Shifts in benthic megafauna communities after glacial retreat in an Antarctic fjord

The study examines how glacial retreat—a prominent manifestation of warming along the West Antarctic Peninsula—alters the structure and function of Antarctic nearshore benthic megafauna communities. Although the Antarctic continent is largely uninhabitable, its seabed supports rich megafaunal assemblages that play key roles in ecosystem functioning. Marian Cove has experienced ~1.9 km of glacier retreat from 1956 to 2017, creating a natural chronosequence where distance from the glacier correlates with time since deglaciation. Prior observations suggest that sedimentation and ice-scouring shape assemblages and that responses vary with depth and distance from glaciers. However, quantitative, community-wide assessments that integrate both taxonomic structure and functional traits across depth gradients remain limited. This study aims to: (1) quantify distributions of benthic megafauna across sites differing in time since deglaciation; (2) assess structural (taxonomic) and functional diversity; and (3) determine how community structure, function, and environmental drivers vary with time since deglaciation and depth.

Previous work in Antarctic fjords has shown that ice-scour and sedimentation strongly influence benthic community structure, often favoring pioneer taxa in disturbed, ice-proximal areas and filter-feeding dominated assemblages across nearshore habitats. Studies in Marian Cove and elsewhere on the WAP have documented ascidian community shifts with habitat stability and distance from glaciers, but many have focused on specific taxa, limited regions or shallow depths, and often lacked quantitative functional assessments. Taxonomic diversity (TD), capturing the average taxonomic distance between individuals, and functional diversity (FD), reflecting trait-based ecosystem functioning, have been proposed as informative indices of ecological responses to environmental change. Despite their utility, few Antarctic nearshore studies have jointly applied quantitative TD and FD to assess climate-driven community responses.

Study area: Marian Cove (King George Island, WAP) is a ~4.5 km long, ~1.5 km wide fjord with three basins, a sill at ~40 m restricting water exchange, and tidewater glaciers at its inner end. Glaciers retreated ~1.9 km between 1956 and 2017, especially during warmer periods. Physical conditions below ~10 m showed small spatial variation in temperature (−0.3 to −0.6 °C) and salinity (~33.9–34.1 psu). Survey design and image acquisition: Five stations (MC2–MC6) spanning a gradient of distance/time since deglaciation were surveyed. MC2–MC5 were imaged in austral summer 2017/2018 at 10, 20, 30, 50, 70, and 90 m; MC6 (closest to glacier; newly exposed <5 years) was imaged in 2018/2019 at 50 and 70 m. A VideoRay Pro4 ROV with GoPro Hero5 (1080p, 60 fps) and a 50×50 cm stainless-steel quadrat captured quantitative seabed images (quadrat spacing ≥5 m). Ancillary sampling included grabs, trawls, and SCUBA at selected stations (Table 1). Image analysis and taxonomy: All recognizable epibenthic megafauna (>~1 cm) in quadrat images were identified to the lowest possible taxon and counted; colonial taxa were counted as one per colony. Visible infauna (e.g., Laternula elliptica siphons) were included, acknowledging underestimation of infauna due to surface-limited imagery. Identifications used published morphologies and WoRMS. Diversity metrics: For each station-depth, number of taxa, taxonomic diversity (TD; quantitative taxonomic distinctness), and functional diversity (FD) were computed. FD was calculated from density data and a species-by-traits matrix encompassing five biological traits with categorical scoring (Supplementary Table 9). Statistics and environmental drivers: Non-parametric tests (Kruskal–Wallis, Mann–Whitney U) assessed differences. Community patterns were examined via cluster analysis and non-metric MDS (Bray–Curtis). Indicator species were identified using IndVal. BIOENV analyses related community structure (composition and abundance), TD, and FD to environmental variables (grain size fractions: gravel, sand, silt, clay; mean grain size; sorting; depth; distance from glacier as proxy for exposure time; sediment organic matter [SOM]; C/N). PRIMER v6.1.16 was used for TD, clustering, MDS, and BIOENV; FD and IndVal were computed in R v3.4.

• Taxonomic scope: 70 benthic megafauna taxa identified from ROV imagery across five stations (MC2–MC6) and six depths (10–90 m). Taxa richness increased toward the outer cove, especially at 70–90 m. Shallow 10 m depths had few taxa and low spatial variation.
• Abundance patterns: Total mean density was lowest at the innermost MC6 (29.0 ind. m−2) and peaked at inner site MC5 (116.3 ind. m−2). At 30 m near the glacier (MC5), density reached 120.0 ind. m−2.
• Successional composition: Pioneer species dominated ice-proximal sites (MC6: 82.1%; MC5: 52.4%; MC4: 50.8%) and declined outward (MC2: 10.4%). MC6 was dominated by Serpulidae spp. (24.4 ind. m−2; 81%). MC5–MC4 were dominated by pioneer ascidians Molgula pedunculata (31.0% and 22.8%) and Cnemidocarpa verrucosa (16.1% and 17.2%). The outermost MC2 had the highest share of late-successional taxa (24.5%), including Anoxycalyx cf. joubini, Rossella cf. podagrosa, Aplidium cf. radiatum, and Ascidia challengeri. At 50–90 m, pioneer taxa decreased and late-successional taxa increased with distance.
• Community groupings: Cluster/MDS separated four assemblage groups: • Group A (10 m all stations; 20 m at MC2): few taxa (11 total), very low density (24.3 ind. m−2); dominated by motile mollusks (Laternula elliptica, Margarella antarctica, Nacella concinna). • Group B (>10 m at innermost MC6): lowest richness (5 taxa), low density (29.0 ind. m−2), dominated by Serpulidae spp. (indicator; 90% contribution). • Group C (20–90 m near glacier: MC3–MC5; plus 30 m at MC2): dominated by pioneer ascidians M. pedunculata and C. verrucosa (IndVal indicators); very high densities near glacier (up to 120 ind. m−2). • Group D (50–90 m at outermost MC2; >60 years since exposure): pioneer species rare (3.4%); dominated by late-successional A. challengeri (21.8 ind. m−2), Rossella cf. podagrosa (14.1 ind. m−2), Aplidium cf. radiatum (13.2 ind. m−2), and large A. cf. joubini (>50 cm). Multiple indicator species identified.
• Diversity metrics: • TD: very low at the innermost site; otherwise similar among stations (except 10 m where TD was lower overall). • FD: lowest at innermost site; increased with distance from glacier and peaked at 30 m at the outermost station. FD was lowest at 10 m across sites.
• Environmental drivers (BIOENV): • Community structure (composition and abundance) best explained by the combination of distance, depth, and sediment texture (grain size, gravel, sand, silt): R = 0.749, p = 0.001; top single variable: silt (R = 0.634). • TD best explained by gravel + sand (R = 0.493, p = 0.032); top single variable: silt (R = 0.430). • FD best explained by distance + SOM (R = 0.291, p = 0.033); top single variable: distance (R = 0.255).
• Depth-specific patterns: At 10 m (frequent ice-scour and high turbidity), communities had low density, low TD and FD, and little spatial variation; at 50–90 m, communities transitioned from pioneer-dominated near the glacier to mature, late-successional assemblages outward.
• Successional stages inferred along the chronosequence: high disturbance, colonization (pioneer dominance, density peak), transition (declining pioneer density, increasing richness), and maturing (late-successional, slow-growing taxa; increased FD).

The findings demonstrate that glacial retreat drives distinct structural and functional trajectories in Antarctic nearshore benthic megafauna depending on depth and time since deglaciation. At shallow depths, persistent ice-scouring, meltwater-induced turbidity, and associated stress maintain low-diversity, functionally monotonous communities regardless of exposure time, constraining succession. In contrast, deeper zones experience reduced disturbance and reveal a clear successional sequence: initial colonization by rapidly recruiting pioneer tube worms and solitary ascidians; a transition phase with declining pioneer dominance and appearance of taxa requiring more stable habitat; and a mature stage dominated by slow-growing/recruiting taxa (colonial ascidians, glass sponges) with larger body sizes and longer lifespans. Mechanistically, community structure (taxonomic composition and abundance) is linked to physical habitat stability and formation time, reflected by sediment texture and depth, while community function (FD) additionally responds to food availability (SOM) and exposure time (distance), peaking where both stability and diverse food sources (e.g., at 30 m) support varied feeding strategies. The divergence in drivers of TD versus FD underscores the need to assess both taxonomic and functional facets to understand ecosystem responses. Ecologically, continued glacial retreat that enhances shallow-water disturbance could suppress diversity and succession, potentially reducing secondary production and benthic carbon storage; conversely, newly exposed deeper areas may accumulate biomass and functional complexity over time, enhancing carbon sequestration by longer-lived taxa.

This study shows that shifts in the structure and function of the benthic megafauna community in a deglaciated Antarctic fjord vary with depth and distance from the glacier. Using ROV surveys across Marian Cove, we found that frequent disturbance limits development at shallow depths, while deeper communities progress from colonization to transition and maturing stages with time since deglaciation. Community structure is primarily influenced by physical habitat stability and formation period, whereas community function is additionally shaped by food availability. Considering both structural (TD) and functional (FD) dimensions is therefore essential for interpreting ecological responses to glacial retreat. Given that distance from the glacier reflects exposure time in Marian Cove, present-day spatial patterns capture past successional processes and provide a basis for predicting future benthic ecosystem responses to ongoing climate change in Antarctica.

• Environmental covariate gaps: Station MC6 lacked environmental measurements and was excluded from BIOENV analyses, limiting inference at the most ice-proximal site.
• Method constraints: Image-based surveys capture surface-visible epibenthos (>~1 cm); infauna are likely underestimated, and small or cryptic taxa may be missed.
• Spatial/depth coverage: MC6 imagery was only at 50 and 70 m; 10–30 m depths at the innermost site were not imaged in this campaign.
• Generalizability: Results derive from one fjord system and observational chronosequence; unmeasured factors and interannual variability may influence broader applicability.