Introduction

Climate change is critically altering coastal and wetland landscapes around the globe1,2,3,4,5,6,7,8,9. Such transformations manifest variably as rising sea levels, warming sea-surface temperatures, more frequent extreme weather events and storm surges, and subsequent changes to salinity and freshwater regimes. Climate change is also rapidly driving coastal erosion, the loss of wetland habitats, and the collapse of estuarine biodiversity and community structures. These impacts are multi-scalar, ranging from effects on microbial communities to regional scale plant and animal communities and landscape formations. For example, experimental studies along the southeastern Atlantic coast of the United States in the state of Georgia, home to one of the most expansive stands of marshland and adjacent maritime hardwood forests in the world10, have demonstrated extreme impacts of saltwater intrusion11,12. At small scales, even moderate influxes of saltwater simulating heightened sea levels had cascading impacts on ecosystem characteristics and geomorphology that only compounded when scaled up. For instance, a loss of root and rhizome density contributed to alteration of nutrient levels and ratios in just a single month, extirpation or replacement of plant communities within 18 months, critical reductions in carbon uptake due to reduced biomass, and an observed loss of nearly two centimeters of elevation in just two years11.

These impacts are not unique to the Georgia coast, and processes of ecosystem and landscape loss and change, and the role of climate change in these processes, are well documented globally. What remains poorly understood in these processes, is the role of deep-time human-environment interactions in shaping modern ecosystems and landscapes, and how historical “baselines,” from which ecologists, conservationists, and other environmental scientists and/or activists measure the degree of difference of modern ecological conditions, too often presume some ideal “state of nature” uninfluenced by human activities (see refs. 13,14,15,16,17,18,19,20,21). Modern ecological and conservation approaches to mitigating the impacts of climate change tend to be predicated on the existence of such “natural” conditions across which conservation efforts and climate change impacts might be assessed (e.g., refs. 22,23). While most scientists now recognize the complexity, and politics, of establishing ecological baselines (e.g., refs. 24,25), and increasingly accept the necessity of using the historical and paleoecological records to construct these baselines (e.g., refs. 26,27,28,29), the majority of this work fails to capture the deep-time, keystone role of people in their models. Importantly, the peoples, actions, histories, and ecosystems often neglected in historical baselines tend to be Indigenous, and in the case of the southeastern Atlantic coast, also enslaved and free Gullah-Geechee (Black) peoples, whose communities are so often disenfranchised from their ancestral landscapes.

Historical baselines are often established as homogeneously “pre-European”26 and are subsequently used as “referents” by which modern ecosystem functions are compared to those of the past28. Such “natural” baselines however were more often than not engendered by deep-time human environment entanglements. Humans, across space and time, actively drive ecosystem processes by way of such activities including landscape management via controlled fire regimes30,31, the long-term management of fisheries and shellfisheries32,33 as well as plant communities34,35,36,37,38,39, and large-scale landscape and landform modifications that drive modern ecosystem functions and coupled natural and human systems40,41. We argue that it is not enough to simply measure the distances of modern ecological conditions from “baseline” conditions. Rather, if we are to build resilience into our modern coastal and wetland ecosystems to mediate the impacts of climate change, we must understand how our “baselines” were produced and how these legacies of long-term human-environment interactions continue to drive modern ecosystems. In other words, what are the “primary legacy drivers” of the contemporary ecosystem?

We define “primary legacy drivers” as the ecological remnants of past human-environment engagements that continue to shape modern ecological conditions and processes. The Millennium Ecosystem Assessment defines drivers as “any natural or human-induced factor that directly or indirectly causes a change in an ecosystem”42. More specifically, ecologists have discussed legacy effects as “indirect effects that persist for a long period of time in the absence of the casual species, or after this species has ceased the casual activity”43. While direct drivers are defined as those that “unequivocally influence ecosystem processes,” indirect drivers cover those that alter direct drivers, including human institutions, sociopolitics, economics, etc. And, while the term legacy effects does capture some of the critical issues here that need to be considered for understanding long-term ecosystem dynamics, for the most part, ecologists employ these ideas in the absence of the broader socio-economic context of such actions. This is particularly problematic when we consider how such processes operated at long-term time scales (e.g., multiple centuries). By not paying attention to the broader social context, we are left with a view that perceives such effects as descriptive drivers, lacking a deeper understanding of how the results of human institutions are uniquely situated to become deeply embedded within the daily functioning of ecosystems and subsequently impacting them over expanded time frames.

We pose that the term “primary legacy driver” captures the holistic range of impacts of deep-time human histories on their ecosystems and defines the material legacy of these indirect drivers on modern, direct ecosystem drivers. Economic systems, political organizations, and other institutions of human activity “indirectly” and, at times, “directly” on shorter timescales drove ecosystem changes in the past. These ecological changes, whether landscape modifications, alterations of soil geochemistry, or the distribution of plant and animal species, exist today as anthropogenic keystone structures40,41 that directly articulate with the functioning of contemporary ecosystems. We consider these to be primary legacy drivers and argue that, from the perspective of conservation and sustainability, they are unique and more complex in their temporality, their historicity, and their anthropogenic entanglements rather than simple concepts of “direct” or “indirect” may suggest. Thus, if we are to understand the resilience and sustainability of these systems in the face of climate shifts, we need to have both an understanding of how such ecosystems function in the present and recent past (ca. last 50 years) and the long-term legacy effects and the primary actions that developed in concert with these ecosystems over millennia or more. Key to such an understanding is not only the impact that such actions had from the past, but also to what degree to these lasting primary legacy drivers have for ecosystems under extreme contemporary environmental shifts, particularly when humans no longer practice the institutions that operated in the past (see ref. 41).

In this paper, as archeologists, we highlight how 5000 years of Indigenous Native American and early industrial Euro-American settlement along the Georgia coast of the southeastern United States continues to shape modern coastal and estuarine ecosystems (Fig. 1). We argue that archeological data is not just a valuable source for paleoecological data, but a record of those past human actions that produced the anthropogenic ecosystems that ecologists often use or view as natural historical baselines. From this perspective, the Georgia coast is the cumulative effects of five-millennia of human-environment interactions that continue to drive certain facets of landscape and ecosystem resilience in the face of climate change across the southern Atlantic coast. To demonstrate, using archeological and modern data, we highlight (1) the expanse and time-depth of human occupation across estuarine landscapes, (2) the cumulative nature of these occupations, (3) the local and landscape-scale impacts of the deposition of trillions of oyster shells onto landforms by Indigenous communities, and (4) the large-scale landscape modifications of Euro-American plantation owners and enslaved communities. Through these examples, we illustrate how even protected or “natural” areas, often used as measuring sticks for modern conservation and sustainability efforts, are the result of substantive anthropogenic intervention. In this regard, we add to discussions regarding “natural” baselines, as pre-human ecosystems, those ecosystems unaffected by deep time human engagements, in many places, simply do not exist (see refs. 44,45,46,47,48,49).

Fig. 1: Regional maps of the Georgia coast.
Fig. 1: Regional maps of the Georgia coast.
Full size image

A The study area. B Distribution of all recorded Indigenous (teal) and historic Euro-American (red) archeological sites across the six Georgia coastal counties. C Southern portion of Sapelo Island showing back-barrier marsh area and marsh islands/hammocks referred to in the text. D Kernel density estimates for radiocarbon dates for two of the sites discussed in this paper (Kenan Field and Middle Place) and a summary of all radiocarbon dates associated with Indigenous occupations across the entire Georgia coast. The two main peaks for the entire coast model are partly a product of research foci, including higher research activity on Late Archaic (c. 5000-3000 cal BP) and later Mississippian period (c. 1000-500 cal BP) sites. The latter peak, when taken in concert with the increase in archeological sites during this period (see ref. 58), does indicate a substantial population increase at this time. This is when the large majority of shells were deposited across the coast. Satellite imagery sources as compiled by Esri ArcGIS Pro: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community.

Results

Pre- and early industrial human settlement along the Georgia coast is extensive and cumulative

The coastal and wetland environments of the Georgia coast include barrier islands, small marsh islands, saltwater marsh, brackish marsh, tidal creeks, river channels, and maritime forests50. These environments are anchored by large, Pleistocene barrier islands that separate the marsh ecosystems, located between the mainland and the barrier islands, from the open ocean. The Pleistocene shoreline that created these barrier islands formed between 110,000 and 25,000 years ago and today is fronted on the east by Holocene islands that formed c. 5000 years ago51. During this mid-Holocene period, back-barrier areas were re-flooded, forming the modern-day salt marsh lagoon system52,53,54. It was during this time that our modern coastal and marsh ecosystems emerged, and as the Holocene islands emerged along with them, Indigenous communities began to people these places in greater numbers immediately41.

Recorded in the Georgia Archeological Site File (the statewide repository for all known archeological sites in the state of Georgia), there are c. 3300 known Indigenous archeological sites across the six counties that make up Georgia’s coast (Fig. 1B). These sites span the last 5000 years up until the 17th and 18th centuries AD. To illustrate the long-term temporality of human impacts, Fig. 1D presents a modeled summary of all radiocarbon data (n = 604) from the Georgia coast. Using kernel density estimate (KDE) models it is clear that human activity in the region accelerated beginning circa 5000 cal BP. We presume that the probability curve is a loose proxy for relative populations over time, but we recognize the probable impact of research bias on the available data (e.g., increased focus on Late Archaic (circa 5000-3000 cal BP) components relative to others). Regardless of relative bias, the results demonstrate a consistent human presence on the coast since the period of earliest human settlement.

Along the immediate coast, including the large barrier islands themselves and the small marsh islands, or hammocks located in the marsh, there are 1047 known Indigenous archeological sites representing settlement and occupation over the last 5000 years, coeval with entire life history of these ecosystems themselves. Of these 1047 sites, roughly two-thirds (n = 710) are located on the large barrier islands while one-third are located across the small hammocks (n = 337)41. In terms of density along the entirety of the 160 km of the Georgia coast, there are c. 4.6 known Indigenous archeological sites per square kilometer of marsh island uplands and 2.1 known Indigenous sites per square kilometer of barrier island upland41. Importantly, this density is not homogeneous across the landscape. Instead, Indigenous occupations were consistently denser in the areas along the mainland sides of the barrier islands, and less dense between islands. As Thompson et al.41 illustrate, in the marshy back-barrier area along the northern Georgia coast, between Skidaway and Wassaw Islands, known Indigenous site density for small marsh islands is c. 33.1 sites per square kilometer of hammock uplands. This heterogeneity of occupation across the landscape has important implications for mosaic, or patchy, landscapes that tend to drive higher rates of biodiversity at landscape scales (sensu55).

Full island surveys have illustrated the sheer density and extent of human occupation across these small marsh islands41,56. Four marsh islands, Little Sapelo (0.47 km2), Pumpkin (0.03 km2), Mary (0.1 km2), and Patterson (0.18 km2) have been subject to full surveys, with test pits dug every 20 m on a regular grid across each island (Figs. 1C and 2). Across each one, c. 65% of all tests yielded evidence for Indigenous occupation while 25%, or one out of every four tests, yielded evidence for substantive shell deposition by Indigenous communities41. Simply, what this means is that in a random survey, if one were to dig a hole for a stratigraphic profile, one out of every four would include a shell unit (deposited by Indigenous people) within the profile. Expanding to any evidence at all for Indigenous occupation (e.g., artifacts, but no shell necessarily), one out of every one-and-a-half tests (65%) would show evidence for Indigenous land use. Importantly, we must note that the main limitations of archeological data in this study are that we are vastly underestimating the anthropogenic legacies of land use on modern landscapes. That is, the total number of archeological sites that have existed along the coast can never be known. Sites have been destroyed for urbanization, for roads, and by shifting sea levels and erosion. As such, our estimates for the extent and intensity of Indigenous land use here in this paper are on the lowest end possible. As we will demonstrate in the next section, the extent of Indigenous occupation on these islands, especially their deposition of oyster shells, critically altered these landforms.

Fig. 2: Survey extent on top of LiDAR maps (elevation) for marsh hammocks.
Fig. 2: Survey extent on top of LiDAR maps (elevation) for marsh hammocks.
Full size image

A Patterson Island, B Mary Hammock, and C Little Sapelo. Each map (A–C) indicates the extent of the shovel test survey conducted on each hammock. Shovel tests are indicated as circles. The sizes of the circles are scaled proportional to the measured thickness of subsurface shell layers that were encountered during the survey. D Depiction of tidal zones, including supratidal, intertidal, and between low tide averages at Patterson Island (top panels) and Mary Hammock (bottom panels) at present-day sea levels (left panels) and in an alternate scenario in which anthropogenic sediments (shell layer thicknesses depicted in (A–C)) are removed from these hammocks (right panels). Results show that large portions of these hammocks would sit lower relative to local tidal averages were it not for intensive shell deposition by past inhabitants of these hammocks. LiDAR data were processed using publicly available data via NOAA. Satellite imagery sources as compiled by Esri ArcGIS Pro: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community.

On Ossabaw Island, one of the most thoroughly surveyed coastal barrier islands, Indigenous sites dating to the last 5000 years span the 36 km2 of wooded upland (Fig. 3A). To date, 220 known Indigenous archeological sites have been recorded on Ossabaw Island, at a density of 6.1 sites per square-kilometer57. In addition to Indigenous occupations, beginning in the 18th century AD, Ossabaw became home to four large plantations that relied on the labor of enslaved Africans (Fig. 3A). As we will discuss below, these plantation economies greatly altered landforms, hydrological processes, and ecosystem functioning. Importantly, though, the modifications and land use patterns favored by Euro-American plantation owners were very frequently conditioned by the ecological legacies of Indigenous occupation and interactions with the environment. For example, based on archeological site sizes recorded by Pearson57 on Ossabaw, while the average size of an Indigenous archeological site is c. 0.95 ha, the average size of the Indigenous sites located closest to or at the four plantations established on Ossabaw is 13.64 ha (Fig. 3C). Additionally, the Indigenous sites closest to the plantations have an average number of occupation components (or time periods represented) of 4, while the average number of components across all Indigenous sites on the island is just 1.57, indicating that not only were these sites the most expansive Indigenous occupations on the island, they were also occupied for longer and more intensively (Fig. 3). These figures demonstrate that the cumulative and intensive Indigenous occupation of landscapes heavily influenced subsequent environmental conditions that were found to be highly favorable for industrialized plantation agriculture. These plantation economies themselves then became part of the continuous, cumulative human engagement with, and structuring of, coastal and wetland ecosystems as plantations extensively modified the landscape with networks of canals, ditches, and agricultural fields.

Fig. 3: Indigenous archeological and historic plantation sites located on Ossabaw Island.
Fig. 3: Indigenous archeological and historic plantation sites located on Ossabaw Island.
Full size image

All (A) Indigenous site locations are digitized from Pearson57. B includes a map of the Indigenous and Historic archeological site at Middle Place, including the surface-mapped shell middens across the eastern portion of the site. Twenty-three of these middens were tested and serve as the basis for shell-count extrapolations presented in the text. The boxplot (C) compares the total distribution of all Indigenous site sizes across Ossabaw Island (bottom box) with the site sizes of the four Indigenous sites located closest to, or underlying, plantations (top box). Red triangles indicate the data points used to create the boxplot (the sizes of the four Indigenous sites upon which plantations were subsequently located). LiDAR data were processed using publicly available data via NOAA. Satellite imagery sources as compiled by Esri ArcGIS Pro: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community.

Indigenous communities shaped ecological and geomorphological processes through the extraction and deposition of trillions of oyster shells

One of the primary reasons that the extent and intensity of Indigenous settlement is so critical to understanding modern ecological legacies is that most Indigenous archeological sites along the Georgia coast represent places on the landscape where masses of oyster shell (and other species of shellfish and organic remains) have been deposited in the form of dense layers of shell atop natural landscapes. Indigenous use, extraction, and deposition of oysters is well-documented and understood. Oysters were a key component of Indigenous foodways and were secondarily used to build shell architecture (e.g., shell construction layers in platform mounds). Indirectly, however, the accumulation of 5000 years of oyster shells across these coastal landscapes has played a “keystone” role in ecosystem functioning40,41. Critically, the deposition of oyster shells altered island topographies (via added elevation) and locally transformed soil geochemistry and thus island biogeographies (see below).

To illustrate the scale of oyster deposition, we draw on two recent archeological field projects that allow direct quantification and estimation of deposited oyster shells. Together, Middle Place (c. 90 ha) on Ossabaw Island (Fig. 3B) and Kenan Field (c. 66 ha) on Sapelo Island (Fig. 4) are two of the largest Indigenous archeological sites on the Atlantic coast. Middle Place was a large, Indigenous town occupied for roughly 350 years between c. AD 1150-1500 (Fig. 2D). Approximately one-third of the site remains intact, untouched by destructive historic plowing during the plantation period (c. AD 1790 s to 1861). In this one-third of the site, there are 168 domestic shell midden-mounds that have been mapped on the surface (Pearson57). Twenty-two of these shell middens were recently tested via small-scale excavations. Using direct counts and weights of oyster shells excavated from these tests, we extrapolated the abundance of oysters across the rest of the 168 middens for which we have reliable dimensions. We estimate that c. 30,000,000 oysters (or 60,000,000 left and right shells) are deposited in just the eastern one-third of the Indigenous town at Middle Place. This roughly calculates to c. 3,000,000 pounds of oyster shell deposited over a roughly 30 ha area.

Fig. 4: Shovel test survey of Kenan Field on Sapelo Island.
Fig. 4: Shovel test survey of Kenan Field on Sapelo Island.
Full size image

In (A) shovel tests are indicated as circles and are sized proportional to the amount (kilograms) of shell recovered from each shovel test. The underlying surface of (A) is LiDAR (elevation) data. B presents shovel tests as small, black circles while shell middens mapped on the surface across the site are indicated in teal. LiDAR data were processed using publicly available data via NOAA. Satellite imagery sources as compiled by Esri ArcGIS Pro: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community.

Similarly, at Kenan Field on Sapelo Island (Fig. 4), a systematic shovel test survey allows the calculation of total estimated deposition of shell (both above and below the surface) and has been used to reconstruct both diachronic patterns of settlement structure and demography58. Radiocarbon dating shows that Kenan Field was occupied consistently since at least 4500 cal. BP, though the primary occupation, and majority of shell deposition, spans the last 1000 years (Fig. 2D). During this time, approximately 1.6 billion oysters, or over 160 million pounds of shell, are estimated to have been deposited across its 66 ha area. These estimates suggest that 11% of the site’s volume is comprised of shells. Extrapolated to the 1047 Indigenous sites located throughout the coastal landscape recorded in the Georgia Archeological Site File, which includes both shell and non-shell bearing sites, the estimated volume of oyster deposits, and the resultant scale of ecological and landscape modifications, becomes staggering.

Similar depositions of shells occur across the small hammocks scattered throughout the marsh. These small marsh islands are critical, keystone structures within these broader ecosystems as they provide resources, shelter, and goods/services that are crucial for a host of plant and animal species41,59. In fact, these isolated uplands located within the marsh affect over 45% of marsh ecosystem processes, providing refuge for specific communities of plants and animals41. As Whitaker et al.60,61 point out, ecologists along the Georgia and South Carolina coasts have even identified “High Calcium Communities” of plants that are hosted exclusively on deposited oyster shells which foster the growth of specific plant species that are otherwise extremely rare across these ecosystems (also see refs. 62,63).

Of the four hammocks surveyed by Thompson et al.41, an average 25% of each island yielded substantive shell deposits. These deposits ranged in thickness between 1 cm and 90 cm (Fig. 2). As such, the modern topographies of these islands are directly the result of Indigenous land-use legacies (Figs. 5 and 6B). Figure 5 includes stratigraphic examples of the different kinds of landscape modifications resultant from Indigenous land use practices including extensive heaps (Fig. 5A), circular shell pilings that enclosed Indigenous villages (Fig. 5B), and individual households trash piles of dense shell that likely number in the hundreds of thousands up and down the coast (Fig. 5C-F).

Fig. 5: Examples of archeological shell deposit profiles.
Fig. 5: Examples of archeological shell deposit profiles.
Full size image

The images here depict three different kinds of Indigenous shell deposits. All represent anthropogenic deposits and include Indigenous artifacts. A is an example of large, extensive shell heaps while (B) is an example of a shell ring: large, circular shell pilings that surrounded Indigenous villages (up to c. 60–100 m across) during the Archaic period (c. 5000-3500 cal BP). C–F are representative of ubiquitous domestic midden-mounds associated with household and daily refuse deposited by Indigenous communities. A A one meter deposition of shell on top of Pumpkin Hammock’s natural surface, which raises the landform to well above modern high tide levels; (B) Profiles from one of the large shell rings on Sapelo Island (adapted from Waring and Larson104 and Thompson105); (C) Profile drawing of excavations into shell midden on Pumpkin Hammock, note the dense shell deposits atop sterile, natural sand landform; (D) Photograph of drawn profile depicted in (C); (E) Photograph of excavations into shell midden on Pumpkin Hammack, drawn profile depicted in (F); (F) Profile drawing of excavations into shell midden on Pumpkin Hammock.

Fig. 6: All potential landscape modifications identified via LiDAR on Ossabaw Island.
Fig. 6: All potential landscape modifications identified via LiDAR on Ossabaw Island.
Full size image

A Roads are indicated as thick, teal lines while other linear anomalies identified in the LiDAR, including ditches and canals, are indicated in white. Photographs illustrate (B) the modification of the landscape via shell mounding, (C) the scale of canals, and (D) the thickness of the shell and organic-rich topsoil at the South End plantation (dark brown layer flecked with white shell overlying sandy yellow substrate). LiDAR data were processed using publicly available data via NOAA.

While these shell-bearing landforms provide substrate for unique plant and animal communities, they also contribute to the overall geomorphological resilience of the coastal landscape. Indeed, in many cases, current (and future projected) high-tide lines would have already overtaken the natural ground surface41 (Fig. 5A). To formally demonstrate the amount of land that would be lost to rising sea levels if it weren’t for the deposition of shell by Indigenous communities, we provide a series of formal topographic models (Fig. 2D). The detailed systematic excavation surveys of the hammocks, whereby small standardized holes were dug at equal intervals across the whole islands, along with topobathymetric data from the U.S. Geological Survey’s (USGS) Coastal National Elevation Database (CoNED64) and average tidal levels at Water Level Stations along the Georgia coast in the National Oceanic and Atmospheric Administration’s (NOAA) National Water Level Observation Network (NWLON65), provided a basis for the construction of a spatial model of the hypothetical scenario in which no anthropogenic deposition occurred on coastal hammocks. A present-day tidal zone baseline can be contrasted to an alternate scenario in which humans have not substantially structured the landscape by digitally removing anthropogenic sediments (shell deposits) from these hammocks. The results of this analysis (Fig. 2D) illustrate the substantial impact of anthropogenic landscape modification along the Georgia coast. In the study areas for each hammock, ca. 4.9% of the area of Patterson Island and ca. 4.5% of the area of Mary Hammock that sits above the highest average daily tides (MHHW) today would be submerged in this alternate scenario. Moreover, what areas remain above MHHW also would sit lower and would regularly be inundated by higher than average tides, such as King Tides and surges from minor tropical storms, for example.

Indigenous and Euro-American communities intentionally modified island landscapes and landforms

This discussion of shell deposition has focused primarily on the unintended ecological and geomorphological consequences of Indigenous engagements with estuarine ecosystems. Separate from these kinds of impacts, however, are those modifications to the landscape by both Indigenous and Euro-American settlers that were meant to intentionally alter ecosystem processes, primarily by way of water management, agricultural production, and the construction of transportation infrastructure.

There are roughly 165 km of known landscape modifications across Ossabaw Island (Fig. 6A). These modifications include substantial landscape alterations such as canals, ditches, roads, causeways, dams, and fences, all of which may influence local ecological conditions, including the movement of water (salt, brackish, and fresh) and the movements of people and plant and animal species across islands. Upon their creation, alterations such as canals and ditches would have moved salt and brackish waters into the interiors of barrier islands to bring nutrients to agricultural fields with the incoming tides66 (Fig. 6C). These alterations continue to exist today and also continue to connect freshwater ponds to brackish estuarine systems, critically altering hydrological regimes, salinity gradients, and both terrestrial and lentic ecosystems.

While most researchers have assumed the majority of these modifications to be Euro-American in origin, preliminary unpublished and ongoing work has begun to hint that at least some of these features, including canals, paths, and other water management features like dams may have been constructed and used by Indigenous communities centuries before Euro-American settlement. Indigenous-built canals are certainly known from other areas of the coastal southeastern United States67,68. It is possible that Indigenous alteration of the landscape, beyond occupation and shell deposition, may have provided the infrastructural bases upon with plantation economies were built and ecological impacts accumulated. Though, more work is needed to explore these historical-ecological entanglements and cumulations.

In addition to infrastructure, the shell- and organic-rich deposits of long-term Indigenous occupations at particular places would have provided prime soil conditions for agricultural production. Indigenous-altered soil conditions were well understood by both Euro-American, enslaved, and Free Black communities. Writing in her autobiography about growing up Gullah-Geechee on Sapelo Island, Cornelia Bailey69 notes that her mother, after moving to the village of Hog Hammock, would complain that “…the soil’s not right, it’s got no shells in it and it’s too shifty, the fields won’t grow the same…” Even modern agricultural practices sometimes include the enrichment of fields with calcium carbonate, specifically crushed or powdered oyster shells, which have been shown to provide a range of benefits (e.g., refs. 70,71,72). Using our estimations from Middle Place and Kenan Field, large Indigenous sites would provide anywhere from 100,000 up to c. 1,000,000 pounds of shell per acre of arable land, while plowing would have crushed and spread oyster deposits and mixed them with an organic-rich matrix of topsoil (Fig. 6D).

The expansive agricultural economies of Sea Island cotton, and to a lesser extent rice, would have intentionally articulated with the ecological legacies of Indigenous occupations, while the preparation and maintenance of fields themselves, along with extensive logging and hydrologic engineering, would have further transformed the upland and lowland marsh ecosystems once again, beginning in the 18th century AD. Not only did the deep-time legacies of Indigenous interaction with environments drive critical changes to local ecosystems, but they also drove further accumulation of ecological impacts by attracting plantation agriculture and conditioning the placement of these Euro-American activities on the landscape, further driving ecological and geomorphological heterogeneity. Indeed, scientific studies highlight the effects of Indigenous shell deposition on such characteristics as soil geochemistry73, and these alterations are reflected in both the historical record and traditional ecological knowledge among Black communities and White plantation owners alike. Examples include Cornelia Bailey’s account of the relationship between growing quality and the presence of shells in the soil on Sapelo Island69 or Theresa Singleton’s74 observation that Indigenous constructed shell landforms on Butler Island were routinely taken advantage of by plantation owners. Additionally, oysters are documented to have been used by historic landowners to pave roads and paths, to create tabby (a type of concrete) for structures, to crush and use as limewash, to crush for fertilizer to be used or sold, and even to supplement the feed of chickens75.

Cumulatively, then, plantation-era activities would have further altered ecosystems, as a new milieu of human behaviors was inscribed geomorphologically and geochemically onto these landscapes (e.g., refs. 76,77). Our historical baselines are the combination of both Indigenous and Euro-American land use over the last 5000 years. Importantly, these ecological legacies aren’t simply additive. Rather, they are entangled with and predicated on one another, creating unique anthropogenic legacies.

Discussion

Primary legacy drivers, the lasting keystone structures produced through deep-time human-environment interactions, play a critical role in the ecosystem, landscape, and geomorphological processes across the Georgia coast today. From topography and the distribution of plant and animal communities to soil chemistry and matrix properties, modern coastal landscapes are far from “natural” in the sense of the word, which neglects human legacies, not just modern actions, as primary ecosystem drivers. As we have shown, some of the islands and landforms of the Georgia coast have never experienced a period in which humans were not present. In fact, for Ossabaw Island, the last permanent resident moved off the island in 2016, which was the first time in its 5000-year history without a permanent human occupation. In contrast to using the state-protected island of Ossabaw as a “natural” referent, it would be more appropriate to use it as an example of what happens when humans, whose institutions were largely responsible for the keystone structures and legacy effects we observe today, are removed from anthropogenically driven and conditioned ecosystems. Importantly, while we focus here on a case study from the southeastern coast of the United States, the kinds of shell deposits we highlight are representative of a global phenomenon of shell use, deposition, and subsequent landscape alteration through long-term anthropogenic land use practices (see ref. 78).

While past human-environmental interactions drive the ecological processes and patterns we might observe today, they have also unintentionally—and perhaps sometimes intentionally on some scales—built geomorphological resiliency into coastal landscapes. While active management of landscapes by both Indigenous and Euro-American inhabitants was critical, whether controlled burning, directing water, draining wetlands, clearing debris and sedimentation from natural and anthropogenic waterways, etc., the cumulative effects of these practices resulted in long-lasting legacies that have persisted beyond those occupations. For the shorelines, tidal creek banks, and marsh islands in particular, shell accumulation is adding critical elevation and a more resilient substrate, both of which are buffering ecological structures against increasing erosional rates, rising sea levels, and saltwater intrusion.

As climatic conditions rapidly change, these anthropogenic legacies continue to shape and drive the structure of ecosystem and landscape heterogeneity and biodiversity. As noted for the Georgia coast in particular, drivers of spatial heterogeneity at the regional scale, especially marsh platform geomorphology, vegetation composition and biomass, and invertebrate patterns are driven primarily by hydrology, elevation, and soil properties79. As we have illustrated, all of these processes are affected locally and at the landscape scale by the legacies of deep-time human-environment interactions. These kinds of discontinuous, fragmented, patchy, or mosaic landscapes drive biodiversity at scale55,80,81,82,83; this heterogeneity drives landscape, species, and ecological resilience in the face of external and internal perturbations84,85,86,87, including the impacts of global climate change.

Regarding conservation, sustainability, assessment of climate change impacts, and the future mediation of these impacts, studying the deep-time human dimensions of modern ecosystem functioning is critical. As Thompson et al.41 point out, many of the shell-modified landscapes, especially the small marsh islands, serve as “keystone” structures, or primary legacy drivers, within broader coastal and estuarine ecosystems by providing critical services, goods, and refuge to a host of species, especially as other landmasses and landforms continue to experience high rates of erosion. In southern Louisiana across the Mississippi River Delta, Mehta and Chamberlain40 similarly point to the role of persistent, human-modified landscapes in generating resilience for past and present coastal communities and ecosystems. In thinking about coastal and marshland landscapes as heterogeneous, we can begin to conceptualize different parts of these landscapes as key nodes in landscape scale networks of disjointed, yet highly connected, habitats. Marsh islands and other anthropogenically modified landforms, especially those characterized by mass shell deposition and Euro-American agriculture, serve as critical, central nodes across these mosaic landscapes by disproportionately increasing and safeguarding biodiversity (sensu88).

As such, these primary legacy drivers, keystone habitats and landforms, and the coupled human-ecosystem processes and histories that produced them, must be incorporated into research and management plans88. At the most basic level, this means incorporating archeological and historical data into both ecological models and projections on the one hand, and in the interpretation of ecological and geomorphological data on the other hand. Simply put, the consideration of anthropogenic land use as a potential independent variable is key if we are going to use these kinds of models to plan for the mediation and resiliency of our local ecosystems. It may be that the actual mechanisms driving a specific ecosystem function or relationship are the result of past anthropogenic alterations that are not routinely considered.

Importantly, the socio-ecological and historical context of these actions also needs to be considered to understand the implications that such processes have moving forward for these ecosystems. In other words, it is not enough simply to document the correlation of these drivers with past activities. To develop plausible scenarios and management plans for contemporary ecosystems we need to understand the full range of how these human institutions functioned to create lasting effects and how they leave environments either open to impacts, more stable to climate shifts, or lend insight into “nature-based solutions” for building resiliency into these ecosystems (see ref. 89).

While the scope of the current study is to demonstrate how coastal landscapes themselves have been altered, added to, and transformed by past anthropogenic activity, using this framework indicates there are many rich, future directions for further research. Most substantive would be formal geomorphological and ecological studies/surveys that explicitly evaluate how anthropogenically transformed landscapes are integrated into, and drive, ecosystem and geomorphological processes. We have cited studies here that provide evidence for unique plant communities growing on shell-rich archeological sites62,63 and have modeled the effects of shell-built topography and elevation on the relationship between tidal fluctuations, sea level, and upland areas of small marsh islands. Future research, by researchers beyond archeology, could and should evaluate these ecosystem and landscape characteristics more closely. For instance, we look forward to ecological work that dives deeper into the effects of anthropogenic land use on plant, animal, and microbial communities, comparing study plots across landforms with known land-use histories. Complementing these studies could be more extensive studies that model differential erosional, hydrological, and geomorphological processes across anthropogenically altered and non-anthropogenically altered landforms.

Finally, the peoples and histories most often erased from historical ecological baselines, as well as from modern conservation efforts and management policy, are Indigenous communities. While we have argued for more attention to be paid to the anthropogenic contexts of historical baselines and sustainable futures, we could refine this to argue that we must pay attention to the Indigenous contexts of past and present ecosystems and to the critical role that Indigenous communities must play in building resiliency and driving sustainability. Unfortunately for places like the Georgia Coast, forced removal, novel diseases, and colonial violence have created critical historical disconnects between modern communities and these ancestral places. The kinds of interdisciplinary research we advocate for here, especially when carried out collaboratively with Indigenous descendant communities, would go a long way not only towards integrating Indigenous knowledge into ecosystem management and theorizing, but towards revitalizing and encouraging Indigenous relationships to the homelands from which they have been long disenfranchised (see refs. 90,91,92,93). Such work would begin to contribute not only to the realm of ecosystem studies, but even further towards discussions and aims of environmental social justice (e.g., refs. 94,95). Globally, and for millennia, Indigenous communities managed landscapes and transformed ecosystems. Rather than being content with building ecological baselines, we must concern ourselves with outlining the contours of socioecological baselines that capture the integrative nature, and key mechanics, of human-ecosystem relationships, especially those that have been retrospectively proven to drive sustainable futures.

Methods

Regional settlement distributions

Regional distributions and counts of Indigenous and historic archeological sites along the Georgia coastal zone were compiled from the Georgia Archeological Site File. All sites in the database with a “Cultural Affiliation” designation that could be categorized as Indigenous or historic Euro-American are included in Fig. 1B. Steps were taken to remove any duplicate entries or sites. All sites listed for Bryan, Camden, Chatham, Glynn, McIntosh, and Liberty Counties were included. All data are housed with the Georgia Archeological Site File at the University of Georgia Laboratory of Archeology. As site locational data are sensitive, site file data is only made available to qualified researchers by way of access permissions.

Ossabaw Island survey

All site survey data for Ossabaw Island was digitized from Pearson’s57 compendium which compiles results of all shoreline and interior-island surveys that have thus far been completed96. As Pearson57 mapped the extent of surface shell distributions, all estimated site boundaries could be digitized to calculate site areas. Beyond site size, the number of archeological components (time periods) represented at each site, indicated by the presence of different kinds of artifacts (namely ceramics), were also recorded. All site sizes for sites included in Pearson’s57 report were included in boxplot summary in Fig. 3C. The four sites located closest to the four plantations, and used to assess size and longevity of occupation for Indigenous sites closest to Euro-American settlements are as follows using Pearson’s57 site designations: OSS 121 (closest to North End Plantation), OSS 24 (closest to Middle Place Plantation), OSS 12 (closest to Buckhead Plantation), and OSS 19 (closest to South End Plantation). All site areas for Ossabaw Island Indigenous sites are included as Supplemental Data 1 archived at Zenodo (https://doi.org/10.5281/zenodo.14883650)97.

Middle Place survey

The calculations for oyster deposition estimation for Middle Place were made using data from excavations conducted in 2023. Twenty-two domestic shell middens were tested via a 50 cm × 50 cm square excavation unit excavated in 10 cm levels. All oyster shells from these excavations were counted and all right oyster valves were used to estimate the minimum number of individuals (MNI) within each excavation unit. Using midden thickness and surface area, we estimated the total percentage of each midden that was excavated and extrapolated our oyster MNI calculations to the rest of the midden. As shell midden thickness is quite standard across all 22 tested middens (c. 30–50 cm), we then used the known surface areas of all 168 mapped shell middens at Middle Place57 to estimate total shell deposition across all known domestic middens. All data for shell quantifications, midden measurements, and calculations for extrapolated estimations are included as Supplemental Data 2 archived at Zenodo97.

Kenan Field survey

Kenan Field, located on Sapelo Island, was surveyed between 2013-2021 via close-interval, systematic shovel test survey98. All shovel tests were excavated in 50 cm × 50 cm squares, 20 m apart across the majority of the site, ultimately covering 66 ha. Tests were excavated in 20 cm levels. At least one profile from each test was drawn and recorded, including the depths and thickness of oyster deposits when encountered. While oyster shell was not retained, all shell was weighed before being reinterred.

To calculate the total estimated weight of deposited shells across the entire Kenan Field site, shell weights from the systematic shovel test survey were extrapolated volumetrically. In total, 21.09 m3 of sediment were excavated and approximately 3875 kg of shell was recovered. The extrapolated total weight of shell deposited at Kenan Field was then calculated from the expected total volume of the site based on the survey’s median depth of cultural deposits (i.e., 60 cm below surface) and the total site extent (i.e., 66 ha). Data and calculations used to make these estimations are included as Supplemental Data 3 archived at Zenodo97. Shovel test data including the presence of shell and its weight are included in Supplemental Data 4 archived at Zenodo97.

Back barrier island (hammock) survey

The hammock survey was led by Victor Thompson and John Turck in 2007. This project comprised full-coverage shovel-test surveys of four marsh islands: Little Sapelo, Mary Hammock, Patterson Island, and Pumpkin Hammock. Across each island, a small test excavation (shovel test) was placed every 20 m, with the objective to reveal the extent of Indigenous occupation and use of each hammock. Each shovel test was 50 cm round and c. 1 m deep. When encountered, all shell was weighed from each unit, and all encountered shell middens were measured for thickness and depth. All material excavated was screened through ¼” screen and materials (e.g., ceramics) were identified and analyzed in the laboratory. All data on artifact presence and shell layer thickness from shovel tests across Litle Sapelo, Mary Hammock, and Patterson Island are included as Supplemental Data 5 archived at Zenodo97.

Hammock elevation and tide inundation modeling

The tidal zone models were developed based on detailed (1 m resolution) coastal topobathymetric elevation data provided by the USGS’ CoNED64, locally adjusted to tidal datums (MLLW, MLW, MHW, MHHW) based on Water Level Stations from the NWLON65. These methods are described in greater detail in Thompson et al. 2024. The hundreds of systematic, standardized units excavated by Thompson et al.38 facilitated the estimation of the depth of anthropogenic sediment across each marsh hammock through interpolation of measured shell deposition layers. By subtracting these sediments from the CoNED DEM and applying the same local tidal adjustments and calculations, we approximate the elevation of areas of each hammock relative to tidal averages. Importantly, this alternate scenario does not account for the different sedimentation processes that would likely occur were anthropogenic sediments to have never been deposited - such analysis is beyond the scope of this paper.

Radiocarbon dating and chronologies of Indigenous occupation

To define the period wherein humans were engaged in landscape and ecological modifications on the Georgia coast a previously assembled regional database of 604 radiocarbon dates99 and newly reported dates from the Middle Place site were placed into three Kernel Density Estimate models100 in OxCal 4.4100 (Supplemental Data 6 archived at Zenodo97). One model was for the entirety of the assembled dates, representing the entirety of the coastal zone of Georgia. Two models were for each of the two case study archeological sites, Middle Place and Kenan Field. Each model was a KDE_Model, wherein a kernel density estimate is produced to “smooth out” the influence of the radiocarbon calibration curve as a preferred method over a simple summed probability density. In OxCal 4.4, the KDE_Model bandwidth size selection is the product of Bayesian inference where the prior bandwidth for each observation point is the composite KDE distribution of the other dates included in the model100,101,102). All dates from archeological contexts were included in the models, with the exception of dates coming from human ancestors and bulk sediment dates. Dates were calibrated via IntCal20 and Marine20, with an updated ΔR of −277 ± 21 based on data provided in Thomas103. OxCal code for replicating these models can be found as Supplemental Code 1 archived at Zenodo97.

Ossabaw Island landscape modification survey

LiDAR data from NOAA was used to identify all potential landscape modifications across Ossabaw Island. LiDAR allows for the identification of high-resolution topographical differences across large areas. The island-wide data was systematically surveyed by enhancing topographical contrast using ArcGIS Pro Dynamic Range visualizations. All linear, grid-like, or unnatural angles of landscape features (e.g., right angles) were digitized. Known landscape features (e.g., known canals, ditches, and roads), as well as information from historical documentation, were used to identify potential unknown landscape modifications.