Highly productive ecosystems on land and at sea provide natural carbon (C)-sinks. Loss of such sinks due to deforestation and land use changes have resulted in 32% of the accumulated global CO₂ emissions^a,b. Marine forests, which contribute to oceanic C-sinks (blue C), are also under global threat due to multiple pressures. However, in the Arctic, these forests and their related C-sequestration may expand with warming and reduced sea-ice cover^c. Macroalgal forests have typically been ignored as C-sinks, as they mainly grow on rocky shores where organic matter does not accumulate. Yet, macroalgae are C-donors to sinks beyond their habitat, and a first-order global estimate suggests that their contribution is substantial^d. However, there are huge knowledge gaps in the distribution, production, and sequestration potential of macroalgae, especially in the vast and inaccessible Arctic.

The aim of CARMA was to test the main hypothesis that burial of C exported from Arctic marine forests provides a globally relevant, and growing, contribution to C-sequestration in marine sediments. The project posed three key questions:

• Q1. Is C from marine forests present in Arctic marine sediments and, if so, what percent of the stock do they contribute?
• Q2. Are there latitudinal and/or temporal trends in the contribution of marine forest to C stocks?
• Q3. How much C do Greenland marine forests sequester?

CARMA applied an interdisciplinary approach at the intersection of marine ecology, geology, oceanography, remote sensing, biochemistry and genetics, with Greenland as a case study. We coupled conceptual advances with emerging techniques, in particular environmental DNA (eDNA) for tracing C from marine forests in Arctic marine sediments and used field observations, remote sensing and model approaches to identify the distribution of Arctic macroalgae, their potential C-export to oceanic sinks and to predict changes. To explore climate-related patterns in the contribution of marine vegetation of Arctic C-sinks, we analysed marine sediments and vegetation along Greenland/Arctic climate gradients, analysed deep sediment cores from the West Greenland coast as historical archives to explore long-term changes in fingerprints of macroalgae in relation to climate proxies, and explored trends in pan-Arctic marine vegetation. CARMA involved primary collaboration between excellent research environments in Denmark, Saudi Arabia, and Germany and expanded the collaboration to a larger international forum during the project period. This scientific environment made CARMA a unique framework for training young scientists.

Developing the eDNA fingerprinting tools for macroalgae in marine sediments was more demanding than expected, and whereas metabarcoding methods proved successful to identify macroalgae to order-level, they did not allow proper quantification of macroalgal DNA. This requires quantitative polymerase chain reaction (qPCR) assays, which were developed by the end of the project for future use. Analyses of amino acid isotopes could also not be completed within the project period but are underway. The relative contribution of Arctic macroalgae to the pool of organic matter in the sediments was instead coarsely quantified based on stable isotopes of sediment organic matter and supplemented by top-down estimates of macroalgal distribution, trends, production, and associated estimates of potential export to C-sinks.
This latter approach was explored further than anticipated and, in addition to the Arctic angle, we engaged in a larger international collaboration to also develop global estimates of macroalgal distribution, production, potential contribution to C-sinks, and to evaluate opportunities and limitations in using macroalgae for climate change mitigation. We thereby reached goals beyond the initial aims and expanded our network.

Marine vegetated habitats in the Arctic/Greenland: CARMA estimated the area distribution of marine vegetation at local- to Greenland- and pan-Arctic scale^1-10. Local-scale distribution was assessed by remotely operated vehicle (ROV) and remote sensing (drones, satellites) for the vast Nuup Kangerlua fjord system in SW Greenland⁷ while species distribution models were used for assessing the potential distribution area of dense kelp forests (>50% cover) at the Nordic scale⁶ and of potential habitats of brown macroalgae and eelgrass at pan-Arctic scale⁹ and at global scale¹¹.
For Greenland, the distribution area of dense kelp forests was modelled at 1304 km^{2 6}. The modelled suitable habitat area of brown macroalgae was markedly larger and varied with modeling approach and resolution. The most updated estimate (based on stacked distribution modeling of individual species and corrected for % rocky coastline) was 11,007 km² with an associated production of 362.9 g C m^-2 y^-1, totaling 4.0 TgCy^{-1 15}. CARMA also identified recent increasing trends in pan-Arctic vegetation based on observational evidence supported by modeling⁹, and the same models predicted continued increases through this Century¹².

Export of macroalgae to C-sinks: CARMA documented large-scale export of floating macroalgae from coastal habitats based on direct observations and satellite imagery. Tracking of export pathways from a large Greenland fjord system suggested that most of the exported algae was retained within the fjord with a fate of degradation, grazing⁸ or sequestration while a small fraction was exported offshore to potentially reach C-sinks in the deep sea⁷. Pathways of floating exported macroalgae off the coast of Greenland were further traced by GPS-marked buoys, satellites, and modeling in the Baffin Bay area, where novel vertical exchange patterns were documented³. CARMA also contributed to a global model of potential sequestration of macroalgal C in the deep sea based on modelled macroalgal production along the global coastline coupled to an ocean model tracking the export of macroalgae and on this basis estimating the transport time to reach beyond the shelf for potential sequestration in the deep sea and the likelihood that macroalgae would reach this sink before they were degraded. Overall, modelled global export of macroalgal POC beyond the shelf was 56 (1-170) TgCy^-1 of which 4-44 TgC y^-1 (1.1-11.8% of NPP) could be retained for >100 years¹⁵. For Greenland, the matching export of macroalgal POC beyond the shelf is 0.98 (0.28 - 2.6) TgC y^-1 with 0.04-0.47 TgCy^-1 potentially retained for >100 years¹⁵. These estimates do not include sequestration of macroalgal particles on the shelf or of dissolved C (DOC). Applying the 1st order estimate of total global macroalgal C-sequestration (avg. 11% of NPP)a yields 0.62 TgC for Greenland. These coarse estimates are our best reply to Q3.

Macroalgal blue carbon: We developed an eDNA fingerprinting tool to trace the contribution of marine forests to C-pools¹⁶ in sediments^17-19. The tool consists of an 18S ribosomal DNA mini-barcode for marine macrophytes and an associated 18S reference library^18,19, including a specific library and testing for key Arctic/North-Atlantic macroalgal species¹⁹. The tool documented macroalgal DNA in 94% of all (89) surface sediment samples distributed around Greenland and Svalbard from the shore to 1460 m depth and to 350 km offshore. The abundance of macroalgal sequences was highest near-shore and dominated by brown algae²⁰. The tool also traced millennial-scale burial of macroalgae in six dated sediment cores distributed along a N-S gradient (79°N-60°N) on Greenland’s West Coast where especially brown macroalgal fingerprints were documented in 86.5% of all samples and dated 2600 years back²¹. The isotopic signal of the organic matter of surface sediments gave a coarse indication of a significant relative contribution of macroalgae to the pool of organic matter (range 21-85%)²⁰. Overall, in reply to Q1, these results underline the prevalent presence of macroalgae in Greenland marine sediments and the potential for long-term storage. But better tools are needed for quantification. In reply to Q2, we did not identify latitudinal- or temporal trends in the contribution of macroalgae to C-pools.
In conclusion, CARMA advanced the knowledge on C-cycling of Arctic marine forests by

1. providing the first quantitative estimates of the potential distribution of Arctic marine forests and their increasing trends,
2. evidencing and documenting export pathways of Arctic macroalgae to potential C-sinks,
3. documenting the prevalent presence of macroalgal C in Arctic surface sediments and millennial-scale preservation of macroalgal C in deeper sediments via new eDNA fingerprinting methods and
4. providing a coarse estimate of potential C-sequestration by Greenland’s marine forests.

These findings partly supported the main hypothesis. Better quantification of the contribution of Arctic macroalgae to C-sequestration is, however, still needed. Although Arctic macroalgae contribute to C-sequestration, increases in their C-burial due to climate change is not eligible as a climate change mitigation strategy, which requires that the increase is linked to management, but the findings can guide the management of these important habitats.

The data, concepts and tools developed by CARMA can be taken to the next level: Upscaling of mapping of tidal- and floating macroalgae to Greenland scale; refinement of macroalgal species distribution models and predictions, better quantification of macroalgal DNA by qPCR through sediment cores to quantify changes over time and experimentally testing links between macroalgal DNA and OC during degradation. Sequestration of macroalgal DOC is also a major unknown.