Bridging scales: Regional coherence between gross primary productivity and boreal tree growth revealed by harmonized flux and dendrochronological data

Gross primary productivity (GPP) is widely assumed to drive annual tree growth, an assumption embedded in many terrestrial ecosystem models. Yet, empirical evidence from boreal forests often shows weak coupling between eddy-covariance GPP and tree growth, a pattern attributed to physiological constraints and mismatched measurement scales. We revisited this relationship using harmonized datasets of eddy-covariance GPP and tree-ring basal area increments (BAI) from six boreal flux sites spanning diverse structural and climatic conditions, combined with a continental-scale tree-ring width network covering Canada encompassing >40,000 tree samples from >4,500 sites. Tree-ring widths were converted to BAI and detrended using species-specific generalized additive mixed models (GAMMs) to isolate abiotic/biotic-driven growth anomalies. Growth anomalies were aggregated within radii optimized to match flux tower footprints and expressed as annual percent deviations from expected growth. Eddy-covariance GPP was derived from long-term flux records following standardized gap-filling and filtering protocols, and annual sums were computed from monthly fluxes. Pairwise correlations between GPP and growth anomalies were assessed using Pearson coefficients with bootstrap confidence intervals to account for temporal autocorrelation. Spatial correlation fields were constructed by correlating site-level GPP time series with gridded growth anomalies to evaluate regional coherence. Contrary to previous findings, five boreal flux sites exhibited significant positive correlations between annual growth anomalies and GPP (r = 0.69–0.85, p < 0.05), while one northern black spruce site showed no association. Spatial correlation fields revealed that growth–GPP coupling extends well beyond flux tower footprints, forming coherent regional patterns across hundreds of kilometers. These results challenge the prevailing view of universal source–sink decoupling and highlight the importance of scale and representativeness in diagnosing carbon allocation. Our findings have implications for improving terrestrial ecosystem models by replacing fixed allocation coefficients with formulations that incorporate sink controls, lags, and storage dynamics. This work demonstrates that harmonizing multi-scale observations can uncover robust linkages between photosynthetic carbon supply and wood production, advancing our understanding of biosphere–atmosphere interactions under climate variability.