Structural and functional responses of harpacticoid copepods to anoxia in the Northern Adriatic : an experimental approach

Combined in situ and laboratory studies were conducted to document the effects of anoxia on the structure and functioning of meiobenthic communities, with special focus on harpacticoid copepods. In a first step, anoxia was created artificially by means of an underwater chamber at 24 m depth in the Northern Adriatic, Gulf of Trieste (Mediterranean). Nematodes were found as the most abundant taxon, followed by harpacticoid copepods. While nematode densities were not affected by treatment (anoxia/normoxia) or sediment depth, these factors had a significant impact on copepod abundances. Harpacticoid copepod family diversity, in contrast, was not affected by anoxic conditions, only by depth. Ectinosomatidae and Cletodidae were most abundant in both normoxic and anoxic samples. The functional response of harpacticoid copepods to anoxia was studied in a laboratory tracer experiment by adding13C pre-labelled diatoms to sediment cores in order to test (1) if there is a difference in food uptake by copepods under normoxic and anoxic conditions and (2) whether initial (normoxia) feeding of harpacticoid copepods on diatoms results in a better survival of copepods in subsequent anoxic conditions. Independent of the addition of diatoms, there was a higher survival rate in normoxia than anoxia. The supply of additional food did not result in a higher survival rate of copepods in anoxia, which might be explained by the presence of a nutritionally better food source and/or a lack of starvation before adding the diatoms. However, there was a reduced grazing pressure by copepods on diatoms in anoxic conditions. This resulted in a modified fatty acid composition of the sediment. We concluded that anoxia not only impacts the survival of consumers (direct effect) but also of primary producers (indirect effect), with important implications for the recovery phase.

Abstract. Combined in situ and laboratory studies were con ducted to document the effects of anoxia on the structure and functioning of meiobenthic communities, with special focus on harpacticoid copepods. In a first step, anoxia was created artificially by means of an underwater chamber at 24 m depth in the Northern Adriatic, Gulf of Trieste (Mediterranean). Nematodes were found as the most abundant taxon, fol lowed by harpacticoid copepods. While nematode densities were not affected by treatment (anoxia/normoxia) or sedi ment depth, these factors had a significant impact on copepod abundances. Harpacticoid copepod family diversity, in con trast, was not affected by anoxic conditions, only by depth. Ectinosomatidae and Cletodidae were most abundant in both normoxic and anoxic samples.
The functional response of harpacticoid copepods to anoxia was studied in a laboratory tracer experiment by adding 13C pre-labelled diatoms to sediment cores in order to test (1) if there is a difference in food uptake by copepods under normoxic and anoxic conditions and (2) whether ini tial (normoxia) feeding of harpacticoid copepods on diatoms results in a better survival of copepods in subsequent anoxic conditions. Independent of the addition of diatoms, there was a higher survival rate in normoxia than anoxia. The supply of additional food did not result in a higher survival rate of copepods in anoxia, which might be explained by the pres ence of a nutritionally better food source and/or a lack of starvation before adding the diatoms. However, there was a reduced grazing pressure by copepods on diatoms in anoxic conditions. This resulted in a modified fatty acid composition of the sediment. We concluded that anoxia not only impacts the survival of consumers (direct effect) but also of primary producers (indirect effect), with important implications for the recovery phase.

Introduction
The dissolved oxygen concentration in bottom waters of many shallow coastal seas and estuaries is decreasing, of ten leading to so-called "dead zones'', areas with too little oxygen to support most marine life (Gray et al., 2002: Rabalais et al., 2002. With worldwide more than 400 systems recognized, covering a total area of ca. 245 300 km2 (Diaz and Rosenberg, 2008), hypoxia (defined here as DO levels < 2 mL L_1) and anoxia (no oxygen) are among the top list of emerging environmental challenges (UNEP, 2004: Rabalais et al., 2010 and have been found to expand rapidly (Con ley et al., 2011). Hypoxia caused by eutrophication is now considered to be amongst the most pressing water pollution problems in the world (Wu, 2002). Water column stratifi cation -the isolation of bottom water from exchange with oxygen-rich surface water -is another main cause of hy poxia (Diaz, 2001). A comparable phenomenon takes place in the archipelago area of the Baltic Sea, where algal mats become stagnant in shallow embayments, covering wide ar eas in whose centre hypoxic and even anoxic conditions de velop rapidly (Arroyo et al., 2012). The same study showed that the negative impact of hypoxia induced by drifting al gal mats (eutrophication) was propagated to almost all levels of the trophic and functional chain, influencing species in teractions even at the lowest levels. Moreover, the response to anoxia and the recovery from it can be size-(macrofauna vs. meiofauna) and species-dependent (Wetzel et al., 2002: Arroyo et al., 2006. In the Northern Adriatic, seasonal oxygen depletions, of ten associated with massive marine snow events, have also been noted periodically for centuries (Crema et al., 1991). In the second half of the 20th century their frequency and sever-ity, however, have markedly increased due to high anthro pogenic input of nutrients (Druon et al., 2004). Studies on the effects of hypoxia/anoxia here have mainly focused on benthic macrofauna organisms (i.e. Stefanon and Boldrin, 1982;Stachowitsch, 1984;Occhipinti-Ambrogi et al., 2002). Little research has been done to detect and quantify stress induced by oxygen depletion on meiofauna (Metazoa passing through a 1 mm sieve but retained on a 38 pm sieve) (but see Barmawidjaja et al., 1995;Travizi, 2000). Elmgren (1978) stated that oxygen-dependent zonation of the fauna can occur as macro fauna (> 1 cm) disappears at higher oxygen levels than some of the meiofauna. Especially nematodes are known to persist in low numbers in areas which have been anoxic for long pe riods. Most information on species-and/or community-level responses (including sensitivity and tolerance) to low DO is mainly provided by aquaculture impact studies (e.g. Grego et al., 2009;Zhou et al., 2012). While aquaculture, how ever, represents a rather long-term impact (weeks to months) for the benthic fauna, the course of oxygen depletion in the Northern Adriatic often takes place within a few days (Sta chowitsch, 1984). The potential short-term impacts (direct and indirect effects) on the meiofauna compartment might be different and are still underdocumented. Besides differences in terms of causes and time frame of anoxia, the ecology and the contribution to ecosystem functioning of the organisms under study are pivotal to evaluate the final impact of anoxia. These parameters will determine the survivors after anoxia and will thus ultimately govern the recovery of the ecosys tem. In contrast to several studies on the effects of anoxia on the structural diversity of biotic communities, little is known about their functional response to anoxia, e.g. in terms of their feeding ecology.
The present study focusses on both the structural and functional responses of meiobenthic organisms to hy poxia/anoxia. Because of their small size, high abundances, short generation times and their close association with the sediment, meiobenthic organisms form an appropriate tool to determine the effects of perturbations in aquatic ecosys tems (Kennedy and Jacoby, 1999). More specifically, the present study targeted harpacticoid copepods as they repre sent, after nematodes, the second-most-abundant meiofauna group (Giere, 2009) and are known to respond rapidly to hy poxic/anoxic conditions (Moodley et al., 1997;Modig and Olafsson 1998). Moreover, harpacticoid copepods form an essential link between primary producers and higher trophic levels (Giere, 2009). Consequently, any impact of low oxy gen levels on this group in terms of lower density/diversity will have an important, cascading effect on the energy flow of the overall ecosystem.
To obtain detailed information on shifts in abundance, di versity, community composition and distribution of harpacti coid copepod species after short-term hypoxia/anoxia (i.e. days), an in situ experiment was performed. This field experiment was designed to better understand the impact of anoxia in the Northern Adriatic by gaining detailed infor mation on the survivors, as this is crucial to interpret post disturbance community patterns and in the resumption of overall ecosystem functioning. Although copepods have gen erally been found to be very sensitive to low oxygen lev els (e.g. Modig and Ólafsson, 1998;Travizi, 2000), it is important to know which copepod families are most sensi tive. Since many hypoxic/anoxic systems are eutrophicationpromoted (Diaz, 2001), the functional response of copepods to additional food supply (using 13C pre-labelled diatoms) was studied in the laboratory. In particular, we tested for dif ferences in diatom uptake by copepods under normoxic and anoxic conditions and whether initial feeding (i.e. before the onset of anoxia) on high-quality food like diatoms resulted in a better survival of and a differential response of the con sumers (copepods) to anoxic conditions. The use of trophic tracers, for example, allows testing the effect of additional food sources over the short term (stable isotopes, see review by Boecklen et al., 2011), while fatty acid (FA) profiling (see review by Kelly and Scheibling, 2012) provides insight into changes in the long run. The combination of both techniques will allow measuring carbon assimilation and contribution of provided diatoms to the energy level in the consumers under oxic vs. hypoxic conditions. Here we used FA profiling to check the quality of the food present in the sediment in the control vs. the treatments; subsequently, stable isotope anal yses were used to trace the food uptake of copepods in nor moxic and anoxic conditions. Data from this experiment will contribute to our knowledge on the energy flow between pri mary producers and consumers in low-DO conditions, which is a main issue in the overall functioning of benthic commu nities in impacted sites.

Study area
The effect of anoxia on meiobenthic organisms, with special focus on harpacticoid copepods, was studied in August 2009 (in situ experiment) and August 2010 (laboratory experi ment) in the Gulf of Trieste, Northern Adriatic Sea (Mediter ranean). The experimental site is located on a sublitoral soft bottom at 24 m depth, near the oceanographic buoy of the Marine Biology Station, Piran (Slovenia) about 2 km off shore (45°32/5.68"N, 13°33/ 1.89//E) in order to avoid any impact of fishery activities.

In situ experimental set-up and sampling
Anoxia was experimentally induced in situ from 8 to 12 August 2009 with a plexiglas underwater de vice (50 X 50 X 50 cm), originally referred to as EAGU (Experimental Anoxia Generating Unit) and described in detail in Stachowitsch et al. (2007). The chamber, originally designed to document macrofauna behaviour during low-DO conditions (see also Riedel et al., 2013), is equipped with a digital camera (Canon EOS 30D; images taken in 6 min intervals), underwater flashes and a sensor array (DO, hydrogen sulphide H2S). The sensors were positioned 2 cm above the sediment and values were recorded in 1-min intervals. Bottom water pH and temperature was measured with a WTW TA 197 pH sensor at the beginning and end of the deployment.
The deployment was terminated on day 5, and triplicate sediment PVC cores (length 20 cm; inner diameter = 4.6 cm; surface =16.6 cm2) were taken from the anoxic sediment in side the chamber by scuba divers. One day before the end of the deployment (i.e. at day 4) normoxic samples (3 repli cates) were taken at ca. 4-5 m distance from the chamber. In the laboratory, the sediment cores were immediately sliced into 5 depth layers: 0-0.5; 0.5-1; 1-1.5; 1.5-2; and 2-3 cm and stored in 4 % formaldehyde (final concentration).
Meiobenthic organisms were extracted by washing the sediment on a 1000 pm sieve mounted over a 38 pm sieve. The sediment retrieved (38 pm sieve) was centrifuged at 3000 rpm for 10 min with Ludox solution (specific density of 1.18 g cm-3 ) (Heip et al., 1985). This step was repeated 3 times to ensure that all meiofauna was extracted. Finally, meiofauna samples were stained with rose bengal and pre served in 4 % formaldehyde (final concentration).

Identification of harpacticoid copepods
Higher meiofauna taxa were counted under a binocular mi croscope (cnidarians were not taken into account). Copepods were picked out and stored in 75 % ethanol. For the identi fication of harpacticoid copepods, all or a minimum of 100 copepods were sorted at random from the extracted samples. From the anoxic and normoxic samples, 3 replicates per layer were identified, yielding a total of about 900 copepods for identification. Picked individuals were mounted on glycerine slides. In order to avoid flattening the specimens and to allow rotation of the animals for better observation, a small piece of a cover glass was placed in the glycerine before closing the slide with a cover glass. Harpacticoid copepods were iden tified at family level using a Feitz laborlux 12 microscope equipped with a drawing tube. Identification keys included Boxshall and Hasley (2004) and Tang (1948Tang ( , 1965.

Laboratory experimental set-up
To test for the effect of available food (here 13 C pre-labelled diatoms) on the response of copepods to anoxia, a lab exper iment was set up.
The cores were taken by scuba divers on 3 August 2010 at the same location as the field samples from 2009 (see be fore). Overall, 24 sediment cores (same dimensions as for field samples) were retrieved and immediately transported to a thermostatic chamber at 17 =b 1 °C and stored in complete darkness. The cores were provided with oxygen through an aeration system (tubes connected to the cores) to give a con- stant air flow in the water column. Air tubes were placed in the water column just above the sediment such as not to dis rupt the sediment of the cores (Fig. 1). Four additional sed iment cores (further referred to as cores taken at To) were collected to estimate the field harpacticoid copepod commu nity composition at the start of the experiment.
The effect of feeding before the onset of anoxia was tested using the epipelic pennate diatom Seminavis robusta as ad ditional food source (see also De Troch et al., 2007). The clone involved in the current experimental work is referred to as F2-44 in the diatom culture collection of the Laboratory of Protistology and Aquatic Ecology, Ghent University, Bel gium (clones from which the lineage was started were iso lated from a sample collected in November 2000 from the "Veerse Meer", a brackish water lake in Zeeland, the Nether lands) . By the time of setting the experiment, the diatom cells were 51-56 pm (n = 20) in length.
The diatom cultures were grown in flat tissue bottles (175cm2 surface) in a climate room at 16-18°C with a 12:12 h light-dark and 25-50 pmol photons m-2 s-1 light regime. The labelling technique resulted in isotope signa tures (513C) of -17.29 %o for untreated and 8949.51 %o in the 13C-enriched cultures. Prior to the experiment, the la belled medium was replaced by autoclaved artificial seawater (salinity: 32) without any additional nutrients (no f/2 growth medium added). To estimate the density of diatom cells in culture, the cells were homogeneously suspended by shaking the bottles in which cultures were grown, and 50 pL of the suspension was transferred into a well of a 96-well plate. After cell set tlement, their number was counted under a Zeiss Axiovert 40C inverted microscope (Zeiss Gruppe, Jena, Germany), al lowing the estimation of the available amount (densities) to apply in the experimental cores. At the start of the exper iment (day 1) 1.25 X IO6 fresh diatom cells were pipetted into the experimental units, corresponding to about 7.5 X IO5 10 cm-2 . The experiment was conducted in complete darkness to avoid additional diatom growth.
2. 7jND: cores incubated for 3 days in nor moxic conditions with 13 C enriched diatoms (D) (1.25 X IO6 cells core 1). This treatment allowed checking for difference in copepod abundance between treatments with and without extra diatoms.
3. 72N: cores collected at day 9 of the experiment (72 ), i.e. 6 days after 7j in normoxic conditions, without extra diatoms. This treatment serves as a control for anoxia without diatoms (treatment 72A).
5. 72 A: cores made anoxic at day 3 (7j) of the experiment and collected at day 10, i.e. after 7 days in anoxic con ditions (72), without extra diatoms. This treatment is a control for anoxic treatment with extra diatoms (72AD).
Short-term anoxia was created by sealing the core with a thick layer of paraffin oil (density: 0.8 g cm-3 ) in order to prevent oxygen diffusion between the seawater and the at mosphere (technique successfully used by Justin and Arm strong, 1983).
The cores were sliced into two layers: 0-1 cm and 1-3 cm. Meiofauna was extracted by means of décantation with fil tered seawater. The sample was put in 5 L of seawater and stirred, with meiofauna remaining suspended. The super natant was collected over a 38 pm sieve; this was repeated 5 times in order to collect all meiofauna from the sediment (Vincx, 1996). Centrifugation with ludox was not applied as we targeted live copepods for the stable isotope analysis. Fur thermore, ludox could impact the final 13 C signal through its osmotic effect on components of low molecular weight. The survival rate was estimated by manually removing dead specimens from the sample; live specimens were preserved at -20 °C prior to further stable isotope analysis. To detect a reliable 13C / 12C ratios in the tissue of the harpacticoids, a minimum of 15 pg C per sample is required. Therefore we used all live copepods from the 0-1 cm layer. There was in sufficient biomass of copepods in the deeper sediment layer.

Stable isotopes, fatty acids and pigment analysis
Copepods were picked out, washed several times with milliQ water, placed in tin capsules ( 8 x 5 mm, Elemental Micro analysis Limited) within 2 h (to avoid label leakage) and des iccated overnight at 60 °C. For each replicate, <$13C values and copepod biomass (total carbon) were measured with a continuous flow isotope ratio mass spectrometer (type Eu ropa Integra) at the UC Davis Stable Isotope Facility (Uni versity of California, USA).
Prior to meiofauna extraction, two aliquots of the sediment were taken from each core, one for FA analysis and one for pigment analysis. Hydrolysis of total lipid extracts of sedi ment (on average 30 pg per sample) and méthylation to fatty acid methyl esters (FAME) was achieved by a modified onestep derivatization method after Abdulkadir and Tsuchiya (2008). The boron trifluoride-methanol reagent was replaced by a 2.5 % H2SO4-methanol solution since BF3-methanol can cause artifacts or loss of polyunsaturated fatty acids (PUFA; Eder, 1995). The FA methyl nonadecanoate C 19:0 (Fluka 74208) was added as an internal standard to allow quantification. Samples were centrifuged (Eppendorf Cen trifuge 5810R) and vacuum dried (Rapid Vap LABCONCO). The FAMEs obtained were analysed using a gas chromato graph (Hewlet Packard 6890N) coupled to a mass spec trometer (HP 5973). All samples were run in splitless mode (1 pL injector per run, injector temperature 250 °C), using a HP88 column (60 m x 25 mm i.d., Df = 0.20; Agilent J&W; Day 9 t2n T A D Day 10 x t 2â TAD 6. 72 AD: cores enriched with diatoms, made anoxic at day Agilent Co., USA) with a He flow rate of 1.3 mLm in-1 . The initial oven temperature was 50 °C for 2min, followed by a ramp at 25 °C min-1 to 175 °C and a final ramp at 2 °C min-1 to 230 °C with a 4-min hold. The FAME were identified by comparison with the retention times and mass spectra of authentic standards and mass spectral libraries (WILEY, NITS05), and analysed with the software MSD ChemStation (Agilent Technologies).
Quantification of individual FAME was accomplished by using external standards (SupelcoTM 37 Component FAME Mix, Supelco # 47885, Sigma-Aldrich Inc., USA). The quan tification was obtained by linear regression of the chromato graphic peak areas and corresponding known concentrations of the standards (ranging from 5 to 250pgm L-1 ).
Pigments were extracted according to the protocol of Wright and Jeffrey (1997) for chlorophylls and carotenoids of marine phytoplankton. The sediment samples were lyophilized and weighted. Two (2) mL acetone 90 % was added as extraction fluid, followed by sonication of the sam ples (30 s in red light) to avoid degradation of pigments. Then 1.5 mL of the overlying fluid was taken with a syringe containing a filter (Syringe Alltech HPLC 13mm 0.2 pm PFTE). The first 0.5 mL was considered waste, while the last 1 mL sample was captured in a dark vial immediately closed afterwards. The vials (extract + standards + blank samples) were then placed in the auto sampler of the HPLC (Gilson Macherey-Nagel column C18 250/4.6 nucleodur 5 pm).

In situ experiment
Meiofauna counts were standardized to densities/10 cm2. The diversity indices of Hill (1973) were calculated as Nq = number of taxa or families; Afi = exp (H'), with Hr as the Shannon-Wiener diversity index (loge based); and M n f = the reciprocal of the proportional abundance of the most common taxa or family. Community structure was analysed through multidimensional scaling (MDS) based on Bray-Curtis similarity calculations. SIMPER (SIMilarity PERcentage) was carried out, and differences amongst iden tified groups were tested with a two-way analysis of variance (two-way ANOSIM), with treatment (anoxia/normoxia) and sediment depth as main factors. Calculations and multivari ate analyses were performed using the PRIMER 6 software package (version 6.1.10) (Clarke and Warwick, 2001).

Lab experiment
Stable isotope ratios are expressed relative to the conven tional standards (Vienna Pee Dee Belemnite (VPDB) for car bon) in units of parts per thousand, according to the formula where X is 13C and R the ratio of 13C / 12C. Incorporation of 13C was expressed as specific uptake, i.e. the difference between the 13C fraction of the con trol (i.e. based on the natural signature of organisms that did not feed on labelled diatoms) and the sample: A 5 13C = 5 13C sampieá 13C control (where á 13C is expressed relative to VPDB standard).
Further standardization o f 5 13 C values w as done follow ing M iddelburg et al. (2000). Incorporation o f 13C is re flected as excess (above background) 13C and expressed as total uptake (I) in m g 13C in d .-1 , calculated as the product o f excess 13C (E ) and individual biom ass (organic carbon). E is the difference betw een the fraction 13C of the control (/'control) and the sam ple (ísam pleL E = F sample -/'control, w here F = 13C/(13C + 12C) = R/(R + 1). The carbon isotope ratio (R) w as derived from the m easured 5 13C values as R = (á13C/1000 + 1) x R v p d b w ith RVPDB = 0.0112372 as S13C is expressed relative to VPDB. This total uptake w as further standardized and expressed per unit carbon o f cope pod.
Differences in uptake were tested by means of twoway analyses of variance (two-way ANOVA) with treatment (anoxia/normoxia) and sediment depth as main factors, using the software Statistica 7.0. Prior to all ANOVAs, Cochran's C test was used to check the assumption of homoscedasticity.

Sensor data
Dissolved oxygen concentration decreased exponentially from initially 5.51 m LL-1 (=7.87m gL-1) to OmLL-1 within 48 h, and anoxic conditions persisted until the end of the deployment on day 5. H2S started to increase soon after onset of anoxia, from 0 pM to final values reaching ~ 29 pM. The temperature in the chamber remained constant (average 20 °C); the bottom water pH dropped from initially 8.1 to a minimum of 7.6. The salinity was 38.

Meiofauna densities and community composition
Overall, a two-way ANOVA on the total meiofauna den sities (ind. 10 cm-2 ) showed a significant effect of sedi ment depth (decreasing abundance with depth, p = 0.002), while the different oxygen treatments (normoxia/anoxia) in terestingly showed no effect (p = 0.05, Fig. 3a). Nema todes were the dominant meiofauna taxon and reached the highest relative abundance ± stdev in the 1-1.5 cm layer in normoxia (97.8 ± 3.5% ) and 1.5-2 cm depth in anoxia (97.0 ± 3.7% ). However, no significant effect of treatment (anoxia/normoxia) or depth layer was found (p = 0.19 for treatment, = 0.13 for depth) for nematodes. Copepods were the second-most-abundant taxon, reaching a maximum relative abundance in the top sediment layers (0-0.5 cm, In contrast to nematodes, there was a significant effect of both treatment (p = 0.03) and depth (p < 0.001) on the rel ative copepod contribution (Fig. 3b). In the latter case, the difference between normoxia and anoxia was only signifi cant for the top sediment layer (0-0.5 cm, post hoc Tukey HSD, p = 0.003). Kinorhyncha represented the third-mostabundant taxon, with 8.3 =b 1.3% at 0.5-1 cm depth (nor moxia) and 12.6 =b 6.8 % in the top 0-0.5 cm layer (anoxia). The percentage proportion of Kinorhyncha differed signif icantly with depth (p <0.001), but there was no effect of treatment (p = 0.84). Remaining taxa present in the normoxic samples included (in order of decreasing abundance) Polychaeta, Ostracoda, Acarina, Bivalvia, Amphipoda, Mysidacea, Turbellaria and Insecta. The anoxic samples contained Polychaeta, Ostra coda, Acarina and Turbellaria. For all these taxa, there was both a treatment (p = 0.04) and a sediment depth (p = 0.004) effect.
On the MDS plot on the relative meiofauna composition (stress = 0.05, Fig. 4a), the top sediment layer (0-0.5cm) grouped separately from the deeper layers (0.5-3 cm), which points to a strong effect of the sediment depth and no clear effect of anoxia on relative meiofauna composition. This separation was further confirmed by ANOSIM (R = 0.651, p = 0.001). A two-way crossed SIMPER analysis showed an average similarity in taxa contribution of 72.4% in the surface layers (0-0.5 cm) and 79.8% in the deeper lay ers (0.5-3cm). Nematodes (60.0%, 95.4%, respectively, in 0-0.5 cm and 0.5-3 cm) and copepods (29.5 % in 0-0.5 cm) contributed most to the dissimilarity between surface and the deeper layers.
The MDS plot (stress = 0.12, Fig. 4b) of the relative copepod families composition revealed a high similarity between the top layers (0-0.5 cm) of normoxic and anoxic samples, while the deeper layers showed a higher variability (i.e. sam ple points are more spread). The difference in relative family composition between surface (0-0.5, 0.5-1 cm) and deeper sediment layers (1-3 cm) was supported by ANOSIM (R = 0.719, p = 0.001).

Diversity patterns
Independent of the treatment, the meiofauna in the surface layers (0-0.5, 0.5-1 cm) showed the highest diversity (Ta ble la). Overall, the Hill diversity indices (Table la) revealed no significant difference in meiofauna diversity between the normoxic and anoxic samples (two-way ANOVA, p = 0.65). However, there is a significant decrease of diversity with in creasing sediment depth (two-way ANOVA, p = 0.02), indi cated by the number of taxa (No) in normoxia and, to a much lesser extent, in anoxia. The decreasing Ainf (Table la) val ues in deeper sediment layers illustrate that the dominance level increased with depth.
For harpacticoid copepods, no significant difference in di versity (No, number of families, Table lb) was found be tween normoxic and anoxic samples (two-way ANOVA, p = 0.47). Sediment depth, however, had a significant effect, resulting in a decrease in copepod diversity with increasing depth (two-way ANOVA, p < 0.0001).
The induction of anoxia yielded a significant decrease of the oxygen levels in the overlying water (one-way ANOVA, p< 0.00001) from initially 6 .6 ± 0 .2 m g L -1 (T\N) and 6.4 ± 0.06 m gL-1 (72N) to 0.58 ± 0.29 mg L-1 (72A) after 7 days of closure of the core, independent of the addition of extra diatoms or not.
Chlorophyll a (Chi a) concentrations were measured at time T\ and Ti and ranged between 0 and 90pgg_1. The addition of diatoms had a highly significant effect on Chi a values (p < 0.0001) as samples without additional diatoms had < 5 p g g 1 Chi a, while treatments with extra diatoms had > 3 5 p g g 1 Chi a. Because of the high variance in Chi a concentrations in treatments with additional diatoms (83.9 ± 52.2 p g g -1 (TiND), 69.8 ± 33.3pgg" 1 (T2ND), 36.6 ± 24.0pgg-1 (72AD)), there was no significant differ ence in Chi a between the different time intervals, T\ and 72, (one-way ANOVA, p = 0.35). Another pigment, Chlorophyll c2 (Chi c2), showed similar patterns to Chi a with maximum 3.5 ± 4.2 pgg-1 (T\N) in treatments without diatoms and up to 8 .7 ± 7 .2 pgg-1 after adding diatoms (72ND). In terms of carotenoids, the concentration of fucoxanthin ranged

Survival rate
The survival rate of harpacticoid copepods was signifi cantly higher in normoxic conditions (93.2 ± 3.7 %) than in anoxic conditions (73.6 ± 16.2%) at T2 (two-way ANOVA, p = 0.005). Additional food supply tended to have a nega tive effect on copepod survival as the average survival rate for treatments with extra diatoms was 79.9 ± 14.5 % in con trast to 86.9 ± 16.1 % in treatments without extra diatoms. Due to the high variance, this effect was, however, not sig nificant (two-way ANOVA, p = 0.2). More specifically, the anoxic treatment with extra diatoms (7)> AD) had a lower sur vival rate (66.9 ± 5.8 %) than the one without extra diatoms (80.3 ± 21.5 %, r 2A). In combination with the effect of the anoxic condition, the former (T^AD) had a significantly lower survival rate than the oxic treatments with or without extra diatoms (72ND: 93.0 ± 2.0 %; 72N: 93.5 ± 5.3 %) (post hoc Tukey HSD, p < 0.01). It is worth noting that there was no effect of food addition on the survival rate in the oxic treatments (72N, 72ND).

Diatom uptake as indicated by biomarkers
The total FA concentration in the sediment was high est in treatments with additional diatoms (note dif ferent scales in Fig. 6a, b), with the highest val ues in TiND (307.6 ± 42.0 pgm L-1) and 72 AD (966.1 ± 272.2 pgm L-1 ). The unsaturated FA C16 : 1&>7, for example, was most abundant in these treatments, and linoleic acid C18:2&>6 was found only in samples with additional diatoms. Independent of the additional food supply, the anoxic samples showed a higher total FA concentration than the normoxic treatments (Fig. 6a, b). Especially the relative contribution of C 16 : 1&>7 increased in the anoxic conditions, in particular when diatoms were added (72A, 72AD, Fig. 6c,  d).

In situ experiment
In the present field experiment, total meiofauna densities were not significantly affected by anoxic conditions. This is in contrast with previous studies that showed a significant de crease of meiofauna densities due to anoxia (Moodley at al., 1997;Travizi, 2000). While studies on macrofauna revealed a peak in mortality at the transition from severe hypoxia to anoxia (Riedel et al., 2012), meiofauna -in general being more tolerant (Moodley et al., 1997) -decreased in density, but some may still be alive. Van Colen et al. (2009) created hypoxic conditions in a tidal mudflat for 40 days. While no macrobenthos survived, nematode diversity and abundances, for example, changed, but no complete mortality occurred. The strong resistance of nematodes to anoxia and the fact that they are the dominant meiofauna group can explain the lack of a clear effect of anoxia on total meiofauna density. Another and probably reliable explanation is the short du ration of the anoxic/sulphidic phase (i.e. 5 days). In contrast to long-term impact studies, e.g. from aquaculture impact (see Grego et al., 2009), the present study aimed to analyse short-term effects since this is often the case in the North ern Adriatic (Stachowitsch, 1984). The present study con cluded that only copepods showed a short-term anoxia re sponse, while nematodes, kinorhynchs and other meiofauna taxa were not affected. In addition to the duration of anoxia, also the way of creating anoxia can be pivotal in the interpre tation of the results. The way the benthic underwater cham ber created anoxia could also explain the lack of response by meiofauna. By sealing a 50cmx50cmx50cm volume off from the surrounding environment (Stachowitsch et al., 2007), the field experiment mimics the situation where water column stratification is the main cause of hypoxia, i.e. the isolation of bottom water from oxygen-rich surface water (Diaz, 2001). As this set-up caused a total cut-off from the food supply, eutrophication as an important factor in cre ating anoxia (Gray et al., 2002) was however neglected. In view of their large biomass and energy storage, macrofauna can generally cope with low-to no-food supply for longer periods (Ott, 1981). Moreover, a 5-day experiment/cut-off is negligible compared to weeks of strong stratification. The re sponse of meiofauna is less well known, and we can merely speculate that their low energy demands may be crucial to overcome food restrictions. In that context, the lab experi ment (see further) could contribute to our understanding of the role of food for the response of meiofauna to anoxia be cause it tested specifically for the effect of food level (normal or elevated). In contrast to the lack of a clear effect of anoxia on meio fauna, it was mainly the sediment depth that determined meiofauna densities and community composition. This cor responds with other meiofauna studies where the upper few centimetres of the sediment harboured more meiofauna than the deeper horizons (see review by Giere, 2009). As there was no significant difference in overall density between the natural (normoxic) and the treated cores (anoxia), we con clude that the impacted meiofauna community corresponded to a "natural, pristine" meiofauna community. Note that the meiofauna counts for this study were based on rose-bengalcoloured individuals and may therefore be an overestimation of the actual number of life animals (see Grego et al., 2013a).
Nematodes are known to dominate in terms of abundance and biomass in most meiofauna samples (Giere, 2009) and were also the most abundant taxon in this study, independent of the treatment (normoxia and anoxia), followed by cope pods. Our results showed that nematode densities were not affected by depth and treatment, whereas copepods were less abundant in anoxic conditions and in the deeper sediment layers. This difference in response can be due to phylogenetic constraints and lifestyle (Wetzel et al., 2001). Typically, a low oxygen demand in combination with a high surface : volume ratio enable some species to survive hypoxia/anoxia for ex tended times.
Studies within a given habitat typically concluded that of all permanent meiobenthic taxa, foraminiferans (see also Langlet et al., 2013) and nematodes are best adapted to hy poxia/anoxia and that crustaceans (e.g. copepods) are the least tolerant (see review by Wetzel et al., 2001). The gen erally high tolerance of nematodes to low oxygen concen trations (see also Moodley et al., 2000) often results in an increase in their relative abundance. One of the mechanisms to escape from temporal anoxia is to migrate from hypoxic sediments into the water column and to settle again to the sediment with the return of normoxic conditions (Wetzel et al., 2001). Harpacticoid copepods, in contrast, represent the meiobenthic taxon most sensitive to low DO levels (Hicks and Coull, 1983;Modig and Ólafsson 1998;Travizi, 2000), which might explain their restriction to the well-oxygenated top sediment layers.
A similar trend was found for meiofauna diversity: it de creased with depth, independent of the treatment (anoxia vs. normoxia). The top sediment layers (0-0.5, 0.5-1 cm) were also characterized by the highest meiofauna taxa diversity.
At lower taxonomic level, there was a clear effect of anoxia on the harpacticoid copepods' family composition. The families Ectinosomatidae and Cletodidae dominated in anoxic conditions, while the abundance of other families, in cluding Longipediidae and Miraciidae, decreased rapidly in anoxic conditions. Very little is known so far about differ ential resistance of harpacticoid copepod families to anoxia (but see Grego et al., 2013b). Overall, harpacticoid copepods have a high sensitivity to low-oxygen conditions (see before), except for so-called worm-like copepod species belonging to the families Paramesochridae and Leptastacidae, thought to be able to survive in anoxic conditions (Hicks and Coull, 1983). The diversity of harpacticoid copepod families dif fered significantly with depth. Again, as for meiofauna taxa, depth was found to be the most important factor structuring the harpacticoid copepod community.

Lab experiment
The lab experiment was conducted to get a better insight into the functional responses of harpacticoid copepods to anoxia. The initial copepod community (at To) differed slightly from the one reported for the field experiment (see before, col lected one year earlier). The community was dominated by the families Cletodidae, Laophontidae and Miraciidae, but a lower share of Ectinosomatidae was found, while they domi nated in the normoxic cores of the field experiment. This can be explained by interannual variability in the benthic com munities. However, the samples collected the year afterwards by Grego et al. (2013b) were also dominated by the family Cletodidae. Cletodidae were also found to dominate in the anoxic cores of the field experiment in the present study. Next to interannual variability, the underestimation of the family Ectinosomatidae can also be due to the extraction via décantation and not by means of centrifugation with Ludox (see field experiment). Since we wanted to use the live cope pods, i.e. the individuals that survived the lab experimental treatment, the use of Ludox was not an option. Species of the family Ectinosomatidae are often rather small and closely as sociated with sediment grains, it is plausible that they were not sufficiently extracted by the décantation method. In spite of these small differences in copepod family composition, there were no major changes in the overall diversity as the av erage number of copepod families in the To cores (6.8 ± 1.0) falls within the ranges reported for the normoxic (7.7 ± 0.6) and anoxic cores (6.0 ± 0.0) of the field experiment.
The use of pre-labelled diatoms (i.e. Seminavis robusta), as a typical food source for many meiofaunal taxa (Giere, 2009), proved to be successful in previous studies on harpacticoid copepods and their contribution to the energy flow in benthic ecosystems (e.g. De Troch et al., 2006. The attractiveness of meiofauna to diatoms can mainly be ex plained by the high content of PLJFA and the FA C16 : 1&>7 (De Troch et al., 2012). First we wanted to test for a func tional response in terms of food uptake by harpacticoid copepods in anoxic conditions. Previous studies indicated that anoxia negatively affected feeding behaviour of plank tonic calanoid copepods Acartia tonsa (Sedlacek and Mar cus, 2005) and of the juvenile blue crabs Callinectes spp. (Das and Stickle, 1993). Widdows et al. (1989), for example, explained the reduced food uptake by oyster larvae (Cras sostrea virginica) in anoxia as a way to save energy because digestion and absorption of food accounted for about 17 % of total metabolic energy expenditure. Studies on harpacti coid copepods showed that organisms became quiescent in order to survive unstable niches (Vopel et al., 1996). The same was observed for planktonic copepods, which entered a form of stasis, shutting down metabolic activity to minimize the energy expense of respiration (Kiorboe et al., 1985). This could explain why the copepods survived low-DO conditions but showed no feeding in the anoxic treatments. They even leaked some of the heavy carbon (13C) into the environment.
As expected, copepods in normoxic conditions had a 1.25times-higher survival rate than copepods in anoxic condi tions. Furthermore, we hypothesized that an additional highquality food source before the start of anoxic conditions might result in a higher survival rate during anoxia. Diet and level of food intake influence oxidative stress (Monaghan et al., 2009), and dietary intake influences the availability of ex ogenous antioxidants. Reduced caloric intake is known to be associated with reduced oxidative damage (Masoro, 2006). Souza et al. (2010), for example, found that certain calanoid copepods species survived better when they received addi tional food at the onset of a stress condition (i.e. ultravioletradiation-induced oxidative stress). In contrast, in our exper iment additional food did not yield any higher survival rate in anoxic conditions. One explanation might be the different types of stress involved in a mesocosm. However, also Sed lacek and Marcus (2005) found that the reduced egg produc tion of the calanoid copepod Acartia tonsa as a response to anoxia could not be mitigated by extra food (dinoflagellates).
Another explanation is that we did not reach complete anoxia in the sediment cores in the laboratory (DO was mea sured in the overlying water), although low oxygen concen trations in the water column can lead to anoxic conditions in the sediments (Diaz, 2001). Alternatively, as decomposi tion of organic matter (here diatoms) typically results in low oxygen concentrations (Gray et al., 2002), the cores with ad ditional diatoms (72AD) were expected to have lower DO concentrations and consequently lower survival rates than those without diatoms. This was, however, not the case in the present study as there was no significant difference in oxygen level in cores with or without extra diatoms. In both cases the oxygen level was as low as 0.58 ± 0.29mg L-1 (= 0.41 ± 0.20 mL L-1 ). This oxygen level can be ranked as medium severe hypoxia because it is just <0.5mLL_1 (Riedel, unpublished data), whereas in the field experiment it dropped down to 0 mL L-1 .
Finally, it is possible that there was already a high amount of initial food present in the sediment and any addition of extra diatoms would not imply any significant difference between both treatments at the onset of anoxia. The con trary, however, was true because the Chi a concentrations increased significantly in the treatments with additional di atoms. Moreover, the FA composition of the sediments with out additional diatoms did not show the presence of any other good food source, as FA concentrations in these cores were relatively low. The increase of the FA 16 : 1&>7, which is typ ical for diatoms, in the treatment with extra diatoms under lined this. More interestingly, however, was the considerable increase of FA in anoxic conditions, independent of the treat ment (additional diatoms or not). This can be explained by a reduced grazing pressure by copepods on diatoms in anoxic conditions. Thus, anoxia does not only impact the survival of consumers (direct effect) but also of primary producers (in direct effect). Moreover, the reduced grazing pressure and the increased diatom density might have important conse quences for the recovery phase after anoxia. Van Colen et al. (2012) stressed that a full recovery of ecosystem function ing after hypoxia implies a complete recovery of organismsediment interactions. The latter mainly points to bioturbating (macro)organisms. Nonetheless, the present study under lines the role of the players at the basis of the food web, i.e. the primary producers and grazers on them, as being es sential for the recovery phase.

Conclusions
In contrast with what we expected, total meiofauna commu nity density was not affected by anoxic conditions, mainly because of the dominance of highly tolerant nematodes. Sed iment depth seemed to be the most important factor struc turing meiofauna communities in our in situ experiment. The lack of effect of anoxia on total meiofauna composition could be due to too-short incubation time of the benthic underwa ter chamber for the effect to be shown on the most abundant meiofauna group, the nematodes. Deployments for longer time intervals (i.e. 1 month, 2 months, 1 yr) have been done but focused on foraminiferans (Langlet et al., this volume) and sediment geochemistry (Koron et al., this volume; Met zger et al., this volume). Moreover, novel approaches, such as the use of vital stains, may allow a more accurate assessment of alive/dead individuals following short-term impact by hy poxia and anoxia (e.g. see Grego et al., 2013a, for copepods and nematodes). Harpacticoid copepods were the most sen sitive meiofauna taxa to low oxygen concentrations. While densities of harpacticoid copepods were impacted by both sediment depth and anoxia, family diversity was affected by depth only. Stable isotope analysis showed that initial feed ing with diatoms did not result in a higher survival rate under anoxic conditions. In fact, copepods stopped feeding under anoxic conditions, yielding increased primary producer lev els. Accordingly, this reduced grazing response has impor tant implications for the recovery phase.