Fatty Acids to Quantify Phytoplankton Functional Groups and Their Spatiotemporal Dynamics in a Highly Turbid Estuary

Phytoplankton community composition expresses estuarine functionality and its assessment can be improved by implementing novel quantitative fatty acid–based procedures. Fatty acids have similar potential to pigments for quantifying phytoplankton functional groups but have been far less applied. A recently created dataset containing vast information on fatty acids of phytoplankton taxonomic groups was used as reference to quantify phytoplankton functional groups in the yet undescribed Guadalquivir River Estuary. Twelve phytoplankton groups were quantitatively distinguished by iterative matrix factor analysis of seston fatty acid signatures in this turbid estuary. Those phytoplankton groups including species unfeasible for visual identification (coccoid or microflagellated cells) could be quantified when using fatty acids. Conducting monthly matrix factor analyses over a period of 2 years and the full salinity range of the estuary indicated that diatoms dominated about half of the phytoplankton community spatiotemporally. The abundance of Cyanobacteria and Chlorophytes was inversely related to salinity and little affected by seasonality. Euglenophytes were also more abundant at lower salinity, increasing their presence in autumn–winter. Coccolithophores and Dinophytes contributed more to phytoplankton community at higher salinity and remained little affected by seasonality. Multivariate canonical analysis indicated that the structure of the estuarine phytoplankton community was most influenced by salinity; secondly influenced by water temperature, irradiance, and river flow; and unaffected by nutrients. Fatty acids are especially suited for phytoplankton community research in high turbid estuaries and generate outcomes in synergy with those derived from classical pigment assessments.

8 153 along its course, and the estuary section between the river mouth and the port of Seville 154 has been converted into an 84 km navigating channel (50 km shorter than the original 155 course). The GRE keeps a minimum depth of 6.5 m by means of periodic dredging and 156 it is completely isolated from the fringing marshes. As a result of this strong regulation, 157 freshwater inputs to the estuary have decreased an average 60%, with extreme reduction 158 in dry years [20]. The Alcala dam is the last hydrological regulation point and sets the 159 upper limit of the estuary, 110 km upstream the river mouth. The estuary has a cross- 190 Identifying microalgae at the genus rank allowed creating phytoplankton functional 191 groups at the class level needed for comparison with results from the inferring models.

192
193 Due to excess of suspended inert material, particle retention in glass fiber filters was 194 inefficient to ascertain adequate sample size for lipid extraction. To overcome such 195 inconvenience, a total of 25 l water sample was centrifuged using a continuous 196 centrifuge RINA SRP 2C (Riera Nadeu SA, Barcelona, Spain) equipped with a 0.2 l 197 clarifying capacity rotor. The centrifuge was operated at 7000 g with an estuary water 198 sample constant inflow of 90 l h -1 . After centrifugation of every 25 l water sample, the 199 centrifuge was stopped and the rotor detached to enable collecting the centrifuged 200 particles from a Teflon sheet covering the inner wall of the rotor. The centrifuged 201 material was easily scraped with a spatula, lyophilized and transferred to storage bottles 202 that were kept frozen (-40ºC) in a nitrogen atmosphere prior to analysis. Samples from 203 the outflowing centrifuged water (10 l) were filtered through pre-combusted (4 h at 204 450ºC) glass fiber filters (Whatman GF/F, 47 mm, 0.7 µm nominal pore size) in order to 205 check for the efficiency of the centrifugation process concerning to total solid, organic 206 matter and lipid retention. Total suspended solids and organic matter content was 207 gravimetrically determined after drying (60ºC) filter-retained material until constant 208 weight and subsequent combustion at 450ºC, respectively. Glass fiber filtered water

218
219 Total lipid extraction was performed on subsamples between 0.8 g and 2.6 g from the 220 lyophilized seston material depending on the total inorganic particle content. Lipids 221 from the dried material were extracted following the method of Folch [25] using three 222 subsequent extractions of a 2:1 chloroform-methanol mixture. Extractions were 223 combined into a single volume and phase separation achieved after adding a 20% 224 volume of a 1 M KCl solution. The upper phase was discarded and the lower phase was 225 evaporated under a stream of nitrogen. The dried total lipid extract was gravimetrically 226 determined and stored in 2 ml Teflon sealed glass vials at a concentration of 10 mg ml −1 227 in a 2:1 chloroform/ methanol mixture containing 0.01% hydroxy-butyl-toluene (BHT).
228 For fatty acid analysis, lipid extracts were trans methylated in an acid catalysed reaction 229 for 16 h at 50°C using 1 mL of toluene and 2 ml of 1% sulphuric acid (v/v) in methanol  250 representing fatty acid biomarkers (S1 Table) was used as input data in an adaptation of   The seston fatty acid matrix for all GRE samples (S1 Table) was also used to infer  Table). To calculate the proportional contribution 287 Following previously published mixing models, the analysis implemented a normal 288 likelihood function. MCMC chains were run as "very long" for 10 6 iterations with 289 5*10 5 iteration burn-in and a thinning rate of 5*10 2 .   Table). The nearly seven percent lipid  Table).  Table 1 shows the relative cell abundance 341 after compiling by season and salinity range those genera belonging to the same class.  (Table 1). During all seasons, there was a clear 352 decrease in Cyanophyceae cell abundance as salinity increased. A similar, although less 353 marked, tendency with regard to salinity was observed for the Chlorophyceae. The 354 presence of this class was, however, more constant among seasons in comparison to the 355 Cyanophyceae. Dinophyceae and Coccolithophyceae species were more frequently 356 observed in winter but their pattern of variation relative to salinity was uncertain ( Table   357 1). All other classes included in Table 1 were intermittently detected and represented 358 less than 5% of total cell abundance.   368 only during winter and summer (Fig 1). When phytoplankton diversity was measured 369 utilizing the higher taxonomic rank functional groups that resulted after CHEMTAX 370 analysis, the Shannon-Weaver index decreased (P<0.05) only at the highest salinity 371 range during spring and winter (Fig 1). 383 group contributing rather constantly around 50% to total phytoplankton community 384 spatial and temporally, with only a slight, although significant (P<0.05), decrease in 385 spring (Fig 2). The microscopically observed inverse relationship between 386 Cyanophyceae and salinity coincided with results from CHEMTAX analysis (Fig 2).
387 Total Cyanophyceae contribution to phytoplankton community estimated from FAs was 388 lower than that microscopically determined. A similar conclusion can be reached for the 389 Chlorophyceae+Trebouxiophyceae group, although with a lower difference between 390 both methods. FAs indicated increased Dinophyceae contribution to phytoplankton 391 community at higher salinity during spring, summer and autumn. A similar positive 392 relation with salinity was found for Coccolithophyceae+Pelagophyceae abundance (Fig   393 2). The presence of Eustigmatophyceae representatives was caught by CHEMTAX 394 analysis whereas species from this class could not be detected under the microscope.
395 The Eustigmatophyceae contributed around 10% to phytoplankton community in 396 winter, spring and summer, and around 5% in autumn, but no evident relationship with 397 salinity could be concluded (Fig 2). Despite significant seasonal effect in the   408 Among the minority taxonomic groups, occurrence of the Raphidophyceae was only 409 perceived after CHEMTAX analysis. It was shown to be the most marked seasonal 410 group (detected only in spring and summer) and it was present only in the higher 411 salinity waters of the estuary (Fig 3). The Euglenophyceae showed an inverse trend, 412 being noticeably more abundant in the autumn-winter period and in lower salinity 413 waters. The Pavlovophyceae was also more abundant during autumn-winter but it 414 favourably proliferated in higher salinity waters (Fig 3). Other interesting seasonal 415 change was the near absence of Pyramimonadophyceae, Mamiellophyceae and 416 Cryptophyceae representatives during winter. The Chlorodendrophyceae was the most 417 seasonally constant group (Fig 3). The classes Chrysophyceae, Xanthophyceae and 418 Synurophyceae were not detected by any of the matrix factor analysis, coinciding with 419 their practical absence in microscopic observations (Table 1).   (Table 2). Salinity explained more percent of total variation than season but the 459 highest source of variation (36.62%) was residual ( Table 2). The CAP performed on the 460 full data set of samples revealed a noticeably more structured pattern across axis 1 461 (δ 1 2 =0.705, δ 2 2 =0.032) when constraining was done on salinity (Fig 4). In this instance, 462 samples from lower to higher salinity clearly scored from left to right on axis 1 with

513
514 In a preliminary DistLM analysis, nutrients were included among all environmental 515 predictor variables and were found to explain low percent variation of phytoplankton 516 composition, namely 1.4% (phosphate), 1.8% (ammonia), 4.4% (silicate) and 5.6% 517 (nitrate). Low nutrient contribution to total variation and their very high levels detected 518 throughout the estuary salinity gradient (Fig 6), discouraged using nutrients as relevant 519 explanatory variables and were thus removed from the analysis. Despite nitrate, 520 phosphate and silicate concentrations being significantly lower in higher salinity waters, 521 their respective minimum values were still too high to be considered as potentially 522 limiting factors of phytoplankton growth [35]. Ammonia showed an inverse trend 523 respecting to other nutrients, increasing with salinity in the estuary, and its relative 524 contribution to the inorganic nitrogen pool was nearly an order of magnitude lower than 525 that of nitrate (Fig 6). Nitrite concentrations were on average half the value of ammonia 526 and no clear spatial or temporal pattern could be found. Nutrient concentrations were 527 unaffected by dam discharge (Fig 7). When seasonality was decomposed into the five 531 rainfall variability occurring in the zone, but having minor influence during the dryer 532 summer and winter periods (Fig 8). The contribution of dam discharge to explain 533 phytoplankton community variability was very constant at around 10-11% from spring 534 to autumn and reached a minimum 2% in winter. The effect of temperature was more 535 marked in summer-autumn respecting to the minimal influence exerted in winter-spring.
536 The relevance of irradiance was maximal in summer and reached the minimum value in 537 autumn (Fig 8).