A new dataset of soil Carbon and Nitrogen stocks and proﬁles from an instrumented Greenlandic fen designed to evaluate land-surface models

. Arctic and boreal peatlands play a major role in the global carbon (C) cycle. They are particularly efﬁcient at sequestering carbon due to their high-water content which makes primary productivity exceed decomposition rates . Though, their :::::: because ::::: their :::: high ::::: water :::::: content ::::: limits :::::::::::: decomposition ::::: rates :: to ::::: levels ::::: below :::: their ::: net ::::::: primary ::::::::::: productivity. :::: Their : future in a climate-change context is quite uncertain in terms of carbon emissions and carbon sequestration. Nuuk-fen site is a well-instrumented greenlandic site of particular interest for testing and validating land-surface models with 5 monitoring of soil physical variables and greenhouse gas ﬂuxes ( CH 4 and CO 2 ) ::: and :: is ::: of :::::::: particular :::::: interest ::: for :::::: testing :::: and :::::::: validating :::::::::: land-surface ::::::: models. But knowledge of soil carbon stocks and proﬁles is missing. This is a crucial shortcoming for a complete evaluation of models, as soil carbon

As some additional minor rephrasing and rewriting have been done following the inputs of professional English writers, it is possible than some English corrections addressed to reviewers concerning the English text have been modified since responses to reviewers have been made.
Here are listed the major changes addressed to the manuscript : • A thorough check of scientific English use.
• A rewriting of the abstract and introduction following reviewers comments.
• In particular, we added paragraphs on greenlandic fens characteristics and on the lack of nitrogen data, as asked by reviewer #7. • According to reviewer #7, manuscript have been rewritten by putting an emphasis on Nitrogen stocks, neglected in the first version of the paper. Two subsections have been added in the Results concerning nitrogen stocks and profiles. • Figure 1 is improved by adding coordinates, proper length bars, compass roses, and highlighting zones of interest of the fen (automatic chambers, flux tower, soil temperature probes) • Methods and Materials section has been fully reorganized according to ESSD criterions and some reviewers comments to improve the logical flow and readability of the paper • Same for Results section.
We want to thank the different reviewers for their inputs that greatly helped improving the manuscript quality, and we hope the changes made will convince the reviewers and the editors that the manuscript reached sufficient quality for publication.
The first transect (T1) roughly follows a N-S axis. The automatic chambers are situated on either side of the transect between 5 plots T 1 0 and T 1 20. The second transect (T2) starts at the last automatic chamber, at the 20 meter plot of the first transect, in the middle of the fen and goes through the fen in its larger axis. The soil temperature probes can be seen between plots T 2 30 and T 2 45.
3.1 Physical site measurement ::: Fen :::::::: physical ::::::::::::: characteristics First, we investigated :::::::: measured : the topography of the fen and the depth of the sediment layer that delimits organic and mineral 10 soil horizons at every plot for both transects. The elevation is ::: was : measured with a topographer rod. The depth of the organicmineral interface (OMI) is ::: was measured with a rigid metallic probe. The probe is ::: was lowered into the ground until a strong resistance, characteristic of mineral soils, is ::: was : encountered.
Soil carbon density ⇢ C (gC.m 3 soil ) was then computed as : Similarly, soil nitrogen density ⇢ N (gN.m 3 soil ) was computed as : 8 A total of n = 135 samples were collected along both transects (n 1 = 65 and n 2 = 70). For each of these samples, values of mass, volume, density, dry mass, bulk density, carbon and nitrogen content (%) and density (kg.m 3 ), and carbon-nitrogen (C/N) ratios were measured and/or calculated. Figure 5 shows distribution histograms for all data, and descriptive statistics (mean, median, upper and lower deciles) are presented in Table 2.
5 Figure 6 presents mean soil profiles of bulk density, water mass fraction, carbon and nitrogen content and density along both transects .
These soil profiles (n = 17) were averaged over depth for both transects , and are presented in Figure 7 :::::: (Figure :: 7). As the 10 fen depth has :::::: showed : a substantial variability along the transects, resulting averaged profiles are noisy. For instance, samples extracted at 50 cm depth may be in a very ::::: purely organic soil horizon or a quasi-mineral one depending on the fen area it was extracted from. Hence, mean soil profiles do not necessarily reflect the vertical distribution of data with respect to the OMI.
To reduce the noise due to the OMI heterogenoity ::::::::::: heterogeneity, we renormalized all the data with respect to the OMI. For 15 a sample extracted at a depth z from a peat core with a : an : OMI depth z OM I ; , : we define its normalized distance from OMI d OM I (%) as : These normalized profiles are shown in the figure 8. ::::: Figure :: 8.

Bulk Density
Variation in bulk density is attributable to the relative proportion of organic and inorganic soil particles, and is a reliable indicator of the mineral or organic nature of a soil. More than 50% of the samples have a bulk density below 0.187 g.m 3 (Figure 5), characteristic of organic-rich material. Samples with bulk density between 0.5 and 1 g.cm 3 corresponds :::::: g.cm 3 :::::::::: correspond to mixed organic-mineral material ( (Loisel et al., 2014). The higher the bulk density, the higher the mineral content. Finally, 15 the 10% remaining samples with bulk densities higher than 0.978 g.cm 3 (Table 2) correspond to the most mineral part of the soil :::: with ::: the :::::: highest ::::::: mineral ::::::: fraction, near or below the OMI, as most mineral soils have bulk densities between 1.0 and 2.0 g.cm 3 (Rezanezhad et al., 2016).
Typical bulk density profiles in peatlands tend do : to : show a gradual increase with depth Quinton et al. (2000) :::::::::::::::::: (Quinton et al., 2000) : as peat decomposition reduces the proportion of large pores by breaking down plant debris into smaller fragments (Rezanezhad et al., 2016), it increases the mass of dry material per volume of peat. Normalized mean profiles of bulk density (Figure 8.b) 25 clearly shows this abrupt transition from mixed organic-mineral material to fully ::: pure : mineral soil below the OMI. ( Figure ::: 6.c), which is coherent with the proportions ::: data : given in Yu (2012).

Carbon mass percentage
As expected, concentration of soil organic carbon in the organic layer is ::: was much higher than in the mineral horizons. High carbon content in the depth of the first transect seems :::::: seemed to indicate a carbon burial in the natural accumulation basin. We also note that the limit between the soil horizons with high carbon content and low carbon content also follows the OMI. In particular, the drop of the sedimentary layer in the first transect is clearly visible, and the variations of the mineral layer of the second transect between T 2 30 and T 2 60 meters as well. Normalized mean carbon content profiles (Figure 8.e) clearly shows ::::::: showed the abrupt decrease in carbon content near the OMI. Below the OMI, carbon contents value are :::::: content :::::: values 5 :::: were below 10 %, which is coherent with the mineral characteristics of the soil horizons below.

Soil carbon density and soil carbon profiles
More than 70% of the soil samples have :: Mean local maximum of soil carbon density can reach :::::: reached ::::: value :: up :: to : 80kg.m 3 : at ::: 80 ::: cm ::::: depth (Figure 7.f). One sample 10 has a particularly high carbon density of 160kg.m 3 (Figure 9.b). This high value may be due to a bad sample handling during the extraction or manipulation, resulting in a sample compaction that artificially increased measured bulk density.
As expected, soil carbon density matches ::::::: matched well the measured organic-mineral interface (Figure 6.e1). The alleged ::::::: assumed carbon accumulation in the accumulation basin on :: of the fen discussed in the previous section is confirmed, as a local maximum of soil carbon density is ::: was : clearly visible at the bottom of the soil plots T 1 25 and T 2 30 (Figure 6.a1). 15 Mean soil carbon density profiles are :::: were : non-monotonous. In the organic horizons, SOC density increases :::::::: increased with depth and reaches :::::: reached : its local maximum between 60 and 80 % of the organic-mineral interface depth (Figure 8.f). Near the OMI, coherently ::::::: coherent with the abrupt decrease in carbon content and increase in bulk density discussed in the previous sections, the soil carbon density decreases :::::::: decreased. Soil carbon density profiles that first increases then decreases :::::::: increased

Integrated soil carbon stocks
For each peat core, total carbon stocks C T (kgC.m 2 ) were calculated by vertically integrating carbon density profiles using the trapezoïdal rule : with z j the sample depth and ⇢ Cj the soil carbon density, calculated using equation (2).
Note that because of the difficulties setting the manual gouge auger much below the mineral-organic interface, the maximum sampling depth varies between the different peat cores. Hence, the integration depth also varies between peat cores. However, the carbon content below this interface does not exceed 7 % except for two unusual samples ( Figure 9) and we can consider that not taking into account the soil horizons below the mineral-organic interface does not underestimate much the calculated 30 total carbon stocks.

C/N Ratios
Carbon/Nitrogen ::::::: nitrogen (C/N) ratios can give useful information about the nutrient content and the quality and humification degree of organic matter : a low C/N ratio is usually equivalent to a high humification level. With a mean value of 21.6, observed C/N ratio are :::: ratios ::::: were in the range of those observed from a variety of field and laboratory studies (Bridgham et al., 1998;Rezanezhad et al., 2016;Wang et al., 2015).
C/N ratios are ::: were : higher in the first centimeters ::: few ::: cm : depth (approx. 25%), potentially indicating less microbial transformation of the peat in the upper layers (Kuhry and Vitt, 1996). In the :: At :::::: greater : depth of the fen, C/N ratio are :::: ratios ::::: were lower because microorganisms slowly consume the carbon and recirculate the nitrogen, resulting in a gradual reduction of C/N values (Rydin and Jeglum, 2013). In northern regions, due to colder temperatures, the decomposition activity is slow, explaining the small difference between maximal and minimal C/N values. The C/N profiles stay :::::::: remained : relatively stable throughout the 5 depth (21.6%) and the OMI does :: did : not seem to distinguish separate zones.
Although bulk density and C/N ratio :::: ratios : are reliable indicator for peat degradation, the lack of ash content data and isotopic measurements does ::: did not allow a quantificiation of carbon accumuluation rate nor carbon loss in the peatland (Krüger et al., 2015). Overall, this new dataset of soil bulk density, carbon and nitrgogen :::::: nitrogen : content, profiles and stocks is :::: were in the range of previous estimates (Yu, 2012;Loisel et al., 2014;Chimner et al., 2014).

Discussions
As noted by Loisel et al. (2014), the accuracy of this type of measurements ::::::::::: measurement mostly depends on sample handling, 15 in particular the care deployed :::: taken : to avoid any peat compaction. Our sample density measurements may be :::: have ::::: been uncertain. On the other hand, mass carbon percentage ::::::::: percentages : are independent of any compression or any physical aleas.

Calculation of 95% confidence interval soil profile
The sampling mean most likely follows a normal distribution. Under this hypothesis, the standard error of the mean (SEM) can be calculated as X (z) = (z) p N (z) with N(z) the number of samples collected at a depth (z) and (z) the standard deviation 5 over those samples. The confidence interval at 95% is defined as I(z) = X(z) ± 1.96 ⇥ X (z).
Author contributions. XM, BUH, MM, CD and BD designed the field campaign. XM and BUH conducted the field work, collected and prepared the samples for analysis. PA conducted the laboratory analysis. XM, CD and BD designed the manuscript. All authors contributed to the writing.
Competing interests. The authors declare that they have no conflict of interest.