Macrobenthic assemblage structure and organismal stoichiometry control faunal processing of particulate organic carbon and nitrogen in oxygen minimum zone sediments

. The Arabian Sea oxygen minimum zone (OMZ) impinges on the western Indian continental margin between 150 and 1500 m, causing gradients in oxygen availability and sediment geochemistry at the sea ﬂoor. Oxygen availability and sediment geochemistry are important factors struc-turing macrofaunal assemblages in marine sediments. How-ever, relationships between macrofaunal assemblage structure and sea-ﬂoor carbon and nitrogen cycling are poorly un-derstood. We conducted in situ 13 C: 15 N tracer experiments in the OMZ core (540 m [O 2 ] = 0.35 µmol l − 1 ) and lower OMZ boundary (800–1100 m, [O 2 ] = 2.2–15.0 µmol l − 1 ) to investigate how macrofaunal assemblage structure, affected by different oxygen levels, and C:N coupling inﬂuence


Introduction
Episodic depositions of dead phytoplankton (phytodetritus) represent a major input of particulate organic matter (POM) to the deep-sea floor, stimulating the feeding and reproductive responses of the faunal and microbial assemblages that control sea-floor carbon cycling (reviewed by Smith et al., 2010). The bathyal continental margins (200-3000 m) account for a mere 7 % of the global sea floor but recycle ∼30 % of oceanic sedimentary POM (Middelburg et al., 1997). Oxygen minimum zones (OMZs) are stable bodies of poorly oxygenated water (<22 µmol l −1 ) persisting along the continental margins of the eastern Pacific Ocean, Arabian Sea and southwest Africa (Levin, 2003). OMZs are maintained by high primary productivity at the oceans surface and poor advective mixing of the water column, resulting in high organic matter flux and biological oxygen demand within the water column; and accumulation of organic matter in OMZimpacted sediments (Devol and Hartnett, 2001;Hartnett et al., 1998). OMZs currently impinge upon ∼ 6 % of the continental margin sea floor (Helly and Levin, 2004) but are predicted to expand as consequences of anthropogenic climate change and ecosystem degradation (Bakun and Weeks, 2004;Stramma et al., 2008).
Metazoan macrofauna (size range: 250-1000 µm) are important contributors to ecosystem processes in deep sea sediments. Macrofauna contribute directly to OM recycling through ingestion, assimilation and respiration (e.g. Aberle and Witte, 2003;Witte et al., 2003a). Indirectly, macrofaunal subduction of POM provides an important route for deposition of organic matter deeper into the sediment (e.g. Levin et al., , 1999, whilst macrofaunal bioturbation enhances sediment oxygenation; stimulates aerobic metabolism by the sediment community (e.g. Kristensen and Holmer, 2001); and provides microhabitats for microbes and meiofauna (reviewed by Giblin et al., 1995). Macrofaunal also influence sediment microorganisms through predation and competition for resources (e.g. Aller, 1994;Wieltschnig et al., 2008;Hunter et al., 2012) . At OMZ-impacted continental margins the macrofaunal assemblages are influenced by depth-dependent changes in ambient oxygen availability and differences in organic matter (OM) availability within the sediments (e.g. Levin and Gage, 1998). Macrofauna are often absent in the OMZ core, where oxygen levels are lowest; whilst high abundance, low diversity assemblages are typical at the OMZ boundaries. By contrast, seafloor outside the OMZ's influence are characterised by low-density, macrofaunal assemblages (Levin et al., 1991(Levin et al., , 2000Ingole et al., 2010). High macrofaunal abundances in the OMZ boundaries are driven primarily by hypoxia-tolerant polychaete genera, such as the ampheretid Linopherus spp., the spionid Prionospio spp. or the cirratulid Monticellina spp. (e.g. Levin and Edesa, 1997;Levin et al., 2000;2009;Ingole et al., 2010). Macrofaunal diversity is positively correlated to oxygen availability, increasing concomitantly with decreases in macrofauna abundance across the OMZ boundary (e.g. Levin et al., 2000Levin et al., , 2009. Quantifying the role of the fauna within ecosystem processes is important in order to develop accurate models of sea floor carbon and nitrogen cycling. Stable-isotope labelling techniques allow pathways and rates of OM processing to be traced over short time periods. So far, these methods have primarily been used to trace the pathways and processing of organic carbon in deep-sea sediments, using 13 C labelled phytodetritus. Following deposition of labelled phytodetritus, macrofauna rapidly ingest the 13 C label and are responsible for its subduction below the sediment surface (e.g. Aberle and Witte, 2003;Witte et al., 2003a). Macrofauna play an important role in the initial community response, both directly consuming OM and mediating its availability to other organismal groups (e.g. Witte et al., 2003b;Moodley et al., 2005). At the OMZ-impacted Pakistan margin, macrofauna significantly influenced the fate of OM within the sediments. The polychaete Linopherus spp., which occurred in high densities, consumed between 45-80 % of the isotopically-labelled OM deposited between 850 and 1800 m, in the OMZ lower boundary (Woulds et al., 2007). However, the relationship between macrofaunal assemblage structure and OM processing remain largely unexplored.
Organic nitrogen availability limits secondary production in marine sediments and may limit carbon cycling at the seabed (Vitousek and Howarth, 1991). Dual-labelling ( 13 C: 15 N) studies simultaneously traced the utilisation of organic C & N in intertidal (Rossi, 2007) and shallow subtidal sediments (Evrard et al., 2010), demonstrating that faunal and microbial N demand are important drivers of OM recycling. However, C & N are physiologically decoupled by individual organisms, which use C-rich (e.g. carbohydrates; lipids) and N-rich (e.g. amino acids) organic compounds differently: for respiration and energy storage; and growth respectively (e.g. Frost et al., 2002;Persson et al., 2010). Faunal processing of particulate organic carbon (POC) and nitrogen (PON) are determined by the internal budgets of individual organisms, controlled by changes in ontogeny and physiology (reviewed in Sterner and Elser, 2002). Therefore, in order to understand macrofaunal contributions to benthic carbon cycling, we must investigate the relationships between macrofaunal carbon and nitrogen processing.
The aim of the present study is to exploit oxygen-driven gradients in macrofaunal assemblage composition across the lower boundary of an OMZ to explore relationships between macrofaunal assemblage structure and short term processing of phytodetritus-derived carbon and nitrogen. Pulses of 13 C: 15 N labelled diatoms simulated a phytodetritus deposition event, and assimilation of phytodetrital carbon and nitrogen by the macrofauna were traced over four and seven days. We hypothesise that oxygen limitation will modify macrofaunal assemblage composition, influencing faunal POC and PON processing. Variations in organismal C:N ratios reflect differences in POC and PON demand within the fauna. Therefore, we predict that differential POC and PON assimilation will reflect the organismal stoichiometric budgets within each macrofaunal assemblage.

Study area
The Arabian Sea OMZ impinges upon the western Indian continental margin between 150 and 1500 m. Ambient oxygen levels at the sea floor fall as low as 0.35 µmol l −1 at 540 m, increasing to 2.2-2.36 µmol l −1 at 800 m and 15.00 µmol l −1 at 1100 m (Hunter et al., 2011). Stableisotope labelling experiments were conducted at four stations on the Indian continental margin, ( Fig. 1; Table 1). These span differing oxygen regimes within the OMZ core (540 m) and lower OMZ boundary (800-1100 m). All stations were visited during the 2008 post-monsoon period (September-November) as part of R/V Yokosuka cruise YK08-11.

Experimental design
Prior to YK08-11, an axenic clone of Thallassiosira weissflogii (CCMP, Bigelow marine Laboratories) was cultured in artificial seawater and L1 culture medium enriched with 99 % 13 C-bicarbonate (NaH 13 CO 3 , Cambridge Isotope Laboratories) and 50 % 15 N-sodium nitrate (Na 15 NO 3 , Cambridge Isotope Laboratories). Algae were cultured at 16 • C (light: dark = 16:8; salinity = 35; duration = 28 Table 1. Environmental conditions. Oxygen, Temperature and Salinity data are as reported in Hunter et al., (2011). Porosity values represent the means (±SD), between 00 and 05 cm sediment depth, of three background sediment cores at each station. Sediment Organic matter and Unprocessed Algal material data are presented as the mean values (±SD) for the upper 5 cm sediment, within each spreader experiment. Experiments were conducted using Oceanlab spreader systems. These are in situ meso-cosms consisting of a transparent polycarbonate tube (diameter: 25 cm, length: 30 cm) and acetal plastic lid. The lid of each spreader contains an injector unit that releases a known dose of isotopically-labelled phytodetritus onto the enclosed sediment surface (0.049 m 2 ). Spreaders were deployed by the submersible Shinkai 6500, which gently pushed each tube into the sediment. A marinegrade plywood collar, approx. 10 cm above the base of the spreader tube, prevented it being pushed too far into the sediment and provided additional stability. Labelled phytodetritus was released by pushing an elastically-tensioned plunger, rupturing the membrane of the injector unit. The spreader lid was left in place for a minimum of two hours, to allow the phytodetritus to settle, and then removed to mitigate against experimentally-induced hypoxia (Riedel et al., 2008). The polycarbonate tube was left in situ for the duration of the experiment. At the end of each experiment the sediment enclosed within a spreader was sampled using push cores and the polycarbonate tube recovered.
Three spreaders were deployed at each station for four day incubations, using the submersible Shinkai 6500. Three replicate spreaders were also deployed for seven day incubations at stations T2 800 m and T2 1100 m (Table 1). Spreaders were positioned approximately 2 m apart on areas of undisturbed sediment and placed into the sediment with minimal resuspension. Experiments commenced with the deposition of a fixed dose of 13 C: 15 N labelled T. Weissflogii slurry (650 mg C m −2 ; 160 mg N m −2 ). Spreaders were left in situ for four or seven days, after which each was sampled with three push cores. Cores were horizontally sectioned in 1 cm intervals to 3 cm depth, then in 2 cm intervals to 7 cm depth, and a 3 cm fraction between 7 and 10 cm. One core from each spreader was allocated for macrofaunal extraction, one core was allocated for extraction of foraminifera and meiofaunal, and one core was used for microbiological and geochemical analysis. Loss of cores during sampling resulted in the recovery of only two replicate 4 day incubations at station T2 1100 m and two replicates from each 7 day incubation (T2 800 m and T2 1100 m). 25 additional background cores were taken to provide natural stable isotope signatures of the fauna (13 between 800 and 1100 m) and sediments (3 at each station).

Sample processing and data treatment
The upper 5 cm of sediment from each macrofauna core was wet-sieved through a 250 µm mesh, using filtered sea water. Sieve residues were picked for macrofauna (under x12 and x20 magnification) and transferred to 4 % buffered formaldehyde solution. Macrofauna were sorted to family (and genus where possible), rinsed in ultrapure water and dried to constant weight at 60 • C. All fauna were decarbonated by addition of 1-2 drops of analytical grade 6 mol l −1 hydrochloric acid (HCl aq ). Sediment samples were centrifuged to extract pore water, homogenized and frozen immediately at −80 • C. Post-cruise, frozen sediments were sub-sampled and 3 g aliquots lyophilised (−60 • C; −0.0001 mbar; 24 hrs). Lyophilised sediments were decarbonated by addition of excess 1 mol l −1 HCl aq , incubated for 24 hrs at 30 • C in an acidfumed environment and dried to constant weight at 60 • C. Mean sediment porosity was estimated from the change in mass and volume during lyophilisation (following Hunter et al., 2011).
Organic carbon and nitrogen concentrations and the isotopic ratios ( 13 C/ 12 C and 15 N/ 14 N) of fauna and sediment were determined using a PDZ Europa ANCA-GSL elemental analyser linked to a PDZ Europa 20-20 IRMS (Sercon Ltd, Cheshire UK). Samples were combusted at 1000 • C with helium (He) used as the carrier gas. Fauna were analysed, with the IRMS and internal standards adapted for low carbon samples. Isotope ratio data, expressed in δ units (‰), was used to calculate the atom % 13 C and 15 N by where at %X sample is the 13 C or 15 N content of the sample (in atom %)X sample is the isotopic ratio ( 13 C or 15 N) of the sample (in δ units) and R standard is an international reference material: Vienna PeeDee Belemnite for C (R VPDB = 0.0112372); atmospheric nitrogen for N (R atmN = 0.0036765). Faunal 13 C values were corrected for formaldehyde preservation effects by adding 1 ‰ to each δ 13 C-value (following Sweetman and Witte 2008a, b;Gontikaki et al., 2011). Excess 13 C and 15 N concentrations in each sample were calculated following Middelburg et al., (2000) as where E is the excess isotopic label, at %X sample is the atom % 13 C or 15 N of the sample, and at %X background is the atom % 13 C or 15 N of the background material. Background 13 C and 15 N values were determined from the natural isotopic signatures of macrofauna sampled at the Indian margin. Incorporation of the 13 C and 15 N was calculated as the product of isotopic label excess and the faunal carbon or nitrogen content. Assimilation of phytodetrital carbon ( phyto C) and nitrogen ( phyto N) was calculated by ( 3) where I phyto is the concentration of phyto C or phyto N, I iso is the amount of 13 C or 15 N incorporated, and at %X phyto is the isotopic content ( 13 C or 15 N) of the phytodetrital substrate, in atom %. Sediment C and N concentrations, and isotopic data where determined as described for the fauna, against the natural isotopic signatures of background sediment samples. Sediment carbon and nitrogen content were used to calculate the molar C:N ratio, as a proxy for sediment organic matter quality (Hedges and Keil 1995) and the percentage of phyto C and phyto N recovered in the sediment samples estimated for each experiment.

Statistical analysis
Environmental data for each station (Table 1) shows that alongside depth-dependent oxygen gradients, localised variations in sediment organic carbon and nitrogen content occur both within and between experimental stations. A scatterplot matrix (Fig. A1) displays the relationships between environmental variables and quantifies correlations using Spearman's rank correlation co-efficient. This matrix shows strong correlations between oxygen availability and most of the environmental variables (Temperature, salinity, POC, PON AlgC, AlgN). Weak correlations were observed between oxygen availability and both sediment porosity and sediment C:N ratios, but these variables were co-linear. Only ambient oxygen availability and sediment C:N ratio, were used as descriptors of each station during analysis. Macrofaunal assemblage structure (C & N biomass) and macrofaunal assimilation of phyto C and phyto N where modelled as responses to changes in sediment C:N ratios and oxygen availability, between stations by Constrained Analysis of Principle Components (CAPSCALE) (Anderson and Willis, 2003). T1 540 m was excluded from analyses because macrofauna were absent at this station. The solitary ascidian at T2 1100 m was excluded from these analyses as an extreme outlier. Biomass C and biomass N were transformed by √ (x + 0.01) to reduce the influence of highly abundant taxa. phyto C and phyto N assimilation data were transformed by 3 √ (x + 0.01), to control for the effects of differences in incubation time at T2 800 m and T2 1100 m and reduce the influence of highly abundant taxa. Ordination models were constructed from Bray-Curtis dissimilarity matrices, with the multivariate normality of each matrix tested using beta-dispersion tests (1000 permutations, p < 0.05) (Anderson, 2006). Model selection was carried out by minimizing the Aikaike Information Criterion (Bedrick and Tsai, 1994). Significance of the final ordination models were tested by permutational analysis of variance (1000 permutations, p < 0.05) (Anderson and Willis, 2003). Multivariate correlations between changes in phyto C and phyto N assimilation to changes in macrofaunal assemblage structure (biomass C and biomass N) were tested by Procrustes rotational tests (PROTEST) between final ordination models. This method compares data matrices using a rotational-fit algorithm to minimise their sum-of square residuals and calculate a goodness-of-fit statistic (m 2 ). The significance of the m 2 statistic was tested using a randomisation test (1000 permutations; p < 0.05) (Jackson, 1995). All analyses were carried out in R2.9.2 (R Development Core Team 2009) using VEGAN (Oksanen et al., 2009) and MASS (Venables and Ripley, 2002) packages.

Macrofaunal assemblage
Description of the macrofauna at each station is based upon both four and seven day experiments, with each spreader treated as a replicate. Sample sizes were n = 3 at stations T1 540 m and T1 800 m, n = 5 at T2 800 m and n = 4 at T2 1100 m metazoan macrofauna were absent at T1 540 m, where oxygen levels were ∼0.36 µmol l −1 . Between 800 and 1100 m, total abundance, biomass C and biomass N were higher at T1 800 m and T2 800 m, compared with T2 1100 m (Fig. 2). In the present study the definition of macrofauna included large nematodes (>250 µm), which contributed ∼30 % of macrofaunal abundance at station T2 800 m, and ∼60 % at T2 1100 m. Nematodes contributed little C or N biomass (Fig. 3). The polychaetes were the dominant macrofaunal taxa, by biomass. However, at T2 1100 m the presence of s single large ascidian in one spreader contributed ∼45 % of C biomass and ∼20 % of N biomass.
Cirratulids and sabellids were the most abundant polychaete families at T1 800 m and oweniids and cirratulids were abundant at T2 800 m and T2 1100 m. Cirratulids and oweniids contributed most to polychaete C & N biomass. The cirratulids were characterised by three genera: Cirratulus spp.; Tharyx spp.; and the mud-ball forming Monticellina spp. (Levin and Edesa 1997), and the oweniids represented by the genus Owenia spp. According to Fauchald and Jumars' (1979) polychaete feeding group classifications, stations T1 800 m, T2 800 m and T2 1100 m were dominated by surface deposit feeding polychaetes, with sub-surface deposit feeding families playing only a minor role. Natural abundance stable isotope signatures of the macrofauna are summarised in Fig. 4. The data exhibits considerably variation across all taxa. A harpactacoid copepod exhibited the most depleted δ 13 C signature (−26 ‰) and an ophiuroid displayed the heaviest δ 13 C signature (−11 ‰). δ 15 N signatures typically ranged from 2 to 12 ‰. However, harpactacoid copepods, nyphtid polychaetes and bivalves all exhibited depleted δ 15 N (<2 ‰). Heavier δ 15 N signatures (>12 ‰) were observed in the turbellaria, as well as phoxocephalid crustacea and goniadid polychaetes. The natural isotopic signatures provide background data for the quantification of the 13 C and 15 N enrichments of macrofauna in each spreader experiment.

Macrofaunal uptake of phytodetrital C and N
Between 800 and 1100 m, mean phyto C and phyto N processed by the metazoan macrofauna ranged from 5540.46 (±2482.74) to 1511.66 (±1862.28) µg C m −2 and 355.50 (±334.82) to 21.96 (±20.04) µg N m −2 , after 4 days, and from 2150.56 (±1463.19) to 373.79 (±414.89) µg C m −2 and 97.41 (±84.42) to 31.64 (±5.44) µg N m −2 , after 7 days. Between 4 and 7 days decreases in both phyto C and phyto N assimilation were observed at T2 800 and T2 1100 m. Assimilation of phyto C and phyto N was dominated by the cirratulids at both 800 m stations (Fig. 5). However, at T2 800 m, there is a shift in phyto C and phyto N assimilation between four and seven days from the cirratulids to other taxonomic groups. At T2 1100 m, differences in phyto C and phyto N assimilation were generally small between taxa. However, the solitary ascidian accounted for ∼ 76 % of the faunal biomass and ∼ 75 % of phyto C assimilation in one of the four day experiments at this station.
Biomass-specific phyto C and phyto N assimilation was relatively constant between macrofaunal taxa (Fig. 6). During the 4-day incubations, cirratulid polychaetes exhibited higher biomass specific assimilation at both T1 800 m and T2 800 m. However, greatest biomass-specific assimilation of both phyto C and phyto N were observed in a eunicid polychaete at T2 800 m. Likewise, at T2 1100 m, relatively high biomass specific assimilation of phyto C and phyto N were observed in a solitary flabelligerid polychaete. Fewer macrofauna were recovered from the 7-day experiments and biomass-specific phyto C and phyto N assimilation exhibited no taxon-specific changes between four and seven days.

Multivariate ordination models
CAPSCALE ordination models tested the significance of environmental changes upon macrofaunal assemblage structure and phyto C and phyto N assimilation, between stations (Fig. 7). The influence of sediment C:N ratios upon assemblage structure (biomass C and biomass N) and feeding responses ( phyto C and phyto N assimilation) could not be isolated  from stochastic variations in the data and were excluded during model selection (Supplement data). CAPSCALE ordinations demonstrated significant relationships between ambient oxygen availability and macrofaunal assemblage structure (Fig. 7a, b). In addition, spatial variability within the ordination models indicates unconstrained heterogeneity in macrofaunal assemblage structure at all three stations. CAPSCALE ordinations showed that changes in oxygen availability to have significant effects upon macrofauna phyto C & phyto N assimilation (Fig. 7c, d). Unconstrained spatial variability was again a feature of the ordinations, suggesting that phyto C and phyto N assimilation by the macrofaunal assemblage is influenced by assemblage structure. Procrustes rotational analysis confirms this by demonstrating significant correlations between macrofaunal assemblage structure (biomass C and biomass N) and both phyto C assimilation (m 2 = 0.896; p < 0.001) and phyto N assimilation (m 2 = 0.989; p < 0.001).

Relationships between C & N uptake
Relationships between macrofauna phyto C and phyto N assimilation were explored by calculation of C:N ratios for the somatic tissues of macrofaunal specimens, their total phyto C and phyto N assimilation and biomass specific assimilation of phyto C and phyto N at each station (Fig. 8). Somatic C:N ratios of the macrofauna ranged from 5 to 20, with outliers between 30 and 40 observed in the nematodes; cirratulid and sabellid polychaetes; and a pelecypod mollusc (Fig. 8a) The solitary ascidian was the most extreme outlier with a somatic C:N ratio of 70. Macrofaunal phyto C: phyto N assimilation ratios were variable between taxa (Fig. 8b). However, most values ranged between 10 and 60 indicating preferential  assimilation of phyto C, relative to the C:N ratio of the labelled phytodetritus (4.04). phyto C: phyto N ratios between 100 and 250 suggest a strong carbon demand by the nematodes; eunicid, sabellid and spionid polychaetes; pelecypod molluscs; and the ascidian. Biomass specific phyto C: phyto N assimilation ratios exhibit considerably less variation, with most values ranging between 2 and 5 (Fig. 8c). These values are close to the C:N ratio of the labelled phytodetritus, suggesting preferential assimilation of phyto N, relative to the somatic C:N ratio of each specimen.

Macrofaunal assemblage
Changes in macrofaunal abundance and biomass were observed across the OMZ-impacted Indian margin. macrofauna were absent in the OMZ core (T1 540 m; [O 2 ] = 0.35 µmol l −1 ), whilst in the lower OMZ boundary (800-1100 m), highest faunal density and biomass were found at 800 m. Subsequently, both macrofauna abundance and biomass decreased between 800 and 1100 m. Previous estimates of macrofaunal abundance and biomass at the western Indian continental margin are lower than those reported in the present study, with Ingole et al., (2010) reporting densities of 424 to 707 m −2 , between 525 and 1524 m. The present study is consistent with macrofaunal abundance estimates between 850 and 1100 m depth, at the OMZ-impacted Pakistan and Oman margins (Levin et al., 2000Hughes et al., 2009). macrofaunal abundances were greater in the present study than those observed in non-impacted bathyal sediments, such as western Norwegian fjords ( Witte, 2000). Thus, the Indian margin macrofaunal assemblage is characterised by high densities of small organisms. This supports Levin's (2003) hypothesis that small-bodied organisms, whose bodies have a greater surface area for gas-exchange, are more prevalent under conditions of persistent dysoxia (low oxygen). An oxygen threshold between 540 and 800 m ([O 2 ] = 0.35-2.20 µmol l −1 ) limited penetration of macrofauna into the OMZ core. This is consistent with observations made at the Pakistan margin, where macrofaunal presence and activity were limited by an oxygen threshold at ∼700 m depth Woulds et al., 2007). For example on the Pakistan margin, the amphinomid Linopherus spp. dominated the macrofauna between 700 and 1100 m , and the spionid Prionospio spp. was observed in high densities in OMZ impacted regions of the Indian and Oman continental margin (Ingole et al., 2010;Levin et al., 2000). In the present study, macrofaunal assemblages were dominated by two polychaete families, the cirratulids (Cirratulus spp.; Tharyx spp. monticellina spp.) and the oweniids (Owenia spp.) between 800 and 1100 m. This contrasts with Ingole et al. (2010) study at the Indian margin, highlighting the influence of spatial and temporal heterogeneity upon benthic macrofaunal assemblages. Within OMZs, zonation of the macrofaunal assemblages is associated with taxon-specific differences in tolerance to dysoxia, causing the abundance of annelids, molluscs, crustacea and echinoderms to increase sequentially (Levin et al., 2000;Levin et al., 2009). The spatial-scale of present study is too small to observe this zonation. However, family-level changes in macrofaunal distribution occured across the lower OMZ boundary exhibiting a significant relationship to differences in oxygen availability. Previously, changes in benthic macrofaunal assemblages have been linked to changes in both oxygen availability and sediment geochemistry, on OMZ-impacted margins (Levin and Gage, 1998;Levin et al., 2009). The present study could not isolate the effects of sediment geochemistry upon macrofaunal assemblage structure from the spatial heterogeneity at each station. Nevertheless, it is feasible that small-scale changes in sediment geochemical parameters, at each station, may be an important factor driving the spatial variability of the macrofaunal assemblages.

Linking Assemblage structure to macrofaunal feeding responses
Traditionally, the feeding ecology of deep-sea macrofauna has been investigated using specimen morphology and comparison with the feeding modes of shallow-water analogues (e.g. Jumars et al., 1990;Pagliosa, 2005). The use of both natural stable-isotopic signatures and stable isotope-labelling experiments provide powerful tools to test these assumptions and trace the flow of carbon and nitrogen through macrofaunal assemblages (Aberle and Witte 2003;Witte et al., 2003a;Witte 2008a, 2008b;Gontikaki et al., 2011). In the present study, the macrofaunal assemblages at both 800 and 1100 m were dominated by polychaetes, such as cirratulids and oweniids, defined as surface-deposit feeders by Fauchald and Jumars (1979). Natural stable-isotopic signatures of macrofauna at the Indian margin support this assumption, with faunal δ 13 C and δ 15 N signatures (Fig. 4) similar to the natural δ 13 C and δ 15 N values of sediment OM at each station (Fig. B1). Natural abundance δ 13 C values were similar to those of marine phytoplankton (Fry and Sherr 1984) indicating the importance of phytodetritus as an OM source for these macrofauna. The natural δ 15 N signatures of most macrofauna were relatively depleted compared to those found in oxygenated bathyal sediments (Sweetman and Witte 2008a;Gontikaki et al., 2011). This indicates that macrofauna at the Indian margin are reliant on small low-quality POM particles as a food source (e.g. Rau et al., 1990). Relatively high variability in the δ 15 N signatures of the nephytid polychaetes and other groups may reflect variations in POM quality within the sediments, whilst heavier δ 15 N signatures in the phoxocephalid crustaceans and goniadid polychaetes is indicative of omnivorous or predatory feeding modes ( Vanderklift and Ponsard, 2003).
The present study compares macrofaunal feeding responses across the OMZ-impacted Indian margin, tracing the uptake of 13 C: 15 N labelled phytodetritus doses (650 mg C m −2 ; 160 mg N m −2 ) at four stations. Feeding responses were observed at the 800 and 1100 m stations, with phyto C and phyto N uptake highest at 800 m, where macrofaunal biomass was greatest. Macrofaunal uptake decreased between four and seven days, similar to the trend observed by Witte et al., (2003b) at the Porcupine Abyssal Plain. However, in the present study decreases in assimilation of phyto C and phyto N may be primarily driven by differences in faunal biomass between experimental incubations. Macrofaunal feeding responses were strongly correlated with changes in macrofaunal assemblage structure, exhibiting significant relationships to station-specific oxygen concentrations between 800 and 1100 m. Thus, oxygen availability may drive changes in macrofaunal assemblage structure that control POC and PON processing across the lower OMZ boundary.
It is hypothesised that family-level differences in trophic ecology and individual variation in feeding behaviour influence the feeding responses of deep-sea macrofaunal assemblages (Witte et al., 2003a;Sweetman and Witte 2008a, b). In deep sea sediments, bulk OM uptake by the macrofauna is primarily driven by the feeding responses of the dominant taxonomic groups to POM sedimentation. In particular, deposit feeding polychaetes dominate macrofaunal uptake of POM at the bathyal continental margins (Sweetman and Witte, 2008a;Gontikaki et al., 2011,). In the present study phyto C and phyto N assimilation were dominated by cirratulid polychaetes at the 800 m, whilst a solitary large ascidian dominated assimilation of phyto C in one experiment at 1100 m. This is consistent with observations that consumption of POM is proportional to consumer biomass (e.g. Middelburg et al., 2000;Woulds et al., 2007).
The present study demonstrates that within OMZimpacted sediments, changes in macrofaunal assemblage structure influence seafloor OM recycling. This confirms the hypothesis that variations in macrofaunal organic matter processing are driven by macrofaunal assemblage structure (Witte et al., 2003a;Sweetman and Witte, 2008a). In the present study, most macrofaunal taxa exhibited relatively constant biomass specific assimilation of phyto C and phyto N, with the exception of cirratulid, eunicid and flabelligerid polychaetes. These taxa are characterised by Fauchald and Jumars (1979) as motile surface-deposit feeders, suggesting they can make at least limited self-directed movements to exploit patchy food resources. This may facilitate higher biomass-specific phyto C and phyto N assimilation and, as a result, the numerically abundant cirratulids dominated assemblage-level POC and PON processing. These cirratulids fulfilled an ecological role at the Indian margin that was similar to that of the highly active amphinomid Linopherus spp. on the OMZ-impacted Pakistan margin (Woulds et al., 2007). Direct comparison between the present study and previous research is challenging because of differences in sample size, incubation time and phytodetrital dose added. However, the biomass-specific phyto C assimilation values indicate that feeding activity by the macrofaunal assemblages is higher on the OMZ-impacted Indian margin compared with oxygenated bathyal (Witte et al., 2003a;Sweetman and Witte, 2008a;Gontikaki et al., 2011) and abyssal (Aberle and Witte, 2003;Sweetman and Witte, 2008b) sediments. Therefore, our results support the hypothesis that OMZ boundary sediments are sites of enhanced macrofaunal activity (Woulds et al., 2009).

C:N coupling of macrofaunal feeding responses
In deep-sea sediments faunal feeding responses are calibrated by the quantity and relative quality of organic matter supplied (Aspetsberger et al., 2007;Buhring et al., 2006). However, the mechanisms driving these responses are poorly studied. The availability of organic matter for biological utilisation (lability) is strongly correlated to its relative organic nitrogen content (Cowie et al., 1999;Hedges and Keil, 1995). The relationship between carbon and nitrogen assimilation provides a useful framework to explore how organismal feeding ecology influences ecosystem-scale carbon and nitrogen cycling (following Sterner and Elser 2002).
The present study is the first to simultaneously trace the uptake of POC and PON by deep-sea macrofaunal assemblages. Previously, dual-labelling ( 13 C: 15 N) experiments have demonstrated the importance of faunal C:N coupling in coastal and estuarine sediments. Evrard et al., (2010) indicated a strong preference for nitrogen in "meio-" faunal organisms (63-1000 µm) following organic matter deposition in a coastal sublittoral sediment, a size fraction which encompasses the common definition of deep-sea macrofauna (>250 µm). Likewise, Rossi et al., (2007) demonstrated organic nitrogen to have reduced residence time within estuarine sediments compared with organic C, with macrofauna preferentially consuming 25 % of the 15 N label from a phytodetrital source versus just 3 % of the 13 C label. This contrasts with the present study, where macrofaunal assimilation of phyto C and phyto N accounted for ∼ 1 % and ∼ 0.4 % respectively of the phytodetrital dose. These results indicate that overall faunal C:N coupling was driven by preferential assimilation of phyto C at the assemblage level. This is a surprising observation, given that nitrogen is a limiting nutrient in marine sediments.
The mechanisms that drive faunal uptake of POM act at the organismal level, balancing demands for organic carbon (as an energy source) and nitrogen (for somatic growth and reproduction) against the need to detoxify and excrete nitrogenous waste products (Raubenheimer and Simpson, 2004). In the present study, family-level C:N coupling indicated preferential uptake of phyto C by the fauna (Fig. 8a). The somatic tissues of both marine and freshwater benthic invertebrates are protein-rich, with C:N ratios ranging between 1.5 and 9 ( e.g. Liess and Hillebrand, 2005;Clarke 2008). However, At the Indian margin macrofaunal C:N ratios exhibited greater variability, ranging between 5 and 20. Whilst the presence of skeletal material (e.g. carbonate) can result in high C:N ratios, all specimens were washed in ultra-pure water and de-carbonated prior to analysis. The highest somatic C:N ratio was recorded in the solitary ascidian as a consequence of its tunicin (carbohydrate) test (Goodbody 1974). Thus, availability of organic carbon is likely to represent an important limiting nutrient for this organism. Overall, higher somatic C:N ratios may indicate accumulation of lactate within faunal tissues. Lactate is an end product of anaerobic metabolism whose accumulation in animal tissue indicates persistent oxygen-stress (Seibel, 2011). Subsequently, we hypothesise that modest energy yield from anaerobic metabolism increases faunal demand for organic C, but limits its mineralisation to dissolved inorganic carbon, resulting in the observed preferential assimilation of phyto C.
Animals are characterised by stoichiometric homeostasis, adapting their feeding responses to maintain nutrient consumption at an optimum level (Frost et al., 2002(Frost et al., , 2005. The phyto C: phyto N ratio for biomass-specific uptake (Fig. 8c) supports the proposed hypothesis. These data show relationships between phyto C and phyto N assimilation were similar across the major taxa, driven by a strong N preference. This indicates that macrofaunal feeding maximises organic N uptake relative to an internal nutrient budget, approximated by individual somatic C:N ratios. Macrofauna contribute to OM remineralisation through respiration (DIC) and ammonification (DIN), and so assimilation of phyto C and phyto N into their tissues accounts for only a fraction of their activity. In marine sediments macrofaunal are ammonotelic, mineralising organic N directly to ammonia which is rapidly excreted (Wright 1995). Therefore, much of the phyto N processed by fauna is excreted, whilst the low-oxygen conditions resulted in phyto C accumulation within tissues. In a complimentary study, Hunter et al., (2012) observed microbial activity was limited by the presence of fauna across the Indian margin OMZ. Although faunal assimilation of phyto C and phyto N were relatively small, they were concomitant with losses of labile OM and phyto N from the sediments. This indicates that faunal activity is an important control upon C and N cycling in low-oxygen sediments.
Within the lower boundary of the Indian margin OMZ, changes in oxygen availability influence the composition of the macrofaunal assemblages. Differences in short term POC and PON processing were strongly correlated to oxygendriven changes in macrofaunal assemblage structure. At present there is a lack of data on macrofaunal feeding behaviour, with much emphasis placed upon the role of broad functional groups (e.g. Fauchald and Jumars, 1979;Pagliosa, 2005). The present study is the first to use stoichiometric ratios of C and N to investigate macrofaunal feeding responses, identifying the importance of individual nutrient budgets as drivers of the macrofaunal pathways for organic matter recycling. Supplementary material related to this article is available online at: http://www.biogeosciences. net/9/993/2012/bg-9-993-2012-supplement.pdf.