Key biogeochemical factors affecting soil carbon storage in Posidonia meadows

Abstract. Biotic and abiotic factors influence the accumulation of organic carbon (Corg) in seagrass ecosystems. We surveyed Posidonia sinuosa meadows growing in different water depths to assess the variability in the sources, stocks and accumulation rates of Corg. We show that over the last 500 years, P. sinuosa meadows closer to the upper limit of distribution (at 2–4 m depth) accumulated 3- to 4-fold higher Corg stocks (averaging 6.3 kg Corg m−2) at 3- to 4-fold higher rates (12.8 g Corg m−2 yr−1) compared to meadows closer to the deep limits of distribution (at 6–8 m depth; 1.8 kg Corg m−2 and 3.6 g Corg m−2 yr−1). In shallower meadows, Corg stocks were mostly derived from seagrass detritus (88 % in average) compared to meadows closer to the deep limit of distribution (45 % on average). In addition, soil accumulation rates and fine-grained sediment content (

Abstract. Biotic and abiotic factors influence the accumulation of organic carbon (C org ) in seagrass ecosystems. We surveyed Posidonia sinuosa meadows growing in different water depths to assess the variability in the sources, stocks and accumulation rates of C org . We show that over the last 500 years, P. sinuosa meadows closer to the upper limit of distribution (at 2-4 m depth) accumulated 3-to 4-fold higher C org stocks (averaging 6.3 kg C org m −2 ) at 3-to 4-fold higher rates (12.8 g C org m −2 yr −1 ) compared to meadows closer to the deep limits of distribution (at 6-8 m depth; 1.8 kg C org m −2 and 3.6 g C org m −2 yr −1 ). In shallower meadows, C org stocks were mostly derived from seagrass detritus (88 % in average) compared to meadows closer to the deep limit of distribution (45 % on average). In addition, soil accumulation rates and fine-grained sediment content (< 0.125 mm) in shallower meadows (2.0 mm yr −1 and 9 %, respectively) were approximately 2-fold higher than in deeper meadows (1.2 mm yr −1 and 5 %, respectively). The C org stocks and accumulation rates accumulated over the last 500 years in bare sediments (0.6 kg C org m −2 and 1.2 g C org m −2 yr −1 ) were 3-to 11-fold lower than in P. sinuosa meadows, while fine-grained sediment content (1 %) and seagrass detritus contribution to the C org pool (20 %) were 8-and 3-fold lower than in Posidonia meadows, respectively. The patterns found support the hypothesis that C org storage in seagrass soils is influenced by interactions of biological (e.g., meadow productivity, cover and density), chemical (e.g., recalcitrance of C org stocks) and physical (e.g., hydrodynamic energy and soil accumulation rates) factors within the meadow. We conclude that there is a need to improve global estimates of seagrass carbon storage accounting for biogeochemical factors driving variability within habitats.

Introduction
The recent focus on carbon trading has intensified the interest in quantifying the capability of a variety of ecosystems to store carbon, since carbon storage provides one means of valuing these ecosystems. The role of seagrass meadows in absorbing and storing carbon dioxide over centennial to millennial scales is being evaluated in the context of climate change mitigation (Fourqurean et al., 2012;Duarte et al., 2013). Seagrasses occupy only 0.1 % of the ocean surface but are considered one of the largest carbon sinks worldwide (Duarte et al., 2005Mcleod et al., 2011). Unlike terrestrial ecosystems, which store organic carbon (C org ) mainly in the living biomass, C org stocks in seagrass meadows are mainly found in their soils, where it can accumulate over millennia . The substrates where seagrasses grow meet the requirements for sediment to be considered a soil , despite marine ecologists broadly referring to seagrass substrates as sediments (Kristensen and Rabenhorst, 2015).
Seagrasses encompass a wide variety of species across a range of depositional environments and water depths (Carruthers et al., 2007), and the variability in the soil C org stocks among seagrass habitats has been found to be high (up to 18-fold; Lavery et al., 2013). However, there has been a tendency to simplify regional and global estimates of C org stocks in seagrass ecosystems from a very limited data set, based on few species and habitats (Nelleman et al., 2009;Fourqurean et al., 2012). Geomorphological settings (i.e., topography and hydrology), soil characteristics (e.g., mineralogy and texture) and biological features (e.g., primary production and remineralization rates) control soil C org storage in both terrestrial ecosystems (Amundson, 2001;De Deyn et al., 2008;Jonsson and Wardle, 2009) and in mangrove and tidal salt marshes (Donato et al., 2011;Adame et al., 2013;Ouyang and Lee, 2014). However, our understanding of the factors regulating this variability in seagrass meadows is limited (Nellemann et al., 2009;Duarte et al., 2010;Serrano et al., 2014).
Based on the terrestrial analogues and the limited research undertaken on seagrasses, it is likely that multiple factors may influence C org storage within seagrass meadows, including biotic and abiotic factors acting in the water column, canopy and the soils. The seagrass itself may exert a primary control on C org storage through its biomass, productivity and nutrient content (Lavery et al., 2013;Serrano et al., 2014;Miyajima et al., 2015), all of which are highly variable depending upon seagrass species and habitat conditions (Alcoverro et al., 1995;Collier et al., 2007). Seagrass density, biomass and productivity are strongly related to the underwater light penetration (Dennison, 1987;Duarte, 1991). Therefore, it can be expected that different irradiance regimes (and therefore depth) would influence the C org storage capacity of seagrasses (Serrano et al., 2014).
Once C org is buried in the soil, biotic and abiotic factors are likely to control the degree of C org accumulation and preservation (Burdige, 2007). The rates of soil accumulation, the sediment structure and the biochemical composition of the organic matter buried may strongly influence C org accumulation and preservation, and are highly variable among seagrass meadows (De Falco et al., 2000;Kennedy et al., 2010;Duarte et al., 2013). Soil accumulation may be a function of the seagrass canopy structure (De Falco et al., 2000;Gacia and Duarte, 2001;Peralta et al., 2008;Hendriks et al., 2010), the availability of suspended particles to settle outside of the water column and the production of biogenic carbonates within the meadow (De Falco et al., 2000;Mazarrasa et al., 2015). If the accumulated sediments are fine, then they are likely to enhance the preservation of C org since they tend to limit oxygen exchange and redox potentials, which reduce remineralization (e.g., Keil and Hedges, 1993). In addition, finally, while both autochthonous (e.g., plant detritus and epiphytes) and allochthonous (e.g., seston and terrestrial matter) sources contribute to the C org pool in seagrass soils (Kennedy et al., 2010), the proportion of seagrass-derived C org may be an important factor controlling C org storage capacity. Seagrass tissues contain relatively high amounts of degradationresistant organic compounds (e.g., lignin and cellulose; Harrison, 1989;Klap et al., 2000;Torbatinejad et al., 2007;Burdige, 2007) compared to seston and algal detritus (Laursen et al., 1996), which are more prone to remineralization during early diagenesis (Henrichs, 1992).
From the above, it is clear that a large number of factors can potentially influence the stocks and accumulation rates of C org in seagrass meadows. Here we studied Posidonia sinuosa meadows across a depth gradient, aiming to highlight key biogeochemical factors affecting C org storage in seagrass soils that need to be accounted for when attempting to produce regional or global estimates of C org storage in seagrass meadows. Previous research at this site (Collier et al., 2007(Collier et al., , 2008 showed significant variation in plant biomass and productivity, water quality and sediment biogeochemistry parameters across this depth gradient. Bare sediments were also sampled and studied in order to determine the "background" C org stocks and fluxes in the absence of a seagrass meadow.

Study site and sampling
The study was conducted at Cockburn Sound in Western Australia (Fig. 1), in dense and monospecific P. sinuosa meadows across a significant depth gradient. Cockburn Sound is a sheltered marine embayment consisting of a deep central basin surrounded by shallow sand banks and seagrass meadows (Kendrick et al., 2002). Four vertical cores were sampled at four water depths in vegetated areas (1.6, 4, 5.7 and 8 m), while a single core at 4 m water depth was col-  Figure 1. Location of the study sites, Cockburn Sound, Western Australia (Australia). White dot points represent the coring sites in seagrass P. sinuosa meadows at 2, 4, 6 and 8 m depth (from west to east). Bare sediment core is indicated by a black dot point. lected from a bare area located at about 2 km distance from the nearest seagrass meadow. It was difficult or impossible to find a "pure control" (as per ecological definition) for this study. Shallow unconsolidated substrates in the study area should be occupied by seagrasses unless anthropogenic disturbances or hydrodynamic energy preclude so. In our case, the reference site was chosen based on the absence of seagrass at least since the 1960s (Kendrick et al., 2002), similar water depth (4 m), and the low likelihood of seagrass detritus from surrounding meadows being exported and accumulated in the area (Skene et al., 2005).
The core barrels consisted of PVC pipes (65 mm inside diameter) with removable coring heads to cut fibrous material and minimize core shortening (compression) during coring . The core barrels were driven into the soil by a hydraulic drill (LHD 23M, Atlas-Copco) that combined percussion and rotation. All cores were sealed at both ends, transported vertically to the laboratory and stored at 5 • C before processing.
The lengths of soil recovered ranged from 57 to 123 cm. Compression of loose soils during coring is an inevitable phenomenon and is routinely corrected by distributing the spatial discordances proportionally between the expected and the observed soil column layers (e.g., Glew et al., 2001). The overall degree of core shortening was low (less than 12 %) in all cases (corrected decompressed depths ranged from 65 to 134 cm). The results reported in this study (i.e., density, soil accumulation rates and C org stocks and accumulation rates) have been corrected for compression.

Laboratory procedures
The cores were cut longitudinally into two halves and sliced at regular intervals (i.e., 1 cm thick slices). Each slice/sample was weighed before and after oven drying to constant weight at 70 • C (DW), and subsequently subdivided for analysis. The C org elemental and isotopic composition of the organic matter was measured in milled subsamples from every second slice. These subsamples were acidified with 1 M HCl, centrifuged (3500 RPM; 5 min) and the supernatant with acid residues was removed using a pipette, then washed in deionized water once, the residues were centrifuged again and the supernatant removed. The residual samples were re-dried (70 • C) before carbon elemental and isotopic analyses. Samples were acid-rinsed to ensure complete removal of inorganic carbon (i.e., carbonates) before C org analysis, despite the fact that this procedure may lead to an underestimation of soil C org stocks (Phillips et al., 2011;Brodie et al., 2011). The C org elemental and isotopic composition was also analyzed in P. sinuosa macro-detritus (i.e., sheaths, roots and rhizomes) collected at different depths along all seagrass cores for the carbon source study. The samples were washed in deionized water, dried at 70 • C and encapsulated, and the C org elemental and isotopic composition was analyzed using a Micro Cube elemental analyzer (Elementar Analysensysteme GmbH, Hanau, Germany) interfaced with a PDZ Europa 20-20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK) at University California Davis facilities. The relative contents of C org were calculated for the bulk (preacidified) samples. Carbon isotope ratios are expressed as δ values in parts per thousand (‰) relative to VPDB (Vienna Pee Dee Belemnite).

Age-depth chronology
The age of the soil along the cores was determined combining 210 Pb and AMS-14 C techniques for the recent (ca. < 100 yr BP) and older (ca. < 500 cal yr BP) material, respectively. Concentrations of 210 Pb were determined by alpha spectrometry through the measurement of its granddaughter 210 Po, assuming radioactive equilibrium between the two radionuclides (Sánchez-Cabeza et al., 1998). Between 150 and Table 1. Average ± standard error (SE) density (in g cm −3 ), C org content (in %), δ 13 C signatures and sediment grain-size content at Cockburn Sound (normalized for ca. 500-year-old deposits).  *** *** *** *** 300 mg aliquots of each sample were acid-digested after addition of 209 Po as spike and polonium isotopes were plated onto pure silver disks, and their alpha emissions were measured by alpha spectrometry. The concentrations of 210 Pb at depths which were found to be constant were used to determine the average supported 210 Pb concentrations, which were then used to obtain the concentrations of excess 210 Pb. A selection of samples of each core was measured for 226 Ra by gamma spectrometry to confirm the validity of the esti-mates of 210 Pb-supported values. Concentrations of 226 Ra were determined using a high-purity Ge well-type detector (CANBERRA, mod. GCW3523) through the 351 keV emission line of 210 Pb. For radiocarbon analyses, four samples of shells and one sample of P. sinuosa sheath remains were radiocarbon-dated at the National Ocean Sciences AMS Facility (Woods Hole Oceanographic Institution, Woods Hole, MA; Table A1 in Appendix A) following standard procedures (Stuiver and Pollack, 1977). Sheaths and shells were washed in ultrapure MQ water in order to remove fine sediment particles, examined under a stereomicroscope for lack of attached reworked materials and dried at 60 • C before radiocarbon dating. The conventional radiocarbon ages were converted into calendar dates in years BP (cal yr BP) using the Calib 7.1 software (Marine13 curve), and the local marine reservoir effect due to the C dissolved in marine water was adjusted by deducting 71 years from the calibrated radiocarbon ages (Ulm, 2006). The calibrated 14 C ages corrected for the marine reservoir effect were used to produce an age-depth model (linear regression; present is 2012).

Numerical procedures
The C org stocks per unit area (kg C org m −2 ) were estimated by computing the cumulative mass of C org accumulated over the last ca. 100 and 500 years (inventories in 13 to 30 and 40 to 75 cm thick deposits, respectively). The short-and longterm accumulation rates (g DW m −2 yr −1 ) of C org were calculated by dividing the C org inventories in the soil by the ages (for 100-and 500-year-old deposits, respectively). The decay rates of soil C org were calculated by fitting an exponential equation to the decreasing trends in C org content (mg C org cm −3 ) with aging. The data reported for seagrass soil properties at different water depths and bare sediments (Average ±SE) were normalized for ca. 100-and/or 500year-old deposits (specified in each case).
A one-way ANOVA (analysis of variance) test was applied to test for any significant effect of water depth on the C org elemental and isotopic composition, C org stocks and accumulation rates and fine sediment content (< 0.125 mm). When significant effects were detected, pairwise a posteriori comparisons were performed using a Tukey's HSD (honest significant difference) test. Data were fourth root transformed Table 3. Soil accumulation rates (SARs), C org accumulation rates and C org inventories in the seagrass cores studied (average ± standard deviation). Estimates over short-term (derived from 210 Pb dating, last 100 years) and long-term (derived from 14 C dating, last 500) periods are provided. The thicknesses of seagrass soils corresponding to 100 and 500 years are provided.

Water
Short-term (100 years) Long-term (500 years) depth (m) Thickness Stock SAR C org acc. rates Thickness Stock SAR C org acc. rates  (Parnell et al., 2010) was used to estimate the contribution of potential sources to the sedimentary C org . The model was run with three sources (seagrass detritus, epiphytes/macroalgae and seston). Separate mixing models were computed for each core and for both 100 and 500 years of accumulation. The δ 13 C values for all sources were assumed to be constant for each core, except the δ 13 C signatures of seagrass detritus. Previous studies showed that the δ 13 C values of P. sinuosa varied along this depth gradient (Collier et al., 2008). To account for this variability in seagrass tissue δ 13 C, the δ 13 C signatures of seagrass detritus measured directly in the seagrass detritus present in each core were used in the corresponding mixing model. Concentration dependence was incorporated into the model because elemental concentrations were different between sources (Phillips and Koch, 2002). We did not consider any fractionation with aging (0 ± 0 ‰) in the model because previous studies suggest small diagenetic shifts for δ 13 C during decomposition (Zieman et al., 1984;Mateo et al., 2010).

Results
The soil characteristics of the P. sinuosa meadows in Cockburn Sound changed significantly with soil depth (and thus age), starting as low-density, highly organic soils that turned into inorganic-dominated material 20 cm below the soil surface (after ca. 50 to 150 years of burial; Fig. A1 in Appendix A). Over 500 years of accumulation, soils in the P. sinuosa meadows closer to the upper limit of distribution (at 2 and 4 m depths) were significantly richer in C org (mean ± standard error of the mean = 1.2 ± 0.2 % C org ) than those from deeper areas (at 6 and 8 m depths; 0.5 ± 0.1 % C org ; Tables 1 and 2). The properties of the bare sediment core were homogeneous with depth/age (Fig. A1) and, on average, the C org content was lower (0.06 %) and the density higher (1.2 g cm −3 ) compared to the vegetated cores (Tables 1 and 2). Medium and fine sands dominated in all seagrass cores (87 % in average), while medium and coarse sands dominated in the bare sediment core (78 % in total; Table 1 and Fig. 2a). The proportion of fine grain-size material (< 0.125 mm) increased from the bare core (averaging 1 %) to P. sinuosa meadows closer to the deeper limit of distribution (4-5 % at 6 and 8 m depth) and meadows closer to the upper limit of distribution (6 to 11 % at 2 and 4 m depth; Tables 1 and 2). Concentration profiles of 210 Pb showed decreasing trends from the surface down to depths of 10 to 16 cm (decompressed depths). The concentrations of 226 Ra (average: 0.4 ± 2.1 Bq kg −1 ) were in agreement with those of 210 Pb in the deepest sections of the cores, indicating absence of excess 210 Pb ( 210 Pb ex ; Fig. 3). All cores had similar concentrations of supported 210 Pb (10.5 ± 0.9 Bq kg −1 ), whereas the 210 Pb ex inventories in the vegetated soils ranged from 427 ± 45 to 723 ± 48 Bq m −2 . Mixing of the upper soil layers was most severe in seagrass cores from the 2 and 6 m depth sites, where mixing was apparent in the top 3 and 7 cm, respectively. Average short-term soil accumulation rates (SARs; ca. last 100 years) for each core were determined by applying the Constant Flux : Constant Sedimentation (CF : CS) model below the base of the mixed layer (Krishnaswamy et al., 1971;Masqué et al., 2002; Fig. 3), ranging from 1.3 ± 0.2 to 3.0 ± 1.1 mm yr −1 (Table 3) concentrations measured in the reference core (i.e., bare sediment) were low (10.1 ± 1.2 Bq kg −1 ) and not statistically different from the supported 210 Pb concentrations measured in the P. sinuosa cores (10.4 ± 1.2 Bq kg −1 ). The absence of excess 210 Pb accumulation in bare sediment suggests negligible recent net accumulation of 210 Pb (and thus sediments) in the absence of vegetation (i.e., last ca. 100 years). According to the age-depth models based on 14 C ages, long-term SARs (ca. last 500 cal yr BP) in P. sinuosa cores ranged from 0.8 to 1.3 mm yr −1 , while long-term SARs in bare sediments averaged 1.5 mm yr −1 (Table 3). Over 100 and 500 years of accumulation, the shallow P. sinuosa meadows (at 2 and 4 m depths) stored more carbon (averaging 4.0 and 6.3 kg C org m −2 , respectively) than the deeper counterparts at 6 and 8 m depths (1.2 and 1.8 kg C org m −2 , respectively; Table 3 and Fig. 4). The lowest C org inventories (500 years of accumulation; 0.6 kg C org m −2 ) and accumulation rates (1.2 g C org m −2 yr −1 over 500 years) were found in the bare sediment core. The soil C org content (mg C org cm −3 ) in the shallower meadows (at 2 and 4 m depth) decreased exponentially at rates of 0.0058 ± 0.0012 yr −1 (R = 0.76) and 0.0043 ± 0.0005 yr −1 (R = 0.86), respectively, while in meadows closer to the deeper limit of distribution (at 6 and 8 m depth) it decreased at 0.0037 ± 0.0014 yr −1 (R = 0.65) and 0.0085 ± 0.0011 yr −1 (R = 0.92), respectively.
The δ 13 C values of sedimentary organic matter in soils from shallow meadows (at 2 and 4 m depths) were higher (−12 ‰) than those from the 6 and 8 m depths (−14 to −16 ‰; Fig. 2b; Tables 1 and 2). Organic carbon in bare sediments was the most depleted in 13 C (overall mean -20 ‰). Carbon isotopic ratios in extant seagrass tissues also varied between cores (Table 4a). On average, δ 13 C signatures of seagrass detritus preserved in the cores at 2, 4 and 6 m water depth were 13 C-enriched (−10 to −11 ‰) compared with those from 8 m depth (−13 ‰). The δ 13 C signatures of living epiphytes and macroalgae at Cockburn Sound averaged −16 and −19 ‰, respectively (Table 4a).
The mixing models applied indicated that seagrass detritus was the most important source of soil C org in all meadows studied (ranged from 43 to 94 %; Table 4b) over 500 years of accumulation, but its contribution decreased with water depth. In meadows closer to the upper limit of distribution (at 2 and 4 m depth), seagrass-derived detritus contributed 80 to 94 % of the sedimentary C org , about 2-fold higher than in deeper meadows (at 6 and 8 m depth; ranging from 43 Figure 4. Inventories of C org (kg C org m −2 ) in P. sinuosa meadows (at 2, 4, 6 and 8 m depth) and bare sediments at Cockburn Sound (normalized for ca. 500-year-old deposits).
The contribution of seston increased with depth, but was always less than the contributions from Posidonia and epiphytes/macroalgae (Table 4b). Bare sediments had the lowest seagrass contribution to the C org pool and the highest proportion from seston (20 and 58 %, respectively; Table 4b).

Discussion
The results show a consistent decline in C org stocks and accumulation rates with water depth in P. sinuosa meadows, where shallow meadows closer to the upper limit of distribution accumulated 3-to 4-fold higher C org stocks and at higher rates than those nearer the depth limits of distribution. We interpret the associated changes in biological (e.g., productivity, cover and density), chemical (e.g., recalcitrance of C org stocks) and physical (e.g., hydrodynamic energy and SAR) factors within the meadows as evidence that the production, trapping and preservation of soil C org in coastal areas are the result of complex interaction among all three sets of factors, as we represent in Fig. 6, and as is discussed below. The findings from this study are consistent and complement earlier findings by Serrano et al. (2014). The analyses of new variables in the same cores (i.e., 210 Pb dating, sediment grain-size, stable carbon isotopes in organic matter) provided new insights into the factors driving differences in C org storage along a depth gradient. We also compare the biogeochemical characteristics of seagrass soils with adjacent bare sediments. Differences in C org stocks and accumulation rates between this and the previous study (Serrano et al., 2014) are related to the new age-depth models obtained in the cores (i.e., based on 210 Pb dating). The results obtained lead us to conclude that in order to assess differences and compare C org storage between seagrass habitats, it is recommended that C org stocks be normalized by a period of accumulation (Rozaimi et al., 2016), rather than soil depth as commonly used (e.g., Serrano et al., 2014). Therefore, we present the results and develop the discussion according to the period of accumulation ( 210 Pb-derived, short-term, last 100 years; and 14 C-derived, long-term, last 500 years).
The results indicate that the P. sinuosa plants themselves play a key role in determining the amount of C org available for burial along the depth gradient. It is well established that accumulation of C org in sediments and soils is strongly affected by net primary production (Cao and Woodward, 1998;Serrano et al., 2014). The decline with depth of C org stocks, C org accumulation rates and seagrass-derived inputs into the sedimentary pool that we observed coincides with reduced seagrass abundance and production reported by Collier et al. (2007). These authors reported 18-24-fold reductions from shallow (2 m) to deep (8 m) P. sinuosa meadows in shoot density (from 1435 to 80 shoots m −2 ), aboveground biomass (from 899 to 47 g DW m −2 ) and belowground biomass (from 1028 to 43 g DW m −2 ) on the same depth gradient. Similar trends in meadow structure and productivity with depth have been found in other Posidonia meadows, linked to reductions in irradiance (West, 1990;Duarte, 1991;Alcoverro et al., 2001;Olesen et al., 2002).
Relationships between water column depth, seagrass canopy structure and C org stocks have been reported for Zostera muelleri and Halophila ovalis meadows (e.g., Samper-Villarreal et al., 2016). However, previous studies based their comparisons on soil thickness rather than C org accumulation rates (e.g., period of accumulation) and rely on the assumption that environmental gradients linked to e.g., anthropogenic disturbances remained constant over the period reconstructed. Seagrass meadow structure (e.g., density, cover, biomass) and even presence/absence can vary over seasonal, annual and decadal timescales, in particular for short-lived and highly dynamic meadows such as those formed by genera Zostera, Halophila and Halodule. The presence of a clear and stable environmental gradient (i.e., depth) over the last millennia (Skene et al., 2005), together with the presence of seagrass remains along the cores studied, provides further evidence for the relationships between biogeochemical factors and seagrass soil C org storage reported in this study.
The higher SAR, fine-grained sediment contents and plant detritus inputs in meadows closer to the upper limit of distribution would contribute to higher accumulation and preser- Figure 5. Biplots showing the relationships among the variables studied in the seagrass and bare sediment cores from Cockburn Sound (normalized for 500-year-old deposits). (a) δ 13 C signatures (‰) plotted against C org stocks (mg C org cm −3 ); (b) sediment grain size < 0.125 mm (%) plotted against C org stocks (mg C org cm −3 ); (c) δ 13 C signatures (‰) plotted against sediment grain size < 0.125 mm (%); and (d) contribution of seagrass detritus (%) plotted against soil C org stocks (kg C org m −2 , over 100 years -small symbols -and 500 years -big symbols -of accumulation).  Figure 6. Influence of biogeochemical factors on the organic carbon storage capacity of seagrass ecosystems. Organic carbon in seagrass soil increases with high SAR, fine sediment content, seagrass NPP and density; and decreases with high hydrodynamic energy and water depth. SARs, soil accumulation rates; NPP, net primary production.

Soil organic carbon
vation of C org after burial. The SAR in seagrass meadows is mainly controlled by the canopy structure, which affects the trapping and retention of sediment particles (Gacia and Duarte, 2001;Peralta et al., 2008;Hendriks et al., 2010), the hydrodynamic energy, the availability of fine-grained suspended particles in the water column and the production of biogenic carbonates within the meadow (De Falco et al., 2000Mazarrasa et al., 2015). High plant biomass and density is associated with greater retention of particles (in particular, fine-grained sediments), lower hydrodynamic energy and higher production of biogenic carbonates within the meadow (De Falco et al., 2000), ultimately enhancing soil accumulation. The presence of a dense rhizome mat underlying shallow meadows may provide a positive feedback mechanism for enhanced SARs (i.e., presence of cavities reducing erosion and increasing soil accumulation; De Falco et al., 2000;Le Hir et al., 2007). The higher content of fine sediments we observed in shallow meadows would contribute to the higher C org accumulation, since fine sediments generally retain more C org compared to medium and coarse sands (Keil and Hedges, 1993;Burdige, 2007), and because rem-ineralization rates tend to be reduced in fine sediments due to lower oxygen exchange and redox potentials (Hedges and Keil, 1995;Dauwe et al., 2001;Burdige, 2007;Pedersen et al., 2011).
The differences in decay rates highlight different levels of C org preservation in the different meadows. This is likely a result of both the sources of C org being buried and the biogeochemical conditions within the soils. Previous studies demonstrated that both autochthonous (e.g., seagrass and epiphyte detritus) and allochthonous (seston and terrestrial matter) sources contribute to the C org pool in seagrass soils (50 % each on average; Kennedy et al., 2010). Here, we observed larger amounts of seagrass-derived C org in shallow meadows (85 % in average), pointing to an important factor driving their higher C org storage capacities compared to that of deeper meadows, namely the carbon preservation potential. Posidonia tissues contain relatively high amounts of degradation-resistant organic compounds in their tissues (e.g., lignin and cellulose; Harrison, 1989;Klap et al., 2000;Torbatinejad et al., 2007) and high C / N ratios (Duarte, 1990;Pedersen et al., 2011;Kaal et al., 2016). In contrast, seston and algal detritus, which contributed as much as 64-75 % of the C org in the deeper sites, have a higher labile C org content (Laursen et al., 1996) more likely to be remineralized during early diagenesis (Henrichs, 1992), potentially explaining the higher soil C org decay rates in the deep (at 8 m) P. sinuosa meadows. However, the soil C org decay rates in P. sinuosa meadows at 6 m depth were in the range of those found at 2 and 4 m depths. This may be due to the limitations of the approach used here. For example, we assumed that C org inputs (i.e., quantity and quality) and decomposition have been constant during the period of accumulation under study, but this may not have been the case. Further, obtaining reliable estimates of C org decay rates is also complicated by the presence of living biomass in the upper part of the soils, which is the case for the seagrass core sampled at 6 m depth, where fluctuations in the concentration of C org are evident.
The C org decay rates of P. sinuosa meadows (0.0056 yr −1 in average) are much higher than those reported for the similarly sized species P. oceanica (ranging from 0.00008 to 0.0005 yr −1 ; Serrano et al., 2012). This may contribute to the up to 16-fold lower C org stocks and accumulation rates in the soil beneath P. sinuosa compared to P. oceanica (Serrano et al., 2014).
Despite the limitations involved in using bare sediments as reference sites (e.g., inherent biogeochemical differences that preclude the settlement of seagrasses in bare sediments), the results suggest that C org stocks and accumulation rates are much higher in seagrass meadows than in adjacent bare sediments. The 3-to 11-fold lower C org storage capacity of bare sediments compared to P. sinuosa meadows at comparable depths is mainly due to the absence of seagrass inputs. However, it may also result from the absence of a canopy that would otherwise enhance the trapping and retention of organic-rich, fine sediment particles (Hendriks et al., 2008), as reflected in the low content of fine-grained sediments. Since all continental margins store C org , there is a need to account for the net C org storage capacity due to the presence of seagrasses when evaluating their role as carbon sinks.
The processes described in this study highlight the importance of meadow structure and productivity for C org accumulation, supporting the hypothesis that the higher production of shallow meadows leads to higher accumulation rates of soil, fine-grained particles and seagrass detritus, which ultimately lead to the higher preservation and accumulation of C org . The relative importance of the biogeochemical factors identified in this study (e.g., hydrodynamic energy, sediment accumulation rates, fine sediment content, water depth, seagrass net primary production and density) in driving C org storage was not addressed, but rather we discussed the reasons why they can play a role in driving organic carbon storage and highlight potential synergistic and/or antagonistic interactions among them. Understanding the factors controlling C org storage in seagrasses is at its onset, and a much better understanding is required before being able to disentangle the relative role/importance of each factor.

Data availability
Data supporting this study are available at Edith Cowan University Research Online portal.