The studied maerl bed is located in the semi-enclosed Bay of Brest (Brittany, NW France), which has a total area of 180 km², a maximum tidal amplitude of 8 m and an average depth of 8 m (Fig. 1). Water-exchange with the shelf waters (Iroise Sea) occurs through a narrow (2 km wide) and deep (40 m) channel (Grall et al. 2006), and two rivers (the Aulne, with a catchment area of 1842 km², and the Elorn, with a catchment area of 402 km²) contribute freshwater inputs (Grall et al. 2006). Tidal action and regular wind-generated swell induce short-term variability in hydrological factors and enhance water mass mixing (Grall et al. 2006). Local hydrodynamics influences the sediment composition, which ranges from muds to coarse gravels, but there are also rocky substrata (Grall et al. 2006). Maerl beds cover 30% of the total surface area (Grall et al. 2006).

The sandy beach site is located in the Anse de Dinan, Bay of Douarnenez, France (Fig. 1) and is characterized by a median particle size ranging from 140 to 200 mm, a low silt content (0.2%) and a fraction of dead shells ranging from 2% to 9% (Baldrighi et al. 2019).

Three replicate samples were collected by SCUBA divers in April 2012 at the maerl bed site (48°17.627′N, 4°26.470′W, 7 m depth, water temperature of 14°C and salinity of 33.9; Fig. 1) using Plexiglas corers (6 cm inner diameter) buried down to 5 cm depth. The sediment cores were sliced into five layers: Maerl Layer at the top (ML, including maerl without sediment), 0-1, 1-2, 2-3 and 3-4 cm. They were preserved in buffered 4% formalin solution and stained with Rose Bengal (0.5 g l^-1). Samples collected below the maerl bed contained poorly sorted, very coarse sands with 3.5% of organic matter. At the sandy beach (48°14.109′N, 4°32.545′W, temperature of 14.4°C, salinity of 35.39; Fig 1), samples were collected in May 2012 at low tide in the intertidal, using Plexiglas corers (3.6 cm inner diameter). The samples consisted of three separates replicates of 15 cm depth cores and contained very well sorted fine sands with 1.5% of organic matter (see Carriço et al. 2013 and Baldrighi et al. 2019 for details). The authors are aware of the different sampling strategy adopted, i.e. a 5 cm sediment depth (maerl bed) vs. a 15 cm sediment depth (sandy beach). However, most meiofaunal organisms inhabit the top 5 cm of sediment (e.g. Ingels and Vanreusel 2013), allowing us to assume that comparisons between the two environments would not be substantially affected.

All samples were sieved through 1000 µm and 32 µm mesh sizes. Meiofaunal organisms were extracted by Ludox centrifugation following Danovaro (2010), counted and classified to the highest taxonomic level using a stereomicroscope. At least 100 nematodes from each replicate were picked out randomly and mounted on permanent slides after formalin-ethanol-glycerol treatment for identification to the genus level with an optical microscope according to Platt and Warwick (1988) and the recent literature available (NeMys database, Bezerra et al. 2021). Species richness (SR), expected number of genera (51) and diversity indices (Shannon, H’, Margalef D, Pielou evenness and J’) were calculated for both meiofauna and nematodes with the DIVERSE routine (PRIMER 6+; Clarke and Gorley 2006). The nematodes were divided into four trophic groups following Wieser (1953): (1A) selective (bacterial) feeders with no buccal cavity or a fine tubular one; (1B) non-selective deposit feeders with a large but unarmed buccal cavity; (2A) epistrate or epigrowth (diatom) feeders with a scraping tooth or teeth in the buccal cavity; and (2B) predators/omnivores with a buccal cavity with large jaws. The index of trophic diversity (ITD) was calculated considering the relative contribution of each trophic group to the total (Gambi et al. 2003). Nematode biomass was calculated by biovolume, estimated from all specimens per replicate (Andrassy 1956). Then, dry weight (µg DW) was estimated by multiplying each body volume by an average density (1.13 g cm^-3), and finally the biomass was expressed as carbon content (µg of C/10 cm²), which was considered to be 40% of dry weight (Feller and Warwick 1988).

Differences in total meiofaunal abundance, total nematode biomass, nematode diversity between layers, total meiofaunal abundance and total nematode biomass between the maerl bed and the sandy beach were assessed by one-way analyses of variance (ANOVA). Prior to the ANOVAs, the homogeneity of variances was assessed by the Anderson-Darling test and, when necessary, data were square-root transformed. Tukey’s HSD test was performed to assess significant between-level effects. The ANOVAs and Turkey’s test were performed using the STATISTICA V.10 software. A one-way analysis of similarities (ANOSIM) was used to assess between-layer differences in meiofauna community structure and nematode composition. The SIMPER routine (cut-off of 90%, on fourth-root transformed data) was used to determine the contribution of each meiofaunal taxon and nematode species to the total dissimilarity. ANOSIM and SIMPER analyses were performed using Primer 6+ (Clarke and Gorley 2006).

The total meiofaunal abundance was 1986±457 ind/10 cm². The highest mean abundance occurred in layer 1-2 (748±466 ind/10 cm²) and decreased with increasing depth, being the lowest in layer 3-4 (179±25 ind/10 cm²) (Fig. 2A). The ML and layers 0-1 and 1-2 showed a significantly higher abundance than the deepest layer (ANOVA, p<0.05).

A total of 12 (8±1) taxa characterized the maerl station, including Nematoda, Copepoda + nauplii, Polychaeta, Ostracoda, Kinorhyncha, Oligochaeta, Amphipoda, Cladocera, Gastropoda, Tardigrada, Isopoda and Foraminifera. From layer 0-1 to 3-4, Nematoda were the most abundant (82%-97%), followed by Copepoda (1%-9%) and their nauplii (1%-7%). Polychaeta accounted for 1% in layers 0-1 and 1-2 and Foraminifera for 1% in layer 0-1. In the ML, Copepoda (31%) with their nauplii (60%) were the most abundant (91%), followed by Nematoda (7%), Polychaeta (2%) and Ostracoda (1%) (Fig. 2B). Kinorhyncha, Oligochaeta, Amphipoda, Cladocera, Gastropoda, Tardigrada and Isopoda accounted for less than 1% of the total abundance at all layers and are together indicated as “other taxa”. The community structure differed significantly between sediment layers (ANOSIM, R=44%; P=0.012) (Table S1A), with the differences being mainly due to the increasing abundance of Nematoda and the decreasing abundance of Copepoda along the sediment profile (SIMPER, Table S2). The sediment layers showed dissimilarities ranging from 30% (0-1 vs. 1-2 cm) to 74% (2-3 vs. 3-4 cm) (SIMPER, Table S1A).

The mean total nematode biomass was 108.2±41.9 µg of C/10 cm². It increased along the sediment profile from 2.7±1.2 µg of C/10 cm² in the ML to 40.3±10.9 µg of C/10 cm² in layer 2-3 cm (Fig. 3A). There were 78 genera and 22 families of nematodes (Table S3). The expected number of genera (51) ranged from 15.2±0.5 in layer 3-4 to 20.5±4.9 in layer 0-1, showing a decreasing trend along the sediment profile (Fig. 3B, Table S4). The Shannon index ranged from 2.3±0.3 in layer 1-2 to 2.7±0.3 in the ML, and the Pielou index ranged from 0.7±0.1 in layers 0-1, 1-2 and 2-3 to 0.9±0 in the ML, indicating that all genera were equally represented (Table S4). There were no significant between-layer differences in any diversity index.

Linhomoeidae was the most abundant family (36%), followed by Desmodoridae (30%), Chromadoridae (9%), Comesomatidae (8%) and Xyalidae (4%). The most abundant genera were Terschellingia (21%). Spirinia (13%) Molgolaimus (14%), Metalinhomoeus (12%) and Sabatieria (7%). Molgolaimus was more abundant (39%) in layer 0-1 and decreased with depth to 1% in layer 3-4, and Spilophorella decreased with depth from 20% in the ML to 0% in layer 2-3. Terschellingia increased in abundance with depth from 1% in the ML to 35% in layer 3-4, and Sabatieria increased in abundance with depth from 1% in the ML to 16% in layer 2-3. Metalinhomoeus also increased in abundance with depth from 4% in the ML to 23% in layer 3-4. Spirinia increased in abundance from 2% in the ML to 28% in layer 1-2 and then dropped to 6% in layer 3-4 (Fig. 3C).

The nematode composition in the ML and layer 0-1 differed significantly from those at all other layers (ANOSIM, Global R=37%: p=0.004, Table S1B), mostly owing to the presence of Terschellingia, Spirinia and Perspiria at the deeper layers and Molgolaimus in the top 1 cm (SIMPER analysis, Table S5). The trophic structure of the nematode assemblage was 1A (40%), 2A (30%), 1B (25%), and 2B (5%), although the contribution of some trophic groups changed along the sediment profile. 1A were less abundant in the ML (18%) and reached the highest abundance in layer 0-1 (53%), while 1B and 2B did not change significantly across layers (Fig. 3D). On the other hand, 2A were dominant in the ML (58%), showing the highest biomass (38.4 µg of C/10 cm²). 1B and 2B showed biomasses of 26.1 and 27.0 µg of C/10 cm², respectively, and 1A showed the lowest biomass (16.7 µg of C/10 cm²).

Total meiofaunal density in the maerl bed was five times higher than on the sandy beach (1.986±457 ind/10 cm² vs. 384±16 ind/10 cm², respectively; ANOVA, p<0.05) (Fig. 4). Twelve higher taxa were identified in the maerl bed vs. 8 on the sandy beach (Fig. 4), where Nematoda was the most abundant (96%), followed by Copepoda (3%). Cumacea, Gastrotricha, Isopoda, Ostracoda, Platyhelminthes and Tardigrada accounted for less than 1%. Total nematode biomass was significantly higher in the maerl bed than on the sandy beach (108.2±41.9 and 47.6±2.1 µg of C/10 cm²; ANOVA, p<0.05). Nematode diversity in the maerl bed was more than three times higher than on the sandy beach, where 11 families and 20 genera were identified (Fig. 4), including Richtersia (51%), Trileptum (9%), Daptonema and Omicronema (8% each) as the most abundant. Overall, 66 and 8 nematode genera were exclusive to the maerl bed and the sandy beach, respectively, with 12 being shared (Table S6). 1A dominated in the maerl bed (40%) and 1B on the sandy beach (72%). Overall, the maerl bed showed a lower ITD (0.32) than the sandy beach (0.62), indicating a greater trophic diversity in the maerl bed (Table S6).

The high maerl meiofaunal abundance recorded in the Bay of Brest was similar to that found in other structurally complex and perennial habitats such as seagrass beds (Novack 1989, Pusceddu et al. 2014) and similar micro-habitat rich environments such as coral sediments (Semprucci et al. 2013). The highest meiofaunal abundance was found in layer 1-2, and it progressively decreased along the sediment profile. Conversely, meiofauna is often concentrated in the first few centimetres of sediment (Giere 2009). In the maerl bed, the meiofauna distribution likely responds to its complex architecture, allowing a higher organic and oxygen content subsuperficially. This complexity also protects the sediment below, because meiofaunal organisms are less exposed to hydrodynamic perturbations. Indeed, the low erosion, combined with organic accumulation and dispersal lowering, may favour an increase in abundance subsuperficially. Conversely, taxa living on the surface or between the maerl thalli are less protected from current or wave actions (Martínez et al. 2021).

Among the 12 taxa identified, nematodes were the most abundant in all sediment layers except in the ML, where the community was dominated by Copepoda with their nauplii. Nematodes are well-known to penetrate deeply into sediments thanks to their specialized morphology and high tolerance to anaerobic conditions (Heip et al. 1985), while hydrodynamism can also contribute to the dispersal of the specimens into deeper layers (Zeppilli et al. 2014). Consequently, nematodes were less abundant in the ML, possibly because of the low shelter inherent to its particular architecture, but probably also to a higher predation pressure (Grall et al. 2006).

Copepods and nauplii were particularly abundant in the ML, resembling coral sediment assemblages (Semprucci et al. 2013). This could be related to the additional food resources (macroepiphytic algae and microphytobenthos production) associated with maerl beds (Grall et al. 2006). In addition, the swimming ability of crustaceans, compared with other meiofaunal groups, allows them to speedily return to the seafloor after being resuspended in the water column by current or waves (e.g. Colangelo et al. 2001, Zeppilli et al. 2015b). However, their numbers decreased along the sediment profile, despite the coarse sediment that has been suggested to promote their presence (Colangelo et al. 2001), likely due to their high sensitivity to oxygen depletion or anoxia, unlike nematodes (Moodley et al. 2000).

Some genera of Chromadoridae are almost entirely restricted to upper sediment layers (Platt 1977). Accordingly, the epistrate feeders Spilophorella and Spirinia showed high abundances in the ML and layer 1-2, respectively, and decreased along the sediment profile, probably because of lowering food availability (Pusceddu et al. 2009). Terschellingia and Sabatiera penetrated deep into the sediment, having their highest abundances in layers 3-4 and 2-3, respectively. Terschellingia, a selective deposit feeder, is usually found in anoxic sediments (Wieser 1960). Sabatieria, a non-selective deposit feeder displaying a wide range of ecological preferences (Steyaert 1999), is also known to be a facultative anaerobic organism, a trait allowing it to inhabit suboxic or anoxic sediment layers, so it is thus a common deeper-living nematode (Jensen 1987). Metalinhomoeus was present in all sediment layers, but was particularly abundant in layer 3-4. This genus is known to be associated with subtidal silt or muddy sediments (Wieser 1960), being frequent in sediments with low oxygen levels (Heip et al. 1985). It also has a body morphology that allows it to dig quickly up and down into the sediment, passing easily from reduced to well-oxygenated zones (Jensen 1987). Surprisingly, we found no families typically associated with 3D complex substrates, such as Epsilonematidae and Draconematidae from corals or hard substrates (Raes et al. 2008, Zeppilli et al. 2015b). Therefore, our data support the hypothesis that maerl nematodes benefit mainly from the sediments below the bed, instead of the 3D rhodoliths, which mainly act as a protective layer for the sediments and assemblages below. Moreover, all trophic groups were equally represented in all layers, allowing us to suggest possible differences in food sources along the sediment profile. This may well help reduce the number of competitive interactions, while allowing the different feeding groups to coexist in small sediment patches.

Maerl beds are known to harbour higher macrofaunal biodiversity than bare sediments (Foster et al. 2013), and meiofauna is no exception. In the maerl bed, its abundance was five times higher, and its taxa diversity 1.5 times higher than on the sandy beach. Obviously, nematodes were also more numerous (three times), as was the number of exclusive genera (66). The functional diversity (i.e. trophic diversity) was also higher in the maerl bed. All trophic groups were represented: bacterial and epistrate feeders (1A and 2A respectively) dominated in the maerl bed, while the non-selective deposit feeders (1B) dominated on the sandy beach. Thus, maerl beds seemed to host a rich meiofaunal community and a unique and peculiar nematode community, which is clearly distinguishable from those inhabiting sandy beach ecosystems. Accordingly, maerl beds seem to harbour a rich and specific meiofaunal community that may serve as potential prey for these predator/omnivore nematodes. The physical presence of complex maerl thalli, combined with the numerous resources provided, may explain the difference in meiofaunal diversity between these habitats. Moreover, the sandy beach samples were collected at low tide in the intertidal. This may have also affected the community composition and richness, since only well-adapted fauna can survive the extreme conditions arising from long-term exposure (e.g. higher radiation, salinity and oxygen availability variation) (Baldrighi et al. 2019 and literature therein).

Our study documented novel findings on the abundance, structure and diversity of the meiofauna and nematode communities characterizing the maerl beds from the Bay of Brest. Our results suggest that maerl beds create a more heterogeneous environment, richer in microhabitats, that promotes highly diversified meiofaunal and nematode assemblages in the sediments below, which proved to be particularly rich when compared with more homogenous environments lacking the protective rhodolith layer (e.g. a sandy beach). Maerl beds, which have been largely neglected in meiofaunal studies, have been shown to harbour high meiofaunal assemblages with a very complex structure and a high functional diversity. Therefore, their protection may be crucial for marine biodiversity preservation.