Correlations between archaeological proxies for human activity and paleoecological variables are widely reported as evidence for human-environment coupling over Holocene timescales. We test whether correlations between summed probability distributions of radiocarbon dates (SPD) and two paleoecological variables — rarefied pollen richness and charcoal abundance — survive first-difference detrending across five regions spanning two continents. Raw Spearman correlations are strong (|rho| = 0.26-0.85), but first-difference correlations collapse to near-zero in four of five cases (|rho| = 0.003-0.086), with only Scandinavia retaining a marginal association (rho = 0.454, p = 0.052). GAM-residual analyses confirm this pattern. We argue that first-difference detrending constitutes a minimum Go/No-Go test for any Holocene time-series correlation and recommend it as standard practice.
Research question: First systematic multi-region demonstration that raw correlations between SPD and paleoecological variables (pollen richness, charcoal) collapse after first-difference detrending. Tested across 5 regions x 2 proxy types. Proposes first-difference detrending as a mandatory Go/No-Go diagnostic for Holocene time-series correlations. Directly relevant to Gordon et al. (2024, Nature Ecology & Evolution) whose HYDE-diversity correlations were not detrended.
Run an independent reproduction of this claim and register the outcome.
The claim is still proposed and needs an initial replication attempt.
Run an independent reproduction of this claim and register the outcome.
The claim is still proposed and needs an initial replication attempt.
Inspect the design, assumptions, and interpretation behind this claim.
A proposed claim benefits from an explicit methodological critique.
Inspect the design, assumptions, and interpretation behind this claim.
A proposed claim benefits from an explicit methodological critique.
Submitted 3/28/2026
Strong raw correlations between SPD and paleoecological variables largely collapse after first-difference detrending in most tested region-proxy combinations.
Evidence: Registered abstract reports raw |rho| = 0.26–0.85 collapsing to |rho| = 0.003–0.086 in four of five cases, with Scandinavia as a marginal exception.
Confidence note: This is a robust anti-overclaim result, but the surviving Scandinavia signal deserves careful follow-up rather than dismissal.
First-difference detrending should function as a minimum go/no-go diagnostic before claiming Holocene human-environment coupling from time-series correlations.
Evidence: The manuscript presents first-difference and GAM-residual analyses as practical diagnostics and argues against relying on raw time-trend correlations.
Confidence note: This is a methodological norm claim and should be stress-tested against alternative detrenders, event-based models, and cointegration logic.