The research, led by Ngoc B. Nguyen of the University of California, Berkeley, together with Miao Zhang and Trevor F. Keenan, set out to resolve a fundamental ambiguity in how ecologists read their own data. Plant carbon uptake depends on the balance between carbon sources, chiefly photosynthesis, and carbon sinks, such as respiration, storage and the construction of new tissue. How much carbon dioxide a canopy absorbs is one thing; how much of that absorbed carbon ends up as lasting woody biomass is another. Previous analyses combining eddy-covariance flux measurements with tree-ring records had found that, on interannual timescales, the two often diverge, a pattern frequently read as evidence for widespread sink limitation, meaning that growth is constrained by factors other than carbon supply, such as temperature, drought or nutrient availability acting directly on cell division and expansion.

To test whether that inference was justified, the team assembled two independent lines of evidence. The first came from long-term observations at eddy-covariance flux towers, instruments that continuously measure the exchange of carbon dioxide between ecosystems and the atmosphere. The researchers drew on gross primary productivity data from the FLUXNET2015 and AmeriFlux networks, paired with on-site tree-ring width measurements at 31 forest sites, allowing them to compare annual photosynthetic carbon uptake with annual woody biomass growth at the same locations. The second line of evidence came from a global ensemble of vegetation models drawn from the TRENDY project, the model intercomparison that underpins the annual Global Carbon Budget, supplemented with detailed output from the Community Land Model version 5.0.

The results confirmed the puzzling pattern that had troubled the field. Across the observations, and in more than half of the ensemble models, the short-term correlations between annual gross primary productivity and the annual woody carbon biomass increment, a quantity the authors denote rΔcWood,GPP, were weak. In other words, a year in which a forest photosynthesized unusually well was often not a year in which it laid down conspicuously more wood, and vice versa. This replicates the decoupling previously reported in the literature and validates the observational signal that had motivated the sink-limitation hypothesis. Had the story ended there, the new study would simply be a confirmation. But it did not end there.

When the authors zoomed out from year-to-year fluctuations to the long arc of recent decades, a strikingly different picture emerged. Between 1951 and 2020, woody carbon biomass increased in proportion to photosynthesis across both the observations and the models. Over these seventy years, the fraction of photosynthesized carbon ultimately converted into lasting wood remained broadly consistent, and forests that took up more carbon over the long run accumulated proportionally more biomass. The decoupling that dominates annual comparisons simply vanished at longer timescales. Perhaps most tellingly, there was no relationship between the strength of the short-term correlation, rΔcWood,GPP, and the long-term sensitivity of biomass to photosynthesis across the model ensemble. A model could exhibit near-zero annual coupling and still show robust, proportional long-term biomass gains from enhanced photosynthesis.

This disconnection between short-term correlation and long-term sensitivity has a logical foundation rooted in how carbon moves through a tree. Photosynthesized carbon does not immediately become wood. It passes through a labyrinth of intermediate pools: sugars in the phloem, nonstructural carbohydrate reserves stored in stems and roots, and allocation decisions that shift with season, water status and phenology. Wood formation itself proceeds on its own schedule, governed by cambial activity that is sensitive to temperature and water potential and that can lag carbon uptake by weeks or months. In any given year, drought can suppress cell expansion while photosynthesis continues at a reduced but nonzero rate, or a favorable growing season can extend cambial activity long after the peak of carbon assimilation has passed. These timing mismatches, storage buffers and allocation lags naturally weaken annual correlations without implying that the overall supply of carbon fails to set the ceiling on how much wood can eventually be built.

The practical implications of this reframing are substantial. If weak annual correlations between photosynthesis and growth do not indicate sink limitation, then the widespread sink limitation inferred from such correlations in earlier studies is probably less prevalent than previously reported. That matters because the source-versus-sink debate sits at the heart of projections of the terrestrial carbon sink, the portion of human carbon emissions that land ecosystems absorb each year. Global carbon budgets assume that rising atmospheric carbon dioxide fertilizes photosynthesis and thereby increases carbon storage in vegetation, a mechanism constrained by evidence from free-air carbon dioxide enrichment experiments and global syntheses. If tree growth were predominantly sink-limited, extra photosynthetic carbon would simply accumulate unused or be returned to the atmosphere, weakening expectations for future land carbon sequestration. The new findings suggest instead that long-term biomass accumulation does track carbon supply, lending support to the view that photosynthetic uptake remains a meaningful constraint on how much carbon forests can bank.

The study also carries a methodological warning for the field. Observational records that link flux towers to tree rings are invaluable, but the authors caution against inferring source-sink control directly from them. A correlation measured at one timescale cannot be straightforwardly extrapolated to another, particularly when the underlying biology involves storage pools and time lags that scramble year-to-year signals. Integrated perspectives on plant carbon balance have increasingly emphasized that source and sink processes interact continuously rather than one simply commanding the other, and the new analysis fits comfortably within that framework: sinks matter for the timing and routing of carbon, but over decades the source still sets the budget.

There remain open questions. The models themselves diverge in how they represent wood formation, allocation and biomass turnover, and more than half of the ensemble reproduced the observed decoupling while others did not, highlighting persistent structural uncertainties in how global vegetation models handle carbon sinks. Branch turnover, mortality and disturbance also shape how much photosynthesized carbon survives as lasting biomass, and biases in forest carbon accounting remain an active area of concern. Still, the central message of the study is clear and consequential: the noisy annual conversation between photosynthesis and growth conceals a much steadier long-term relationship. Forests, over decades, do convert additional carbon uptake into additional wood, and the short-term decoupling that has intrigued and confused ecologists is not evidence that the global carbon sink has hit a growth-imposed wall.

Subject of Research: The decoupling of short-term photosynthetic carbon uptake from woody biomass growth in forests and its implications for source-sink limitation of the terrestrial carbon sink.

Article Title: Long-term biomass growth unimpeded by short-term photosynthetic decoupling

Article References: Nguyen, N. B., Zhang, M., & Keenan, T. F. (2026). Long-term biomass growth unimpeded by short-term photosynthetic decoupling. Nature Plants. https://doi.org/10.1038/s41477-026-02418-1

Image Credits: AI Generated

DOI: 10.1038/s41477-026-02418-1

Keywords: forest carbon, photosynthesis, sink limitation, biomass growth, eddy covariance, tree rings, TRENDY models, gross primary productivity, terrestrial carbon sink, carbon allocation, Nature Plants, wood formation