Integrated Shallow Geophysical Surveys at Two Caddo Period Archaeological Sites within the Limits of a Water Reservoir in Northeastern Texas, USA

Abstract

The newly constructed Bois d’Arc Lake Reservoir in Fannin County, Texas, USA, inevitably flooded a large ground surface area (67.34 km²) when the reservoir began impounding water in April 2021. Inside this (now) flooded area, land-based archaeological data recovery investigations discovered and documented several archaeological sites, now registered in the state of Texas; though, only two neighboring sites, namely, 41FN178 and 41FN244, are examined here. The first phase of archaeological testing at these sites included shovel testing, test unit excavations, and geoarchaeological trenching that yielded archaeological artifacts suggesting that Middle Caddo Indian peoples (AD 1200–1400) might have occupied this landscape. As the sites were recognized before the reservoir’s impoundment phase, this merited a non-invasive, non-destructive, high-resolution near-surface geophysical study to map strategic areas within sites 41FN178 and 41FN244 that might yield potential shallow targets of archaeological context. The adopted geophysical survey comprised 3D direct current electrical resistivity imaging (ERI) and land horizontal magnetic gradiometry (HMG), each mapping a total surface area of 2133 and 15,640 m², respectively. The combination of 3D ERI and land HMG surveys was instrumental in rapidly mapping the horizontal and vertical extent of shallowly buried anomalies within a large area prior to the completion of the dam and the beginning of water impoundment. Based on the geophysical insights, the outline of several Caddo houses with functional internal and external features (e.g., burnt cooking surfaces, storage pits, refuse pits, fired soil, ditches, a dump site, and a compound fence) are thought to exist within the uppermost 2 m of the Quaternary stratigraphy at both sites. At site 41FN244, 3D ERI found numerous resistive anomalies surrounding a conductive anomaly, collectively interpreted as a group of post-holes surrounding the remains of a Caddo house’s inner clay floor. It also found a cluster of several resistive anomalies interpreted as midden or middens. The HMG survey carried across areas from which archaeological test units also yielded positive findings, at sites 41FN178 and 41FN244, identified numerous scattered monopolar and dipolar anomalies interpreted as post-molds of Caddo houses, compound enclosures or fences, and adjacent middens. Archaeological excavations guided by the geophysical results yielded significant cultural material and post-mold features at site 244, which validate the geophysical interpretation in a preliminary context. Additionally, several dispersed magnetic anomalies are thought to be shallowly buried hearths, burn cooking surfaces, storage pits, and ditches. The mapped magnetic anomalies agree with the location and distribution of previously found archaeological artifacts and the extent of resistive and conductive resistivity anomalies. Follow-up archaeological excavations of these geophysical anomalies have preliminarily confirmed interpretations.

1. Introduction

The cultural and archaeological tradition commonly recognized as Caddo or southern Caddoan-language speakers flourished around AD 800–900 in Northeast Texas, Southeast Oklahoma, Southwest Arkansas, and Northwest Louisiana (Schultz [1]). The prehistoric Caddo culture periods thrived between ca. AD 900 to AD 1680 [Early (AD 900–1200), Middle (AD 1200–1400), and Late (AD 1400–1680)], passing into its historical period from AD 1680 to 1860, following Story [2]. The early European accounts from the 17th century and the archaeological record indicate that there was a wide range in size, shape, form, and use of architectural space in the Caddo-built environment of Eastern Texas (Schultz [1]). Caddo people lived in dispersed but sedentary communal settlements situated along significant rivers and minor streams and tributaries; hence, they closely interacted with the environment and landscape. These settlements ranged from isolated farmsteads and small hamlets to villages and large community centers (see Figure 1). The Caddo-built environment has been identified through archaeological excavations (Schultz [1] and references therein) and geophysical prospecting (e.g., Schultz [1]; Perttula et al. [3]; Walker and McKinnon [4]). The Caddo’s architecture was a sustainable, true ecosystem because they collected natural raw materials (e.g., dirt, rocks, clay, timber, and grass), used them in their natural state, and transformed them into functional structures for shelter and habitable purposes. Caddo architecture includes structures that range in shape from round or elliptical to rectangular or square. These structures range from small features, covering an area of less than 2 m², that likely represent the footprints of granaries or storage platforms, to large structures covering over 200 m² that are linked to earthen mounds (Schultz [1]). According to Schultz [1], physical evidence for these wooden pole grass and cane-covered structures includes post-holes or post-molds, wall trenches, wattle-impressed daub fragments, hearths, pits, the occasional prepared floor, and the occasional well-preserved burned superstructure of a building that includes structural elements such as charred timbers. Buildings have a variety of structural attributes, including partitions and extended entranceways, and may be found isolated or associated with plazas or earthen mounds.

The construction of large, engineered water dams in the United States of America triggers archaeological data recovery investigations in the framework of cultural resource management projects. The recent completion (August 2021) of the Bois d’Arc Lake Reservoir in northeastern Texas, USA, by the North Texas Municipal Water District inundated 67.34 km² of Bois d’Arc Creek’s floodplain and Quaternary terraces. Designed to treat and deliver up to ~2.65 × 10⁸ L of water daily for 2 million people in the Dallas Metropolitan area, the lake reservoir began to impound water in April 2021. In the inundated surface area, archaeological investigations took place at several registered archaeological sites scattered within the limits of the dam. Subject to this investigation are archaeological sites 41FN178 and 41FN244 (see Figure 2). Shovel testing and test unit excavations at sites 41FN178 and 41FN244 yielded positive results. The artifact assemblage from the shovel testings consisted of ceramic sherds, debitage, burnt clay, daub, and fire-cracked rock; hence, they became targets for near-surface geophysical testing. The objective of this shallow geophysical investigation was to locate and characterize any anomalies of possible archaeological context within specific areas at sites 41FN178 and 41FN244. Three-dimensional (3D) electrical resistivity imaging (ERI) and land horizontal magnetic gradiometry (HMG) were the geophysical methods of choice. Their results successfully guided archaeological excavations, which uncovered several activity areas, particularly in site 41FN244.

2. Geological and Geomorphological Setting

The topography of the study area is relatively flat. The ground surface has been incised by the Red River and its tributaries, including the southwest-northeast-trending Bois d’Arc Creek. The stratigraphy of the area comprises Upper Cretaceous bedrock unconformably overlain by Quaternary clastic deposits (Shelby et al. [5]). From bottom to top, the stratigraphy consists of the Eagle Ford Formation, the Bonham Formation, the Blossom Sand, the Brownstown marl, Quaternary fluviatile terraces (Qt4 and Qt2) of varying textures, and Quaternary alluvium (Shelby et al. [5]). Figure 3 shows a schematic stratigraphic model of the area, and the geological map is shown in Figure 4.

In the area, the Eagle Ford Formation is approximately 107 m thick. It is comprised of medium-to-dark-gray shale, with occasional occurrences of bituminous, selenitic, and calcareous concretions and marine megafossils. The thickness of the overlying greenish-gray Bonham Formation varies from 114 to 160 m, is composed of marl and clay, and has marine megafossils. The westward-thinning Blossom Sand pinches from 75 to 6 m. It is mainly brown but weathers brown and red and is characterized by quartz sand, calcareous, glauconitic, ferrigenous, and thin clay interbeds. The Brownstown marl is the uppermost level of the local bedrock formation. Marl, clay, and lesser sand make up this unit. It is glauconitic at the base, phosphatic, and has some calcareous septaria. The Brownstown Marl is dark grey, and its weather is yellowish gray with some marine megafossils. The Quaternary fluviatile terrace deposits are dated as Pleistocene (Shelby et al. [5]). Shelby et al. [5] subdivide them into five sub-units (Qt1 to Qt5), but only two occur within the area under investigation (e.g., Qt2 and Qt4). The Qt4 terrace is a ~10 m thick fining-upward clastic sequence made of gravel, sand, and silt. It contains freshwater and terrestrial molluscan faunas and is generally dissected with exposed bedrock at the edges. The Qt2 terrace is a ~10 m thick, fining-upwards clastic sequence with a well-sorted, cross-bedded basal gravel passing into well-bedded sand, silt, and occasional silty clay. The sequence is mainly red to reddish tan, not extensively dissected, with immature soils and freshwater and terrestrial molluscan faunas. The Quaternary alluvium is characterized by floodplain deposits deposited by the Red River drainage system.

Geoarchaeological investigations at sites 41FN176, 41FN177, 41FN178, and 41FN244 were conducted at the proposed Bois d’Arc Lake reservoir by Cross Timbers Geoarchaeological Services coeval with the present 3D ERI survey (see Kibler, this report). The goal of these geoarchaeological investigations was to document the soil stratigraphy of the four sites, determine the age of the deposits hosting the archaeological materials, and understand their contextual integrity. The investigated sites are located on a late Quaternary alluvial terrace and its sloping or eroded margin. The terrace is mapped as Qt2 within the Lower Bois d’Arc Creek valley, following Shelby et al. [5]. The investigations were completed through the excavation and examination of 14 backhoe trenches with an average length of ~11 m and a mean depth of 0.70 m. In general, the trenches were placed on the top of the Qt2 terrace and its sloping margin, except for two trenches located on the distal margin of the Bois d’Arc Creek floodplain between sites 41FN176 and 41FN178. All trenches ended on or within deposits of late Pleistocene age, except for the two floodplain trenches that encountered the water table at <1 m below the ground surface. Detailed soil-stratigraphic descriptions for the uppermost 1.0 m relevant to ERI sites 244-area1, 244-area2, and 178-area3 are found in Kibler’s report. The consensus reached by the geoarchaeological investigations establishes that the Holocene to upper Pleistocene deposits consist of a complex mixture of clays, silts, and very fine to fine sands that host archaeological sites in the Holocene mantle. Given the textural nature of these deposits, a relatively low range of electrical resistivity values at the proposed 3D ERI sites is anticipated.

3. The Caddo Nation

The Caddo People, or Caddo Nation, have lived in the diverse ecoregions of the Arkansas and Red River basins in modern-day Arkansas, Louisiana, Oklahoma, Texas, and Missouri since time immemorial (Perttula [7,8]; Trubitt [9]). The Caddo People established in an environment of hardwoods, pines, and interspersed prairies that are distinct from tall-grass prairies to the north, west, and south and different from the floodplain environments of the Lower Mississippi Valley embayment (Perttula [7]). The Caddo People were a society of farmers, warriors, potters, priests, and traders. They were powerful and populous people linked by language, relationship, and beliefs organized into 25 bands of Caddoan-speaking peoples who aligned themselves to larger tribes like the Cadohadacho of the great bend of the Red River, the Natchitoches of the lower Red River Valley, and the Hasinai of the Neches and Angelina River Valleys (Perttula [8]; Carter [10]; Swanton [11]). As a centralized earthen mound society, the Caddo People are widely considered to be the most western great chiefdom of the American Southeast (Perttula [7,8]). They dominated a far-reaching interconnected system of trade and exchange between intertribal Caddo polities and neighboring Plains, southeastern, and southwestern cultural complexes (Perttula [8,12]; Schambach [13]; Story [14]). Due to European diseases and conflict, the population size of the Caddo was decimated during the Late Caddo period (A.D. 1400–1680) and the Historic Caddo period (A.D. 1680–1860+) (Carter [10]; Swanton, [11]; Perttula [12]).

The Caddo cultural tradition is archaeologically defined by the temporally diagnostic cultural materials that belong to a distinctive population of people who share geographic, economic, social, religious, and political systems (Perttula [8,12]). Caddo occupations are temporally distinguished by changes in the environment, the adoption and adaptation to new lithic and ceramic technologies, and the use and construction of households (Perttula [8,15]; Perttula and Rogers [16]). Archaeological investigations at the Hurricane Hill site (41HP106) and the Oak Hill Village site (41RK214) in Northeast Texas documented intensively occupied Early, Middle, and Late Caddo period occupations (Perttula [15]; Perttula and Rogers [16]). These sites supported large year-around communities comprised of households who subsisted on the cultivation of maize, squash, gourds, native seeds, nuts, and tubers, gathered aquatic resources, and hunted deer, bison, and bear (Perttula [8,15]; Swanton [11]). The style of architecture found at these two sites varies between large open-walled arbors, or ramadas, with slanted roofs; dome-shaped grass houses with or without extended entrances; granaries; and sub-rectangular, or rectangular buildings with clay-cemented walls and floors (Trubitt [9]; Perttula [12]; Perttula and Rogers [16]; Newkumet and Meredith [17]; Perttula [18]).

Settlement types in Northeast Texas are dispersed along the Red River and its major tributaries and include extractive camps, base camps, hamlets, farmstead villages, and civic mound centers (Perttula [8]; Trubitt [9]; Story [14]). Year-around Hamlet and farmstead-village settlements are defined by multiple discrete activity areas associated with household structures and extractive activities like cooking and storing food and the manufacture of tools and raw materials (Binford [19]; Kelly [20]; Vehik et al. [21]: Pg. 11; Perttula [8,15]). Archaeological structures are formed by the consistent repetition of residential activities within a bounded space over a long period (Perttula [8]; Kelly [20]). Residential houses can be identified by the presence of construction materials like post-molds, burned clay, and daub concentrations and habitation features like hearths, pit features, artifact clusters, residential middens, and sandy clay-packed floors (Perttula [8,15]; Trubitt [9]; Vehik et al. [21]).

Since 2001, a wealth of information about the ethnohistory and archaeology of the Caddo Nation has become publicly available through the Texas Beyond History (TBH) website (https://www.texasbeyondhistory.net/tejas/index.html (accessed on 11 May 2024)). Most of the timeline of the Caddo Nation is understood by early works, including those of Perttula [7], Parsons [22], Griffith [23], Gregory [24], Bolton [25], Sabo [26], and Smith [27].

Numerous archaeological excavation works have documented the architectural signature of Caddo houses and the built environment. To name a few, they include the works of Schultz ([1] and references therein), Webb [28], Jelks and Tunnell [29], Spock [30], Rogers [31], Sabo [32], Story [33], and Taormina [34]. Several well-documented archaeological sites are scattered in valleys and tributaries of the Ouachita, Red, Sulphur, Sabine, Arkansas, and Neches rivers in what is today Northeast Texas, Northwest Louisiana, Southwest Arkansas, and Southeast Oklahoma (Perttula [7]; https://www.texasbeyondhistory.net/tejas/index.html), as shown in Figure 5.

3.1. Timeline of the Caddo Nation

People from the Caddo Nation date back a few thousand years before the present. Story [2] proposed a chronological framework for the Caddon Nation, which has been refined through archaeological, historical, and ethnographic research throughout the decades. The TBH presents a current Caddo timeline, summarized in

Table 1. The Caddo Nation timeline (https://www.texasbeyondhistory.net/tejas/fundamentals/timeline.html, accessed on 11 May 2024).

Period		Date(s)
Pre-Columbian times	Late Archaic	2000 B.C. to 200 B.C.
	Woodland (Early Ceramic)	500 B.C. to A.D. 800
	Emerging Caddo	A.D. 800–1000
	Early Caddo	A.D. 1000–1200
	Middle Caddo	A.D. 1200–1400
	Late Caddo	A.D. 1400–1600
Colonial times	European Invasion	1542–1730
	European Colonization	1730–1800
Historical times	Anglo-American Conflict	1800–1859
	Louisiana Treaty of Cession	1835
	Brazos Reserve, Texas	1855–1859
	Civil War	1860–1867
	Resettled in Oklahoma	1868
	Caddo Tribe	1874
	Allotment	1889–1901
	Tribal Charter	1936
Modern times	NAGPRA (Native Americans Grave Protection and Repatriation Act) Enacted	1990

. Several centuries of the Caddo timeline pre-date the European invasion (pre-Columbian times). The Caddo Nation interacted for a few centuries with the colonists (Colonial times). Collectively, the Caddo-speaking peoples formed a society that early Spanish explorers highly regarded as civilized and friendly in comparison to many of their neighbors. This initial respect did not spare the Caddo from the common fate of so many of the native societies who came into contact with European diseases, guns, land dispossession, and desires. The Caddo Nation survived post-colonialism as its population declined. In historical times (between ~1800 and 1936), the Caddo population and territory continued to reduce while interacting with the Anglo-American settlers significantly. In less than two centuries, the Caddo Nation was reduced to a few hundred refugees who were assigned tiny parcels of land in the Oklahoma Indian Territories, several hundred kilometers northwest of their homeland. In modern times (after 1936), the Caddo Nation of Oklahoma has over 4000 members at its official tribal headquarters in Binger, Oklahoma. It was here, in and around the towns of Anadarko, Binger, and Fort Cobb, that the Caddo settled during and after the Civil War. This final relocation was preceded by over a century of turmoil during which Caddo groups were forced to give up their home territories in Northeast Texas, Northwest Louisiana, Southwest Arkansas, and Southeast Oklahoma. The cultural periods of the Caddo Nation are only briefly described as follows.

3.1.1. Pre-Columbian Times of the Caddo Nation

In the Late Archaic period (2000 B.C. to 200 C.C.), some of the Late Archaic groups in the Caddo Homeland were hunter–gatherers and may have begun small-scale gardening. The intensive harvesting of hardwood nuts, such as hickory and walnut, combined with deer hunting and a host of other food resources, apparently provided enough surplus food for people to begin staying longer in one place.

In the Woodland (Early Ceramic) period (500 B.C. to A.D. 800), Caddo ancestors gradually shifted from being mobile hunter–gatherers to increasingly settled villagers who planted domesticated crops to supplement wild foods, and they produced and stored surplus food. Long-distance trading of tools and foods with people from other regions began. Pottery making was introduced from the southeast, as well as the bow and arrow (probably from the southwest).

In the Emerging Caddo period (A.D. 800–1000), early Caddo society began to form as one of the earliest Mississippian cultures in the southeast. Among the many villages, some emerged as ritual centers where religious and political leaders lived. Temples and other special buildings sometimes stood on top of earthen mounds. Temple and burial mounds were sometimes arranged around open plazas, where the people gathered on solemn and festive occasions. Complex religious and social ideas took hold, including the notion that some people and certain lineages were more critical than others.

In the Early Caddo period (A.D. 1000–1200), Caddo society entered its prime, an era of unprecedented wealth, population, and prestige. They developed a distinct pottery tradition, produced outstanding pottery, and buried their dead. Major Caddo ritual centers in most parts of the Caddo Homeland, especially along the Red River, were the principal places of small, independent societies.

In the Middle Caddo period (A.D. 1200–1400), the Caddo people grew more numerous, and more and more villages, hamlets, and farmsteads were established throughout the Caddo area. Corn became the mainstay crop for most Caddo groups. People lived among their cornfields. The strategically located site of Spiro on the Arkansas River reached its peak as an essential trading and ritual center. Caddo potters experimented a great deal with different shapes and designs.

In the Late Caddo period (A.D. 1400–1600), the Caddo population peaked after A.D. 1400, with Caddo settlements built throughout the Caddo Homeland, including many places that had not been settled before. The ritual mound centers became less critical in some areas. The east–west trade brought small quantities of marine shells, turquoise, cotton, and southwestern pottery to the Caddo Homeland from as far west as the Pacific Ocean, as well as trade pieces from the Mississippi Valley.

3.1.2. Colonial Times of the Caddo Nation

In the European Invasion period (1542–1730), the first Europeans who set foot in the Caddo Homeland were Spaniards, members of the De Soto entrada in 1542. The Spaniards did not stay long, but they returned over a century later. In the intervening period, profound change began to reach the Caddo Nation: Old World diseases, plants (such as peaches and watermelons), animals (especially horses), and metal tools and weapons. In the late 1600s, the Spanish entered the region from the southwest and the French from the Mississippi Valley. They established missions and trading posts and competed with one another for control over the Caddo domain. Recurring diseases (like smallpox) continued to decimate Caddo populations. Rival Indian groups, now equipped with guns, encroached from the east.

In the European Colonization period (1730–1800), as Europeans and their descendants colonized North America, Caddo societies dealt with catastrophic changes caused by rapid population loss, incursions of enemies from the north and east, mounted raiders from the west, and a changing economy. Caddo groups became intermediaries and active partners in trade, especially with the French and French allies. Caddo groups formed alliances in an attempt to cope with massive population loss and threats from intruding enemies.

3.1.3. Historical Times of the Caddo Nation

In the Anglo-American Conflict period (1800–1859), the relentless push of Anglo-American settlers from the east forced the Caddo to abandon much of their homeland as their population plunged, with the remnant groups banding together for survival.

In the Louisiana Treaty of Cession period (1835), the Caddo groups gave up a million acres of their traditional lands in present-day Louisiana and Arkansas and moved westward into Texas in exchange for modest payments, only some of which were ever made. The forced exodus began a 20-year period during which the Caddo had no permanent home. Relentless Anglo settlement from the east pushed Caddo groups westward out of their homeland into North-Central Texas.

In the Brazos Reserve, Texas period (1855–1859), the new state of Texas set aside a small reserve on the Brazos River about 75 miles west of Fort Worth for the dwindling groups of Caddo, Wichita, and other tribes. Hostile settlers in the area soon forced the Caddo to flee to the Indian Territory (today’s Oklahoma), where they were to be given land with the Wichita.

During the Civil War period (1860–1867), distrustful of Southerners, most of the surviving Caddos moved to Kansas during the war. Some stayed in Oklahoma.

During the Resettled in Oklahoma period (1868), the Caddo people returned to Indian Territory to find that most of their lands had been given to Plains Indian groups. The Caddo finally settled down on the remaining land near the towns of Binger, Fort Sill, and Anadarko in the state of Oklahoma.

In the 50 years following the Civil War, the Caddo learned to live in West-Central Oklahoma, often intermarried with members of other tribes, slowly increased their numbers, and struggled to cope with assimilation into American society.

In the Caddo Tribe period (1874), for the first time, the Caddo were recognized as a single tribe or nation, a change brought about by the necessity of dealing with the United States government.

During the Allotment period (1889–1901), on the order of the United States government, Caddo tribal lands, like those of certain other tribes, were parceled out to each adult Caddo, 160 acres each, while white settlers were given everything left over (most of the Caddo land). This action was a deliberate strategy intended to seize more Indian lands, prevent tribal reorganizations, and force Indian peoples to assimilate into American society.

In the Tribal Charter period (1936), the Caddo Nation adopted a Tribal Charter and set up a formal government with an elected chairman and tribal council. Against all odds, the Caddo survived their nineteenth-century low point and today number over 4000. The members of the Caddo Nation of Oklahoma do not live on a reservation, and many of them do not even live in Oklahoma; they live in houses and apartments in America’s cities and rural areas. Like many Americans, Caddo people strive to maintain and revive their cultural traditions amid the hassled pace and changes of modern life.

3.1.4. Modern Times of the Caddo Nation

During the NAGPRA Enacted (1990) period, the enactment of the Native Americans Grave Protection and Repatriation Act (NAGPRA) gave the Caddo Nation a greater voice in their cultural patrimony and deciding the fate of the bones, grave goods, and sacred items of Caddo ancestors found on federal and tribal lands or held in federally funded institutions.

3.2. Architectural Spaces of the Caddo Nation

For the Caddo, identity and community cohesiveness appear to have been encoded in architecture (Sabo [32]). Based on documentary, archaeological, and geophysical (magnetometry) data, Schultz [1] provides an excellent examination of the structuring of architectural space by Caddo groups living in eastern Texas. Caddo structures varied in size, shape, form, apparent use, and overall treatment, both temporally and geographically. The shapes of structures ranged from round or ovoid to square or sub-square with rounded or diagonal corners or rectangular and had different types of entrances. They had a variety of associated features and were built in varying types of locations for varying purposes. There were mound and non-mound Caddo structures throughout the Caddo realm.

Caddo structures included domiciles, granaries, public-gathering plazas, and special-purpose structures that may have been directly associated with religious, political, or ceremonial uses. These structures or buildings were consistently made with a rigid frame of wooden poles entirely covered with grass and cane (Figure 6) and occasionally plastered with clay mud, and some were open. Archaeological evidence for these wooden pole grass and cane-covered structures includes post-holes or post-molds, wall trenches, wattle-impressed daub fragments, hearths, pits, the occasional prepared floor, and the occasional well-preserved burned superstructure of a building that includes structural elements such as charred timbers. Buildings had a variety of structural attributes, including partitions and extended entranceways, and may be found isolated or associated with plazas or earthen mounds. These structures range from small circular features (covering an area of less than 2 m²) that likely represent the footprints of granaries or storage platforms to large structures covering over 200 m² that are linked with earthen mounds. The possible duration of Caddo structures was between 10 and 15 years, which provides an estimate for the duration of occupation at a given site of 30 to 80 years based on structure overlapping and reuse of space (Schultz [1]). Caddo structures were burnt at the end of their use (Webb [28]).

The prevailing structures among the Southern Caddo groups were grass lodges with wattle and daub walls and grass-thatched roofs. The special-purpose structures were further subdivided into subclasses based on size (unusually small structures, less than 2 m in diameter), earthen mound association (on mound platforms or pre-mound structures), and non-mound structures with distinctive architectural form or associated interior features (Spock [30]). Specialized structures included mortuaries, meeting halls, temples, elite residences, and other public buildings. Structures with extended entrances are often considered specialized structures and are not as common as those with non-extended entrances (Schultz [1]).

The Caddo’s domicile structures were not related to mounds and commonly lacked structural attributes such as partitions and extended entrances (Spock [30]). Non-mound-related circular structures were the signature architecture of the Caddo house, while square-to-rectangular structures were the anomaly in many instances, likely a special-purpose building or primarily associated with mounds (Schultz [1]). As Caddo structures were burned at their end of use, their footprint can be mapped by geophysical methods, primarily magnetometry and electrical resistivity. While magnetometry yields an anomaly map, electrical resistivity imaging provides in-depth information about the anomaly. Therefore, the combination of both techniques permits the three-dimensional investigation of the structure. For instance, Walker [35] addresses the use of archaeogeophysics as primary archaeological data and provides an excellent discussion of the structures recorded from the magnetometer surveys at the George C. Davis (41CE19) and Hill Farm (41BW169) sites. Fundamental architectural features can be mapped with magnetometric data, including shape, type of entranceway, and internal functional features like hearths and storage pits (Walker [35]).

4. Archaeological Context of the Site

Separated by an inter-site ground distance of nearly 85 m and about 500 m west of the southwest–northeast trending Bois d’Arc Creek, the surface area of sites 41FN244 and 41FN178 is 26,099 m² and 9746 m², respectively (see Figure 1). Both sites are generally shallow, low-density cultural deposits, primarily in terrace sediments of sandy loam over clay, with lower elevations extending into floodplain soils. However, fieldwork at both sites in 2019 identified spatial locations with high densities of Caddo cultural materials likely associated with intact, subsurface archaeological features. At 41FN178, shovel tests and systematic 1 × 1 m and 50 × 50 cm excavation units recovered materials sufficient to delineate three spatially separated activity areas (Activity Areas 1–3). Similar fieldwork at 41FN244 identified seven activity areas. Of those, Activity Areas 1, 2, and 3 appeared to comprise the site core, while the other four may be associated with a variety of activities expected at a village site (food processing, hide tanning, public use space, etc.). The overall artifact assemblage from each site suggests a relatively intensive Formative to Middle Caddo occupation. While it was unclear how this might be reflected in the archaeological features from 41FN178, given that most of the artifacts were associated with the sociocultural transitional Formative Caddo Period, the artifacts and features from 41FN244 were consistent with what one would expect from a Middle Caddo farmstead or hamlet site. The presence of maize and radiocarbon dates from 41FN244 lends further evidence to this interpretation. In Activity Areas 1, 3, and 7, the densest artifact concentrations and deepest units are associated with what appear to be mima mounds. The creation of these mounds is typically associated with bioturbation, specifically from small vertebrates such as gophers. Evidence of bioturbation in those units is clear. The creation of the mounds may impact the vertical distribution of artifacts within them. Despite this, the site maintains horizontal integrity within all the activity areas to varying degrees.

The artifact assemblage from the site is typical of what one would expect from a Middle Caddo farmstead or hamlet (Perttula [8]). Most of the ceramics from the site are grog-tempered, and though complete attribute analysis has yet to be completed, the presence of some decorated sherds and a pipe bowl fragment suggests the possibility that fine wares similar to those found at Hurricane Hill and other East Texas Middle Caddo sites are present at 41FN244 (Perttula [8,15,36]). The stone tool and macrobotanical assemblages are also reflective of a Middle Caddo occupation. Botanical analysis of two vertical columns has indicated the presence of corn at the site (Rutherford [37]). Of the 15 levels analyzed, it was present in 13; however, it was not found in high densities. Radiocarbon dates from the corn yielded a range from AD 1300–1450, which is expected as increased corn consumption began in Middle Caddo sites throughout East Texas (Perttula [8]). There is some question as to when East Texas Caddo peoples relied on corn as a staple subsistence crop (Perttula et al. [38]), but given the ubiquity of corn identified from testing at 41FN244, it seems clear it was used as a staple crop during site occupation. The artifact assemblage from 41FN178 primarily consisted of faunal remains, fire-cracked rocks, stone tools, and lithic debitage. The only analyzed botanical material came from Activity Area 1. The botanical assemblage was minimal but mostly comprised of burnt hickory nuts, and radiocarbon dates from that material ranged from AD 800–1000. Artifacts collected from Activity Area 3 more than doubled the ceramic assemblage for the site as a whole. This led to the hypothesis that there might be an Early to Middle Caddo household or village spatially and chronologically, separate from the activities occurring in Area 1.

Given the size of each site and the relatively even horizontal extent of the assemblage, it is possible that multiple phases of Caddo occupation occurred along the terrace. There is clearly an intensive, tightly dated Middle Caddo occupation in Activity Area 3 at 41FN244. This is less clear in Activity Areas 1, 4, and 6 from the same site, but the assemblages appear to be Early to Middle Caddo chronologically. Those Activity Areas may represent special-purpose areas associated with a contemporaneous occupation in Activity Area 3 or vice versa. These could include superstructures or plazas built for public gatherings. They may also represent earlier occupation episodes with rectangular, square, or circular household units (Walker and McKinnon [4]). Artifact assemblages from 41FN178 suggest a clear spatial and chronological distinction between Activity Areas 1 and 3, with the possibility of additional occupation in Activity Area 2, but the distinctions were not as clear in excavations as they were at 41FN244.

5. Geophysical Methods

Ground-penetrating radar (GPR), land magnetometry, and direct current (DC) electrical resistivity imaging (ERI, in America) or electrical resistivity tomography (ERT, in Europe) are considered the three most frequently used near-surface, non-destructive geophysical techniques in archaeology and cultural heritage (Batayneh [39]; Martorana et al. [40]), not necessarily in such order. The recent work of Martorana et al. [40] provides an in-depth discussion of these methods in archaeological applications. GPR is commonly done in 2D or 3D mode, and with a suitable antenna and a tight (~20 cm) in-line spacing can yield a very high resolution. Land magnetometry can be done in standard mode using a fixed base station and a rover or mobile station or in gradiometer mode, where two or more magnetic sensors are positioned side-by-side (horizontal mode) or one sensor above the other (vertical mode), and the closer the line spacing, the higher the resolution. ERI can be done in 2D, 3D, or even time-lapse mode. One-dimensional (1D) electrical resistivity is instead a vertical-sounding method and does not produce 2D or 3D images, so it is not appropriate for archaeological prospections. For this survey, we used 3D ERI and land magnetic horizontal gradiometry (HMG), a combination of methods effectively previously used in archaeology (i.e., Batayneh et al. [41]; Young and Droege [42]; Drahor et al. [43]; Al-Saadi et al. [44]).

Two-dimensional/three-dimensional ERI is extensively utilized in archaeological research, environmental and engineering investigations, hydrogeological studies, and mineral exploration campaigns (Reynolds [45]; Loke et al. [46]). The method has proven successful in both archaeological and geoarchaeological (Ortega et al. [47]; Teixidó et al. [48]; Papadopoulos et al. [49]) research worldwide because of its non-destructive, cost-effective, and rapid ability to image subsurface human-made structures of archaeological contexts that are shallowly buried in alluvial deposits (Loke et al. [46]; Griffiths and Barker [50];). In particular, the method can determine the depth and geometry of structures in the vertical section (Griffiths and Barker [50]) with a certain degree of resolution. As the non-invasive electrical resistivity method can save time, costs, and effort in archaeological prospection and yield detailed images of the subsurface anomalies, the method was selected for this project. Satisfactory results are obtained with 2D resistivity surveys (Batayneh et al. [41]; Griffiths and Barker [50]; Noel and Xu [51]; Papadopoulos et al. [52]; Tsokas et al. [53]); however, the best results are obtained using a 3D survey (e.g., Al-Saadi et al. [44]; Leucci et al. [54]; Berge and Drahor [55,56]; Wake et al. [57]; Getaneh et al. [58]; Moník et al. [59]).

Archaeological surveys typically use magnetic measurements for detecting burnt objects (fireplaces, kilns, and fired clays), burnt ‘cooking’ surfaces, remains of room structures, post-holes, storage units, irrigation canals, agricultural fields, and other massive human-made objects at shallow depths and for the planning of subsequent archaeological excavations (Sharma [60]; Schmidt [61]; Kaya et al. [62]; Urban et al. [63]; Skrame et al. [64]; Cajigas [65]). Under proper soil conditions, archaeological features such as post-holes, storage pits, and fire hearths vary from the surrounding soil matrix and can be detected by distinctive spatial signatures in geophysical data sets from archaeological sites (Perttula et al. [3]; Schmidt [61]; Kaya et al. [62]). Several cultural processes can induce change in the magnetic variation of the archaeological soil record (Kvamme [66]; Rego and Cegielski [67]). These processes include human-induced firing, construction with fired materials (e.g., bricks), human exacerbation of the magnetic enrichment of topsoil and the accumulation of topsoil due to human construction (e.g., mounds and earthworks), human constructions, removing topsoil (e.g., ditches, pits, etc.), human-imported stone, some of which may be more magnetic than local stones and or the surrounding soil material, and, finally, iron artifacts made by humans.

Magnetic gradiometers, either with a pair of proton-precession sensors or cesium-vapor sensors, are particularly sensitive to the near-surface and detect shallow targets more effectively and faster than single-sensor magnetometers; hence, they are attractive in archaeological investigations (Mussett and Khan [68]). The magnetic gradiometry method has proven successful in archaeological studies within urban (Kaya et al. [62]; Sauck et al. [69]; Abdallatif et al. [70]; Drahor et al. [71]) and rural sites (Perttula et al. [3]; Drahor et al. [43]; Kaya et al. [62]; Urban et al. [63]; Skrame et al. [64]; Cajigas [65]; Rego and Cegielski [67]; Rabbel et al. [72]; Godio and Piro [73]; Vafidis et al. [74]; Seeliger et al. [75]) because of the following features, which make it ideal for this project: (a) non-destructive, non-invasive, and cost-effective character; (b) high sensitivity to absolute field changes of less than one nT; (c) higher tolerance to high-gradients caused by shallowly buried targets; (d) rapid ability to map shallowly buried objects and structures of archaeological context; (e) provides valuable results that assist excavation strategies and complement input from other geophysical methods. In Caddo archaeology, the magnetic method has been widely used to map the remains of the footprints of architectural structures (Schultz [1]; Walker and McKinnon [4]).

5.1. 3D Electrical Resistivity Imaging (ERI)

5.1.1. 3D ERI Data Acquisition

Due to the large surface area of archaeological sites 41FN178 and 41FN244, only the strategic areas of the sites were selected for targeted 3D ERI measurements. True-3D and pseudo-3D ERI measurements were performed in three grids, here referred to as zones 178-area3, 244-area1, and 244-area2, within sites 41FN178 and 41FN244, respectively, as shown in Figure 7. True-3D ERI measurements consist of deploying the multi-electrode cable in a snake-like pattern while conserving a 3D geometry (square or rectangular). In this case, transmitting electrodes and receiving electrodes can be in different electrode strings, so there is no interpolation between the electrodes’ strings, and all the raw data are stored in a single file appropriate for 3D inversion. The pseudo-3D ERI measurement consists of deploying multiple parallel transects of a multi-electrode cable. The measurement along each transect is acquired in a profile mode (i.e., current electrodes and potential electrodes are always distributed along the same cable transect), and, subsequently, all the acquired transects are merged into a merged 3D spread suitable for 3D inversion. Figure 8 illustrates the deployed 3D electrode grid at each site. Each ERI grid had a unique geometry, size, and terrain conditions, yielding a unique 3D electrode layout and a range of horizontal and vertical resolutions. The surveyed areas for sites 178-area3, 244-area1, and 244-area2 were 796.25 m² (17.5 × 45.5 m), 507 m² (19.5 × 26 m), and 830 m² (41.5 × 20 m), respectively.

The ERI data were measured with the SuperSting R8/IP/SP Resistivity meter with a 56-electrode capability switchbox by AGI (Advanced Geosciences, Inc., Austin, TX, USA). The measurement settings were 2 Amps current, 1.2 s measure time, a 2 to 10% maximum error, and two cycles. True-3D ERI data were acquired in resistivity mode using a mixed electrode array comprised of radial dipole–dipole, and a radial gradient array was used at zones 244-area1 and 178-area3. Due to time constraints, pseudo-3D ERI measurements were done at zone 244-area2 with the combination of eleven (11) 2D parallel profiles measured with the strong-gradient electrode array.

Table 2. Summary of the 3D ERI electrode grids.

Criteria	Zone 178-Area3	Zone 244-Area1	Zone 244-Area2
Electrode layout	True-3D	True-3D	Pseudo-3D
Electrode array	Mixed dipole gradient	Mixed dipole gradient	Strong gradient with 75% transmitting electrode overlap
Electrode capability	56	56	56
Maximum number of electrodes	84	196	336
Electrode spacing (m)	3.5	1.5	0.75
In-line spacing (m)	3.5	2.0	2.0
Number of electrodes along the x/y axis	4/14	4/14	14/0
Number of roll-along or slide-along used	1 slide-along	5 roll-along	No
Terrain correction	Yes	Not needed	Yes
Horizontal (x/y) resolution (m)	1.750/1.75	1.000/0.75	0.375/1.0
Vertical (z) resolution (m)	1.75	1.25	1.068
Observations:	They were completed in two full work days due to long measurement times for the roll-along sections.	Roll-along failed because the instrument overheated, so slide-along was implemented.	Eleven parallel 2D resistivity profiles were merged into a 3D grid, all acquired with the same criteria.

summarizes the details of each 3D electrode grid’s ERI survey layout. The geographic coordinates and elevation of the electrodes were measured with a global positioning system (GPS). Figure 9A shows the field setup configuration of the resistivity meter.

The archaeological sites are located immediately west of the Bois d’Arc Creek and directly south of the Coffee Mill Lake. Bluestem prairie, Blackland prairie, and Oak–Hickory forest vegetation communities dominate the surveyed area. The terrain conditions were generally flat within the archaeological site boundaries. However, some areas had a gentle gradient associated with the geomorphological nature of the Quaternary terrace deposits. At the sub-meter scale, localized topographic rises can be detected within the 3D ERI electrode grids.

5.1.2. 3D ERI Data Inversion

Topographic corrections were applied in the 3D electrical resistivity inversion to raw data files collected at zones 178-area3 and 244-area2 because of noticeable cm-scale elevation differences within the electrode grids. Zone 244-area1 did not require topographic corrections due to a flat ground surface. The 3D ERI data processing included merging the electrode array raw data files into a single raw data file and its corresponding terrain file using the Earthlmager3D^® software (version 1.5.3) by AGI. Both resistivity 3D models’ inversion-stopping criteria were based on a root-mean-square (RMS) percent error of 3%. The smooth inversion algorithm was used. For the forward modeling method, the finite element method was chosen for zones 178-area3 and 244-area1, and the finite difference method for site 244-area2, all with two mesh divisions between electrode pairs.

5.2. Magnetic Gradiometry

5.2.1. Basic Concepts

Relevant to archaeological prospection, three types of magnetometers can be used in archaeogeophysical surveys: proton-precession magnetometers, alkali-vapor magnetometers, and fluxgate magnetometers (Kvamme [66]). Martorana et al. [40] provide a comparison between the characteristics of these magnetometers in terms of sensitivity, pros and cons, possible applications, and the degree of suitability for their use in archaeological or cultural heritage applications. The fluxgate and the proton-precession magnetometers are the typical choices in most archaeomagnetic surveys, leaving the alkali-vapor magnetometer as the infrequent choice probably because they are subject to heading errors (Martorana et al. [40]), an aspect that is fixed during the data-processing workflow.

Fluxgate magnetometers measure the Earth’s magnetic field in the x–y–z directions and their magnitude (a vector measurement). Proton-precession magnetometers and alkali-vapor magnetometers only measure the magnitude or intensity of the Earth’s total magnetic field (a scalar measurement). However, in surveys where greater precision is required, like in archaeology, high-sensitivity alkali-vapor magnetometers, either in single-sensor mode (a base station and a rover station) or dual-sensor (gradiometer) mode, are preferred (Sharma [60]; Mussett and Khan [68]) over fluxgate and proton-precession magnetometers.

The alkali-vapor magnetometers operate on the principle of optical pumping, a development in radiofrequency spectroscopy based on irradiating alkali metal (e.g., cesium or rubidium) vapor atoms with spectral laser beams of appropriate frequency (Sharma [60]; Milsom and Eriksen [76]). The effects of electrical noise and high field gradients are less severe than with proton-precession magnetometers, and measuring times are very short. The alkali-vapor magnetometers continuously measure the intensity of the total magnetic field, and the orientation of the instrument is not critical. The precision and sampling rate of the device is much higher (10 samples per second) than that of proton-precession magnetometers. As the output frequency (100–300 kHz) is determined with great accuracy, the high sensitivity of the order of <0.01 nT can be obtained (Martorana et al. [40]; Geometrics [77]). The fast-sampling mode, automatic and quick data logging, and high sensitivity make alkali-vapor magnetometers ideally suited for use as part of a land or airborne magnetic gradiometry survey.

Single-sensor magnetometer surveys are limited in that they measure the total magnetic field, including effects from archaeological materials, the underlying geology, and the diurnal fluctuations of the Earth’s magnetic field (Cajigas [65]). Conversely, measurements of the vertical and horizontal magnetic gradients have significant advantages over total magnetic field surveys with single-sensor magnetometry. These include the reductions in regional magnetic gradients, the elimination of short-term diurnal fluctuations, and the attenuation of longer wavelength anomalies due to deeper magnetic sources; thus, smaller-scale shallow anomalies are enhanced by removing the masking effects of longer wavelength anomalies (Young and Droege [42]; Marcotte et al. [78]), improving the resolution of minor anomalies, and resolving complex anomalies (Cowan et al. [79]). Another advantage of using a gradiometer is that the transient signals of the Earth’s magnetic field are the same for the two sensors, so it completely cancels out any differences in signals that are unaffected by diurnal variations (Sharma [60]; Mussett and Khan [68]; Milsom and Eriksen [76]; Cowan et al. [79]). For this reason, there is no need to correct for diurnal variations, which requires a fixed base station and a rover station (i.e., single-sensor magnetometer). Magnetic (vertical or horizontal) gradient anomalies are always shaper and narrower than the associated total magnetic field anomaly. In addition, it is possible to reconstruct the total field from measured gradient data, leading to improved leveling of the total field data (Cowan et al. [79]).

Absolute field changes of less than one (1) nT and higher tolerance of high gradients are significant in archaeology, and sensitivity is always crucial in gradiometry (Mussett and Khan [68]). A gradiometer operates with two sensors to record the differences in the strength of the Earth’s total magnetic field (Sharma [60]; Mussett and Khan [68]; Milsom and Eriksen [76]). Magnetic gradient readings, either horizontal or vertical, are usually taken over a grid formed by closely spaced traverses, typically 2.0 to 0.25 m (Abdallatif et al. [70]). Naturally, the tighter the line spacing, the higher the lateral resolution. Magnetic gradiometry detects soil disturbances near the subsurface, typically to a depth of two meters (Rego and Cegielski [67]). The sensors in a gradiometer are fixed a distance apart; either one is next to the other or one above the other. If positioned side-by-side at the same altitude, the horizontal magnetic gradient is measured. If one is directly above the other, the vertical magnetic gradient is measured. In either case, the result is given either as the difference in nT or as the gradient of the magnetic field in nT/m, which is the difference between the two readings measured by the sensors divided by the distance between the sensors. The vertical distance (d) between the gradiometer and the buried target is a crucial factor in detecting magnetic anomalies, and detection ability is inversely proportional (decreases) with the cube (d³) of its distance (Drahor et al. [43]; Kvamme [66]). Additionally, the closer an object is to the ground surface, the greater the difference between the magnetic fields it produces at the two sensors, and so the greater the magnetic gradient, whereas deep objects or bodies produce only a small gradient. Hence, in magnetic gradiometry surveys, the closer the sensors are to the ground surface, the stronger the signals are. The horizontal gradient measurements require the two sensors to be at the same altitude and, hence, at the same distance from the ground, something that the vertical gradient does not provide. Since archaeological targets are shallow and the anomalies are minor, often a few nT, a gradiometer has the advantage over a single-sensor magnetometer.

5.2.2. Justification of the Land Horizontal Magnetic Gradiometry (HMG) Method

Following Kvamme [66] and Rego and Cegielski [67], several cultural processes can induce change in the magnetic variation of the archaeological record. These processes include human-induced firing, construction with fired materials (e.g., bricks, clay, and stones), human exacerbation of the magnetic enrichment of topsoil and the accumulation of topsoil due to human construction (e.g., mounds and earthworks), human constructions, removing topsoil (e.g., ditches, pits, trenches, etc.), human imported stone, some of which may be more magnetic than local stones and or the surrounding soil material, and, finally, iron artifacts made by humans.

Land magnetic surveying is a passive, non-invasive method based on the measurement of localized perturbations to the Earth’s magnetic field caused by the presence of buried targets. Gradiometers, either with a pair of proton-precession sensors or alkali-vapor sensors, are notably more sensitive than fluxgate magnetometers (Martarona et al. [40]), are sensitive to the near-surface anomalies and detect shallow targets more effectively and faster than single-sensor magnetometers; hence, they are attractive in archaeological surveys (Kaya et al. [62]; Skrame et al. [64]; Mussett and Khan [68]).

In general, both the magnetic vertical and horizontal gradiometry methods have proven successful in archaeological studies within urban (Sauck et al. [69]; Kaya et al. [62]; Abdallatif et al. [70]; Drahor et al. [71]) and rural sites (Perttula et al. [3]; Drahor et al. [43]; Urban et al. [63]; Cajigas [65]; Rego and Cegielski [67]; Rabbel et al. [72]; Godio and Piro [73]; Vafidis et al. [74]; Mekkawi et al. [80]; Seeliger et al. [75]) because of the following reasons: (1) it is non-destructive, non-invasive, and cost-effective; (2) grants high sensitivity to absolute field changes of less than 1.0 nT; (3) offers higher tolerance to high gradients caused by shallowly buried targets that might be younger than the archaeological ones; (4) provides a rapid ability to map shallowly buried objects and structures of archaeological context with only one person; and (5) affords valuable and fast results that assist excavation strategies and complement input from other geophysical methods.

In particular, the horizontal magnetic gradiometry method was chosen over the vertical magnetic gradiometry method because of the following reasons:

• Estimates the location of abrupt lateral changes in magnetization or mass density (Kaya et al. [62]) because the measurements are taken across a ~1.0 m wide swath, defined by the alignment of two sensors at the same altitude, perpendicular to the survey direction (see insert in Figure 9B). Conversely, vertical magnetic gradient measurements do not offer this sampling feature because they do not sample across a swath perpendicular to the survey direction; instead, both sensors sample over the same coordinate along the same survey line at different altitudes, requiring closer in-line spacings.
• Provide additional information on gradients between survey lines, providing several advantages, including reducing the line dependency of anomalies, improving the resolution of features sub-parallel to the survey line direction, and reducing the aliasing involved when the survey line spacing to source depth rations is not optimum, which, in turn, allows for fine-tuning filter parameters, resulting in sharper, clearer images, especially for high-resolution texture filters (Marcotte et al. [78]; Cowan et al. [79]).
• Improves the leveling of total magnetic intensity data, and the need for tie-lines is optional (Cowan et al. [79]).
• Can be used to develop advanced map products, including (1) to derive a total field that is free of diurnal effects; (2) to derive the magnetic gradient tensor; (3) to estimate the depth of magnetic sources; (4) to derive the analytic signal amplitude, which emphasizes the source edges effects, reduces the interference effects of the anomalies, and yields an enhanced image of the anomaly boundaries (Bournas and Baker [81]); (5) to generate input for the automatic interpretation of horizontal magnetic gradient schemes; (6) to enhance the total field grids generated from profile data; (7) to ensure that isolated features are imaged and placed closer to their correct locations rather than close to the traverse lines; and (8) to indicate which side of the survey lines interline features will peak despite the fact that their location is not necessarily that accurate. Such advantages provide more interpretable products and attributes stemming from quantitative computations, which are possible only by measuring the horizontal magnetic gradient (O’Connell et al. [82]).
• Can be used to detect magnetic source boundaries in high resolution using the magnetic horizontal gradient operator, which emphasizes the source effects, reduces the interference effects of the anomalies, and yields an enhanced image of the source boundary locations; the locations are more precisely determined compared to those obtained with a vertical magnetic gradient (Skrame et al. [64]).

5.2.3. Land HMG Data Acquisition

The land HMG data were collected with a Geometrics G-858GAP alkali (cesium)-vapor gradiometer, with a resolution of 0.01 nT at a sampling rate of 10 readings per second. The two cesium-vapor sensors remained fixed (vertically upwards) during the entire survey. The horizontal sensor separation was kept fixed at 0.90 m. The orientation of the sensors was kept the same during the entire survey; however, the orientation of alkali-vapor magnetometers is not critical. The horizontal distance from the GPS antenna to the midway distance between the sensors was 1.2 m. The vertical distance from the sensor staff to the GPS antenna was 1.6 m. This geometry was input for the GPS offset correction during the magnetic data processing. During surveying, the distance between the ground surface and the sensors was ~0.60 m. The center of the hand-held counterbalanced forward staff was carried along the survey line, allowing for each sensor to extend 45 cm on either side. Figure 9B illustrates the field setup configuration of the gradiometer. Both archaeological sites were typically subdivided into 30 m × 30 m search blocks and surveyed with 2.0 m in-line spacings. The measurements were done bi-directionally along north–south trending survey lines. The terrain conditions were generally smooth within the search blocks. However, some areas have a gentle gradient associated with the Quaternary terrace deposits.

5.2.4. Land HMG Data Processing

For each magnetic gradiometry data set, the positions of both start and endpoints were verified, GPS offset corrections were applied, data points were migrated into UTM coordinates (Universal Transverse Mercator, NAD83 Zone 15N), and both filtering (the removal of spikes, dropouts) and de-stripping were applied. The filtered and de-stripped data were exported into Surfer^® software (version 10.7.972) for final gridding, gradient computation, visualization, and interpretation. Although both minimum curvature and kriging gridding methods are widely used in the Earth sciences, the minimum curvature method was chosen for gridding the magnetic gradient data sets (Lloyd and Atkinson [83]; O’Connell et al. [82]; Reford et al. [84]; Wang and Qiu [85]), which contain a large point population (N > 1000 observations). Conversely, kriging works well with small data sets (<250 points). After testing kriging, it was noticed that kriging yields jagged anomaly edges. Contrarywise, minimum curvature generates smoother surfaces because it honors the extensive data set as much as possible. Minimum curvature gridding fits the data to a constrained minimum curvature surface (Smith and Wessel [86]). For this project, to compute the gradient in units of nT/m, the HMG was estimated as the difference between the measurements from sensor 2 and sensor 1 divided by their horizontal distance (0.90 m) and then gridded using the minimum curvature method at 0.10 m grid spacings (x and y directions) before contouring. Equation (1) shows the mathematical expression for the calculation of the horizontal magnetic gradient:

$$HMG={\displaystyle \frac{M_{{sensor}_2}-M_{{sensor}_1}}{d_{\overline{s2s1}}}}$$

(1)

where M_sensor2 and M_sensor1 are the measured magnetic fields by sensor 2 and sensor 1, respectively, and $d_{\overline{s2s1}}$ is the horizontal distance between both sensors.

6. Results and Interpretation

6.1. 3D ERI Results and Interpretation

In general, the natural site conditions at all sites presented favorable ground contact resistance levels (<1500 Ω). The measured contact resistance ranges for zones 244-area1, 244-area2, and 178-area3 are 250–1500 Ω, 250–1250 Ω, and ~120–600 Ω, respectively (Figure 10). The quality factor of all the inversion results is considered excellent because the achieved root-mean-square (RMS) percent error is ≤4%. The estimated surveyed depths for zones 244-area1, 244-area2, and 178-area3 were 4.5, 9.5, and 12.4 m, respectively.

Table 3. Summary of the 3D ERI inversion results for all surveyed zones.

Zone	Topographic Correction	RMS %	L2-Norm		Iteration Number	Surveyed Depth (m)
			Yes/No	Value
178-area3	Required	3.62	Yes	0.2	1	12.4
244-area1	Not required	2.28	No	0.2	1	4.5
244-area2	Required	3.54	Yes	0.5	5	9.5

summarizes the inversion results from each 3D ERI grid.

6.1.1. Zone 178-Area3

The 3D inverted electrical resistivity model for this zone, shown in Figure 11A, yields a narrow resistivity range between 6 and 15 Ω-m. Figure 11A shows that the higher resistivity values (9–15 Ω-m), in shades from green to red, grade abruptly into lower resistivity values (6–9 Ω-m). This sudden transition defines the juxtaposition of a high resistivity (resistive) anomaly against a low resistivity (conductive) anomaly. The resistive anomaly represents a deep and elongated resistive structure roughly ~7 × 30 m and up to ~7.5 m thick, has undulating borders, and has a central maximum of resistivity that decreases outwards. Fine-grained clastic materials like clay, silts, and mud are conductive materials with typical low resistivity values (1–100 Ω-m) (Gunn et al. [87]), including under oversaturated conditions. The 3D inverted electrical resistivity model is sliced horizontally in increments of 0.50 m for visualization and analysis purposes, as shown in Figure 11B. Figure 11A,B clearly show that the interpreted silty-clay lens dominates the resistivity volume until it shrinks and thins out somewhere between an elevation of 144 and 143 m (a.s.l.), which corresponds to a depth of 7.5 m. Conductive materials dominate below this depth.

Positive findings from shovel tests and test unit excavations at the archaeological site came from depths of <1 m; thus, this resistive anomaly is too deep to be considered an archaeological target. Together, the depth and shape of the resistive structure, its resistivity value relative to the surrounding materials, and the geologic setting in which it occurs suggest that it is likely a deep silty-clay lens within the mapped Quaternary fine-grained terrace deposits and characterized by low resistivity. This interpretation is consistent with surface geology (Shelby et al. [5]) and ground-truthing by shallow geoarchaeological trenching (Kibler [88]) performend in parallel with the 3D ERI survey.

6.1.2. Zone 244-Area1

The 3D resistivity inversion results reveal a constricted electrical resistivity range between 16 and 22 Ω-m for this zone. The resistivity range is subdivided into three groups, mainly low resistivity (16–17 Ω-m), intermedium resistivity (17–19 Ω-m), and high resistivity (19–22 Ω-m), which, in turn, translate into three types of electrical resistivity anomalies. The 3D inverted electrical resistivity model for this zone is shown in Figure 12A. It reveals various conductive anomalies (total of 11) in shades of blue and numerous resistive anomalies (total of 10) in shades of yellow to red. In the plan view, the resistive anomalies dot the ground surface as semi-concentric ring structures on the left side of the electrode grid. The concentric ring resistive structures have a central maximum of resistivity, and their dimensions range from ~2 m to as large as 4 m in diameter. Collectively, the spatial distribution of the resistive anomalies forms approximately a circular pattern that envelopes a large conductive anomaly of approximately 10 m in diameter (see Figure 11A). These smaller (~2 m), ring-shaped conductive anomalies also occur around the large conductive anomaly.

Several subsurface features support the idea of the remains of a Caddo house structure. Caddo houses in the Caddoan region were made with a wooden frame bent into a beehive shape and thatched with long prairie grass (TBH [89]). These were large buildings, sometimes more than 10–12 m tall and 8–10 m in diameter. Several interior wooden posts supported the grass-thatched roof. The house walls were made of closely spaced thin poles set within a foundation trench. Daub (dried mud) plaster sealed the walls. The extended entranceway helped keep heat from escaping in the winter. A layer typically made the interior floor of the Caddo houses of compacted fine-grained earth materials such as clay, silt, sand, or a mixture of the three (TBH [89]). For visualization and analysis purposes, the conductive and resistive anomalies are also displayed with horizontal depth slices every 0.25 m (Figure 12B). Figure 12B shows that the sizeable conductive anomaly and its outlying resistive anomalies become more roundish closer to the subsurface. Though some outlying resistive structures vanish somewhere between 1.75 and 2.50 m depth, the main conductive structure is still preserved beyond that depth.

The long electrical resistance measurements (Shortle and Smith [90]; Elliot et al. [91]) and electrical resistivity imaging (Hagrey [92]) of tree trunks suggest that high resistance and strong local resistivity anomalies are related to wood decomposition causing wet, rotting, or dry cavities. Fine-grained clastic materials like clay silts behave as conductive materials, showing typically low-resistivity values (1–100 Ω-m) (Gunn et al. [87]), including under oversaturated conditions. This interpretation is consistent with the local geology (Shelby et al. [5]) and ground-truthing by shallow geoarchaeological trenching. Under these criteria, both the major conductive anomaly and the surrounding resistive anomalies can be interpreted as having a possible archaeological context.

The following subsurface features detected with 3D electrical resistivity imaging support the idea of a Caddo house structure at this site: (a) recent archaeological shovel test and test unit excavations at the site yielded artifacts; (b) relatively large, concentric conductive anomaly (low resistivity) interpreted as the remains of an interior clay-floor of a Caddo house; (c) concentric, ring-shaped resistive anomalies roughly evenly spaced surround the conductive anomaly (e.g., the floor) and are interpreted as post-holes with decomposed wood, embedded in a less-resistive, fine-grained matrix. The post-holes held the wooden frame of a Caddo house; (d) both conductive and resistive anomalies project into the subsurface and are within a depth range of archaeological interest. Alternatively, the detected conductive and resistive anomalies could be geological in origin. Shallow backhoe trenching in the area found fine-grained materials in the uppermost layer of the terrace deposits, which is consistent with the geologic descriptions of the terrace deposits (Shelby et al. [5]). Therefore, the conductive and resistive anomalies could be explained by the possible presence of clay-rich and silty-sand-rich lenses, respectively, deposited in a fluvial environment.

6.1.3. Zone 244-Area2

Figure 13A,B show the final 3D inverted electrical resistivity model of this zone as a cube and in horizontal slices spaced every 0.50 m. The results reveal a wide electrical resistivity range between 5 and 148 Ω-m within the estimated surveyed depth of 9.5 m. The resistivity range is subdivided into three groups, mainly low resistivity (5–20 Ω-m), intermedium resistivity (20–60 Ω-m), and high resistivity (60–148 Ω-m), which, in turn, translate into three types of electrical resistivity anomalies. Low-resistivity and high-resistivity values are interpreted here as conductive and resistive anomalies, respectively. The uppermost 1.50 m contains the highest concentration of intermediate and high resistivity anomalies (in shades of yellow to red), and the remaining 8.0 m is strongly conductive material without anomalies.

The distribution of the resistive anomalies forms three clusters (see Figure 13A), all with a central resistivity maximum. The resistive anomalies in cluster 1 are few. The resistive anomalies in clusters 2 and 3 are numerous individual and closely spaced, comprising a large elongated and semi-concentric resistive anomaly roughly 16 × 12 m and 20 × 14 m, respectively (Figure 13A), whose thickness does not seem to exceed 2 m, as suggested by the horizontal slices in Figure 13B. Between 2.4 and 9.5 m depth, this zone is strongly dominated by conductive materials, here interpreted as clay, silt, and fine-grained sand or a mixture of the three, consistent with the geology (Shelby et al. [5]) and ground-truthing by the geoarchaeological trenching.

The results of 3D electrical resistivity imaging from archaeological sites worldwide (Al-Saadi et al. [44]; Papadopoulos et al. [52]; Tsokas et al. [53]; Leucci et al. [54]; Berge and Drahor [55,56]; Wake et al. [57]; Getaneh et al. [58]; Moník et al. [59]) consistently reveal that the human-made remains (e.g., remains of wall foundations, ceramic pots, etc.) of archaeological context buried in alluvium (typically conductive) yield strong local resistivity anomalies. Under these criteria, the large resistive anomaly (clusters 2 and 3) can be considered of archaeological interest. The following subsurface features detected with 3D electrical resistivity imaging support the idea of buried archaeological remains at this site: (a) recent archaeological shovel test and test unit excavations at the site yielded archaeological artifacts; (b) two relatively large, semi-concentric resistive structures (high resistivity) embedded in topsoil conditions project into a shallow depth range (<2.4 m); (c) the resistive anomalies correspond with topographic highs roughly 30 cm high in clusters 2 and 3, which can represent a midden or middens comprised of resistive materials such as lithic reduction debris (e.g., chert and quartzite), ceramic sherds, and bones. Alternatively, the detected conductive and resistive anomalies could have a geological nature. Shallow backhoe trenching also found fine-grained materials in the uppermost layer of the terrace deposits, which agrees with the geologic descriptions of the terrace deposits (Shelby et al. [5]). The resistive and conductive structures could be explained by the presence of silty-sand-rich and clay-rich lenses, respectively, within a fluvial setting.

6.2. Land HMG Results and Interpretation

The magnetic gradiometry survey identified a total of 507 monopolar magnetic anomalies and a total of 160 dipolar magnetic anomalies of potential archaeological interest within a total mapped area of 15,640 m². In order to analyze the magnetic anomalies, the magnetic gradient map of each search block is plotted with its magnetic gradient range; otherwise, multiple anomalies are masked if the same gradient is applied to all search blocks within the archaeological sites. Several of the magnetic anomalies are interpreted in an archaeological context. For instance, monopolar anomalies, either negative or positive, typically denote buried negative soil structures, including (storage) pits, ditches, embankments, walls, earth mounds, empty cavities, irrigation canals, and associated agricultural fields. Conversely, dipolar anomalies typically represent subsurface highly burned materials such as hearths, kilns, storage pits, bricks, burned cooking surfaces, burned clay, burned rocks, circular or linear structures, or metallic objects (Perttula et al. [3]; Schmidt [61]; Kvamme [66]; Urban et al. [63]; Rego and Cegielski [67]; Vafidis et al. [74]). It is assumed that the interpreted archaeological features lie immediately beneath the surface (i.e., <2.0 m depth) without exceeding the late Holocene–late Pleistocene boundary mapped by Kibler [88]. Schultz [1] studied the nature of the prehistoric and historic architectural space by Caddo groups living in eastern Texas through a detailed examination of documentary, archaeological, and geophysical (magnetic) data to understand better how Caddo cultural space was created, maintained, destroyed, and altered, as well as how this relates to the broader Caddo society. Most of the excavated Caddo structures have a concentric shape, whose diameters vary from a couple of meters to dozens of meters.

The entire magnetic anomaly data set is classified according to the amplitude and the size of the anomaly in units of nT/m and m, respectively. Amplitude is categorized as high, medium, and low according to the values > 10 nT/m, between 5 and 10 nT/m, and <5 nT/m, respectively. Size is classified as small or large if the anomaly is ≤1 m and >1 m, respectively. The combination of these two properties provides a magnetic anomaly class number ranging from 1 to 6 (see

Table 4. A horizontal magnetic gradient anomaly class was adopted for this study. The classification criteria are based only on the amplitude and size of the dipolar or monopolar magnetic anomalies. The amplitude refers to the maximum absolute value of the anomaly. The size refers to the maximum length of the anomaly.

Anomaly Class	Amplitude	Size
1	high amplitude (>10 nT/m)	large size (>1 m)
2	high amplitude (>10 nT/m)	small size (<1 m)
3	medium amplitude (5–10 nT/m)	large size (>1 m)
4	medium amplitude (5–10 nT/m)	small size (<1 m)
5	low amplitude (<5 nT/m)	large size (>1 m)
6	low amplitude (<5 nT/m)	small size (<1 m)

for details). The shape of the anomaly is not a diagnostic feature, so it is not part of the classification system. The magnetic data identify numerous distinctive spatial patterns interpreted as archaeological features and metallic objects and are further discussed below. Figure 14 shows the uninterpreted horizontal magnetic gradient map of zones 178-area1, 178-area2, and 178-area3, surveyed within archaeological site 41FN178. Zones 178-area1 and 178-area3 were surveyed in one 30 × 30 m search block, but zone 178-area2 was subdivided into four 30 × 30 m search blocks.

6.2.1. Zone 178-Area1

The observed gradient varies from −30 to +30 nT/m in zone 178-area1 (Figure 14). Seven monopolar and ten dipolar magnetic anomalies occur in the northern and western parts of this block, which are categorized as class 1, 5, and 6 geophysical anomalies. The distribution of some anomalies defines a geometrically concentric pattern, approximately 10 m in diameter, that encircles a strong dipolar anomaly in the center. The monopolar anomalies are weakly magnetized (−2.3 to +2.5 nT/m) in comparison to the adjacent dipolar anomalies, which reach magnetization levels of ±30 nT/m. Several archaeological artifacts have been recovered from test units within this area. Two geoarchaeological trenches (Kibler [88]) indicate that the first ~1 m of the terrace deposits is comprised of a 0.12 to 0.22 m thick, very friable fine-grained sandy loam of late Holocene age passing into late Pleistocene alluvium composed of ~0.20 m thick brown clay loam over a ~0.40 m thick clay to clay loam of varying tones.

Together with the present magnetic data, the observed concentric arrangement of monopolar and dipolar anomalies likely represents a circular feature consistent with the remains of several post-molds linked to a Caddo house. The strong dipolar anomalies immediately south of the concentric structure appear to be a projection roughly radiating to the south that may be an extended entranceway common in Caddo houses. The encircled strong dipole is interpreted as a burnt cooking surface developed on late Holocene’s fine-grained sandy loam, following the trench profile descriptions of Kibler [88]. One weakly magnetized monopolar anomaly next to the cooking surface is reminiscent of a storage pit. The magnetic anomalies situated east of the interpreted Caddo structure are possible refuse pits. Figure 15 shows the anomaly class map on which an archaeological interpretation is based.

6.2.2. Zones 178-Area2 and 178-Area3

The observed magnetic gradient in the neighboring zones 178-area2 and 178-area3 varies from −35 + 30 nT/m (Figure 14). Zone 178-area2 yields a mixture of numerous closely spaced monopolar (85) and dipolar (25) magnetic anomalies, which are categorized as class 1, 3, 5, and 6 geophysical anomalies and are plotted in Figure 14, on which an archaeological interpretation is based. Seven monopolar and fourteen dipolar magnetic anomalies of potential archaeological context occur in the central-southeast area of zone 178-area3. Numerous soil stratigraphy descriptions from neighboring geoarchaeological trenches (Kibler [88]) consistently place the late Holocene–late Pleistocene boundary at ~0.20 m deep in the area. Overall, brown, very friable, fine-grained sandy loam comprises the unconsolidated late Holocene sediments, which pass into unconsolidated late Pleistocene alluvium made of brown clay loam to clay. Preliminary shovel tests and test units within zones uncovered archaeological material within the late Holocene sediments. The complete archaeological interpretation of the horizontal magnetic gradiometry data for zones 178-area2 and 178-area3 is illustrated in Figure 15 and described next.

The massive assemblage of numerous closely spaced dipolar and monopolar magnetic anomalies in zone 178-area2 makes up an elongated highly magnetic structure that emerges from the background magnetic gradient. The feature strikes northwest–southeast, has an irregular perimeter, is nearly 75 m long × 30 to 40 m wide, and covers a surface area of nearly 2250 m². From an archaeological context, this elongated structure represents a sizeable communal midden possibly comprised of refuse materials such as lithic reduction debris, ceramic sherds, and bones. Additionally, a large assembly of multiple widely spaced monopolar anomalies and one dipolar anomaly wraps around the interpreted midden and is understood as the wide distribution of peripheral refuse pits.

The mapped concentric arrangement of monopolar and dipolar anomalies in zone 178-area3 suggests a main concentric structure, which is surrounded by secondary features that might have been functionally related to the main structure. The central concentric structure is consistent with the remains of several equally distributed post-molds of a Caddo house. The dipolar and monopolar anomalies immediately north of the concentric structure resemble a structural projection extending northwards, reminiscent of a prolonged entranceway of a Caddo house. The inner southern highly magnetized dipolar anomaly is interpreted as a burnt cooking surface developed on late Holocene fine-grained sandy loam. Two weakly magnetic monopolar and one dipolar anomaly next to the interpreted cooking surface are explained as storage pits. An external, strong dipolar anomaly, nearly 3 m west of the proposed Caddo house, is described as a highly fired soil (sandy loam), possibly an outdoor hearth, or a larger cooking surface. The other marginal magnetic anomalies surrounding the interpreted Caddo house outline are possible peripheral refuse pits.

6.2.3. Zone 244-Area1

Two nearby geoarchaeological trenches at about 115 m east of this block reveal the late Holocene–late Pleistocene boundary between 0.25 and ~0.30 m deep (Kibler [88]). Kibler [88] describes the uppermost 0.75 m of the soil stratigraphy within this area. In general, pale brown to very dark grayish brown, friable, very fine-grained sandy loam with an occasional fine oxidized Fe masses comprises the unconsolidated, ~0.30 m thick late Holocene sediments, which pass into unconsolidated late Pleistocene alluvium. The uppermost ~0.45 m of the alluvium consists of a firm to very firm, yellowish-brown to light brownish-gray clay loam to strong-brown clay, with clay films and silt and sand coats on ped faces and medium to coarse oxidized Fe masses.

Figure 16 depicts the interpreted horizontal magnetic gradient map for zone 244-area1, which yields a magnetic gradient from −40 to +30 nT/m and reveals twelve monopolar anomalies and fourteen dipolar anomalies that classify as class 1, 3, 5, and 6, as shown in Figure 16, from which archaeological interpretation is based. Collectively, the magnetic anomalies define a geometrically concentric pattern nearly 16 m in diameter. The circular pattern encircles three weak monopolar anomalies that emerge from the even weaker magnetic background. Several (dipolar and monopolar) magnetic anomalies dot the surroundings of the concentric pattern. Test units within this block yielded archaeological materials.

Several anomalies, here interpreted as burnt or highly fired post-molds, clearly define the ~16 m wide concentric structure, which is, in turn, interpreted as a Caddo house. The three anomalies inside the house are understood as pits. Two adjacent strongly magnetized dipolar anomalies directly south of the proposed Caddo house are construed as a hearth. The strongly magnetized dipolar anomalies in the northwest and east are interpreted as highly fired soil and as a wide burnt cooking surface. The remaining monopolar anomalies and one dipolar anomaly immediately outside the Caddo structure likely represent ditches and refuse pits. Figure 17 provides the archaeological interpretation of the magnetic map of zone 244-area1.

6.2.4. Zone 244-Area2

Several test units within this search block have yielded archaeological materials. The late Holocene–late Pleistocene boundary is at ~0.30 m deep (Kibler [88]). The late Holocene unconsolidated sediments are comprised of pale brown to very dark grayish brown, friable, very fine-grained sandy loam, which passes into unconsolidated late Pleistocene alluvium composed of clay loam to clay.

This zone yields a horizontal magnetic gradient from −28 to +26 nT/m and reveals 47 monopolar anomalies and 8 dipolar anomalies (see Figure 16) that classify as class 1, 3, 5, and 6, as shown in Figure 17, from which archaeological interpretation is based. The assembly of westerly magnetic anomalies, mainly class 5 and 6 and one class 1, clearly define a circular pattern nearly 10 m in diameter, which encloses two closely spaced strong dipolar anomalies. One class 3 dipolar anomaly occurs about 5 m east of the concentric structure. Multiple class 5 and 6 monopolar magnetic anomalies surround the concentric pattern. The eastern class-type 5 and 6 magnetic monopoles delineate a broad and irregularly shaped feature striking roughly northeast–southwest, measuring approximately 6 to 23 m wide × 16 m long, covering a surface area of almost 430 m².

The monopolar anomalies and one strong dipolar anomaly that make up the circular pattern are consistent with the close distribution of post-molds in a late Holocene soil horizon, which is, in turn, interpreted as a Caddo structure or house. The dipolar and monopolar anomalies immediately west of the concentric structure bear a resemblance to a structural projection extending to the west, which is indicative of an extended entrance to a Caddo house. The strong dipolar anomalies inside the proposed house are construed as an inner burnt cooking surface. The only class 3 dipolar anomaly behind the Caddo house is likely an outdoor hearth. The multiple monopolar magnetic anomalies surrounding the proposed Caddo house are interpreted as peripheral refuse pits and ditches. The broad eastern feature comprised of multiple closely spaced weak magnetic monopoles is possibly a dump site or midden, with two highly fired sites located in the southernmost limit. Figure 16 shows the archaeological interpretation of the magnetic anomaly classes from this zone.

6.2.5. Zone 244-Area3

Test units within this zone yielded archaeological materials. Soil stratigraphy descriptions from two neighboring geoarchaeological trenches (Kibler [88]) put the late Holocene–late Pleistocene boundary between 0.25 and 0.30 m deep. A friable, very dark grayish brown to pale brown to very dark gray, very fine-grained sandy loam comprises the unconsolidated late Holocene sediments, which change into unconsolidated late Pleistocene alluvium made of yellowish-brown firm clay loam to strong brown very firm clay.

The horizontal magnetic gradient map reveals a yield gradient from −16 to +16 nT/m with 126 monopolar and 21 dipolar magnetic anomalies (see Figure 16) that are classified as class 1, 3, 5, and 6, as shown in Figure 17, from which archaeological interpretation is based. An irregular accumulation of numerous closely to widely spaced monopolar anomalies mixed with lesser dipolar anomalies forms a large and unevenly shaped feature. The size of this feature is nearly 90 m long by 50 m wide (longest axes) and covers a surface area of approximately 4500 m².

From an archaeological context, the large size of this feature, its asymmetrical shape, the random distribution, and the high abundance of magnetic anomalies indicate a sizeable communal midden or dumpsite. In particular, there is a high concentration of strong magnetic dipoles inside the significant feature that might represent an area with highly fired soil sites. Six low-to-high-amplitude dipolar anomalies, also considered highly fired soil sites, occur inside this midden. Numerous peripheral magnetic anomalies, interpreted as refuse pits, some of which are highly fired sites, occur outside the midden. Figure 16 depicts the contextual archaeological interpretation of this zone’s magnetic map.

6.2.6. Zone 244-Area4

Soil stratigraphy descriptions from two ~1.0 m deep geoarchaeological trenches within a 45 m radius from this block place the late Holocene–late Pleistocene boundary between a depth range of 0.30 to 0.50 m (Kibler [88]). A friable, very dark grayish-brown to pale brown, very fine sandy loam comprises the late Holocene unconsolidated sediments, which pass into unconsolidated late Pleistocene alluvium comprised of a firm to very firm, yellowish-brown to light brownish-gray clay loam to strong brown clay.

Figure 16 shows the horizontal magnetic gradient map for this zone and yields a magnetic gradient from −22 to +22 nT/m with 65 monopolar anomalies and 23 dipolar anomalies under classes 1, 3, 5, and 6, as shown in Figure 17, from which archaeological interpretation is based. Four strong dipolar anomalies and seven monopolar anomalies, equidistantly located, form a 13 m wide concentric structure in the southern search block. Two dipolar anomalies immediately southwest of the concentric structure resemble a structural projection extending to the southwest. Three moderately spaced magnetic monopoles occur inside the circular structure, and multiple closely to widely spaced monopolar and lesser dipolar anomalies cluster outside. On the northern block, three dipolar anomalies and seven monopolar anomalies define a ~19 m wide concentric pattern, which encloses one magnetic monopole in the center. Numerous closely spaced monopolar and lesser dipolar anomalies occur immediately around the periphery of the concentric structure. Eight equidistant and linearly organized magnetic anomalies with moderate to high amplitudes define a ~20 m long × ~4 m wide north–south striking linear structure ~4 m east of the round feature.

The magnetic anomalies delineating the concentric structures in the southern and northern blocks suggest the post-mold remains of Caddo houses. The proposed Caddo house in the south block likely had a typical entranceway projecting southwestward, but the house in the north block might have faced northeast without an extended entranceway. The monopolar anomalies observed inside the proposed Caddo structures represent storage pits. The closely spaced strong dipolar anomalies immediately outside the entrance of both Caddo houses are interpreted as burnt cooking surfaces. The numerous weak-to-strong monopolar and dipolar anomalies distributed around the proposed Caddo houses are interpreted as highly fired soil sites, refuse pits, and few ditches. The broad spatial distribution of the refuse pits observed next to each Caddo house suggests that each house might have had a dumpsite with several peripheral refuse pits. The observed linear feature east of the northern Caddo house is probably a compound fence or enclosure. The archaeological interpretation of the magnetic map from these search blocks is shown in Figure 16.

6.2.7. Zone 244-Area5

Following the soil stratigraphy description of three ~1.0 m deep geoarchaeological trenches within a 75 m radius from this block, the late Holocene–late Pleistocene boundary undulates between a depth range of 0.20 to 0.50 m (Kibler [88]). A very dark grayish brown to very dark grey, friable, very fine-grained sandy loam makes up the unconsolidated late Holocene sediments, which, in turn, transition into unconsolidated late Pleistocene alluvium composed of clay loam to strong brown clay.

The horizontal magnetic gradient in this zone varies from −16 to +16 nT/m (see Figure 16) and yields 72 monopolar anomalies and 30 dipolar anomalies, which are classified as classes 1, 3, 5, and 6, as shown in Figure 17, from which archaeological interpretation is based. However, most of the anomalies occur on the east side. Five weak dipolar anomalies and three weak monopolar anomalies located on the west side define a ~11 m wide concentric pattern. Two moderately spaced magnetic dipoles occur inside this circular structure, and also multiple closely-to-widely spaced magnetic monopoles and dipoles arise outside. A north–south striking, ~32 m long × ~3.5 m broad assemblage of seven dipolar and thirteen monopolar anomalies arranged linearly crosses the search blocks immediately east of the circular structure observed on the west side. On the east side, three moderate-to-strong dipolar anomalies and eleven weak-to-strong monopolar anomalies also define a ~17 m wide circular pattern, which encloses three magnetic dipoles and three monopoles and is also bounded by multiple closely to widely spaced magnetic monopoles and dipoles in the north and south. Another north–south striking, ~24 m long × ~1.5 m wide linear accumulation of four equidistant monopolar anomalies runs east of this concentric structure.

The magnetic anomalies delineating the concentric structures suggest post-mold remains, which, in turn, support the presence of two neighboring Caddo houses, one of which might have had a distinctive entranceway. The two weak monopolar anomalies observed in the center of the smaller Caddo structure represent storage pits. Inside the larger Caddo house is a possible burnt cooking surface next to four storage pits. The numerous weak-to-strong monopolar and dipolar anomalies distributed around the proposed Caddo houses are interpreted as hearths, highly fired soils, ditches, and refuse pits. The two observed linear features adjacent to the proposed Caddo houses are likely compound fences or enclosures. Figure 14 explains the archaeological interpretation of the magnetic map in this zone.

6.2.8. Zone 244-Area6

Soil stratigraphic descriptions from three geoarchaeological trenches within a 70 m radius from this block place the undulating late Holocene–late Pleistocene contact between 0.20 and ~0.50 m deep (Kibler [88]). A friable, dark grayish brown, very fine sandy loam comprises the unconsolidated late Holocene sediments within the site location. The uppermost ~0.50 m of the late Pleistocene alluvium consists of a firm to very firm, brown solid clay loam.

Figure 16 shows the horizontal magnetic gradient map for this zone. The magnetic gradient varies from −15 to +15 nT/m and reveals 49 monopolar anomalies and nine dipolar anomalies that are categorized as class types 1, 3, 5, and 6, as shown in Figure 17, from which archaeological interpretation is based. Together, the magnetic anomalies clearly define a geometrically concentric pattern nearly 14 m in diameter. The circular pattern encircles five monopolar anomalies. Multiple (dipolar and monopolar) magnetic anomalies border the edge of the concentric pattern. A group of monopolar anomalies immediately northeast of the circular pattern resembles a structural northeastward projection that may be an extended entranceway common in Caddo structures.

Thirteen anomalies, here interpreted as post-molds, clearly define the 14 m wide concentric structure, which is, in turn, understood as the blueprint of a Caddo house. The five monopolar anomalies observed inside the proposed house are taken as storage pits. Four strongly magnetized dipolar anomalies located in the northwest corner of the block are interpreted as an 8 m wide outdoor hearth or burnt cooking surface. The strongly magnetized dipolar anomalies in the northwest and east are interpreted as a highly fired soil site and as a wide charred cooking surface. The remaining monopolar anomalies outside the Caddo structure likely represent a group of dispersed ditches and refuse pits. The dipolar anomaly facing the structure’s entrance resembles a highly fired soil. Figure 16 provides the archaeological interpretation of the magnetic gradient map of this zone.

6.2.9. Zone 244-Area7

Figure 16 exhibits the horizontal magnetic gradient map. The magnetic gradient varies from −12 to +12 nT/m and yields 37 monopolar anomalies and six dipolar anomalies. The anomalies are categorized as classes 1, 3, 5, and 6. Together, eight equidistant magnetic anomalies outline a ~9 m wide geometrically circular arrangement. The circular feature circumscribes two monopolar anomalies located almost in the center of the circular structure. Multiple monopolar magnetic anomalies arise outside the concentric structure. A northwest–southeast striking, ~38 m long × ~4 m wide, assembly of six monopolar and five dipolar anomalies arranged linearly traverses the search block immediately behind (east) the proposed circular structure. Multiple monopolar anomalies are dispersed around the proposed circular and linear structures. East of the magnetically low linear structure is an irregular accumulation of widely spaced monopolar anomalies with a horizontal magnetic gradient varying from −1 to +3 nT/m. The irregularly shaped magnetic feature is nearly 20 m long × 10 m wide, covering a surface area of 200 m². The eight anomalies that delineate the concentric structure are diagnostic of equidistantly positioned post-molds, which, in turn, are interpreted to represent the physical remains of a Caddo house without a protrusive entrance. The two weak monopolar anomalies observed in the center of the house represent adjacent storage pits. The various weak-to-moderate monopolar anomalies scattered around the Caddo house are possible ditches and refuse pits. The long linear feature is expected to be a Caddo compound fence or enclosure, whereas the immediately adjacent irregularly shaped magnetic feature is a potential small dumpsite or midden. This archaeological interpretation is shown in Figure 17.

7. Preliminary Results from Archaeological Excavations

Archaeological investigations at site 41FN244 excavated and documented seven Activity Areas associated with an intensive Middle Caddo period (A.D. 1250–1400) hamlet or farmstead-village community (Rutherford [37]). The location of excavation blocks was based on the high-potential areas and anomalies identified in the geophysical investigations (Rutherford [37]). A large anomaly was found in block 580 E 823 N (Figure 18A–E), in Activity Area 3 within the surveyed area 244-area3, extended across 16 excavation units and included 22 cultural features associated with lithic technologies and residential activities. These cultural features include a sizeable residential midden, a sandy clay-packed floor (Figure 18A), two hearths (Figure 18B), three artifact clusters, six post-molds (Figure 18C), five pit features, and two thermal pit features. The lithic artifact assemblage from these features includes 1669 pieces of debitage, 62 flake tools, 18 bifaces, 15 complete projectile points, 11 projectile point fragments, 10 cores, 4 handstones, 7 hammerstones, 5 modified stones, and 1 abrader.

All artifact types are primarily concentrated within the midden area in the northeast corner of the block from 673 to 703 cm below datum depth (cmbd). Ground stone and debitage were present in the central hearth located in the center of the block and the northwest hearth located on the western periphery of the midden fill (683–706 cmbd) (Figure 18D). Post-molds are located outside of the midden area and are found in a straight line in a northwest-to-southeast orientation (663–703 cmbd) (Figure 18E). The presence of debitage, a biface fragment, and flake tools in post-molds suggests lithic work areas were either swept clean, and the artifacts were pushed against the walls, or tool maintenance occurred along the walls. Artifact clusters and pit features represent discrete depositional patterns of tool recycling or trashing. Pits and artifact clusters are mainly located within the midden area and on the southern and southeastern peripheries of the midden area. The floor space is located across the entire block surface and extends from 693–723 cmbd; all artifact types are diffusely distributed across all excavation units, with concentrations located near the central hearth area.

The lithic assemblage recovered from block 580 E 823 N reflects activity areas that process, produce, manufacture, and store work materials, tools, and foods. The toolkit consists of a wide range of expedient and curated technologies used to undertake a variety of cutting, sawing, scraping, slicing, chopping, and grinding activities. The preliminary results of the cultural features and lithic assemblage suggest the occupants of the structure practiced a hunter–fisher–farmer economy that manufactured and maintained a wide range of versatile, maintainable, and reliable stone technologies year-round. Furthermore, these preliminary results suggest that discrete work areas associated with in situ artifacts and cultural features were used as activity areas for processing, manufacturing, and maintaining technologies and material goods.

8. Conclusions

In Caddo archaeological investigations, land magnetometry has been the most widely used exploratory technique due to its effectiveness in mapping the existence of thermal features and post-molds of Caddo houses. Conversely, not new to archaeological prospecting, the use of 3D ERI has been underused in Caddo archaeology despite its ability to quickly map both the horizontal and vertical occurrences of architectural structures.

The successful integration of non-destructive, high-resolution, true-3D and pseudo-3D ERI and land HMG surveying successfully supported archaeological data recovery investigations within the critical 30 × 30 m search blocks of archaeological sites 41FN178 and 41FN244 prior to the Bois d’Arc Lake Reservoir’s impounding phase. Three-dimensional ERI mapped a total surface area of 2133 m², reaching a depth of investigation greater than four meters with a high resolution. Land HMG mapped a total surface area of 15,640 m² down to ~2 m depth and was helpful in mapping the occurrence and distribution of zones with a higher probability of being ancient human activity areas and targets for archaeological excavation. Archaeological excavation supports the integrated interpretation based on the 3D ERI and land HMG methods, as shown by the preliminary results from surveyed area 244-area3 (Activity Area 3, discussed in Section 7). The geophysical survey results support the archaeological interpretation of the site, which is that it was a Middle Caddo farmstead or hamlet site. It also aided in the design and implementation of strategic archaeological excavations, mainly at site 41FN244. At site 41FN244, 3D ERI found numerous resistive anomalies surrounding a conductive anomaly, collectively interpreted as a concentric group of post-holes surrounding the remains of a Caddo house’s inner clay floor. It also found a nearby cluster of several resistive anomalies interpreted as midden or middens. The HMG survey carried across both sites identified numerous scattered monopolar and dipolar anomalies interpreted as concentric assemblages of the post-molds of Caddo houses and adjacent compound enclosures or fences, as well as middens.

Additionally, several dispersed magnetic anomalies are thought to be shallowly buried hearths, burn cooking surfaces, storage pits, and ditches. The mapped magnetic anomalies agree with the location and distribution of previously found archaeological artifacts and the extent of resistive and conductive resistivity anomalies. The combined near-surface geophysical methods were instrumental in rapidly mapping the horizontal and vertical extent of shallowly buried anomalies within a large area. Archaeological excavations of these geophysical anomalies have preliminarily confirmed interpretations.