Metabolomics and traditional indicators unveil stress of a seagrass (Cymodocea nodosa) meadow at intermediate distance from a fish farm

20 Seasonal variation of structural, physiological and growth indicators and the metabolome of 21 the seagrass Cymodocea nodosa, as well as biogeochemical conditions of underlying sediment 22 were studied in two meadows growing at increasing distance downstream from a fish farm in 23 the Aegean Sea in order to assess seagrass performance under stress. Horizontal rhizome 24 production decreased significantly with proximity to the fish farm (0.67 and 1.57 g DW m d 25


Introduction 43
Seagrasses are indicators of environmental deterioration, as meadow declines often point to 44 disturbances that affect the entire ecosystem (Orth et al., 2006). ' values (Marbà et al., 2006). 118 In this paper we assess the performance of C. nodosa meadows at intermediate distance from 119 a fish farm using traditional seagrass-based indicators and metabolomics. We do so by 120 quantifying the seasonal variability in nutrient availability in water, pore water and sediment, 121 along with structural (density and biomass), physiological (elemental and isotopic composition 122 of carbon, nitrogen and sulfur) and growth indicators (rhizome production) and the metabolome 123 of seagrasses growing at intermediate distance and away from a fish farm in the Aegean Sea 124 (Greece). We also explore the relationships of C. nodosa metabolites of stress and growth with 125 horizontal production, to examine if C. nodosa has obstructed performance. The combination 126 of traditional ecological indicators and metabolomics provides more understanding of how 127 environmental stressors affect seagrass stress response. This is the first work to combine 128 traditional seagrass-based indicators and metabolomics to reveal stress of seagrass meadows. 129 7

Materials and Methods 131
Sampling Strategy 132 Sampling was carried out at Vourlias Bay (Peloponnese, Greece), which is an allocated 133 aquaculture zone with intense fish-farming activity (9 farms). The site was located 134 approximately 2 km east of the main bay, separated by a headland where there is a fish-farm 135 with 28 cages (37° 27' 08" N 23° 04' 20" E) (Fig. 1). Apart from this aquaculture facility, there 136 were no other features (e.g. rivers, towns) that could cause a disturbance to the environment. 137 Two sampling stations were selected which will henceforth be referred to as 'Station A' and 138 'Station B' at 440 m and 780 m distance from the fish-farm, respectively. The seagrass meadow 139 at Station A was the closest we could find to the fish-farm. Phosphorous sedimentation in 140 seagrass meadows of the Mediterranean, which has been proved a reliable indicator of fish-141 farm waste, has been shown to decrease exponentially with distance from the cages and the 142 sedimentation rates were roughly 100% and 40% higher at 250 m and 500 m from the fish-143 farm, respectively, than at non-impacted sites (Holmer et al., 2008). Yet, seagrass meadows 144 growing 800 m downstream from fish farms have shown no evidence of impact (Marbà et al., 145 2006). Based on the above, it could be expected that there is an effect at 440 m (Station A) and 146 that this effect should be low at Station B, and therefore we can expect to detect differences 147 between stations. The predominant coastal currents in the area move from the North-West, 148 travelling through Vourlias Bay and towards our sample sites, passing the adjacent fish farm. 149 Station A is situated in a small bay where the currents circulate, whereas the currents pass 150 Station B and move down the coast to the South-East (data taken from model described in 151 Tsagaraki et al., 2011). 152 The site was visited twice, during a warm (June 2017) and a cold (March 2018) period, when 153 water temperature was 29° C and 15° C respectively. The water depth was 8 m at both stations.

ACCEPTED MANUSCRIPT
bundles from each of the 5 replicates were rinsed using milliQ water and scraped free from 180 epiphytes before they were put in liquid nitrogen and later stored at -80 °C, to be analysed for 181 metabolites and nutrient content. The remaining seagrass tissue was kept frozen. 182

Laboratory Analysis 183
Ammonium analysis of water samples was carried out using the indophenol blue method 184 (Ivancic & Degobbis, 1984) and determination of phosphate was carried out as described by 185 Strickland and Parsons (1972 190 Sediment was dried at 60 °C for 48 hours to attain density, water content and porosity 191 measurements. The dried sediment was pulverised before being measured for elemental and 192 isotopic composition. Total carbon, nitrogen and 15 N were analysed in sediment as it was. The 193 sediment was then acidified for the determination of organic carbon (OC) and 13 C. The 194 inorganic carbon (IC) was estimated as the difference between total carbon and OC. The 195 elemental and isotopic composition was expressed in % and δ unit notation (‰ deviations from 196 the established international standards ratio), respectively. 197 For sediment 34 S sulfide measurements we used the two-step distillation method (Fossing and 198 Jorgensen, 1989) to obtain the chromium reducible sulfur (CRS) and acid volatile sulfur (AVS) 199 fractions. A 5-10 g fraction of homogenised sediment was combined with 10 mL 50% ethanol 200 in a reaction flask, and degassed with N2 for 10 minutes. The AVS fraction was attained by 201 distilling the mixture for 30 minutes at room temperature with 8 mL 12M HCl. The CRS 202 fraction required reduction using 1M Cr2 + in 0.5M HCl, and then boiling and distilling for 30 203 minutes. The sulfide extract was collected as Ag2S precipitate in traps and δ 34 S was analysed 204 as described below. Granulometry assessment of the sediment was carried out by removing all 205 grains larger than 63 µm (sand) by wet sieving. The remaining sediment was then categorised 206 using a Sedigraph (Micromeritics 5100) into < 63 µm (silt) and > 0.1 µm (clay). 207 The young seagrass tissue was freeze-dried to be analyzed for total sulfur (TS) and the δ 34 S 208 isotopic fraction and their metabolic profile. The remaining (leaves, rhizomes and roots) tissue, 209 which had not been quenched in liquid nitrogen, was dried at 60 °C for 48 hours and 210 homogenized with the young tissue to determine dry weight. Young and remaining tissue were 211 ground to a powder to be used for analysis of elemental and isotopic composition on total 212 carbon (TC), nitrogen (TN), δ 13 C, δ 15 N and δ 34 S using elemental analyser combustion 213 continuous flow isotope ratio mass spectroscopy. The δ 34 S measurement required the ground 214 up seagrass tissue to be weighed into tin capsules along with vanadium pentoxide. Horizontal rhizome production (g DW m -2 y -1 ) was calculated from the measured rhizome 228 internodal lengths. First a 30% running average was applied to the data, in order to account for which allowed an estimation of annual horizontal rhizome elongation rate per apex. The 233 following formula was then applied to the data: 234 Where annual production (cm y -1 ) is each annual cycle measured from the consecutive minima, 239 horizontal weight (g) is the total dry horizontal rhizome weight of the ramet, apex density (m -240 2 ) is the number of apical shoots per metre, and horizontal length (cm) is the total length of the 241 ramet. We display here the mean production measurements of the two sampling campaigns. 242 Biomass (g DW m -2 ) was estimated as the dry weight of the seagrass tissue; leaves, rhizomes 243 or roots, divided by the sampled area. Shoot density (shoots m -2 ) was estimated as the number for each PC. Heatmaps were used to visualise how stress and growth-related metabolites 268 differed between stations. Linear regression analysis was carried out on the Fsulfide data in order 269 to identify sulfide intrusion and also in order to detect possible relationships between 270 metabolites and growth. Correlations were carried out using Spearman's rho to assess whether 271 stress and growth associated metabolites showed positive or negative trends with horizontal 272 rhizome production. The statistical significance of these correlations were tested using one-273 tailed and two-tailed t-tests. As it was possible to predict the trend direction in advance, using 274 one-tailed t-testing was appropriate. As two-tailed t-tests protect against type-I errors and 275 cognitive biases, we also report it. 276 Horizontal rhizome production and root biomass were decreased two-fold at Station A 291 compared to Station B (Table 2). Rhizome biomass had an interaction effect between station 292 and season, with the biggest variation between seasons at Station A, twice as high in the warm 293 period. The majority of the structural variables were enhanced during the warm period (shoot 294 density by 83 %, leaf biomass by 218 %, total biomass by 113 % and above:below ground 295 biomass ratio by 100, Table 2). 296

ACCEPTED MANUSCRIPT
The majority of the measured elemental concentrations and isotopic values were significantly 297 elevated at Station A compared to Station B: Leaf C by 1.5 %, rhizome N by 27 %, root N by 298 37 %, rhizome S by 9.2 %, root S by 30 %, leaf δ 13 C by 11 %, rhizome δ 13 C by 2.4 %, root 299 δ 13 C by 1.3 %, leaf δ 34 S by 0.9 %, root δ 34 S by 2.8% (Table 2). Some variables (S of leaves 300 and rhizomes, δ 15 N of all tissues, and rhizome δ 34 S) also had a station effect, but this depended 301 on the season (Table 2). Seasonality was shown by many of the variables being elevated either 14 ACCEPTED MANUSCRIPT in the warm (C of all tissues and leaf δ 34 S) or the cold period (N of all tissues, Table 2). Fsulfide 303 did not show significant differences between stations or sampling periods. Fsulfide linear 304 regressions of leaves and rhizomes (R 2 = 0.58, p < 0.01) and of rhizomes and roots (R 2 = 0.39, 305 p < 0.05) were positively related indicating sulfide intrusion in the plants from the roots. 306 PCA analysis on metabolite levels in C. nodosa leaves revealed clear separations between 307 stations but not seasons (Fig. 2). Metabolites in rhizome tissue showed a seasonal difference, 308 and root metabolites separated mainly by season. However, there was also separation between 309 Station A and B during the winter. 310 Mass spectrometry detected 203 leaf metabolites. A portion of them were related to seagrass 311 stress or growth and varied between stations (Fig. 3). Analysis of the C. nodosa metabolome 312 was carried out on leaves and roots. Particular focus was given to the leaves as this is the most 313 active metabolic tissue in plants which has shown the biggest changes in metabolite 314 composition in response to abiotic stress (Obata et al., 2015). Growth and stress-related 315 metabolites varied between stations (Fig. 3, Table 3). Growth-promoting metabolites including 316 sugars and sterols were lower at Station A. The growth metabolite with the largest difference 317 between stations was fructose (2.5 times lower at Station A). Several stress-indicating 318 metabolites including some amino acids were higher at Station A, and two metabolites known 319 to decrease during abiotic stress (2-oxoglutaric acid and stearic acid) were lower at Station A 320 (Fig. 3). Remarkably putrescine was 112 times higher at Station A. Spearman correlations 321 between horizontal rhizome production and growth-promoting metabolite levels in the leaves 322 yielded positive correlations (Table 4), the strongest significant (p<0.05) correlation being with 323 the sugar myo-inisitol (rs = 0.647). Spearman correlations between horizontal rhizome 324 production and stress-indicating metabolites yielded significant negative correlations except 325 for 2-oxoglutaric acid, which showed a positive correlation. The strongest correlation was with 15 ACCEPTED MANUSCRIPT relationship between horizontal rhizome production and alanine, and a significant positive 328 relationship between horizontal rhizome production and sucrose (Fig. 4). 329 Of the 20 most significant (ANOVA) metabolites in root tissue, a large proportion of amino 330 acids were higher at Station A during cold period (Fig. 5). We found significant negative 331 relationships between leaf glutamate and horizontal rhizome production (R 2 = -0.45, Fig. 6 Ongoing abiotic stress is expected to lead to reduced plant performance, and therefore less 391 growth, which is in accordance with the negative correlations found between 'stress 392 metabolites' and horizontal rhizome production. 393 The increase of amino acids in the roots at Station A during the cold period, coinciding with 394 higher pore water ammonium levels, could be explained by increased ammonium assimilation. 395 Rapid entry of ammonium into the cells causes many changes to seagrass physiology including 396 impaired ATP production and photosynthetic electron transport (Alexandre et al., 2018; 397 Villazán et al., 2013). To minimize these effects, seagrasses rapidly assimilate ammonium and 398 convert it into amino acids, which depletes carbon skeletons (Brun et al., 2002, 2008). 399 The GOGAT cycle, where nitrate or ammonium are ultimately integrated into amino acids, is 400 recognised as the most important route of nitrogen assimilation in higher plants (Robinson et Table Legends  793   Table 1. Ammonium, phosphate and δ 34 S sulfate of bottom water and pore water for each 794 station. Organic (OC) and inorganic (IC) carbon and total nitrogen (TN) pools and δ 34 S sulfide 795 of seagrass sediments. Significant differences (P < 0.05) between stations are given by capital 796 letters and between season by small letters, * indicates station X season effect. 797 798 Table 2. Shoot density, leaf, rhizome and root biomass, above to below ground biomass ratio, 799 total biomass and horizontal rhizome production, and average TC, TN, TS, δ 13 C, δ 15 N, δ 34 S, 800 Fsulfide AVS and Fsulfide CRS for leaves rhizomes and roots at the two stations during the warm 801 and cold periods. Significant differences (P < 0.05) between stations are given by capital letters 802 (A) and between seasons by small letters (a), * indicates station X season effect. 803 804 Table 3. Two-way ANOVA testing of stress-associated and growth-associated metabolites in 805 seagrass leaves. 806 807 Table 4. Spearman correlations between horizontal rhizome production and metabolites 808 associated with growth or stress in seagrass leaves in both sampling periods.