Key role of bacteria in the short‐term cycling of carbon at the abyssal seafloor in a low particulate organic carbon flux region of the eastern Pacific Ocean

The cycling of carbon (C) by benthic organisms is a key ecosystem function in the deep sea. Pulse‐chase experiments are designed to quantify this process, yet few studies have been carried out using abyssal (3500–6000 m) sediments and only a handful of studies have been undertaken in situ. We undertook eight in situ pulse‐chase experiments in three abyssal strata (4050–4200 m water depth) separated by tens to hundreds of kilometers in the eastern Clarion‐Clipperton Fracture Zone (CCFZ). These experiments demonstrated that benthic bacteria dominated the consumption of phytodetritus over short (~ 1.5 d) time scales, with metazoan macrofauna playing a minor role. These results contrast with the only other comparable in situ abyssal study, where macrofauna dominated phytodetritus assimilation over short (2.5 d) time scales in the eutrophic NE Atlantic. We also demonstrated that benthic bacteria were capable of converting dissolved inorganic C into biomass and showed that this process can occur at rates that are as high as the bacterial assimilation of algal‐derived organic C. This demonstrates the potential importance of inorganic C uptake to abyssal ecosystems in this region. It also alludes to the possibility that some of the C incorporation by bacteria in our algal‐addition studies may have resulted from the incorporation of labeled dissolved inorganic carbon initially respired by other unstudied organisms. Our findings reveal the key importance of benthic bacteria in the short‐term cycling of C in abyssal habitats in the eastern CCFZ and provide important information on benthic ecosystem functioning in an area targeted for commercial‐scale, deep‐sea mining activities.

Benthic POC cycling is an important ecosystem function and service in the deep sea as it plays a key role in influencing the amount of C that is ultimately sequestered in seafloor sediments (Dunne et al. 2007;Thurber et al. 2014). To date, the role of POC in deep-sea food webs has largely been explored through gut pigment (Drazen et al. 1998;Wigham et al. 2003;Wigham et al. 2008), fatty acid (Hudson et al. 2004), and natural-abundance stable-isotope signature analyses (Iken et al. 2001;Drazen et al. 2008). However, these methods do not allow the quantification of the rates of phytodetritus C cycling through the benthos to be determined. Other studies have used food web modeling (Rowe et al. 2008;van Oevelen et al. 2012) or introduced stable or radioactive isotope labeled substrates to quantify assimilation of organic matter by different sediment-dwelling organisms (Rowe and Deming 1985;Cahet and Sibuet 1986;Middelburg 2014Middelburg , 2018. These experiments, referred to as pulse-chase studies, allow the most important metazoan and microbial groups involved in C cycling to be identified and rates of C turnover to be assessed (Boschker and Middelburg 1997;Boschker et al. 2014;Middelburg 2014Middelburg , 2018. Microbial community structure is known to fluctuate with energy availability (Kanzog et al. 2009;Bienhold et al. 2012;Hoffmann et al. 2017), and can change in cold (0 C) deep-sea environments within periods as short as 1 week when exposed to fresh organic material (Kanzog et al. 2009). Hence, pulse-chase experiments can be supplemented with methodologies designed to detect changes in microbial community composition (e.g., 16S rRNA analysis) resulting from the addition of organic matter. To date, only five in situ pulse-chase experiments have been carried out at the abyssal seafloor, and all have been conducted under relatively eutrophic waters masses, including the NE Pacific (Sweetman and Witte 2008;Enge et al. 2011;Jeffreys et al. 2013), SE Pacific (Stratmann et al. 2018), and NE Atlantic (Witte et al. 2003a). However, only Witte et al. (2003a) quantified the collective response of bacteria and macrofauna to a fresh phytodetritus input event in an nondisturbed abyssal setting. Moreover, in contrast to shipboard studies that have shown that bacteria are important regulators of short-term degradation of organic material in the deep sea (Rowe and Deming 1985;Lochte and Turley 1988;Boetius and Lochte 1996;Kanzog et al. 2009;Hoffmann et al. 2017), the study by Witte et al. (2003a) identified macrofauna as the dominant group of organisms responsible for the initial stages of organic matter remineralisation. Such conflicting results highlight our limited understanding of the role of different organism size classes in the degradation of POC at the abyssal seafloor.
Direct measurements of POC supply and consumption have revealed that the input of POC is, at times, not sufficient to fulfill the C demands of abyssal ecosystems (Smith and Kaufmann 1999;Smith et al. 2002Smith et al. , 2013Smith et al. , 2014. While episodic food falls (e.g., gelatinous zooplankton carcasses) and aggregates can explain much of the discrepancy between POC supply and demand in areas close to the continental margin (NE Pacific; Smith et al. 2013Smith et al. , 2014, it is possible that organic C produced in situ through inorganic C fixation may provide another C source to deep-sea communities (Middelburg 2011;Molari et al. 2013). Christensen and Rowe (1984) and Middelburg (2011) both suggested that inorganic C fixation processes (e.g., nitrification) may produce organic material for benthic communities, but the rates of inorganic C fixation in deep-sea sediments were estimated to be relatively minor compared to incoming POC fluxes (Christensen and Rowe 1984) and shallower environments (Middelburg 2011). Brunnegard et al. (2004), however, showed that nitrification was the most important N-cycling process in sediments of the abyssal NE Atlantic accounting for 65% of the incoming N flux. More recently, a study in the abyssal NE Atlantic and Mediterranean Sea showed that deep-sea benthic prokaryotic communities could incorporate inorganic C into their biomass at very high rates (equivalent to approximately 20% of the heterotrophic C-production) through chemoautotrophic or mixotrophic processes (Molari et al. 2013). However, this study did not assess the transfer of fixed microbial C to metazoan consumers, and was undertaken on abyssal samples incubated at atmospheric pressure, which may have led to pressure-related artifacts (e.g., an increase in the activity of nonpiezophilic microbes). We know of no in situ study that has quantified inorganic C fixation at the abyssal seafloor or assessed C transfer through the metazoan community. Thus, the importance of this process in abyssal ecosystems is still unclear.
The abyssal seafloor underlying the eastern equatorial Pacific receives an annual POC flux equivalent to~20% of the flux at the equator (e.g., 0.4 g C m −2 yr −1 at 10 N vs. 2.2 g C m −2 yr −1 at the equator; Smith et al. 1997;Smith and Demopoulos 2003). The low POC flux to the seabed here allows the precipitation of manganese and iron oxides from the water column and sediment pore waters, forming polymetallic nodules at the seafloor that are rich in nickel and copper (e.g., exceeding 1.6% by weight; International Seabed Authority 2010a), as well as other commercially important metals (e.g., cobalt, molybdenum, and lithium) (International Seabed Authority 2010a). In the Clarion-Clipperton Fracture Zone (CCFZ) of the equatorial Pacific Ocean, nodules are found over an area of~5 × 10 6 km 2 at weights ranging from < 1 to > 35 kg m −2 (International Seabed Authority 2010a). Because of the sheer volume of nodules in the CCFZ and the increasing difficulty of extracting the important metals they contain from land-based mines, the CCFZ has become a prime area of interest for future seabed mining. Currently, 16 contractors have claimed exploration rights in this region, with contractors conducting surveys to gather baseline data on biodiversity and genetic connectivity across mine claim areas. If mining proceeds in the CCFZ, nodule extraction will significantly disturb the seafloor environment (e.g., 600-800 km 2 of seabed per mining operation per year affected by direct nodule removal and sediment plumes; Smith et al. 2008b;Levin et al. 2016). Thus, this represents a serious environmental issue, heightened by the lack of present knowledge about abyssal seafloor biology and ecology in this region.
Because C cycling and eventual C sequestration are key ecosystem functions and services in the deep sea (Thurber et al. 2014), the fate of POC should be included in baseline surveys, impact assessments, and later monitoring surveys at mine claim sites, so mining-related changes in this important ecosystem function can be detected. We undertook in situ pulse-chase studies in the eastern CCFZ to quantify C flow through benthic bacteria and macrofauna, determining how the abyssal benthos processes fresh POC beneath mesotrophic waters and whether seafloor processes in the CCFZ differ relative to other deep-sea environments. We also assessed the amount of inorganic C fixation that occurs in the seafloor of the CCFZ, and possible short-term changes in microbial community composition following the addition of organic and inorganic C. Our study focused on two different locations in the UK Seabed Resources UK1 claim area, and a location in the Ocean Minerals of Singapore claim area, in the eastern end of the CCFZ (Fig. 1). The most northerly and southerly sites were separated by over 200 km, which allowed us to address regional patterns in benthic processes. Specifically, we tested the following hypotheses. (1) Macrofauna dominate the short-term assimilation of phytodetritus at the abyssal seafloor of the eastern CCFZ. (2) Abyssal benthic bacteria actively assimilate inorganic C into their biomass in the eastern CCFZ.
(3) Microbial community composition changes significantly over short-term time scales following the addition of organic and inorganic C.

Study sites
In situ experiments were carried out in the eastern CCFZ ( Fig. 1) in two separate 30 x 30 km blocks (Strata A and B) of the UK1 claim area, as well as s 30 × 30 km block at the southerly end of the Ocean Minerals of Singapore (OMS) claim area 1 (Fig. 1). The UK1 strata (ca. 4100-4200 m water depth, bottom temperature approximately 1.5 C) were located at the northern and southern end of the UK1 contract area. The more northerly UK1 Stratum A was visited in October 2013 during the AB01 cruise aboard the R/V Melville, while the southern UK1 Stratum B and OMS stratum (ca. 4040 m water depth, bottom temperature approximately 1.5 C) were visited in February/March 2015 during the AB02 cruise aboard the R/V Thomas G. Thompson. The seafloor in the UK1 and OMS areas is characterized by abundant polymetallic nodules and silty clay sediments (radiolarian oozes).

Labeled algae preparation for phytodetritus-addition experiments
An axenic clone of the diatom Phaeodactylum sp. was used as an isotopically labeled food source in our phytodetritusaddition experiments. This species was chosen as a suitable food source because it belongs to a widely distributed diatom genus that occurs throughout the Pacific Ocean (De Martino et al. 2007) and possibly sinks to the seafloor in phytodetrital aggregates, although this requires confirmation. The phytoplankton culture was grown in F/2 algal medium in artificial seawater (Guillard 1975;Grasshoff et al. 1999) and isotopically labeled by replacing 50% of 12 C bicarbonate in the culture Fig. 1. Location of the UK1 and OMS study areas in the polymetallic nodule province of the eastern CCFZ.

Sweetman et al.
Key role of benthic bacteria in abyssal food webs medium with NaH 13 CO 3 . To harvest the algae, the cultures were filtered onto a 0.45-μm filter (cellulose acetate) and then rinsed with unlabeled F/2 medium to remove excess 13 C-labeled bicarbonate. The algal material was then centrifuged (404 g x 5 min) and washed with unlabeled F/2 medium five times, then frozen and freeze-dried. Isotope ratio mass spectrometry (IRMS) analyses on the algal biomass revealed that the diatom cells had a mass C content of 18.6% and a 13 C content of 37.8 atom %. This algal material was fresh and was likely significantly less degraded than the phytoplankton detritus that is typically deposited at the abyssal seafloor. As a result, the overall benthic responses observed maybe different from responses to naturally occurring phytodetritus at the seafloor (Aspetsberger et al. 2007). However, short-term (1.5-4 d) labeling studies testing C cycling processes using identically grown and harvested algal sources have been carried out in the deep sea (Witte et al. 2003a;Sweetman and Witte 2008;Jeffreys et al. 2013;Stratmann et al. 2018), allowing direct comparisons to be made with our data. Moreover, fresh algal cells, including diatoms, have been found at abyssal depths where they are likely available to benthic organisms (Lochte and Turley 1988;Agusti et al. 2015).

Benthic lander deployments Phytodetritus-addition experiments
In situ experiments were undertaken with a deep-sea benthic chamber lander (KUM GmbH). Four lander experiments were carried out in the UK1 Stratum A (2013), three in the UK1 Stratum B (2015), and two in the OMS Stratum (2015) ( Table 1). One additional lander deployment (without labeled algae, and hereafter referred to as a background deployment) was carried out in 2013 to collect sediment samples for background isotope measurements. A background deployment was also undertaken in 2015 to measure background sediment community oxygen consumption (SCOC) rates.
The lander carried three independent, autonomous, square benthic chambers (484 cm 2 ) separated by approximately 0.5 m. Immediately before each algal-addition lander deployment, approximately 47 mg of labile Phaeodactylum sp. culture (equivalent to approximately 180 mg C m −2 or 18% of the annual POC flux; Amon et al. 2016) were hydrated with cold (4 C), filtered (0.2-μm) seawater and placed in an algal injector mounted in the lid of each benthic chamber. The lander was then deployed and descended to the seafloor in approximately 2 h. After landing at the seafloor, the benthic chambers were slowly driven into the sediment by chamber motors that were commanded by an on board computer. Two hours after its arrival at the seafloor, the 13 C labeled algal biomass was injected into each chamber and homogeneously distributed by stirrers mounted on the chamber lids. Approximately 1 min after algal injection, the chamber stirrers were switched off for 1.5 h to allow the labeled algae to sink to the sediment surface (sinking rates were determined from previous laboratory experiments). Stirrers were then turned on, and the chamber waters gently mixed (60 rpm) for the duration of the 36 h (AB01) and 34 h (AB02) in situ incubations. Contros Hydroflash ® oxygen optodes were mounted on the lid of each chamber to measure SCOC rates at the seafloor during the AB02 expedition. Unfortunately, SCOC could only be measured in a single algal-addition and single background experiment due to two malfunctioning optodes. At the end of the experiments, the chambers were sealed and the lander was acoustically recalled from the seabed, arriving at the sea surface approximately 1 h later. Once the lander was back aboard the ship, the depth of the overlying water in each chamber was measured to calculate SCOC rates, the top water was siphoned off and filtered through a 63-μm mesh, and the sediment and nodules enclosed in each chamber were recovered from 0 to 2 cm and 2 to 5 cm deep sediment horizons. If nodules were found to cross the 0-2 cm or 2-5 cm sediment boundary within a chamber, they were classified as coming from the sediment layer where more than 50% of the nodule was present. Sediments from each vertical stratum were homogenized by mixing thoroughly with a spackle knife. From this homogenate, approximately 40 mL of sediment was removed and placed in 50 mL plastic vials for bacterial phospholipid-derived fatty acid (PLFA) analysis and frozen at −20 C. Samples (5 mL) for determining microbial community structure were collected from pooled 0-5 cm sediment samples using a sterile 20 mL syringe, and frozen at −80 C. The remaining sediment was sieved for macrofauna using a 300-μm sieve and 4 C, 0.2-μm-filtered seawater. Polymetallic nodules from each sediment layer were washed with a 300-μm sieve to remove any attached fauna (e.g., hydrozoans/nodule crevice-dwelling animals, such as nematodes), and the nodule volume was later calculated by water displacement. Sieved macrofauna from the polymetallic nodule residues (including any nodule biota) were pooled with the macrofaunal sample from the same sediment layer, and preserved in 4% buffered formaldehyde seawater solution. Macrofauna and bacteria from the background lander deployment from the AB01 expedition were sampled using the same procedures as above.

Dissolved inorganic carbon assimilation experiments
To quantify the assimilation of dissolved inorganic carbon (DIC) by benthic bacteria, and its subsequent transfer to the metazoan macrofaunal community, seafloor sediments were incubated in situ in two benthic experiments with 46 mL of 20 mM DI 13 C-labeled seawater (1.7 g L −1 of 99 atom % NaH 13 CO 3 ) at one UK1 Stratum B and one OMS stratum site during the AB02 cruise (Table 1). The DI 13 C solution was injected into each chamber approximately 2 h after the lander reached the seafloor from a 60-mL plastic syringe mounted in a syringe injection system located on the outside of the benthic chambers. The sediments were then incubated for 34 h before the chambers and enclosed sediments were retrieved by the lander, and the lander was recalled to the surface. All sediments were sampled in the same way as described above.
Sample processing and analysis of microbial community structure and diversity Short-term changes in microbial community structure and diversity following the addition of the labeled phytodetritus or DIC were assessed by comparing microbial community structure in the experimental chambers to measurements made in background sediments (0-5 cm) collected using an OSIL Bowers and Connolly mega-corer (with 12 x 10-cm diameter tubes) in the same study area. DNA was extracted from all sediments using the FastDNA Spin Kit for Soil (MP Biomedicals) as previously described (Shulse et al. 2017). For the samples collected during the AB01 cruise, the V4 region of the 16S rRNA gene was polymerase chain reaction (PCR)-amplified using the oligonucleotide primer pair F515/ R806, which included the Illumina flowcell adapter sequences (Caporaso et al. 2011). For experiments performed during the AB02 cruise, amplicon processing was performed as described in Lindh et al. (2015) using primers 341F and 805R targeting the V3-V4 region of the 16S rRNA gene (Herlemann et al. 2011). Sequencing of the PCR products from all experiments was performed on an Illumina MiSeq at the Hawaii Institute of Marine Biology Genetics Core Facility. Illumina MiSeq sequences obtained from the AB01 cruise experiment were processed as described in Caporaso et al. (2011), while data obtained from the AB02 experiments were processed as described in Lindh et al. (2015). Operational Taxonomic Units (OTUs) from all experiments were delineated at 97% 16S rRNA gene identity. DNA sequences have been deposited in the National Center for Biotechnology Information Sequence Read Archive under accession numbers SRP057408 and SRP078396.

Isotopic sample preparation and analysis
In a shore-based laboratory, frozen sediment samples for PLFA analysis were freeze dried and subsequently ground with a mortar and pestle. Lipids were then extracted from approximately 3 g of dried sediment using a Bligh and Dyer extraction procedure in which lipids were sequentially isolated by rinsing on a silicic acid column with chloroform, acetone, and methanol (Sweetman et al. 2010(Sweetman et al. , 2016. The lipid extract was then derivatized to volatile fatty-acid methyl esters and measured by gas chromatography IRMS (GC-IRMS) for PLFA concentration and delta 13 C (δ 13 C) signatures. The C-isotope ratios were corrected for the single methyl group inserted during derivatization. Bacterial biomass in the algal-addition and DICaddition experiments was calculated from the weightedaverage PLFA (iC 15:0 , aiC 15:0 ) concentration (μmol) mL −1 sediment/(a x b), where a is the average PLFA concentration in bacteria (for this study, we assumed 0.056 g C PLFA g −1 biomass; Brinch-Iversen and King 1990) and b is the average fraction-specific bacterial PLFA encountered in sediment dominated by bacteria (0.07; calculated after Rajendran et al. 1993Rajendran et al. , 1994. The prefixes "i" and "ai" refer to "iso" and "antiso," respectively. Areal estimates of bacterial biomass (mg C m −2 ) were then corrected for the fraction of nodule-free sediment.
Metazoan macrofauna were removed from formaldehyde solution by gently sieving on a 300-μm mesh in cool, filtered seawater in the laboratory. The sieve residue was then transferred to a petri-dish containing cold, filtered seawater, and macrofauna were picked out and identified under a dissecting microscope. Macrofauna from the AB01 expedition were grouped into polychaetes, crustaceans, and molluscs. In the AB02 samples, polychaetes were identified to family, while crustaceans and other taxa were identified to order or major taxon. Separate sorting utensils were used for background and enriched-isotope samples to avoid contamination with enriched 13 C. Due to insufficient biomass being available for measuring isotope signatures of specific animals, macrofauna were pooled into total macrofauna for the AB01 samples, and polychaete family, crustacean order, and major taxa for the AB02 samples. Organisms were washed to remove attached organic debris in cooled, filtered seawater before being pooled in silver cups, frozen, and freeze dried. All freeze-dried cups were then acidified using the methods of Sweetman et al. (2010Sweetman et al. ( , 2014Sweetman et al. ( , 2016 and dried at 45 C for 7 d prior to isotopic analysis. The isotopic ratios ( 13 C : 12 C) and biomass of macrofaunal organisms from the algae-and DIC-addition experiments were measured using a Thermo Flash EA 1112 elemental analyzer (EA; Thermo Fisher Scientific) coupled to a DELTA V Advantage IRMS (Thermo Fisher Scientific). Total incorporation of 13 C by macrofauna (mg 13 C m −2 ) from the algal-addition experiments was calculated following the methods of Sweetman et al. (2010Sweetman et al. ( , 2014Sweetman et al. ( , 2016. Total bacterial assimilation (mg 13 C m −2 ) of 13 C from the algal-addition experiments was calculated from label incorporation into the bacterial fatty acids (iC 15:0 , aiC 15:0 ) following the methods of Sweetman et al. (2010Sweetman et al. ( , 2014Sweetman et al. ( , 2016 using an average fraction-specific bacterial PLFA encountered in sediment dominated by bacteria (0.07; calculated after Rajendran et al. 1993Rajendran et al. , 1994. The 13 C-assimilation values (mg 13 C m −2 ) for macrofauna and bacteria were then converted to daily C-assimilation rates (mg C m −2 d −1 ) by accounting for the fractional abundance of 13 C in the added algae as follows: C assimilation = 13 C incorporated (mg 13 C m −2 )/fractional abundance of 13 C in algae, and dividing by 1.5 (AB01 experiments) or 1.42 (AB02 experiments) depending on the length of the experiment. Macrofaunal biomass (mg C m −2 ) and daily C assimilation rates (mg C m −2 d −1 ) for bacteria and macrofauna for each depth interval were then corrected for the fraction of nodulefree sediment.
The uptake of DI 13 C into macrofauna biomass and bacterial PLFA was calculated by measuring specific uptake (Δδ 13 ) of macrofauna and the bacterial PLFAs iC 15:0 , aiC 15:0 , C 16:1ω7c , iC 17:1ω7c , 10-Methyl C 16:0 , C 17:1ω8c , C17:1 ω6c , and Cy-C 19:0 . Specific uptake of bacterial PLFA was calculated as excess (above background): Δδ 13 C = δ 13 C PLFA sample -δ 13 C PLFA background . Δδ 13 -values of macrofauna were calculated as Δδ 13 C = δ 13 C macrofauna sample -δ 13 C macrofauna background . Positive Δδ 13 C values indicated the incorporation of 13 C label into a specific bacterial PLFA and into macrofauna. Assimilation rates of DI 13 C by bacteria (mg 13 C m −2 ) were calculated from label incorporation into the bacterial fatty acids iC 15:0 , aiC 15:0 , C 16:1ω7c , iC 17:1ω7c , 10-Methyl C 16:0 , C 17:1ω8c , C17:1 ω6c , and Cy-C 19:0 following the methods of Sweetman et al. (2010Sweetman et al. ( , 2014Sweetman et al. ( , 2016, using an average fraction-specific bacterial PLFA encountered in sediment dominated by bacteria (0.43; calculated after Rajendran et al. 1993Rajendran et al. , 1994. To estimate total daily assimilation of DIC into bacterial biomass (mg C m −2 d −1 ), we accounted for the fractional abundance of 13 C in the water phase overlying the sediment in each chamber as follows: C assimilation = 13 C incorporated (mg 13 C m −2 )/fractional abundance of 13 C in water phase and dividing by 1.42. The fractional abundance of 13 C in the water phase ranged between 0.06 and 0.1, and was dependent on the volume of water within each chamber. It was calculated from the concentration (20 mM) and amount of 13 C label (99%) in the injected bicarbonate medium, and an average background DIC concentration at the seafloor of 2.4 mM, which was derived from four sites in the same locality as our AB02 study sites using the World Ocean Database and Ocean Data View (Version 4).

Data analysis
All data analysis was carried out in R. Prior to statistical analysis, data were checked for normality and heteroscedasticity. Data were transformed, when necessary, to meet parametric assumptions. For datasets meeting parametric assumptions, differences in biomass and daily C-assimilation rates were tested using a two-way ANOVA with organism group (macrofauna and bacteria) and cruise as factors. Significant differences were explored further using Tukey HSD post hoc tests. If datasets failed to meet parametric assumptions after transformation, nonparametric Kruskal-Wallis tests were used, and significance tests were explored via nonparametric multiple comparison procedures. An alpha level of 0.05 was used as a criterion for statistical significance throughout. For microbial community structure and diversity analyses, all data were analyzed using the R-package "vegan." Prior to community structure analyses, the 16S rRNA gene dataset was normalized by dividing the observed number of sequence reads for each individual OTU by the total number of sequence reads in the entire sample to calculate relative abundances (i.e., percentages of total sequences) of OTUs. To compare community structure between samples, we calculated Bray-Curtis distances using the function "vegdist" in the R-package "vegan" and performed permuational ANOVA using the function "adonis" in "vegan." Graphical outputs for microbial data were made using the R package "ggplot2."

Benthic community structure
Mean bacterial biomass measured in benthic chambers that were incubated with algae or DIC during the AB01 and AB02 campaigns were 530 AE 104 mg C m −2 (n = 4, standard error of the mean [SEM]) and 277 AE 52 mg C m −2 (n = 6, SEM), respectively (Fig. 2). Mean macrofaunal biomass was less in in the 0-5 cm sediment layer of sediments collected during the AB01 and AB02 cruises. Different letters denote significant differences (p < 0.05) between means. Error bars denote AE 1 SEM (n = 4 for AB01 and n = 6 for AB02).

Phytodetritus carbon assimilation by the benthic community
Mean total (summed bacteria and macrofauna) daily C-assimilation rates were not significantly different (Welch two-sample t test, p = 0.14) between the AB01 (0.97 AE 0.18 mg C m −2 d −1 , n = 4, SEM) and AB02 experiments (0.60 AE 0.10 mg C m −2 d −1 , n = 4, SEM), although the statistical power of the test was low (due to between chamber variability being large). The pooled background δ 13 C signature for abyssal macrofauna sampled during the AB01 cruise was −18.6‰. Druffel et al. (1998) showed that POC settling at a depth of 3450 m in the water column in the NE Pacific possessed a δ 13 C signature of~−21‰. Assuming a δ 13 C enrichment of 1‰ per trophic step (Fry and Sherr 1984), it seems plausible that phytodetritus was the main source of C to macrofauna in the AB01 samples, and possibly the AB02 samples as well given the relatively close proximity of the AB01 and AB02 study sites (Fig. 1). Mean δ 13 C values for macrofauna from the algal-addition chambers for AB01 and AB02 were −10.0 AE 3.9‰ and −6.1 AE 10.1‰, respectively (n = 4, SEM), suggesting minimal uptake of 13 C-labeled phytodetritus. This was also reflected in mean daily C assimilation rates of macrofauna in the AB01 and AB02 experiments, namely 0.9 AE 0.4 x 10 −3 mg C m −2 d −1 (n = 4, SEM) and 5.0 AE 4.0 x 10 −3 mg C m −2 d −1 (n = 4, SEM), respectively (Table 2; Fig. 3). The single large vermiform animal had a δ 13 C signature of −18.9‰ indicating no uptake of labeled phytodetritus. In contrast, mean daily C-assimilation rates for bacteria were several orders of magnitude higher at 0.97 AE 0.18 mg C m −2 d −1 (n = 4, SEM) and 0.59 AE 0.10 mg C m −2 d −1 (n = 4, SEM) for the AB01 and AB02 studies, respectively (Fig. 3). Mean daily C-assimilation rates by bacteria during both cruises were significantly greater (two-way ANOVA, p < 0.001) than daily macrofauna C-assimilation rates (Fig. 3) from the AB01 and AB02 cruises, but no significant cruise effect was identified, although the power of the statistical test was low. The dominant macrofaunal taxa involved in C assimilation in the AB02 experiments were nereid polychaetes, followed by ostracods, gastropods and tanaids (Table 2; Fig. 4). Subsurface and surface deposit-feeding polychaetes (e.g., orbinid and spionid polychaetes, respectively), as well as carnivores (e.g., pilargids and pisionids) and other omnivorous polychaetes (e.g., lumbrinerids) consumed little or no algal C (Table 2; Fig. 4). SCOC rates that were measured in a single background and algal-addition experiment during the AB02 expedition were 2.

Evidence for chemoautotrophy at the abyssal seafloor
Positive Δδ 13 values indicative of assimilation of DI 13 C were identified in six out of eight bacterial-specific PLFAs studied (Fig. 5). The highest Δδ 13 values were detected in the bacterial PLFA iC 17:1ω7c , 10-Methyl C 16:0 , and C 17:1ω8c (Fig. 5). No 13 C uptake was detected in the PLFA C 17:1ω6c or Cy-C 19:0 . Δδ 13 values of bacteria-specific PLFA were generally higher in the 0-2 cm sediment layer compared to 2-5 cm sediment depth (Fig. 5). The mean Δδ 13 values of macrofauna from the DICincubation experiments was −3.2 AE 0.2‰ (n = 2), indicating no transfer of labeled C from bacterial biomass into the metazoan macrofauna after 34 h. Mean daily incorporation of inorganic C into benthic bacteria was 1.20 AE 0.35 mg C m −2 d −1 (n = 2), which was 1.2 and 2.0 times greater than the mean daily bacterial assimilation rates of labeled algal C from the AB01 and AB02 cruises, respectively. A statistically significant difference was detected (Kruskal-Wallis, p = 0.0015) between median daily algal C-incorporation rates by fauna and bacteria (AB01 and AB02 data pooled) and daily inorganic C-incorporation rates by bacteria. Median daily inorganic C fixation by bacteria was significantly greater than median daily algal C incorporation by metazoan macrofauna, and median daily algal C incorporation by bacteria was significantly greater than median daily algal C incorporation by metazoan macrofauna. No significant difference was detected between median daily algal C or inorganic C assimilation into bacterial biomass.

Microbial community structure and diversity
Alpha-diversity was very stable for both the AB01 and AB02 experiments with an average Shannon-Weiner diversity index of 6.25 AE 0.11 (n = 5, standard deviation [SD]) and 6.65 AE 0.08 (n = 14, SD), respectively. Analysis of microbial community composition by Permutational ANOVA (PERMANOVA) using Bray-Curtis distances confirmed that there were no statistically significant differences in microbial taxonomic composition when sediments were incubated with fresh phytodetritus compared to background sediments (PERMANOVA, p > 0.05, n = 9-15; Fig. 6A,B). Analysis of the taxonomic composition among OTUs binned at phylum/class level in background and algal-addition treatments revealed highly similar relative abundances (% of total sequences) of the major microbial groups (Fig. 7A,B). This pattern was consistent for both the AB01 and AB02 experiments (Fig. 7A,B). At the OTU level, the top 10 most abundant OTUs detected in the AB01 experiments included the Acidobacteria AT-s2-57 (OTU 4425583), which had the highest observed relative abundance (~5%) in treatments amended with algal biomass (Table 3). Most of the top 10 OTUs, such as OTU 619389 affiliated with Flavobacteriales, had either similar or lower relative abundances in the algalamended compared to the background samples from AB01 (Table 3). Forty percent of the top 10 most abundant OTUsfound in the sediments from the AB02 experiments had similar relative abundances in the algae treatment relative to background samples (Table 3) Fig. 3. Mean daily bacterial and macrofaunal algal C-assimilation rates (mg C m −2 d −1 ) in the 0-5 cm sediment layer of benthic chamber sediment samples collected during the AB01 and AB02 cruises. Different letters denote significant differences (p < 0.05) between medians. Error bars denote AE 1 SEM (n = 4). Note: Different y-axes for bacteria and metazoan macrofauna algal C-assimilation rates. 50 (Planctomycetes), OTU 15 (Rhodospirillaceae), OTU 6 (Nitrospirales), OTU 12 (Chloroflexi), the gammaproteobacterial OTUs affiliated with JTB255 marine benthic group (2 + 19), which had slightly higher relative abundances in algal-amended compared to background samples. Microbial community structure was similar in sediments amended with labeled DIC and in background samples (PERMANOVA, p > 0.05, n = 5; Fig. 6C). The top 10 most abundant OTUs observed in the DIC-addition experiment were the same as for the AB02 algal-addition experiments (Table 3). Nevertheless, OTUs 2 and 19 affiliated with JTB255 marine benthic group bacteria and OTU 50 and 15 affiliated with Planctomycetes and Rhodospirillaceae, respectively, had slightly higher relative abundances in the DIC-incubation treatments compared to background samples (Table 3).

Discussion
Bacteria dominated the short-term assimilation of fresh phytodetritus in both the AB01 and AB02 experiments, consuming 99.9% AE 0.0% (n = 4, SEM) and 98.8% AE 1.1% (n = 4, SEM) of the 0.97 AE 0.18 mg C m −2 (n = 4, SEM) and  0.60 AE 0.10 mg C m −2 (n = 4, SEM), respectively, that was traced into benthic (macrofauna and bacteria) biomass over 34-36 h. Thus, our first hypothesis that macrofauna dominate the short-term assimilation of phytodetritus in this region of the Pacific can be rejected. We found similar responses by the abyssal benthic community across all three strata that were   separated by tens to hundreds of kilometers, so our results may be generalizable over large spatial scales in the eastern CCFZ. We did not consider the role played by metazoan meiofauna (32-300 μm), meiofaunal protists, nanobiota, fungi, or Archaea in C cycling during our experimental studies. Previous investigations in the abyssal central Pacific have shown that the metazoan meiofaunal taxa (32 to 300-μm), protists, and nanobiota constitute a large fraction of the infaunal abundance and biomass (Snider et al. 1984). While the exclusion of metazoan meiofauna could have altered our results somewhat, this likely had less effect on our results than the exclusion of the Foraminfera. Previous investigations have shown metazoan meiofauna do not respond as strongly to phytodetritus C input as macrofauna or Foraminifera (Gooday et al. 1996;Moodley et al. 2000Moodley et al. , 2002Nomaki et al. 2005;Guilini et al. 2010;Pozzato et al. 2013;Mevenkamp et al. 2017;Stratmann et al. 2018). Among the Foraminifera, monothalmids dominate assemblages in the eastern CCFZ, including the UK1 and OMS areas (Nozawa et al. 2006;Goineau and Gooday 2017). There is some evidence that, as a group, these "primitive," single-chambered Foraminifera are less responsive to pulsed inputs of labile organic matter than multichambered taxa, particularly some rotaliid species (Bertram and Cowan 1999;Enge et al., 2011). However, the high dominance of monothalamid Foraminifera at our study sites and the rapid response by bacteria to phytodetritus input may lead to monothalamids playing a much larger role in C cycling in the eastern CCFZ than other areas (e.g., NE Pacific; Enge et al. 2011) as exclusive feeding on bacterial biofilms by allogromiid species has been previously documented (e.g., Bernhard and  Bowser 1992). Therefore, the role of the meiofaunal Foraminifera community, in particular the monothalamid forams, should be investigated in follow-up studies. The contribution of macrofauna-sized Foraminifera (> 300 μm), which in the CCZ mainly comprise monothalamids, such as komokiaceans and tubular forms, to short-term C-uptake was assessed during the AB01 experiments. The amount of C-uptake by these forams was insignificant (3.3 AE 1.7 x 10 −3 mg C m −2 d −1 ) compared to that of bacteria as also found by Sweetman et al. (2009). As a result, we did not include these large Foraminifera in the AB02 experiments. Conversely, some small rotaliids (notably Alabaminella weddellensis and Epistominella exigua) that are known to feed on fresh phytodetritus in the North Atlantic Ocean (Gooday 1988) are a minor component of sediment assemblages in samples from the UK1 and OMS areas. There is evidence that pulsed inputs of phytodetritus occur occasionally in the eastern CCZ (Radziejewska 2002). If such an event occurred at one of our study sites, it is likely that species such as A. weddellensis and E. exigua would consume the phytodetritus, rapidly increase their population density and undertake an active role in processing this labile material.

Community responses to phytodetritus
Rapid responses by bacteria to the input of phytodetritus/ fresh organic material have been shown at a variety of deepsea sites (Rowe and Deming 1985;Lochte and Turley 1988;Boetius and Lochte 1996;Kanzog et al. 2009;Hoffmann et al. 2017). For example, Rowe and Deming (1985) found rapid (< 5 d) utilization of 14 C-labeled glutamic acid by benthic microbes that had been collected from abyssal depths in the Bay of Biscay and on the Demerara Abyssal Plain (DAP). Lochte and Turley (1988) also documented significant bacterial growth after 2 d on freshly deposited detrital material recovered from 4500 m depth and incubated at in situ pressure. More recently, Kanzog et al. (2009) andHoffmann et al. (2017) found significant positive responses by microbial communities from the bathyal Arctic Ocean to the addition of chitin and phytodetritus in 7 d in situ and 23 d ex situ experiments, respectively. Finally, unpressurized shipboard studies using 13 C-labeled detritus revealed that abyssal bacteria can process much more phytodetritus C than macrofauna (Moodley et al. 2005). Despite our in situ abyssal data from the CCFZ being consistent with these findings, pressurerelated artifacts associated with sediment retrieval from depth (e.g., decompression of organisms, macrofaunal death, and an increase in the activity of nonpiezophilic microbes) may significantly alter benthic activities and resource partitioning among organism groups even if sediments are repressurized following seabed sampling. Care should therefore be taken when comparing results from ex situ (shipboard) to in situ (seabed) experiments.
To our knowledge, only the study of Witte et al. (2003a) on the Porcupine Abyssal Plain (PAP, 4800 m) in the eutrophic NE Atlantic has assessed the in situ short-term response of both benthic bacteria and macrofauna to the addition of identically grown and harvested, fresh phytodetritus in an undisturbed abyssal setting. Interestingly, as in our study, bacterial biomass in the NE Atlantic was much greater (2.5 x 10 3 mg C m −2 ) than macrofaunal biomass (120 mg C m −2 ) (Witte et al. 2003a). Yet, in the NE Atlantic, macrofauna consumed more than 95% of the total amount of C that was assimilated by the benthos over 2.5 d, with an almost undetectable contribution by bacteria (Witte et al. 2003a). Moreover, mean total daily C-assimilation rates by the benthic bacterial and macrofaunal community in the CCFZ were 39-64% of those recorded from the abyssal PAP by Witte et al. (2003a) and 47-79% of the total modeled heterotrophic organic C-utilization rates recorded from the DAP by Rowe and Deming (1985) (summed bacteria and macrofauna C-assimilation rates: 0.6 to 1.0 mg C m −2 d −1 for the CCFZ vs. 1.52 mg C m −2 d −1 for the PAP after 2.5 d [Witte et al. 2003a] and 1.23 to 1.29 mg C m −2 d −1 for the DAP [Rowe and Deming 1985]). It is unlikely that the lower response by macrofauna we detected (relative to the PAP and DAP studies) was an artifact of our experimental design, as very little organic material would need to be consumed by the macrofauna to detect positive C-uptake into the macrofaunal biomass (e.g., 0.6 μg C would need to be consumed by 1 mg C of macrofaunal biomass to detect a positive shift in δ 13 of 20 ‰ assuming the algae were labeled with 37.8 atom % 13 C). Although the lower total C-assimilation rates measured in our study were partly due to daily bacterial and macrofaunal C-assimilation rates being normalized to account for the presence of nodules, a number of other factors could possibly account for the differences in the benthic response we observed. First, the shorter duration of our experiments compared to those of Witte et al. (2003a) may have been a factor. Short-term in situ tracer studies are logistically simpler to run as they require less ship-time, are not as prone to de-oxygenation artifacts, and useful for identifying uptake by fast-growing organisms at the bottom of the food web (e.g., microbes) (Middelburg 2014). However, they do not always allow adequate detection of transfer from basal resources via intermediates to eventual consumers (Middelburg 2014). As a result, they can show completely different responses by the benthic community compared to longerterm in situ investigations. In a 1.5 d in situ study at 1250 m in Sognefjorden (Norway), Witte et al. (2003b) showed more C incorporation by macrofauna relative to bacteria, and lower amounts of C assimilation, with the relative roles being reversed and more C being processed over a 3 d period. This response was also seen in the abyssal NE Atlantic (Witte et al. 2003a). Thus, the shorter duration of our experiments may have been one cause for the different organism responses we observed compared to those seen by Witte et al. (2003a). Longer studies (weeks to months) may thus reveal differences in the dominant C-processing organisms. Second, the smaller amounts of phytodetritus added to sediments in our study may have also played a role. Our algal C-addition was equivalent to~18% of the annual POC flux to the seafloor estimated for the CCFZ study area (1 g C m −2 yr −1 ; Amon et al. 2016). The amount added in the NE Atlantic study was much larger in absolute (1 g C m −2 ) and relative (33-50% of the annual POC flux at their study site) terms (Lutz et al. 2007). Modeling studies simulating benthic C cycling at abyssal depths have revealed that the bacterial contribution to total benthic C cycling increases under low POC flux relative to high POC flux conditions (Dunlop et al. 2016), which can be explained by competition for substrates between macrofauna and bacteria. In a shallow-water experiment, macrofauna exploited labile C-sources to a greater extent than bacteria when food was more concentrated in seafloor sediments but were outcompeted by bacteria when food was less concentrated (Van Nugteren et al. 2009), though this is not always evident (e.g., Sweetman et al. 2016). Therefore, lower amounts of C added may explain the more significant and consistent bacterial response at our study sites. If our results reflect abyssal community responses to smaller phytodetritus pulses that are typical for many mesotrophic and oligotrophic abyssal sites, they suggest that bacteria may play a key role in the shortterm cycling of phytodetritus at the abyssal seafloor of the CCFZ. Finally, differences in benthic community structure between study sites could also explain the lower amount of C processed, and lower macrobenthic response to phytodetritus in our study compared to at the PAP. At our study sites, total benthic biomass was lower (mean summed biomass: 0.44 g C m −2 ) and known surface deposit-feeding fauna (e.g., spionids) contributed little to macrofaunal abundance (14% of polychaetes) and a negligible amount to C assimilation (Table 2; Fig. 4). In contrast, benthic biomass at the PAP was much greater (summed bacterial and macrofaunal biomass: 2.6 g C m −2 ) and surface-feeding cirratulid and spionid polychaetes were very abundant (50% of all polychaetes), and they contributed heavily to macrobenthic biomass (57% of polychaete biomass) and C assimilation (Aberle and Witte 2003;Witte et al. 2003a).
Assuming that our results can be generalized to the greater CCFZ, our findings have implications for the extraction of polymetallic nodules in this region (Smith et al. 2008b;Levin et al. 2016). Previous abyssal investigations assessing the response of the abyssal benthos to low-intensity, small-scale disturbance (e.g., at the DISCOL, Peru Basin and Inter-Ocean Metals, CCFZ study sites where < 11 km 2 of seafloor was disturbed) have shown that deep-sea benthic prokaryotic abundances and activities remain depressed relative to control areas for decades following disturbance (Weaver et al. 2016;Stratmann et al. 2018). Full-scale mining will be significantly greater in intensity and scale (600-800 km 2 of seafloor will be disturbed per year per mining operation, Smith et al. 2008b;Levin et al. 2016) compared to the disturbance created at the DISCOL and IOM study sites (Jones et al. 2017). Consequently, the disturbance effects on benthic communities are likely to be greater and persist for much longer (Miljutin et al. 2011;Levin et al. 2016;Jones et al. 2017). Our findings, which confirm the key role of deep-sea benthic bacteria in the initial stages of organic matter mineralization as shown by others (Rowe and Deming 1985;Lochte and Turley 1988;Boetius and Lochte 1996;Moodley et al. 2002), together with findings showing limited microbial recovery from small-scale disturbance in the abyss (Weaver et al. 2016;Stratmann et al. 2018), imply that key benthic ecosystem functions (e.g., bacterial C remineralization and C-burial efficiency) could be significantly impacted for decades, and perhaps even longer if polymetallic nodule mining commences. Benthic biodiversity and community structure assessments are a requirement for baseline surveys in mine claim areas in the CCFZ (International Seabed Authority 2010b). However, studies assessing C flow through the benthic community are not specified in the ISA regulations for the assessment of possible environmental impacts arising from polymetallic nodule extraction in the Area (International Seabed Authority 2010b). A power analysis based on our data reveals that only a small number of experiments (n = 10) would be required at both impacted and control sites to detect a 50% change in microbial C-cycling rates (one-sided test, power = 0.8). We therefore strongly recommend that studies assessing organic C processing by the benthic community be undertaken during baseline and monitoring surveys in mining claim zones, as they may provide a powerful tool for detecting changes in benthic ecosystem functioning resulting from mining.

Uptake of dissolved inorganic carbon by bacteria
Although there was considerable variability in the Δδ 13 values among the extracted bacterial lipids, the positive Δδ 13 values for six bacteria-specific PLFA in the chambers where DI 13 C was injected indicated active assimilation of inorganic C into bacterial biomass. This is consistent with our second hypothesis that "abyssal benthic bacteria actively assimilate inorganic C into their biomass in the eastern CCFZ." Estimated daily DI 13 C-incorporation rates were statistically indistinguishable to the bacterial algal-C assimilation rates we measured, which suggests the possibility that some of the algal-C incorporation we detected in the bacteria could have resulted from the incorporation of labeled DIC produced during the remineralisation of algal material by other unstudied organism groups, such as meiofauna or protists. Our 13 Cincorporation rates from the labeled DIC study were 24 to 44 times greater than the estimated chemoautotrophy rates reported for open ocean sediments (0.04 mg C m −2 d −1 ) by Middelburg (2011), but were similar to daily inorganic C-fixation rates measured using sediment cores that were recovered from abyssal depths and incubated at atmospheric pressure by Molari et al. (2013) (e.g.,~0.24-0.96 mg C m −2 d −1 , recalculated from fig. 4A in Molari et al. 2013, assuming a sediment density of 4 mL g dry sediment −1 and a sediment depth of 2 cm). The rate of inorganic carbon assimilation by bacteria was equivalent to 31% to 57% of the incoming POC flux estimated for our study area (2.7 mg C m −2 d −1 ; Amon et al. 2016), suggesting inorganic C assimilation by deep-sea bacteria, and the in situ production of organic C may be avery important process in abyssal benthic food webs. Although the lack of positive Δδ 13 values in the macrofauna indicated zero transfer of microbial-assimilated C into the metazoan macrofaunal food web over the short duration of our experiments, it is possible that this does occur but over longer durations than studied here. Clearly, further fieldwork using longer incubation times are required to confirm the generality of these findings, as well as to quantify C transfer from DIC via microbial intermediates to benthic metazoan communities.
It is unclear what energy source was utilized by bacteria to incorporate inorganic C into their biomass. Christensen and Rowe (1984) proposed that nitrification could play a role in supplying organic C to the deep-sea benthos, and Brunnegard et al. (2004) showed that nitrification can be a key N-cycling mechanism in abyssal sediments. However, it is unlikely that nitrification was the main inorganic C fixation process in our experiments. Assuming a molar stoichiometry of CO 2 fixed : NH 3 oxidized of 0.05-0.1 (based on Christensen and Rowe [1984], Middelburg [2011], and Könneke et al. [2014]), a NH 3 oxidized : O 2 utilized stoichiometry of 0.5 and 35% of oxygen consumption is due to nitrification (Christensen and Rowe 1984) would mean that 0.88-1.75 mol of CO 2 would be fixed per 100 mol of O 2 consumed. Our background SCOC rates were 0.22 mmol O 2 m −2 d −1 , implying the amount of CO 2 fixed by nitrification would be equivalent to 1.9 x 10 −3 to 3.9 x 10 −3 mmol C m −2 d −1 or 2.3 x 10 −2 to 4.6 x 10 −2 mg C m −2 d −1 . As our rates were much larger (1.20 AE 0.35 mg C m −2 d −1 ) this suggests only a minor role for nitrification in the bacterial incorporation of inorganic C that we detected. The limited role of nitrification in inorganic C fixation is also apparent if we assume that all of the particulate organic nitrogen (PON) transported to the seafloor is oxidized by nitrification. For example, if the C flux to the seafloor is 1000 mg C m −2 yr −1 (Amon et al. 2016) or 2.74 mg C m −2 d −1 and the C: N ratio of the particulate organic material reaching the seafloor is 12:1 (at 4000 m depth at Station Aloha; Karl et al. 2012), the daily PON flux to the seafloor would be 0.23 mg N m −2 d −1 or 1.6 x 10 −2 mmol N m −2 d −1 . If all of the PON was then nitrified at the seafloor and 0.05-0.1 mol of CO 2 were fixed per mole of NH 3 oxidized, then the CO 2 fixation rate would be 0.8 x 10 −3 to 1.6 x 10 −3 mmol C m −2 d −1 or 1.0 x 10 −2 to 2.0 x 10 −2 mg C m −2 d −1 , which is 1-2 orders of magnitude lower than our mean inorganic C fixation rate. Providing the C incorporation we detected was due to only chemoautotrophic C-fixation processes, sulphide oxidation could also serve as an energy source. However, again it is unclear where the source of sulphide would come from at our study sites since POC fluxes here are low (Lutz et al. 2007;Amon et al. 2016). Nitrite oxidation (Pachiadaki et al. 2017), hydrogen oxidation resulting from the radiolysis of water (D'Hondt et al. 2009), and oxidation of reduced iron or manganese particles that have been deposited from nearby hydrothermal vent sites (Resing et al. 2015;Tully and Heidelberg 2016) could offer alternative energy sources for at least some of the inorganic C fixation. Alternatively, it could be the result of heterotrophic bacteria using anaplerotic C-fixation mechanisms. However, estimates of the contribution of anaplerotic C fixation to biomass production suggest this pathway supplies only 5% to 10% to total biomass production (Boschker et al. 2014). Thus, if this process was the main cause for the inorganic C-fixation we measured, mean bacterial biomass production would range from 12 to 24 mg C m −2 d −1 , which is 4.4 to 8.8 times greater than the incoming POC flux, and 1.5 to 3.1 times greater than published deep-sea bacterial biomass production estimates (e.g., 7.76 mg C m −2 d −1 based on a global bacterial biomass production of 0.88 Pg C yr −1 in Danovaro et al. [2015] and an open-ocean area of 31.07 x 10 13 m 2 in Middelburg [2011]). Clearly inorganic C fixation by abyssal microbial communities requires further study, including an assessment of the dominant mechanisms involved, and identification of the major energy sources being used.

Microbial diversity and assemblage composition responses
Deep-sea benthic microbial community structure and diversity can rapidly change following exposure to organic matter. Hoffmann et al. (2017) documented a significant decrease in microbial diversity (number of OTUs) over a 23 d period following the addition of phytodetritus. Similarly, Kanzog et al. (2009) showed that the addition of chitin led to significant shifts in microbial community structure in a 7 d in situ experiment. Despite bacteria dominating the assimilation of phytodetritus and incorporating DIC into their biomass, we did not observe a significant shift in microbial community structure or diversity (relative to background sediments) in any of the lander experiments over a period of 34-36 h. Thus, our third hypothesis that "microbial community composition changes significantly over short-term time scales following the addition of organic and inorganic C" can be rejected. Previous microbial community structure studies at Station M in the more eutrophic NE Pacific (POC flux~3 g C m −2 yr −1 ) have shown that abyssal benthic bacterial community structure remains relatively stable over long periods of time (weeksyears) despite fluctuations (4-5X) in POC flux (Moeseneder et al. 2012). We suspect the lack of a significant overall community response in our experiments was due to the short incubation times (34-36 h).
Although no significant shift in microbial community structure was seen during the AB02 experiments, we did detect small shifts in the relative abundance of some OTUs between treatments with and without added algae. The OTUs belonging to the Gammaproteobacteria JTB255 Marine Benthic Group (e.g., OTU 2 and 19), Nitrospirales (OTU 6), Chloroflexi (OTU 12), Planctomycetes (OTU 50), and Rhodospirillaceae (OTU 15) increased slightly in relative abundance when exposed to fresh phytodetritus (Table 3). Hoffmann et al. (2017) showed an increase in the relative abundance of 12 bacterial families from Arctic deep-sea sediments following exposure to phytodetritus at cold temperatures (0 C) over a period of 23 d. There were also slight OTU changes following the addition of labeled DIC. For example, OTUs affiliated with JTB255 marine benthic group bacteria (OTUs 2 + 19), Planctomycetes (OTU 50), and Rhodospirillaceae (OTU 15) had approximately or greater than twice as high relative abundances in the DIC experiments compared to background conditions after~1.5 d. As the added algae were axenic, it is unlikely that the slight positive responses we observed in the algae-addition experiments were due to the addition of algaeassociated microbes. Therefore, the slight increase in the relative abundance of some OTUs could have been caused by specific microbial groups responding positively to the new conditions created by our additions. However, Deming (1985) showed bacteria doubling times of weeks to months when sediment samples from abyssal depths were incubated under in situ conditions. Moreover, Alongi (1990) showed slow bacterial growth rates (0.001-0.12 d −1 ) in surface sediment samples collected from 10 bathyal and abyssal (695-4350 m) sites in the Solomon and Coral Sea. It is therefore more probable that the cold temperatures at the seafloor in our study region (~1.5 C) limited changes in in situ microbial community composition, and the slight differences we observed in the relative abundance of specific OTUs were simply due to natural variability in the sediments. Our deep-sea study area in the eastern CCFZ is also replete with DIC (bottom water concentration:~2.4 mM) so our small DIC addition (~5-10% of total DIC concentration) should not have led to significant changes in DIC concentration at the seafloor and significant shifts in microbial community composition. We recommend that more long-term investigations, as well as the use of other methods (e.g., dual stable isotope probing; Wegener et al. 2012) should be undertaken to identify the nutritional pathways and feeding responses of specific microbes in abyssal sediments.

Conclusions
Our results reveal a key role for bacteria in the initial degradation of fresh phytodetritus at the abyssal seafloor of the eastern CCFZ, which is consistent with findings from earlier investigations from other deep-sea regions. The consistent nature of our results suggest that these characteristics may be generalizable over large spatial scales (distances of tens to hundreds of kilometers). The results presented highlight the differences that may occur in short-term C-cycling processes in abyssal environments characterized by different POC fluxes (i.e., mesotrophic vs. eutrophic systems). Our in situ DICaddition experiments indicated that fixation of inorganic C by benthic bacteria does occur at the abyssal seafloor in this region. The daily DIC assimilation rates were statistically indistinguishable to bacterial assimilation rates of phytodetritus POC, which further highlights the importance of benthic bacteria in abyssal sediments, and points to the possibility that DIC fixation by bacteria may provide a different source of C to abyssal benthic communities in this region. Given the large changes that are predicted to occur in the eastern Pacific over the next century due to climate change (e.g., declining POC flux and benthic biomass; Sweetman et al. 2017) and mineral extraction (Jones et al. 2017), more short-and longterm studies are now needed in other abyssal areas to determine the generality of our findings, and assess the ecosystem consequences that could arise from significant changes to benthic microbial structure and ecosystem functioning in the abyss.