Oaks Keep Breathing After They Stop Growing — A Climate-Model Reckoning

TL;DR

• The finding. Rao et al., Science Advances (June 2026, DOI 10.1126/sciadv.ady7139) — 137 tree-ring sites across 8 Quercus species show photosynthesis (carbon uptake) and wood growth (carbon storage) are decoupled at the seasonal and annual scale, with correlation to vapour-pressure deficit variability at r=0.86.
• Why it matters. Climate models, IPCC assessments, and voluntary carbon markets largely assume the two track together. They don't. Wood-based carbon storage is the durable pool — leaves, roots, and respiration release most uptake back within months to years.
• Scale of the gap. The paper doesn't put a single global number on it, but the mechanism implies temperate broadleaf forest carbon-sink estimates may be overstated by a materially non-trivial margin — the direction is clear, the magnitude is the next research question.
• The quieter story. Timing of drought matters more than total precipitation. A wet year with a dry June can produce less wood growth than a drier year with well-timed rain — a finding that breaks how most vegetation models are parameterised.
• What's affected. Nature-based carbon credits (Verra, Gold Standard, ACCUs), corporate net-zero pathways relying on afforestation offsets, national GHG inventories using biomass-growth proxies, and any climate model treating GPP as a wood-storage proxy.
• What it isn't. Not a claim that forests aren't carbon sinks. Not a Southern Hemisphere finding — no Eucalyptus data. Not a replacement for direct biomass measurement, which is exactly what the paper argues we now need more of.
• Bottom line. If your climate strategy, portfolio, or policy leans on modelled forest carbon uptake as a storage proxy, that assumption just got weaker. Ask for measured biomass, not modelled GPP.

Between 26% and 36% of the carbon that oak trees pull out of the atmosphere every year arrives after the tree has finished making new wood for the season. The photosynthesis keeps running; the growth has already stopped. That gap — quiet, seasonal, and almost entirely absent from the equations underneath most climate projections — is the reason a paper by Mukund Palat Rao and 27 collaborators is spreading through science media this week and unsettling a lot of people who model forests for a living.

The paper appeared in Science Advances (vol. 12, issue 24, DOI 10.1126/sciadv.ady7139) and its findings are being amplified through a fresh ScienceDaily wave that broke overnight. The core claim is deceptively small: in oaks, carbon uptake and wood growth are not the same process, and they do not obey the same climate signals. The implications are not small at all.

What the team actually did

Rao's group — anchored at Columbia's Lamont-Doherty Earth Observatory but drawing in NASA JPL, UC Davis, CREAF Barcelona (Josep Peñuelas), and a long roster of tree-ring labs across North America — layered four independent measurement systems on top of the same forests:

The species set covered eight Quercus — alba, bicolor, montana, rubra, coccinea, palustris, falcata across the eastern United States, and douglasii in California — spanning humid temperate and Mediterranean climate regimes. The four high-frequency sites contributed seven site-years of matched flux-and-growth data, which is the empirical spine of the paper.

Multi-scale, multi-method, multi-region. That is what makes the result hard to argue with.

The finding, in one paragraph

Oaks in the eastern US grow between May and July. Their leaves keep photosynthesising until October. Californian Q. douglasii grow between December and April, cease by August, and photosynthesise into the following autumn. In both regions, roughly a third of the year's carbon uptake happens after the tree has already finished making new wood. That post-growth carbon does not disappear — it goes into foliage, roots, defensive chemistry, non-structural carbohydrates (starches and sugars stored as reserves), and metabolic maintenance to keep cells alive through winter. Some of it is used to jump-start growth the following spring. Very little of it becomes new trunk.

Why this breaks a modelling assumption

Most Earth System Models (ESMs) — the machinery underneath IPCC scenarios, national carbon budgets, and every credible net-zero pathway — treat photosynthesis and above-ground growth as tightly coupled. Gross primary productivity (GPP) goes up, wood mass goes up, long-term carbon storage goes up. This is the mechanical basis of the "CO₂ fertilisation effect" that appears, in various strengths, inside CMIP6-generation models and downstream integrated assessment models.

Rao's data say the coupling is conditional, seasonal, and increasingly leaky as the climate becomes more variable. Specifically:

• Wood formation requires turgor pressure. Cells build new xylem by expanding under internal water pressure. When the atmosphere becomes hot and dry — high vapour-pressure deficit (VPD) — trees lose that pressure and growth stops within hours.
• Photosynthesis is more tolerant. Leaves can keep fixing carbon at a slightly reduced rate even as growth halts. The environmental window for capturing carbon is wider than the window for turning it into wood.
• The decoupling worsens with variability. Across sites, the strength of the photosynthesis-growth decoupling correlates with inter-annual VPD variability at r = 0.86, p < 0.05. The more the local climate whipsaws between wet and dry, the more carbon the tree captures without converting to wood.

That last point is the one that should make policymakers pause. Climate variability is the direction of travel. The models assume the coupling holds; the observations say the coupling weakens exactly when variability rises.

The mechanism, made visible

The pathway on the left of the fork is what climate models assume dominates. Rao's team is saying the pathway on the right is 26–36% of the annual flux — and rising with climate variability.

The stakes for carbon accounting

This is where the paper stops being a botany story and starts becoming a financial one. Three specific pressure points:

1. Forest carbon-credit methodologies. Most nature-based-solution credit protocols — Verra VCS, Gold Standard AFOLU, ART TREES — either assume or default to biomass-growth-based sequestration proxies. If a meaningful fraction of measured GPP does not enter woody biomass but instead cycles back through leaves, roots, and respiration, then the durable, credit-worthy fraction of measured carbon uptake is lower than baseline assumptions suggest. The direction is unambiguous; the magnitude is what the market now needs to price.

2. National inventory reporting. Countries that lean heavily on their forest sink in NDC accounting — the US, EU-27, Canada, Russia, Brazil, Australia, and China all do — use ESM-derived or empirically-calibrated sink estimates. The paper's own conclusion is blunt: "by assuming tight coupling between photosynthesis and woody biomass, current earth system models may overestimate long-term carbon sequestration in forests."

3. Corporate net-zero pathways. Any pathway that relies on natural-climate-solution offsets to close a residual-emissions gap now carries a modelling-basis risk that was not on most Boards' radars. The finding does not invalidate forest carbon; it re-prices it.

The quieter story wedge

Read the paper carefully and there is a second, less-quoted finding that matters. Rao's team notes that growth is insensitive to climate variability after mid-summer. In other words, once the growth window closes, no amount of favourable weather in August or September makes up for a bad June. The wood-building season is short, water-limited, and non-recoverable. This makes forest carbon storage far more sensitive to timing of drought than to total precipitation — a subtlety that most annual-averaged model runs cannot see.

For anyone modelling forest sinks under future scenarios, this reframes the question from "how much rain will there be" to "when will the dry stretches fall relative to the growth window." Those are different questions with different answers.

Uncertainty ledger — what this study does not prove

Intellectual honesty matters here. Five things the paper deliberately does not claim:

1. It is not a global result. The empirical base is North American oaks — a single genus in two climate regimes. Extrapolation to conifers, tropical hardwoods, or boreal forests is not established. Rao acknowledges this.
2. The magnitude of overestimation in ESMs is not quantified. The paper flags the direction (models likely overestimate long-term sink) but does not put a percentage on it. That work is next.
3. Some post-growth carbon does become wood the following year via non-structural carbohydrate reserves. The 26–36% figure is not a pure loss to short-cycle carbon — some fraction is deferred, not diverted.
4. CO₂ fertilisation is not disproven. It is shown to be less efficient at producing durable wood than the simple coupling assumption implies.
5. Field-measured decoupling ≠ model correction. ESM developers will need to build turgor-limited growth submodels before any of this shows up in projections. That is a multi-year exercise.

What to watch over the next six to eighteen months

• ESM community response. Expect a wave of follow-up papers testing whether the decoupling holds in conifers (Beier, Peñuelas groups already positioned) and tropical species (Amazon FACE network).
• Voluntary carbon market methodology revisions. Verra and ART have both signalled openness to revised biomass-vs-GPP treatment; this paper will accelerate that conversation.
• IPCC AR7 drafting. The first-order draft opens in 2027. Rao et al. is exactly the kind of high-signal, single-mechanism finding that gets pulled into WG1 Chapter 5 (carbon cycle).
• Australian relevance. No Quercus data from the Southern Hemisphere in this paper. The equivalent question for Eucalyptus — Australia's dominant carbon-sink genus, with a very different water-use physiology — is genuinely open and would be a high-impact PhD or postdoc programme for a well-resourced Australian group.

The sentence to remember

Photosynthesis is not growth. Growth is not storage. And the assumption that one leads cleanly to the next — quietly embedded in every net-zero pathway we currently believe in — just got a lot harder to defend.

Sources

• Rao, M. P. et al. (2026). Decoupled carbon assimilation and growth responses to aridity in temperate deciduous oaks. Science Advances 12 (24), eady7139. DOI: 10.1126/sciadv.ady7139
• Data & code: Dryad, DOI 10.5061/dryad.m63xsj4h4
• ScienceDaily wave, 9 July 2026 (secondary-coverage surge)
• Columbia State of the Planet, Phys.org, EurekAlert, Wood Central (June 2026 primary coverage)