Add organic carbon to soil and you might lose more carbon than you add. This is the priming effect: fresh, easily metabolized organic inputs — root exudates, crop residues, compost — stimulate microbial activity so strongly that the microbes don’t just consume the new material. They consume the old material too. The locked-up carbon that had been stable for decades or centuries becomes food, because the fresh input provides the energy the microbial community needed to break down the recalcitrant stock.

The accounting is perverse. A farmer adds a tonne of compost. The microbes eat the compost and, now energized, eat two tonnes of soil organic matter that would otherwise have stayed put. Net result: minus one tonne. The intervention intended to sequester carbon released it.

The mechanism is metabolic co-activation. Soil organic matter is not chemically inert — it is merely energetically unfavorable to decompose. The stable carbon is stable because breaking it down requires more metabolic investment than microbes can sustain from the recalcitrant substrate alone. Add a labile energy source and the investment becomes affordable. The fresh carbon subsidizes the decomposition of the old. The microbes that produce metabolites increasing organic carbon retention in some soils can decrease it in others, depending on whether the labile input primes more decomposition than it contributes.

The through-claim: stability is not a property of the material but of the energetic context. The same soil organic molecule is stable when the surrounding microbial community is energy-limited and unstable when that community is subsidized. What changed was not the molecule but the energy landscape around it. Carbon sequestration strategies that ignore this — adding organic matter without understanding the priming dynamics of the receiving soil — can accelerate the problem they were designed to solve.

The trap is in the logic: more input, less stock. The relationship between addition and accumulation runs backward when the addition unlocks what was already there.