Technical Comments

Comment on “Late Upper Paleolithic occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago”

See allHide authors and affiliations

Science  10 Apr 2020:
Vol. 368, Issue 6487, eaaz4695
DOI: 10.1126/science.aaz4695


Davis et al. (Research Articles, 30 August 2019, p. 891) report human occupation at Cooper’s Ferry, Idaho, USA, ~16,000 years ago, well before Greenland Interstadial 1 (GI-1). Critical review suggests that this early date is not supported by the evidence. Human occupation might have begun in the mid-16th millennium before the present, but would have been more likely after ~15,000 years ago, coeval with GI-1.

Davis et al. argue for human occupation at the Cooper’s Ferry site in western Idaho, USA, starting between 16,560 and 15,280 calibrated years before the present (cal yr B.P., base year 1950) at 95.4% highest posterior density (hpd) (generalized as ~16,000 years ago) (1). However, critical examination of the data and analysis presented gives rise to a degree of caution.

The start date for human occupation at Cooper’s Ferry is not directly attested (1). It is the start Boundary range calculated by the OxCal Bayesian chronological modeling software (2, 3) around a uniform-probability phase containing 12 14C ages and one optically stimulated luminescence (OSL) age for the LU-3 stratigraphic unit, excluding data from within-pit features. This start Boundary is necessarily before the earliest measured age in the model, OxA-38051, nonmodeled 95.4% calendar age range 15,945 to 15,335 cal yr B.P. (1). Nonetheless, Davis et al. argue that debitage, bone fragments, and a piece of charcoal in stratigraphic positions within LU-3 below OxA-38051 justify the use of such an earlier date estimate (1). However, there is no quantifiable information, and the stratigraphically lower OxA-38106 in fact yielded a much more recent age, raising issues of bioturbation (1).

Is the earlier date justified? The oldest attested age, OxA-38051, is flagged as a possible outlier (residual, in-built age?) (1). The available LU-3 14C dates range widely [midpoints, even excluding three major outliers, differ by 1660 14C yr (1)]. Thus, modeling will determine too early a start Boundary simply given the data scatter. This is especially the case where the 95.4% limits are cited, because with no older constraint, the probability tail is long in this direction (Fig. 1). For example, if we use the midpoints of the calendar age ranges reported for the nine non–major outlier 14C data from LU-3 (1) as “known” ages, simulations of a LU-3 phase place the beginning of a calculated start Boundary substantially before the oldest attested date (Fig. 1A). Likewise, the cited start of the 95.4% range (16,560 cal yr B.P.), or even its approximate midpoint, 16,000 cal yr B.P. (1), are substantially earlier than OxCal First queries for the initial dated event, or the median of the hpd ranges, from multiple runs of the Davis et al. (1) models, both with and especially without OxA-38051 (Fig. 1B).

Fig. 1 Modeled start Boundaries for LU-3 (1).

(A) Fifty simulations of the start Boundary of an OxCal uniform-probability phase containing 9 LU-3 14C ages representing the approximately known real calendar ages [using R_Simulate from the midpoints of the respective calibrated calendar age ranges in table 1 of (1), rounded with measurement error treated as ±50 14C yr]. The beginning of the 95.4% probability age range for the start Boundary is variously 1478 to 658 years (mean ± SD: 881 ± 201 years) older than the oldest “known” age in the phase (15,640 cal yr B.P.); the median of the overall start Boundary ranges is 360 to –27 years older. (B) Start Boundaries and median ranges from the dating model in (1) with and without OxA-38051, 10 iterations each. The cited 95.4% hpd range for the start Boundary from (1) is indicated. Data are from OxCal (2, 3, 5) using IntCal13 (13).

Definite association of all dated samples with direct human activity is also imperfect. Heavily fragmented bone samples without clear human-use associations could include material from prior, or later, carnivore and rodent scavenging and mobilization, as well as subsequent human digging [Archaic pit digging is noted (4)] and trampling, or other natural movement. The unidentified charcoal could derive from (prior) nonhuman processes and also likely includes in-built age (5, 6). This applies especially in a river canyon setting such as Cooper’s Ferry, with a surrounding higher-elevation environment where, in addition to immediate riparian resources, long-lived species of trees (e.g., conifers) are accessible (7). The in-built age issue for charcoal samples is further exacerbated by probable human exploitation (scavenging) of snags or dead fallen logs and branches, including those likely available near the site carried by the Salmon River from erosion episodes upstream. The mix in ages evident within LU-3 implies that some additional processes were at work. Pervasive bioturbation has previously been noted as a feature of the site (8) [whether anthropogenic or natural—and, despite best efforts by the excavation team, rodent burrows are a major challenge at the site, with three of the 14C outliers associated with such features despite prior screening against this specific problem (1)]. Nor is more recent intrusive material the only concern. Some older (residual) bone or charcoal material could become incorporated into the stratigraphic units with human presence (8), especially as pits cut into deeper deposits (1). The F134 pit with two 14C ages more than 3000 14C yr apart (1) could be an example, as could the old outlier from LU-4, OxA-X-2792-42 (1). Clear hearth features and directly associated charcoal and bone, and replicated dates on these materials, are widely recognized as primary evidence of undeniable human presence at early sites (8, 9). The earliest such context at Cooper’s Ferry is F129, interpreted as a hearth. Three 14C dates on charcoal are directly associated, and this context, running across to F143, appears to define a contemporaneous-use surface (1).

Revision of the Davis et al. dating model (1) offers a more appropriate age estimate. This excludes the five very large, >90% probability, outliers identified (1) and applies the OxCal Charcoal Outlier_Model (5) to the dates on charcoal samples to allow approximately for in-built age (5, 6). The OxCal Combine function approximates the age of the F129 use episode (perhaps the earliest secure human presence), and the OxCal First query estimates the date range for the initial, attested, activity associated with the LU-3 phase (Fig. 2). Although still an important early Paleolithic site in the northwest United States, this would suggest (average ± SD 95.4% hpd ranges from 10 model runs, rounded outward) an initial (First query) LU-3 date of ~15,935 ± 75 to 15,130 ± 20 cal yr B.P. including OxA-38051, and ~15,735 ± 20 to 14,740 ± 90 cal yr B.P. without. The F129 hearth is placed ~14,980 ± 5 to 14,130 ± 10 cal yr B.P. with OxA-38051, and ~14,985 ± 5 to 14,205 ± 10 cal yr B.P. without. The latter, providing secure evidence of human presence, is proximate to the beginning of Greenland Interstadial (GI) 1e to 1a, starting ~14,692 (±4) (maximum error ±186 years) B2K = 14,642 yr B.P. (10), and the earliest hypothesized opening of an ice-free corridor (IFC) southward from Beringia at ~14,800 cal yr B.P. (1). Within the OxCal model (1), only OxA-38051 and OxA-X-2792-48 indicate earlier ages. Both are associated approximately with the base of the surface running from F129 to F143 (1) but lack clear cultural association. OxA-X2792-48 had low %C (<30%) and is noted as an OxA-X date (1), indicating a research measurement using nonstandard or experimental methods (11). Details are not given. The C:N ratio stated (3.3) would indicate a standard date on ultrafiltered collagen (satisfactory range ~2.9 to 3.5), and not use of potentially more accurate hydroxyproline (satisfactory values ~5.0) (12). OxA-38051 is the oldest age but was flagged as a possible outlier (1). Thus, neither of these two key dates are perhaps best evidence. Yet the claim of very early (pre-IFC, pre–GI-1) human presence depends on them. The other LU-3 data appear approximately coeval with the GI-1 interval (and subsequent LU-4 with the early Holocene). Critical review suggests that more and better evidence is required to be confident of human presence at Cooper’s Ferry before ~15,000 years ago.

Fig. 2 A revised Bayesian age model for Cooper’s Ferry, area A.

The model uses data and stratigraphic information, excluding the five large (>90% probability) outliers, from (1). OxCal Boundaries are in magenta; dates on charcoal are in black, dates on bone in blue; the OSL dates are in orange. Solid distributions are the modeled probabilities (the crosses indicate the median values) with the 68.2% and 95.4% hpd ranges below; hollow distributions show the nonmodeled probabilities. The outputs (O) of the OxCal General Outlier_Model (posterior/prior) are shown next to the bone (blue) 14C dates and OSL (orange) dates; the OxCal Charcoal Outlier_Model is applied to the 14C dates on charcoal (black) (output, O, always 100/100 for this model). The vertical green band represents the span of Greenland Interstadial 1 (10); the vertical yellow band indicates Greenland Stadial 1 (10). Modeling uses OxCal 4.3 (2, 3, 5) with IntCal13 (13).


View Abstract

Stay Connected to Science

Navigate This Article