Aquatic food webs in restored marshes: a stable-isotope approach in the Gironde estuary (SW France)
Author contributions: LC conceived and designed the research, produced and analyzed the data, wrote original manuscript; MV, JL, BL, NS produced and analyzed the data, edited the manuscript; MK, SB, ML produced the data, edited the manuscript; HB and HC edited the manuscript.
Abstract
Intertidal marshes are important habitats for nekton. However, historical draining and dyking hampered European coastal wetlands. Marsh restoration is therefore critical not only to improve their capacity to protect coastal lines but also to rehabilitate their ecological functionalities. The benefits of intertidal marsh restoration for nekton community composition and feeding ecology are examined in a case study within the largest macrotidal estuary in Western Europe (Gironde). The structure and functioning of nekton food webs are addressed using stable isotopes of carbon (δ13C) and nitrogen (δ15N). δ13C values of suspended particulate matter illustrated the influence of the tidal connectivity with the adjacent estuary in restored habitats. δ15N values of nekton and its resources evidenced little difference in food-web complexity, but spatial or seasonal variations for some resources and consumers, related to a combination of temporal and tidal effects. The European eel (Anguilla anguilla) and the introduced freshwater topmouth gudgeon (Pseudorasbora parva) dominated food webs in the fully and partially connected marsh habitats, respectively. The isotopic niche and diet composition of A. anguilla also varied between habitats, as did the diet of other nekton species (Chelon ramada and Palaemon longirostris). This study informs the rehabilitation process of important functionalities of restored aquatic habitats for nekton.
Implications for Practice
- Stable isotope data allowed the analysis of the structure and functioning of aquatic food webs in restored marshes subjected to different management modalities.
- Trophic interactions between locally available resources and nekton species varied spatially and temporally, highlighting seasonal and tidal influences on the properties of the restored habitats for nekton feeding.
- This study improves our understanding of the processes shaping the rehabilitation of an essential ecological functionality for fish and macrocrustaceans in one of the most threatened ecosystems worldwide.
Introduction
Among estuarine habitats, intertidal marshes are of particular importance for nekton (i.e. fish and macrocrustaceans), which uses them as refuges from predation and as feeding grounds at various life stages (Whitfield 2017). However, intertidal marshes have been severely degraded worlwide, affecting coastal fish and macrocrustacean populations and associated fisheries (Gedan et al. 2009; Baker et al. 2020). Hydromorphological recalibration is one of the major causes of the decline of intertidal marshes in Europe, through drainage and dyking for the purpose of agriculture, urbanization, or harbor development (Goeldner-Gianella 2007). This resulted in the isolation of marshes from the adjacent coastal or estuarine environment. Drained and diked lands—named polders—were covering about 15,000 km2 of coastal wetlands in Europe in the twentieth century, including 1400 km2 in France (Goeldner-Gianella 2007). Efforts toward “depolderization” (sensu Goeldner-Gianella 2007), or “de-embankment” (sensu Wolters et al. 2005), have emerged since the ratification in 1971 of the Ramsar Convention on Wetlands of International Importance. The tidal restoration of coastal marshes can rely on diverse modalities of marsh reconnection to the adjacent marine or estuarine environments (see Debue et al. 2022 for a review) and has recently been developing in some European countries (especially in England, Belgium, the Netherlands, or France; Wolters et al. 2005).
The tidal restoration of coastal and estuarine marshes consists in reestablishing the influence of tides in previously drained areas. This is expected to lead to the restoration of marsh ecological functionalities and resulting services, including climate regulation, food provision, carbon retention, protection from risks of submersion, and support for native biodiversity (Gilby et al. 2020; Lepage et al. 2022). The effects of marsh tidal restoration (and more generally of “coastal realignment”) on biodiversity were reviewed by Debue et al. (2022). Globally, nekton abundance tends to increase quickly after the tidal restoration although responses in terms of species richness and community composition are variable depending on specific contexts, in particular the degree of connectivity with the adjacent estuarine or marine environment before the restoration (Debue et al. 2022).
The trophic functionality of intertidal marshes for nekton relies on the provision of specific habitat conditions supporting favorable feeding conditions (Cattrijsse & Hampel 2006). The inclusion of fish gut content observations and assessments of prey availability and their consumption in restored marshes is thus recommended (Neckles et al. 2002). This study aims at characterizing the structure of nekton food webs in fully (open) and partially (sluiced) tidally connected habitats of the restored marshes of Ile Nouvelle, which is located in the oligohaline area of the largest macrotidal estuary in Europe, the Gironde (Lobry et al. 2003). The stable isotope compositions of fish, macrocrustaceans, and their potential resources in the marshes were analyzed during two campaigns characterized by variable seasonal and hydrological contexts. The hypothesis addressed is that nekton assemblages and resources available to them in the marshes would vary in composition depending on the tidal connectivity allowed with the adjacent estuary. As such, we expected more complex aquatic food webs in the fully connected marsh, resulting from a higher diversity of resources and the presence of estuarine specialized consumers, whereas food webs in the sluiced marsh would resemble simplified freshwater systems used by more generalist consumers.
Methods
Study Area
«Ile Nouvelle» extends over about 300 ha and has been used for agricultural activities until the 1950s (Fig. 1A & 1B). A restoration program was implemented in 1991 to create a marsh in the previously drained lands by building channels, ditches, and ponds. Restoration was designed differently in the Northern and Southern parts of the island. In the south, the marsh is inundated with rainwater and punctual inflows of estuarine waters managed with sluice gates. These sluices limited tidal intrusion (except for punctual channel curation events) until 2013 when they were redesigned in compliance with the European eel French National Action Plan. Since then, tides enter the marsh channel for water heights above 3 m, which happens for about 60% of annual tidal cycles over a year in the Gironde estuary (SHOM REFMAR data base: https://data.shom.fr/donnees/refmar/—Pauillac-Trompeloup tide gauge). In the meantime, the dikes confining the northern part were left unmaintained, and a breach appeared in the outer dike during the Xynthia storm in 2010. Since then, estuarine waters flood the northern part of the marshes at each rising tide through the breach and a regression channel. Almost no secondary channels are connecting to the main channel except for previous ditches that were pre-existing before the breach. A secondary breach was created in 2016 in an inner dike, forming today four management units characterized by contrasted hydrological connectivity regimes (Fig. 1C): “FTC1”, in the northern part, is fully tidally connected; “FTC2,” in the north-central part, is also fully although indirectly tidally connected as it receives estuarine water from FTC1 through the internal breach located on the central inner dike at every tidal cycle; “PTC1” and “PTC2” are sluiced and more confined marshes, in which the connectivity with the estuary is restricted to periods of medium to spring tides when water heights reach 3 m or more in the oligohaline area of the estuary. In FTC1 and FTC2, estuarine waters thus fill the channel and submerge the mudflats and parts of the vegetated low shore at each rising and high tide, the immersed areas varying depending on water heights throughout the year. The main channel is always accessible for nekton except in about a third of its inner section, which dries up when water heights decrease below 1.5 m. In the sluiced areas PTC1 and PTC2, water levels are relatively stable all year round, and tidal exchanges are only possible through sluice gates specifically designed to allow European eel crossing for water heights above 3 m.
Sampling Strategy
Nekton and its potential food sources were sampled in the marshes during two campaigns characterized by variable seasonal and tidal contexts (2019 and 2021; Table 1). For each campaign, the sampling periods of nekton and its potential food sources were separated by several weeks to months to consider the incorporation times of food source isotopic compositions within fish tissues (Guelinckx et al. 2007; Buchheister & Latour 2010). Nekton was sampled first in September 2019, during a neap tide when water heights varied between 0.93 and 5.18 m at lowest and highest tides (average tidal range of 3.78 m), and second in April 2021, during a spring tide when water heights varied between 0.33 and 6.04 m during lowest and highest tides (average tidal range of 5.55 m). Water temperature (°C) and salinity (psu) were measured at each station during fish retrieval with a WTW3630 manual probe.
| Trophic group | Stations | Campaigns | |
|---|---|---|---|
| 2019 | 2021 | ||
| Suspended particulate matter | Downstream estuary (Royan) | 6/28 | 3/22 |
| Dordogne (Vayres) + Garonne (Portets) = watershed | 7/03 | 7/27 | |
| Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 6/06 + 6/17–18 + 9/10 | 3/09 | |
| Plant detritus | Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 7/11–12 | 3/09 |
| Primary producers (microphytobenthos, riparian vegetation, and macroalgae) | Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 7/11–12 | 3/09 |
| Benthic macro- and meiofauna | Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 7/11–12 | 3/29–31 |
| Zooplankton | Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 7/17–18 | 3/29–31 |
| Nekton (fish and macrocrustaceans) | Ile Nouvelle (FTC1, FTC2, PTC1, and PTC2) | 9/9–12 | 4/27–30 |
Collection of Samples
Suspended particulate matter (SPM) was collected in Ile Nouvelle and in the Dordogne and Garonne Rivers (at Vayres and Portets, respectively, see Selleslagh et al. 2012 for stations location), later grouped as “watershed” stations (WS), and in the downstream area of the Gironde Estuary (at Royan, later named “downstream estuary” [DE]). Three replicate water samples were filtered from one to four times per sampling campaign (Table 1). Water was collected with a Niskin bottle (bottom water at rising tide in Royan) or with a bucket (surface water in marshes and at ebb tide in rivers). Water samples were filtered on precombusted GF/F Whatman filters. The filtered volumes ranged from 35 to 390 mL depending on SPM load (SPM concentrations up to 10 g/L in the Gironde estuary; Sottolichio & Castaing 1999).
Dominant species of the riparian vegetation were sampled in the marshes by cutting out a few leaves at the bottom of three shoots from the same taxa growing at the aquatic-terrestrial interface (see Table S1 for species list and Table 1 for sampling dates). The rare macroalgae growing on rocks or at the surface of sediments were handpicked. Microphytobenthos (MPB) was sampled by collecting superficial sediment in areas emerged at low tide. Sediment samples were collected in triplicates with a spoon by scraping about 0.5 cm deep and within a 30 cm radius area.
Benthic macro- and meiofauna were collected from three replicates of 5 L sediment samples collected with a small shovel. The mud was sieved through a 200 μm mesh sieve. All the organisms and plant detritus retained by the sieve were collected with tweezers and placed in glass containers pre-filled with dechlorinated tap water (Table S1 for species list; Table 1 for sampling dates).
Zooplankton was sampled with a 60 cm diameter-opening plankton net (60 μm mesh) held facing the current for 10 minutes. Depending on water depth, the net was either held from a boat (FTC1), from a small jetty (PTC1), from a float tube (PTC2), or manually (FTC2). Moreover, zooplankton sampling was performed at PTC1 and PTC2 stations in 2021 by directly sieving the water on a 200 μm mesh, as organisms were then highly concentrated in restricted shallow lentic systems. All the samples were placed in jars pre-filled with dechlorinated tap water.
Fish and macrocrustaceans were collected with double fyke nets, set up for a half or a full tide cycle (6–12 hours). Fyke nets were positioned longitudinally in the channel in the fully connected areas or across deep ponds in the sluiced areas, so as to ensure the organisms remained immersed even during low tides. Fish, shrimps, and crabs were quickly retrieved alive from the fyke nets, separated per species, and stored in tanks containing field water. A maximum of 19 individuals per fish species, station, and sampling time was selected for tissue sampling (supernumerary individuals were released). Selected fish were measured to the nearest cm (standard length for Perciformes and Clupeiformes; total length for Anguilliformes) and euthanized by cerebral percussion (Table S1 for species list and size ranges).
Laboratory Work
Plant leaves, macroalgae, and plant detritus were rinsed in successive baths of ultrapure water and 10% hydrochloric acid (HCl) to remove any potential epibionts or exogenous organic matter as well as inorganic carbon. MPB was extracted from the sediment following the method of Bolch (1997) modified by Liénart et al. (2017) and adapted to estuarine water. Briefly, a density gradient was created in centrifuge tubes using the combination of deionized water (density of 1) and a solution of sodium polytungstate (density of 2). After centrifugation, the MPB, isolated at the interface between deionized water and the solution of sodium polytungstate was recovered and filtered on a precombusted (4 hours at 450°C) GF/F Whatman filter.
Glass containers containing meiofauna, macrofauna, and zooplankton in dechlorinated tap water were stored in a fridge for 24 to 72 hours to allow for the evacuation of gut contents. Macrofauna and meiofauna were then sorted out by coarse taxonomic group (subphylum, class, sub-class or family level). Zooplanktonic organisms were sorted out by size class in 2019 (200–500, 500–800, 800–1400, and ≥1400 μm), the taxonomical content of each size class being roughly described (sub-order, class or sub-class level). As preliminary statistical analyses highlighted that organisms of different size classes had similar isotopic compositions, samples from 2021 were only sorted out by taxonomic levels for simplicity.
Fish, shrimps, and crabs were rapidly dissected to collect dorsal muscle for fish, abdominal flesh for shrimps, and muscle from the major claw for crabs. All the samples, including filters, were stored at −20°C until further processing. All the samples collected on filters (MPB, SPM) were dried in an oven at 50°C for 8 to 12 hours. Carbonates were then removed using HCl fumigation under a vacuum-enclosed chamber for 24 to 48 hours. The surface of the filters was scraped with a scalpel to recover a powder and reduce the amount of glass fibers in analyzed samples. All other samples were freeze-dried, ground into a fine powder using a mortar and pestle. For the zooplankton, invertebrates with exoskeletons, and small fish analyzed whole (gobies Pomatoschistus sp., mosquitofish Gambusia holbrooki), powdered samples were acidified (10% HCl followed by rinsing with ultrapure water) to remove carbonates. Lipid extraction was not performed (it was later shown that proportion of carbon in the samples varied little among species with differences between end members, here SPM and Anguilla anguilla, remaining largely below 10‰ for all conditions; Post et al. 2007). The powdered samples were transferred into tin (raw samples) or silver (acidified samples) capsules.
Determination of Isotopic Compositions
Data Analysis
δ13C and δ15N values of resources and consumers were compared between stations at each campaign using non-parametric Kruskal–Wallis tests. When a significant difference between stations was observed (p < 0.05), Dunn tests were applied to compare sites by pairs.
The food chain length was computed for each station and campaign following Post et al. (2000) and Post (2002). This calculation was based on the full range of δ15N values (max – min) divided by the value of the trophic fractionation factor between each trophic level (Δδ15N), the value of which (2.68 ± 1.13‰) was retrieved from previously published estimations in our study area (Ballutaud et al. 2019).
Anguilla anguilla was the most abundant nekton species observed in at least one fully connected and one sluiced station during both campaigns. The variation in its isotopic niche among conditions was thus addressed using indices of similarity (ISim) and nestedness (INes) applied from Cucherousset and Villéger (2015).
Bayesian mixing models “MixSIAR” developed by Stock et al. (2018) were applied to isotope data of European eel (A. anguilla), thinlip mullet (Chelon ramada) and Delta prawn (Palaemon longirostris) to compare the relative contributions of potential food sources to their diet between conditions. These three species are known to use estuarine habitats for relatively long periods (juvenile stage for the Thinlip mullet; the yellow-stage that can last 5–15 years for the European eel; whole life cycle for the Delta prawn). More «transient» species, that is, nekton species that move in and out of the estuary on a shorter or seasonally more restricted time frame, such as anchovies (Engraulis encrasicolus), were not considered further in the stable isotope analysis. The trophic fractionation factors (TFF) used in the models were 1.43 ± 1.66‰ for δ13C and 2.68 ± 1.13‰ for δ15N (Ballutaud et al. 2019). Models were run using three Monte-Carlo Markov chains in parallel, with 100,000 iterations as burnin, and then 50,000 for the determination of posterior distributions of resources contributions. The convergence of the chains was checked using Gelman and Rubin (1992)'s test, with a value ≤1.05 considered reliable.
The resulting detailed list of sources included in each model is provided in Table S3. Outputs from the mixing models were analyzed through the distributions (mean, median, 50 and 95% confidence intervals) of predicted contributions of every selected potential food source to one consumer's diet for each sampling condition (station and campaign). All data are available in an online repository (https://doi.org/10.57745/XGJDWH) with access upon request.
Results
Sampling Conditions
The water temperatures during nekton sampling were 23.5, 24.0, 20.3, and 24.0°C at PTC1, PTC2, FTC1, and FTC2, respectively, in 2019 (neap tide) and were 16.9, 15.4, 12.8, and 12.0°C at similar respective stations in 2021 (spring tide). Salinities were typically fresh (0.8–0.9) in the sluiced marshes PTC1 and PTC2, and brackish (1.7–2.1) in the fully connected marshes FTC1 and FTC2 during the first campaign and ranged between 2.4 and 2.7 at the four stations during the second campaign.
Composition and Environmental Affinities of Assemblages
Plants were dominated by reeds (Phragmites australis and Phalaris arundinacea) in all marsh areas, with sedges (Carex sp.) and yellow iris (Iris pseudacorus) also observed. Sea clubrush (Bolboschoenus maritimus) only occurred in the fully connected stations (Table S1). Benthic fauna composition also varied between stations, with Chironomidae, Odonata, and Oligochaetes dominating in the sluiced marshes, and Gammarids, Polychaetes, and Gastropods dominating in the fully tidally connected habitats (Table S1). The taxonomic composition of zooplankton samples varied markedly between the fully connected and the sluiced marshes during the second campaign, with communities dominated by the freshwater cladoceran Daphnia spp. and aquatic insect larvae in sluiced stations, and estuarine copepods dominating the communities in the fully connected stations (Table S1).
The diversity of nekton species also differed between the fully connected and the sluiced marshes, with 11 species collected at PTC1 and PTC2, and 6 in FTC1 and FTC2 (Table S1). However, 7 out of 11 species from sluiced areas (PTC1 and 2) were introduced species displaying freshwater affinities (Ictalurus melas, Carassius carassius, Ctenopharyngodon idella, Cyprinus carpio, Gambusia holbrooki, Lepomis gibbosus, and Pseudorasbora parva), while only one nonindigenous species (P. parva) was observed in the fully connected stations (FTC1 or 2). Marine-associated species such as the thinlip mullet and spotted sea bass (Dicentrarchus punctatus) were only observed in the fully connected stations, but the anchovy, the goby (Pomatoschistus sp.), and the European eel (Anguilla anguilla) were sampled in both areas (Table S1).
Isotopic Compositions of Resources and Consumers: Spatial Variations
During the 2019 campaign, the δ13C values of SPM were significantly higher in DE than in PTC2 (Fig. 2A). Similarly, SPM δ15N was lower in DE than in PTC2 (Fig. 2C). In 2021, the SPM isotopic composition was similar at all sites (Fig. 2B & 2D). Zooplankton δ13C and δ15N values displayed significant spatial variations during both campaigns, although variations between sites were not consistent among periods (Fig. 2E–H). Zooplankton δ15N values were, however, lower at PTC2 in 2021 (Fig. 2H).
Regarding other resources, MPB and plant detritus isotopic compositions were similar at all sites and campaigns (Table S2). Macrophytes δ13C values were lower at FTC1 (average −30.5‰) compared to PTC1 and PTC2 (averages of −27.6 and −27.5‰, respectively; Table S3) in 2019. Macrophytes δ15N values were only different between FTC1 and FTC2 in 2019, and no significant variation between sites was detected for any isotope in 2021 (Table S2). Benthic invertebrates also displayed significant spatial variations for both isotopes in 2019 (Table S2), although patterns of variations were different depending on the isotopic ratio considered (Table S2). The isotopic compositions of benthic invertebrates were similar at all sites in 2021 (Table S2).
Regarding nekton secondary and tertiary consumers, European eel δ13C values were highest in FTC1 during both campaigns (Fig. 2I & 2J). Its δ15N values were similar at all sites in 2019 (Fig. 2K), and lower in PTC2 as compared to FTC1 in 2021 (Fig. 2L). The thinlip mullet also had spatially variable δ13C values in 2019, suggesting marine enrichment at FTC1 although sample numbers are small (Table S2). Its δ15N values also varied spatially in 2021, although only PTC1 and PTC2 differed (Table S2). Both δ13C and δ15N values of the Delta prawn varied between stations in 2019 (Table S2), with differences between FTC1 and FTC2 only for carbon, and between FTC2 and PTC1 for nitrogen (Table S2). The Delta prawn δ15N values were similar at all sites in 2021. Significant variations between stations were also detected for δ13C values of the common ditch shrimp (Palaemonetes varians) in 2021, with significant differences between PTC1 and PTC2 (Table S2). The isotopic composition of the topmouth gudgeon (P. parva) varied between sites in 2019, with distinct δ13C values in PTC2 relative to FTC2 and PTC1 (Table S2). No other significant spatial variation was detected for the remaining fish species with sufficient observation numbers for the analysis to be carried out (i.e. common carp C. carpio and gobies Pomatoschistus sp.; Table S2).
Food-Webs Structure and Complexity
Overall, the δ13C values characterizing sampled aquatic food webs at Ile Nouvelle ranged from −33.9 to −14.8‰, and the δ15N values ranged from −0.5 to 14.5‰ (Fig. 3). Benthic invertebrates and zooplankton had very variable isotopic compositions at each station and campaign (large error bars; Fig. 3), related to their large taxonomical diversity (Table S1) potentially participating in the aforementioned variable spatial patterns of their mean isotopic values.
The higher species richness of secondary and tertiary consumers of nekton at sluiced stations PTC1 and PTC2 than at fully connected stations FTC1 and FTC2 (Table S1) resulted in a higher density of data and groups and an apparent higher complexity of associated food webs (especially at DK1 in 2021; Fig. 3F). However, no major shift was highlighted in terms of the food web overall structure (Fig. 3), notwithstanding previously addressed spatial variations in the isotopic compositions of specific sources (i.e. SPM and macrophytes), primary consumers (i.e. zooplankton, benthic invertebrates) or higher-level nekton consumers (i.e. P. varians, Palaemon longirostris, A. anguilla, Chelon ramada, P. parva).
Anguilla anguilla, Pomatoschistus sp., Engraulis encrasicolus, and P. longirostris had the highest δ15N values in the fully connected areas (FTC1 and 2; Fig. 3A–D), while P. parva, L. gibbosus, or G. holbrooki had the highest δ15N values in the sluiced areas (PTC 1 and 2; Fig. 3E–H). The range of δ13C values of European eel indicated a freshwater to marine enrichment from sluiced PTC2 toward fully connected FTC1 marshes (Fig. 3B, 3F, & 3H). Based on overall ranges of δ15N values characterizing whole food webs, the food chain length varied little among stations and campaigns, from a minimum of 3.9 at FTC2 in 2021 to a maximum of 5.4 at PTC1 in 2019 (Table 2).
| Station | 2019 | 2021 |
|---|---|---|
| FTC1 | 5.2 | 4.6 |
| FTC2 | 4.0 | 3.9 |
| PTC1 | 4.3 | 5.4 |
| PTC2 | 4.3 | 5.3 |
Diet Variations in Key Nekton Species
The diet of the European eel at TC1 was dominated by macrocrustaceans and zooplankton, with a lower but notable contribution of fishes and benthic invertebrates (Fig. 4A & 4E). At PTC1, eels mainly consumed fish in 2019 and invertebrates in 2021 (Fig. 4B & 4F). At PTC2, zooplankton dominated the eel's diet during the first campaign, while invertebrates were mostly consumed during the second one (Fig. 4C & 4G). The size ranges of eels also differed between stations, with smaller individuals generally more frequently sampled in the sluiced marshes, although large specimens also occurred (Table S1). The mean sizes and ranges were 37.9 (26.7–53.4) cm, 28.8 (24.4–33.4) cm, and 26.0 (9.8–49.2) at FTC1, PTC1, and PTC2, respectively, during the first campaign; 38.1 (29.5–45.5) cm, 28.4 (13.2–61.3) cm, and 19.6 (11.6–27.8) cm at FTC1, PTC1, and PTC2, respectively, during the second.
Benthic invertebrates and zooplankton dominated the diet of the Delta prawn at FTC1, FTC2, and PTC1 stations, with macrophytes and SPM also being consumed at PTC1 and FTC2 (Fig. 4H–J). Thinlip mullet, the diet of which could only be assessed in the fully tidally connected areas (FTC1 and FTC2) in 2019, appeared to feed upon MPB at both stations, followed by benthic invertebrates at FTC1 and zooplankton at FTC2 (Fig. 4K & 4L). Mullets from both stations had similar sizes, with 15.1 cm on average (8.0–41.9) at FTC1 and 15.7 cm on average (5.9–31.9) at FTC2 (Table S1).
Isotopic Niche of Anguilla anguilla
The isotopic niche of European eels in FTC1 was completely different from the one observed at PTC1 and PTC2, with a total niche distinction for PTC1 (ISim and INes = 0) at both campaigns and for PTC2 in 2021 (Table 3). During the 2019 campaign, the isotopic niches of eels were largely overlapping between PTC1 and PTC2 (Table 3). Conversely, the isotopic niches of eels hardly overlapped between campaigns at PTC1 (ISim = 0.01; Table 3), while they remained similar among campaigns at FTC1 (INes = 0.88; Table 3).
| Compared stations | 2019 | 2021 | Station | 2019 versus 2021 |
|---|---|---|---|---|
| FTC1 versus PTC1 | ISim = 0 INes = 0 |
ISim = 0 INes = 0 |
FTC1 | ISim = 0.198 INes = 0.878 |
| FTC1 versus PTC2 | ISim = 0.075 INes = 0.152 |
ISim = 0 INes = 0 |
PTC1 | ISim = 0.01 INes = 0.118 |
| PTC1 versus PTC2 | ISim = 0.036 INes = 0.97 |
ISim = 0.058 INes = 0.279 |
PTC2 | ISim = 0.175 INes = 0.402 |
Discussion
Results from this study converge with previous findings regarding the responses of nekton community composition to tidal restoration in marsh habitats of the Gironde estuary (Lechêne et al. 2018a, 2018b), while highlighting the trophic interactions at play and providing quantitative comparisons of the niche sizes and dietary responses of specific consumers. Findings reflect the spatial and temporal variations in resource availability and access conditions for the mobile consumers in the marshes. Although it is not possible to disentangle the relative effects of time (season or year) and tides on all trophic interactions due to a limited sampling effort, our results overall suggest the tidal connectivity influences assemblage composition and associated trophic functionalities for nekton in restored marshes of Ile Nouvelle.
As regards to the composition of assemblages, zooplankton and nekton taxa observed during the two campaigns evidenced a freshwater affinity in the sluiced habitats and an estuarine to marine affinity in the fully tidally connected habitats. It was already shown that nekton assemblages in these tidally connected habitats rapidly converged toward those of reference natural intertidal habitats in the Gironde estuary, and this only a few years after the tidal restoration occurred (Lechêne et al. 2018a, 2018b). Our study suggests the tidal reconnection of the northern marsh also improves the suitability of the restored habitat for estuarine and marine nekton species using intertidal marshes as feeding grounds as either adults or juveniles. These results agree with our hypothesis that the fully tidally connected restored marshes support more typical estuarine food webs, although those food webs were not found to be characterized by a higher complexity.
Meanwhile, the confinement of waters in the south sluiced marshes, especially PTC2, was highlighted by SPM isotopic composition, which was more 13C-depleted there than at any other station in 2019, when tidal ranges were lower (neap tide). On the contrary, the SPM δ13C values were higher and similar in other restored areas, the DE and its watershed. This particularity of SPM δ13C values in PTC2 was not observed in 2021 (spring tide conditions allowed exchanges with the surrounding estuary in all marsh habitats). This result may reflect a contrasted origin of basal resources sustaining aquatic food webs in the different management units of Ile Nouvelle during specific tidal conditions. When tides are low and water heights are limited to 3 m or less, the aquatic environment in diked marshes is mostly influenced by the impluvium and consists of confined freshwater shallow channels and pools. Conversely, pelagic estuarine basal resources dominate the aquatic ecosystem in the tidally connected area. In the Gironde Estuary, the δ13C value of SPM is typically −25.2‰ and reflects the large dominance of refractory terrestrial organic matter, more negative values reflecting the input of litter (δ13C = −28.7‰) and/or freshwater phytoplankton (δ13C = −34.5‰) (Savoye et al. 2012). The depleted δ13C of SPM in restored marshes thus indicates an important contribution of litter and/or freshwater phytoplankton, especially during periods of low tidal connectivity. SPM in restored marshes, however, is enriched with marine-influenced estuarine waters in fully tidally connected areas and in the sluiced areas during higher tides, when water heights reach 3 m or more (SPM δ13C values are then similar in all marsh areas, the estuary and its watershed). Furthermore, SPM in the Gironde estuary was shown to consist of 98% of terrigeneous material from the estuary turbidity maximum zone, close to which Ile Nouvelle is located, with only 2% of phytoplankton from freshwater, estuarine, or marine origin (Savoye et al. 2012). Resources fueling the restored marshes from Ile Nouvelle throughout the year are therefore primarily of terrigeneous origin.
As regards to food web complexity, the number of trophic levels varied between 3.9 and 5.4, which suggests a difference of less than one trophic level between stations and campaigns (considering a TFF of 2.7‰ between two trophic levels; Ballutaud et al. 2019). The food chain length was therefore comparable among conditions. These values fall within the ranges usually observed in aquatic environments, which are often characterized by less than six trophic levels (Christensen & Pauly 1993). The values observed herein are thus overall indicative of an intermediate level of complexity, with, for instance, a longer chain length than in tidal marshes of Delaware (2.7–2.8; Wainright et al. 2000) and similar to coastal lagoons of Camargue in the Mediterranean Sea (4.4; Persic et al. 2004). This food chain length can, however, be expected to evolve with time, as the ecological functionalities in restored habitats are known to depend on the maturation process of newly or restored marshes, which may take several decades (Hampel et al. 2003). Nevertheless, neither the observed food chain length observed herein nor the overall structure of the food-webs, as described by δ13C versus δ15N biplots, agrees with the hypothesis of more complex aquatic food-webs in the fully connected than in the sluiced restored marshes at this point. Which leaves to be tackled the question of the nature and diversity of resources used by consumers as well as the level of diet specialization of those consumers.
Regarding the resources available in the marshes of Ile Nouvelle, the δ13C and δ15N values of zooplankton varied between stations at both campaigns. This variability reflects seasonal patterns characterizing planktonic communities in drained marshes and the adjacent Gironde estuary (David et al. 2005, 2020). Zooplankton in the estuary is indeed dominated by copepods (especially Eurytemora affinis) which are more abundant at very low salinities (around 0) as their distribution downstream is limited by very high SPM concentrations (David et al. 2005). Conversely, zooplankton assemblages in freshwater drained marshes vary greatly seasonally in relation to environmental fluctuations and human control on sluice gates (David et al. 2020). From spring to summer, planktonic communities in freshwater drained marshes are dominated by a succession of large pelagic cells, small cells, and then taxa with alternative food strategies due to nitrogen limitations and phosphorous desorption from the sediment leading to eutrophication processes and to the dominance of r-strategists' species (David et al. 2020). In our study, the zooplankton communities in the fully connected habitats were dominated by estuarine copepods, while cladocerans dominated the sluiced marshes. In addition to previously discussed results on SPM, those observations on zooplankton confirm that current management of the tidal connection through sluice gates in the south marsh results in a freshwater drained-type dominated system. It also implies that improved water replenishment through a more flexible management of sluice gates during warm periods may be beneficial in terms of eutrophication risks and replenishment of planktonic communities in this habitat (David et al. 2020).
Isotopic data and mixing models also revealed differences in diet composition of nekton consumers across stations and campaigns, some of which may partly be influenced by tidal connectivity in the marshes. The diet of the thinlip mullet could only be compared between the two fully connected stations from the northern marsh (FTC1 and FTC2), as this species does not occur, or rarely so, in the sluiced marshes (Lechêne et al. 2018b; four isolated small and damaged individuals during the second campaign in this study). The thinlip mullet was shown to rely on MPB, benthic invertebrates, and zooplankton during the first campaign, which is in line with previous descriptions of its diet in estuarine and coastal environments (Lebreton et al. 2013; Carpentier et al. 2014). However, the same resources, which may plausibly contribute to its diet during the second campaign, were found inconsistent with the isotopic composition of locally available resources. This may indicate that the specimens analyzed for this particular campaign did not feed in the marsh before they were captured, for a period of at least a few weeks to months based on known incorporation times of δ13C and δ15N in fish tissues (Guelinckx et al. 2007; Buchheister & Latour 2010). These fish were more likely to be simply exploring the marsh at the time rather than residing there for feeding (as other species spending less time in marsh estuarine habitats throughout the year, such as anchovies). The specimens of mullets concerned (n = 5) were also particularly large (sizes comprised between 32.5 and 41.0 cm total length). The hydrodynamic conditions in the fully connected stations in 2021 (especially very high flow speeds in the channel during the ebb) may also have complicated their feeding on benthic resources such as MPB or invertebrates, and this may also partly explain the importance of zooplankton in their diet during this campaign.
The isotopic niche displayed by European eels in the fully tidally connected area clearly contrasted with that observed in the sluiced marshes, foreseeing a notable difference in terms of feeding habits, a difference also suggested by mixing models. Eels were shown to rely on macrocrustaceans, zooplankton, benthic invertebrates, and fishes at all stations, but proportions of macrocrustaceans were generally higher in the fully-reconnected area (FTC1) while freshwater benthic invertebrates (mostly Chironomidae) and freshwater fishes (mostly exotic species) contributed more in the sluiced marshes. The eels sampled also varied in size between stations, with smaller individuals in the southern stations (means of 37.9–38.1 at FTC1, 28.4–28.8 at PTC1 and 19.6–26.0 at PTC2). Therefore, variations in their diets can be attributed to both ontogenic and habitat-related conditions, together affecting the availability and accessibility of resources for eels. Importantly, the diversity of freshwater fishes in the sluiced marshes of Ile Nouvelle may confer advantageous feeding conditions for small individuals of this critically endangered species, in a part of the estuary where access to other favorable intertidal or shallow habitats is complicated due to the scarcity of available natural wetlands (Carassou & Boët 2020). However, eels are there competing with other carnivorous introduced fish species such as Lepomis gibbosus, Ictalurus melas, or Pseudorasbora parva, that are not found in the fully tidally reconnected marshes where eels dominate the food web with little or no competitors. These introduced species, which also include primary and secondary consumers (Carassius carassius, Cyprinus carpio, and Gambusia holbrooki) are known to occur in the oligohaline area of the Gironde estuary (Lobry et al. 2003). They could have colonized the marsh from the start of the restoration program, when ponds and ditches were created in the previously drained land, or during punctual sluice gate openings for curation works. After 2013, they could also use the adjusted sluice system as long as water heights reached 3 m or more (which happens for about 60% of annual tidal cycles). They could then persist in the sluiced marsh area due to its relatively stable environmental conditions, while they were filtered out in the fully tidally connected habitats where environmental conditions are more variable (salinity and hydrodynamic conditions in particular). This highlights the benefits of full tidal reconnection for the rehabilitation of the trophic functionalities of estuarine marshes for marine-euryhaline species, while sluiced marshes, even if they benefit the critically endangered European eel, also favor the establishment of exotic species.
While the trophic interactions highlighted herein appear plausible, it remains possible that the fish sampled in the fully tidally restored marshes may have consumed resources in the adjacent estuary before their sampling in the marsh. However, they may then have done so in subtidal habitats, as there is very limited access to other intertidal mudflats or low shores in the oligohaline section of the estuary. Although it is possible to distinguish benthic resources from subtidal and intertidal environments based on their δ13C values in the polyhaline area of the Gironde estuary (Selleslagh et al. 2015), we could not sample the adjacent oligohaline subtidal habitats in this study. Only future direct or indirect observations would thus confirm if the fishes exploring the restored marshes of Nouvelle, especially in fully tidally connected habitats, display a feeding activity in the marsh; present stable isotope data, however, suggest so and this assumption appears consistent with rather favorable abiotic and biotic conditions. The trophic interactions highlighted herein therefore illustrate the ongoing restoration process of important ecological functionalities of the habitat for nekton in estuaries, some of the most threatened ecosystems worldwide (Costanza et al. 1997).
Ackowledgements
This study was supported by the EU LIFE project Adapto coordinated by the Coastal Protection Agency (https://www.lifeadapto.eu/). It benefited from funding by New Aquitaine Region, The Adour-Garonne Water Agency, and INRAE. Fishing permits were obtained from Direction Interrégionale de la Mer Sud Atlantique, France. We thank S. Cardonnel, B. Augizeau, and G. Campet (Conseil Départmental de la Gironde) for their support with field sampling, and F. Ferchiche (PhD student, University of Bordeaux, CNRS, Bordeaux INP, EPOC, UMR 5805) for the preparation of suspended particulate matter samples for isotopic analysis. P. Camoin (INRAE Centre Nouvelle-Aquitaine Bordeaux) designed Figure 1. A.-M. Lassallette helped proofread the manuscript for English. This work was carried out with the support (equipment and staff) provided by the XPO scientific infrastructure (LIFE research infrastructure—Living in Freshwater and Estuaries, INRAE) of EABX research unit, and we would like to thank C. Gazeau-Nadin and R. Le Barh for their contributions. We also thank an anonymous reviewer for a very thorough and constructive assessment of the manuscript.
