Unveiling in situ oxygen, carbon and nutrient cycling of a sponge-driven biological hotspot in the arctic

TL;DR

This study shows that deep-sea sponge grounds in the Arctic are hotspots for carbon and nutrient cycling, with high respiration rates and nutrient release linked to sponge biomass.

Contribution

The study provides in situ evidence that deep-sea sponge grounds are hotspots of carbon and nutrient cycling, contrasting with previous ex situ assumptions.

Findings

Respiration rates in sponge grounds were 7–21 times higher than in surrounding soft sediments.

Nutrient fluxes of ammonium, phosphate, nitrite, and nitrate correlated with sponge biomass.

Sponge grounds suggest a high organic carbon demand beyond surface-derived supply.

Abstract

Deep-sea sponge grounds are habitat-forming benthic communities characterized by high biomass and structural complexity. Despite their ecological significance, their role for the deep-sea environment remains poorly understood and their functioning is often inferred from ex situ studies. We hypothesized that deep-sea sponge grounds exhibit substantially higher respiration and nutrient turnover than surrounding soft sediments, making them hotspots of carbon and nutrient cycling in the deep sea. Integrated respiration and nutrient cycling were quantified in a sponge ground on the summit of an Arctic seamount (Schulz Bank, ~ 580 m depth). We used in-situ incubation chambers measuring oxygen consumption, prokaryotic cell removal, and inorganic nutrient fluxes. Respiration rates ranged from 0.13 to 0.93 mmol O₂ m⁻² h⁻¹, which is comparable to cold-water coral reefs and up to 7–21 times higher than reported for soft sediments of the Arctic deep sea. This indicates a high organic carbon demand exceeding surface-derived supply, suggesting the uptake of additional food resources. All incubations showed net release of ammonium, phosphate, nitrite and nitrate, with fluxes correlating with sponge biomass. Our results demonstrate that deep-sea sponge grounds function as hotspots of carbon and nutrient cycling and suggest distinct functional contributions of sponge groups and their microbiome.

The online version contains supplementary material available at 10.1038/s41598-026-41798-4.

Figures

Prokaryotic cell abundance during the different in-situ incubations, with their R^2^ and p- value. Green = Low sponge biomass, blue = high sponge biomass of low or high microbial abundance sponges. Only low sponge biomass incubation no. 2 showed a significant exponential regression (bold, y = 210703 *e^− 0.0013x^).

Inorganic nutrient fluxes of all incubations divided in low (green) and high (blue) sponge biomass incubations (in their consecutive order) including average of low or high biomass (light grey ± SD) and average of all incubations (grey ± SD). Fluxes of HMA and LMA dominated communities are indicated in dark and light blue, respectively. (A) DIN (dissolved inorganic nitrogen), (B) NH_4_^+^ (ammonium), (C) NO_3_^-^ (nitrate) (D) NO_2_^-^ (nitrite), (E) PO_4_^3-^ (phosphate).

Community Respiration Rates of soft sediment data (black dots^[52](#CR52)^, as well as deep hotspots like coral reefs (blue) and sponge ground (pink/red): Rockall Bank (blue triangle^[46](#CR46),[47](#CR47)^—Træna Reef (pink tringle^[40](#CR40)^, (pink circle^[44](#CR44)^, (blue square^[40](#CR40)^—Mingulay Reef (blue circle^[48](#CR48)^—Stjernsund reef (blue diamond^[48](#CR48)^— Vazella ground (pink diamond^[20](#CR20),[45](#CR45)^—Schulz Bank (red star, this study). Adapted from^[40](#CR40)^. The solid line shows an empirical model of the double-exponential decrease with water depth, fit su

Full text

Introduction

Deep-sea benthic communities are found in extreme conditions, including darkness or cold temperatures, without any photosynthetic organisms to directly fuel their food web. Despite these challenges, biological hotspots with high biodiversity and biomass thrive in many areas of the deep ocean, like cold-water coral reefs and deep-sea sponge grounds^1,2^. Each of these communities must have developed unique adaptations to survive in their extreme surroundings, yet many of the processes that enable their survival remain unknown. Furthermore, while their role in global element budgets was previously considered minor, recent studies have shown that due to their widespread distribution and high nutrient cycling capacity, the influence of sponge grounds, particularly on carbon, nitrogen and silicon cycles should not be underestimated^3–5^. With increasing deep-sea exploration efforts, sponge grounds are being documented with growing frequency worldwide, particularly in the North Atlantic and both polar regions^6^. They have been documented on continental shelves, slopes, seamounts, and other topographic features, and in some regions can extend over thousands of square kilometres (e.g. ~8000 km² on the Scotian Shelf^7^. However, comprehensive global area estimates are currently lacking due to incomplete mapping and uneven survey efforts as well as the unknown exact environmental preferences needed for distribution models^8,9^. Sponges are one of the most common and biomass-rich megafaunal organisms in the deep sea, often with a scattered distribution^10^. In some areas they dominate benthic communities, with biomass stocks exceeding 200 g C m⁻²^1^, forming dense assemblages characterized by either monospecific or mixed species assemblages^11^.These sponge grounds can even create reef-like structures through deposition of dense, tens of cm-thick, layers of skeletal (glass-) needles, so-called spicule mats^11–13^. They provide important ecosystem services by enhancing habitat heterogeneity and providing spawning, nursery and foraging areas for many fish and invertebrate species^7,13^. In addition, sponges affect the water column in their immediate surroundings due to their ability to pump and process large volumes of water (upper estimate around 35 mL min^− 1^ (cm sponge)^−3^^14,15^ with a 70–99% retention efficiency^15,16^. This way they play an important role in benthic-pelagic coupling, extensively filtering particulate organic matter (POM) from the water column (e.g., bacterio- and phytoplankton, e.g.^15,17,18^. Recent studies have shown that common North-Atlantic sponge species can also take up dissolved organic matter (DOM), that constitutes the largest organic carbon stock in the ocean^19^ and is especially important for the deep sea, where POM from photosynthetic production is scarce. Dissolved organic carbon (DOC) can account for over 90% of their carbon diet and is thought to be primarily oxidized to support respiration and meet the energetic demands of basal metabolism, rather than being incorporated into biomass for growth or reproduction^20,21^. Unlike most other animals, sponges can satisfy the majority of their carbon requirements from DOC because their high-volume filtration system and specialized choanocyte cells, together with microbial symbionts in many species, allow efficient capture and intracellular processing of dissolved organic compounds that are believed to be too dilute to support typical particulate-feeding of the majority of aquatic animals^22^. Through the so-called sponge loop^3^, sponges transform DOC into particulate organic carbon (POC) and transfer it to higher trophic levels by detrital pathways fuelling diverse benthic ecosystems^4,23,24^.

The highly efficient uptake of dissolved and particulate nutrients by sponges is further supplemented by metabolic pathways that are facilitated by the abundant sponge microbiome^25^, which takes up dissolved nutrients and transfer accessible forms of (particulate) nutrients to the sponge. The host will, on the other hand, return metabolic waste products back to its microbiome or free-living microbes in their surrounding environment resulting in highly efficient recycling of host-waste products^4,26^. Sponges typically contain a diverse community of microbial symbionts, like bacteria and archaea^27,28^, which enable them to use inorganic substrates such as ammonium (NH_4_^+^) or nitrite (NO_2_^−^ ) via chemoautotrophic pathways to generate energy^29–31^. In fact, microbial biomass in so-called high microbial abundance (HMA) sponges makes up 20‒30% of the sponge holobiont, while low microbial abundance (LMA) sponges contain fewer and smaller-sized microbes with a lower diversity, more similar to that found in the ambient seawater^28^. Microbial symbionts play an important role in nutrient recycling and the processing of host waste products in both LMA and HMA sponges, where the metabolic activity of HMA sponges is strongly shaped by their abundant microbial symbionts and LMA sponges rely more on direct filter-feeding of particulate food and exhibit a metabolism that is largely driven by the host itself^32,33^.

Despite their widespread distribution, key aspects of deep-sea sponge communities, such as respiration rates and the exchange of organic and inorganic nutrients, remain poorly understood due to technical restraints of deep-sea work. Advanced technologies, e.g. the use of remotely operated vehicles (ROVs) for deployments of experimental devices on the deep-sea floor, are starting to unravel their unique role in the environment revealing their ecological significance^4,11,12^.

We hypothesize that deep-sea sponge grounds function as active biogeochemical hotspots in the deep sea, characterized by high rates of organic carbon processing and inorganic nutrient cycling. We conducted in-situ incubation experiments on a densely populated deep-sea sponge ground at the Schulz Bank summit, a seamount along the Arctic Mid Ocean Ridge (73°50’N, 7°34’E). The in-situ net uptake and release rates of oxygen, organic carbon (i.e., bacterioplankton) and inorganic nutrients (i.e. ammonium, nitrate, nitrite and phosphate) within this sponge-driven biological hotspot will demonstrate the role of deep-sea sponge communities to nutrient and carbon cycling, which were identified as important ecosystem services. Anthropogenic activities including deep-sea mining and bottom trawling threaten these unique ecosystems, with potential cascading consequences for nutrient availability, microbial community structure and food supply to higher trophic levels throughout the wider deep-sea ecosystem.

Materials and methods

Study site and setting

The Schulz Bank seamount (73°50′N, 7°34′E), part of the Arctic Mid-Ocean Ridge between the Norwegian and Greenland Basins, rises 2,700–3,000 m from the basin floor to a summit at ∼560 m depth (Fig. 1A, B). It hosts diverse, dense benthic communities with vulnerable marine ecosystem (VME) indicators like sponges (Geodia spp.) and corals^13,34^, driven by a “seamount effect”^35^. Strong near-bed currents (up to 76 cm s⁻¹) generate a Taylor cap or anticyclonic circulation that isolates summit waters, promotes upwelling of nutrient-rich deep water, and enhances retention of organic matter from seasonal phytoplankton blooms, alongside oscillatory tidal flows that sustain bentho-pelagic coupling and elevated near-bed particle concentrations^35^. Previous measurements reveal bottom water temperatures of -0.5–1 °C, salinities of 34.9–35.0, and dissolved oxygen ∼307 µmol L⁻¹ and labile carbon peaks supporting sponge ground development year-round^35^. The summit of the Schulz seamount is completely covered with a continuous spicule mat with a high abundance of sponges and associated fauna at water depths between 579 and 1100 m, forming a reef-like ecosystem^13,34,36^. This sponge ground has a community density of around 120 individuals (ind) m^− 2^, where sponges make up around 57% of the total morphotaxa density. Large structure-forming sponges have a density of around 9 ind m^− 2^^13^. On the seamount flanks (1,100–15000 m water depth), mixed sediment types with patches of hard substratum and spicule mats we found, forming a transitional zone incorporating similar taxa as found on the summit^34^. The lower slope until 2400 m depth shows predominantly bedrock walls with a high density of fauna. Soft sediments were found between 2000 and 2800 m depth with lower abundances of sponge fauna and associated fauna. The different sponge cover is likely connected to the hydrodynamic conditions around the seamount. The summit of the seamount lays at the interface between two water masses (~ 500–600 m), which allows internal waves of up to 200 m height to break at the summit. This produces strong currents and turbulences at the summit but less at the slope and foot of the seamount^35^.

The benthic megafaunal community of the summit is dominated by hexactinellids (phylum Porifera, Class Hexactinellida, common name “glass” sponges) and demosponges (Porifera, Demospongiae) associated with other fauna including ascidians (Ascidiacea spp.), cnidarians (soft corals), echinoderms and demersal fish species^13^. The primary structure-forming hexactinellid species are Schaudinnia rosea, Scyphidium septentrionale, Trichasterina borealis, and Asconema foliata (vase-shaped sponges) and the main demosponge taxa are Geodia spp., Stelletta rhaphidiophora (massive, subglobular) and Lissodendoryx complicata (branched). Schaudinnia rosea was protruding further into the water column (up to about 1 m) and often growing on top of the near-bottom globular sponges like Geodia spp. and S. rhaphidiophora. The sponges S. rosea,* S. septentrionale*,* T. borealis*,* L. complicate* and A. foliata are considered LMA species, whereas Geodia spp. and S. raphidiophora are considered to be HMA sponges, with Chloroflexi and Acidobacteria as the main microbes associated with Geodia spp.^27^.

In-situ incubations

To determine oxygen, bacterioplankton, and inorganic nutrients net fluxes (NH_4_^+^, NO_3_^−^, NO_2_^−^, PO_4_^3−^), two 50 × 50 × 50 cm (125 L volume) acrylic incubation chambers with open bottoms were deployed at ~ 580 m. The chambers were designed by the NIOZ Royal Netherlands Institute for Sea Research and are modified version of the chambers used by Stratmann et al.^37^. Each chamber can be operated up to 6,000 m water depth and is equipped with a pre-programmable 6 × 60 mL syringe water sampling port, a stirring plate (adjustable speed) ensuring homogenous mixing, as well as a dissolved oxygen sensor (Advantech-ARO-USB) to continuously monitor oxygen concentrations and temperature (20 s interval) during the incubations (Fig. 1C). A lead rope attached to the outside bottom weighted the chambers to prevent that they were topping over due to strong currents and were kept submerged up to 10 cm in the sediment, creating an enclosed chamber space/volume with negligible leakage. Limited exchange with the surrounding environment can not completely be excluded due to the porose nature of the sponge spicule mat and potential release of pore water into the chamber. Nevertheless, we did not record any obvious inflow of surrounding water (e.g. no unexpected changes in oxygen were recorded).

Fig. 1(a) Schulz Bank seamount as part of the Mid Atlantic Ridge (GEBCO WMS, https://wms.gebco.net/mapserv?) (b) Summit of the seamount with in-situ incubation positions (see Table 1 for details), (c) In-situ incubation chamber with main components indicated, (d) Chamber deployed on the sponge ground incubating a seamount community including a large LMA sponge.

The ROV Ægir 6000 carefully positioned the two incubation chambers at three random locations at the Schulz Bank summit between 577 and 583 m water depth (73°50’N, 7°34’E; Fig. 1B, D; Table 1) during a G.O. Sars research cruise in 2017 (July 20th – August 6th, 2017). Two chambers were deployed simultaneously, enclosing benthic communities with or without one or more prominent sponge individual(s), to exclude any effect due to temporal variations in the environmental conditions. Incubations without large prominent sponges represented a community control in opposite to the sponge dominated incubations (Supplementary Fig. S1). The continuous benthic cover did not allow for a (soft) sediment negative control without sponges and other megafauna on the seamount summit. Therefore, controls included benthic community without prominent sponges. Three incubations contained a high sponge biomass (> 240 g of sponge DW) and the remaining three incubations contained a low biomass (< 30 g of sponge DW, from here on referred to as “high” and “low” sponge biomass, respectively; Table 1). These incubations also contained a variable associated megafauna community which was of a similar composition between incubations, including anemones, ascidians and soft corals (Table 1). After chamber placement the hatch was kept open for the first 10–15 min while starting the mixing plate, to allow potentially resuspended material to be washed out and settle. Then the hatch was closed and the sampling program was started by the ROV through a lever on the outside of the chamber (Fig. 1C). After 2–5 min the first sample of 60 mL (t_0_) was taken. During the 4–6 h (Table 1) incubations in total six consecutive water samples were taken autonomously at equal time intervals (48 to 72 min, depending on the overall deployment time). At the end of each incubation, the hatch was opened again and the chambers were carefully lifted. Visible incubated megafauna including the sponges (Table 1) were retrieved with the titanium arm of the ROV and stored in its bioboxes. On board the research vessel, sponge biomass (dripping wet weight) was assessed. Tissue samples were flash-frozen in liquid nitrogen, and stored at − 80 °C. Dry weight (DW) was later measured in the laboratory after a 24 h freeze-drying process. Water samples were processed within 2 h after recovery of the chamber and subsampled for inorganic nutrients (NH_4_^+^, NO_3_^−^, NO_2_^−^, PO_4_^3−^; 5 mL) and prokaryotic cell abundance (1 mL). No technical replicates were taken due to the limited amount of sampling water. Benthic megafauna in the chambers was analysed using video images made by the ROV (Supplementary Fig. S1) and samples retrieved by the ROV.

Table 1. Stations and characteristics of the in-situ incubations with dry weight and wet weight of sponge biomass indicated as well as the presence of associated fauna as observed from video image analysis.NoStationIncubation time (h)Major incubated sponge speciesSponge biomass(kg ww)SpongeBiomass(g dw)Other visible megafauna1GS2017-110-34J4No large sponge fauna< 0.113x AnemonesTunicatesLow spongebiomass2GS2017-110-38F6No large sponge fauna< 0.1305x TunicatesAnemone Small L.complicata Small *Craniella sp.*3GS2017-110-40J6No large sponge fauna< 0.1 1

L. complicata

Tunicates 4GS2017-110-34F4 Geodia parva (HMA)

1.8 357Encrusting yellow sponge4x Soft coralsTunicatesAnemonesCraniella sp.High spongebiomass5GS2017-110-38J6Schaudinnia rosea (LMA)Geodia hentscheli (HMA)1.190.6240TunicatesSmall L. complicataSmall Craniella sp.6GS2017-110-40F6Schaudinnia rosea (LMA)Geodia hentscheli (HMA)2.10.281303Small L. complicata

Laboratory analysis

Inorganic nutrient analysis

Samples for dissolved inorganic nutrient analysis were filtered through sterile polycarbonate membrane syringe filters (0.2 μm with attached 0.8 μm prefilter, Whatman Nuclepore), collected in 6 mL high-density polyethylene ponyvials and stored at -20 °C until further analysis. Inorganic nutrient concentrations were determined by colorimetric analyses using a QuAAtro Gas Segmented Continuous Flow Analyser (Seal Analytical Ltd., UK) simultaneously on four channels. Measurements were calibrated against standards diluted to known nutrient concentrations with low nutrient seawater. A “nutrient cocktail” with PO_4_^3−^ and NO_3_^−^ was used as a standard in every run to monitor the performance. Precision and accuracy of the standard showed between run reproducibility better than 1.5%, but typically 0.7% of its average value.

Prokaryotic cell abundance analysis

Unfiltered water samples (1 mL) for prokaryotic cell abundance from the benthic chamber were collected in 2 mL cryovials and fixed on board with 20 µL 25% EM grade glutaraldehyde solution for 15 min at 4 °C. After fixation, the samples were snap frozen in liquid nitrogen and stored at -80 °C until further analysis. Bacterioplankton abundances were analyzed using a FACS Calibur flow cytometer (Becton Dickinson,* San Jose*,* Calif.*) with a 15 mW 488-nm air-cooled argon-ion laser and a standard filter setup^38^. Upon thawing, samples were 2x diluted in Tris-EDTA buffer (pH 8.0, 10 mM Tris-HCl, Roche Diagnostics; 1 mM EDTA, Sigma-Aldrich) to avoid electronic coincidence and were stained with SYBR Green I (Molecular Probes; 1:10,000 final concentration). Samples were left in the dark for 15 min at room temperature until they were analyzed for 1 min at 49 µL min^− 1^. Prokaryotic cells (including bacteria and archaea) were identified based on green fluorescence and side scatter signals. Events with low fluorescence and scatter were excluded to minimize the contribution of SYBR-stained non-biological particles. Picoeukaryotic cells, characterized by higher fluorescence intensity and scatter, were not included in the prokaryotic gate. Counts were corrected using procedural blanks consisting of Tris-EDTA buffer stained with SYBR Green I to account for background fluorescence and non-specific staining.

Data was analyzed using Flowing Software 2.5.1 freeware.

Oxygen, bacterioplankton and inorganic nutrient fluxes

The oxygen flux was calculated from a linear regression of oxygen concentration changes during the incubation. The first ~ 50 min of each incubation were excluded from the calculation due to initial non-linear noise related to chamber equilibration, disturbance, or other transient effects (Supplementary Fig. S2).

Oxygen removal rates were used to estimate carbon consumption assuming a balanced respiratory quotient (RQ) of 1: 1 mol C respired = 1 mol O_2_ removed^20,39^. This value was used since no further information about the RQ in deep-sea ecosystems is yet available, whereas it can be expected that the value lays between 0.8 and 1.6, since it depends on the elemental composition of the substrate and the fraction of substrate that is respired^23,29,40,41^. This introduces ± 20–30% uncertainty in absolute C-respiration rates but does not alter interpretation of metabolic dominance patterns or sponge ground significance, which are robust across the RQ range.

Prokaryotic cell net removal rates were estimated assuming an exponential clearance of cells in the incubations over time^20^, which corresponds to the observation that the benthic megafauna is dominated by filter-feeding organisms (Table 1). This exponential model was fitted to each individual incubation. Initial net removal rates were estimated based on the tangent of each exponential model at t_0_. In order to estimate the amount of carbon delivered by bacterioplankton, we assumed that prokaryotic cells are comparable to bacterioplankton abundance and amounted 12.4 fg C for each cell, which is an average of values measured for open ocean bacteria (2–23 fg C^42^.

To calculate inorganic nutrients over time, a linear regression analysis was performed for each individual incubation including all time points. Therefore, fluxes were calculated from the slope of a linear regression fitted to the observed concentration change of each incubation. All fluxes were corrected for incubation volume (125 L) and normalized to the incubated surface area (0.25 m^2^) to quantify ecosystem-scale metabolic rates (µmol m⁻² h⁻¹) across varying sponge biomass densities. This area-based normalization accounts for higher organismal density in sponge-rich chambers while enabling comparison with regional deep-sea benthic flux studies. Due to the small sample size (n = 3 per biomass group, n = 1–2 within sponge groups), results of statistical tests should be interpreted with caution. We performed a two-sample t-test to compare the mean slope between groups, as is standard practice, but acknowledge the limited statistical power and will discuss the results accordingly.

Results

Oxygen flux

The bottom water temperature inside the incubation chambers was on average 0.15 ± 0.01 °C (average ± SD throughout text). The sponge ground community on the Schulz Bank summit respired on average 0.63 ± 0.31 mmol O_2_ m^− 2^ h^− 1^ (Fig. 2A). The high sponge biomass incubations (No. 4–6, > 1 kg WW) had a significantly, more than two-fold, higher oxygen uptake (0.89 ± 0.04 mmol O_2_ m^− 2^ h^− 1^) compared to incubations with a low biomass of sponges (No. 1–3, 0.37 ± 0.22 mmol O2 m− 2 h− 1, t-test, p = 0.042). There was a significant positive linear fit between incubated sponge biomass and respiration (y = 0.0017x + 0.36, r^2^ = 0.8, p = 0.016, Fig. 2B). The slope indicates sponge-specific respiration of 0.0017 mmol O₂ g⁻¹ DW h⁻¹ (equivalent to 1.7 µmol O₂ g⁻¹ DW h⁻¹). The y-intercept (0.36 mmol O₂ m⁻² h⁻¹) can be expected to represent background respiration from associated megafauna and microbes, representing about 57% of the average community respiration. In high sponge biomass settings, sponges drive > 60% of the total metabolism.

Fig. 2(a) Dissolved oxygen uptake of all benthic incubations divided in low and high sponge biomass incubations including average of low or high biomass (light grey ± SD) and average of all incubations (grey ± SD) (incubations in their consecutive order), (b) Benthic oxygen consumption per sponge biomass with linear fit. A linear fit is shown but sample size is too small for true statistical inference. Green = low sponge biomass, dark blue= high biomass of high-microbial abundance (HMA) sponges, light blue= high biomass of low-microbial abundance (LMA) sponges.

Prokaryotic cell abundance fluxes

The initial prokaryotic cell abundance at the start of the incubations ranged from 5.6 × 10^4^ to 2.4 × 10^5^ cells mL^− 1^ (average 1.5 × 10^5^ cells mL^− 1^, Fig. 3). The in-situ incubations did not show a significant decrease in 5 out of 6 incubations, of which the net removal rates were set to 0. A significant net removal rate of 5.5 × 10^6^ cells m^− 2^ h^− 1^ (r^2^ = 0.96, p < 0.01) was found for one of the low sponge biomass incubations (Fig. 3, No. 2), which corresponds to a bacterioplankton carbon removal rate of 1.4 × 10^− 5^ mmol C m^2^ h^− 1^.

Fig. 3. Prokaryotic cell abundance during the different in-situ incubations, with their R^2^ and p- value. Green = Low sponge biomass, blue = high sponge biomass of low or high microbial abundance sponges. Only low sponge biomass incubation no. 2 showed a significant exponential regression (bold, y = 210703 *e^− 0.0013x^).

Inorganic nutrient fluxes

Five out of six in-situ incubations showed a linear increase of inorganic nutrient concentrations during the incubations (Supplementary Fig. S4). One incubation showed a slight decrease in NO_2_^−^ concentrations over time (No.1, low sponge biomass) and one incubation did not show a linear trend due to a switch from release to uptake of NO_3_^−^ (No.3, low sponge biomass).

The sponge ground community showed, on average, a net release rate of 39.9 ± 24.3 mmol DIN m^− 2^ h^− 1^ (DIN: dissolved organic nitrogen; the sum of NH_4_^+^, NO_3_^−^ and NO_2_^-^) during the in-situ incubations (Fig. 4A). The three high sponge biomass incubations showed an almost six times higher average net release rate of DIN (29.2 ± 24.6 mmol m^− 2^ h^− 1^) than low biomass incubations (4.8 ± 2.7 µmol m^− 2^ h^− 1^), albeit because of the large differences between incubation with different sponge groups, this offset is not significant (t-test, p = 0.20; Fig. 4).

The net release of NH_4_^+^ by the sponge ground community was on average 18.6 ± 21.1 µmol m^− 2^ h^− 1^ (Fig. 4B). Incubations with a higher sponge biomass appeared to have a higher release of average NH_4_^+^ (28.0 ± 23.6 µmol m^− 2^ h^− 1^), compared to incubations with a lower biomass of sponges (4.63 ± 2.58 µmol m^− 2^ h^− 1^, t-test, p = 0.17; Fig. 4), although this difference was not significant. The two incubations with a high biomass of LMA sponges (No. 5 and 6, light blue, Fig. 4B) released around 10 times more NH_4_^+^ (39.9 ± 16.3 µmol m^− 2^ h^− 1^) than the incubation with a high biomass of HMA sponges (No.4, dark blue, 4.1 µmol m^− 2^ h^− 1^) or low sponge biomass incubations (green, 4.6 ± 2.5 µmol m^− 2^ h^− 1^).

NO_3_^−^ was released in all incubations with an average of 26.6 ± 7.6 µmol m^− 2^ h^− 1^ (Fig. 4C). Incubations with a higher sponge biomass show a trend towards a lower release of NO_3_^−^ (22.1 ± 5.8 µmol m^− 2^ h^− 1^) compared to incubations with a lower biomass of sponges (31.2 ± 6.9 µmol m^− 2^ h^− 1^, t-test, p = 0.16). There was a lower net NO_3_^−^ release rate within the two high biomass incubations (No. 5 and 6) dominated by LMA sponges (18.8 and 20.6 µmol m^− 2^ h^− 1^, respectively) compared to the high biomass of HMA incubation (28.7 µmol m^− 2^ h^− 1^, No. 4) and low sponge biomass incubations (31.1 ± 6.9 µmol m^− 2^ h^− 1^).

NO_2_^-^ net release in all incubations (average of 1.3 ± 1.4 µmol m^− 2^ h^− 1^, Fig. 4D) was approximately one order of magnitude lower compared to the other nitrogen fluxes. Similar to NH_4_^+^, highest release of NO_2_^-^ was observed in the incubations with a high biomass of LMA sponges (2.9 ± 0.9 µmol m^− 2^ h^− 1^). The release was on average about double the flux of incubations with a high biomass of HMA (1.3 µmol m^− 2^ h^− 1^), and 10 times the flux of low sponge biomass incubations (0.3 ± 0.7 µmol m^− 2^ h^− 1^).

Fig. 4. Inorganic nutrient fluxes of all incubations divided in low (green) and high (blue) sponge biomass incubations (in their consecutive order) including average of low or high biomass (light grey ± SD) and average of all incubations (grey ± SD). Fluxes of HMA and LMA dominated communities are indicated in dark and light blue, respectively. (A) DIN (dissolved inorganic nitrogen), (B) NH_4_^+^ (ammonium), (C) NO_3_^-^ (nitrate) (D) NO_2_^-^ (nitrite), (E) PO_4_^3-^ (phosphate).

In general, PO_4_^3−^ was released in all incubations, with an average of 4.9 ± 3.4 µmol m^− 2^ h^− 1^ (Fig. 4E). PO_4_^3−^ fluxes appeared to have a 2 times higher net release rates in the high sponge biomass incubations (6.6 ± 4.1 µmol m^− 2^ h^− 1^) compared with low biomass incubations, yet this difference was not significant (3.3 ± 2.8 µmol m^− 2^ h^− 1^, t-test, p = 0.32). Incubations with a high biomass of LMA sponges tended to have a 20 to 40% higher release rate of PO_4_^3−^ (8.5 µmol m^− 2^ h^− 1^) compared to the incubation with high HMA biomass (1.9 µmol m^− 2^ h^− 1^) and low sponge biomass incubations (3.5 µmol m^− 2^ h^− 1^), respectively.

The nitrogen-to-phosphorus ratio (N: P) in the dissolved inorganic nutrient pool (DIN: PO_4_^3−^) of the water surrounding the sponge ground community (15.3:1) was slightly below the Redfield elemental ratio of 16:1 (Supplementary Fig. S4&5). During the incubations, the ratio of DIN: PO_4_^3−^ decreased over time, showing a relatively higher release of P over N. The ratio between carbon consumption (respiration) and phosphate and DIN release were very similar between all incubations and was on average 90:1 for ΔC: ΔPO_4_^3−^ (Redfield ratio = 106:1) and 10:1 for ΔC:ΔDIN (Redfield ratio 6.6:1, Supplementary Fig. S5).

Discussion

Here, we present the first observations of in-situ oxygen, prokaryotic cell, and inorganic nutrient dynamics of an Arctic deep-sea sponge ground. Due to the technical and logistical challenges of conducting in situ experiments in the deep sea, our study is subject to several methodological constraints that should be considered when interpreting the results: (1) limited replication (n = 1–3 per treatment) precludes formal statistical comparisons, requiring interpretation of trends rather than robust treatment effects; (2) continuous sponge cover prevented sediment-only controls, limiting separation of sponge-mediated fluxes from sediment and associated megafauna contributions; (3) open-bottom chambers inserted into porous spicule mats may permit limited porewater exchange and hydrodynamic artefacts; (4) single samples per time point constrained flux resolution for all analytes; and (5) unmeasured DOC prevents complete carbon budget reconstruction.

Although regression analyses reveal strong sponge biomass-dependent trends, incubations captured whole-community metabolism rather than isolated sponge physiology. Future studies using higher replication, sediment controls, direct DOC measurements, and benthic eddy covariance would strengthen sponge-specific quantifications. Nevertheless, observed trends remain consistent with prior work and provide valuable context for nutrient cycling in these ecosystems. Later sections elaborate these interpretive constraints.

Oxygen consumption as well as inorganic nutrient cycling of the Schulz Bank benthic community was driven by sponge biomass, whereas inorganic nutrient cycling seemed to be additionally driven by the type of sponges, i.e. the marked differences in abundance and biomass of microbial symbionts of sponges. Oxygen, ammonium, nitrite and phosphorus net fluxes showed positive correlations with increasing sponge biomass (Fig. S3). Fluxes were altered by sponge biomass, despite comparable associated megafauna communities (tunicates, ascidians, brittle stars) across incubations whose biomass contributions could not be quantified (cf. regression partitioning: sponges ~ 43% of average community respiration). Prokaryotic cell abundances did not significantly change during most incubations. We hypothesize that prokaryotic abundances were already depleted prior to the start of the incubations by the filter-feeding benthic community (see discussion below). We observed that sponges changed the rates of respiration, nitrogen, and phosphorus species up to 10-fold thereby rivaling respiration rates of well-established biological hotspots, such as deep-sea cold-water coral reefs^40^. Accordingly, despite limitations in replication and methodological resolution, our data suggest that deep-sea sponges provide important ecosystem services beyond serving as habitat and adding structural complexity: they significantly alter biological activity and drive ecological processes such as biochemical nutrient cycling, warranting future studies with higher replication to confirm their value which we are increasingly beginning to appreciate as new research emerges.

The Schulz Bank sponge ground exhibited area-specific community fluxes from 3.2 to 23 mmol O_2_ m^-2^ day^-1^ (average 15.0 ± 7.6 mmol O_2_ m^-2^ day^-1^). These respiration rates were up to 21 times higher than soft sediment incubations from the Arctic deep sea (2467 m depth, − 1.1 ± 0.1 (SD) mmol O_2_ m^− 2^ d^− 1^^43)^ and up to 7 times higher than the bare sediment close to the Norwegian Træna Coral Field at 300 m depth (3.4 ± 0.6 mmol C m^-2^ day^-1^^37)^.

Instead, these respiration rates are comparable to other known sponge-dominated ecosystems, such as the demosponge grounds near the shallower Norwegian Træna Coral Field with a density of 0.55 ind m^2^ and a biomass of 6.3 ± 2.9 kg WW m^− 2^^40,44^. Their estimates were based on an ex-situ approach within aquaria and showed that large demosponges respired similar amounts of oxygen (3–34 mmol O_2_ m^− 2^ day^− 1^). Non-invasive in-situ Aquatic Eddy Correlation (AEC) measurements in the same area showed slightly higher, but overall comparable oxygen uptake rates (45.3 ± 20.9 mmol O_2_ m^− 2^ day^− 1^^40)^. Yet, the respiration rate of the sponge ground community at Schulz Bank is lower compared to the estimated respiration of the Vazella sponge ground on the Canadian shelf (150–200 m depth; average temperature 10 °C ) of 54.8–77.1 mmol O_2_ m^− 2^ day^− 1^^21^, assuming up to 4 ind m^− 2^^5^ and a (ex situ) respiration rate of 1.89–2.66 µmol O₂ h⁻¹ g DM⁻¹^45^ with an estimated dry weight of 300 g per sponge^20^.

The Schulz Bank sponge ground of this study thus showed slightly lower oxygen fluxes, despite higher density (9 ind m^− 2^) and biomass (8.4 kg WW m^− 2^) of sponges^34^. However, the sponge ground community at the Schulz Bank occurred at 7–9 °C lower temperatures (average 0.15 °C) than the aforementioned North-Atlantic areas, which aligns with Q_10_ temperature effects reducing metabolic rates by 50–70%^20^. Notably, the sponge-specific respiration (1.7 µmol O₂ g⁻¹ DW h⁻¹) remains within the typical range for cold-water hexactinellids (0.3-4.0 µmol O₂ g⁻¹ DW h⁻¹), suggesting physiological adaptations, such as psychrophilic enzymes in sponges or their microbial symbionts, maintain metabolic performance despite low absolute community fluxes characteristic of polar deep-sea ecosystems.

Respiration rates also fall in the same range as that of other deep-sea hotspots as for example scleractinian cold-water coral reefs of the Træna Coral Field (81.7 ± 9.8 mmol O_2_ m^− 2^ day^− 1^^40^, the Rockall Bank (between 15 and 57 mmol O_2_ m^− 2^ day^− 1^; assuming an RQ of 1^46,47)^, the Mingulay reef (28 mmol O_2_ m^− 2^ day^− 1^^48)^ or the Stjernsund reefs (25 mmol O_2_ m^− 2^ day^− 1^^49)^. These cold-water coral respiration rates are known to be 9–20 times higher than those of surrounding soft sediments and indicate strong organic matter consumption as well as recycling processes and thereby optimize resource gains over losses^49^. Respiration rates of sponge-dominated ecosystems are therefore comparable to established deep-sea biological hotspots, like cold-water coral reefs and are + 66% above the assumed empirical modelled oxygen uptake of deep-sea sediments (@580 m depth = 4.94 mmol O_2_ m^− 2^ day^− 1^, Fig. 5^50^) . The high sponge metabolic activity has important implications for carbon and nutrient cycling, and therefore benthic-pelagic coupling, potentially even on a global scale, considering sponges are increasingly found in high abundances during research expeditions in the deep sea^8,51^.

Fig. 5. Community Respiration Rates of soft sediment data (black dots^52^, as well as deep hotspots like coral reefs (blue) and sponge ground (pink/red): Rockall Bank (blue triangle^46,47^—Træna Reef (pink tringle^40^, (pink circle^44^, (blue square^40^—Mingulay Reef (blue circle^48^—Stjernsund reef (blue diamond^48^— Vazella ground (pink diamond^20,45^—Schulz Bank (red star, this study). Adapted from^40^. The solid line shows an empirical model of the double-exponential decrease with water depth, fit suggested by Andersson et al. (y = 38*[(1-0.17) e^-0,018x^+ 0.17e ^-0.00046x^]^50^, .

These high sponge ground community energy requirements need to be covered by sufficient food input to sustain a thriving ecosystem. They rely on organic matter from surface ocean productivity in the form of phytodetritus after the yearly phytoplankton bloom [53]. Additionally, sponge holobionts, possess the capacity to take up DOC, which can make up to 99% of their daily total carbon intake [20,54,55]. Deep-sea sponges are also well established as very efficient filter-feeders of small phyto- and bacterioplankton [17,20,39], even digesting their symbiotic microbes [18], or by assimilating their metabolic products [26]. The yearly vertical flux of POC supplied to the sponge ground community at the Schulz Bank has been determined in an earlier study and is on average 9.3 ± 7.6 mg C m^− 2^ day^− 1^ [35], which equals on average < 5% of the respiratory carbon demand that was established in this study. Note that, besides meeting their respiratory carbon demand, sponges require even more carbon to support anabolic processes, including growth, reproduction, chemical defence, immune responses and healing. Only after the summer phytoplankton bloom period the sponge ground community of the Schulz Bank temporarily receives higher amounts of POC for several weeks, with close to 12% of the respiratory carbon demand being covered by the vertical POC flux (28.4 mg C m^− 2^ day^− 1^ [35]) . The remainder of the year the organic carbon deficits rapidly increases. This large discrepancy between vertical flux of POC and estimated minimal respiratory carbon demand was observed in many deep-sea ecosystems, such as Vazella sponge grounds at the eastern Canadian shelf [56], the glass sponge reefs on the western Canadian shelf [17] as well as a cold-water coral ecosystems [49]. POC therefore likely represents only one component of the total food supply, with unquantified DOC and microbial contributions.

The much higher assimilation efficiency of bacterioplankton (68–97%) compared to DOC (32–77%) suggests that microbial prey provides an especially suitable and essential resources to sponges in the deep-sea, including a source of vitamins and other key metabolites [21]. The feeding of bacterioplankton in the in-situ incubations of this study could not be quantified, likely reflecting methodological limitations. No sediment-only controls were conducted, chambers remained open-bottomed during deployment, and single time-point sampling per incubation prevented robust flux calculations. Combined with low replication and variable initial procaryotic cell concentrations, the observed prokaryotic uptake in one incubation likely represents an artefact. These constraints highlight the challenges of quantifying fine-scale procaryotic grazing. Highly efficient procaryotic cell removal explains the very low concentrations of bacteria in the t_0_ water samples of less than 1.5 × 10^5^ bacteria per mL on average in all incubations [20,21]. Reducing the time after closing the incubation, and reducing the incubation time (to for example 30–60 min) might have increased the chances to determine prokaryotic uptake rates. This will likely also hold true for DOC, which was not measured in this study.

The prevailing strong hydrodynamics on the summit of the seamount will increase the delivery and replacement of prokaryotic cells and DOC, which can significantly increase fluxes toward the benthic community. These processes cannot be recorded by a closed chamber set-up but need to be connected to hydrodynamical measurements. Other technologies, such as the measurement of concentration differences throughout the water column in combination with water flow measurement (e.g. Eddy co-variance), or the use of stable isotope enriched food should be deployed to measure benthic community food uptake. In conclusion, the lack of both DOC and prokaryotic cell uptake rates in this study limits our ability to close the carbon budget for this deep-sea sponge ground, whereas main parts of their diet could potentially be fuelled by the available prokaryotic cells and DOC. Future studies should prioritize measuring these carbon sources and their fluxes in situ, while simultaneously increasing sample size and replication to strengthen statistical power and interpretation.

When particulate organic matter is degraded by heterotrophic organisms, NH₄⁺ can be released as a by-product of protein and nucleic acid metabolism. In our study, NH_4_^+^ release rates varied substantially, independent from the incubated biomass of sponges. Interestingly the two incubations with a high LMA sponge biomass showed particularly high release rate (39.9 ± 16.3 µmol m⁻² h⁻¹), exceeding those found in other deep-sea benthic systems, with high biomass of filter- or suspension feeders, such as cold-water coral reefs (3.6 to 6.2 µmol m⁻² h⁻¹ [47]) . In contrast, the single incubation dominated by HMA sponges or the three incubations with low sponge biomass had comparable lower NH₄⁺ release rates (4.63 ± 2.58 µmol m⁻² h⁻¹). These lower rates may reflect either genuinely lower excretion or rapid consumption of NH₄⁺ within sponge tissue by ammonia-oxidizing archaea (AOA) and bacteria (AOB), as shown for Geodia sp [26,29,57]. The resulting NO₂⁻ can then be further oxidized to NO₃⁻ by nitrite-oxidizing bacteria (e.g., Nitrospinae), which are common in deep-sea environments and may contribute to carbon fixation through the reverse tricarboxylic acid cycle [58,59]. Consistent with this, we observed NO₃⁻ release across all treatments, indicating active nitrification regardless of sponge biomass. The highest NO₃⁻ release occurred in incubations without large sponges (Fig. 3), which may suggest that free-living microbes in the spicule mat may play a key role, or alternatively, that sponge-associated microbes further process NO₃⁻ through denitrification, dissimilatory nitrate reduction to ammonium (DNRA), or by releasing it as N₂ gas [60]. However, denitrification and anammox likely represent minor pathways in deep-sea systems, as suggested by previous studies [60].

Our community-level approach was not designed to disentangle sponge-specific nitrogen transformations, particularly since many processes overlap between members of the sponge-ground benthic community, but was able to show that nitrogen cycling processes are markedly different between sponge types. Additionally, we have indications that sponge biomass has a major influence on the availability of organic and inorganic resources within the surrounding environment (Fig. S3). Even within sponge tissue, steep internal oxygen gradients can spatially, temporally and functionally couple aerobic and anaerobic nitrogen cycling, allowing aerobic nitrification and anaerobic processes to occur simultaneously [60–62]. This interplay highlights the heterogeneity of nitrogen cycling at the community level.

Because nitrogen cycling cannot be considered in isolation, we also examine phosphorus dynamics and N: P ratios, which together shape nutrient stoichiometry and reveal how sponge functional groups influence the broader ecosystem balance.

Our in-situ incubations showed a net release of PO₄³⁻, with a trend toward higher release when containing a higher biomass of LMA sponges. Phosphorous is an essential element required by all organisms for biological synthesis of phospholipids, DNA/RNA and energy transfer processes (ATP). Animals, including sponges, generally retain phosphorous efficiently and release excess phosphorous in form of inorganic PO_4_^3−^, if sufficient P is taken up from the particulate diet [25,63]. As a result, animal-dominated nutrient regeneration typically results in N: P ratios near or above the Redfield values. In contrast, microbial metabolism exhibits comparatively higher cellular phosphorous demand associated with rapid growth and elevated nucleic acid content, resulting in proportionally lower N: P ratios and higher phosphate regeneration during microbial turnover and remineralization [64]. Release of PO₄³⁻ by sponges was also described in other studies, including HMA and LMA sponges [15,63], reflecting the combined metabolic activity of the sponge host and its associated microbial community.

While our data set is limited to two LMA- and one HMA-dominated sponge ground community, an interesting pattern appears to emerge: The inorganic N: P ratio within our incubations that contained a high biomass of LMA sponges showed a decrease over time due to a proportionally addition of more P than N to the pool of dissolved nutrients. This may indicate that LMA sponges, which rely more strongly on particulate food sources and microbial prey, ingest and digest P-rich microbial biomass, leading to disproportionate release of inorganic phosphate. Alternatively, or additionally, elevated microbial phosphorus turnover within the sponge holobiont may contribute to enhanced phosphate regeneration or internal microbial recycling of NH₄⁺ within the LMA sponge holobiont leading to a proportionally higher release of PO₄³⁻ (Fig. S4). This pattern suggests that LMA sponges primarily act as sources of inorganic nutrients (NH₄⁺ and PO₄³⁻), whereas HMA sponges may recycle these dissolved nutrients, resulting in only a small net release, as previously modelled for Geodia barretti [29]. The stability of the N: P ratio in the presence of HMA sponges implies that they contribute to maintaining nutrient stoichiometry in the surrounding environment, counterbalancing nutrient release from LMA sponges and potentially supporting ecosystem stability.

We note, however, that interpretation of N: P and C: N ratios must be made with caution, as background remineralization by sediments or release by non-sponge fauna in the incubations could alter NH₄⁺ or PO₄³⁻ concentrations independently of sponge activity.

Across all incubations, the relative balance between carbon respiration, phosphate, and dissolved inorganic nitrogen (DIN) remained consistent (Fig. S5). This indicates that food with a similar C: N:P ratio was consumed and that the system as a whole depends on recycling of the same (limited) food sources. This complex nutrient cycling might be responsible for nutrient retention and use efficiency and prevent the accumulation of toxic intermediates (NH₄⁺, NO₂⁻). Microbial nitrogen transformations, such as nitrification and denitrification release also significant free energy (tens to hundreds of kJ per mole of substrate oxidized or reduced [57,65]), suggesting that nitrogen cycling not only regulates nutrient availability but also fuels microbial primary production and sustains host–microbe interactions in energy-limited deep-sea environments. Ambient microbial planktonic communities themselves can, in turn, be an organic nitrogen source for filter-feeding communities. In addition, sponges are known to release detritus that fuels members of their food web [3,4,23,66] showing the potential for a productive ecosystem, surrounded by resource-poor waters. All nutrient fluxes derived from our incubations should also be interpreted with caution, as the reduced sample size and limited replication weaken the statistical power and thereby constrain the robustness of our conclusions.

Although our study provides only limited in-situ data of resource cycling within a deep-sea sponge-dominated ecosystem, it contributes to a growing body of evidence that sponges play a pivotal role in deep-sea biogeochemistry. Importantly, our results highlight that nutrient fluxes are not uniform and the dominance of individual sponge species and functional groups within the community (i.e. LMA versus HMA) can act as varying sources of inorganic nutrients, and reveal the complexity of nitrogen and phosphorus cycling. Through their high filtering capacity and symbiotic microbes, sponges collectively influence benthic nutrient dynamics and even affect microbial plankton communities hundreds of meters above sponge ground [67], with potential consequences for local and global nutrient budgets. It is important to note that part of these effects observed in our incubations can also be attributed to the associated benthic community, including microbes, meiofauna, and other small organisms, capturing the integrated activity of the entire community. Unfortunately, we were not able to disentangle these components since we concentrated this study on the community dominated by sponges.

Looking ahead, the resilience of sponge-mediated nutrient cycling faces multiple pressures: Climate-driven warming, acidification and deoxygenation have been shown in shallow-water sponges to alter pumping rates, filtration efficiency, respiration and holobiont nutrient fluxes, as well as restructuring microbial communities and stimulate dysbiosis and shifts in nitrogen pathways such as enhanced denitrification (e.g. [68,69]). Experimental warming of cold‑water and deep‑sea sponges similarly increases respiration and nitrogen efflux, indicating that even modest temperature rises can substantially modify sponge‑driven carbon and nitrogen cycling [70]. In addition to these drivers, projected changes in the North Atlantic Meridional Overturning Circulation are expected to shift key water mass interfaces and modify flow regimes along the Mid‑Atlantic Ridge [71], with likely consequences for boundary‑layer hydrodynamics and lateral particle transport. Such changes could reduce the delivery of suspended particulate and dissolved organic matter to Schulz Bank, potentially lowering the food supply sustaining the sponge ground and thereby weakening sponge‑mediated nutrient cycling at this site. Superimposed on these physiological sensitivities, deep‑sea mining and other seafloor disturbances threaten to remove or fragment sponge grounds entirely [72,73]. The loss of sponge grounds would likely reduce nutrient and organic matter transfer across benthic and pelagic food webs, with potentially cascading impacts on higher trophic levels, including commercially important fish. Given the slow growth and limited recovery of many deep-sea sponge species, incorporating sponge grounds into marine biogeochemical models and conservation strategies is essential to safeguard their role in global ocean processes.

Conclusion

The Schulz Bank sponge ground community represents a hotspot of carbon and inorganic nutrient cycling in the deep sea, with respiration rates comparable to other deep-sea hotspot ecosystems despite lower ambient temperatures. Although these results must be interpreted with caution due to methodological constraints, our data suggest that sponges are the primary drivers of local carbon and inorganic nutrient fluxes, with distinct contributions from HMA and LMA sponges to different aspects of the nitrogen cycle, as well as interactions with the surrounding environment.

By facilitating the transfer of energy and nutrients between the water column and the seafloor, sponge grounds, such as those at Schulz Bank, not only shape local ecosystem productivity, but also influence broader biogeochemical processes. The high metabolic activity and nutrient cycling of the Schulz Bank sponge ground highlight its importance as a deep-sea carbon and nutrient cycling hotspot. This underscores the need to incorporate these complex ecosystems into marine biogeochemical models and conservation strategies, as their loss or alteration could have cascading effects on biodiversity, nutrient dynamics, and the resilience of deep-sea environments.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1