PUBLIC RESEARCH PREVIEW Research in progress · uncertainties and source access are recorded throughout.

Locality 2 · Research extension · October 28, 2026

Did movement change?

Connect the tracks to their surfaces, water evidence and observation windows before interpreting a trend.

Open the dating ledger →

What this comparison establishes

A preservation-adjusted change in movement is not yet demonstrated. We can now compare exposure and identify observation biases, rather than treating the number of recorded prints as a population curve. The missing ingredients are consistent trackway geometry, mapped preservation masks, and the duration for which each surface could receive tracks.

This review covers 20 chronological units, 19 counting records, and 15 aquatic-microfossil samples at White Sands Locality 2. It links the existing chronology to the original 2026 supplement. It does not assign Locality 2 ages to other White Sands trackways. Review date: October 28, 2026; editorial state: draft.

What changed when we accounted for area

The following values are our calculations from the published count and area columns in Table S1. They describe recovered human-print density within the listed exposure, not people per year, visits per year, or population size.

SurfaceHuman printsListed area (m²)Calculated prints/m²Published rounded prints/m²
TH4376.705.526
TH3191.3514.0714
TH(-3)70.2626.9227

TH4 has over five times as many recorded prints as TH(-3), but TH(-3) has about 4.9 times the density after dividing by area. The known observation areas range from 0.20 to 6.70 m², a 33.5-fold difference. This is a useful reversal of a raw-count ranking, not evidence that traffic decreased over time. Heavy trampling at TH(-3), different excavation windows, and unknown recording duration still prevent that inference.

Rules used for the comparison

Surface-identity review: cross-publication joins involving TH2, TH3 and TH4 are unresolved. The values below retain the 2026 source labels; the disputed photograph cannot independently identify their count windows or ages. Read the scope audit.

Horizon-by-horizon comparison

Rows run from older/lower to younger/upper. A dash means unreported. Joint marks the shared TP10–TP11 observation. Counts, areas and taxa come from Table S1; modeled ages remain those in the existing horizon ledger. The sediment and water columns explicitly separate surface observations from nearby depositional context.

Horizon; modeled age ±2σ (cal ka BP); basisHuman count / area; calculated prints/m²Sediment contextStanding-water evidence and limitExcavation and preservationAnimal reports and movement evidence
TH(-3); 23.49 ± 0.32; interpolated7 / 0.26 m²; 26.92Transition near the top of lower lacustrine clay, with a thin sand bed nearby.Aquatic microfossils in sample 18.1r, 0.04 m below. Burrowed lake clays and larger aquatic assemblages occur farther below. This does not establish water depth while walking.Base of the main trench. Heavy trampling and weak morphology.Animals not listed in Table S1. Possible small-foot shuffling/rotation in Fig. S14; tentative event interpretation, not a trend.
TH(-2); 23.40 ± 0.29; interpolated— / —; —Clay-rich interval above the lower lake deposits.Sample 18.1p, 0.05 m above, contains Limnocythere; 18.1r below also contains it. Intermittent water is compatible with this interval; the track surface itself is unsampled.Documented then removed to expose the trench bottom. Poorly defined trample; possible handprint unconfirmed; area/count unreported.Animals not listed in Table S1. Presence reported; missing counts must not become zeros.
TH(-1); 23.22 ± 0.24; interpolated1 / 0.20 m²; 5.00Fine-grained beds with sand interbeds in the transitional succession.Between microfauna-bearing 18.1p and barren 17.1j. Broad transitional context; no horizon-specific standing-water observation.Lower trench. Trampling and weak morphology.Animals not listed in Table S1. No quantified movement geometry in Table S1.
TH0; 23.14 ± 0.22; interpolated4 / 0.33 m²; 12.12Sand/fine-bed contact within the post-lake alluvial succession.Between microfauna-bearing 18.1p and barren 17.1j. Broad transitional context; no horizon-specific standing-water observation.Mapped surface in Fig. S13. Trampling and weak morphology; saw cuts removed parts of the exposed surface.Animals not listed in Table S1. Several print sizes visible; no standardized movement metric.
TH1; 22.70 ± 0.11; interpolated5 / 0.32 m²; 15.62Near the gypsiferous clay interval associated with short-lived ponding.Main text interprets nearby gypsiferous clay as brief ponding around 22.7 ka. Sample 17.1j at 1197.09 m is barren: absence alone does not establish dryness.North–south trench extension. No standardized preservation score.Animals not listed in Table S1. Trench orientation is not evidence of walking direction.
TH2; 22.66 ± 0.10; interpolated45 / 3.82 m²; 11.78Indurated dolomite-rich horizon; positive and negative track relief.Dolomite follows a ponding/evaporation interval in the authors’ interpretation. Water chemistry history does not determine conditions during each footfall.Only the part adjacent to the main trench is counted. Dolomite cement and fills; both positive and negative relief.Proboscidea, Carnivora (canid), Aves. Linear features and human/animal associations are reported. Fig. S12’s groove-to-horizon assignment is disputed; its geometry cannot be assigned to this dated row without reconciliation.
TH3; 22.61 ± 0.09; interpolated19 / 1.35 m²; 14.07Heterogeneous deformed middle interval; original and composite logs differ in detail.Post-ponding lake-margin/alluvial context. The two nearest microfauna samples are barren and widely spaced; local wetting at this surface is unresolved.Only the part adjacent to the main trench is counted. No standardized preservation score; animal overprinting can obscure human anatomy.Proboscidea, Carnivora (canid). Canid over human print in Fig. 2A; linear features also reported.
TH4; 22.57 ± 0.08; interpolated37 / 6.70 m²; 5.52Fine-grained bed over the heterogeneous, deformed middle interval.Post-ponding lake-margin/alluvial context. The two nearest microfauna samples are barren and widely spaced; local wetting at this surface is unresolved.South shoulder. No standardized preservation score.Animals not listed in Table S1. Linear feature indicated in Fig. S21; Table S1 has no directional measurements.
TH5; 22.52 ± 0.07; interpolated3 / 0.34 m²; 8.82Sand/fine-bed contact in the composite section.Post-ponding lake-margin/alluvial context. The two nearest microfauna samples are barren and widely spaced; local wetting at this surface is unresolved.Floor and south shoulder. Few prints; no standardized preservation score.Animals not listed in Table S1. No quantified movement geometry in Table S1.
TH6; 22.43 ± 0.06; interpolated1 / 0.25 m²; 4.00Bedded silt/clay interval with sand interbeds nearby.Post-ponding lake-margin/alluvial context. The two nearest microfauna samples are barren and widely spaced; local wetting at this surface is unresolved.Trench floor. One human print; no standardized preservation score.Animals not listed in Table S1. A single print cannot determine a trackway distribution.
TH7; 22.41 ± 0.06; interpolated1 / 0.20 m²; 5.00Sand/fine-bed contact in the composite section.Post-ponding lake-margin/alluvial context. The two nearest microfauna samples are barren and widely spaced; local wetting at this surface is unresolved.Trench floor. One human print; no standardized preservation score.Proboscidea. Human and proboscidean presence; animal count unavailable.
TP8; 22.33 ± 0.07; interpolated8 / 0.58 m²; 13.79Clay/silt-dominated package with interbeds.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.South-side bench. Strong trampling and relief; weak anatomical detail.Animals not listed in Table S1. Trampling recorded; no standardized turning or direction measurements.
TP9; 22.22 ± 0.09; interpolated4 / 0.35 m²; 11.43Alternating sand and fine beds; an infilled linear feature is exposed.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.Mapped exposure in Fig. S11. Filled groove changes surface relief.Animals not listed in Table S1. Human prints overlie an infilled linear trace; relative order is observable.
TP10; 22.08 ± 0.10; interpolatedJoint: 4 / 0.33 m²; 12.12Bedded fine-grained interval; the counted exposure also includes TP11.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.Two closely spaced surfaces pooled in one Table S1 row. Hinged-block excavation illustrated in Figs. S3E and S10; no calibrated recovery rate.Animals not listed in Table S1. Do not allocate the four prints or 0.33 m² separately to TP10 and TP11.
TP11; 21.97 ± 0.14; extrapolatedJoint: 4 / 0.33 m²; 12.12Clay/silt interval; the counted exposure also includes TP10.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.Two closely spaced surfaces pooled in one Table S1 row. Hinged-block excavation illustrated in Figs. S3E and S10; no calibrated recovery rate.Animals not listed in Table S1. Do not allocate the four prints or 0.33 m² separately to TP10 and TP11.
TP12; 21.91 ± 0.17; extrapolated11 / 0.96 m²; 11.46Sand/silt beds; the bench intersects several layers.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.South-shoulder bench intersects several layers. Hard layers are difficult to separate.Animals not listed in Table S1. Different orientations visible in Fig. S9; no azimuth sample published in this table.
TP13; 21.86 ± 0.19; extrapolated5 / 1.00 m²; 5.00Clay-rich band in the composite section; indurated exposure.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.Irregular upper face of an indurated block. At least one poorly defined print.Animals not listed in Table S1. No quantified movement geometry in Table S1.
TP14; 21.82 ± 0.20; extrapolated3 / 0.69 m²; 4.35Indurated fine-grained upper sequence; exact contact grain size not resolved here.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.North-wall recess. Shallow outlines in a hard surface; difficult excavation. Proposed sandal impression remains unconfirmed.Animals not listed in Table S1. Three tracks do not supply a directional distribution.
TP15; 21.78 ± 0.22; extrapolated14 / 0.80 m²; 17.50Thin gypsum-rich sheets with dolomite in parts of the surface and track fills.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.South wall; at least three superposed surfaces. A top print may belong to TP16. Indurated gypsum sheets and dolomite fills; layer mixing. Listed area 0.80 m² differs from the approximate 0.5 × 1 m description; retain the tabulated area.Animals not listed in Table S1. Multiple surfaces cannot be treated as one movement event.
TP16; 21.68 ± 0.25; extrapolated2 / 1.00 m²; 2.00Upper bedded sand/silt/clay sequence; desiccated surface.Upper lake-margin/wet-playa/alluvial setting; the listed upper microfauna samples are barren. No per-surface water-depth measurement.Trench top; adjacent animal traces extend beyond the human-count window. Shallow prints; cracking, erosion and weak contrast impair detection. Possible ursid identification is unconfirmed.Proboscidea, Folivora (ground sloth), Carnivora (canid), Carnivora (ursid; unconfirmed). Desiccation records an exposed phase, not permanent dryness.

How the water evidence is connected

The 15 sample records in the accompanying data retain the original elevations, modeled ages and six abundance columns. Each horizon links to the closest samples above/below in the same published Locality 2 composite section, with signed vertical offsets; a shared rounded elevation is stored separately. These are contextual joins, not claims of identical surfaces or synchronous conditions. Sample thickness and lateral offsets are not quantified in this extraction. Shared field-name roots do not establish identical sampling positions: Table S15 places proxy 22WHSA1-18.1r at 1196.22 m, while Table S4 lists the seed sample with that field ID at 1196.26 m. Each table’s position and measurement type remains intact.

The strongest pre-track lake evidence occurs below TH(-3): burrowed clay and aquatic assemblages. Sample 22WHSA1-18.1r is 0.04 m below TH(-3) and contains 55 adult and 11 juvenile Limnocythere valves/g. Sample 18.1p, 0.05 m above TH(-2), has 8 and 2. The three upper samples in Table S15 are barren. That sparse sampling leaves long intervals unresolved; notably, the gap between 18.1p and 17.1j is 0.70 m. Absence in a sample is not proof of continuous dryness throughout an interval.

The authors interpret a brief ponding interval near 22.7 ka followed by evaporative conditions and the TH2 dolomite-rich horizon. That offers a local environmental contrast to investigate, while the model uncertainties and shared chronology prevent precise year-by-year alignment with behavior. Their climate correlations also helped evaluate the chronology, so agreement with those same correlations is not a new independent timing test.

Research ideas that can now be tested

These are proposals, not findings. The most promising outcome is a reproducible change in movement geometry, with an explanation that survives observation controls.

Priority and hypothesisObservable predictionCompeting explanation and controlResult that would advance or weaken it
1. Shoreline change redirected routes.Trackway orientations, path concentration or turning locations shift relative to independently mapped shoreline/wet-ground facies.Narrow excavation windows favor some orientations. Map the full observed polygon, damaged areas and detection quality; simulate tracks crossing the same window. Compare single surfaces before packages.A directional change persisting across comparable masks and plausible shoreline/age reconstructions advances the idea. A change that disappears after window correction weakens it.
2. Some surfaces record stopping or repeated local use.More turning, short loops, repeated overlap or local shuffling relative to traversing trackways.Soft substrate, slipping, long exposure or superposed layers can produce apparent clustering. Treat individuals/trackways as clusters; vary plausible exposure durations and exclude uncertain layer assignments.Replicated movement geometry on independent, well-resolved surfaces advances the idea. TH(-3)‘s tentative shuffling remains an event-level candidate, not a time-series result.
3. Humans and animals changed how they shared routes.Directional alignment and human–animal crossings change beyond what their mapped densities predict.Larger animal prints are easier to detect; overprinting destroys earlier prints. Use taxon-specific visibility masks and randomize paths within the observed polygons while preserving length/orientation distributions.A residual association, with explicit cross-cutting order, supports a spatial relationship. It still does not establish hunting or attraction to water.

TH2–TH3 follow-up completed

Read the TH2–TH3 movement assessment. The original 3D archive is now inventoried, and a visually matching surface has conflicting horizon labels in the 2025 and 2026 figures. Local trace analysis is feasible; a preservation-adjusted cross-horizon movement change remains unestablished. The initial proposal below is retained as the rationale for selecting this pair, with its implementation constrained by that follow-up.

Initial comparison proposal

Start with TH2 and TH3 as a feasibility pair for geometry extraction, using Fig. S12, the published orthomosaics, and the original mapped records where accessible. They have relatively large documented exposures and human/animal associations. They are not yet a matched experiment: TH2’s dolomite and relief differ, and the counted windows are only the trench-adjacent portions. The nominal model ages differ by about 50 years, with overlapping uncertainties; the stratigraphic order is more secure than the elapsed time.

For each surface, create one record per print and one per connected trackway: relative position in the mapped exposure, direction and direction uncertainty, connection confidence, taxon, relief, anatomical quality, missing/damaged area, overprint order, excavation method, and the layer-assignment confidence. Link every record to the observed polygon. Public records can use relative coordinates without disclosing protected site coordinates. Do not treat multiple prints from one trail as independent walkers.

Then evaluate whether both surfaces retain enough comparable geometry after excluding damaged and ambiguous areas. If they do, compare orientation and turning with uncertainty estimated at the trackway level. If they do not, the useful result is a precise statement of the missing exposure or documentation, rather than a fitted climate–population relationship. The available aggregate table cannot yet support that statistical test.

For a later temporal analysis, start with the 11 discrete horizons, retaining the uncounted TH(-2) as missing. Analyze the nine packages separately until their individual surfaces can be resolved. Use stratigraphic order first, then sensitivity analyses across age models; do not divide counts by differences between model midpoint ages to obtain a visitation rate. Those differences are not measured durations of track-forming exposure.

Provenance and unresolved gaps

Primary source: Springer et al. (2026), DOI 10.1126/sciadv.aeh0707, existing source S5. Open article · supplement archive from Europe PMC. Table S1 (PDF pp. 46–47), Table S15 (p. 66), and Figs. S6, S21 and S26 were visually inspected. Figs. S7–S14 supply descriptive preservation and movement context. Ages use Table S9d through the unchanged repository ledger.

The source data release, DOI 10.5066/P13W7HTM, is identified, but its machine-readable download was not retrieved in this pass. Values here were transcribed from the original supplementary PDF. No calibrated detectability, per-horizon surface duration, standardized animal counts, or complete direction/trackway dataset was recovered in this review. The existing S8 chronology arithmetic flags remain unresolved and unchanged.

Canonical numerical/context records: research/data/white-sands/horizon-context.json; chronology: research/data/white-sands/horizons.json. Derived density calculations and integrity checks: src/lib/white-sands-context.mjs. The original workbook remains a frozen version-1 snapshot. The new comparison is a research extension, not a reanalysis that establishes a behavioral or population trend.

Inspect the published counting columns

Each Table S1 row appears once here. Calculated density uses the tabulated count and area. An arithmetic flag does not replace the published value or produce a revised demographic estimate.

Table S1 observations; S5. Null entries mean unreported.
RecordArea m²Human printsPublished prints/m²Calculated prints/m²Published MNIPublished I/m²MNI/area audit
TP161.00222.00111.00
TP150.80141817.50556.25Does not reproduce printed I/m² by nearest-integer rounding
TP140.69344.35222.90Does not reproduce printed I/m² by nearest-integer rounding
TP131.00555.00111.00
TP120.96111111.46323.13Does not reproduce printed I/m² by nearest-integer rounding
TP11-TP100.3341212.12123.03Does not reproduce printed I/m² by nearest-integer rounding
TP90.3541111.43112.86Does not reproduce printed I/m² by nearest-integer rounding
TP80.5881413.79578.62Does not reproduce printed I/m² by nearest-integer rounding
TH70.20155.00115.00Does not reproduce printed I/m² by nearest-integer rounding
TH60.25144.00114.00Does not reproduce printed I/m² by nearest-integer rounding
TH50.34398.82112.94Does not reproduce printed I/m² by nearest-integer rounding
TH4Surface-label review6.703765.52410.60
TH3Surface-label review1.35191414.07533.70Does not reproduce printed I/m² by nearest-integer rounding
TH2Surface-label review3.82451211.78611.57Does not reproduce printed I/m² by nearest-integer rounding
TH10.3251615.63133.13
TH00.3341212.12123.03Does not reproduce printed I/m² by nearest-integer rounding
TH(-1)0.20155.00115.00Does not reproduce printed I/m² by nearest-integer rounding
TH(-2)UnreportedUnreportedUnreportedUnreportedUnreportedUnreportedUnreported
TH(-3)0.2672726.9241315.38Does not reproduce printed I/m² by nearest-integer rounding

MNI: minimum number of individuals inferred by the authors. I/m²: the published individuals-per-square-meter column. Neither is a population census.

Source record S5 → · White Sands dossier →