For many animals, evolution has selected for complex visual systems despite the high energetic demands associated with maintaining eyes and their processing structures. Therefore, the metabolic demands of visual systems make them highly sensitive to fluctuations in available oxygen. In the marine environment, oxygen changes over daily, seasonal and inter-annual time scales, and there are large gradients of oxygen with depth. Vision is linked to survival in many marine animals, particularly among the crustaceans, cephalopods and fish, and early life stages of these groups rely on vision for prey capture, predator detection and their distribution in the water column. Using in vivo electroretinogram recordings, we show that there is a decrease in retinal sensitivity to light in marine invertebrates when exposed to reduced oxygen availability. We found a 60–100% reduction in retinal responses in the larvae of cephalopods and crustaceans: the market squid (Doryteuthis opalescens), the two-spot octopus (Octopus bimaculatus), the tuna crab (Pleuroncodes planipes) and the graceful rock crab (Metacarcinus gracilis). A decline in oxygen also decreases the temporal resolution of vision in D. opalescens. These results are the first demonstration that vision in marine invertebrates is highly sensitive to oxygen availability and that the thresholds for visual impairment from reduced oxygen are species-specific. Oxygen-impaired retinal function may change the visual behaviors crucial to survival in these marine larvae. These findings may impact our understanding of species’ vulnerability to ocean oxygen loss and suggest that researchers conducting electrophysiology experiments should monitor oxygen levels, as even small changes in oxygen may affect the results.
Phototransduction is the process by which the energy from photons of light is translated into neural signals by photoreceptor cells (Rayer et al., 1990). The neural signaling requires the constant depolarization and repolarization of photoreceptors and downstream neurons, making vision one of the most energetically expensive processes in many animal systems (Ames, 2000; Pepe, 2001). These metabolic demands increase as temporal and spatial resolution increase and the visual system becomes more complex (Niven and Laughlin, 2008; Wong-Riley, 2010). As in terrestrial vertebrates, many marine invertebrates possess complex visual systems with a range of contrast and light sensitivities as well as sufficient temporal resolution for executing vital tasks such as prey capture or predator evasion (Warrant and Johnsen, 2013). One technique used to measure visual physiology is the electroretinogram (ERG), which records the summed activity of photoreceptors and downstream neurons in the eye in response to visual stimulation (Brown, 1968). The ERG is commonly used to measure the response of retinas in both vertebrate (Brown, 1968; Chen and Stark, 1994; Chrispell et al., 2015) and invertebrate systems (Cohen et al., 2015; Cronin and Forward, 1988; Frank, 1999; Lange and Hartline, 1974). Oxygen effects on vision have been extensively studied in humans and other terrestrial vertebrates, where a decline in oxygen (hypoxia) is known to cause a decrease in both sensitivity to light (Linsenmeier et al., 1983; McFarland and Evans, 1939) and temporal resolution (Fowler et al., 1993). In these vertebrate experiments, diminished sensitivity to light is demonstrated by a decrease in amplitude of visual responses after exposure to lower levels of oxygen, whereas reduced temporal resolution is seen as an inability of the retina to respond to high-frequency flashes of light. Both light and sufficient levels of oxygen are thus required for normal visual function.
In the marine environment, large gradients of irradiance and oxygen exist with depth, and changes in the partial pressure of oxygen (PO2) with water depth in the ocean can be up to 10-fold greater than the changes in atmospheric PO2 over terrestrial altitude (McCormick and Levin, 2017). For example, oxygen content with depth can decline by as much as 35% between 7 and 17 m depth off the coast of California (Frieder et al., 2012), and varies over time with diurnal/diel cycling, seasonal hypoxia and El Niño–Southern Oscillation cycles (Levin et al., 2015). Organisms that experience large changes in both irradiance and oxygen include those in regions with coastal hypoxia and diel oxygen cycling (Altieri and Gedan, 2015; Tyler et al., 2009), shallow oxygen minimum and oxygen limited zones (Gilly et al., 2013; Wishner et al., 2013; Wishner et al., 2018), and shallow embayments or fjords (Hansen et al., 2002).
Understanding how the visual systems of marine organisms respond to reduced oxygen availability will provide information on constraints on habitat preference in marine organisms. Highly visual marine organisms include cephalopods (e.g. squid, octopus), arthropods (e.g. crabs, krill) and fish (McCormick and Levin, 2017). These groups support major world fisheries (FAO, 2018), and the larval stage of marine organisms is a crucial bottleneck for survival for recruitment to the fishery and reproductive population. Early life stages of arthropods, cephalopods and fish rely on vision for behaviors essential to their survival, including prey capture and predator avoidance, and as a cue for diel vertical migration (Forward, 1988; Robin et al., 2014). Vision may also be one of the sensory modalities used in larval choice of settlement site and detection of conspecifics in species with an adult benthic stage (Lecchini et al., 2010; Lecchini, 2011). Hypoxia is known to affect many physiological processes in marine organisms (Grieshaber et al., 1994; Wu, 2002), but to our knowledge, the effects of hypoxia on visual physiology in marine invertebrates, and specifically their larvae, have not been studied. Here, we determined how exposure to low oxygen affects (1) visual sensitivity to light, (2) the dynamic range (range of irradiance that can be detected visually) and (3) the temporal properties of vision in larvae of the market squid (Doryteuthis opalescens), the two-spot octopus (Octopus bimaculatus), the tuna crab (Pleuroncodes planipes) and the graceful rock crab (Metacarcinus gracilis). These species are representative of highly visual invertebrates of both economic and ecological interest. We hypothesized that exposure to reduced PO2 in marine invertebrate larvae would decrease the magnitude of the ERG response to light stimuli, and that reduced PO2 would decrease the temporal resolution of the eye.
California Cooperative Oceanic Fisheries Investigations
photon flux density (μmol photons m−2 s−1)
partial pressure of oxygen
power spectral density
PO2 with 10% of retinal function
PO2 with 50% of retinal function
PO2 with 90% of retinal function
MATERIALS AND METHODS
To investigate how retinal function changed in response to a decline in PO2, we recorded in vivo ERGs in tethered, intact larvae while controlling the PO2 of pH-buffered seawater flowing over the animal (Fig. 1A). To control for differences in the shape of the ERG waveform across different species, we measured the size of the ERG response by integrating over the entire response. PO2 in the recording chamber was measured throughout the experiments using a fiber optic probe. Partial pressure units (kPa) are presented to best represent the oxygen available for animal tissues (Seibel, 2011), but wherever reasonable, a conversion to oxygen concentration (μmol kg−1 O2) was also calculated. The term ‘normoxia’ is used to describe surface ocean oxygen levels, approximately 100–105% saturation for the given temperature and salinity. To compare the magnitude of visual stimulation across species, we report light stimuli as a species-specific irradiance [species photon flux density (PFD; μmol photons m−2 s−1)], which is the irradiance of light at the plane of the animal's eye weighted to the spectral sensitivities for each species (Fig. S1).
Larvae of Octopus bimaculatus Verrill 1883, Metacarcinus gracilis (Dana 1852) and Pleuroncodes planipes Stimpson 1860 were obtained by conducting plankton tows at ∼30 m depth with a 325-μm mesh net in the Southern California Bight off the coast of La Jolla, CA, USA, at a recurrent market squid egg bed site (McGowan, 1954) (CA collection permit: SCP-13633; 32°51′30.13″N, 117°16′25.93″W) during the natural reproductive periods of each species (August 2017–April 2018). After collection, larvae of interest were kept in 0.5-liter tanks under a strict 13 h:11 h light:dark cycle, with a consistent feeding schedule [Zeigler Larval AP100 dry food (Gardners, PA, USA; crabs) or live copepods (cephalopods)], and in near-constant water temperature (16°C) prior to testing.
To obtain larvae for Doryteuthis opalescens (Berry 1911), divers collected freshly laid egg capsules at the same site in La Jolla (∼30 m water depth) during the full and new moon (±2 days) in January–March 2018 during the spawning period. Capsules were placed in 4-liter tanks and water chemistry was monitored throughout development until hatching (∼3–4 weeks) using Honeywell Durafet® pH sensors (Phoenix, AZ, USA) and Aanderaa oxygen optodes (model 4531; Bergen, Norway). Squid egg capsules were maintained at constant PO2 (∼22 kPa/∼260 μmol kg−1), temperature (11°C) and pH (∼8.2). After hatching, paralarvae were placed in smaller tanks (0.5 liters) and maintained as described for the other species.
All recently collected/hatched larvae were held for a minimum of 24 h prior to testing, and only individuals that appeared healthy upon inspection were chosen for experimentation. Only individuals of a single larval stage were used for each species (paralarvae for D. opalescens and O. bimaculatus, Stage II for P. planipes, and megalopae for M. gracilis), as different larval stages may have distinct oxygen tolerances (Yannicelli et al., 2013) and visual capabilities and/or structures (Feller et al., 2015). Here, these stages are referred to collectively as ‘larvae’ when individuals of multiple species are described in the text, but only individuals of the specified life stage for each species were tested. Owing to the challenge of determining sex in larvae, differences between sexes were not quantified in this study.
All procedures were in compliance with the Institutional Animal Care and Use Committee (IACUC) of the University of California San Diego, and care was taken wherever possible to reduce the stress and discomfort of the animals (e.g. transportation at a cool temperature in darkness, etc.). All experiments were conducted within 6–10 h after sunrise each day so individuals were at the same stage of circadian rhythm. All experiments were performed on dark-adapted (30 min) individuals of each species. Infrared (IR) light (940 nm) and video microscopy were used to prepare animals for recording and view animals during recording; IR light is beyond the sensitivity of most marine organisms at low intensities (Cronin and Forward, 1988; Fernandez, 1973; McCormick and Cohen, 2012). The electrophysiology recording equipment was housed in a light-tight enclosure and, with the exception of the controlled light stimuli and IR illumination for imaging, animals were kept in complete darkness for the duration of the experiment.
The recording chamber on the microscope stage was constantly perfused (∼4 ml min−1) with a solution of sterile seawater (Instant Ocean, Blacksburg, VA, USA; 27 g l−1 of ultrapure water, salinity 33.3) and Hepes (Fisher Scientific, Hampton, NH, USA; final concentration: 20 mmol l−1). The PO2 of this solution was adjusted by changing the gas concentrations and flow rate of aeration (standard aquarium pump) in a solution reservoir; seawater was cooled to ecologically relevant temperatures [range of average experimental temperatures=14.1–16.4°C (D. opalescens); 13.9–15.8°C (O. bimaculatus); 13.9–15.9°C (M. gracilis); 13.9–15.3°C (P. planipes)] in an ice bath and controlled on the stage by a heater. The average pH of the solution in all experiments was 8.04±0.04. This range is within the diel variability of nearshore Southern California Bight waters; however, under natural conditions, pH would decline with decreasing oxygen in the water column (Frieder et al., 2012). Oxygen concentration was measured in the recording chamber during all experiments using a dip optical probe (DP-PSt7; PreSens, Regensburg, Germany; Fig. 1A). Oxygen was converted to PO2 (kPa) with the ‘Respirometry’ package in R using the corresponding temperature and salinity from each experiment. All results are given in PO2 values, with concentrations (μmol kg−1) given in parentheses; all concentrations are averages of all individual trials within each species, corresponding to the temperature conditions reported above for each species.
Each larva was immobilized and its dorsal surface was attached to a borosilicate pipette using cyanoacrylate (Loctite superglue, Westlake, OH, USA) and then immediately submerged into solution on the recording chamber (using micromanipulator b2; Fig. 1A). ERG recordings were made using a whole-cell patch clamp electrophysiology amplifier to detect and record extracellular changes in potential in the eye through a standard extracellular glass electrode. Electrodes were borosilicate pipettes pulled to a resistance of 3–6 MΩ and filled with external recording solution. Electrode tips were placed into the eye using a digital micromanipulator (micromanipulator b1; Fig. 1A; Scientifica, Clarksburg, NJ, USA). No suction or pressure was applied to the recording electrode at any point during the experiment. The ERG signal was recorded in voltage-clamp mode using a Multiclamp 700B amplifier (Molecular Devices, San Jose, CA, USA) low-pass filtered at 4 kHz (Bessel), digitized at 20 Hz using an Instrutech ITC-18 A/D board (HEKA Elektronik, Holliston, MA, USA), and saved to a computer hard drive using the custom acquisition software writing in IgorPro (WaveMetrics, Lake Oswego, OR, USA). After obtaining an ERG recording and before beginning PO2 manipulation, larvae were held in the perfusion reservoir at normoxia (equivalent to 100–105% O2 saturation at ∼15°C and 33 salinity) to ensure there was a stable ERG response. After a stable baseline was obtained, the PO2 was decreased in the recording chamber with the addition of nitrogen gas (N2) to the solution reservoir. After obtaining a minimum PO2 value, the PO2 was then increased to normoxia by adding air (21% O2) to the reservoir (Fig. 1B).
Light stimuli were generated using a collimated green super-bright T-1 3/4 package LED (525 nm; 35 nm FWHM; Thorlabs LED528EHP; Newton, NJ, USA) focused through a 2× air objective that illuminated the entire stage; stimulus irradiance was adjusted by pulse width modulation (20 kHz duty cycle) through a computer-controlled constant current driver. Irradiance (photon flux) was measured at the experimental plane with a radiometer (Thorlabs), and converted into a species-specific irradiance in units of equivalent PFD (μmol photons m−2 s−1) for the spectral sensitivity of each species (e.g. squid PFD). Data for spectral sensitivities were obtained or modified from existing literature for the same species, or a taxonomically related species with similar life history and habitat depth: D. opalescens from sensitivity of Doryteuthis pealeii (Hubbard et al., 1959); O. bimaculatus from sensitivity of O. vulgaris (Brown and Brown, 1958); P. planipes (Fernandez, 1973); and M. gracilis from Cancer irroratus (Cronin and Forward, 1988). Spectral sensitivity curves were multiplied against the spectrum of the experimental light (LED) to obtain a species-specific irradiance for each species (Fig. S1). All light stimuli were presented from a dark background (no visible light), and animals were held in darkness between stimulus presentations. The term ‘darkness’ for this study refers to both the absence of light stimuli and the absence of environmental light within the light-tight experimental enclosure.
Three experimental irradiance manipulations were used. The time series test recorded ERG responses to a 1 s square step of light at a constant irradiance of 3.56 μmol photons m−2 s−1 repeated every 20 s, providing a nearly continuous measure of ERG response during the experimental manipulation of PO2. Two additional tests were conducted at specific oxygen conditions [normoxia (∼22 kPa/∼265 μmol kg−1), intermediate reduction of PO2 (∼6.5 kPa/∼95 μmol kg−1) and low PO2 (∼3.5 kPa/∼55 μmol kg−1); Fig. 1B]. The light series test consisted of square step pulses of irradiance (1 s light stimulus every 7 s) increasing from dim light to bright light (0.056–3.53 μmol photons m−2 s−1) at nine equally spaced irradiance increments, repeated three times with 20 s in between each series at each oxygen condition (Fig. 1C). The temporal test consisted of a chirped (1–20 Hz) square wave modulated between darkness and an irradiance of 3.26 μmol photons m−2 s−1 (Fig. 1D). Time series and light series tests were completed on larvae of all four species. The temporal test was completed on D. opalescens paralarvae and P. planipes larvae due to the lack of availability of O. bimaculatus paralarvae or M. gracilis megalopae at the time of experiments. All oxygen values were measured directly on the stage throughout visual tests. Experiments were conducted in vivo, with 100% survival of all larvae throughout the duration of the experiment.
Analysis of results
All electrophysiology data were analyzed using the software IgorPro (WaveMetrics). All waves were down-sampled to 2 kHz, digitally filtered with a binomial smoothing algorithm (IgorPro) with a corresponding Gaussian filter cut-off frequency of 40 Hz, and digitally notch filtered (60 Hz) before analysis. For all square waves, the amplitude of the response and the integrated area under the waveform were calculated. Within a species there was no difference in results when the measurement of the amplitude or the area was used, but because of the differences in waveform shapes between species, the integrated measurement was used for all final results.
For time series data, all measurements were normalized to the average of the ERG response in normoxia during the 5 min prior to the initiation of oxygen decline. Each normalized ERG measurement was matched to the corresponding oxygen measurement, and ERG responses were averaged over every minute to smooth the data. Oxygen metrics for retinal function, V90, V50 and V10, defined as the PO2 where there was 90%, 50% and 10% retinal function, respectively, were calculated for each trial and averaged across individuals. Statistical differences between metrics (V90, V50 and V10) were determined using Kruskal–Wallis one-way ANOVAs within each species (d.f.=2 for D. opalescens, O. bimaculatus and M. gracilis; d.f.=1 for P. planipes). Pairwise differences between metrics (e.g. V90 versus V50, etc.) were determined using Dunn’s test with Bonferroni correction for multiple comparisons (d.f.=2 for D. opalescens, O. bimaculatus and M. gracilis; d.f.=1 for P. planipes).
For light series data, three repeated tests were averaged at each oxygen condition. Both the amplitude of and area under the response waveform were calculated, and the integrated area was used for final analysis as explained for the time series data. ERG responses to stimuli at each irradiance were normalized to the maximum response (during normoxia at the highest irradiance). Response–irradiance curves were fit with a Hill equation for the averages of each species, as is often used to describe visual response–irradiance functions (Shapley and Enroth-Cugell, 1984; Oesch and Diamond, 2011). Within each species, Kruskal–Wallis one-way ANOVAs were conducted to determine differences between ERG responses at each oxygen condition (normoxia, ∼22 kPa/∼265 μmol kg−1; intermediate reduction of PO2, ∼6.5 kPa/∼95 μmol kg−1; and low PO2, ∼3.5 kPa/∼55 μmol kg−1) at each irradiance (d.f.=2; Table S1). To determine whether changes in retinal function in different oxygen conditions were consistent across irradiance (i.e. whether the shape of the response changed with oxygen condition), values were also normalized to the maximum value (ERG response at highest irradiance) within each oxygen condition. Oxygen values presented are averages from all trials within each species.
For temporal response analysis, three presentations at each oxygen concentration were averaged. In some cases, low oxygen reduced the amplitude of the response so that it became indistinguishable from the baseline noise. Therefore, only data where the light response was greater than 2 standard deviations of the baseline noise was used. Power spectral densities (PSDs) were calculated using the fast Fourier transform (window size=4000). The resulting PSD was normalized to the value at 1 Hz and converted to gain (dB). Results were analyzed for significance using a Kruskal–Wallis one-way ANOVA on the cut-off frequency (–6 dB) at each oxygen condition for each of the two species (D. opalescens and P. planipes) tested. Oxygen values (in partial pressure and concentration) presented are averages from all trials within each species.
The potential for loss of retinal function from PO2 in the environment in the Southern California Bight was calculated using the physiological threshold data collected in this study and the oxygen concentration data collected via CTD (conductivity–temperature–depth) casts made during California Cooperative Oceanic Fisheries Investigations (CalCOFI) cruises. Data from springtime (March–May) cruises conducted between 2005 and 2017 were downloaded from the CalCOFI website (calcofi.org) and casts closest to the animal collection site for these experiments (line 93.3 station 26.7 and 28) between 2005 and 2017 were averaged. Oxygen data from these casts were converted from concentration (µmol kg−1) to PO2 (kPa) units using the R package ‘AquaEnv’ and code from Hofmann et al. (2011). ERG response data from physiology experiments and the corresponding PO2 were fit with the best-fit model for each species (linear for D. opalescens and M. gracilis and nonlinear for O. bimaculatus and P. planipes) to calculate the predicted retinal function at each 1-m depth bin.
Sensitivity to light
During a continuous decline in PO2 from 22 kPa (280 μmol kg−1 O2=normoxia) to ∼3 kPa (∼45 μmol kg−1), the amplitude of the ERG to a 1 s square step pulse of light from darkness to a constant irradiance of 3.56 μmol photons m−2 s−1 decreased by 60–100% relative to responses in normoxia in all species (‘time series test’; Fig. 1B). The magnitude of retinal impairment and the PO2 at which the decline began differed among species (Fig. 2). The calculated oxygen metrics for retinal function show declines across all species as PO2 decreases, with significant differences between V90, V50 and V10 within a species in D. opalescens (P=0.012) and M. gracilis (P=0.006), but not in O. bimaculatus (P=0.156) or P. planipes (P=0.655, Kruskal–Wallis tests; Fig. 2).
Surprisingly, retinal function (V90) began declining at relatively high PO2 (only 1–2 kPa/20–30 μmol kg−1 below oxygen saturation) in D. opalescens (V90=22.2 kPa/258 μmol kg−1) and M. gracilis (V90=19.4 kPa/229 μmol kg−1). In contrast, oxygen thresholds for vision in O. bimaculatus (V90=11.5 kPa/133 μmol kg−1) and P. planipes (V90=5.7 kPa/68 μmol kg−1) were at lower PO2 values. Retinal function continued to decline with further reductions of oxygen in D. opalescens, O. bimaculatus and M. gracilis, and the PO2 where only 50% ERG function remained (V50) for each species was 13 kPa (151 μmol kg−1), 7.2 kPa (85 μmol kg−1) and 10.2 kPa (121 μmol kg−1), respectively (Fig. 2). In all cases, the ERG response returned to at least 50% of the maximum response (relative to the initial responses in normoxia) after re-oxygenation of the solution (Fig. 3), indicating the decline in ERG response during exposure to reduced PO2 was not from the death of the retinal tissue.
To determine whether the oxygen effects were dependent on light level, we presented light steps over a range of irradiance from 0.056 to 3.53 μmol photons m−2 s−1 at three different oxygen conditions (normoxia, intermediate reduced PO2 and low PO2; ‘light series test’; Fig. 1C). There was a decrease in ERG amplitude across all irradiances tested in all species as PO2 decreased (Fig. 4), similar to what was observed in the first experiment, indicating that declines in ERG responses observed in low PO2 were not irradiance-dependent (Fig. 4A,C,E,G). At each irradiance tested, ERG responses were significantly different between the ERG response at normoxia, intermediate reduced PO2 and low PO2 (P<0.05, Kruskal–Wallis tests; Fig. 4, Table S1), with the exception of the lowest irradiance (0.056 μmol photons m−2 s−1) in larvae of D. opalescens, O. bimaculatus and P. planipes (P=0.246, 0.301 and 0.105, respectively). To quantify the response–irradiance relationship (ERG response at each irradiance), ERG responses at each oxygen condition were fit with a Hill equation (Fig. 4A,C,E,G). To examine how the shape of the response–irradiance relationship was influenced by oxygen, we scaled the responses to the maximum ERG response within each oxygen condition (Fig. 5). Small changes in the shape of the response–irradiance relationship were seen in the cephalopods (D. opalescens and O. bimaculatus), but differences in the ERG response across oxygen conditions at each irradiance were not statistically significant in any species, indicating the response–irradiance relationships were stable at different oxygen conditions (P>0.05, Kruskal–Wallis tests; Table S1).
To determine how PO2 affects the temporal properties of the larval ERG response, we presented square wave linear chirp stimuli (frequency modulated between 1 and 20 Hz) between darkness and a constant irradiance of 3.56 µmol photons m−2 s−1 in larvae of D. opalescens and P. planipes (‘temporal test’; Fig. 1D) and measured the corresponding ERG response. With this recording of ERG response to flashes of light at multiple frequencies, we computed the PSD to determine the power of the visual signal at each frequency. As expected, there was a decline in the power of the response as frequency increased, indicative of the natural temporal resolution limit for the species (Fig. 6). We quantified the temporal resolution using a cut-off frequency (frequency at which power drops below −6 dB). During exposure to low PO2, we observed a steeper decline in the power of the response in larval D. opalescens, with a cut-off frequency decreasing from 4.6 Hz at normoxia (23.1 kPa) to 2.8 Hz at low PO2 (3.8 kPa; P=0.009, Kruskal–Wallis test; Fig. 6). No significant change in temporal resolution was observed with a change of PO2 in P. planipes (7.5 Hz at 23.4 kPa to 6.8 Hz at 3.6 kPa; P=0.755, Kruskal–Wallis test; Fig. 6). This indicates that the temporal resolution of vision in paralarvae of D. opalescens was reduced when exposed to low PO2, but that larvae of P. planipes were not significantly affected within the range of PO2 tested here.
Based on studies in terrestrial vertebrates, we expected that large decreases in oxygen would reduce ERG responses; however, for most marine species, the magnitude of the reduction in oxygen needed to impact retinal function is unknown. It is also unknown whether the decline in retinal function occurs at some threshold oxygen concentration or more continuously as oxygen is decreased below some critical threshold. To our knowledge, no direct comparisons of oxygen effects on visual physiology between different groups of marine invertebrates exist. These results demonstrated major retinal impairment in three species of marine invertebrate larvae after exposure to surprisingly minor amounts of oxygen decline (decreases of 1–2 kPa/20–30 μmol kg−1 from oxygen saturation). Interestingly, there were large differences in visual sensitivity to low PO2 among species, with almost 100% loss of retinal function in larvae of D. opalescens, O. bimaculatus and M. gracilis at low PO2 (∼3 kPa), whereas retinal function in larvae of P. planipes was relatively unaffected (i.e. ERG responses never declined enough to define a V10 within the range of PO2 tested). Additionally, during exposure to a decline in PO2, the retinal responses decreased continuously in D. opalescens paralarvae and M. gracilis megalopae, whereas O. bimaculatus paralarvae and P. planipes larvae were able to maintain ≥90% retinal responses until ∼11 and ∼8 kPa, respectively, before ERG responses began showing the effects of decreased oxygen availability (Fig. 2).
The two species tested for temporal resolution also had very different responses to exposure to low PO2. Paralarvae of D. opalescens showed a strong decrease in the power of the retinal response at higher frequencies during exposure to low PO2, whereas the larvae of P. planipes showed very little change in temporal resolution. These differences in retinal responses across species are likely due to different metabolic tolerances to low oxygen; however, data for critical oxygen thresholds for metabolism (as in Seibel et al., 2016) do not yet exist for larvae in these species. In addition, the decrease in retinal function after only minor declines in PO2 suggest that oxygen effects on vision may be an important sublethal physiological effect of low oxygen that may not be captured completely by a metric such as the critical oxygen thresholds for metabolism.
The declines in retinal responses for these invertebrate larvae during exposure to reduced oxygen are comparable to what is reported for terrestrial mammals. For example, the ERG response in cats began decreasing almost immediately after a decline in oxygen started, at PO2 values similar to what would be experienced if a human were to drive from sea level to approximately 2000 m elevation (e.g. Lake Tahoe in California) (Linsenmeier et al., 1983; Steinberg, 1987). The effects of oxygen on the visual system of the cat were noted to occur at PO2 values much greater than when effects would be observed in other neural circuits (Linsenmeier et al., 1983). The PO2 values that cause a reduction in marine invertebrate larval vision are well within the range of variability they experience in their natural environment. For example, the average daily range of oxygen (caused by both biological and physical forcing) in coastal areas of the Southern California Bight is ∼63 μmol kg−1 (∼4 kPa at 15°C) at 7 m depth (Frieder et al., 2012); our results suggest that this magnitude of variability (ΔPO2=4 kPa), even at high PO2 (21–17 kPa), could cause a 10–20% decrease in retinal function in larvae of D. opalescens and M. gracilis (Fig. 2) if they did not move upward to better-oxygenated waters. In addition, using the visual sensitivity of each species to PO2 reported above (e.g. Fig. 2), we calculated the decrease in retinal function with depth under present-day (2005–2017) average springtime ocean oxygen conditions in the Southern California Bight. The decline in PO2 alone from 0 to 30 m depth would decrease retinal function by 15–59% in larvae of D. opalescens, M. gracilis, O. bimaculatus and P. planipes in coastal Southern California (Fig. 7), even without including the effects of decreasing irradiance on the ERG response. Such changes in oxygen levels could significantly reduce larval fitness through loss of retinal function; however, the manifestation of the physiological impairment in visual behavior is unknown.
Many marine larvae remain at depth during the day to avoid visual predation, often at the depth near their threshold for light detection; they then migrate vertically to near-surface waters at night when predation pressure is reduced (diel vertical migration) (Forward, 1988; Sulkin, 1984; Zeidberg and Hamner, 2002). In all species, but especially in larvae of D. opalescens, O. bimaculatus and M. gracilis, retinal sensitivity to light declined during exposure to reduced oxygen (Fig. 2). Larvae of O. bimaculatus also experienced a decline in the range of irradiance that was physiologically detected, indicating that the sensitivity to changes in light irradiance (contrast) will be reduced during exposure to reduced PO2. For example, the shadow of a predator may become undetectable. Marine larvae require sufficiently high temporal resolution (the ability to distinguish between stimuli varying in time) to detect and appropriately respond to predators or prey (Frank, 1999). The decline in temporal resolution under reduced PO2 in larvae of D. opalescens would reduce the ability to detect high-frequency movements, such as the burst-swimming pattern of the copepods they feed on. Decreased light sensitivity and temporal resolution induced by low oxygen may introduce a greater risk for predation, increase vulnerability to starvation (if prey detection is reduced) and/or potentially weaken detection of the light cue for vertical migration entirely if the decrease in retinal response is sufficient to change visual behaviors. All individuals recovered some level of retinal function with re-oxygenation of the solution after the acute exposure time of the experiments (∼30 min at reduced oxygen; Fig. 3), indicating that the retinal response may recover during the time required for a small larva (∼1.5–3 mm) to swim a few meters.
These data also have important practical implications for researchers studying these organisms in a laboratory setting. The significant changes in retinal function after even small depletions in oxygen availability for some species suggest that it is important to monitor and maintain appropriate oxygen levels during in vitro and in vivo experiments.
Knowledge of how the retinal function of marine invertebrate species changes during exposure to reduced PO2 may help define species-specific vulnerabilities and resilience to future oxygen loss in the ocean. Globally, oxygen declines have resulted from warming (Schmidtko et al., 2017) and from nutrient and organic loading (Breitburg et al., 2018), and these losses can be exacerbated in areas with naturally occurring coastal hypoxia and upwelling (Altieri and Gedan, 2015; Levin and Breitburg, 2015). Given the apparent high visual sensitivity and species-specificity of visual tolerance to low oxygen, documenting how larval (and adult) vulnerabilities to low oxygen are manifested in visual behaviors and ecology will be important for predicting responses to global and local declines in ocean oxygen.
We thank P. Zerofski for help with larval collections, and J. H. Cohen, D. Deheyn, T. Martz, F. Powell and M. Tresguerres for comments on data and analysis, as well as two anonymous reviewers, who helped improve the manuscript.
Conceptualization: L.R.M.; Methodology: L.R.M., N.W.O.; Software: L.R.M., N.W.O.; Validation: L.R.M., N.W.O.; Formal analysis: L.R.M.; Investigation: L.R.M.; Resources: L.R.M., L.A.L., N.W.O.; Writing - original draft: L.R.M.; Writing - review & editing: L.R.M., L.A.L., N.W.O.; Visualization: L.R.M.; Supervision: L.A.L., N.W.O.; Funding acquisition: L.R.M., L.A.L., N.W.O.
This research was supported by a Charles H. Stout Foundation grant, a Frontiers of Innovation Scholars Program grant from the University of California San Diego, and the Scripps Institution of Oceanography education office for support to L.R.M., a National Science Foundation Graduate Research Fellowship grant DGE-1144086 to L.R.M., and a National Science Foundation grant OCE-1829623 to L.A.L. and N.W.O.
The authors declare no competing or financial interests.