Diel vertical migration into anoxic and high-pCO2 waters: acoustic and net-based krill observations in the Humboldt Current

A select group of marine organisms can enter the Oxygen Minimum Zones (OMZs) and even anoxic waters, while performing diel vertical migration (DVM). DVM of the euphausiid Euphausia eximia off northern Chile in the spring of 2015 was documented based on acoustic measurements using an echo sounder along with net samplings. Dissolved oxygen (DO) concentrations were obtained using a vertical profiler, and water samples were collected to obtain in situ nitrite (NO2−) concentrations as well as pHT, total alkalinity (AT), and therefore carbon dioxide partial pressure (pCO2) was estimated. Krill were found to migrate up to the surface (0–50 m) during the night and returned to ca. 200–300 m depth during the day, spending between 11 and 14 h at these layers. At the surface, DO and NO2− concentrations were 208 and 0.14 μM respectively, while pHT was 8.04 and 405 μatm pCO2. In contrast, at the deeper layers (200–300 m), DO and NO2− were < 3 and 6.3 μM respectively, with pHT 7.53 and 1490 μatm pCO2. The pHT and high pCO2 values at depths represent the conditions predicted for open ocean waters in a worst-case global warming scenario by 2150. The acoustic scatter suggested that > 60% of the krill swarms enter the OMZ and anoxic waters during the daytime. These frequent migrations suggest that krill can tolerate such extreme conditions associated with anoxic and high-pCO2 waters. The inferences drawn from the observation of these migrations might have strong implications for the current oceanic carbon pump models, highlighting the need for understanding the molecular and physiological adaptations allowing these migrations.

Climate change is undeniably causing profound changes on Earth 1 . The consequences for life forms are far from being understood, but it is expected that the physiological diversity among species will determine a broad array of responses. Oceans are not the exemption and are likely to suffer vast climate change driven alterations. For example, the ocean acidification (OA) and carbon dioxide partial pressure (pCO 2 ) levels, which are known to be associated with Oxygen Minimum Zone (OMZ) systems, are likely to intensify for open-ocean and coastal waters under the future scenarios of increasing atmospheric carbon dioxide release 2 . Unique biological forms that live in close association with these OMZs offer a unique opportunity to understand the adaptions required for open ocean inhabitants to overcome such projected future changes.
In the northern Humboldt Current System (NHCS), a few euphausiid species (e.g. Euphausia mucronata) have been described as entering to OMZs during their diel vertical migration (DVM) 3 . This OMZ (upper boundary of ∼45 µM or 1 mL dissolved oxygen L −1 ) has become one of the shallowest and most intense OMZs in the world's oceans 4,24 , with dissolved oxygen (DO) concentrations falling below the sensor-specific detection limits (~ 3 nmol L −1 ) in a significant fraction of the OMZ waters. Such extreme oxygen-deficient waters have recently Vertical migration into anoxic/high-pCO 2 waters. Algorithms applied to the datasets of 38 and 70 kHz detected high zooplankton abundance in the echograms at both frequencies along the transect. Acoustic records from both 38 and 70 kHz showed active DVM of zooplankton reaching the OMZ and even the AMZ (Figs. 3 and 4). Acoustic records along the coast (23-29°S) denoted a high eco-intensity signal, i.e. volume backscattering strength (S v , dB re 1 m −1 ), in the deeper layer (mostly at 200-300 m), with an S v range from − 71 to − 65 dB during day, and strong acoustic signals at the surface during night (≤ 50 m). The echograms obtained on 7 December 2015 at Stn L3 provided a detailed description of DVM patterns of zooplankton (Fig. 3 www.nature.com/scientificreports/ explaining the differences in NASC values. However, significant interaction between these factors was found when considering the time period along with the oxygen conditions (Table 2).

Discussion
Acoustic scatters, confirmed by net samplings, provided clear evidence that E. eximia migrate to surface layers (< 50 m) at night, concentrating high abundance and biomass near the surface. The net-based night/day abundance ratio obtained was comparable to the day/night scatter ratio estimated to be 2.7 considering the AMZ layer. These acoustic observations cannot be attributed exclusively to E. eximia DVM. Certainly, other migrant organisms could contribute to the observed patterns; however, net catches confirmed that E. eximia was the dominant species in both night and day conditions. Therefore, it is plausible to assume that E. eximia represented mostly the DVM observed in echograms (see "Methods" section for details on acoustic processing). During the  where it was found that the larval development of another krill species, E. pacifica, was strongly reduced at pH 7.69. Several studies have attempted to understand the effects of OA on krill species in different regions, mainly based on laboratory conditions. Some findings have predicted severe detrimental effects on embryonic development, larvae and post-larvae growth, reproduction, fitness, behaviour, metabolism, survival and inter-moult period 21,[30][31][32]62 , while others suggest the capacity of these organisms to tolerate high CO 2 waters 33,34 . However, all these studies have only manipulated CO 2 levels under normoxic (air-saturated) conditions and, therefore, without truly representing natural OMZ and AMZ conditions. It has been suggested that the efficiency of the aerobic respiration is dependent on the ratio pO 2 /pCO 2 63 , pointing out that the effects of low DO in some oceanic areas could be exacerbated by the increase in CO 2 . The physiological basis for the arguments by Brewer and Peltzer 63 comes from the fact that haemoglobin has an optimal pH for carrying oxygen (O 2 ) 35 . Further, Mayol et al. 36 demonstrated that high pCO 2 further challenges respiration in the NHCS. However, recent evidence in fish shows that they are better adapted to tolerate hypoxia under elevated CO 2 conditions as compared to hypoxia alone 64 . Certainly, the effects of pCO 2 and pO 2 cannot be considered independently, as they co-occur in nature and are both involved in respiratory processes. Although there is no consensus on mechanisms nor on the interactive effects of simultaneous changes of O 2 and CO 2 on animal metabolism, it is clear that E. eximia has adapted to survive up to 14 h in these conditions. When integrating information from Figs. 3 and 4, it became clear that the E. eximia individuals were capable of crossing the OMZ and reaching the AMZ where anoxic/high-pCO 2 waters prevailed. They tolerated this condition for at least 11-14 h during their DVM activity. Some species of zooplankton, such as E. eximia, are clearly adapted to inhabit OMZs during the day 3,22,29 . For instance, Tremblay et al. 70 suggested that some Humboldt krill species are able to depress their metabolism based on a regulation index. Our observations indicated that E. eximia can tolerate the AMZ in the NHCS for extended periods. During daytime, krill should have little aerobic scope in the OMZ and AMZ; since environmental DO partial pressures (pO 2 ) are below DO critical partial pressure (P crit ) (i.e. pO 2 < P crit ) to sustain aerobic metabolism. Therefore, our observations suggest that E. eximia´s metabolism is supressed or maintained at its minimum, perhaps fuelled by anaerobic pathways. In their study, Seibel et al. 22 found that E. eximia exhibited metabolic suppression during DMV, tolerating a pO 2 of 0.8 kPa (or ∼15 µM DO at 10 °C) for at least 12 h. Our observations and estimations agree with this previous report for E. eximia, although the acoustic records and NO 2 − measurements at Stn L3 showed that this species can even stay between 11-14 h in anoxic waters. This confirms the ability of E. eximia to suppress their metabolism, which might be more advantageous in the long term, than an enhanced anaerobic metabolism, which might still play a crucial role in short term.  Figure 2D: The red dotted horizontal line represents the OMZ upper limit (1 mL DO L −1 ), whereas the white isolines indicate NO 2 − concentrations from discrete profiles taken along transect. Stn L3 indicates the location where hydrographic, nutrient and carbonate system profiles were selected. Plots were created using Surfer software (version 16.6.484, https ://www.golde nsoft ware.com/produ cts/surfe r).

Scientific Reports
| (2020) 10:17181 | https://doi.org/10.1038/s41598-020-73702-z www.nature.com/scientificreports/ From the DO vertical profile, a pO 2 of 0.23 kPa (or ∼3 µM DO at 11.7 °C) was estimated at the AMZ core (300 m). This pO 2 value could be even lower or 0, according to NO 2 − measurements (as anoxia proxy), which would make this krill species the first to be reported to temporarily enter such extreme natural conditions. Kiko et al. 65 previously reported that E. mucronata survived several hours in anoxia at 13 °C under laboratory conditions, with a P crit of 0.6 kPa at 13 °C. Oxidative stress indicators have been proposed to explain, in part, the ability of E. eximia to live in the OMZ. This species has shown an elevated superoxide radical production, but very low lipid peroxidation 71 . These authors 71 suggested that E. eximia antioxidant enzyme activity was elevated enough to avoid oxidative damage during highly contrasting oxygen conditions during the species' DVM 71 .
Other respiratory proxies, such as the electron transport system (ETS) [72][73][74] , have been reported to be significantly higher in the OMZ during the day than at night in the E. distinguenda in the Eastern Tropical Pacific off Mexico, supporting evidence for metabolic suppression 72 . This enhanced enzymatic activity during the day would coincide with the DVM in the OMZ. These authors suggested that higher enzyme activity would allow the organisms to tolerate environments with low oxygen in deep waters 72,73 .
The DVM speeds estimated from the acoustic scatter were comparable to those in other studies with euphausiids using a similar methodology 40 . By assuming E. eximia to represent mostly the DVM observed in echograms, the swimming speeds estimated for E. eximia ranged from 24.78 to 99.28 m h −1 , in agreement with previous reports for E. mucronata (60-234 m h −1 ) 66 . Using the relationship between swimming speed and oxygen consumption rate proposed by Torres and Childress 67 for E. pacifica, we have calculated that E. eximia might need to consume between 0.065 to 0.098 μM O 2 mg DW −1 h −1 during its DMV within AMZ and OMZ.
Other krill species co-inhabiting the central and NHCS, such as E. mucronata, have been shown to contribute significantly to the carbon fluxes by migrating into the OMZ 3 , especially through the export of carbon contained in faecal pellets. Many studies have reported that euphausiids feeding on a diatom-rich diet at the base of the oxycline may have a significant role in carbon export flux [37][38][39] . Acoustics-based observations should be treated   www.nature.com/scientificreports/ with caution when determining the habitat compression or the upper limit of the OMZ, as the patterns observed could be highly species-specific, and not shared by other species. Krill migrating into AMZ are suitable sentinels of the effects of the ongoing climate change in this "natural lab", as this ecosystem is characterized by extreme pO 2 /pCO 2 conditions. Further studies will be required to better understand the physiological and genetic plasticity as well as adaptability of these krill species, which allow them to survive such extreme environmental conditions. Future experiments should consider environmentally relevant exposure times, i.e. when organisms stay under extreme environmental conditions, rather than unrealistic exposure times defined a priori. The study of the reaction norms, 41 described as the phenotypic expression of a single genotype across a range of environmental conditions (e.g. temperature, pO 2 , and/or pCO 2 ) might be a useful experimental tool for addressing such scientific questions.
Although limited in both spatial and temporal terms, our observations during the cruise, showed that euphausiid species inhabiting the NHCS, such as E. eximia, migrate into anoxic and high-pCO 2 waters. These water masses already have pH and pCO 2 values that are predicted for open ocean waters in the future (for the year ~ 2150-2200) 1,84 based on the trajectories of atmospheric CO 2 release by anthropogenic activities 23 . OMZs have expanded over the past 50 years, and this expansion is predicted to continue as the climate becomes warmer worldwide 25,26 . During our research cruise, DO concentrations were close to zero within the OMZ core, where even high NO 2 − concentrations established an anoxic condition 5,27 . The vertical extension of the zooplankton epipelagic community is usually limited by the presence of the OMZ 3,20,28,29,59 , but our findings not only offer acoustic confirmation of the results of previous net-based studies 42 showing that krill migrate to a particular depth. E. eximia can tolerate sub-critical oxygen levels for many hours and, but also, can temporally tolerate naturally anoxic and high-pCO 2 conditions. Furthermore, a large proportion of krill swarms (> 60% NASC) stayed in the AMZ, which has strong implications for carbon flux and further modelling.

Methods
Physical-chemical measurements. Samples were collected during the LowpHox I expedition to the OMZ off northern Chile, which was conducted in December 2015, on board the R/V Cabo de Hornos. Hydrographic variables were continuously measured at stations along the transect using a Pump Profiler System (PPS 27 ). This equipment pumps water directly from the desired depth (at resolution 1 m) while profiling the water column with an attached conductivity-temperature-depth (CTD) system (Seabird SBE-25) that provides continuous determination of salinity, temperature, and depth. DO was measured by an oxygen sensor (Seabird SBE 43) attached to CTD. Filtered seawater was collected for nutrient analysis in plastic flasks (30 mL) and was analysed on board for NO 2 − . Nutrient concentrations were obtained by automated colorimetric methods 69 using a high-resolution Auto Analyzer (SEAL Analytical, model AA3HR).
The pH samples were collected (in triplicate for each station and depth stratum) in 50 mL syringes and immediately transferred to a 25 mL thermoregulated cell at 25 ± 0.1 °C for standardisation, and pH was measured in a Metrohm pH meter equipped with a combined, double-junction, glass Ag/AgCl electrode (Metrohm model 6.0258.600). The electrode was calibrated with 8.089 Tris buffer solution at 25 °C (material supplied by Andrew Dickson, Scripps Institution of Oceanography, San Diego, USA). The pH values were reported on the total scale (pH T ). Samples (500 mL) for alkalinity (A T ) were poisoned with 50 µL of saturated HgCl 2 solution, stored in borosilicate bottles, and kept in darkness at room temperature. A T was determined by the open-cell titration method 43 using an automatic Alkalinity Titrator Model AS-ALK2 Apollo SciTech. All the samples Table 2. PERMANOVA test applied to the relative krill abundance as Nautical Area Scatter Coefficient (NASC m 2 nmi −1 ) in relation to time and oxygen conditions at Stn L3 during Lowphox cruise, December 7, 2015. OXY, oxygenated layer (> 1 mL DO L −1 ); OMZ, oxygen minimum zone (< 1 mL DO L −1 ); and AMZ, anoxic marine zone (> 0.5 µM NO 2 − ). df, degree of freedom; SS, Sum of squares; MS, mean square; P(perm), permutationbased p value. *, significant value; ** very high significant value; ns, not significant value. www.nature.com/scientificreports/ were analysed at 25 °C ± 0.1 °C. The accuracy was controlled against a certified reference material (Batch 140, A T 2232.58 ± 0.80 µmol kg −1 ) and repeatability averaged 2-3 µmol kg −1 . pCO 2 estimates were performed using CO2SYS software for MS Excel 44 set with Mehrbach solubility constants 45 refitted by Dickson and Millero 46 . The KHSO 4 equilibrium constant determined by Dickson 47 was used for all calculations. In addition, satellite-based sea surface temperature (SST) and chlorophyll-a (Chl-a) concentrations were evaluated over the study period. SST data were obtained from GOES 11/12 (5-km at coastwatch.pfeg.noaa.gov) and surface Chl-a from Aqua MODIS (L3SMI, 4-km at coastwatch.pfeg.noaa.gov). Both datasets were 8-day composites.
Acoustic analysis. The acoustic information was obtained from the R/V AGS 61 "Cabo de Hornos". This vessel is equipped with the scientific echo sounder model EK60 (Simrad, Kongsberg) with six transducers (Splitbeam), previously calibrated at frequencies of 18,38,70,120,200 and 333 kHz. These transducers were installed on a retractable keel that reaches 9 m below the water surface. The echo-integration system was calibrated following the manufacturer's specifications 49 . The calibration consisted of the insonification of a known target to correct the gain of the transducer for the estimates of target strength (TS) and volumetric backscatter force (S v ) for the echo-integration. The target used corresponded to a copper sphere of known white strength, specific for each frequency, which was located under the vessel in the centre of the acoustic beam and outside the near field of the transducer. The system parameters during the acoustic radiation emission were set according to the CCAMLR standards (Table S1).
Acoustic data were collected during the LowpHox cruise off northern Chile (Fig. 1A). The cruise track was a coastline parallel transect with a bottom depth ranging from 800 to 1200 m. The vessel speed was between 6 and 12 knots. The full dataset covers 991 km of acoustic recording in oceanic waters from 7 December (23.5°S; 70.8°W) to 13 December 2015 (32.1°S; 71.7°W). Frequencies of 38 and 70 kHz were used to detect and separate zooplankton (euphausiids) from other dispersants in the water column. Analyses were carried out with Echoview software (version 9.1). The flow chart for the process of noise reduction is shown in Fig. S1. We excluded data from the surface (5 m depth from the transducer, i.e. "ringdown" effect) and the bottom echo. Data were filtered to remove all noise sources in an efficient and semi-automated way (user-independent), thereby avoiding the altering of acoustic measurements and biases when manually removing noise sources 76 . To identify and compensate for the attenuated pings, the "attenuated signal filter" algorithm described by Ryan et al. 77 was used. Noise generated by interference from other acoustic equipment was removed using the impulse noise removal function 77 . Background noise was removed using the algorithm-based operator formulated by Robertis and Higginbottom 78 . Finally, the number of pings function was used to resample acoustic signals in order to better visualise daily migrations of the zooplankton swarms. The concept of noise must be understood as any acoustic signal, either biological or mechanical, and/or electrical interference not of interest or representing false measurements 48 .
We defined and limited our analysis to an effective sampling range between 5 and 400 m, given the physical limitations from near-field and sound-absorption effects related to the echo-sounder frequencies used (38 and 70 kHz). Absorption is greater for the 70 kHz frequency, which exhibits the shortest range but has a greater vertical resolution than the 38 kHz echo sounder. The 38 kHz frequencies, on the other hand, exhibit a much longer range (> 1000 m) but with limited resolution for small zooplankton scatters. However, both frequencies have proved to be efficient for studying the macro-zooplankton distributions (e.g. euphausiids) [50][51][52][53] . Díaz-Astudillo et al. 54 used 75 kHz frequency to study the zooplankton structure and vertical migration, detecting a high abundance of macro-zooplankton groups in the fjord water of Chilean Patagonia. In this work, a maximum threshold value of − 70 dB was established 53 to separate the zooplankton layers from other scatters (e.g. fish).
Volume backscattering strength (Sv, dB re 1 m −1 ) values from the single-frequency were re-scaled into the customary index "nautical area scattering coefficient" (NASC, in units of m 2 n mi 2 ) 55 , using a grid of 50 m (depth) by 1 h (distance). Since NASC lies in the linear domain, it can be considered proportional and suitable for indexing target abundance 56 . NASC measurements have been successfully used to estimate krill distribution and relative abundance [79][80][81] . Although these measurements may contain other scatterers of similar size, several studies have suggested that NASC reflects the abundance and distribution of adult krill 79,82,83 . Net sampling. Simultaneously with the acoustic data collection, zooplankton was collected with an opening/closing Tucker Trawl net of 1 m 2 mouth area and 300-μm mesh. A calibrated General Oceanics flowmeter was suspended across the centre of the net mouth to estimate the volume of seawater filtered by the net. Oblique tows were from depth strata 0-200 and 200-600 m during the day as well as night. The towing speed was 2 knots per 20-40 min. The sampling device was positioned at the krill aggregation maximum scattering quantity depth, corresponding to the maximum krill individual density. The depth and scatter sensors of the ship allowed us to accurately place the sampling device in the location where the krill aggregations were detected.
Once on board, krill samples were concentrated for taxonomical analysis. A stereomicroscope was used on live krill individuals at the taxonomic level of species. Euphausia eximia was identified, following Baker et al. 75 , in both the daytime and nighttime samples. A random subsample was made from each catch to assess the body length, sex composition, and maturity status of the sampled individuals. E. eximia body lengths were measured from the front of the eye to the telson tip. The samples were then preserved in liquid nitrogen for further analysis. E. eximia abundance, as a representation of the entire sample, was standardised to individuals per m 3 based on the water volume filtered by the net.

Statistical analysis.
To test the differences in the NASC values with regard to the time of day (i.e. nighttime versus daytime) and between the different oxygen conditions (i.e. OXY, OMZ, and AMZ), a distance-based permutational multivariate ANOVA analysis (PERMANOVA test, set at 9,999 permutations) was performed using Scientific Reports | (2020) 10:17181 | https://doi.org/10.1038/s41598-020-73702-z www.nature.com/scientificreports/ PRIMER-e software (version 7.0.13) 57 . PERMANOVA is a routine for testing the simultaneous response of one or more variables to one or more factors in an analysis of variance based on a resemblance measure. Datasets were averaged for each 38 and 70 kHz paired cell and fourth root transformed before analysis. The Bray-Curtis similarity of NASC values was used as a resemblance measure among sample groups 68 . Chemical conditions, considered a categorical factor, were set according to the chemical values in the water column: OXY, > 1 mL DO L −1 ; OMZ, < 1 mL DO L −1 ; and AMZ: > 0. 5