By 13,000 BP human populations were present across North America, but the exact date of arrival to the continent, especially areas south of the continental ice sheets, remains unclear. Here we examine patterns in the stratigraphic integrity of early North American sites to gain insight into the timing of first colonization. We begin by modeling stratigraphic mixing of multicomponent archaeological sites to identify signatures of stratigraphic integrity in vertical artifact distributions. From those simulations, we develop a statistic we call the Apparent Stratigraphic Integrity Index (ASI), which we apply to pre- and post-13,000 BP archaeological sites north and south of the continental ice sheets. We find that multiple early Beringian sites dating between 13,000 and 14,200 BP show excellent stratigraphic integrity. Clear signs of discrete and minimally disturbed archaeological components do not appear south of the ice sheets until the Clovis period. These results provide support for a relatively late date of human arrival to the Americas.
Citation: Surovell TA, Allaun SA, Crass BA, Gingerich JAM, Graf KE, Holmes CE, et al. (2022) Late date of human arrival to North America: Continental scale differences in stratigraphic integrity of pre-13,000 BP archaeological sites. PLoS ONE 17(4): e0264092. https://doi.org/10.1371/journal.pone.0264092
Editor: Raven Garvey, University of Michigan, UNITED STATES
Received: October 14, 2021; Accepted: January 26, 2022; Published: April 20, 2022
Copyright: © 2022 Surovell et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting information files.
Funding: BTW, Adelphi University (https://www.adelphi.edu/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript CEH, PLR-9112174, National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript CEH, PLR-0540235, National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript JAMG, Pennsylvania Historic and Museum Commission (https://www.phmc.pa.gov/Preservation/Grants-Funding/Pages/default.aspx), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript JAMG, 1556318, National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript MLL & MK, BCS-91095914, National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript RLK, 05-14863 National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript RLK, 07-10868 National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript RLK, 19-39019 National Science foundation (https://www.nsf.gov/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript RLK, George C. Frison Institute for Archaeology and Anthropology (https://www.uwyo.edu/anthropology/frison-institute/), Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript MK, University of Wyoming (https://www.uwyo.edu/) Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript MK & MLL, Wyoming Archaeological Foundation (http://www.wyomingarchaeology.org/wyoming-archaeological-foundation.html) Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript MK & MLL, 11-14S, Wyoming Cultural Trust Fund (https://culturaltrust.wyo.gov/) Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript MK & MLL, Roy J. Shlemon Center for Quaternary Studies (http://www.uwyo.edu/quaternary/) Sponsor role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
No consensus has been reached among archaeologists about the date of initial human arrival to the Americas, but all agree that human populations were distributed across the North American continent by 13,000 BP, as evidenced by fluted Clovis projectile points and associated artifacts from surface and buried contexts [1–5]. Before 13,000 BP, the clearest evidence for human presence comes from eastern Beringia, the unglaciated portions of Alaska and the Yukon Territory [e.g, 6–13]. Evidence for humans south of the Laurentide and Cordilleran ice sheets prior to 13,000 BP remains sparse and controversial despite more than a century of fieldwork in the region [e.g., 14–26]. Archaeologists widely accept the critical appraisal of artifacts from firmly dated pre-13,000 BP contexts as a valid means of evaluating claims for early human presence in the Americas [27–32]. When artifacts are found in buried contexts pre-dating 13,000 BP, there are at least two possible explanations for their occurrence—humans were present before 13,000 BP, or humans were not present but younger artifacts have intruded into older sedimentary contexts. Many debates regarding the peopling of the Americas can be reduced to distinguishing between these two possibilities.
Unambiguous association between artifacts and the strata from which they are derived has been a hallmark of establishing the antiquity of humans in the Americas since the earliest days of American archaeology [27,31], yet archaeological tools for evaluating these associations remain crude. Here, we develop a simple means of evaluating the stratigraphic integrity of association between buried archaeological occupations and dated strata and then evaluate the early North American archaeological record using existing provenience data to distinguish between an early or late arrival. The classic Clovis-first model posits that humans arrived in eastern Beringia sometime prior to the Clovis period and breached the continental ice sheets around or just before 13,000 BP [33,34]. If so, the oldest evidence for intact occupations should first occur in eastern Beringia and then appear south of the ice sheets in Clovis times. If there was a widespread pre-13,000 BP population south of the ice sheets, there should be clear evidence for it in the form of unambiguous occupations from archaeological contexts reliably pre-dating 13,000 BP.
Modeling stratigraphic mixing
To develop expected differences between vertically mixed and unmixed multicomponent sites, we created a simulation for R v. 4.1.1  that combines a depositional history, an occupational history, and a disturbance model (S1 File). For our simulations, we hold depositional and occupational history constant, but our model can be customized to accommodate any such history. In our simulation, sediments accumulate over 18,000 years at a constant rate of 0.1 mm per year. Seven occupations take place, each spaced 2,000 years apart with the first occurring at 12,500 BP and the last at 500 BP. During each occupation, 500 artifacts are deposited on the ground surface at the time, and each artifact is assigned a random arbitrary horizontal provenience. Each year, all artifacts are moved randomly up or down a fixed distance determined by a maximum dispersal rate and depth. If any artifacts breach the ground surface, they are placed on the surface that exists at the time. It is assumed that artifacts closer to the current ground surface are more susceptible to post-depositional disturbance and displacement and therefore should move more than those at greater depth. Following from that assumption, we modeled actual artifact dispersal rates (r) as a logistic function of depth (d) and maximum dispersal rate (rmax): (1)
This function is shown graphically in S1 Fig. If the maximum dispersal rate is set to 1 mm per year, for example, an artifact 5 cm beneath the ground surface will move 0.97 mm each year up or down. An artifact 50 cm beneath the surface will move 0.56 mm per year. An artifact buried 1 m in depth will move 0.19 mm, and one at 2 m of depth will move around 0.01 mm per year. We do not mean to imply that this function can be used to describe all archaeological cases, but it is intended to be a reasonable approximation of the relative degree to which artifacts at different depths are subject to post-depositional disturbance processes originating from the ground surface such as bio- and cryoturbation.
In Fig 1, we illustrate four outputs of the simulation while varying maximum dispersal rate (rmax). Specifically, we show resulting backplots and vertical density histograms. With rmax set to a low rate of 0.1 mm per year, each occupation is vertically discrete with culturally sterile zones separating occupations. Artifacts have not moved far beyond the surfaces on which they were originally deposited. When viewed as a vertical density histogram, the minimally disturbed case exhibits a very peaked, multimodal distribution with clear stratigraphic integrity (Fig 1a and 1b).
Horizontal black lines on backplots show the original stratigraphic position of occupations before post-depositional disturbance. (a&b) rmax = 0.1 mm/yr. (c&d) rmax = 0.5 mm/yr. (e&f) rmax = 1.0 mm/yr. (g&h) rmax = 3.0 mm/yr.
When rmax is set to a much higher rate (e.g., 3 mm per year), individual occupations are not stratigraphically discrete and become visibly mixed (Fig 1g and 1h). As others have found, with significant mixing archaeological distributions become vertically homogenized [36,37], although artifacts from different occupations generally maintain their stratigraphic order. In this situation, it would be challenging to assign individual artifacts to specific occupations. It would also be difficult using vertical distributions alone to determine how many occupations occurred. In the highly mixed simulation, artifacts can be found well above and below the surfaces on which they were deposited. At depth, artifact densities slowly decline to zero. Near the surface, a sharp mode in artifact count is evident because the most recent occupation has had the least time to be dispersed leaving somewhat high densities at the uppermost levels. When viewed as a vertical density histogram, the overall distribution shows no gaps, and little multimodality. Overall, the vertical density distribution has a down-skewed appearance. The two intermediate cases show intermediate attributes of the minimally and maximally mixed cases, with varying degrees of multimodality (Fig 1c–1f).
To illustrate the effect of increased mixing on the apparent age of the deepest artifacts in this system, we varied rmax from 0.1 to 5 mm per year and ran ten iterations of each model to determine how the apparent age of the deepest artifact increases with greater disturbance (Fig 2). Apparent age in this context refers to the age of the sediments from which the deepest artifact was recovered. With increased mixing, artifacts move progressively downward into older sediments producing the false impression of the presence of humans on a site hundreds to thousands of years prior to the initial occupation.
Dashed line shows the age of the initial occupation at 12,500 BP.
ASI: Apparent Stratigraphic Integrity Index
In our simulated multicomponent archaeological cases, the degree of vertical mixing is clearly reflected by changes in vertical artifact distributions. We developed a statistic that we call the Apparent Stratigraphic Integrity Index (ASI) that captures this variability. The use of the term “apparent” is because frequently reoccupied but minimally disturbed sites could theoretically exhibit similar stratigraphic artifact distributions to highly disturbed sites (S2 File; S2 Fig). Likewise, the vertical artifact distribution from a highly disturbed site could exhibit properties typical of a minimally disturbed locality. To determine with greater certainty the extent to which artifacts have moved vertically in an actual archaeological site would require additional kinds of data (e.g., systematic refitting). All things being equal, however, we expect minimally disturbed multicomponent sites to show dramatic changes in artifact density from stratum to stratum (Fig 1a and 1b) and heavily disturbed sites to show gradual changes in density from level to level (Fig 1g and 1h).
The ASI is based on changes in the artifact frequency between adjacent excavation levels. In general statistical nomenclature, this statistic could be called the relative mean absolute successive difference and is similar to the unstandardized mean absolute successive difference (MASD) . Before calculating the ASI, consecutive levels with zero artifacts are collapsed to single zero values. After that transformation, the calculation of ASI uses the mean absolute change in artifact frequency (f) between all adjacent levels divided by two times the mean artifact frequency for all levels: (2)
For most cases, the ASI varies between zero and one with higher values implying greater stratigraphic integrity. Consecutive sterile levels are reduced to a single zero value because if left in, they would result in a lower ASI value when they should theoretically have little bearing on the question of stratigraphic integrity. To calculate an ASI for a site, an excavation unit, or any systematically excavated area, the function requires an array of artifact counts sorted by excavated level (S1 File). One disadvantage of the ASI as we have constructed it is that its value can vary depending upon an arbitrary choice of thickness of excavation levels or elevation bins. For our application, however, due to the nature of the available data, we use artifact counts from standardized 5 cm levels. This choice is also justified because the use of 5 cm levels is fairly standard practice in excavations of hunter-gatherer archaeological sites. It is important to note that comparison of ASI values among sites is best done using standardized level thicknesses, since coarse units of inquiry (i.e., thicker levels) will exhibit lower ASI values than fine ones.
In Fig 3, we show ASIs for six simulated vertical artifact distributions. Generally speaking, highly homogenized vertical distributions have low ASI values, but those with gaps representing sterile or low density strata separating occupations exhibit high ASIs. In the examples shown, relatively intact distributions with little mixing have ASIs from 0.4 to 1.0. Highly mixed sites tend to show values less than 0.3.
In addition to sensitivity to bin width (or the width of excavated levels), the ASI is also affected by small sample size (S3 Fig). For several levels of disturbance, we randomly sampled between ten and 3,500 artifacts from simulated sites and calculated the resulting ASIs (S2 File). This exercise showed that low density sites with very few artifacts will exhibit inflated ASI values. Sites that show less than a mean of approximately 30 artifacts per level will appear more intact than they actually are when measured by ASI. This problem reinforces the idea that it is difficult to determine the stratigraphic integrity of any archaeological site with a small assemblage. On the opposite end of the spectrum, ASI values should be most reliable for sites with high artifact densities.
The apparent stratigraphic integrity of paleoindian sites
We compiled vertical density data from a series of North American multicomponent sites that contain Paleoindian occupations (Fig 4, S2 File). No permits were required for the described study, which complied with all relevant regulations. Artifact density data are either first reported here or were taken from published literature. The sample includes eight sites with components argued to pre-date 13,000 BP, five from eastern Beringia and three from south of the ice sheets. Beringian pre-13,000 BP sites include Broken Mammoth [7,39,40], Dry Creek [10,12], Holzman South [9,13], Swan Point [6,8], and Owl Ridge [41,42]. All Beringian sites occur within a 200 km reach of the Tanana River valley and its tributaries in eastern Alaska and are buried in Pleistocene loess. The oldest components in all of these sites are argued to date between ca. 13,000 and 14,200 BP. Pre-13,000 BP sites south of the ice sheets include Cooper’s Ferry [19,43,44], Debra L. Friedkin [18,45], and Gault [46,47]. The oldest occupations at these sites are hypothesized to date to at least 15,500 to 18,500 BP [18,19,45,47]. We considered the inclusion of other potential pre-Clovis sites [e.g., 16,20,21,48], but appropriate data were not available, the site was insufficiently buried, or artifact counts were too low. Four multicomponent sites with Paleoindian occupations post-dating 13,000 BP are also part of the sample. They include Alm Shelter , Helen Lookingbill , Hell Gap, Locality I [51–53], and Shawnee-Minisink [54–56]. Importantly, Shawnee-Minisink has a Clovis component, and due to the nature of the excavations at Shawnee (S2 File), we only have vertical artifact distribution data for the strata surrounding and including the Clovis component.
(1) Dry Creek and Owl Ridge, (2) Holzman South, Swan Point, and Broken Mammoth, (3) Cooper’s Ferry, (4) Alm Shelter; (5) Helen Lookingbill; (6) Hell Gap; (7) Gault and Debra L. Friedkin, (8) Shawnee-Minisink. Light blue polygons show the estimated extent of the continental ice sheets at 14,900 BP from Dalton et al. . Digital elevation model of North America is from the USGS Global Multi-resolution Terrain Elevation Dataset .
To generate vertical density profiles, for all but two sites (Gault and Friedkin), we binned the elevations of piece-plotted artifacts into 5 cm levels after adjusting for stratigraphic tilt. When using this method, we omitted artifacts recovered from screens due to lower precision provenience. To account for sloping stratigraphy when present, we isolated portions of sites where at least one planar artifact concentration could be discerned within a three-dimensional backplot and fit a plane through it using multiple linear regression (S4 Fig). We then transformed absolute artifact elevations to relative elevations above or below that plane. If no stratigraphic slope could be discerned, we binned artifacts into 5 cm levels using elevations. We do not have piece-plot data for Gault and Debra L. Friedkin, so we compiled vertical density profiles from previously published 5 cm level counts [18,47] and could not account for stratigraphic tilt.
All five eastern Beringian sites show multiple stratigraphically discrete archaeological components (Fig 5). Backplots of these Alaskan sites show easily defined planar archaeological components separated by low density or sterile stratigraphic units. Vertical density histograms are unambiguously multimodal. The pre-13,000 BP sites south of the continental ice sheets display very different patterns. Cooper’s Ferry and Friedkin exhibit similar density profiles to each other, even though artifact densities between the two sites differ by three orders of magnitude. Both sites exhibit traits expected of vertically mixed deposits with smoothed and homogenized distributions slowly tailing off with depth. The profile from Gault, however, shares traits with both the Beringian and southern pre-Clovis sites. It is clearly multimodal, but it lacks clear gaps or sterile zones between archaeological components. To what extent the Gault profile is affected by sloping stratigraphy, we do not know.
We calculated ASI values for all pre- and post-13,000 BP sites (Table 1; Fig 6). There are clear differences between the pre-13,000 BP sites north and south of the continental glaciers. All five Beringian sites show relatively high levels of apparent stratigraphic integrity with ASI values ranging from 0.367 at Broken Mammoth to 0.546 at Owl Ridge. The Owl Ridge ASI is likely inflated due to relatively low artifact densities, but its deepest component is stratigraphically discrete. Still, all early Beringian sites cluster together. Early sites south of the ice sheets also form a cluster, but on the low end of the ASI scale. The Debra L. Friedkin site has the lowest ASI of any observed site at 0.045. The Gault site shows the highest apparent stratigraphic integrity of these three at 0.219. Cooper’s Ferry has an intermediate value, which could be inflated due to low artifact counts. Differences between the mean ASIs of early northern and southern sites are significant (Welch t-test, t = 5.359, df = 3.303, two-tailed p = 0.01). While the post-13,000 BP sites south of the ice sheets show extremely variable ASI from 0.343 (Helen Lookingbill) to 0.676 (Shawnee Minisink) (S5–S8 Figs), they are higher than all supposed pre-Clovis sites south of the ice sheets. They are also not of demonstrably different magnitude than the pre-13,000 BP sites north of the ice sheets (Welch t-test, t = 0.418, df = 3.87, two-tailed p = 0.698), but they are significantly greater than the southern pre-13,000 BP sites (Welch t-test, t = 3.096, df = 4.863, two-tailed p = 0.028).
Integrity of association between artifacts and the dated stratigraphic contexts from which they derive is a baseline standard for establishing the antiquity of humans in the Americas. Our systematic evaluation of some of the earliest buried archaeological sites in North America supports a relatively late arrival of humans to areas south of the continental ice sheets. In our sample, the oldest sites demonstrating relatively unmixed and discrete occupations prior to 13,000 BP all occur in eastern Beringia, which is noteworthy considering these sites are located in an area commonly affected by cryoturbation [12,42]. The archaeologists working at these Alaskan sites were aware of and can identify the problems that bioturbation, solifluction, and other phenomena can bring to the excavation, data recovery, and its analysis and interpretation; excavators frequently consulted with pedologists and Quaternary geologists throughout. The Alaskan sites analyzed here contained unambiguous minimally disturbed archaeological components. For example, at Broken Mammoth, Swan Point, and several other sites the stratigraphy was straightforward and the radiocarbon dates were in correct order . At Dry Creek  and Broken Mammoth [60,61], the oldest cultural component occurred beneath a thin sand layer that physically separated that component from the younger components. At the Holzman site, up to 20 cm of culturally sterile loess deposits separate components 4 from component 5a and sterile bands of sand deposits separate components 5a from 5b [9,13].
All of the pre-13,000 BP sites south of the continental ice sheets display patterns of significant mixing. South of the ice sheets, the first evidence for a discrete occupation is from the Clovis period at the Shawnee-Minisink site, which given its latitude and age, could have also experienced cryoturbation. Stratigraphically discrete occupations regularly occur in the archaeological record from the Clovis period onward. Furthermore, Shawnee-Minisink is one of several Clovis sites to exhibit a stratigraphically discrete cultural occupation [e.g., 62–69]; this is a trait clearly associated with the Clovis complex and not associated with any sites pre-dating Clovis in the New World, except those in Beringia. To some extent, data availability for the oldest purported sites in the Americas undermined this study by limiting sample size of site south of the Canadian ice sheets. Sites claimed to be older than 13,000 BP are few and data supporting their status as sites have been poorly disseminated. Given the status of available data regarding these sites, we must question whether there are any sites in the Americas south of the ice sheets that exhibit an unambiguous and stratigraphically discrete cultural occupation with sufficient numbers of artifacts of clear human manufacture.
One site that might be argued to meet those criteria is Monte Verde, Chile [24,70,71], but many researchers question whether most objects from the MV-II Pleistocene peat bog are truly artifacts, especially the organic items. There are only about six items that satisfy most critics as undoubtedly of human manufacture [27,72]. Other sites, like Page-Ladson and Paisley Caves have very small numbers of artifacts, such that evaluating stratigraphic integrity is challenging [73,74]. Even where large numbers of questionable artifacts are argued to be present as at Chiquihuite Cave or Pedra Furada, they do not occur in discrete identifiable archaeological components [14,75].
Another site that might preserve a stratigraphically discrete occupation below Clovis is Cactus Hill, Virginia, which is argued to have an archaeologically sterile zone separating the Clovis and pre-Clovis levels [20,76,77], but data demonstrating stratigraphic separation have not been published. Furthermore, no archaeologically sterile levels separating the Clovis and pre-Clovis components are evident in the vertical artifact distribution data that have been published . Because excavated levels at Cactus Hill were not dug in uniform thicknesses and no provenience data for piece-plotted artifacts are available, it is difficult to compare ASIs from Cactus Hill to other sites. Nonetheless, using vertical artifact distribution data from Areas A/B and B of the site , it is possible to calculate ASIs, and like other pre-Clovis sites south of the ice sheets, Cactus Hill does not preserve discrete archaeological components, and ASIs generally fall in the range of 0.156 to 0.309 (S9 Fig). Furthermore, artifact counts gradually decline with depth (S9 Fig). Both findings suggest that Cactus Hill has likely experienced some mixing, and the excavators have noted that “downdrift from upper levels” has affected lower levels of the site .
All three of the pre-Clovis sites outside of Alaska included in this study have been argued to be stratigraphically intact, and questions have been raised about two of them [18,19,45,47,78,79]. Our analysis supports the hypothesis that mixing has affected these sites, and furthermore, we note that it would be a simple matter to establish whether that is in fact the case. The ideal way to do so would be systematic refitting of chipped stone artifacts with the goal of searching for refits between components to determine to what extent items are moving vertically in these sites. There is a long tradition of such studies in archaeology, including Paleoindian archaeology both north and south of the ice sheets [39,42,46,80–85]. If little vertical distance separates refitting and conjoining artifacts, it might be shown that artifacts have not moved significantly. Notably, a stratigraphically restricted study of Clovis technology from a different excavation area at the Gault site than the one analyzed herein found that artifacts were moving vertical distances of at least 19 cm from a sample of only 27 refits . Examination of stratigraphic differences in artifact size, long axis dip/inclination, lithic raw materials, technology, or burning of artifacts could provide independent evidence for stratigraphic integrity. Of great importance is the presence of cultural features in sealed stratigraphic units associated with unambiguous occupation surfaces marked by sufficient artifact counts to allow their identification. In short, there is no evidence for a stratigraphically discrete archaeological occupation with large numbers of artifacts before 13,000 BP in the New World except those in Beringia.
These observations provide support for the hypothesis that the first arrival of humans to areas south of the Laurentide and Cordilleran glaciers occurred near in time to 13,000 BP. It is possible humans colonized the New World thousands of years before 13,000 BP, but if they did, they should have produced stratigraphically discrete occupation surfaces, some of which would be expected to have large numbers of artifacts. That they did so in Beringia but failed to do so south of the continental glaciers suggests that either there was something fundamentally different about pre-Clovis human behavior and/or geomorphology south of the ice sheets or that the evidence indicating the presence of humans south of the ice sheets has been misinterpreted. At a minimum, it shows that when stratigraphically discrete occupations are not present, additional studies must be performed to demonstrate that stratigraphic integrity of association between artifacts and dated strata exists.
The oldest evidence for archaeological sites in the New World with large numbers of artifacts occurring in discrete and minimally disturbed stratigraphic contexts occur in eastern Beringia between 13,000 and 14,200 BP. South of the ice sheets, the oldest such sites occur in association with the Clovis complex. If humans managed to breach the continental ice sheets significantly before 13,000 BP, there should be clear evidence for it in the form of at least some stratigraphically discrete archaeological components with a relatively high artifact count. So far, no such evidence exists. These findings support the hypothesis that the first human arrival to the New World occurred by at least 14,200 BP in Beringia and by approximately 13,000 BP in the temperate latitudes of North America. Strong evidence for human presence before those dates has yet to be identified in the archaeological record.
S1 Fig. Dispersal rate function used in mixing simulation.
Shown for rmax = 1 mm/yr.
S2 Fig. Results of a simulation illustrating a case of low ASI and high stratigraphic integrity illustrated as a backplot (left) and vertical density histogram (right).
In this model, occupation intensity gradually increases over time and no vertical mixing occurs. In the backplot, artifacts are colored by occupation.
S3 Fig. Average number of artifacts per level vs. ASI for simulated multicomponent site while varying the maximum rate of artifact dispersal (rmax).
S4 Fig. Example of method used to correct artifact elevations for stratigraphic tilt.
Three-dimensional scatterplot of artifacts from Holzman South with a plane fit by multiple linear regression to component 5a.
S5 Fig. Backplot and vertical density histogram for Locality I of the Hell Gap site, Wyoming for N 1481 to 1482 m and E 1294.5 to 1296.1 m.
S6 Fig. Backplot and vertical density histogram Alm Shelter, Wyoming for N98 to 99 and E 99 to 100.
S7 Fig. Backplot and vertical density histogram for the Helen Lookingbill site, Wyoming for N 1014 to 1015 m and E 974 to 976 m.
S8 Fig. Backplot and vertical density histogram of the Clovis component from the Shawnee-Minisink site, Pennsylvania for N 154.89 4 to 158.374 m and E 152.37 to 155.39 m.
S9 Fig. Artifact counts by level for four areas of the Cactus Hill site.
a. Block B, Unit 16; b. Block B, Unit 17; c. W165 N100; d. W115 N70. Excavation levels at Cactus Hill were not dug in uniform thicknesses.
S1 Table. Chipped stone artifact counts by 5 cm level for unit N98 E99 of Alm Shelter.
S2 Table. Artifact and bone counts by 5 cm level for N 96 to 100 m and E 110 to 199 m of the Broken Mammoth site.
Relative elevation in the distance above or below a plane fit through the LP component.
S3 Table. Artifact and bone counts by 5 cm level for stratum LU3 of Area A of the Cooper’s Ferry site.
S4 Table. Debitage and tool count by 5 cm level for Block A of the Debra Friedkin site.
S5 Table. Chipped stone artifact counts by 5 cm level for N 14 to 16 m and E 21 to 23 m of the Dry Creek site.
Relative elevation in the distance above or below a plane fit through Component 2.
S6 Table. Counts of flakes by 5 cm level from Area 15 of the Gault site.
S7 Table. Chipped stone artifact and bone counts by 5 cm level for N 1014 to 1015 m and E 974 to 976 m from the Helen Lookingbill site.
Relative elevations are the distance above or below a plane fit through all artifacts between elevations 98.7 and 99.1 m.
S8 Table. Counts of chipped stone artifacts, ocher, and bone by 5 cm level from N 1481 to 1482 m and E 1294.5 and 1296.1 m of Locality I of the Hell Gap site.
S9 Table. Artifact and bone counts by 5 cm level for N 185 to 192 m and E 502 to 506 m from the Holzman South site.
Relative elevations are the distance above or below a plane fit through artifacts from Component 5a.
S10 Table. Artifact counts by 5 cm level for N 87 to 89 m and E 110 to 111 m from the Owl Ridge site.
S11 Table. Artifact counts by 5 cm level for N 158.9 to 158.4 m and E 152.4 to 155.4 m from the Clovis component of the Shawnee-Minisink site.
S12 Table. Counts of artifacts and bone 5 cm level from N 90 to 95 m and E 98 and 99 m from the Swan Point site.
Relative elevations are distances above and below a plane fit through all artifacts deeper than 1.18 m below datum.
S1 File. R code used in the study including function for calculation of ASI and simulation code.
Kelly Graf was very generous to contribute data from Owl Ridge and Dry Creek. We are grateful to Jessi Halligan and two anonymous reviewers for their thoughtful reviews. We would like to highlight our appreciation of the efforts of one reviewer who was especially helpful in thinking about the use and significance of the Apparent Stratigraphic Integrity Index.
- 1. Anderson DG, Miller DS, Yerka SJ, Gillam JC, Johanson EN, Anderson DT, et al. PIDBA (Paleoindian Database of the Americas) 2010: current status and findings. Archaeology of Eastern North America. 2010; 63–89.
- 2. Anderson DG, Echeverry D, Miller DS, White AA, Yerka SJ, Kansa E, et al. Paleoindian settlement in the Southeastern United States: the role of large databases. In: Thulman D, Garrison I, editors. Early Floridians: new directions in the search for and interpretation of Florida’s earliest inhabitants. Gainesville: University Press of Florida; 2019. pp. 241–275.
- 3. Haynes G. The Early Settlement of North America: The Clovis Era. Cambridge: Cambridge University Press; 2002.
- 4. Waters MR, Stafford TW Jr. Redefining the age of Clovis: Implications for the peopling of the Americas. Science. 2007;315: 1122–1126. pmid:17322060
- 5. Waters MR, Stafford TW, Carlson DL. The age of Clovis—13,050 to 12,750 cal yr BP. Science Advances. 2020;6: eaaz0455. pmid:33087355
- 6. Holmes CE, VanderHoek R, Dilley TE. Swan Point. In: West FH, editor. American Beginnings. Chicago: University of Chicago Press; 1996. pp. 319–322.
- 7. Holmes CE. Broken Mammoth. American Beginnings: The Prehistory and Paleoecology of Beringia. Chicago: University of Chicago Press; 1996. pp. 312–318.
- 8. Holmes CE. Tanana River Valley Archaeology circa 14,000 to 9000 B.P. Arctic Anthropology. 2001;38: 154–170.
- 9. Wygal BT, Krasinski KE, Holmes CE, Crass BA. Holzman South: A Late Pleistocene Archaeological Site along Shaw Creek, Tanana Valley, Interior Alaska. PaleoAmerica. 2018;4: 90–93.
- 10. Powers WR, Guthrie RD, Hoffecker JF. Dry Creek: Archaeology and paleoecology of a late Pleistocene Alaskan hunting camp. Goebel T, editor. College Station: Texas A&M University Press; 2017.
- 11. Graf KE, Buvit I. Human dispersal from Siberia to Beringia: Assessing a Beringian standstill in light of the archaeological evidence. Current Anthropology. 2017;58: S583–S603.
- 12. Graf KE, DiPietro LM, Krasinski KE, Gore AK, Smith HL, Culleton BJ, et al. Dry Creek Revisited: New Excavations, Radiocarbon Dates, and Site Formation Inform on the Peopling of Eastern Beringia. Am antiq. 2015;80: 671–694.
- 13. Wygal BT, Krasinski KE, Holmes CE, Crass BA, Smith KM. Mammoth Ivory Rods in Eastern Beringia: Earliest in North America. American Antiquity. 2021; 1–21.
- 14. Ardelean CF, Becerra-Valdivia L, Pedersen MW, Schwenninger J-L, Oviatt CG, Macías-Quintero JI, et al. Evidence of human occupation in Mexico around the Last Glacial Maximum. Nature. 2020;584: 87–92. pmid:32699412
- 15. Bennett MR, Bustos D, Pigati JS, Springer KB, Urban TM, Holliday VT, et al. Evidence of humans in North America during the Last Glacial Maximum. Science. 2021;373: 1528–1531. pmid:34554787
- 16. Halligan JJ, Waters MR, Perrotti A, Owens IJ, Feinberg JM, Bourne MD, et al. Pre-Clovis occupation 14,550 years ago at the Page-Ladson site, Florida, and the peopling of the Americas. Science Advances. 2016;2.
- 17. Holen SR, Deméré TA, Fisher DC, Fullagar R, Paces JB, Jefferson GT, et al. A 130,000-year-old archaeological site in southern California, USA. Nature. 2017;544: 479–483. pmid:28447646
- 18. Waters MR, Keene JL, Forman SL, Prewitt ER, Carlson DL, Wiederhold JE. Pre-Clovis projectile points at the Debra L. Friedkin site, Texas- Implications for the Late Pleistocene peopling of the Americas. Science advances. 2018;4: eaat4505. pmid:30397643
- 19. Davis LG, Madsen DB, Becerra-Valdivia L, Higham T, Sisson DA, Skinner SM, et al. Late Upper Paleolithic occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago. Science. 2019;365: 891–897. pmid:31467216
- 20. McAvoy JM, McAvoy LD. Archaeological Investigations of Site 44SX202, Cactus Hill, Sussex County Virginia. Richmond, Virginia: Research Report Series No. 8, Virginia Department of Historic Resources; 1997.
- 21. Adovasio JM, Donahue J, Stuckenrath R. The Meadowcroft Rockshelter radiocarbon chronology 1975–1990. American Antiquity. 1990;55: 348–354.
- 22. Waters MR, Stafford TW, Kooyman B, Hills LV. Late Pleistocene horse and camel hunting at the southern margin of the ice-free corridor: Reassessing the age of Wally’s Beach, Canada. Proceedings of the National Academy of Sciences. 2015;112: 4263–4267. pmid:25831543
- 23. Gustafson CE, Gilbow D, Daugherty RD. The Manis mastodon: Early man on the Olympic Peninsula. Canadian Journal of Archaeology. 1979;3: 157–164.
- 24. Dillehay T. Monte Verde: A Late Pleistocene Settlement in Chile. Volume 2: The Archaeological Context and Interpretation. Washington D.C.: Smithsonian; 1997.
- 25. Sistiaga A, Berna F, Laursen R, Goldberg P. Steroidal biomarker analysis of a 14,000 years old putative human coprolite from Paisley Cave, Oregon. Journal of Archaeological Science. 2014;41: 813–817.
- 26. Waters MR, Stafford TW Jr, McDonald GH, Gustafson C, Rasmussen M, Cappellini E, et al. Pre-Clovis mastodon hunting 13,800 years ago at the Manis Site, Washington. Science. 2011;334: 351–353. pmid:22021854
- 27. Meltzer DJ. The Great Paleolithic War. Chicago: University of Chicago Press; 2015.
- 28. Haynes CV Jr. The earliest Americans. Science. 1969;166: 709–715. pmid:17776753
- 29. Chamberlin TC. The criteria requisite for the reference of relics to a glacial age. The Journal of Geology. 1903;11: 64–85.
- 30. Holmes WH. Stone Implements of the Potomac-Chesapeake Tidewater Province. 15th Annual Report of the Bureau of American Ethnology to the Secretary of the Smithsonian Institution, 1893–94. Washington, D.C.: Government Printing Office; 1897. pp. 13–152.
- 31. Lyman RL. A historical sketch on the concepts of archaeological association, context, and provenience. Journal of Archaeological Method and Theory. 2012;19: 207–240.
- 32. Hrdlicka A. Early Man in South America. Washington D. C.: Bureau of American Ethnology Bulletin No. 52, Smithsonian Institution; 1912.
- 33. Martin PS. The discovery of America. Science. 1973;179: 969–974. pmid:17842155
- 34. Haynes CV Jr. Clovis Progenitors: From Swan Point, Alaska to Anzick site, Montana in less than a decade? In: Robertson EC, Seibert JD, Fernandez DC, Zender MV, editors. Space and Spatial Analysis in Archaeology. Calgary: University of Calgary Press; 2006. pp. 253–267.
- 35. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2021. URL https://www.R-project.org/.
- 36. Brantingham PJ, Surovell TA, Waguespack NM. Modeling post-depositional mixing of archaeological deposits. Journal of Anthropological Archaeology. 2007;26: 517–540.
- 37. Perreault C. The Quality of the Archaeological Record. Chicago: University of Chicago Press; 2019.
- 38. Ebner-Priemer UW, Kuo J, Kleindienst N, Welch SS, Reisch T, Reinhard I, et al. State affective instability in borderline personality disorder assessed by ambulatory monitoring. Psychological medicine. 2007;37: 961–970. pmid:17202005
- 39. Krasinski KE. Intrasite spatial analysis of Late Pleistocene/Early Holocene archaeological material from the Broken Mammoth site. M.A. Thesis, Department of Anthropology, University of Alaska Anchorage. 2005. https://www.proquest.com/docview/305360958/abstract/1D68804CD7854A75PQ/1.
- 40. Krasinski KE, Yesner DR. Late Pleistocene/early Holocene site structure in Beringia: a case study from the Broken Mammoth site, interior Alaska. Alaska Journal of Anthropology. 2008;6: 27–41.
- 41. Gore AK, Graf KE. Technology and Human Response to Environmental Change at the Pleistocene-Holocene Boundary in Eastern Beringia: A View from Owl Ridge, Central Alaska. In: Robinson E, Sellet F, editors. Lithic Technological Organization and Paleoenvironmental Change. Springer; 2018. pp. 203–234.
- 42. Graf KE, Gore AK, Melton JA, Marks T, DiPietro L, Goebel T, et al. Recent excavations at Owl Ridge, interior Alaska: Site stratigraphy, chronology, and site formation and implications for late Pleistocene archaeology and peopling of eastern Beringia. Geoarchaeology. 2020;35: 3–26.
- 43. Davis LG, Schweger CE. Geoarchaeological context of late Pleistocene and early Holocene occupation at the Cooper’s Ferry site, western Idaho, USA. Geoarchaeology. 2004;19: 685–704.
- 44. Davis LG, Nyers AJ, Willis SC. Context, provenance and technology of a western stemmed tradition artifact cache from the Cooper’s Ferry Site, Idaho. American Antiquity. 2014;79: 596–615.
- 45. Waters MR, Forman SL, Jennings TA, Nordt LC, Driese SG, Feinberg JM, et al. The Buttermilk Creek Complex and the Origins of Clovis at the Debra L. Friedkin Site, Texas. Science. 2011;331: 1599–1603. pmid:21436451
- 46. Waters MR, Pevny CD, Carlson DL. Clovis Lithic Technology: Investigation of a Stratified Workshop at the Gault Site, Texas. College Station: Texas A&M Press; 2011.
- 47. Williams TJ, Collins MB, Rodrigues K, Rink WJ, Velchoff N, Keen-Zebert A, et al. Evidence of an early projectile point technology in North America at the Gault Site, Texas, USA. Science Advances. 2018;4: eaar5954. pmid:30009257
- 48. Kooyman B, Hills LV, McNeil P, Tolman S. Late Pleistocene horse hunting at the Wally’s beach site (DhPg-8), Canada. American Antiquity. 2006;71: 101–121.
- 49. Ostahowski BE, Kelly RL, MacDonald DH, Andrefsky W, Yu P-L. Alm Rockshelter Lithic Debitage Analysis: Implications for Hunter-Gatherer Mobility Strategies in the Big Horn Mountains, Wyoming. Lithics in the West. Missoula: University of Montana Press; 2014. pp. 120–141.
- 50. Kornfeld M, Larson ML, Rapson DJ, Frison GC. 10,000 years in the Rocky Mountains: The Helen Lookingbill site. Journal of Field Archaeology. 2001;28: 307–324.
- 51. Irwin-Williams C, Irwin H, Agogino G, Haynes CV. Hell Gap: Paleo-indian occupation on the High Plains. Plains Anthropologist. 1973;18: 40–53.
- 52. Larson ML, Kornfeld M, Frison GC. Hell Gap: A Stratified Paleoindian Campsite at the Edge of the Rockies. Salt Lake City: University of Utah Press; 2009.
- 53. Pelton SR, Kornfeld M, Larson ML, Minckley T. Component age estimates for the Hell Gap Paleoindian site and methods for chronological modeling of stratified open sites. Quaternary Research. 2017;88: 234–247.
- 54. McNett CW. Shawnee Minisink: a stratified Paleoindian–Archaic site in the upper Delaware Valley of Pennsylvania. Academic Press; 1985.
- 55. Gingerich JAM. Shawnee-Minisink Revisited: Re-Evaulating the Paleoindian Occupation. Department of Anthropology. University of WyomingEditor. 2007.
- 56. Gingerich JAM. Down to seeds and stones: A new look at the subsistence remains from Shawnee-Minisink. American Antiquity. 2011;76: 127–144.
- 57. Dalton AS, Margold M, Stokes CR, Tarasov L, Dyke AS, Adams RS, et al. An updated radiocarbon-based ice margin chronology for the last deglaciation of the North American Ice Sheet Complex. Quaternary Science Reviews. 2020;234: 106223.
- 58. Danielson JJ, Gesch DB. Global multi-resolution terrain elevation data 2010 (GMTED2010). Washington D. C.: US Department of the Interior, US Geological Survey Open File Report 2011–1073; 2011.
- 59. Potter BA, Holmes CE, Yesner DR. Technology and economy among the earliest prehistoric foragers in interior eastern Beringia. In: Graf KE, Ketron CV, Waters MR, editors. Paleoamerican odyssey. College Station, Texas: Texas A&M University Press; 2013. pp. 81–103.
- 60. Dilley TE. Late Quaternary Loess Stratigraphy, Soils, and Environments of the Shaw Creek Flats Paleoindian sites, Tanana Valley, Alaska. Ph.D. Dissertation, The University of Arizona. 1998.
- 61. Yesner DR, Holmes CE, Crossen KJ. Archaeology and paleoecology of the Broken Mammoth site, central Tanana Valley, interior Alaska, USA. Current Research in the Pleistocene. 1992;9: 53–57.
- 62. Mackie ME. Paleoindian-Proboscidean Interactions in the Terminal Pleistocene. Department of Anthropology. University of WyomingEditor. 2019.
- 63. Haynes CV Jr, Huckell BB. Murray Springs: A Clovis Site with Multiple Activity Areas in the San Pedro Valley, Arizona. Tucson: Anthropological Papers of the University of Arizona, Number 71, The University of Arizona Press; 2007.
- 64. Haury EW. Artifacts with mammoth remains, Naco, Arizona. American Antiquity. 1953;19: 1–14.
- 65. Haury EW, Sayles EB, Wasley WW. The Lehner mammoth site, southeastern Arizona. American Antiquity. 1959;25: 2–42.
- 66. Hannus LA. Clovis Mammoth Butchery: The Lange/Ferguson Site and Associated Bone Tool Technology. College Station: Texas A&M University Press; 2018.
- 67. Leonhardy FC, Anderson AD. The archaeology of the Domebo site. In: Leonhardy FC, editor. Domebo: A Paleo-Indian Mammoth Kill in the Prairie-Plains. Lawton, Oklahoma: Contributions of the Museum of the Great Plains, No. 1; 1966. pp. 14–26.
- 68. Sanchez G, Holliday VT, Gaines EP, Arroyo-Cabrales J, Martínez-Tagüeña N, Kowler A, et al. Human (Clovis)-gomphothere (Cuvieronius sp.) association ~13,390 calibrated yBP in Sonora, Mexico. Proceedings of the National Academy of Sciences. 2014;111: 10972–10977. pmid:25024193
- 69. Frison GC, Todd LC. The Colby Mammoth Site. Albuquerque: University of New Mexico Press; 1986.
- 70. Dillehay TD. Monte Verde: A Late Pleistocene Settlement in Chile. Volume 1: Paleoenvironment and Site Context. Washington D. C.: Smithsonian Institution Press; 1989.
- 71. Dillehay TD, Ocampo C, Saavedra J, Sawakuchi AO, Vega RM, Pino M, et al. New Archaeological Evidence for an Early Human Presence at Monte Verde, Chile. PLOS ONE. 2015;10: e0141923. pmid:26580202
- 72. Fiedel SJ. Artifact Provenience at Monte Verde: Confusion and contradictions. Special Report: Monte Verde Revisited. El Paso, Texas: Scientific American Discovering Archaeology; 1999. pp. 1–12.
- 73. Jenkins DL, Davis LG, Stafford TW, Campos PF, Hockett B, Jones GT, et al. Clovis Age Western Stemmed Projectile Points and Human Coprolites at the Paisley Caves. Science. 2012;337: 223–228. pmid:22798611
- 74. McDonough K, Luthe I, Swisher ME, Jenkins DL, O’Grady P, White F. ABCs at the Paisley Caves: Artifact, Bone, and Coprolite Distributions in Pre-Mazama Deposits. Current Archaeological Happenings in Oregon. 2012;37: 7–12.
- 75. Guidon N, Delibrias G. Carbon-14 dates point to man in the Americas 32,000 years ago. Nature. 1986;321: 769–771.
- 76. Macphail RI, McAvoy JM. A micromorphological analysis of stratigraphic integrity at Cactus Hill, an early Paleoindian and hypothesized Pre-Clovis occupation in South-Central Virginia, USA. Geoarchaeology. 2008;23: 675–694.
- 77. Wagner DP, McAvoy JM. Pedoarchaeology of Cactus Hill, a sandy Paleoindian site in southeastern Virginia, U.S.A. Geoarchaeology. 2004;19: 297–322.
- 78. Fiedel SJ, Potter BA, Morrow JE, Faught MK, Haynes CV Jr, Chatters JC. Pioneers from Northern Japan in Idaho 16,000 Years Ago? A Critical Evaluation of the Evidence from Cooper’s Ferry. PaleoAmerica. 2020;7: 28–42.
- 79. Morrow JE, Fiedel SJ, Johnson DL, Kornfeld M, Rutledge M, Wood WR. Pre-Clovis in Texas? A critical assessment of the “Buttermilk Creek Complex”. Journal of Archaeological Science. 2012;39: 3677–3682.
- 80. Hofman JL. Vertical movement of artifacts in alluvial and stratified deposits. Current Anthropology. 1986;27: 163–171.
- 81. Hofman JL. Defining buried occupation surfaces in terrace sediments. In: Hofman JL, Enloe JG, editors. BAR International Series. Oxford: BAR International Series 578; 1992. pp. 128–128.
- 82. Villa P. Conjoinable pieces and site formation processes. American Antiquity. 1982;47: 276–290.
- 83. Laughlin J. 149 Refits: Assessing site integrity and hearth-centered activities at Barger Gulch Locality B. M.A. Thesis, Department of Anthropology, University of Wyoming. 2005.
- 84. Surovell TA, Waguespack NM, Mayer JH, Kornfeld M, Frison GC. Shallow site archaeology: Artifact dispersal, stratigraphy, and radiocarbon dating at Barger Gulch, Locality B, Middle Park, Colorado. Geoarchaeology. 2005;20: 627–649.
- 85. Gómez Coutouly YA, Holmes CE. The microblade industry from Swan Point CZ4b: Technological and cultural implications from the earliest human occupation in Alaska. American Antiquity. 2018;83: 735–752.