Constructing Greater Cahokia: Site Concentrations and Visibility Surrounding America’s First City

2.1.1. Study Area: Greater Cahokia

The Central Cahokian polity flourished between 1050 and 1350 CE, reaching an urban population numbering several thousand people (Emerson, Koldehoff and Brennan 2018; Milner 1998; Pauketat and Lopinot 1997). Central Cahokia is located approximately twelve kilometers (seven miles) from contemporary downtown St. Louis, Missouri, straddling Madison and St. Clair counties in Illinois, encompassing an area of about 16 square kilometers (Dalan and Holley 2003; Fowler 1977; Mink and Iseminger 1999). Most of Central Cahokia is protected by Cahokia Mounds State Historic Site and is a designated UNESCO World Heritage Site.

Greater Cahokia encompasses the regional villages and sites surrounding Central Cahokia, situated within the Mississippi floodplain known as the American Bottom. It developed concurrently with the city of Cahokia and had a population numbering in the tens of thousands. (Betzenhauser and Pauketat 2019; Emerson, Koldehoff and Brennan 2018). The exact reasons for this expansion are not fully understood but likely tied to concomitant advances in food production, cultural cohesion, and religious systems (Baltus et al. 2022; Fowler 1978; Kelly 2002; Pauketat 2004). Greater Cahokia today exists among the modern urban developments of St. Louis and is therefore subject to increased risk of damage, destruction, and looting.

An important element for Cahokian culture was mounds, where ceremonies would take place, ancestors were honored, and the elite would reside (Pauketat 1993). Many mounds at Greater Cahokia, primarily platform mounds, had large wooden buildings atop them, and both temple and residential areas usually contained a central courtyard and ancestor pole (Pauketat 2009). Central Cahokia had the most impressive array of plazas and mounds, namely the Grand Plaza (Holley, Dalan and Smith 1993; Vilbig, Sagan and Bodine 2020) and Monks Mound (Romain 2017; Schilling 2012) (Figure 2), but this was a design repeated throughout the region (Emerson and Lewis 1991).

Figure 2

Archaeologists have long known about site specialization and differentiation at Greater Cahokia and Central Cahokia (Emerson, Koldehoff and Brennan 2018; Fowler 1978; Griffin and Spaulding 1951; Kelly 2002; Milner 1990; Munson 1971; Pauketat 2003). Considering the intentional razing of buildings and features at Greater Cahokia (Baltus and Wilson 2019; Pauketat 2009; Pauketat et al. 2013), the Cahokian elite and nonelite consciously reshaped their landscape to meet their needs. This regular redevelopment of their cultural landscapes meant Cahokians could enact new social standards to encourage political order and engagement in socio-religious ceremony.

Previous work has shown an elite imposition of order and landscape manipulation within Central Cahokia and the outlying mounds group (Alt, Kruchten and Pauketat 2010; Betzenhauser and Pauketat 2019; Pierce and Matisziw 2018; Rolingson 1996). The primary mounds at each mound group would have had a commanding view of the surrounding landscape and functioned as a visual landmark for the population (Dalan and Holley 2003; Pierce and Matisziw 2018; Reed, Bennett and Porter 1968). Others have theorized that this could have been done for the purpose of surveillance and the need for Cahokian elites at Central Cahokia to make things seeable while imposing a sense of order (Pierce and Matisziw 2018). This idea builds on Jeremy Bentham’s Panopticon model which focuses directly on how landscape structure, in terms of buildings and layout, imposed surveillance and control from a particular point (Bentham 1791).

The organization of labor necessary to build the mounds and plazas with such a coherent design suggests a “top-down imposition of order” (Betzenhauser and Pauketat 2019, p. 141). The level of construction expertise and organization necessary for building the monumental centers of Cahokia and the other mound groups (Alt and Pauketat 2011; Cobb and Butler 2017; Pauketat 2009) may indicate that the elites of Cahokia were more than just figureheads and likely had meaningful influence over labor efforts across the region.

Connected to theories of labor and site organization, there has been much debate over the purpose and complexity of the civilization at Greater Cahokia. The dominant view today holds that Greater Cahokia functioned as a complex chiefdom (Beck 2006; Butler and Welch 2006; Byers 2006; Cobb 2003; Trubitt 2000), where satellite mound groups and village sites functioned more as independent polities with a primarily ceremonial and kinship connection to Central Cahokia. This view holds that the ruling elite at Central Cahokia held limited direct influence over the population in the hinterlands, who were primarily connected through a shared cultural, socio-religious ideology, and/or kinship identity. This debate continues to evolve as further research at the site uncovers more about the unique and complex sociopolitical, administrative, and economic systems of Greater Cahokia, which remain challenging to fully explain (Alt 2010, 2011; Blitz 2010; Holt 2009; Pauketat 2007).

It should also be mentioned that construction designs at Greater Cahokia and other Mississippian centers exhibit strong cosmological connections and alignments (Baires 2015; Baltus and Baires 2012; Chappell 2002; Kelly and Brown 2012; Pauketat, Alt and Kruchten 2017). The significance of these cosmological alignments at Central Cahokia and the satellite mound groups are now well studied (Baires 2014; Dalan and Holley 2003; Fowler 1997; Rolingson 1996; Romain 2017; Smith 1992) and therefore not the focus of this work.

Chronology for Cahokia is divided into four phases: Lohmann (1050–1100 CE), Stirling (1100–1200 CE), Moorehead (1200–1275CE), and Sand Prairie (1275–1350 CE) (Fowler 1978, 1997; Hall 1975; Iseminger 2011). These phases fall into the greater period referred to as Mississippian, which aligns with the height of Cahokian political power in the region. The Mississippian period corresponds to the construction timing of mounds, plazas, causeways, and densest residential areas at both Central Cahokia and Greater Cahokia, with local variations. Most large public construction projects at Central and Greater Cahokia occurred during the Lohmann and Sterling phases, meaning most of this infrastructure was in place early in Cahokia’s development. This research looks at the full Mississippian period and not individual phases due to limitations in the database coding scheme. Any conclusions are thus restricted to Mississippian patterns broadly and not specific Cahokian phases.

2.1.2. Archaeological Intervisibility and Community Design

Archaeologists have long understood the importance of geography, natural and human made, in the interpretation of archeological information (Gillings, Hacıgüzeller and Lock 2020; Llobera 2007; McCoy and Ladefoged 2009; Ogburn 2006; Wheatley 1995). Geographic variables exert a strong influence on human behavior; this was true as much in the past as it is in the present. Greater Cahokia was no exception, with many associated sites located along natural ridges in the landscape, particularly on the east-west ridge running from central Cahokia to the Mississippi riverfront (Fowler 1978; Harn 1971), referred to as the Central Administrative-Political Complex (Pauketat 1994).

The use of planned construction strategies to project power, encourage social cohesion, and convey information was a key component of cultures found throughout the world and throughout time. This “Geography of Power” (Randle 2011) was of immense importance to ancient societies, with archaeological structures strategically constructed at a site for prominence, resource access, religious importance, and environmental advantage (Bernardini and Peeples 2015; Bilotti et al. 2024; Carrero-Pazos 2019; Jones 2017). According to this theory, spatial arrangement of residences and ceremonial structures were important considerations as they served to unify a population through repeated interactions, shared rituals, and daily exchanges (Malouchos and Betzenhauser 2021). A more clustered and denser built environment suggests a society organized around communal ideals and collective efforts; understanding these designs informs our understanding of community patterns within a complex society (Peterson and Drennan 2005; Thompson, Meredith and Prufer 2018).

There are many examples throughout the world of cultures planning their settlements with a priority placed on intervisibility among site locations (Brughmans and Brandes 2017; Gillings and Wheatley 2020; Lake and Woodman 2003; Wheatley 1995). Leaders throughout the pre-Columbian Americas planned site intervisibility for multiple reasons. Diverse examples from North American sites range from small, such as a series of basic hearths which lie along the barren rim of Seminole Canyon in Texas (Turpin 1984), to the more elaborate hilltop platforms in Northwest Mexico (Swanson 2003) or Kivas in the Chacoan area of New Mexico (Kantner and Hobgood 2016). Often, cultures build structures in visually prominent ways to demarcate boundaries and influence (Bernardini and Peeples 2015; Llobera 2003) and for defensive advantage (Jones 2006). Outside of North America, there are extensive examples highlighting the importance of visibility in cultural landscapes (Bongers, Arkush and Harrower 2012; Brughmans, van Garderen and Gillings 2018; Earley-Spadoni 2015; Gillings 2015; Kay and Sly 2001; Lake and Woodman 2003; Llobera 2007; Lopez-Romero Gonzalez De La Aleja 2008; Marsh and Schreiber 2015; Smith and Cochrane 2011; Wheatley 1995; Wright, MacEachern and Lee 2014).

These cases suggest that the use of geography, both planned and natural, to establish power, communication, and social cohesion was a common component of human cultures across time and space. This article investigates how Greater Cahokia may have been built with similar considerations, a topic that other researchers have speculated but did not have the data and technology to assess empirically across the region (Fowler 1978). Determining the reason for this visual preference is a complicated issue, and this research instead first establishes that a visual network indeed exists across the Greater Cahokia landscape. It is also important to note the limitation of visibility studies in archaeological settings, as time, natural processes, and human hands have altered these landscapes.

2.2.1. Archaeological Points and Random Point Datasets

To effectively evaluate intervisibility, two sets of point data are required: archaeological site locations and randomly generated but constrained control points. Comparing these points allows us to determine whether intervisibility was more strongly associated with archaeological sites than would be expected by chance.

The archaeological point data for Illinois was compiled and updated by the Illinois State Historic Preservation Office (ISHPO; https://dnrhistoric.illinois.gov/preserve/archaeology.html) and was generously shared with this research team. The Illinois Inventory of Archaeological Sites (IIAS) database was subset to the 809 sites Mississippian period sites located within both Madison and St. Clair Counties, which was the study area for this study and approximately follows the extent of Greater Cahokia (Figure 3). The IIAS data is the authoritative catalog of archaeological sites in the state and a fascinating quantitative dataset for analysis. Although this dataset is authoritative, it is limited in its temporal resolution and only identifies features within the Mississippian period and not the phases within this period. This limitation means we cannot say that all features surveyed had coeval intervisibility, but rather these findings apply to general landscape visibility trends at Greater Cahokia yet do not model exact, specific realities Cahokians encountered.

Figure 3

Data for IIAS comes from excavations, pedestrian surveys, and citizen scientists. As with all archaeological data, it cannot be considered complete due to intrinsic limitations of archaeological fieldwork and thus risks misrepresenting Greater Cahokia’s true settlement patterns and extent. Regardless, these data are as comprehensive and current as exists today. It should also be noted here that this point data will necessarily mislead readers – Cahokia existed in the region not as individual points on the landscape but as a rich and interconnected tapestry of settlements, farmsteads, cultivated areas, markets, plazas, and mounds. This work therefore modelled observers positioned on built features as they look towards structures and mounds.

As this study was interested in the spatial patterns of residential and ceremonial locations, the dataset was further subset into sites identified as either a habitation, a mound, or both. Of the 809 sites, 679 were classified as habitation, 88 were mounds, and 70 were both mound and habitation sites. This subset of data was then used to create two feature classes in ArcGIS Pro 3.5.2.

To aid our interpretation of the results and add control, a dataset was created of an equal number of randomly distributed points within the mound groups’ viewshed (discussed in section 2.3.2.) for both habitations (638 points) and mounds (85 points). Placement of these random points excluded riverways and lakes that do not represent viable areas for construction. Water areas come from the Census Bureau’s TIGER waterbodies shapefiles for both counties. Although the landscape’s waterways would have differed one thousand years ago, this current data still includes general low-lying areas that would have been present during the Mississippian period. With the ‘Near’ tool, the minimum distance between mound sites and habitation locations was ascertained; 146.388 meters was the minimum mound separation, and 48.358 meters was the minimum separation of habitation sites. Using these parameters and the boundary of the mound groups’ viewshed, the ‘Create Random Points’ tool creates an appropriately constrained dataset of points representing random locations for mound and habitation sites. This allows for a direct comparison between locations in the landscape, thereby identifying trends in the archaeological record as compared to randomness.

2.2.2. Mound Groups

Mound groups (Figure 3) were important public and ritual centers at Greater Cahokia, containing plazas surrounded by multiple mounds that served as seats of both ceremony and power (Topping 2010). Understanding mound group interactions within the regional context of Greater Cahokia provides important insights into the political structure present and speaks to levels of state organization at Cahokia. Mississippian mound groups have been variously identified in the literature (Fowler 1997; Moorehead 2000). This study uses the mound groups as specified by the National Park Service’s “Cahokia Mounds Reconnaissance Survey” from 2016 (http://npshistory.com/publications/srs/cahokia-mounds-rs-2016.pdf), which provides a good synthesize of Cahokian archaeology while also identifying high priority features for preservation protections. All identified mound groups were constructed early in the Mississippian period, either in the Lohmann or Sterling phase, and as such would have existed during the Mississippian occupation of Cahokia.

The NPS report focuses on the mound groups that represent the best opportunity for preservation in a region-wide cultural park and not necessarily a comprehensive list of every mound group in the region. This report identifies five known peripheral mound group sites: St. Louis, East St. Louis, Mitchell, Pulcher, and Emerald. Emerald notwithstanding, these mound groups are what Fowler (1978) termed “Second-Line Communities”. These communities correspond to high density population and ritual areas at Greater Cahokia. This selection of mound groups therefore falls within current county lines of Greater Cahokia but is not exhaustive of all possible Mississippian mound groups. Selected mound groups nonetheless provide insight into connections between the central groups that are of most interest for future preservation and regional park planning efforts.

For this study, Central Cahokia was treated as a mound group comparable to others in the region, since all shared a mound-plaza layout and high population density. However, Central Cahokia stood apart in both magnitude and influence. To some degree, each of these mound groups have been impacted by modern development, whether through destruction to make way for a railroad, as with the St. Louis Group, or extensive agricultural plowing, as with the Pulcher Group. In the case of mound groups that had been leveled, the best height estimate based on the NPS survey and excavation maps was used to recreate an observer point atop the primary temple. The mound-group layer was thus composed of an observer-point location for each of the peripheral mound groups and Monks Mound at Cahokia proper, making six observer points in total that recreated elite viewpoints of the surrounding landscape (Pauketat 1993; Pierce and Matisziw 2018).

2.2.3. LiDAR Data and Digital Elevation Model Creation

This research was built on previous viewshed studies, applying this methodology to understand Greater Cahokia’s visibility design. The analysis focused on the boundaries of Madison and St. Clair counties and the riverfront of St. Louis to include all the primary mound groups. This focus was chosen because the IIAS archaeological sites were subset to these two counties to roughly align with the boundaries of Greater Cahokia in the American Bottom. To study intervisibility, LiDAR data for the study area was obtained and transformed into a raster terrain surface.

Light Detection and Ranging (LiDAR) data were used for the surface model. LiDAR is an active remote sensing system that emits laser pulses, usually in the infrared frequency, to scan a target. These pulses bounce off targets, which are recorded as corresponding intensity and time-to-return data. Pulse return data detail target footprints, height, and sometimes composition. These data files are then interpolated into a 2.5D raster terrain surface, called a Digital Elevation Model (DEM). LiDAR has become immensely popular with archaeologists, leading to major discoveries that have greatly expanded our understanding of the complexity of various sites and civilizations (Canuto et al. 2018; Chase et al. 2012; Fisher et al. 2016; Rostain et al. 2024).

The LiDAR data used in this study came from two sources that are publicly available. The Illinois data are from the Illinois Height Modernization Project. Data for Missouri came from the United States Geological Society’s 3D Elevation Program. These point clouds were transformed into raster surfaces individually before being mosaicked into one in a bare earth Digital Terrain Model (DTM, Figure 3). To do so, LiDAR return values within the dataset were filtered to only ground returns, i.e. excluding vegetation and buildings. The resulting raster has a 1-meter pixel resolution.

During the Mississippian period, tree abundance in the region declined significantly, but not entirely (Munoz et al. 2015; Rankin, Barrier and Horsley 2021), leading us to conclude that a bare earth model was a reasonable attempt at modeling the past landscape. While recognizing that this is an imperfect representation of the past landscape, we believe it to be a justifiable model given our present constraints and knowledge of Greater Cahokian land use practices. Additional details on the LiDAR data and DTM creation for this study are in the Appendix.

2.3.1. First-Order Spatial Analysis

To understand the arrangement and clustering of mound and residential spaces at Greater Cahokia, we began by analyzing first-order properties of Average Nearest Neighbor and Kernel Density Estimation to measure the overall distribution of our archaeological points. First-order spatial analysis provides insights into the spatial organization and behavior of past populations and helps archaeologists address research questions related to settlement patterns, land use, resource exploitation, and social interaction (Bivand, Pebesma and Gómez-Rubio 2013; Cressie 2015; Diggle 2013). The goal was to identify specific areas of higher use and importance for regional habitation to guide both future excavation and preservation work.

Average Nearest Neighbor (ANN) measures the distance between each point in a dataset and its nearest neighbor’s location to assess the degree of clustering or dispersion in the data (Kintigh and Ammerman 1982; Thompson et al. 2022; Thompson, Meredith and Prufer 2018). An area input was necessary to limit the extent of analysis and ensure a fair comparison of the observed and expected values. The equation for ANN is the ratio:

Where ${\underset{¯}{D}}_O$ is the observed mean distance between each point and its nearest neighbor and ${\underset{¯}{D}}_E$ is the expected mean distance for the points of a given random distribution. Thus, if the ANN ratio is less than one, we can say the data are clustered, and if it is greater than one it is dispersed. ANN gives a z-score to indicate averaged standard deviations of observations from the mean of the dataset and a p-score for the confidence interval. ANN was performed for both mound sites and habitation sites in the dataset.

Kernel Density Estimation (KDE) is another prevalent first-order analysis technique in archaeology due to its ability to improve spatial interpretation of a set of distributed observations. KDE can map spatial clustering and reveal areas of varied utilization (Baxter, Beardah and Wright 1997; Bonnier, Finné and Weiberg 2019; Carrero-Pazos 2019; Thompson et al. 2022). KDE uses kernel smoothing for probability density estimation, calculating an intensity or magnitude per unit area for a variable using a kernel function to fit a smoothed surface to each observation. The resulting continuous surface map displays gradients of the variation in the intensity for the variable and is easily interpreted. For KDE in a two-dimensional space, the equation is:

Where f(x, y) is the estimated density at a location, n is the number of point observations, h is the bandwidth or smoothing parameter the controls the size of the kernel, (x_i, y_i) are coordinates of the i-th point, and K is the kernel function. KDE is performed for both mound sites and habitation sites in the dataset. In ArcGIS, kernel density estimates for mound and habitation sites from the entire IIAS database are created using the geodesic method with output cell size determined by the extent of the data layer. The search radius was set to the observed mean distance of each layer’s ANN analysis (see Table 1).

2.3.2. Linear Line of Sight Analysis

With line-of-sight analysis, it is possible to explore visibility in a landscape. This work models visibility to and from key locations at Greater Cahokia, i.e. at the densely populated mound groups, to understand the region’s visibility. Viewshed analysis has previously been performed at Central Cahokia (Pierce and Matisziw 2018), but this was the first regional viewshed analysis ever done for Greater Cahokia. Defining a viewshed and sight limits was not straightforward, so this research leans on the work of others (Brughmans, van Garderen and Gillings 2018; Gillings and Wheatley 2020; Jones 2006; Kantner and Hobgood 2016; Lake and Woodman 2003; Marsh and Schreiber 2015; Ogburn 2006; Pierce and Matisziw 2018; Wheatley 1995; Wright, MacEachern and Lee 2014) while acknowledging its own limitations.

Line of sight analysis allows us to understand the visual elements and networks across the Greater Cahokian landscape. Sight lines were restricted to 20-kilometers based on previous work to assess visual limits in landscapes (Kantner and Hobgood 2016; Ogburn 2006). Other visual limits have been proposed and used in other studies (Gillings and Wheatley 2020; Higuchi 1985; Ogburn 2006), as there is variation in visual limits depending on the modeling scenario. This work models intervisibility between mounds, temples, and house locations, e.g. structures that are several meters wide, making a 20-kilometer visual limit an appropriate distance for the Greater Cahokia setting (Kantner and Hobgood 2016). Therefore, buffer polygons of 20-kilometers around the mound groups were created and clipped to the extent of the counties using the ‘Pairwise Buffer’ and ‘Clip’ tools. This visual limit assumes photopic vision, meaning well-lit conditions and a viewer with good vision, resulting in 723 sites – 638 habitation sites and 85 isolated mounds.

The line-of-sight analysis was based on viewpoints from atop primary mounds at each mound group and divided into two sections; visibility of mound sites compared to an equal number of random points and visibility of residential sites compared to an equal number of random points. Intervisibility between mound groups and individual mounds and residential locations was done to understand how Cahokia elites, who lived on and visited mound tops (Dalan and Holley 2003; Emerson 1997; Pierce and Matisziw 2018), would be able to view the cultural landscape before them while those in that landscape would be able to simultaneously view activities, like ritual events and celestial alignments, happening atop the mound groups. In this way, mound groups served as cultural nodes that connected the population across the landscape.

The visibility of sites started with clipping the archaeological sites to exclude locations outside of the 20-kilometer buffer. Then, datasets of random points equal to the number of Mississippian mounds (85 mounds) and habitations (638 sites) within the 20-kilometer buffer were created. This allows for a more accurate comparison between archaeological sites and random points. Using the ‘Linear Line of Sight’ tool, sight lines were assessed for an observer atop each mound group. Observer heights are set to the primary mound group locations, and target sites are set to the IIAS archaeological dataset points. Observer height from the mound groups was set to 4-meters above ground, as the primary mound groups have been levelled and are now at or close to ground height. This allows an approximate mound viewpoint, as most mounds were approximately between 3- and 6-meters (Bushnell 1904; Fowler 1997). Target height for archaeological sites was set to 2-meters to approximate human height – ArcGIS only allows integers, so the decision was made to round up to 2-meters.

2.3.3. Radial Line of Sight Analysis

In addition to linear line of sight, the ‘Radial Line of Sight’ tool was used to study the total viewshed present at Greater Cahokia. Radial line of sight maps areas that are visible from one or more observer locations. This creates a cumulative viewshed (Wheatley 1995), where the total visible landscape from specific cultural points, i.e. mound groups at Greater Cahokia, was revealed. The same parameters were used as in linear line of sight, with a 20-kilometer radius and 2-meter observer height at each mound group. This tool outputs a polygon shapefile that indicates visible areas and the count of mound groups to which they are visible based on the input surface raster. This highlights visibility across the entire landscape rather than just at specified points, which can reveal not just what was visible but also what was hidden from observer locations, specifically the elite present atop mounds at Greater Cahokia.

2.3.4. Analytic Synthesis

To develop a more complete picture of the Greater Cahokia landscape, we combine the site clustering and viewshed results. The output polygons from the KDE analysis and radial line of sight were overlapped to identify areas of higher site density with visible and hidden areas at Greater Cahokia, highlighting landscape use patterns. Through this synthesis we can map areas of high ceremonial and household use and their relation to visibility and invisibility from the mound, or elite, centers within the Greater Cahokia region. These results reveal patterns of visibility and site use across the landscape, indicating locations of high importance for future research and conservation.

3.1.1. Average Nearest Neighbor

Outputs from ANN are presented in Table 1. The area was constrained to the combined land surface area of the Illinois counties (excluding water surface) and converted to square meters to match the map projection. This allows ANN to work under the assumption that the points may be freely located within the study area. Results for both mound and habitation are less than 1, indicating that the locations of each site type exhibit clustering. The p-value for both site groups was less than 0.01, meaning that this pattern was not due to random chance. The very low (negative) z-score values for both groups also indicate that the observed spatial distribution of these points are statistically significant and very unlikely to be random, as they are in the extreme tails of a normal distribution.

3.1.2. Kernel Density Estimation

Kernel Density Estimation results are a valuable tool to identify areas of clustering in a dataset. KDE is the process of grouping together spatial data points into clusters based on their geographic proximity. KDE creates an intensity gradient that identifies high use areas in the dataset. In our IIAS dataset, KDE reveals significant spatial clustering of mound and habitation sites (Figure 4). Among habitation sites, the areas of highest clustering occur 4–5 kilometers southwest of the Mitchell Mond Group, along Horseshoe Lake just northwest of the Monks Mound Group (central Cahokia), 10 kilometers northeast of the Pulcher Mound Group along the bluff line in present-day Shiloh, Illinois, and around the Emerald Mound Group. It may be appropriate to consider these clusters as important neighborhoods at Cahokia, due to the intensity of settlement sites located in each area (Betzenhauser and Pauketat 2019).

Figure 4

For the mound sites, there was significant clustering around central Cahokia, extending along the Central Political-Administrative Complex towards the East St. Louis and St. Louis groups (Emerson 2002; Pauketat 1998), and at various locations along the bluff lining the America Bottom floodplain. We also observe mound clusters around each of the primary Mound Groups studies. It was interesting to note that aside from the sites of central Cahokia and the Emerald Mound Group, there was variable overlap between habitation and mound sites around Greater Cahokia.

3.2.1. Habitation Visibility Network

Within the IIAS dataset, there are 638 habitation sites within 20-kilometers of a mound group. This fact alone was interesting, as the total record of habitation sites in the dataset equals 679. This means that only 6.1% of habitations (41 sites) exist further than 20-kilometers away from a mound group, but a total of 41.3% of the land area (151,104 hectares of the total 366,206 hectares from the two counties) was outside of these viewsheds. This tells us that, despite needing land for cultivation, Cahokian residents preferred homesteads that were closer to ceremonial and civic life. Such nucleation would help consolidate and direct resources like food and ceramics made in villages. From the linear line of sight analysis, of the 638 habitation locations, 453 were visible from a mound group (71%) whereas 185 were hidden (29%) (Figure 5a).

Figure 5

Using the same area and number of observations with random points, the results are again telling; 280 random points are visible (43.9%) but 358 are not seen (56.1%) from a mound group (Figure 5b). As with any visibility interpretation presented here, any conclusions of planned design must be tempered by model limitations. We can therefore only state with confidence that the Mississippian sites in this study were likely intervisible with one another; however, they were not all built or occupied at the same time, even though the mound groups were built early in the Mississippian Period.

3.2.2. Mound Visibility Network

It was crucial to understand if intervisibility was key for mound sites, as these were connected networks of ceremonial locations potentially used to maintain and encourage participation in the social order. For the 85 mounds in the IIAS dataset that are within twenty radial kilometers of a mound group, 72 are visible (84.7%) to at least one mound group and only 13 are not visible (15.3%) (Figure 6a).

Figure 6

Compared to the same number of random ‘mound’ sites in the landscape (Figure 6b), there was an obvious discrepancy: 48 random ‘mounds’ are visible (56.5%) while 37 are not visible (43.5%) (Table 2). This indicates that locations that are more visible from mound groups were preferred by the builders of Greater Cahokia for construction of mound features. We therefore can confirm the presence of intervisibility in the Greater Cahokia built landscape, which may have been an important factor in deciding locations for both residential and ceremonial features.

3.2.3. Radial Line of Sight Analysis

Interpreting output from a radial line of sight analysis displays the total viewshed for the mound groups within the Cahokia DTM. The resulting polygon (Figure 7) identifies visible locations and the number of mound groups from which that location was visible, as well as the areas that are not visible. The area was again constrained by a 20-kilometer radius to simulate the extent of human visual acuity. The primary findings from this analysis are that bluff ridges and the St. Louis riverfront are among the most visible locations within the landscape, being generally visible by multiple mound groups. We also see that the American Bottom floodplain has good visibility, though much was nonetheless hidden or only seen by one mound group. Given that it is a relatively flat flood plain, it was an interesting finding that there are many non-visible locations. This is not a ubiquitously flat landscape, likely due to both natural topographic variation and some modern landscape changes. Also of interest was the area surrounding the Emerald Mound Group. With the bluff line creating a visual barrier from the other mound groups, the area around Emerald was only visible to Emerald. This identifies the Emerald Group as a singular group with unique visual characteristics.

Figure 7

3.3.1. Habitations

Overlaying the results of KDE and radial line-of-sight analysis revealed areas of high use and visibility at Greater Cahokia, highlighting where populations lived and whether these areas were intervisible with mound groups or concealed from them (Figure 8). Among the habitation KDE, we located residential clusters related to high and low visibility areas. High habitation and high visibility were present around Horseshoe Lake, near the bluffs due south of East St. Louis, and the areas surrounding Emerald Mound Group. We also see residential clustering but low visibility in the uplands east of the Pulcher Group and the East St. Louis Group, as well as areas north of Mitchell. There were also some habitation clusters that exist outside of the visual limits of the mound groups, to the extreme north of Mitchell and southwest of Emerald, within the Richland Complex.

Figure 8

Residential clustering with low visibility was present in the uplands east of the Pulcher Group and the East St. Louis Group, north of Mitchell, and in the uplands southeast of central Cahokia (an area called the Richland Complex). The Richland Complex was of particular interest, as it was the location of many migrant homesteads that had a different material culture and homeland than the elite of Cahokia (Alt 2002). Their location in the bluff uplands was spatially close to central Cahokia, but in many ways they were very distant from the elite activities at Cahokia. It has even been submitted that the people more likely to be sacrificed at Cahokia were drawn heavily from the migrant groups living in the uplands (Ambrose, Buikstra and Krueger 2003). The current research reveals another curious fact about the Richland Complex – its population was restricted from regularly viewing the ceremonial activities occurring at the mound groups. As they were generally a resettled and, in some ways, marginal group at Cahokia (Alt 2002; Pauketat 2003), it may be reasonable to posit that viewing ritual activities happening atop mounds was a component of the social hierarchy that was controlled based on social standing. However, residents of the Richland Complex—along with other non-elite groups—were also invited to participate in rituals at the mound center (Pauketat 2003; Pauketat et al. 2002), complicating any straightforward interpretation of visual exclusivity. Since it was roughly a day’s walk from the Richland Complex to central Cahokia (Alt 2002), the frequency of visits, combined with controlled visibility to the mound groups’ ceremonies, may have played a role in reinforcing the social hierarchy.

There were additional residential clusters that exist outside of the visual limits of the mound groups, to the extreme north of Mitchell and southwest of Emerald. These habitation clusters around Emerald were curious as Emerald was notable for having a low-density occupation (Pauketat, Alt and Kruchten 2017). These clusters were thus a result of small but dense habitation, further highlighting the uniqueness of Emeral. Greater Cahokia’s residential clusters were clearly widespread and had a large presence on the landscape.

We also see areas of high visibility that were lacking any site clusters. Most obvious are the bluffs overlooking the American Bottom, where there are some sparse residential clusters but only small and sporadic. These bluff-top mounds are frequently mortuary mounds with associated villages (Emerson, Hedman and Fort 2024), but the villages were modest and therefore do not register as significant habitation clusters at this regional scale. High visibility areas between East St. Louis and Central Cahokia existed, with small residential clusters present. Further, the areas around Monks Mound exhibited good visibility but have modest habitation clusters due to a lower density of habitation. We may thus interpret these areas as being reserved for elite members of Cahokia society, corresponding to a smaller population and fewer houses, which aligns with excavations from the area (Iseminger 2011; Pauketat 2009), but where residents may have been more visible and in the public eye.

As a final remark on KDE and neighborhoods, there was general variability between mound groups and the presence of habitation clusters. There were documented habitations in the vicinity of each mound group, yet only Central Cahokia and Emerald have notable habitation densities present. This was likely due to the regional scale of this analysis and the excavation histories of each group. It was interesting, nonetheless, that Emerald exhibits a strong habitation density, as it was a pilgrimage site with a relatively small permanent population (Pauketat, Alt and Kruchten 2017). We suspect it was likely that Emerald shows a higher density of habitation in part due to its concentrated hilltop location and attendant population (Pauketat, Alt and Kruchten 2017; Skousen and Huber 2018). Considering that Emerald was also the most removed mound group (being the furthest distance from the next nearest group) and its unique religious role, the societal functions and needs of Emerald are clearly differentiated from the other mound groups.

3.3.2. Mounds

KDE of mounds revealed clustering with high visibility to be present at Central Cahokia, along the bluff line of the American Bottom (particularly in the north), and centered around the Mitchell Group (Figure 9). There was some clustering around each of the other mound groups, but the clustering was slight, and visibility areas were both spotty and generally limited in number of observers. As mounds often served a ceremonial purpose for the elite (Iseminger 2011), and these mound clusters were in highly visible areas, we may conclude that these mound clusters were used for important rituals meant to be seen widely by the Cahokian populace. We speculate that each of these mound groups also offered a vantage point for the elite to observe the population, allowing them to regularly monitor people and the landscape (Pierce and Matisziw 2018).

Figure 9

The only locations of mound clusters with low visibility occurred to the west and southwest of the Emerald Mound Group and far north of Mitchell. All locations only exhibited small density values, but they are nonetheless interesting to consider. As mentioned in section 3.3.1, Emerald served a unique function, as a religious pilgrimage site. These mound clusters to the west of Emerald may have served as part of a religious or ceremonial complex, greeting and observing pilgrims from Greater Cahokia as they entered the sacred Emerald site. As these mound clusters were positioned strategically between Emerald and the other mound groups, they may have even served as locations for arcane and specialized religious activities. Being in the uplands, they were not designed to be clearly visible to the other mound groups, further suggesting an intentionally removed, discreet, and ceremonial purpose to these mounds. Expanding excavation efforts in these areas could offer critical evidence to either support or refute these claims.

The mound clusters north of Mitchell were all positioned east or southeast of the Mississippi-Missouri river confluence. Given the use of rivers as trafficways during the Mississippian period (Wood 2018), this would have been a critical location for Cahokian infrastructure. Further, the radial line of sight analysis confirmed that mound groups Mitchell, East St. Louis, St. Louis, and Pulcher have multiple local riverways within view. These may have served as sentinel locations from which to observe river traffic, sending a clear message to travelers of the grandeur and status of Cahokia. No person traveling on these waterways would have escaped the gaze from atop these mounds, nor would they be able to deny the power and importance of Cahokia projected from such a concentration of impressive mound temples.