Identifying patterns in the multitrophic community and food-web structure of a low-turbidity temperate estuarine bay

Food web dynamics outline the ecosystem processes that regulate community structure. Challenges in the approaches used to capture topological descriptions of food webs arise due to the difficulties in collecting extensive empirical data with temporal and spatial variations in community structure and predator–prey interactions. Here, we use a Kohonen self-organizing map algorithm (as a measure of community pattern) and stable isotope-mixing models (as a measure of trophic interaction) to identify food web patterns across a low-turbidity water channel of a temperate estuarine-coastal continuum. We find a spatial difference in the patterns of community compositions between the estuarine and deep-bay channels and a seasonal difference in the plankton pattern but less in the macrobenthos and nekton communities. Dietary mixing models of co-occurring dominant taxa reveal site-specific but unchanging food web topologies and the prominent role of phytoplankton in the trophic base of pelagic and prevalent-detrital benthic pathways. Our approach provides realistic frameworks for linking key nodes from producers to predators in trophic networks.


Supplementary Methods
Description of the study site. Gwangyang Bay is a semi-closed coastal embayment with an area of ca. 145 km 2 , which is located on the southern coast of Korea ( Supplementary Fig. 1). It is surrounded by highly populated areas and important industrial complexes and supports large commercial fisheries and mariculture activities. Broad intertidal muddy sand flats at the north end of the bay have been reclaimed for the construction of shipping facilities and industrial complexes, and the area has been dredged to support ship navigation since 1982. Because of this increasing development pressure, the bay requires management for economic and ecological sustainability. The Seomjin River has a catchment area of ca. 5 × 10 3 km 2 , consisting of agricultural and forested land, and flows directly into the north end of the bay 1 . The river discharge carries an annual mean of ca. 120 m 3 s −1 of freshwater into the bay but displays substantial seasonal variability, from 30-95 m 3 s −1 in winter (base flows) to 300-400 m 3 s −1 in the summer monsoon season. The bay has a semidiurnal tidal cycle, with maximum tidal ranges Longitudinal patterns of macrobenthic assemblages were previously observed 3-5 . The species number and density of polychaetes, the most dominant taxonomic group of the microbenthic community, decrease drastically from the polyhaline zone of the deep bay channel through the mesohaline zone to the oligohaline zone, with an apparent shift in the dominant taxa. The sources and dynamics of dissolved inorganic nutrients 6,7 and the distributions of phytoplankton 8-10 , zooplankton 8,11 and fish assemblage 12,13 have seasonal and longitudinal heterogeneities. Furthermore, ecologically diversified habitat types are developed along the main channel within the bay, e.g., a common reed Phragmites australis community colonized in the upper estuarine supralittoral wetland, an intertidal muddy sand flat in the northern part, and eelgrass Zostera marina beds along the low tide line on the fringe of the intertidal muddy sand flat. Resuspended benthic microalgae on the intertidal sandflat (an annual mean of 35.3 ± 33.5 mg chlorophyll a m -2 in the top 5 mm of the sediment) 14 is known to support a considerable part of the nutrition of adjacent subtidal macrobenthic consumers 15 .
By contrast, direct and indirect connectivity between the main channel and other habitats is still unresolved. The above description shows the highly heterogeneous conditions in the system but represents the initial stage of our knowledge of patterns of ecosystem dynamics.
Laboratory processing and analyses. After transportation to the laboratory, the water for SPM quantification was filtered onto pre-combusted and pre-weighed Whatman glass fibre filters (GF/F; 47 mm, 0.7 μm pore size). The filters were then dried for 48 h at 60°C in a drying oven and reweighed after recovering to room temperature in a desiccator. The total SPM concentration was calculated by weighing the filter before and after filtration of a known volume of water.
The water for the chlorophyll a measurement was also filtered onto a pre-combusted Whatman GF/F filter, extracted in 90% acetone for 24 h in the dark at −20°C, and its concentration was determined using a fluorometer (Turner Designs Model 10 AU 005, Sunnyvale, CA, USA) according to Holm-Hansen et al. 16 .
Water samples for inorganic nutrient analysis were thawed overnight at low temperature (2C) and brought to room temperature prior to analysis. Concentrations of phosphate (PO4), ammonium (NH4), nitrite (NO2), nitrate (NO3), and silicate (SiO2) were colorimetrically determined using a QuAAtro nutrient analyser (SEAL Analytical GmbH, Norderstedt, Germany). These measurements followed the procedures developed by Murphy and Riley 17 for PO4, Hansen and Grasshoff 18 for NO2, NO3, and SiO2, and Helder and de Vries 19 for NH4.
Phytoplankton species were identified and counted in 100200 μl of the 20 ml concentrate under 200 or 400 magnification using a Nikon's Eclipse E600 microscope (Nikon, Tokyo, Japan).
In the laboratory, the species of mesozooplankton were identified and counted under a Nikon's SMZ 645 stereoscopic microscope (Nikon, Tokyo, Japan) using a Bogorov counting chamber after a resuspension of the sample with a known volume of filtered seawater.
Macrobenthic invertebrates, once sorted in the laboratory, were identified to the species level using a Nikon's SMZ 645 stereoscopic microscope (Nikon, Tokyo, Japan) and then counted. When necessary, the biomass of organisms was estimated by weighing after drying in an oven for 48 h at 60°C.
Nekton samples were sorted in the laboratory, transferred to 70% ethanol, and identified to the lowest possible taxon. When necessary, a dissecting stereomicroscope was used to identify small individuals to the species level. Once counted, individual fish were measured (mm, standard length) and weighed (wet weight, g) to allow abundance, biomass, and frequency of occurrence analyses.
For defatting zooplankton and fish tissues before isotope analysis, while mathematical correction models were designed to adjust the effects of lipids on δ 13 C values based on the tissue C:N ratio 20,21 , our δ 13 C values of lipid-free tissues varied substantially from the mathematically corrected values 22,23 . To avoid this effect, zooplankton and fish tissue samples were defatted in a mixture (2:1:0.8) of methanol, chloroform and water before isotope analysis 24 . Because such a consistent difference in the δ 13 C values was undetectable for other invertebrates whose lipid contents were very low (< 5% of dry tissue weight) compared to those (range 4−34%; > 10% in most cases) in fish 25 , the defatting of invertebrate tissues was excluded in the present study.

Supplementary Tables
Supplementary Table 1 | Frequency and abundance description of individual clusters. Frequency (F, the number of samples) and logtransformed [ln(abundance + 1)] abundances (A, median and range) of taxa accounting for at least 5% of the total abundances of plankton, macrobenthos, and nekton assemblages in each cluster. MAWC, median abundances for whole clusters. Abundance units are Cells l −1 for phytoplankton; Individuals m −3 for zooplankton; Individuals m −2 for macrobenthos; and Individuals 0.03 km −2 for nekton.  of isotope values of deposit feeders in macrobenthos-community clusters. Differences in the  13 C and  15 N values between suspension and deposit feeders in the estuarine channel were significant (Mann-Whitney U test, U = 6.5, P < 0.001 for  13 C; Student's t test, t18 = −3.737, P = 0.002 for  15 N). Differences in the  13 C and  15 N values of benthic baselines (i.e., deposit feeders) between the estuarine channel and the deep bay were significant (Student's t test, t42 = 16.637, P < 0.001 for  13 C; t42 = −2.990, P = 0.005 for  15 N).

Estuarine-channel clusters
Deep-bay clusters

Supplementary Figures
Supplementary Fig. 1 | Visualization of spatial distribution patterns of the SOM units.