Learn

1. The land sink and net primary production

Net primary production (NPP) is the carbon plants and other photosynthesisers fix, minus what they respire themselves. A widely cited estimate puts global NPP at 104.9 Pg C per year, with “roughly equal contributions from land and oceans”. Source: Field et al. (1998), doi:10.1126/science.281.5374.237 · independent peer-reviewed paper

Most of that carbon returns to the air within days to decades as plants, animals and microbes respire. What matters for the climate is the small net difference that stays on land — the land sink.

• Global Carbon Budget 2024: the land CO₂ sink was 3.2 ± 0.9 GtC/yr during 2014–2023, rising from 1.2 ± 0.5 GtC/yr in the 1960s, with interannual variations of up to 2 GtC/yr, and a weaker sink in El Niño years. In 2023 it was 2.3 ± 1 GtC/yr. Source: Friedlingstein et al. (2025), Global Carbon Budget 2024, doi:10.5194/essd-17-965-2025 · peer-reviewed community synthesis (not independent of IPCC AR6)
• IPCC AR6: over 2010–2019, 31% of the carbon emitted by human activities “was stored by vegetation in terrestrial ecosystems (3.4 ± 0.9 PgC yr⁻¹)”. Source: Canadell et al. (2021), doi:10.1017/9781009157896.007 · IPCC assessment (AR6 WGI, Chapter 5)
• At the same time, land-use change releases carbon: 1.6 ± 0.7 PgC/yr in 2010–2019 (AR6, “net land-use change”) and 1.1 ± 0.7 GtC/yr in 2014–2023 (GCB 2024, “LULUCF”). Source: Canadell et al. (2021), doi:10.1017/9781009157896.007 · IPCC assessment (AR6 WGI, Chapter 5) Source: Friedlingstein et al. (2025), Global Carbon Budget 2024, doi:10.5194/essd-17-965-2025 · peer-reviewed community synthesis (not independent of IPCC AR6)

The two land-sink figures cover different decades and use different methods; they are shown side by side, not combined.

2. Soils and peatlands

Soils hold far more carbon than the plants growing on them. One global analysis estimated soil organic carbon at 1,502 Pg C in the top metre and 2,344 Pg C in the top three metres. Source: Jobbágy & Jackson (2000), doi:10.1890/1051-0761(2000)010[0423:TVDOSO]2.0.CO;2 · independent peer-reviewed paper

Peatlands are waterlogged soils where plant remains build up faster than they decay, sometimes over thousands of years. Estimated peatland carbon accumulated since the last ice age:

Source: Yu et al. (2010), doi:10.1029/2010GL043584 · independent peer-reviewed paper

When peat is drained, for example for farming, the exposed organic matter oxidises and releases CO₂:

• More than 25 million hectares of organic soils have been drained for agriculture worldwide; their emissions are “nearly one billion tonnes CO2eq annually”, of which CO₂ is “about 780 million tonnes” (≈ 0.21 Pg C/yr by our conversion¹). Source: Tubiello et al. (2016), doi:10.3390/su8040371 · independent peer-reviewed paper
• Another study estimated contemporary greenhouse-gas emissions from drained peatlands at 1.91 (0.31–3.38) Gt CO₂-eq per year, and that drained peatlands will cumulatively release 80.8 Gt carbon and 2.3 Gt nitrogen. Source: Leifeld & Menichetti (2018), doi:10.1038/s41467-018-03406-6 · independent peer-reviewed paper

The two drained-soil estimates use different scopes (agricultural organic soils vs. all drained peatlands) and units (CO₂ vs. CO₂-equivalents), so they are not directly comparable.

3. Land to ocean: inland waters and rivers

Streams, rivers, lakes and estuaries are not just pipes. They receive carbon from soils, then breathe much of it back out, bury some, and pass the rest to the sea. Researchers call this the land–ocean aquatic continuum.

• Inputs from land. AR6 describes transfers from soils to freshwater systems of “2.4–5.1 PgC yr⁻¹”. Source: Canadell et al. (2021), doi:10.1017/9781009157896.007 · IPCC assessment (AR6 WGI, Chapter 5) A synthesis suggests landscapes “may deliver upward of 5.1 Pg of C annually” to inland waters. Source: Drake et al. (2018), doi:10.1002/lol2.10055 · independent peer-reviewed paper A coupled land–water model estimates loadings of 2,281 Tg C/yr (2.281 Pg C/yr) in the 2010s, up 461 Tg C/yr since the 1800s. Source: Tian et al. (2023), doi:10.1029/2023GB007776 · independent peer-reviewed paper
• Outgassing. Inland waters release about 2.1 Pg C/yr as CO₂ — 1.8 from streams and rivers and 0.32 from lakes and reservoirs. Source: Raymond et al. (2013), doi:10.1038/nature12760 · independent peer-reviewed paper
• Burial. Lakes and reservoirs bury about 0.15 Pg C/yr of organic carbon (0.06–0.25), roughly 40% of it in reservoirs, equal to about 20% of their carbon emissions. Source: Mendonça et al. (2017), doi:10.1038/s41467-017-01789-6 · independent peer-reviewed paper
• Export to the ocean. One reassessment revised the global riverine carbon flux “from 0.45 to 0.78 PgC yr⁻¹”. Source: Resplandy et al. (2018), doi:10.1038/s41561-018-0151-3 · independent peer-reviewed paper AR6 estimates the net export from the terrestrial domain to the open ocean at 0.80 PgC/yr (medium confidence). Source: Canadell et al. (2021), doi:10.1017/9781009157896.007 · IPCC assessment (AR6 WGI, Chapter 5)
• Loops. Before industrialisation, carbon taken up on land, carried to the ocean and outgassed back to the air amounted to 0.65 ± 0.30 Pg C/yr (±2σ). Ignoring these land-to-ocean fluxes overestimates land carbon storage by 0.6 ± 0.4 Pg C/yr. Source: Regnier et al. (2022), doi:10.1038/s41586-021-04339-9 · independent peer-reviewed paper

4. Biological nitrogen fixation, cryptogamic covers and soil crusts

Plants need nitrogen, but the nitrogen gas that makes up most of the air is unusable until certain microbes “fix” it into reactive forms. This biological nitrogen fixation (BNF) helps set how much carbon land ecosystems can take up.

• A global analysis gives a median BNF in natural terrestrial ecosystems of 88 Tg N per year (range 52–130), and finds free-living fixation (not in partnership with plant roots) is at least a third of the total. Source: Davies-Barnard & Friedlingstein (2020), doi:10.1029/2019GB006387 · independent peer-reviewed paper
• Cryptogamic covers — communities of cyanobacteria, algae, fungi, lichens and bryophytes (mosses) growing on soils, rocks and plants (on soil they include biological soil crusts) — take up an estimated 3.84 Pg C per year (90% CI 2.06–7.35; abstract: “around 3.9 Pg”), “around 7%” of terrestrial-vegetation NPP, and have a nitrogen uptake of “around 49 Tg per year” (supplementary best estimate 49.2, 90% CI 26.6–98.6), which the authors suggest is “nearly half of the biological nitrogen fixation on land”. Source: Elbert et al. (2012), doi:10.1038/ngeo1486 · independent peer-reviewed paper

The “nearly half” comparison in Elbert et al. is made against an older global BNF reference (107 Tg N/yr, per the paper’s supplementary table). Against the newer median of 88 Tg N/yr the share would differ; both global totals are uncertain.

Units. 1 Pg (petagram) = 1 Gt (gigatonne) = 10¹⁵ g = 1,000 Tg. “PgC” and “GtC” are the same unit of carbon. Fluxes (per year) and stocks (amounts) are kept in separate tables and never placed on the same axis.

¹ Our conversion, not a figure printed by the source: mass of CO₂ ÷ 3.664 = mass of carbon (44.01/12.011). 780 Mt CO₂ ÷ 3.664 ≈ 213 Mt C ≈ 0.21 Pg C. The source reports total emissions from these soils of “nearly one billion tonnes CO2eq annually”; CO₂-equivalents include other gases and cannot be converted to carbon this way. Back