Dimensions of Maya (R)urbanism

Despite the growing acceptance of ancient Maya urbanism, research has generally focused on localized analyses of cities, often imbuing greater significance to urban contexts than their rural counterparts. In part this might be explained by the urban prestige effect (Smith Reference Smith2023:7–8) but could also result from Maya archaeology’s reluctance to engage in comparative, synthetic analysis (Ek Reference Ek2020). Such trends, however, do not negate the extensive body of scholarship that focuses on rural life and hinterland communities in ancient Maya society (e.g., Hutson et al. Reference Hutson, Dunning, Cook, Ruhl, Barth and Conley2021; Iannone and Connell Reference Iannone and Connell2003; Lamb Reference Lamb2022; Lemonnier and Arnauld Reference Lemonnier and Arnauld2022; Lohse and Valdez Reference Lohse and Valdez2004; Robin Reference Robin2012; Scarborough et al. Reference Scarborough, Valdez and Dunning2003; Yaeger and Robin Reference Yaeger, Robin, Lohse and Valdez2004). Much of this research stems from foundational fieldwork involving intersite surveys and household archaeology conducted by an earlier generation of scholars (e.g., Fedick Reference Fedick1988; Ford Reference Ford1981; Puleston Reference Puleston1974; Wilk and Ashmore Reference Wilk and Ashmore1988). Increasing numbers of recent large-scale lidar studies further reveal variable patterns of settlement density over much broader areas (Canuto et al. Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018; Chase et al. Reference Chase, Chase, Awe, Weishampel, Iannone, Moyes and Yaeger2014; Estrada-Belli et al. Reference Estrada-Belli, Canuto, Šprajc and Fernandez-Diaz2025; Golden et al. Reference Golden, Scherer, Schroder, Murtha, Morell-Hart, Diaz and del Pilar Jiménez Álvarez2021; Inomata et al. Reference Inomata, Triadan, Pinzón, Burham, Luis Ranchos, Aoyama and Haraguchi2018; Ringle et al. Reference Ringle, Gallareta Negrón, Ciau, Seligson, Fernandez-Diaz and Ortegón Zapata2021; Schroder et al. Reference Schroder, Murtha, Golden, Hernández, Scherer, Morell-Hart, Almeyda Zambrano, Broadbent and Brown2020). While there is clearly no lack of empirical record, there is little agreement over what models of urban-rural settlement organization are most appropriate for the lowland Maya.

Traditional interpretations tend to emphasize one configuration over the other, viewing urban and rural as distinct and separate entities. For example, some argue that ancient Maya cities were discrete centers of attraction for more distant rural populations (e.g., Hutson Reference Hutson2016) whereas others interpret rural settlements as independent—largely disconnected from the centralized administration of urban centers (Arnauld et al. Reference Arnauld, Michelet and Nondédéo2013; Lemonnier and Vannière Reference Lemonnier and Vannière2013; Nondédéo et al. Reference Nondédéo, Arnauld and Michelet2013). In contrast, Marken’s (Reference Marken, Marken and Fitzsimmons2015) delineation of settlement zones at El Perú-Waka’ calls for greater attention to urban-rural interactions within Classic Maya polities. Similarly, Garrison and colleagues introduce the concept of “conurbation” to emphasize a continuous social landscape wherein urban and rural populations were fully integrated and “part of a comprehensive, interpretable system built on landesque investment and a high degree of political coordination” (Reference Garrison, Houston and Firpi2019:135). Still others contend that alternative categories such as “town” should be employed to differentiate settlements that exhibit urban functions in otherwise rural landscapes (Eppich and Ball Reference Eppich and Ball2024; Nondédéo et al. Reference Nondédéo, Lemonnier, Hiquet, Purdue, Castanet, Dussol, Testé, Marken and Arnauld2023). Although recent approaches eschew binary frameworks (Lamb Reference Lamb2022; Marken and Arnauld Reference Marken, Arnauld, Marken and Arnauld2023), the pervasive usage of urban and rural perpetuates categorical thinking that may not accurately reflect indigenous distinctions (Hirth Reference Hirth and Sanders2003; Marcus Reference Marcus, Vogt and Leventhal1983). An alternative term that may prove beneficial is the blended concept “rurban,” referring to a mix of large rural and small urban areas with multiple land use functions linked or nested within broader geopolitical systems (Cyriac et al. Reference Cyriac, Mohammed Firoz, Rajendran and Brears2022). While we agree that Maya archaeology must reckon with its conceptual terminology and models of urban-rural settlement (see also Garrison Reference Garrison, Marken and Arnauld2023), such efforts require a unified analytical framework to systematically compare and synthesize these different datasets before new terms and models are adopted.

Smith (Reference Smith2020) defines several key dimensions of urbanism that establish empirical criteria for the analysis of ancient cities. Expanding on these dimensions to accommodate the rural-urban continuum of lowland Maya settlement, Marken and Arnauld (Reference Marken, Arnauld, Marken and Arnauld2023:13–21) offer a revised framework that organizes these analytical categories (urban form, life, meaning, and function) according to their archaeological accessibility, emphasizing the recursive relationship between urban dimensions (Table 1). We find this approach promising for promoting synthesis research in Maya archaeology and examining this unique “rurban” tradition. Thus, our study focuses on two basic aspects: (1) settlement size, including spatial extent, demographic estimates, and approximations of population density; and (2) urban form, comprised of architectural features, site layout, and settlement planning in relation to the natural landscape. We first introduce the study area and summarize findings from previous investigations at Altar, highlighting the urban features found in the site core. We then describe our UAS survey and the broader settlement patterns of the UUCZ according to the analytical dimensions outlined in the preceding text. Contextualizing our results with settlement data from other remote sensing surveys allows us to examine organizational aspects of past urban life at Altar and reevaluate the function of this “principal center” in relation to other Classic Maya cities. We conclude with a discussion of broader implications for studying ancient Maya urbanism, focusing on the connection between cities and more rural settlements.

Table 1.

Analytical Framework of Urban Dimensions for Maya Archaeology based on Marken and Arnauld (Reference Marken, Arnauld, Marken and Arnauld2023) and Smith (Reference Smith2020).

Table 1 Long description

The figure is a shaded relief map of a region showing terrain elevation in green lowlands and brown higher ground, crossed by blue rivers and waterways. Numerous archaeological sites are marked with small triangles and labelled with place names distributed across the map. An inset locator map in the upper left shows the wider geographic context, with a highlighted area indicating the location of the main map. A compass rose and a scale bar in the lower left provide orientation and distance reference.

Historical and Environmental Context

The UUCZ encompasses the region where the Pasión and Salinas-Chixoy Rivers join to form the Río Usumacinta in the western Maya lowlands along the modern-day border of Guatemala and Mexico (Figure 1). The landscape is dominated by meandering river system landforms including seasonally flooded lagoons, oxbow lakes, and infilled cut-off channel remnants in low-lying floodplains flanked by slightly elevated terraces. Unlike the narrow, bedrock channels of the Pasión and lower reaches of the Usumacinta (Aliphat Reference Aliphat1994), the UUCZ is defined by alluvial reaches with wide floodplains and frequent channel migration, resulting in a dynamic fluvial landscape that changed dramatically during the Holocene. While these riparian environments offered vital resources to sustain early human populations (Sharpe et al. Reference Sharpe, Palomo, Inomata, Triadan, Pinzón, Curtis and Emery2024) and rich alluvial soils suitable for nonintensive agriculture (Horseman et al. Reference Horseman, Morell-Hart, Golden and Scherer2024), they also posed significant risks due to flooding (Schroder et al. Reference Schroder, Murtha, Broadbent and Zambrano2021) as recently observed in our study area after back-to-back hurricanes in 2020.

Figure 1.

Overview map of the Maya region with sites mentioned in the text. Inset map shows the location of Altar de Sacrificios and the UUCZ study area shaded in purple. (Color online)

Altar—the largest recorded site in the UUCZ—has been the focus of multiple investigations since the late nineteenth century. Among the early explorers were Maudslay, Maler, and Morley who each visited on separate occasions to record the carved monuments—including the one for which the site is (mis)named—and to produce sketch maps of some of the largest structures in the epicenter (Willey and Smith Reference Willey and Smith1969:4–7). A 1937 expedition sponsored by the Carnegie Institution of Washington made more systematic efforts to map the site, later published alongside a summary of the hieroglyphic inscriptions (Morley Reference Morley1938:309–340). These early documentary efforts provided an important basis for the first large-scale excavation project in southwest Petén conducted by the Peabody Museum of Harvard University between 1958 and 1963.

Not surprisingly, questions about settlement patterns and site location were central to this new project as project director Gordon Willey had just completed the Belize Valley survey (Willey Reference Willey1973:6). Based on Altar’s location at the junction of two major rivers that conjoin to form the Usumacinta, it was hypothesized that the site could be a potential crossroads for highland-lowland exchange (Willey et al. Reference Willey, Smith, Bullard and Graham1960). However, the Harvard team never capitalized on this line of inquiry due to insufficient comparative data (Willey Reference Willey1973:11). Moreover, questions about trade and exchange were not major research topics for American archaeology at the time. Instead, the Harvard project pursued a more traditional culture-historical research program, documenting the site chronology through an intensive excavation program, architectural investigations, and ceramic analysis (Adams Reference Adams1971; Smith Reference Smith1972). William Bullard prepared a more detailed map of the site core and adjacent settlement during the first season (Figure 2), which helped guide the excavation strategy (Smith Reference Smith1972). Altar’s epicenter contains a number of monumental buildings, including temple pyramids and ceremonial platforms, plazas of various sizes, carved stone monuments, a ballcourt, and two palace complexes—architectural features that are now considered typical of urban areas in other Maya cities. These buildings are separated into two main architectural groups (A and B) that have distinct forms, layouts, and construction histories as described in Smith (Reference Smith1972) and summarized in the following text.

Figure 2.

Digitized map of Altar de Sacrificios site core including urban architectural features in Groups A and B (source: Bullard map published in Willey and Smith Reference Willey and Smith1969).

Figure 2 Long description

The figure is a black and white archaeological site plan showing many rectangular building outlines scattered across the area, representing structures and platforms. A blue river runs along the upper right edge of the map. Two main clusters of structures are circled and labelled, including Group A and Group B, with additional labelled features such as plazas and named structures. A legend identifies the mapped symbols, and a north arrow and scale bar in the upper area give orientation and distance. A credit note indicates the map is based on an earlier survey and digitised by the author.

Group B forms a small and compact architectural complex that was the center of localized ritual and ceremonial activity from the Middle Preclassic to the Early Classic period. Buildings are arranged around a small rectangular plaza measuring about 50 m² and faced with a locally available red “sandstone” blockFootnote ¹ in its final architectural phase dated to the Protoclassic Salinas phase (AD 100–450). Str. BI, the tallest building at the site, is a temple pyramid located on the south side of the plaza that was gradually built up over seven construction episodes beginning in the Middle Preclassic San Felix phase (600–300 BC). Early platforms were raised from packed earth later covered with almeja clamshells procured from the local rivers and set into clay mortar. In its final form, Str. BI rose 13 m above the plaza level in multiple terraces where several inscribed stelae and altars carved from the same red “sandstone” material were placed along with dozens of ceramic vessel cache offerings. Structures BII and BIII, on the west and north sides of the plaza, respectively, are smaller temple pyramids with a similar construction history. Enclosing the eastern edge of the plaza is Str. BIV, a broad and low palace complex with a terraced upper platform and court arrangements that yielded a similar stratigraphy as the other Group B buildings. Notably, the area to the east of Group B seems to have been retained as an open space throughout the Preclassic and may have served as an informal plaza or gathering place along the river, at least until construction began in Group A.

Less than 200 m eastward, Group A is more than twice the size of Group B measuring about 350 × 150 m and is divided into North and South Plazas of about equal size. In contrast to the early establishment of Group B, the buildings in Group A mostly date to the Late Classic period (AD 600–900) with the earliest monument (Stela 8) dedicated on 9.9.15.0.0 (AD 628). Departing from earlier architectural traditions, most of the buildings in Group A were constructed with a rubble core and faced with limestone blocks and a plaster finish in the standard central Petén style. The North Plaza is delineated by Str. AI—a large palace complex with terraced platforms that had largely been destroyed by riverbank erosion prior to Harvard’s investigations—and Strs. AII and AIII. The latter was a 10 m high funerary temple on the east side of the plaza that contained multiple graves including the remains of Burial 128, a royal woman who was accompanied with the most elaborate grave goods recovered by the Harvard project including the famous Altar vase (Haviland Reference Haviland1971; Smith Reference Smith1972:266–268). Three large circular altars were found atop Str. AII, a range structure measuring 110 m long, 40 m wide, and 10 m tall that forms the west side of the North Plaza. The South Plaza is slightly larger but less formally defined with various platforms of different sizes and shapes that may have been the foundations for elite residences surrounding the plaza area. Although no standing walls, vaulted structures, or roomed buildings were documented in Group A, many of these structures likely supported perishable dwellings constructed from wattle and daub similar to those found in the outlying areas. A ballcourt separating the North and South Plazas was a focal point for ceremonial activity in Group A. Although access to the Group A plazas was restricted, the layout of monuments and buildings suggests that rituals, ballgames, and other royal performances likely provided formal contexts for periodic communal gatherings in Altar’s urban core during the Late Classic.

While Bullard’s original map remains remarkably accurate, Harvard’s limited explorations beyond Groups A and B documented a sparse distribution of small earthen mounds presumed to be the remains of residential dwellings called “house mounds” in the original report (Smith Reference Smith1972:128–129). Although test pits were placed in all 41 house mounds, the goal was to recover ceramics for the purpose of developing a site chronology rather than an expansive household study. Despite the relatively small number and size of structures documented outside the site core, Altar was originally interpreted as a “principal center” based primarily on the ceremonial architecture, hieroglyphic monuments, and elaborate cache deposits and burial goods recovered from Groups A and B—features similar to those found at other nearby major ruins (Willey Reference Willey1973:6–7).

Subsequent publications make only passing reference to the original Altar reports. Drennan’s (Reference Drennan, Wilk and Ashmore1988) study of residential density across Mesoamerica is one of the only comparative analyses to use Altar’s original survey data, which helped confirm the dispersed settlement patterns of the lowland Maya—a point we return to in the following text. Despite limited decipherment of its poorly preserved inscriptions, Altar has featured in a handful of other accounts related to Maya political organization. Based on its hieroglyphic details and archaeological context, Mathews and Willey (Reference Mathews, Willey and Culbert1991) describe Altar as an independent polity that emerged early and maintained its position by existing on the margins and between more powerful polities like Dos Pilas and Yaxchilan. In another case, Altar is identified as a secondary center with local rulers who were never under the sway of a regional power but consolidated their communal authority through water and agricultural rituals (Lucero Reference Lucero and Lisa2006). Still, in other studies, Altar has been left out of political interaction models altogether (Martin and Grube Reference Martin and Grube2008). To reevaluate Altar’s position on the geopolitical landscape, we must understand how its urban core articulates with the surrounding settlement and landscape.

UAS Survey in the Upper Usumacinta Confluence Zone

This statement characterizes the archaeological situation around Altar during the mid-twentieth century, but the same cannot be said for the region today. The dense canopy and thick patches of bamboo that once enclosed this landscape have largely been replaced by plowed agricultural fields and pastures used for grazing cattle (Figure 3), though remnant river channels and low-lying areas still flood during the rainy season. Such vast deforested landscapes document the alarming rate of change within this tropical ecosystem, but at the same time offer fortuitous opportunities for archaeologists to systematically inventory large stretches of terrain using relatively inexpensive and efficient remote sensing methods (Munson et al. Reference Munson, Mejía Ramón, Aragón, Arroyo, Salinas and Álvarez2019).

Figure 3.

Photograph of the modern UUCZ landscape showing the junction of the Salinas-Chixoy (top left), Pasión (bottom left), and Usumacinta (right) rivers, looking west from Guatemala into Mexico. Photograph courtesy of Jessica Munson. (Color online)

The Proyecto Arqueológico Altar de Sacrificios (PAALS) began in 2016 with the goal of understanding the history of human occupation and landscape change in the UUCZ through extensive settlement survey, geoarchaeological research, and household investigations. From 2017 to 2019, PAALS conducted a series of UAS surveys that generated high-resolution topographic models to identify cultural features and natural landforms (Mejía Ramón Reference Ramón and Andrés2019; Munson et al. Reference Munson, Mejía Ramón, Aragón, Arroyo, Salinas and Álvarez2019). Additionally, geoarchaeological investigations of floodplain deposits exposed in riverbanks and sediment cores along with geomorphic mapping of floodplain landforms have aided in the reconstruction of river channel migration across the study area (Onken Reference Onken2023; Figure 4). This work, combined with ongoing analyses from household excavations, provides an important foundation for understanding the social contexts and natural landscapes in which people pursued diverse livelihoods throughout the Preclassic and Classic periods (Munson and Scholnick Reference Munson and Scholnick2022).

Figure 4.

Relative age of geomorphological surfaces in the UUCZ. Because the estimated age of these landforms postdate site abandonment, these areas were excluded from the structure density estimates and definition of district boundaries. (Color online)

Figure 4 Long description

The map illustrates the relative age of post-250 CE landforms in a defined area, with active channels highlighted in blue. The legend categorizes landforms into ′Early′, ′Late′ and ′Active channel′. ′Early′ areas are marked in pink, ′Late′ areas in yellow and active channels in blue. The map includes labeled locations such as ′Peten, Guatemala′ and ′Machaquila, Mexico′. Study sites are outlined, with ′UUCZ survey sites′ shown in gray and ′Harvard study sites′ in blue. A scale bar at the bottom indicates distances up to 10 kilometers. The map is oriented with a north arrow. The source is credited to DigitalGlobe.

Over three field seasons, PAALS surveyed a total of 54 km² on both sides of the Guatemala-Mexico border using a DJI Phantom 3 Professional quadcopter with a 12-megapixel camera and gimbal stabilization. We collected 26,000+ images that were processed using Structure-from-Motion (SfM)-based photogrammetry to generate high-resolution (≤15cm/pixel) digital elevation models (DEMs) of the UUCZ landscape. To generate the DEMs, photographs were taken with the UAS controlled using an iPad Mini 2 running the iOS application Map Pilot versions 2.5–2.6. The UAS was flown at 15 m/s at an elevation of 85 m the first year (to maximize resolution) above the main peninsula, and at the maximum then-allowable elevation in years thereafter to maximize total areal coverage (150 m the second year in Guatemala and 120 m the third year in Mexico). Routes and photographic intervals were programmed such that there would be a 75% front and side overlap ensuring that every ground point would be visible in at least nine images from different angles.

Photogrammetric reconstructions were generated twice—once in the field at a high resolution (∼30 cm depending on the sector), and again at the conclusion of fieldwork with the highest resolution (∼15 cm)—using Agisoft Metashape versions 1.3.2–1.5.2. Given the scale of the area surveyed, the ground conditions in parts of the survey region, and the limited amount of time available for mapping, it was not feasible to produce ground control points. As such, the resulting DEMs suffer from significant barrel distortion given the UAS’s truncated fisheye-style camera. To correct these aberrations, we assumed that SRTM-based elevations are distributed about the true elevation if at a lower scale. Thus, we resampled the SfM-based DEM to 30 m and subtracted that from the SRTM-based DEM. From these differences we regressed a twelfth-order polynomial raster, producing a topographic distortion error model. This error model was resampled to the SfM model’s original resolution, and subtracted from the raw elevations producing the corrected raster used for identification and analysis.

Archaeological features were detected and digitized by the first two authors—relying on hill-shaded elevation rasters (and detrended versions thereof)—in separate analyses to ensure replicability and were later ground verified by the PAALS team. In total, 430 mounds were recorded including 352 new structures that were not previously documented (Figure 5). Some architectural clusters, however, were noted during earlier explorations and appear on several published maps of the broader region (Tourtellot Reference Tourtellot1988; Tovalín and Ortiz Reference Tovalín and Ortiz V2005; Willey and Smith Reference Willey and Smith1969). Morley (Reference Morley1938:310–314), for example, noted a small group of mounds and one inscribed monument along the south bank of the Usumacinta about 3 km downstream from Altar at a site he called El Pabellón. In addition, Bullard visited ruins near Laguna Ixcoche during the early years of the Harvard project, describing “eight to 12 mounds forming a small ceremonial group . . . substantially larger than El Pabellón” although no monuments were observed (Willey and Smith Reference Willey and Smith1969:35). Another group of mounds located at the opposite end of Laguna Ixcoche was also reported to the Harvard team but went unvisited. The descriptions of these mound locations are consistent with features observed in our survey, so we retain these names to reference the settlement clusters as described in the following text and indicated on Figure 5. In general, mounds are dispersed across the fluvial landscape, in many cases aligned with relict channels. However, the river courses have shifted considerably throughout the Holocene, dramatically changing the landscape and damaging or obliterating an unknown number of archaeological features, thus complicating analysis and interpretation of ancient settlement systems (Munson et al. Reference Munson, Scholnick, Mejía Ramón and Aragon2023). We attempt to account for some of these factors in our population estimates and interpretations of the settlement patterns.

Figure 5.

Settlement map showing the location of identified mounds and UAS-surveyed areas in the UUCZ study region with site names based on earlier investigations. (Color online)

Figure 5 Long description

The map displays the UUCZ study area, highlighting archaeological mounds and surveyed regions. Mounds from the HV project are marked with purple triangles, while mounds from the UUCZ survey are indicated with yellow circles. The map includes various channel types: active river channels, infilled cutoff channels, intermittent flow cutoff channels and oxbow lake cutoff channels, all marked in blue. The map covers parts of Chiapas, Mexico and Petén, Guatemala, with labeled sites such as ′Second Inscription′ and ′El Pabellón′. The scale bar at the bottom indicates distances in kilometers, ranging from 0 to 10 km. The compass rose in the bottom left corner shows the north direction.

Settlement Size: Area, Density, and Population Estimates

The UUCZ survey area is relatively small (54 km²) compared to other remote sensing surveys in the Maya lowlands (Table 2; μ = 214 km²), but many were acquired with airborne lidar that can cover much larger tracts than our drone-mounted system. Nonetheless, we can still make systematic comparisons between regions using common methods. Different approaches, however, have been employed by Maya archaeologists to calculate population estimates, a nontrivial exercise that has generated significant debate (for selected reviews, see Chase et al. Reference Chase, Chase, Chase, Chase, Chase and Chase2024; Culbert et al. Reference Culbert, Kosakowsky, Robert, William, Culbert and Don1990; Webster Reference Webster2018). Here we use individual structure count—a representation of a single-family unit—for the basis of our population estimates, following methods described in Canuto et al. (Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018) to facilitate regional comparison. This technique multiplies the total number of structures in the study area by a population index (PopIndex) that takes into account variation in the average number of household occupants, contemporaneity, and structure function, among other factors (see Canuto et al. Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018:Table 5). In this case, we report population as a range using the low (2.55) and high (3.97) PopIndex values recognizing the inherent imprecision in estimating the size of past populations. While most of the comparative cases in Table 2 derive from the Pacunam Lidar Initiative (PLI), structure density and population estimates for other regions are also reported as ranges according to the publications cited.

Table 2.

Population Estimates and Settlement Density Comparison of Selected Remotely Sensed Archaeological Surveys in the Maya Lowlands.

Table 2 Long description

The table compares remotely sensed survey regions by mapped area, counted structures, structure density, estimated low and high population totals, and corresponding population densities, with sources. Caracol has the highest reported population density, about 309 to 507 people per square kilometer, though its structure counts and structure density are not available. Among entries with structure counts, El Pilar is the densest settlement by structures at about 93 structures per square kilometer and has a population density near 246 people per square kilometer, but only a low population estimate is provided. Several large centers cluster at relatively high densities: Bolonchén and Muluchtzekel, Naachtun, and Tikal each report roughly 81 to 89 structures per square kilometer and population densities around 214 to 354 people per square kilometer depending on low versus high estimates. Midrange examples include Peru, Xultun, Zotz, Tintal, and Uaxactun, with structure densities around 31 to 45 per square kilometer and population densities roughly 79 to 179 per square kilometer. Larger-area regions such as Ceibal and Holmul show lower densities, about 60 to 110 and about 60 to 93 people per square kilometer respectively, while Usumacinta and Lacanja is lower still at about 40 to 63 people per square kilometer. The two UUCZ variants report similar densities near 27 to 43 people per square kilometer, reflecting different inclusion rules for landforms and estimated destroyed mounds. The lowest densities occur in Env 1 and Env 2, each around 4 to 7 people per square kilometer with about 1.7 structures per square kilometer. Comparisons should be interpreted cautiously because methods, coverage, and reporting differ across studies, and some cells are missing or provide only one bound of an estimate.

Population estimates for the UUCZ also consider local landscape changes associated with river channel migration that have compromised archaeological visibility. We calculated the density of structures excluding post-AD 950 landforms (see Figure 4), using this value (10.8 structures/km²) to estimate the number of mounds that may have been obliterated (n_? ≈ 153) in the study area. We then use this extrapolated structure count (n_tot = 583) to calculate the regional population estimates employing the method described in the preceding text. Accounting for taphonomic processes yields a more realistic population estimate for the UUCZ than if one assumes a static landscape. Thus, estimates for the Late Classic population in the UUCZ range from about 1,500 to 2,300—orders of magnitude less than other regions (see Table 2).

Although the UUCZ’s low population estimates may be a function of the smaller area surveyed in comparison to larger regional lidar surveys, population densities for the UUCZ are comparable to macro-settlement densities in the Corona-Achiotal region of northwestern Petén. However, comparisons between neighboring regions reveal that population density in the UUCZ is substantially lower than Ceibal (∼40%; Inomata et al. Reference Inomata, Triadan, Pinzón, Burham, Luis Ranchos, Aoyama and Haraguchi2018) and the Usumacinta-Lacanja (∼60%; Golden et al. Reference Golden, Scherer, Schroder, Murtha, Morell-Hart, Diaz and del Pilar Jiménez Álvarez2021), suggesting that this was not a major hub for permanent population despite its hypothesized importance as a crossroads for exchange. It is possible though that this region experienced large numbers of passers-by who may have made short stops along the way. Such transient encounters, however, would have left ephemeral settlement traces—a point we discuss further below.

At the macro-settlement scale, cluster analysis of the PLI survey blocks indicates that these can be grouped into three urbanization classes with Corona-Achiotal and the three “control” regions (Yala, Env 1, Env 2) classified as rural based on their low overall density and limited political integration (Canuto et al. Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018:7–8). Accordingly, the sparsely settled UUCZ would also fall into this rural category. However, populations were not evenly distributed across any of these landscapes, so we must examine patterns of variation at different scales and in relation to other topographic and environmental factors to better understand density continua and their implications for urbanization processes (Canuto and Auld-Thomas Reference Canuto and Auld-Thomas2021).

We generated a settlement density map for the UUCZ (Figure 6) following the same kernel density protocols used by Canuto et alia (Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018) ensuring consistency in classificatory schemes. Settlement distribution in the UUCZ is indeed sparse with most settled areas classified as “rural” based on the criterion of 10 to 60 structures/km². Notably, areas designated as rural are more sparsely occupied zones on the alluvial plain with settlement located close to previously active river channels. This rural pattern, however, is discontiguous with large “vacant” zones of settlement interspersed across the landscape. Such local variability is similar to the patchy settlement patterns observed in the Corona-Achiotal region (Canuto and Auld-Thomas Reference Canuto and Auld-Thomas2021), although we recognize that some of these areas may have housed a small number of ephemeral structures that left minimal traces on the landscape (Johnston Reference Johnston2004). Crop stains identified in multispectral imagery obtained by the PAALS survey may be consistent with such non-mounded, perishable structures (Munson et al. Reference Munson, Mejía Ramón, Aragón, Arroyo, Salinas and Álvarez2019). Although we were unable to further test these features, we tentatively interpret them as temporary-use structures or other possible campsites that may have been used intermittently by transient populations.

Figure 6.

Kernel density settlement map with values expressed as structures/km². (Color online)

The eastern part of the survey area surrounding Groups A and B has a slightly higher occupation density than other parts of the settlement. Another such small area is found a few kilometers to the west on the north side of the Usumacinta River near the current location of the Pasión-Salinas confluence. These areas of elevated density would be defined as “periurban” according to the classes defined by Canuto and colleagues (Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018; 61–150 structures per km²); however, this term usually refers to the urban fringe surrounding more densely settled cities (Simon Reference Simon2008). All the PLI survey blocks with known archaeological sites have sizable “urban” areas and many have even higher density “urban cores” with true periurban zones surrounding them (Canuto et al. Reference Canuto, Estrada-Belli, Garrison, Houston, Acuña, Kováč and Marken2018:Figure 7; see also Simon Reference Simon2008:171–172). Because there is no high-density epicenter in the UUCZ, we employ the term “rurban” to describe this macro-settlement pattern that combines traditionally urban and rural characteristics paralleling other frontier settlements and anomalous cases that do not easily fit into dichotomous classifications (Cyriac et al. Reference Cyriac, Mohammed Firoz, Rajendran and Brears2022; Raja and Sindbæk Reference Raja and Sindbæk2022).

Architectural Form, Function, and Spatial Organization

Size, layout, and distribution of residential and public architecture offer important descriptive attributes to characterize urban form and settlement planning (Marken and Arnauld Reference Marken, Arnauld, Marken and Arnauld2023:16–18). For the UUCZ, we examine these aspects of the built environment and undefined open space to consider potential land use activities carried out in the rural zones—and how these relate to Altar’s urban function and life in the countryside.

We classified structures into two categories based on criteria established by Bullard (Reference Bullard1960) and the original Harvard project (Smith Reference Smith1972). Residential structures are generally low earthen mounds whereas nonresidential “special structures” are nearly twice as tall and more massive architectural complexes with different forms (Table 3). Bullard’s original survey identified the former as “house mounds” based on their simple form, relatively small size, and dissociation from Groups A and B. The latter are more similar to Bullard’s (Reference Bullard1960) description of minor ceremonial centers, but we recognize these buildings may have more diverse and overlapping functions. Most of the mounds that were ground-verified by our team yield little to no evidence of masonry construction, unlike the monumental buildings in Groups A and B and some of the “special structures” documented in the rural zones. Most mounds identified in our survey are classified as residential (n = 390; 90.7%) and likely served some domestic function (Figure 7). These low earthen platforms offered an elevated, level surface for the construction of perishable buildings generally above the flood zone. Alluvial clay and mud deposits along with bamboo growing nearby offered ample, easily accessible materials for the construction of wattle-and-daub dwellings. Although Harvard’s investigations recorded few details of the domestic architecture, the PAALS excavations have documented the remains of walls, postholes, clay floors, and pit features associated with refuse consistent with domestic activities (Paiz and Munson Reference Paiz and Munson2023; Viñals and Munson Reference Viñals and Munson2024). Unlike the plazuela configuration typical of other lowland Maya settlements, residential platforms in the UUCZ are single, solitary features that generally align with the contours of previous river channels. Depending on their size, most mounds probably could have accommodated two to four perishable structures. Therefore, we consider each mound equivalent to a household unit for the purpose of estimating population and calculating household wealth inequality (Munson et al. Reference Munson, Scholnick, Mejía Ramón and Aragon2023).

Table 3.

Mound Classification and Size Attributes.

Table 3 Long description

The table summarizes mound categories by count and average size attributes: height in meters, base area in square meters, and volume in cubic meters. Residential mounds are most common with 390 total, averaging 2.4 meters tall, about 929.8 square meters in base area, and about 1449.3 cubic meters in volume. “Special” nonresidential mounds are fewer at 27 but larger, averaging 4.9 meters tall, about 2801.4 square meters in base area, and about 13159.5 cubic meters in volume. Focal nodes are rare at 13 yet largest overall, averaging 7.9 meters tall, about 4704.6 square meters in base area, and about 17290.2 cubic meters in volume. Unclassified mounds total 13, but their average height, base area, and volume are not provided. Overall, as mound type shifts from residential to focal nodes, average height, base area, and volume increase while the number of mounds decreases.

Figure 7.

Regional settlement pattern in the UUCZ relative to existing topographic and hydrologic features showing the location of mounds classified as residential and “special structures.” (Color online).

Figure 7 Long description

A thematic map showing a meandering river system across a floodplain, with mapped structure locations and mapped channel features. North is indicated by a north arrow at the lower left. A scale bar labeled 0, 2, 4, 6, 8 kilometer appears along the bottom. Legend items and symbols shown on the map: Non-residential structure is shown with a triangle symbol. “Special” structure is shown with a square symbol. Residential structure is shown with a circle symbol. Unclassified is shown with a dot symbol. Floodplain extents are shown as an outlined area. Active river channel is shown as a line. Infilled river channel is shown as a different line style. Cut-off channel (intermittent flow) is shown as a separate line style. Cut-off channel (active flow) is shown as a separate line style. Structure distribution relative to mapped channels: Most structure symbols appear in clusters along the main meandering channel belt and around multiple curved cut-off channel shapes. Several clusters occur near the central part of the map where the river bends tightly and where multiple cut-off channel segments are drawn. Additional smaller clusters appear toward the left side near another large bend and toward the right side near the channel as it approaches the map edge. Structure symbols are sparse in areas away from the mapped channel belt. Elevation legend text: A legend labeled SRTM DEM shows two values: 155 m and 100 m.

A total of 27 mounds (6.3%) were classified as “non-residential special structures” in our survey. Of these, 14 are buildings in Groups A and B that were previously identified as temple pyramids, ballcourts, administrative buildings, or dedicated funerary mounds. The remaining “special structures” in the outlying rural zone are large platforms or pyramid-shaped structures more than 6 m tall—greater than the height of most residential mounds. Notably, all “special structures” in the peripheral zone are within 400 m of water sources, and most are situated along paleochannels, which provided access to riparian resources and transportation corridors yet avoided riverbanks prone to flooding (see also Ford et al. Reference Ford, Clarke and Raines2009). Many are also part of small architectural clusters. Although we have not yet excavated any “special structures” in the rural zones, we recovered numerous figurine whistles and several small plain whistles from surface collections on Strs. 38 and 67. The acoustic properties of these objects may have served important roles in ceremonies and/or as more mundane communication devices. The location of these prominent structures in an otherwise sparsely occupied landscape may have provided important interregional functions controlling navigation along these waterways, although we do not exclude the possibility that these structures also served additional organizing functions for the resident population. Future investigations will help clarify these points, but for now we interpret the “special structures” in the rural zone, along with the principal temple and palace in Groups A and B, as focal nodes.

Focal nodes have been identified at a number of Maya settlements and are interpreted as integrating features for rural or dispersed populations (Burham et al. Reference Burham, Inomata, Triadan, MacLellan, Houk, Arroyo and Powis2020; Hutson Reference Hutson2016; Ingalls and Yaeger Reference Ingalls and Yaeger2022; Lemonnier Reference Lemonnier, Arnauld, Manzanilla and Smith2012; Peuramaki-Brown Reference Peuramaki-Brown2013; Walden et al. Reference Walden, Ebert, Hoggarth, Montgomery and Awe2019). Although this term embraces a wide range of variation, we use it broadly to refer to built spaces that facilitated social interaction and organization across this rurban landscape (Ashmore Reference Ashmore and Ashmore1981; Hutson Reference Hutson2016). In this study, the “special structures” we identify as focal nodes are evenly distributed, proximal to water sources, and associated with architectural clusters that we identify as districts (Table 4).

Table 4.

Size and Architectural Form Attributes for Focal Nodes in the UUCZ.

Table 4 Long description

Rows list focal nodes by district and structure ID, with height in meters, basal area in square meters, volume in cubic meters, and the architectural form with function notes. Heights range from 3.3 meters at PB 226 (platform) to 13.0 meters at ALS BI (temple pyramid), with other tall pyramids at TP2 163 (9.8 meters) and TP 100 (9.5 meters). Basal area is highest for IX3 220 (7037.3 square meters) and IX 182 (6946.2 square meters), both platforms, followed by GF2 50 (5534.9) and GF 38 (5431.3). Volume values are provided for several districts and peak at IX3 220 (22809.5 cubic meters) and IX 182 (22117.1), with GF 38 also high at 17880.4; PB 226 is the smallest reported volume at 11485.8. Platforms dominate the IX, IX2, IX3, PB, and GF2 entries, while pyramids appear in GF, GF2 structure 67, TP, and TP2; ALS includes a mix of platform, range structure, and temple pyramid forms. Several ALS and TP entries have missing basal area or volume, so comparisons across all districts are limited to the fields that are reported.

Districts are defined as residential zones that have some kind of administrative function and/or particular social identity that is distinct from other sectors of a settlement (Smith Reference Smith2010). According to this definition, districts tend to be larger than neighborhoods and may include public architecture and spaces for social gathering absent within more private settings. Although we previously defined neighborhoods for the UUCZ as a convenient unit for comparative analysis (Thompson et al. Reference Thompson, Walden, Chase, Hutson, Marken, Cap and Fries2022, Reference Thompson, Munson, Ortman, Mejía Ramón, Feinman, Quequezana, Cruz, Green, Lawrence and Roscoe2025), we introduce districts in this article as a means to analyze intrasettlement patterns and social organization. District boundaries were delineated based on the settlement density map, distribution of focal nodes, geomorphological reconstructions, and consideration of the contours of relic river channels and oxbow lakes that may have formed natural boundaries in the past. While the actual boundaries are probably fuzzier and more fluid than those conveyed in Figure 8, the social scale of districts provides a more appropriate unit of analysis to examine aspects of urban function than neighborhoods given the region’s sparse occupation. Table 5 lists the size, structure counts, and population estimates for each district including the estimated number of obliterated mounds. As expected, districts with periurban zones (Altar de Sacrificios and El Trapiche) have the highest population estimates within the UUCZ. However, with only ∼200 people per district they are very small communities. The most sparsely occupied districts (Ixcoche III and Grupo Frontera) had only a few dozen inhabitants, but these are adjacent to landforms that postdate AD 950 and may include obliterated mounds, so populations may have been higher in the past. Overall, the small size of these districts suggests that most members probably were familiar with each other given the natural community size of many human social networks (Hill and Dunbar Reference Hill and Dunbar2003). In contrast to “urban” settings, these districts would have been populated by people who had a high degree of social familiarity but lower intensity of face-to-face interactions than those generated by the effects of energized crowding in city centers (Lamb Reference Lamb, Hutson and Ardren2020; Smith Reference Smith2023). Therefore, we propose that focal nodes in these rural districts may have helped formalize and regularize social interactions through labor organization, communal ceremony, and other activities.

Figure 8.

Regional settlement map showing districts and focal nodes. (Color online)

Figure 8 Long description

The figure is a map showing a meandering river system rendered in shades of blue, with the active river channel in dark blue and infilled or former channels in lighter blue. Curving cutoff channels and oxbow lakes trace the river's past courses across the floodplain. Green shaded areas and small symbols mark mounds and focal nodes, several grouped within dashed outline zones and labelled. A legend in the lower left identifies the features, including focal node, mound, active river channel, infilled cutoff channel, cutoff channel, and oxbow lake. A north arrow and scale bar provide orientation and distance reference.

Table 5.

Population Estimates and Settlement Pattern Summary for Defined Districts within the UUCZ.

Table 5 Long description

District-level figures list area in hectares, counts of structures and focal nodes, and low, medium, and high population estimates. Among named districts, Altar de Sacrificios has the most structures at 64 and the highest population range at 163 to 254, with 4 focal nodes despite being only 41.8 hectares. El Pabellon covers the largest area at 227.7 hectares but has 53 structures and a lower population range of 135 to 210. El Trapiche is similar in structures at 55 with a population range of 140 to 218, while El Trapiche 2 is smaller at 18 structures and 46 to 71 people. The smallest named district by structures is Ixcoche 3 with 9 structures and 23 to 36 people; Grupo Frontera has 12 structures and 31 to 48 people. An unassigned category contributes 131 structures and a population range of 334 to 520, indicating a substantial share not tied to a named district. A subtotal for the UUCZ excluding post 950 AD landforms reports 1,097 structures and a population range of 1,359 to 1,707. The table also estimates 153 mounds destroyed, associated with 390 to 607 people, and the overall total estimated UUCZ population is 1,487 to 2,315. Population values are estimates presented as low, medium, and high ranges, so comparisons should be treated as approximate rather than exact counts.

Rurban Function and Agrarian Land Use

Lastly, we shift our attention to the “vacant” spaces in between these sparsely settled areas to consider potential agrarian activities in the countryside. Frequent floods in the UUCZ deposit rich alluvial soils across this broad plain, making it well-suited to cultivation (Horseman et al. Reference Horseman, Morell-Hart, Golden and Scherer2024). However, such low populations would not have placed great demand for intensive agriculture as in other parts of the Maya lowlands. The PAALS survey identified only one possible drainage canal and no ditched fields or terraces like those documented in more densely settled regions downstream (Golden et al. Reference Golden, Scherer, Schroder, Murtha, Morell-Hart, Diaz and del Pilar Jiménez Álvarez2021; Schroder et al. Reference Schroder, Murtha, Golden, Hernández, Scherer, Morell-Hart, Almeyda Zambrano, Broadbent and Brown2020). Given the settlement’s close proximity to water, riparian resources procured from fishing and collecting may have made significant contributions to the local diet (Sharpe et al. Reference Sharpe, Palomo, Inomata, Triadan, Pinzón, Curtis and Emery2024) in addition to cultigens other than maize (Morell-Hart et al. Reference Morell-Hart, Dussol and Fedick2023).

Another possibility we entertain is that these unmodified, “vacant” areas could have also supported silviculture of cacao and other fruit-bearing trees. Agroforestry was a productive and sustainable strategy of resource management for the ancient Maya (Ford and Nigh Reference Ford and Nigh2009, Reference Ford and Nigh2016). However, cacao is a finicky plant requiring particular conditions for cultivation: consistently moist deep alluvial soils, high temperatures, heavy rainfall, and a shady canopy offering protection from sun and wind (Bletter and Douglas Reference Bletter, Douglas and McNeil2006)—settings not unlike those found in the greater UUCZ. Indeed, there is historical evidence of cacao production along the Pasión and Machaquila rivers during the sixteenth and seventeenth centuries (Caso Barrera and Aliphat Reference Caso Barrera and Aliphat2006). Agricultural censuses of Mexico from 1950 and 1960 respectively estimate 1,400 and 5,000 fruit-bearing cacao trees in the UUCZ’s southern sectors (Dirección General de Estadística 1957, 1965), which could have produced 1,300 to 2,000 kg of dried beans annually. A few local residents still have cacao and other fruit-bearing trees growing in their home orchards today, attesting to the continued presence of silviculture in the region.

During the Classic period most cacao-growing regions were located along the coast, but the UUCZ is one of the few lowland interior zones where cacao could have been cultivated and easily transported due to its environmental setting and navigable waterways (Baron Reference Baron2018:Figure 1). Other lowland Maya riverine settlements, like those in the Sibun Valley and along the Macal River in Belize, were also locations of cacao production during Classic and Postclassic times (McAnany et al. Reference McAnany, Thomas, Morandi, Peterson, Harrison, Masson and Freidel2002; Muhs et al. Reference Muhs, Kautz and MacKinnon1985). Tentatively, we hypothesize that the “vacant” areas in this sparsely occupied fluvial landscape may have contained cacao groves and other valuable tree crops that could have been cultivated for local consumption and export, especially as demand for cacao increased during the Late Classic period (Baron Reference Baron2018; McAnany Reference McAnany2010). Focal nodes in the rural districts may have helped coordinate and manage these activities by storing surplus or arranging shipment of these goods. This interpretation might help explain the paradoxical pattern of such low-density settlement alongside the concentration of such elaborate urban forms in the Altar epicenter, but additional lines of evidence are needed to further test these hypotheses.