Linking sediment biodegradability with its origin in shallow coastal environments

. Abstract In coastal areas and estuaries, such as those encountered in the western part of France (Brittany region), the recycling of carbon and nutrients from sediments can participate in the development of micro and macro-algal blooms with harmful consequences for these ecosystems. One of the main processes controlling this recycling is 15 the microbial mineralization of sedimentary organic matter (SOM). Mineralization is controlled by the origin, quantity and accessibility of the SOM, three factors whose relative importance remain, however, poorly quantified, mainly due to the great diversity of OM sources in coastal areas. The first objective of the present work was to assess the variability of the SOM origin at the regional scale representative of the complexity of the sources likely to be involved. The second objective was to determine the link between the SOM origin and its biodegradability, 20 and how the OM sources can drive nutrient dynamics at the sediment-water interface. To this end, a broad sediment sampling campaign was carried out on Brittany mudflats, particularly affected by the eutrophication, during the spring period. A total of 200 samples were collected at 45 sites. They were characterized by their porosity and grain-size, as well as their chemical composition through elemental, isotopic and molecular biomarker analysis. A wide range of OM sources were identified in the sediments, including both natural (bacteria, algae, macrophytes, 25 terrestrial plants), and anthropogenic (combustion products, crude oil, petroleum products - e.g. from the processing of crude oil at refineries- and fecal matter) sources. Sediment slurry incubations were carried out to determine the spatial variability of potential mineralization rates under oxic conditions. In addition, the measurements of NH 4+ and PO 4 fluxes at the sediment-water interface were made from sediment core incubations under realistic redox conditions of sediment. The physical and chemical sedimentary characteristics explained 58 30 % of the variability of mineralization rates under oxic conditions, with a negligible independent effect of the SOM origin (3 %). Conversely, under insitu redox conditions, the prevalent role of SOM origin over quantity/accessibility on the sediment biodegradability was highlighted with a significant effect 5 and 1.5 fold higher on the PO 4 and NH 4+ fluxes respectively. The anthropogenic inputs from the watershed to the coastal sediment, through agricultural runoff and/or sewages discharge, seem to significantly drive the nutrient dynamics 35 at the sediment-water interface. Higher values of NH 4+ and PO 4 fluxes were measured for the sediment with a chemical composition impacted by human activities. with dichloromethane using an accelerated 155 solvent extractor (Dionex™ASE™ 200) according to the following conditions: 33 mL cells, 5 min heating at 100°C and 60 bars, 10 min static phase, completed with 80% flush and 20-s purge with nitrogen. Elemental sulfur was removed from the total lipid fraction by reduction on metallic copper. The total lipid extract was then weighed and fractionated into aliphatic hydrocarbons, aromatic hydrocarbons and polar compounds on a silica column by successive elution with cyclohexane, cyclohexane/dichloromethane (2/1, v/v) and methanol/dichloromethane (1/1, 160 v/v). After this fractionation step, the mass of each fraction was weighed. Polar fractions were analyzed by capillary gas chromatograph-mass spectrometer (QP2010SE GC-MS, Shimadzu) after derivatization using a mixture of N,O-bis(trimethylsilyl) trifuoroacetamide (BSTFA) and trimethylchlorosilane (TMSC) (99/1, v/v), while aliphatic and aromatic fractions were analyzed without further treatment. The injector used was in splitless mode and maintained at a temperature of 310°C. The chromatographic 165 separation was performed on a SLB-5MS capillary column (length= 60m, diameter=0.25mm, film thickness=0.25µm) under the following temperature conditions: 70°C (held for 1 min) to 130°C at 15°C/min, the 130°C to 300°C (held for 15 min) at 3°C/min. The helium flow was maintained at 1 ml/min. The chromatograph was coupled to the mass spectrometer by a transfer line heated at 280°C. The analyses were performed in fullscan

of fossil fuels, burning of biomass or oil spill (e.g. Benlahcen et al., 1997;Sporstol et al., 1983, Yunker et al., 2002. Most investigations on the spatial variation of the origin and composition of the SOM have been carried out so far at local scale, within a bay (e.g. Dubois et al., 2012;Gu et al., 2017), in lakes (e.g. Dunn et al., 2008;Fang et al., 80 2014) or along an estuary (e.g. Kumar et al., 2020;Meziane et al., 2006). There are only few studies on the regional scale e.g. Lee et al., 2019). Here we are interested in this scale through the case of Brittany, a region in the west of France particularly affected by coastal eutrophication phenomena (e.g. Perrot et al., 2014;Schreyer et al., 2019).
In this region, marine intertidal mudflats are influenced by river discharges and large tidal fluctuations. The agricultural intensification and the urbanization of watersheds have led to the coastal eutrophication, where 85 macroalgal "green tides" are regularly occurred in spring (Ménesguen et al., 2019;Perrot et al., 2014;Schreyer et al., 2019). We therefore expect that the isotopic and lipid markers tools used in the SOM characterization allow to describe the variability of OM sources -natural or anthropogenic -which affect the coastal and estuarine systems in Brittany.
The range of possible sources and hence the great potential diversity of the chemical structure of SOM makes it 90 difficult to relate the biodegradability to SOM composition and ultimately, the biodegradability to the source of SOM. Biopolymers such as proteins have been shown to be more biodegradable than refractory biopolymers specific to vascular plants (e.g. cutan) (Tegelaar et al., 1989). It is thus widely assumed that the terrestrial OM is more refractory than the marine algal OM due to its chemical composition both dominated by higher resistant compounds, and impacted by the aging during the transport (Arndt et al., 2013). The SOM mineralization is thus 95 controlled, in part, by its origin but also by the deposition rate of OM in the sediment (Khalil et al., 2018). To our knowledge, no investigation has dissociated and quantified the respective effects of OM quantity and source on the sediment biodegradability. Recently, Albert et al., 2021 have attempted to highlight the prevalent role of the OM quality on the benthic processes in coastal sediments, but in a context dominated rather uniform OM source materials (spring and summer plankton).

100
The aim of this work was i) to assess the variability of OM composition within coastal sediments of Brittany, at the scale of the entire region, ii) to use this variability to go back to the variability of OM sources, and finally iii) to determine how changes in OM source affect its biodegradability . To this end, 200 sediments were sampled at 45 sites in macrotidal mudflats on the Brittany coast. The SOM characterization was carried out by combining bulk elemental, isotopic and chemical biomarkers analysis. To determine the link between the origin and 105 degradability of SOM, mineralization rates from sediment slurry incubations under oxic conditions and benthic nutrient fluxes (NH4 + and PO4) from core incubations were carried out for all the sediments. In addition, the sediments were characterized by their physical properties (grain size distribution and porosity), which impact the accessibility of SOM for the microorganisms, as well as their element contents (C, N, P). Canonical redundancy analysis (RDA) and variance partitioning were used to explain the spatial variability of the NH4 + and PO4 fluxes 110 and the mineralization rates into independent parameters related to 1) the variation of SOM composition and origin and 2) the variation of the chemical composition and physical properties of sediment.
All study sites were macrotidal mudflats, located in Brittany, north-western France, and have been already fully 115 described in a companion paper (Louis et al., 2021). Overall, 200 sediment samples were collected on 45 sites and classified into 12 mudflats (Fig. 1) during the spring period (year 2019). Sediment cores were sampled with one PVC core (diameter = 6 cm, h = 20 cm) in the upper 10 cm sediment layer for the measurements of nutrient benthic flux, and another one core (diameter = 9 cm, h = 5 cm) was sampled for carrying out sediment slurry incubations as well as for characterization of the surface sediment (elemental, isotopic and lipidic composition; grain-size 120 distribution and porosity).

125
A sample of known volume and weight taken in the upper 5 cm layer of the core sediment was maintained at 4°C and used for porosity measurement. The porosity was calculated by using the amount water determined after drying the sample at 60°C, with the sediment density set at 2.55 (Berner, 1980). The particle size distribution (< 2 mm) was measured using a laser diffraction instrument (Malvern Mastersizer). "Mud" is defined as the sum of the clay and silt particles, with a diameter of less than 63 µm (e.g. Keil and Hedges, 1993;Pye and Blott, 2004)  The total organic carbon and nitrogen (TOC and TN) contents and the carbon and nitrogen isotopic ratios (δ 13 C and δ 15 N) were determined using an element analyzer (FLASH™ EA 2000 IRMS) coupled with an isotopic ratio mass spectrometer (DELTA V™ plus).
An aliquot of freeze-dried and crushed sediment was acid-treated with 2N HCl to remove the carbonate and was 135 subsequently rinsed with deionized water. After centrifugation, the carbonate-free sample was dried at 60°C, and ground before being placed into a tin capsule for the TOC and δ 13 C analysis. A second aliquot without an acidification treatment was used for the TN and δ 15 N analysis. All isotopic data were expressed in the conventional delta notation: The reference is (Pee Dee Belemnite) PDB for δ 13 C and atmospheric N2 for δ 15 N. Analysis uncertainty was less than 0.2‰.
The phosphorus (P) speciation in the surface sediment was made from determining the iron oxide-bound P (Fe-P)

145
and organic P (Org-P) contents, which represented the pool of potentially bioavailable P. The analytical method was totally described in the companion paper Louis et al. (2021). Briefly, the Fe-P content was determined using a Dithionite-Bicarbonate solution (Ruttenberg et al., 1992), and the Org-P content was quantified by calculating the difference between the total P and inorganic P (sums of Fe-bound P, Ca-bound P and detrital P) contents as described in Andrieux-Loyer et al. (2008).

150
The TOC, TN, Org-P and Fe-P contents were expressed as the mass of the carbon, nitrogen and phosphorus in the total dry mass of the sediment.
The C:N and TN:Org-P ratios (mol:mol) were calculated from the TOC, TN and Org-P contents.

Lipid analysis
Approximately 30g of freeze-dried and crushed sediment was extracted with dichloromethane using an accelerated  et al. (2008).
These analyses provided the total concentration of lipid compounds in the dry mass sediment, as well as the concentration of each lipid marker. The results were also expressed as the proportion of each lipid marker against the sum of all lipid compounds quantified in the sediment.

Nutrient benthic fluxes from sediment core incubations
The measurement method of NH4 + and PO4 fluxes was totally described in the companion paper Louis et al. (2021).
Briefly, the sediment cores were incubated in the dark during 4 hours directly on site in a mobile laboratory under controlled temperature (19 ± 2°C) within one hour upon sampling. The overlying water was replaced by 150 mL of nutrient-free artificial seawater ([NaCl] = 33 g.L -1 , [NaHCO3] = 0.2 g.L -1 , pH ≈ 8) and gently aerated and stirred 180 by bubbling in order to preserve the oxic conditions in the overlying water and to prevent the build-up of concentration gradients at the sediment-water column interface. The NH4 + and PO4 fluxes (µmol.m -2 .h -1 ) across the sediment-water interface were estimated by using the change in the molar concentration of the solute in the known volume of overlying water as a function of incubation time and the surface area of the sediment core.

185
The assessment of the biodegradability of the SOM was carried out by sediment slurry incubations in hermetically closed glass flasks, placed in the dark during 4 hours directly on site in a mobile laboratory under controlled temperature (19 ± 2°C), agitation (150 rpm.min -1 ) and oxic conditions. A wet aliquot sediment (m=5g) was used and mixed with a known volume of artificial seawater (V=20mL). The mineralization rate of SOM by microbial degradation, was determined by the measurements of the CO2 concentrations in the headspace of flasks before and 190 after incubation by micro gas chromatography (Agilent ™ 3000A micro GC). The mineralization rate (µmol CO2.g -1 .h -1 ) was calculated as follows: Vm × Vair + Vliq × bunsen coef t×m with: [CO2] the concentration of CO2 measured in the headspace before and after incubation (µmol.mol -1 ); Vm the molar The sediment reactivity can be expressed as the first-order rate constant (e.g. Berner, 1980), calculated as follows: with: R the mineralization rate (µmol CO2.g -1 .yr -1 ); CTOC the content of organic carbon in the dry mass of sediment (µg.g -1 ); and Mc is the molar mass of carbon (Mc = 12).

Co-inertia analysis between the data discriminating OM sources
A co-inertia analysis was carried out to assess how the proportion of lipid markers in SOM varies with the isotopic and elemental signature of sediment (Dray et al. 2003). Unlike constrained analyzes, co-inertia has no causal link 220 (no multiple regression) and just represents the two descriptor datasets in the common space of samples after two simple ordinations. It allows to determine how it varies together by superimposing the two points clouds of stations.
The link between the two tables is thus performed a posteriori, by maximizing the square of covariance between initial projections of samples in the separate ordinations. Being unconstrained geometric data analysis, there is no limitation in the variables from each table. A representation of the isotopic and elemental ratios was thus made in

Canonical Redundancy Analysis (RDA) and variance partitioning
In order to explain spatial variability in NH4 + and PO4 fluxes and mineralization rates (response variables), two constrained analyses were carried out in parallel. For the first analysis, explanatory variables were SOM sources (elemental and isotopic ratios and the proportions of lipid markers groups), hereafter called "SOM origin".
Explanatory variables for the second analysis were sediment composition (the contents of TN, TOC, Fe-P, Org-P, In order to determine the linear relationships between all previously selected variables and the potential 245 biodegradability of sediment (mineralization rates and nutrient fluxes), a RDA was performed and tested by the permutation test. A hierarchical cluster analysis was then run from the projection of all sediment samples on this canonical ordination by the Ward2 Algorithm (Murtagh and Legendre, 2014).
With the selected explanatory variables, the variance partitioning was then used to quantify the variance proportion of NH4 + and PO4 fluxes and mineralization rates independently explained by both the variables "SOM origin" and 250 the "physic-chemical composition" (Borcard et al. 1992). Therefore, a comparison between the effects of the SOM origin on the sediment biodegradability with those of other known sedimentary characteristics was established.
The significance of each fraction of interest was tested by the permutation test.
The "rda", "vif.cca", "anoca.cca" and "varpart" functions of the "vegan" package (R software) were used to perform the RDA, the VIF calculation, the permutation test and the variance partitioning respectively. The

255
"dist.dudi" function of the "ade4" package and the "hclust" function (R software) were used to perform the hierarchical cluster analysis. Each data was previously standardized with the "scale" function before multivariate analysis.

260
Over all sediment samples, the percentage of mud averaged 68 ± 17 % with the clay and silt particles representing respectively 5 ± 2 and 64 ± 15 % of the particle size distribution (Table 2). A large proportion of the surface sediment collected was classified as sandy mud (38 %) and slightly sandy mud (49 %) (Flemming, 2000). The average porosity was 70 ± 10 %. The TOC and TN contents averaged 2.2 ± 1.2 % and 0.24 ± 0.12 % of dry mass, respectively. The pool of bioavailable P (sum of Fe-P and Org-P contents) represented, on average, 0.033 ± 0.018

Elemental ratio
The C:N and TN:Org-P ratios (mol:mol) averaged 10.0 ± 2.1 and 23.1 ± 6.9 respectively over all sediment samples.
The highest C:N ratios were measured in the sediments collected in the Goulven Bay (site #21, mean = 17.8 ± 3.9). The Lorient Bay presented also relatively high values of C:N ratio over all sampling sites (mean = 11.94 ± 1.32). Lowest C:N ratios were measured in the Ria Etel, particularly at the site #6 (mean = 4.8 ± 0.7). At the 275 regional scale, no difference in C:N ratio was evidence between the samples collected in bays, in lower estuaries and or middle estuaries (Fig. 2a). At the local scale (scale of a given mudflat), rather systematic variations of the C:N ratio are observed depending on the location of the sampling station. In the Trieux Estuary, for example, samples taken further upstream of the estuary yielded systematically the highest C:N ratios (site #2, mean = 11.8 ± 0.4; site #4, mean = 7.4 ± 1.1). This is also true for the Aber Wrac'h were sediments collected in the lower estuary 280 (site #23) gave higher C:N ratios (mean = 11.5 ± 1.3) than those collected directly in the bay (Figure 2A). With regard to the TN:Org-P ratios, the sediments sampled in the Rance Estuary presented lowest values over all sampling sites (mean = 16.0 ± 7.6). In the Lorient Bay, the sites located further downstream of the estuary presented sediments with higher TN:Org-P ratios (Fig. 2b), particularly for sites #10 (mean = 36.3 ± 19.8) and #11 (mean = 36.7 ± 4.8). Elevated values of TN:Org-P ratio (up to 25) were also measured in sediments collected in the bay of

OM sources from lipid markers
A wide range of OM sources were identified in the sediments, including both natural (bacteria, algae, macrophytes, terrestrial plants), and anthropogenic (combustion products, crude oil, petroleum products -e.g. from the 305 processing of crude oil at refineries-and fecal matter) sources. Overall, lipid markers originating from terrestrial plants were the most important quantified markers, representing, on average, 31.3 ± 9.9 % of the lipid pool (Fig.   S1a). In contrast, the bacterial OM markers represented only a small proportion of the lipid pool for all sediments, with an average of 1.3 ± 0.9 % (Fig. S1b). The sites in the upstream section of the Lorient Bay were distinguishable from all other sites at the Brittany scale by the highest proportions of terrestrial plants markers (site #13, mean = 310 47.1 ± 5.6 %; site #14, mean = 45.9 ± 2.8 %), coupled to the highest concentrations of these compounds in the sediment (Fig. S1c).
Lipid markers specific to algae and macrophytes represented on average 8.5 ± 3.3 % of the total lipid pool. Higher proportions of microalgae (pelagic and benthic) markers were measured in the sediments collected in the Gulf of Morbihan (mean = 5.7 ± 2.3 %), the Morlaix Bay (mean = 5.2 ± 1.5 %) and the Rance Estuary (mean = 5.7 ± 2.3 315 %) (Fig. S2a). For the Morlaix Bay, this was mainly due to a higher proportion of phytoplankton biomarkers (mean = 4.2 ± 1.5 % against 2.5 ± 1.5 % for overall sediments). In the sediments of the Morbihan Gulf, the high proportions of microalgae markers were clearly associated to larger concentrations of these compounds in the sediments (mean = 1,952 ng.g -1 against 741 ± 777 ng.g -1 ) (Fig. S2b). Largest proportions of green macroalgae markers were observed in the Goulven Bay, the Trieux Estuary and in the downstream sites of the Lorient Bay 320 (Fig. 4a). In these areas, the sediment composition was directly related to the highest amounts of green macroalgae, given the positive relationship between the proportions and concentrations of these biomarkers over all sediment samples (Fig. 4b). In addition to the natural OM sources, anthropogenic OM markers were present in the sediments, with an average proportion in the lipid pool of 19.2 ± 15.5 %. The high spatial variability of anthropogenic markers was mainly 330 due to a high variation of marker proportions of petroleum products (e.g. creosote) (mean = 12.3 ± 15.4 %). A second group of oil and by-products markers was specific to crude oil (Table 1). This proportion over all samples averaged 3.0 ± 3.5 %. No correlation was observed between the marker proportion of petroleum products and crude oil (Fig. 5a), and the samples could be divided into two classes according their location: north and south coast (Fig. 5b). The sediments collected on the southern Brittany coast were characterized by highest proportions 335 of petroleum products markers, while those collected in the northern Brittany coast had lower proportions. For the crude oil markers, the opposite was observed: the sampling sites located at the north Brittany coast presented higher proportions than those located at the south (Fig. 5b). In general, an increase in proportion of petroleum product markers resulted from an increase in these compounds in the sediment (Fig. 5c), as well as for the crude oil markers (Fig. 5d). In the present study, 4 stanols specific to human and animal waste were identified and 340 quantified in the SOM: coprostanol, epicoprostanol, 24-ethylcoprostanol and 24-ethylepicoprostanol, with relative abundances of 28 ± 9 %, 8 ± 5 %, 44 ± 8 % and 21 ± 6 % respectively. The proportion of these fecal markers in the lipid pool of the sediment samples was on average 1.7 ± 0.9 %. Relative higher values were measured in the Goulven Bay (mean = 4.3 ± 1.1%), the Lorient Bay (mean = 2.4 ± 1.2%) and the Rance Estuary (mean = 2.1 ± 0.7 %) (Fig. 6). Except for the Rance estuary, this was mainly due to an enrichment of fecal matter in the sediment 345 ( Fig S3). Markers of combustion products were also detectable, with an average proportion of 2.3 ± 4.2 %. In the sediments collected in the Auray River, the Lorient Bay and Ria Etel, both largest proportions and concentrations of these combustion markers were measured (Fig. S4).

Potential SOM biodegradability
The slurry sediment incubations allowed assessing the SOM biodegradability under oxic conditions. The 360 mineralization rate was determined from the CO2 production measured during the incubation from the microbial respiration of SOM. The average CO2 production over the 200 sampling points was 0.40 ± 0.58 µmol C.g -1 .h -1 .

385
To assess how the proportion of lipid markers in SOM varies with the isotopic and elemental signature of sediment, a co-inertia analysis (see Materials and Methods 2.7.2) was done by comparing the two descriptors' datasets represented in the common space of sampling sites (Fig. S5).
The co-inertia analysis between the isotopic and elemental ratios and the lipid marker proportions, indicated a correlation coefficient (RV) of 0.13. The first and second axis represented 60.1 and 18.8 % respectively of the co-

395
TN:Org-P contributed 52 % of the second axis. The correlations between the lipid groups mentioned above, and the elemental and isotopic ratios, are presented in the Table 3. The groups "Fecal" and "Plants 3" were positively correlated with the C:N ratio (pearson coef = 0.39 and 0.29, p<0.001), while they were negatively correlated with δ 13 C (pearson coef = -0.32 and -0.29, p<0.0001). For the markers of petroleum products, the opposite was To relate biodegradability to OM sources considering sediment properties, the significant variables from the different datasets were selected (see Materials and Methods 2.7.3). In the data matrix "SOM origin", the selected variables were δ 13 C, δ 15 N, and the lipid markers groups "Fecal", "Petroleum products", "Microalgae", "Plants 3" and "Plants 4". Regarding the data matrix "physico-chemical composition", the selected variables were the TN,

410
Fe-P, Org-P and total lipids contents, the percentage of mud and the porosity. With these selected variables in both data matrices, the variance partitioning of mineralization rate, NH4 + and PO4 fluxes was carried out and the results are presented in Fig. 8. The selected parameters from "SOM origin" and "physico-chemical composition" explained 58 % of the variance of the sediment mineralization rates. A large part was the result of the significant and independent effect of the "physico-chemical composition" variables (F=28.2, p=0.001) (Fig. 8). On the 415 contrary, the variations of NH4 + and PO4 fluxes were more correlated with the "SOM origin" than with the "physico-chemical composition" variables ( Fig. 8). The variables "SOM origin" significantly explained 15 and 14 % of variance of NH4 + fluxes (F=4.8, p=0.003) and PO4 fluxes (F=5.2, p=0.001) respectively, compared to 9 and 3 % with the "physico-chemical composition". The variables "SOM origin" and "physico-chemical composition" shared between 10 and 16 % of the variance partitioning of NH4 + and PO4 fluxes respectively.

425
In order to explain the sediment mineralization rates and nutrient fluxes in relation to the selected variables, a canonical redundancy analysis (RDA) was performed. The canonical ordination constrained by both selected variables "SOM origin" and "physico-chemical composition" significantly explained 42 % of the responses of sediment (F = 9.3, p = 0.001), of which 84 % was carried out by the first two axis. The NH4 + flux was mainly correlated with the second axis (correlation coef = -0.92). This second axis was associated to the isotopic ratios 430 δ 13 C and δ 15 N, the proportion of microalgae biomarkers and, to a lesser extent, the proportion of petroleum products markers (Fig. 9a, b). The mineralization rates were mainly correlated with the first axis (correlation coef = -0.80), and with the third axis (correlation coef = 0.53). The opposite was observed for the PO4 flux, which was mainly correlated with the third axis (correlation coef = -0.84) and with the first axis (correlation coef = -0.47).
On the first axis, the mineralization rate and the PO4 flux were associated to the variables related to the sediment 435 composition (TN, Org-P, Fe-P and total lipids contents) and physical properties (percentage of mud and porosity) ( Fig. 9a, b). On the axis 1 and 3, they were associated to the proportion of fecal biomarkers.
The RDA discriminated four groups of sediments from the hierarchical cluster analysis (see Materials and Method 2.7.3) (Fig. 9c). A first cluster was related to the highest mineralization rates, Org-P and TN contents, and porosity https://doi.org/10.5194/bg-2021-318 Preprint. Discussion started: 29 November 2021 c Author(s) 2021. CC BY 4.0 License.
(cluster "I"). Another cluster represented the lowest mineralization rates, nutrient fluxes, porosity and percentage 440 of mud, and the sediments with the lowest contents in TN, Org-P, Fe-P and total lipids (cluster "II"). The cluster "III" was intermediate between the clusters "I" and "II". The last cluster represented the highest NH4 + fluxes, microalgae markers proportions and isotopic ratios (cluster "IV"). The spatial distribution of these groups is presented in Fig. 9d. The sediment collected in the Goulven Bay and Pont L'Abbé would represent those for which the selected variables "SOM origin" and "physico-chemical composition" would explain both higher 445 mineralization rates and PO4 fluxes (clusters "I" and "II"). Some of sediments collected in the Lorient Bay (site #10), the Gulf of Morbihan and the Vannes Estuary (site #41), would represent those of which the selected variables would explain the highest NH4 + fluxes (cluster "IV").  The increasing gradient of δ 13 C along the estuary-coastal zone continuum generally illustrates the mixture of OM from continental to marine origin (e.g. Cook et al., 2004;Li et al., 2016;Ogrinc et al., 2005), which was observed here at both the regional and local (mudflat) scale (Fig. 3). In the Lorient Bay, for example, the upstream sites 465 were characterized by the lowest δ 13 C values (≈ -23 ‰), dominated by the largest inputs of terrestrial plants (Fig.   S1), resulting in elevated sedimentary C:N ratios (>12) (Fig. 2a). In this Bay, downstream sediments show an enrichment of OM from green macroalgae with high concentrations and proportions of these biomarkers in the lipidic pool (Fig. 4). The presence of macroalgae resulted in an increase in δ 13 C seaward of Lorient Bay (up to -19 ‰; Fig. 3). This is in line with the δ 13 C signature of Ulva sp. of around -15.8 ± 4.4 ‰ (Berto et al., 2013;Dubois 470 et al., 2012;Riera et al., 1996), which is significantly higher than the δ 13 C associated to C3 terrestrial plants (δ 13 C = -28.7 ± 0.5 ‰; Davoult et al., 2017;Meyers, 1994). In these downstream sites, the C:N ratio of sediments still remains high (mean = 11.5 ± 1.4), in accordance with the C:N ratio of Ulva sp. around a value of 13 (Dubois et al., 2012;Liénart et al., 2013).
A large contribution of 13 C enriched marine OM sources in the sediment composition was typically observed in 475 the samples collected directly in a bay, e.g. in the Gulf of Morbihan and the Morlaix Bay (Fig. 3). In the Morlaix Bay, the higher proportions of phytoplankton biomarkers could explain the δ 13 C values of -20.6 ± 0.4 ‰ in the sediments, close to the isotopic signature associated to marine phytoplankton (-21.3 ± 1.2 ‰; Liénart et al., 2017).
Sediments collected in the Gulf of Morbihan were characterized by the highest δ 13 C values observed at the regional scale (mean = -17.0 ± 0.6 ‰) (Fig. 3) and a significant contribution of microalgae in SOM composition (Fig. S2).

480
Here, we assume that the major source of OM is microphytobenthos, for which δ 13 C is around -18.2 ± 1.7 ‰ (Dubois et al., 2012;Riera et al., 1996) instead of marine phytoplankton. Over all sediments of the Gulf of Morbihan, the mean C:N ratio was 11.1 ± 1.6, which is slightly higher than the C:N ratio widely used as signature of microphytobenthos (mean = 9.6 ± 1.0) (Dubois et al., 2012;Liénart et al., 2013).

485
In addition to the natural sources of OM, eutrophicated coastal systems receive anthropogenic matter from urban discharges, agricultural and industrial activities impacting the SOM composition. The analysis of oil and byproducts markers in the sediments allowed to discriminate the pollutions between the north and the south of the Brittany coast (Fig. 4). The north sites seem to have been impacted by oil spills leading to the enrichment of crude oil markers in the sediments. For the south sites, the sediments were characterized by an enrichment of petroleum 490 products, particularly those collected in sites classified in the group "bay". No significant difference in δ 15 N was observed between northern and southern Brittany, despite the contrasting isotopic signature of crude oil (δ 15 N < 2 ‰) and petroleum products (δ 15 N > 9 ‰) (Rumolo et al., 2011 and references therein). While the crude oil and by-products are largely depleted in 13 C (δ 13 C < -28 ‰) (Rumolo et al., 2011 and references therein), the 13 Cenriched sediments collected in the present study were associated to higher proportions of petroleum products 495 markers (Table 3). This can be explained by the fact that the composition of sediments collected directly in the bay in the southern Brittany was also impacted by the 13 C-enriched marine OM inputs as discussed earlier (section 1.1). The petroleum inputs seem to have a minor impact on the isotopic signature of sediments. In addition to oil pollution, urban discharge from sewage treatment plants are widely detected by a large increase in δ 15 N in the bulk sediment (δ 15 N ≈ 8-10 ‰) (e.g. Rumolo et al., 2011;Savage, 2005). The current study shows 500 the impact of sewage discharges on the isotopic composition of the sediment at the local scale. For example, the highest δ 15 N values were measured in the sediments collected downstream of a wastewater treatment plant (WWTP) (for ~ 60.000 person equivalent) in the Vannes estuary (site #41). The sediments of this site were also characterized by an enrichment of fecal matter through, in particular, higher concentrations (208 ± 107 ng.g -1 dry sediment) and proportions (0.9 ± 0.2 of the lipid pool) of coprostanol. This marker represents around 60 % of the 505 total sterols in human faeces (Leeming et al., 1996, Harrault et al., 2019. The effect of sewage effluent on the SOM composition was gradually diluted through a decrease in δ 15 N of ≈ 1.7 ‰ measured around 2-3 km from the site #41 (sites #40 and #42; Fig. 1). This is more pronounced than the 1 ‰ decrease reported by Savage (2005) within 10 km of the outfall of a WWTP (for ~ 250.000 person equivalent) in a bay of Sweden. In contrast, a high δ 15 N gradient was exhibited by Rumolo et al. (2011) in the harbor of Naples (≈ 10 ‰) within 1.5 km of the outfall 510 of a WWTP. As the authors claimed, the spatial effect of sewage effluents on sediment composition would be under the control of the hydrodynamics of the coastal system which itself determine the sedimentation rate of OMrich fine particles. Consistent with the dilution of the δ 15 N value between the sites #40 and #42 in the present study, the concentrations and proportions of fecal markers in the SOM declined by ≈ 60 % between these two sites.
High values of δ 15 N in the sediment were also observed in the mid estuary of the Rance (from 7.7 to 9.2 ‰). The

515
proportions of fecal markers were significantly higher than those measured on the regional scale (Wilcoxon test, p < 0.005), while larger concentrations of these markers were not observed. Previous studies showed an increase in δ 15 N in aquatic systems impacted by agricultural activities, where animal waste is often used as a fertilizer (Finlay and Kendall, 2007 and references therein). Since the watersheds in Brittany are particularly impacted by intensive agricultural activities (e.g. Morand and Briand, 1996), it is assumed that the high δ 15 N values measured 520 in the sediment of the Rance estuary are due to the high proportion of cultivated areas in the watershed. The TN:Org-P ratio, a second sedimentary characteristic confirms this hypothesis. The sediments collected in the Rance Estuary presented the lowest TN:Org-P ratios (Fig. 2b), with a mean value of 16.0 ± 7.6 against 23.1 ± 6.9 overall sediments. We suggest that a low TN:Org-P ratio could result from the agricultural runoff. Indeed, animal manure such as pig slurry, used for crop growth, tend to decrease the TN:Org-P ratio of soil through a high retention 525 of P in soil and a rapid leaching of N into groundwater and rivers (Penuelas et al., 2009;Toth et al., 2006). This could thus lead to a decrease in the TN:Org-P ratio of OM inputs from watersheds in the estuary-coastal zone continuum.

530
After characterizing the SOM composition at the regional scale, the objective of this work was to quantify the relative significance of the SOM origin on the sediment reactivity, as well as to identify the SOM sources enabling the sediment to act as a nutrient source for the overlying water.

Mineralization rates under oxic conditions
In the present study, the role of the SOM origin and sediment physico-chemical composition on the sediment of these mineralization rates was explained respectively by the independent effect of "physico-chemical composition" and "SOM origin" variables (Fig. 8). The potential sediment mineralization rates were assessed under oxic conditions; it is widely accepted that the presence of O2 would enhance the organic carbon reactivity 540 in many marine sediments (LaRowe et al., 2020 and references therein). Indeed, the presence of O2 would allow both the aerobic respiration and the cleaving of the non-hydrolysable bonds in the more refractory organic compounds (Burdige et al., 2007). The mineralization process would thus be slightly affected by the OM quality regardless its quantity. This is line with our results highlighting that the spatial variability of OM sources would have a weak independent impact on the mineralization rate of the sediment when it is not constrained by the anoxic 545 conditions. To a lesser extent, the variables "SOM origin" and "physico-chemical composition" shared 16 % of the variance partitioning of the mineralization rates. Considering that this effect reflected the coupled OM quantityorigin, it means that a large deposition of OM from specific sources would enhance the SOM biodegradability.
According to the positive association between the proportion of fecal matter markers and the mineralization rate determined by the RDA, we suggest that the coastal sediment composition impacted by human or animal waste 550 would have a greater potential biodegradability.
The sediment reactivity k, calculated from the mineralization rate and the TOC content, ranged from 0.8 to 11 y-1 for 75 % of data (Fig. 7b). Since the SOM biodegradability seems to be only weakly impacted by its origin as discussed above, we suggest that the spatial variability of k is likely explained by the residual part of the variance partitioning of the mineralization rate reaching 42%. This can include the biological parameters not measured in 555 this work, such as the microbial abundance and diversity.

Nutrient fluxes at the sediment-water interface
The benthic nutrient fluxes are driven by chemical (e.g. precipitation/dissolution; adsorption/desorption), biological (e.g. bacterial respiration) and physical (e.g. diffusion) processes (e.g. Capone et al., 2008;Ekholm and Lehtoranta, 2012;Santschi et al., 1990). The SOM mineralization is involved in the nutrient fluxes through the 560 regeneration of NH4 + and PO4 in the porewater of sediment. Therefore, in addition to the measurements of mineralization rates, the sediment biodegradability was also assessed through its capacity to act as a nutrient source to the overlying water. The NH4 + and PO4 fluxes at the sediment-water interface were measured from the sediment core incubations in the dark and under controlled temperature (19°C). The realistic redox conditions of the sediment were preserved during the core incubations. In the eutrophicated coastal areas, as is the case in Brittany 565 coast, the sediments are often constrained by the hypoxia, and thus the SOM mineralization is limited by a low O2 penetration in the sediment of few millimeters (Middelburg and Levin, 2009). Under these redox conditions, we expect that the mineralization process is more driven by the lability of the SOM rather than by its quantity/accessibility. The OM sources control the SOM composition and likely its lability. We thus suggest that the SOM origin could significantly drive the regeneration of NH4 + and PO4 in the sediment and consequently their 570 effluxes. The variance partitioning of NH4 + and PO4 fluxes confirms this hypothesis (Fig. 8). The percentage of the variance of NH4 + and PO4 fluxes only explained by the variables "SOM origin" reached 15 and 14 % respectively, against 3 and 9 % for the "physico-chemical composition". All parameters measured and selected in the present study explained 34 and 33 % of the variation of PO4 and NH4 + fluxes. The RDA allowed to determine the relationships between these parameters and the benthic nutrient fluxes, and particularly to assess which SOM 575 sources can enhance the PO4 and NH4 + fluxes. (Fig. 9a and b). The PO4 fluxes were positively linked to the proportion of fecal markers in the sediment. This suggests that anthropogenic inputs, such as sewage discharges or agricultural runoff, rich in fecal matter would be positively impacting the PO4 fluxes from the sediment. The RDA also showed a positive relationship between the PO4 flux and the Org-P and Fe-P contents, as previously observed for the same sediments (Louis et al., 2021). This may highlight the concomitance of Org-P mineralization 580 and Fe-P dissolution on the PO4 effluxes under specific anoxic conditions enhanced by anthropogenic pressures (Lehtoranta et al., 2009). This is in line with a work carried out in the Bay of Brest (Brittany), showing both larger mineralization rate and a larger dissolution rate of Fe-P, as well as a larger PO4 flux, where the sediment was influenced by higher urban and farming activities in the watershed (Ait Ballagh et al., 2020).
The NH4 + fluxes were positively related to both the isotopic signature of the bulk sediment (δ 15 N and δ 13 C) and 585 the proportions of microalgae and petroleum markers. The NH4 + fluxes seem to increase when the SOM composition is influenced by 13 C-enriched OM marine inputs, and particularly from microalgae. In addition, high NH4 + fluxes would characterize the sediments collected in coastal areas under anthropogenic pressures. The positive relationship with the proportion of petroleum product markers means that this would correspond to the sediments with a composition particularly impacted by industrial, shipping/docking activities, such as in the bays 590 in the southern Brittany (Fig. 5)

605
In Brittany, a main symptom of the coastal eutrophication, fueled by anthropogenic pressures, is the proliferation of green macroalgae (Ulva sp.). Previous studies highlighted the stimulation of the mineralization rate and nutrient effluxes during the decomposition of green macroalgae in the surface coastal sediment (Corzo et al., 2009;Garcia-Robledo et al., 2013;Lomstein et al., 2006;Trimmer et al., 2000). An increase in NH4 + and PO4 fluxes for the sediments characterized by a high contribution of these algae in their composition might thus have been expected.

610
However, in the present study, no significant link was shown between the lipid markers specific to green macroalgae and the benthic nutrient fluxes. Only the sediments collected seaward of the Lorient Bay (sites #10 and #11) showing a green macroalgae enrichment (see discussion section 4.1.1) presented, in parallel, NH4 + and PO4 fluxes significantly higher than that measured on the regional scale (Wilcoxon test, p < 0.005). In the present study, we focused on the link between the benthic nutrient fluxes and the sediment composition,

615
with a comparison of significant effect of SOM origin vs quantity. Nevertheless, we must keep in mind that a large part of the variance of benthic NH4 + and PO4 fluxes on the regional scale remains unexplained here (66-67 %) (Fig. 8). As mentioned earlier, the microbial abundance and diversity were not assessed in the current study, and the abundance and/or diversity may very well play an additional role in the degradation of the OM. In addition, the bioturbation, mediated by the macrofauna activities, was likely preserved in the sediment core incubations and 620 therefore may be involved in the spatial variability of NH4 + and PO4 fluxes (Graf and Rosenberg, 1997;Welsh, 2003;Kristensen et al., 2012).

Conclusion
This is the first study to our knowledge that has described the variability in SOM composition through a broad sampling campaign of marine mudflats at the regional scale, and made the link with sediment potential 625 biodegradability and nutrient release. The sediment slurry incubations allowed to determine the spatial variability of potential mineralization rates under oxic conditions. The physical and chemical sedimentary characteristics explained 58% of the variability of mineralization rates, with a negligible effect of the SOM origin. The sediment core incubations allowed to assess the spatial variability of sediment biodegradability under realistic redox conditions of the sediment through the measurements of NH4 + and PO4 fluxes at the sediment-water interface. The

630
physical and chemical sedimentary characteristics explained 34 and 33 % of the variation of PO4 and NH4 + fluxes.
The sediment core incubations showed the prevalent role of SOM origin over quantity on the benthic nutrient fluxes, with a significant effect 5 and 1.5 fold higher on the PO4 and NH4 + fluxes respectively. The present study also highlighted the potential impact of human activities on the nutrient dynamics at the sediment-water interface, through agricultural runoff and/or sewages discharges.

635
A significant fraction of the spatial variability of benthic nutrient fluxes could not be explained here by the sedimentary characteristics, suggesting that the spatial variability of benthic microbial diversity/abundance should be considered for further investigations. It would be interesting to link the chemical composition to genetic information of sediment in the aim to understand the variability of benthic microbial function of coastal areas under anthropogenic pressures.

640
In this work, all mudflats were under a macro-tidal regime, and therefore another recommendation would be to determine to what extend the tidal forcing can change the sediment composition and thus its biodegradability.