825 (UOC-23598) and two others from the stratified refuse deposit. One sample from the deposit’s upper layer, just below the slumped third course of the exposed platform terrace (A4, stone C2), dates to cal AD 720–820 (UOC-12673), while another from the lower portion has a similar date range, at cal AD 715–820 (UOC-12684). Both dates extend from the late facet of the Late Classic into the Terminal Classic period. These dates are corroborated by temporally diagnostic ceramics recovered from the deposit, including Dolphin Head Red dishes and bowls, British Honduras Volcanic Ash Wares, Hondo Red, and other Late and Terminal Classic forms and modes, including a solid nubbin foot (Fauvelle et al. Reference Fauvelle, Pitcavage and Braswell2012; Gifford Reference Gifford1976; Hammond Reference Hammond1975; Jordan and Prufer Reference Jordan and Prufer2017; Jordan et al. Reference Jordan, Peuramaki-Brown, Chiac, Saqui and Tzib2021, 5). A broad boundary estimate suggests subsequent refuse deposition continued until some point between cal AD 720–840. Unlike other structures in the group, ALA-002A has no confirmed examples of Early Postclassic pottery. Combined with evidence of burning observed in the stratified refuse deposit, this suggests that the platform was abandoned or decommissioned during the Terminal Classic period while others in the group remained in use.
In parallel, the second overlapping phase utilized data from shovel test excavations across the formal raised boulder and cobble core plaza. Excavations identified three features containing datable carbon. Two of these features, clusters of habitation debris, date to cal AD 755–830 (UOC-16317) and cal AD 735–830 (UOC-12579), respectively (Figure 5). While this latter sample exhibits low agreement (A=53.6%), there is no justification for rejecting this date, as the context is secure beneath the plaza core. The sample likely represents a deposit formed just prior to plaza construction activities (94.9% probability of cal AD 755–830). Another feature, located near the plaza’s southern margin, consists of a large mass of daub surrounded by river cobbles with charcoal and dates to cal AD 775–825 (UOC-12576).
Figure 5. Plaza shovel test excavations: (a) STP69, possible posthole/daub feature, showing enclosure (left) and daub mass (right); (b) STP166, remains of artifact/carbon cluster after excavation; (c) examples of shovel test excavations, on plaza (top [STP197]) and off plaza (bottom [STP92]).
Isolated refuse deposits beneath the plaza core contained moulded-carved ceramics, which first appeared in this part of the Maya lowlands at the onset of the Terminal Classic, no earlier than 800 CE (Helmke and Reents-Budet Reference Helmke and Reents-Budet2008; Rice and Forsyth Reference Rice, Forsyth, Demarest, Rice and Rice2004; Ting and Helmke Reference Ting and Helmke2013, 43; Ting et al. Reference Ting, Martinón-Torres, Graham and Helmke2015). While incorporating ceramic priors into Bayesian models is generally discouraged unless the ceramics bear hieroglyphic dates, moulded-carved pottery is a highly diagnostic decorative style with well-documented temporal attributes. For instance, Ahk’utu’ Moulded-carved pottery, the most widely distributed type in the eastern Peten and central Belize, includes a patronage statement referencing Ixolom, or “Lady Olom,” an elite figure whose rise to power between 810 and 830 CE is recorded in stone inscriptions at Uaxactun, providing a temporal anchor for the earliest of these vessels (Helmke and Reents-Budet Reference Helmke and Reents-Budet2008, 43). Although the moulded-carved sherds in this context could not be identified to type, their secure stratigraphic placement beneath the plaza core, combined with known temporal attributes, strongly supports their inclusion as a terminus post quem in the model. However, we adopted a conservative approach, using a cut-off of 800 CE to represent the earliest probable appearance of moulded-carved ceramics in the region (Ting et al. Reference Ting, Martinón-Torres, Graham and Helmke2015). Incorporating this prior into the model constrains an AMS 14C date from the plaza ballast to cal AD 795–840 (UOC-16254). Plaza activity is estimated between cal AD 795–860, though it likely continued into the Early Postclassic associated with activity at ALA-002B and ALA-002C (discussed below). The overall ALA-002A sequence concludes with a boundary marking the latest probable activity at the structure, estimated to be between cal AD 795–915, but likely between cal AD 805–850 (1-sigma), based on data from both the structure and plaza.
4.1.3. ALA-002B
The sequence for ALA-002B includes platform construction events beginning in the Late Classic, occupation extending through the Early Postclassic, and reoccupation in the Late Postclassic (Table 4, Figures 6 and 7). Artifact assemblages and architectural features suggest the platform functioned as a domestic structure, likely serving as the group’s principal residence during the Late Classic to Early Postclassic occupation. The initial boundary estimate for the earliest possible activity is poorly constrained between 100 cal BC and cal AD 645, although a Late Classic period estimate is likely (cal AD 400–630, 1-sigma). A boundary representing an occupation horizon, marked by a ca. 5 cm-thick tamped, olive-brown, loamy sand layer extending from in front of the platform stair to beneath the bottom course of Step 1, is estimated between cal AD 435–650. Initial platform construction activity is estimated to be between cal AD 590–655, during the early facet of the Late Classic.
Table 4. Modeled results for the ALA-002B stratigraphic sequence
Amodel = 89.8; Aindex = 91.5
Figure 6. Bayesian model for ALA-002B. Blue shading shows commands used for Bayesian modeling in OxCal. Letters correspond to radiocarbon samples plotted on profiles on Figure 7.
Figure 7. Profiles of excavations at ALA-002B showing locations of radiocarbon samples: (a) front face, stair, and adjacent plaza pavement; (b) top of mound; and (c) back of mound. On all profiles outliers are in red. Colored stones are in place or slightly slumped.
A 1-m-wide profile exposure in the front stair suggests at least two phases of platform construction. The earliest architecture consists of two single-course high stone alignments (A2 and A4, visible in profile) atop the occupation horizon (Figure 7a). These alignments may be remnants of a step or low terrace from an earlier platform. The area between these alignments, at the level of the top of the physically lowest line of stones, yielded abundant pottery, including several large, well-preserved, horizontally laying sherds, suggestive of a refuse deposit. This assemblage includes Late or Terminal Classic diagnostic forms and types, including an example of a Benque Viejo Polychrome bowl or dish, a type that virtually disappeared by the Terminal Classic (LeCount Reference LeCount1999; LeCount et al. Reference LeCount, Yaeger, Leventhal and Ashmore2002). Carbon samples from this context could not be dated. Since it is unclear whether this deposit is linked to an earlier platform predating the stair or if it is simply a well-preserved pocket of secondary refuse within the construction core, we modeled the platform construction as a single, unordered phase. Eight AMS 14C samples from the construction core date to the early facet of the Late Classic or shortly thereafter (Table 4). A boundary estimates the onset of platform occupation between cal AD 685–760. Interestingly, OxCal’s Difference function calculates a 35 to 150-year gap between the boundary estimates for the onset of platform construction and platform occupation, supporting the hypothesis that the platform was constructed in two phases. Outlier dates from the construction core (UOC-23616: 1490 ± 20 BP, cal AD 550–640; UOC-23613: 1480 ± 20 BP, cal AD 560–640) provide additional evidence for earlier activities at the settlement unit, potentially related to the probable penultimate phase of occupation.
As at ALA-002A, much of the first riser of the front stair of the platform was covered by the boulder and cobble core of the plaza pavement, indicating that it predates the plaza (Figure 7a). Thus, we used the same modeling approach as ALA-002A (see above), treating platform habitation debris and spatially separated plaza contexts as separate but overlapping phases of activity. Using the same radiocarbon data and stratigraphic sequence for the plaza phase provides an independent check on the date estimates provided by the ALA-002A model. The results are similar, though they allow for the possibility of slightly earlier activity (compare Table 3), with features associated with the penultimate earthen plaza surface returning modeled dates of cal AD 735–830 (UOC-16317), cal AD 725–830 (UOC-12579) and cal AD 770–835 (UOC-12576). The estimated timing of plaza construction is also similar (cal AD 795–845; UOC-16254), as is the estimate for plaza activity, between cal AD 800–860.
We modeled three AMS 14C dates as an unordered phase representing refuse accumulation at the platform. These dates include a charcoal sample recovered from just below the plaza core immediately in front of the S1 riser, dated to cal AD 720–775 (UOC-12582), and two charcoal samples found in pockets of refuse on the back of the platform among fallen granite ashlars from the collapsed rear platform face, dated to cal AD 720–775 (UOC-23615) and cal AD 705–780 (UOC-23618), respectively. Ceramics associated with these refuse deposits comprise a mixture of Late or Terminal Classic types and forms. These include an example of McRae Impressed, a distinctive serving vessel with a notched apron that dates to the Terminal Classic (LeCount et al. Reference LeCount, Yaeger, Leventhal and Ashmore2002) and a sherd from a Fat Polychrome vessel, a type characterized by its distinctive P-shaped bolster rim and bold black-and-red painted designs, originating from regions further north during the Terminal Classic period (Masson and Mock Reference Masson, Mock, Demarest, Rice and Rice2004, 387; Mock Reference Mock1994, 106–107; see also Harrison-Buck Reference Harrison-Buck2024, 419, their fig 11c–f). We also recovered Early Postclassic forms, such as sherds from thumb-impressed fillet bowls, a tall ring base, and a scroll foot. These materials indicate that occupation continued into the early 10th century, supporting the final boundary estimate for the latest possible platform use between cal AD 815–1250. However, the 1-sigma estimate (cal AD 830–1060) is more likely.
We identified Late Postclassic reoccupation of the platform through a layer of near-surface ceramics and other artifacts stratigraphically separated from earlier habitation debris and surfaces by a thin lens of humus. The reoccupation assemblage is diverse, featuring well-preserved pottery resembling crude Postclassic and Historic period materials found to the south in the Toledo District and coastal sites to the north. It also includes notched ovoids or “date-seed” net sinkers (Graham Reference Graham1994, 47, 305, their fig 8.18v-w; see also MacKinnon Reference MacKinnon1989b, 501, 637), as well as numerous small side-notched projectile points made of recycled chert and obsidian (for comparison see Oland Reference Oland2013, 89, their fig 4; Simmons Reference Simmons1995, 139, their fig 6, 2002, 55, their fig 4). Eight AMS 14C dates from charcoal stratigraphically associated with this assemblage confirm a Late Postclassic reoccupation date, beginning between cal AD 1325–1385 and ending between cal AD 1430–1475 (Table 4). An additional AMS 14C date from this context excluded from the model as an outlier at 62% probability (UOC-23611: 335 ± 15, cal AD 1490–1635) may suggest continued habitation or visitation to ALA-002B after Spanish contact.
Figure 8. (a) Profile of excavation at ALA-002C, showing locations of radiocarbon samples. Letters correspond to radiocarbon samples plotted on profile. Outliers mentioned in the text are in red. Colored stones are in place or slightly slumped; (b) Bayesian model for ALA-002C, modeled in OxCal. Blue shading shows commands used for Bayesian modeling in OxCal.
4.1.4. ALA-002C
The low, elongated L-shaped architectural form of ALA-002C and its domestic artifact assemblages suggest this platform may have been an ancillary building or secondary residence at the group, likely beginning in the Late Classic period. Due to a limited number of datable carbon samples from non-habitation debris contexts, stratigraphic models could only provide broad estimates for the timing of platform construction. An Early Classic date from the construction core (UOC-23621: 1710 ± 20 BP, cal AD 255–410) likely reflects redeposited secondary charcoal. The earliest pottery associated with platform activities dates to the Late or Terminal Classic and was recovered from a 25–30 cm-thick deposit of habitation debris adjacent to the exterior of the platform’s buried three-course-high north (rear) face (Figure 8a). This deposit contained diagnostic British Honduras Ash Wares, Hondo Red, and body sherds from an unidentified, imported, painted red-and-black-on-buff/cream thin-walled polychrome vessel.
We modeled five AMS 14C dates from this deposit in an unordered phase within a sequence (Table 5). Four of these dates stemmed from two samples physically large enough to be split into two samples each (UOC-23624 and UOC-23625, UOC-23622 and UOC-23623). We aggregated these groups of dates before calibration using Oxcal’s R_Combine function and verified their internal consistency with a chi-square test. This confirmed firm Late Classic dates of cal AD 670–775 (UOC-23622 and UOC-23623) and cal AD 655–775 (UOC-23624 and UOC-23625). The fifth sample dates much later, entirely within the Early Postclassic, between cal AD 1025–1155 (UOC-23619). As this sample was the stratigraphically lowest in the deposit, it likely reflects some mixing of contexts due to bioturbation or other disturbances. Early Postclassic ceramics recovered from mixed habitation debris and colluvium contexts near the platform surface suggest this sample was initially associated with these contexts. The initial and closing boundaries in the model are poorly constrained and uninformative when estimating the start and end of activity.
Table 5. Modeled results for the ALA-002C stratigraphic sequence
Amodel = 94; Aindex = 94
Similar to ALA-002B, a stratigraphically higher level separated from earlier habitation debris by a thin layer of humus at the rear of the ALA-002C platform yielded a dense deposit of Postclassic notched ceramic ovoids (n = 59). A charcoal sample from this layer returned a Historic period date (UOC-23626: 335 ± 15 BP, cal AD 1490–1635). While this date might accurately reflect the age of this deposit, additional data is needed to conclusively determine an association with the more securely dated Late Postclassic occupation context described for ALA-002B.
4.2. Settlement Unit ALA-045
ALA-045 is a small patio group consisting of three orthogonally arranged platforms: ALA-045A, ALA-045B and ALA-045C (Figure 3b). Excavations at this settlement site included both horizontal and stratigraphic investigations at ALA-045A (0.75 m tall) and ALA-045C (0.5 m tall; Morton and Delos Reyes Reference Morton, Delos Reyes, Peuramaki-Brown and Morton2018; Pennanen and Peuramaki-Brown Reference Pennanen, Peuramaki-Brown and Peuramaki-Brown2016). These investigations revealed the material remains of a commoner residential site featuring low granite-faced platforms and utilitarian objects such as chipped-stone tools and pottery, dating from the Late or Terminal Classic to the Early Postclassic period. ALA-045B remains unexcavated and is barely visible on the surface, primarily identified by concentrations of artifacts and architectural blocks. Only excavations at ALA-045A yielded sufficient AMS 14C dates to develop a stratigraphic sequence model.
4.2.1. ALA-045A
The ALA-045A model incorporates four AMS 14C dates and uses stratigraphic and contextual information to clarify the timing and nature of the platform’s construction and occupation (Table 6, Figure 9). Initial activity at the platform is estimated between cal AD 735–880. The construction process mirrors patterns observed at other sites in Alabama’s settlement zone, with builders first levelling and tamping the underlying occupation horizon. This layer consists of compact sandy clay loam with fine gravel inclusions, distinct from the surrounding construction core and visible in profile extending beneath the earliest platform construction. Ceramics from this context included a well-preserved Pabellon Moulded-Carved sherd, dating to no earlier than 800 CE, firmly within the Terminal Classic period (Helmke and Reents-Budet Reference Helmke and Reents-Budet2008, 43). This ceramic prior constrains a radiocarbon date from the occupation horizon to cal AD 795–870 (UOC-16251).
Table 6. Modeled results for the ALA-045A stratigraphic sequence
Amodel = 100; Aoverall = 99
Figure 9. (a) Profile of excavation at ALA-045A, showing locations of radiocarbon samples. Letters correspond to radiocarbon samples plotted on profile. Outliers mentioned in the text are in red. Colored stones are in place or slightly slumped; (b) Bayesian model for ALA-045A, modeled in OxCal. Blue shading shows commands used for Bayesian modeling in OxCal.
ALA-045A underwent at least three distinct phases of construction and renovation. Initially, builders erected a small, low platform featuring a single course of unshaped granite boulders forming a southern face atop the occupation horizon between cal AD 805–875. Remnants of a coarse sand-like material suggest that this platform featured an exterior crushed granite surface, similar to one documented at Maintzunun in the northern end of the sub-region (Graham Reference Graham1994, 133). Subsequently, between cal AD 815–890, builders added a low, curved stone alignment of stacked unshaped cobbles and boulders adjacent to the south face of the original platform. The presence of Terminal Classic to probable Early Postclassic diagnostic ceramics within this construction suggests it was built toward the latter end of this posterior estimate. Initially thought to mark the edge of a platform or an uncapped burial cist, further excavation revealed its full areal extent, ruling out an interment function. The nature of this feature remains uncertain, although the arced line of cobbles may delineate a construction cell or stabilization feature for a larger, effaced platform. However, such configurations have yet to be documented at Alabama. Alternatively, it could represent the remnants of a small circular platform extension intended as a ritual node within the residential group. The platform construction core yielded two small, hardened nodules of copal resin and a whole quartz crystal, symbolically charged ritual objects (Brady et al. Reference Brady, Cobb, Garza, Espinosa, Burnett, Prufer and Brady2005; Brady and Prufer Reference Brady and Prufer1999). Circular structures dating to the Preclassic period (ca. 1000 BCE-300 CE) are known across the Maya lowlands (Aimers et al. Reference Aimers, Powis and Awe2000; Hendon Reference Hendon2000; MacLellan and Castillo Reference MacLellan and Castillo2022). However, circular shrine architecture also began appearing at sites near the Caribbean coast of Belize starting in the early 9th century CE, thought to reflect increasing interactions with Chontal-Itza groups of the northern Yucatan Peninsula (Harrison-Buck Reference Harrison-Buck2012, Reference Harrison-Buck2024; Harrison-Buck and McAnany Reference Harrison-Buck and McAnany2013; Ringle et al. Reference Ringle, Gallareta Negrón and Bey1998; Rosenswig and Masson Reference Rosenswig and Masson2002). If this feature represents an expression of Yucatec-influenced ritual architecture, it suggests that Alabama was part of this broader regional phenomenon. This interpretation is bolstered by recovered pottery that is stylistically and petrographically similar to that found in northern Belize during this period (Howie and Jordan Reference Howie, Jordan, Peuramaki-Brown and Morton2018, 81–82). This circular feature was buried within the terminal ALA-045A platform between cal AD 830–895 (UOC-16249). An additional radiocarbon date recovered from this construction core is an outlier stemming from old charcoal deposited with quarried earthen construction material (UOC-16250: 11600 ± 35 BP, 11,630–11,405 cal BC).
The occupation of the terminal platform began between cal AD 850–960, either in the Terminal Classic or the Early Postclassic. The final two AMS 14C dates from the sequence come from a terminal habitation debris deposit located off the southern edge of the platform. We modeled these samples in an unordered phase, yielding similar age determinations: cal AD 885–985 (Beta-457820) and cal AD 885–990 (Beta-457818). A boundary estimate for subsequent refuse deposition between cal AD 890–1045 is supported by Terminal Classic and Early Postclassic ceramics found in terminal habitation debris and post-abandonment fall contexts. The latest use of ALA-045A is estimated between cal AD 885–1240, though cal AD 900–1010 (1-sigma) is more likely.
4.3. Settlement Unit ALA-047
ALA-047 comprises four platforms, three of which are arranged orthogonally around a central patio: ALA-047A (ca. 2 m tall), ALA-047B (1 m tall), and ALA-047C (0.8 m tall). The fourth platform, ALA-047D (1.2 m tall), is a short distance to the south (Figure 3c). Despite disturbances from orchard activities, we conducted horizontal and stratigraphic excavations at each platform (Morton et al. Reference Morton, Oliveira, Williams and MM2016). These investigations yielded artifact assemblages, including everyday domestic items and implements such as mano and metate fragments and utilitarian pottery, as well as imported and exotic goods such as a jadeite pendant, large quantities of obsidian, fine chert bifaces, moulded-carved ceramics, and fragments of effigy censers. While ALA-047A was effaced, the granite facing stones of ALA-047C were noted for their excellent quality. These findings suggest a commoner household of greater than modest means active during the Late or Terminal Classic and into the Early Postclassic. We obtained sufficient AMS 14C data at ALA-047A and ALA-047B to produce stratigraphic sequence models. However, we could not create a model for ALA-045C, though charcoal associated with its habitation debris dates to the Late Classic period (Beta-457816: 1360 ± 30 BP, cal AD 650–775).
4.3.1. ALA-047A
The sequence for ALA-047A estimates the timing of platform construction and occupation (Table 7, Figure 10). Limited radiocarbon data result in poorly constrained initial and closing model boundaries, rendering them largely uninformative. A boundary estimate for the platform’s occupation horizon was not included in the final version of the model as it extended at least two millennia before any likely cultural activity at the settlement site. The boundary estimate for the onset of platform construction (cal AD 100–770) is similarly uninformative, though the 1-sigma estimate (cal AD 535–750) is plausible. Small quantities of pottery in the construction core can be stylistically dated anywhere from the Early Classic to the Terminal Classic. We modeled two AMS 14C dates from charcoal in the construction core as an unordered phase. The stratigraphically lowest sample dates to cal AD 665–830 (UOC-16248). The other sample (Beta-457817) dates to cal AD 615–775 and was recovered alongside a cluster of ceramics just below a concentration of cobbles at the junction of the construction core and the effaced south face of the platform. This deposit included two crude censer fragments, stylistically attributable to any time between the Early Classic and the Early Postclassic periods, and an Early Classic jar fragment with an outcurving rim and rounded lip.
Table 7. Modeled results for the ALA-047A stratigraphic sequence
Amodel = 92.4; Aindex = 91.9
Figure 10. (a) Profile of excavation at ALA-047A, showing locations of radiocarbon samples. Colored stones are in place or slightly slumped; (b) Bayesian model for ALA-047A, modeled in OxCal. Letters correspond to radiocarbon samples plotted on profile. Blue shading shows commands used for Bayesian modeling in OxCal.
The model estimates occupation at ALA-047A between cal AD 680–1160. This broad timeframe is supported by pottery found in habitation debris contexts, which include types and forms from the Late Classic, Terminal Classic, and Early Postclassic, including a hand-modeled foot from a Postclassic effigy censer. An AMS 14C date from charcoal associated with slumped granite blocks near the top of the platform dates to cal AD 1040–1215, providing radiometric evidence for Early Postclassic activity.
4.3.2. ALA-047B
The sequence for ALA-047B includes four AMS 14C dates from various strata associated with platform construction and occupation (Table 8, Figure 11). The initial model boundary estimates the onset of activity between cal AD 575–820. We obtained two AMS 14C dates from the platform’s occupation horizon: a bulk sediment sample dated to cal AD 665–820 (UOC-16316) and a charcoal specimen dated to cal AD 670–820 (Beta-456254) from the juncture of the occupation horizon and the base of Feature 1 (discussed below). Diagnostic ceramics from the occupation horizon, stylistically dating to the Late or Terminal Classic, support these posterior probability distributions. A boundary estimates that the ALA-047B platform was constructed atop the occupation horizon between cal AD 680–880. Habitation debris subsequently accumulated off the platform’s edge, forming a ca. 25–30 cm thick layer of refuse between cal AD 735–990. This deposit included Terminal Classic pottery, including clear examples of Ahk’utu’ Moulded-Carved vase sherds (Helmke and Reents-Budet Reference Helmke and Reents-Budet2008; Ting Reference Ting2018; Ting et al. Reference Ting, Martinón-Torres, Graham and Helmke2015). These materials may have been pushed downwards by bioturbation from Feature 1, which overlies the habitation debris and features moulded-carved ceramics. A tap root hole containing cultural materials, visible in profile, tapers through sterile sediment, potentially indicating mixing (Figure 11a).
Table 8. Modeled results for the ALA-047B stratigraphic sequence
Amodel = 110.8; Aindex = 109.7
Figure 11. (a) Profile of excavation at ALA-047B, showing locations of radiocarbon samples. Colored stones are in place or slightly slumped; (b) Bayesian model for ALA-045A. Letters correspond to radiocarbon samples plotted on profile. Blue shading shows commands used for Bayesian modeling in OxCal.
Feature 1 is a large (ca. 70 cm E-W, 50 cm N-S), amorphously shaped concentration of daub situated just off the westernmost platform face. This feature could represent a collapsed daub structure, such as a kiln or an oven, or a collection of refuse potentially linked to ritual activities. Excavators carefully removed this feature in four quadrants, revealing two distinct strata: a compact bottom layer of daub and an upper layer of loose daub and soil. The base of Feature 1 was distinct from the underlying occupation horizon and habitation debris. The initial construction or deposit of the compact base of this feature is estimated between cal AD 835–1030. An AMS 14C sample (UOC-16246) from the looser upper layer dates to cal AD 900–1035, firmly situating it within the Early Postclassic period. This upper layer yielded a dense deposit of artifacts, including nine obsidian prismatic blades and diagnostic pottery, such as a moulded-carved vase sherd, two hollow oven feet, and an Early Postclassic sherd with incised and excised “international style” geometric patterns. Habitation debris subsequently accumulated around the feature, suggesting that if this was a small daub construction, it was no longer used by cal AD 965–1130. The final AMS 14C date in the ALA-047B sequence, from charcoal associated with habitation debris near the western profile wall next to Feature 1, dates to cal AD 1020–1150 (UOC-16245). The final boundary estimates that activity at the platform concluded sometime between cal AD 1000–1290.
5. Discussion
Given the high demands on the quality of archaeological information to justify the placement of radiocarbon data within a Bayesian framework, developing stratigraphic sequence models at Alabama begins with a highly systematic approach to excavation and sampling. Alabama’s alluvial clay-core platforms are susceptible to post-depositional processes such as weathering, erosion, and heaving, which can cause discrete contexts to meld together (Peuramaki-Brown et al. Reference Peuramaki-Brown, Morton, Longstaffe and Jordan2023; see also Brouwer Burg et al. Reference Brouwer Burg, Runggaldier and Harrison-Buck2016). As such, a careful and methodical approach to excavation is needed to identify subtle differences between construction cores and other contexts, such as variations in the matrix and the character and density of artifact assemblages. Achieving a high degree of chronological resolution relies on obtaining radiocarbon dates from these various stratigraphically ordered contexts, beginning with excavation levels representative of the onset of occupation or events before platform construction. These contexts’ data are critical for constraining modeled AMS 14C probability distributions and boundary estimates for undated events earlier and later in the sequence. Moreover, these data offer crucial prior information to address challenges posed by secondary charcoal within overlaying construction cores and aid in constraining long or dual-peak probability distributions associated with the Late and Terminal Classic radiocarbon plateau/reversal. Although unsealed habitation debris contexts are usually the most disturbed, they nonetheless offer valuable insights into the timing of occupation when corroborated with other lines of evidence, such as ceramic data (Wylie Reference Wylie, Leonelli and Tempini2020; c.f. Webster et al. Reference Webster, Freter, Storey, Demarest, Rice and Rice2004).
Stratigraphic chronological modeling of residential groups in Alabama’s outlying settlement zone has given us a more nuanced picture of settlement development and occupation at the townsite. Overall, the results enhance the chronological resolution of boomtown development during the Late and Terminal Classic periods and confirm suspicions that parts of the settlement were occupied into the Early Postclassic. While our analysis primarily focuses on stratigraphic architectural contexts as they provide constraints for Bayesian modeling, the radiocarbon data presented in this article also illuminate the timing of earlier and later occupations in the Early Classic and Late Postclassic periods, respectively. These findings prompt further consideration of the complex social, political, and economic relationships between Alabama’s inhabitants and neighbouring regions during these dynamic cultural periods.
5.1. Situating the developmental history of Alabama’s settlement zone
Radiocarbon data suggest an Early Classic occupation was present at ALA-002 sometime between cal AD 345–545. Cultural deposits from this occupation lie buried beneath a thick layer of sterile sediment, suggesting a period of abandonment at the site before the onset of occupation in the Late Classic. Few artifacts were recovered from these deposits, limiting deeper analyses of the social, political, and economic relationships between Alabama and neighbouring settlements. Nevertheless, we improve our understanding of the regional chronology by merely identifying this occupation (Peuramaki-Brown et al. Reference Peuramaki-Brown, Morton and Jordan2020). Expanding our view northward, Mayflower and Maintzunun along Silk Grass Creek and the Colson Point sites in the North Stann Creek Valley provide evidence of a deeper history for the region, dating back to the Preclassic period (Graham Reference Graham1994, 133–134). Sites in these areas flourished during the Protoclassic and Early Classic (ca. 100–600 CE), evidenced by finds such as the famous jades at Pomona and Kendal (Kidder and Ekholm Reference Kidder and Ekholm1951; MacKie Reference MacKie1985) and more systematic excavations conducted by Elizabeth Graham (Reference Graham1994). Small-scale commercial shell-lime production thrived along the Placencia Peninsula adjacent to the Caribbean coast during this period (MacKinnon and May Reference MacKinnon and May1990), and populations were present at the Colson Point sites (Graham Reference Graham1994). Areas to the south in East-Central Belize lack substantial evidence of Early Classic activity. However, settled communities were established by this time further south in the Southern Belize Region (SBR) at Nim li Punit (Braswell Reference Braswell and Braswell2022b, 104–105; Irish and Braswell Reference Irish and Braswell2015; Prager et al. Reference Prager, Volta, Braswell and Braswell2014) and Uxbenka and its surrounding area (Culleton et al. Reference Culleton, Prufer and Kennett2012; Jordan and Prufer Reference Jordan and Prufer2017; Novotny Reference Novotny2015; Thompson and Prufer Reference Thompson and Prufer2019).
Activity in Alabama’s settlement zone began at ALA-002B during the early facet of the Late Classic, prior to cal AD 655. However, most platform construction, including at other structures in ALA-002 and ALA-047, appears to have begun in the latter half of the Late Classic, with renovations at these settlement units continuing into the Terminal Classic period. Initial activities at ALA-045 appear to have begun in the Terminal Classic, with the ALA-045A platform reaching its final configuration between cal AD 850–965. The extensive remodeling of ALA-002’s plaza during the Terminal Classic period indicates that its residents possessed the necessary resources and social capital to carry out such a significant construction project. The scale and nature of the group’s buildings relative to surrounding settlement units, coupled with the formal raised boulder-and-cobble-core plaza, may suggest that in addition to being a residential site, this group took on a more public function during the Terminal Classic, potentially as an integrative community “focal node” (Hutson Reference Hutson2016, 80; Ingalls and Yaeger Reference Ingalls and Yaeger2022; Peuramaki-Brown Reference Peuramaki-Brown2013).
Evidence from the monumental core, including recent and legacy radiocarbon and obsidian hydration dates, as well as excavation and ceramic data, suggests it was established and constructed rapidly around the same time that the settlement was developing (Peuramaki-Brown and Morton Reference Peuramaki-Brown and Morton2019b, 5–9). Late and Terminal Classic Alabama was integrated into a continuum of settlement types in East-Central Belize, comparable to other high-level civic-ceremonial towns with multiple formalized plazas and specialized architecture, such as Pearce-Huntal Mo’ (Dunham et al. Reference Dunham, Murray, Brooks, Reynolds, Cookro and Jacobs1995; Peuramaki-Brown and Morton Reference Peuramaki-Brown and Morton2019a). Slightly smaller, middle-level sites such as Pomona and Kendal were also present in the sociospatial landscape, along with numerous low-level sites comprising isolated monumental platforms and surrounding settlement units. Coastal sites remained necessary for resource procurement at this time (Graham Reference Graham1994; MacKinnon Reference MacKinnon, McKillop and PF1989a; Sills Reference Sills2016; see also McKillop and Sills 2021, Reference McKillop and Sills2022), and Placencia Caye may have been a node in a coastal canoe trade route, linking traffic to inland centres via the lagoon and rivers (MacKinnon Reference MacKinnon1986, Reference MacKinnon1989b, Reference MacKinnon1990). In the SBR, small autonomous kingdoms centred at Pusilha, Uxbenka, and Nim li Punit were important high-level centers throughout the Late Classic. However, they experienced pronounced declines during the Terminal Classic. Lubaantun, founded slightly later, likely by colonists from Uxbenka, was the last of these centers to be abandoned around 900 CE (Braswell Reference Braswell2017; Irish and Braswell Reference Irish and Braswell2015; Jordan and Prufer Reference Jordan and Prufer2017; Prager et al. Reference Prager, Volta, Braswell and Braswell2014; Prufer et al. Reference Prufer, Thompson, Meredith, Culleton, Jordan, Ebert, Winterhalder and Kennett2017; Thompson and Prufer Reference Thompson and Prufer2019, Reference Thompson, Prufer, Marken and Arnauld2023; see also Braswell Reference Braswell and Braswell2022b, Reference Braswell and Braswell2022c). The decline of these centers may have been exasperated by climatic volatility, which led to a severe drought beginning around 835 CE (Prufer et al. Reference Prufer, Thomspon, Wickert and Kennett2023).
While there is limited evidence for the continued occupation of Alabama’s monumental core beyond the Terminal Classic period, it is evident that the townsite’s outlying settlement zone remained inhabited into the Early Postclassic. This finding contributes to a broader pattern observed across the Maya lowlands, where commoner residences in the settlement zones of towns and cities continued to be inhabited for several generations after the cessation of urban core-focused construction following the collapse of Classic-period political institutions (e.g., Ashmore et al. Reference Ashmore, Yaeger, Robin, Demarest, Rice and Rice2004, 321–322; Iannone et al. Reference Iannone, Chase, Chase, Awe, Moyes, Brook, Polk, Webster, Conolly and Iannone2014; Longstaffe and Iannone Reference Longstaffe and Iannone2011, Reference Longstaffe and Iannone2022; Mixter Reference Mixter2017, Reference Mixter2020, 226; Palka Reference Palka1997, Reference Palka, Inomata and Webb2003, 132–136; Peuramaki-Brown Reference Peuramaki-Brown2012, 228–240; Thompson and Prufer Reference Thompson and Prufer2021). Elsewhere in East-Central Belize, Early Postclassic activity has been documented at both inland and coastal sites, implying widespread continuity of settlement in this region (Graham Reference Graham, Chase and Rice1985, Reference Graham1994; MacKinnon Reference MacKinnon, McKillop and PF1989a; MacKinnon and Kepecs Reference MacKinnon and Kepecs1989).
Following multiple generations of abandonment, people returned to Alabama during the Late Postclassic, at some point between cal AD 1325–1385 and 1430–1475, as evidenced by well-defined contexts at ALA-002. Additional Late Postclassic contexts can probably be found elsewhere in the settlement zone, although they have likely been disturbed by orchard activities. Late Postclassic activity has been documented elsewhere in East-Central Belize, including stelae re-erection and associated pottery deposits at Mayflower and Kendal (Graham Reference Graham1994, 134). While further analyses are needed to gain a deeper understanding of Alabama’s Late Postclassic occupation, it is evident that the site played a role in the cultural dynamics of the region in the centuries leading up to Spanish contact. During this period, substantial populations inhabited inland areas and the coastal margins of northern Belize (Badillo Reference Badillo2021; Masson Reference Masson1999; Masson and Rosenswig Reference Masson and Rosenswig2005; Rosenswig and Masson Reference Rosenswig and Masson2002; Simmons et al. Reference Simmons, Pendergast and Graham2009). It is plausible that some of these populations moved into East-Central Belize through the Hummingbird Gap, recognized as an essential trade corridor during the early Spanish contact period. Maritime traders, well documented in the SBR, also influenced the Late Postclassic dynamics of the region (McKillop Reference McKillop2005, Reference McKillop2024). However, by this time, coastal trade routes had shifted outwards from the coast, utilizing cayes of the outer reef edge (MacKinnon Reference MacKinnon1989b). Whether Late Postclassic activity in Alabama’s settlement zone represents a permanent occupation or the more periodic accumulation of material remains, perhaps suggestive of a waystation used by traders moving back and forth between northern and southern regions, requires further consideration beyond the scope of this article.
Conclusions
This article presented AMS 14C data for the settlement zone of the ancestral Maya townsite of Alabama, East-Central Belize. These data, modeled within Bayesian stratigraphic sequences, provide new insights into Alabama’s development, growth, and occupation and establish a framework for future chronological modeling of earthen core architecture in this region. Three distinct periods of occupation have been identified, beginning with Early Classic activity between cal AD 345–545. Following a period of abandonment, the townsite’s population boomed throughout the Late and Terminal Classic periods, evidenced by the construction of multiple platforms within residential groups across the settlement. Although activity in Alabama’s monumental core appears to have ceased in the Terminal Classic, occupation in the outlying settlement persisted well into the Early Postclassic period, indicating the resilience of this community amidst broader social, political, economic, and ideological transformations in the Maya lowlands. Late Postclassic reoccupation is evident in parts of the settlement, beginning as early as cal AD 1325 and extending potentially beyond cal AD 1475. Together, these findings offer the first detailed deep history perspective for East-Central Belize, with implications for understanding the evolving cultural dynamics of this poorly understood frontier region of the Maya lowlands.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2025.10116.
Acknowledgments
This work was funded by grants from the Social Sciences and Humanities Research Council of Canada (SSHRC; file numbers 430-2015-00193, 435-2018-0091), an Athabasca University Research Incentive Grant and Office Academic Research Fund Grant, and generous supporters through Experiment.com. In addition, this research was conducted with support from a SSHRC Doctoral Fellowship awarded to Matthew Longstaffe and various funding from the University of Calgary Department of Anthropology and Archaeology and Faculty of Graduate Studies. We thank the Belize Institute of Archaeology and Drs. John Morris and Melissa Badillo for permits to conduct archaeological research at Alabama and export carbon samples, the property owners for allowing us access to private lands where Alabama is located, and permissions from representatives of Maya Mopan Village. This research was only possible with the efforts of the SCRAP team, including many local collaborators from Maya Mopan Village. Additionally, we thank the Athabasca University Faculty of Humanities and Social Sciences and Research Office, the Department of Anthropology & Archaeology at the University of Calgary, and the University of Calgary Study Abroad program. Finally, we thank Takeshi Inomata and two anonymous reviewers for their constructive feedback and comments on an earlier draft of this manuscript. Any mistakes are our own.
Data availability statement
The data for this paper are included as Supplemental Online Material and can be made available to the authors upon request.
Competing interests
The authors declare none.
Open access
