Aerial single target acuity of harbor seals ( Phoca vitulina ) for stationary and moving targets of varying contrast

Harbor seals ( Phoca vitulina ) need to detect single objects for example when orienting to landmarks or hunting prey. The detection of single objects, described by the single target acuity (STA), cannot be deduced from formerly determined grating acuity (GA) as different mechanisms underlie STA and GA. Thus, we assessed STA for stationary and moving single targets with varying contrast in two harbor seals in a first approach in air. In a two-alternative-forced-choice discrimination task, the seals had to indicate whether the single target was presented in a left or right stimulus field on a monitor. The STA for full-contrast stationary targets was determined as 0.27 deg of visual angle for both experimental animals. Contrary to our expectations, neither adding motion nor reducing contrast had a strong impact on STA. Additionally, we also determined GA in the two harbor seals (1.2 and 1.1 cycles/deg or 0.42 and 0.45 deg for a single stripe of the grating at threshold) to be slightly inferior to STA. Our results are in good correspondence with contrast sensitivity and allow calculating viewing distances in the context of for example visual orientation.


Introduction
Harbor seals (Phoca vitulina) are amphibious marine mammals with well-developed eyes.Although numerous researchers have questioned the relevance of vision for seals, recent systematic research revealed that the harbor seals' eyes have specific adaptations to achieve good resolution while at the same time providing the sensitivity for dim light vision (for review see Hanke and Reichmuth, 2022).Most likely vision, along with the other senses, serves to detect objects in the animal's environment.Objects in the environment that a harbor seal could benefit from detecting include landmarks, predators, conspecifics, or prey, and even perhaps objects as small as stars (Mauck et al., 2005;Mauck et al., 2008) or particles dissolved in the water column and the motion pattern elicited when passing by the particles (Gläser et al., 2014), described as optic flow (Gibson, 1950;Gibson, 1958).
The visibility of an object is influenced not only by its contrast but also by its visual angle; the visual angle is defined as the angle under which an object of a specific size is seen by an observer from a designated distance.Previous research including harbor seals considered both parameters.The harbor seal's contrast sensitivity (CS) function was documented (Hanke et al., 2011) as a measure for 'real-world vision', meaning the ability to see objects encountered in daily life (see for example Owsley and Sloane, 1987;Woods and Wood, 1994).Additionally, the harbor seal's ability to resolve gratings in air and underwater (grating acuity; GA) revealed the harbor seal eye to resolve equally well in air under bright light conditions and underwater in clear water; the maximum GA amounted to 5.5 cycles/deg (Schusterman and Balliet, 1970;Weiffen et al., 2006;Hanke and Dehnhardt, 2009).
The good correspondence between aerial and underwater GA was unexpected due to the main problem of amphibious vision.In contrast to underwater, corneal refractive power contributes to vision in air.Thus, if a harbor seal was emmetropic, normal-sighted, underwater, the contribution of the cornea to aerial vision should render the eye myopic, short-sighted, in air.However, for harbor seals, pupillary and corneal characteristics were assumed to solve the problem of amphibious vision under specific light conditions (Hanke et al., 2006(Hanke et al., , 2009)): if the pupil is closed to a small pinhole in bright light, and even when it opens slightly to a vertical slit with decreasing ambient light, light is only entering the eye through the centrally flattened cornea (Hanke et al., 2006).Additionally, a small pupil maximizes visual acuity in general.Under these conditions, harbor seals were shown to achieve an aerial GA comparable to the underwater GA (Schusterman and Balliet, 1970;Weiffen et al., 2006;Hanke and Dehnhardt, 2009).Only when the pupil dilates circularly under very dim light conditions, aerial GA decreases (Hanke and Dehnhardt, 2009), most likely as the light can also enter the eye through the peripheral cornea, which is curved in the horizontal and vertical meridian, and as generally any irregularities of the optics maximally affect visual acuity negatively.We would expect underwater GA to be less affected by ambient light comparable to findings in California sea lions (Schusterman and Balliet, 1971); a topic, however, that still needs to be addressed in harbor seals.
The studies on GA and contrast sensitivity in harbor seals involved the presentation of gratings, and thus describe how well a harbor seal can resolve details in a complex natural scene.However, single objects of interest to harbor seals may not always be integrated in a cluttered scene and could instead stand out from plain background.The detection of single objects cannot be deduced from GA, as different mechanisms are responsible for resolving gratings versus single objects.The resolution of gratings is limited by the size of the receptive fields of the retinal sampling units (Land and Nilsson, 2012).However, if a single object is smaller than the receptive field of the sampling unit, its contrast determines whether it will be perceived or not (O'Carroll and Wiederman, 2014).The minimal size at which a single target is perceptible is defined by a species' single target acuity (STA).
In this study, we set out to assess aerial STA for two harbor seals using a circular dark single target on a bright background.To tie in with various previous studies on aerial vision in harbor seals (Scholtyssek, Kelber and Dehnhardt, 2008;Hanke and Dehnhardt, 2009;Hanke et al., 2011), we determined STA in air as a first approach.We used a circular dark single target, as studies on other organisms had tested STA with this type of target (see for example Busch and Dücker, 1987;Chaib et al., 2019;Chaib et al., 2021;Spratte et al., 2021).As STA is commonly determined with circular targets, STA is often also called 'dot acuity'.
We analyzed four aspects of or relating to harbor seal STA that we will elaborate on in the following three paragraphs.We documented the effect of contrast (motivation 1) and motion (motivation 2) on STA in harbor seals.STA for compact targets of different contrast is informative with respect to the visibility of objects in the seal's environment, as natural objects rarely have fully contrast to their background.Assessing the STA for targets varying in contrast additionally provides a meaningful comparison between a harbor seal's STA and their CS function, which, however, was assessed with gratings (Hanke et al., 2011).The determination of STA for moving compared with stationary targets seems interesting, as a harbor seal might want to detect moving as well as stationary single targets.Additionally, results from numerous studies indicate that a harbor seal's visual system is well-adapted to perceive motion (Hanke et al., 2008;Gläser et al., 2014;Weiffen et al., 2014).Overall, STA determinations from this study will allow the calculation of viewing distances under diverse external conditions for harbor seals.
Our third intention was to compare harbor seal STA with GA (motivation 3).Thus, we additionally determined GA in the same two harbor seal individuals under experimental conditions in which only stimulus presentation varied.We wanted to assess whether STA would be superior to GA, as for humans (Hecht, Ross and Mueller, 1947), or whether STA results would correspond well with GA, as previously documented in budgerigars (Melopsittacus undulates; Chaib et al., 2019), South African fur seals (Arctocephalus pusillus) and South American fur seals (Arctocephalus australis; Busch and Dücker, 1987).We measured pupil dilation during STA and GA testing to verify that any putative difference in STA and GA results had not been caused by a difference in pupil dilation.
We also determined STA with circular targets to compare with a previous study involving harbor seals (Jamieson and Fisher, 1970).The former researchers asked two harbor seals to detect a gap between two vertical bars which could be considered a determination of STA for a linear object.The threshold amounted to 0.03 deg of visual angle, irrespective of whether testing occurred in air or underwater, and was thus determined to be far superior than GA (Schusterman and Balliet, 1970;Weiffen et al., 2006;Hanke and Dehnhardt, 2009).The presentation of a linear object might have resulted in the very low harbor seal STA value, which corresponds well with results obtained in studies involving human participants (Hecht and Mintz, 1939;Hecht, Ross and Mueller, 1947); human STA for linear objects was superior to both the STA for square objects and the GA.Thus, with our approach involving a circular object, we could additionally assess (motivation 4) whether a change of stimulus type would lead to an aerial STA result different from or comparable to the results obtained by Jamieson and Fisher (1970).

Experimental animals
The study was conducted at the Marine Science Center of the University of Rostock, Germany.The experimental animals were two male adult harbor seals named "Nick" (born in 1999) and "Luca" (born in 2002).These two harbor seals were housed together with 9 harbor seals, two California sea lions (Zalophus californianus) and a South African fur seal (Arctocephalus pusillus) in a large seawater enclosure and were experiencing a natural day-and-night cycle.Both harbor seals were experimentally experienced as they had already taken part in different studies on vision, timing, or orientation (see for example Mauck et al., 2008;Scholtyssek, Kelber and Dehnhardt, 2008;Heinrich, Dehnhardt and Hanke, 2016;Heinrich, Ravignani and Hanke, 2020;Maaß and Hanke, 2021).Thus, from the previous experiments, the seals were familiar with the following aspects which are relevant for our study: (1) operant conditioning, (2) two-alternative-forced-choice (2AFC) experiments which include (2a) stationing in an experimental station, (2b) responding to a left or right response target, (3) visual experiments including (3a) working in an experimental chamber, (3b) paying attention to a monitor on which optic stimuli are being presented, (3c) circular optic stimuli, and (3d) responding to these stimuli according to the respective task.
The animals were fed 1-4 kg of herring and sprats per day depending on season and motivation.The main food amount was given during experimental sessions which were performed up to 6 days a week during daylight hours.
The experiments carried out in this study were in accordance with the European Communities Council Directive of September 22nd, 2010, (2010/63/EU) and the German Animal Welfare Act of 2006.The individuals used in the study were not subjected to pain, suffering or injury, therefore no approval or notification was required.

General experimental setup for STA and GA determination
The experimental sessions were performed in an experimental chamber (3 m deep, 2 m wide, 2 m high) to achieve a constant ambient illumination of 80 lx in the experimental area (Multifunctional Environment Measuring Instrument 4 in 1, VOLTCRAFT, Hirschau, Germany).During an experimental session, before the first trial and in between trials, the harbor seal had to place its head into a metal hoop station (Fig. 1).Stationing in the hoop ensured a constant viewing distance of 60 cm to the experimental monitor, on which the stimuli were presented (LG 24MB35PM, 24′' LED-Display, resolution: 1920x1080 pixel, contrast: 5.000.000:1,response time: 5 ms, refresh rate: 60 Hz, LG Electronics Deutschland GmbH, Eschborn, Germany).The monitor filled 47 deg of the visual field, thus 66 % of the binocular visual field of a harbor seal (Hanke, Römer and Dehnhardt, 2006).Two response targets were attached at a distance of 10 cm to the left and right of the hoop station respectively (Fig. 1).In every trial, the animal was moving its head to either the left or right response target to indicate the position of the positive stimulus (S+; see 2.3.) in either the left or right stimulus field on the monitor, in line with a 2AFC task.
During experimental sessions, the experimenter stayed in a separate observation room adjacent to the experimental chamber to avoid providing secondary cues.All technical equipment was located in this room, therefore experimental sessions could be operated entirely from it.To observe the harbor seal and its response behavior during the experimental sessions, a camera (Logitech C270 HD Webcam, max.resolution: 720p, diagonal field of view: 55 • , Logitech GmbH, München-Aubing, Germany) was installed in the experimental chamber (Fig. 1) and was connected to a laptop computer (BE163628 Lenovo ThinkPad T420, 14′' display, resolution: 1366x769 pixel, Lenovo Deutschland GmbH, Stuttgart, Germany) in the observation room.The observation room and experimental chamber were linked by an observation window, which could be closed by a black, opaque slider during trials.It was opened to communicate with and reward the harbor seal after a correct response.

Stimuli
The stimuli for all experimental sessions were programmed and displayed with Matlab (The MathWorks, Natick, Massachusetts, USA) and the associated Psychophysics Toolbox 3.0 (Brainard, 1997;Pelli, 1997;Kleiner et al., 2007).The photometric brightness of stimuli and background was measured with a luminance meter (LS-110, Minolta, Langenhagen, Germany).

Stimuli for STA determination with stationary targets
For the determination of the STA with stationary targets (needed to address motivation 1-4), a circular single target with square wave luminance profile was presented on a white background (luminance of 80 cd/m 2 ) in the center of the left or right stimulus field on the monitor (S + ), corresponding to the left or right half of the monitor.The other stimulus field remained white (S-).At the beginning of the experimental session and between each trial, the monitor was gray (luminance of 17 cd/m 2 ).
We used single targets of four different contrasts C1-C4 to the background (Table 1).For the full contrast (C1) target, a black single target was displayed on a white background (needed to address motivation 1-4).The lower contrast levels C2-C4 (needed to address motivation 1) were achieved by modifying the gray value of the single target, while the background always remained white.Contrasts were calculated on the basis of luminance measurements as Weber contrast, which is commonly used to describe contrast of single objects to their background (O'Carroll and Wiederman, 2014).
For STA determination, we used six single targets with preset diameters ranging from 3 to 51 pixels.The expected stimulus diameter in millimeters was calculated from the stimulus size in pixels, taking the monitor width of 527 mm and the preset resolution of 1280 pixels into account (Supplement Tab. 1).
As a control, the diameter of a single target was determined as full width at half maximum (FWHM) from luminance profiles of photos taken of each stimulus on the experimental monitor (preset resolution: 1280x720 pixel) with a digital camera (Canon Powershot V10, Canon Deutschland GmbH, Krefeld, photo resolution: 2400x1600 pixel, JPGformat).The smoothing property of the JPG-compression is advantageous for the determination of the brightness profile, compared to lossless formats like DNG (RAW) or PNG.
Each single target was photographed three times with millimeter paper attached to the monitor for scaling.The luminance profile of each photo was extracted three times in order to extract the gray values along the midline of the single target using ImageJ (version 1.53c, Wayne Rasband, National Institutes of Health, Bethesda Maryland, USA).The luminance profiles were then exported in OriginPro 2018b (version 9.55, OriginLab Corporation, Northampton, USA) for smoothening, sigmoidal curve fitting to luminance profile and determination of the FWHM.Please note, the photos were not calibrated in brightness, as absolute luminance values are not required for assessing the FWHM.Diameters of the single targets, as a mean of three measurements for each photo, were then converted in visual angles taking the viewing distance into account (see 2.4.1.,Table 2).The final measured stimulus sizes are in line with the expected stimulus sizes calculated by monitor width and resolution (Supplement Tab. 1).

Stimuli for STA determination with moving targets
For the determination of the STA with moving targets (needed to address motivation 2), a circular single target with square wave luminance profile was moving semi-randomly in a stimulus field of 100x100 pixels, corresponding to 4.1x4.1 cm or 3.9x3.9deg of visual angle (mfile adapted on the basis of a file kindly provided by S. Chaib, Lund).The stimulus fields were positioned in the center of the right and left half of the monitor.When the single target reached the invisible boundary of the stimulus field, it changed its direction smoothly.The direction taken by the target in a specific frame was normally distributed around the direction it had travelled during the previous frame.Stimulus velocity was set to 1.2 deg/s.The moving target was presented on a white background (luminance 80 cd/m 2 ).The target moving in one stimulus field on the monitor was defined as S +.The alternative stimulus field remained white (S-).
All other aspects of stimulus presentation were as described for STA determination with stationary targets (see 2.3.1.).

Stimuli for GA determination
To determine the GA (needed to address motivation 3), a horizontal and a vertical full contrast square wave grating (Table 1) were simultaneously displayed in two stimulus fields of 720x720 pixel size (19.7x19.7 cm or 18.7x18.7 deg of visual angle) on the right and left halves of the monitor, separated by a gap of 11 cm.The stimulus fields were surrounded by a gray frame (28 cd/m 2 ).The horizontal grating was defined as the S+, the vertical grating as the S-.At the beginning of the experimental session and after each trial, the stimulus fields turned light gray (12 cd/m 2 ).
The gratings presented during threshold determination had spatial frequencies ranging from 0.53 to 2.38 cycles/deg (see Table 2 for corresponding single stripe widths).The spatial frequency of a grating was Fig. 1.The general experimental setup for testing aerial STA and GA in harbor seals.A| side view, B| frontal view.The harbor seals were stationed in a hoop station at 60 cm distance from the experimental monitor (M) on which the stimuli (not shown) were presented.During the experimental sessions, the experimenter stayed in a separate observation room to prevent secondary cue giving.The room was connected to the experimental chamber by an observation window (OW).This window could be closed by a black, opaque slider during trials.To observe the actions taking place in the experimental chamber from the observation room, a camera (C) was installed in the experimental chamber.10 cm to the left and right side of the hoop station, response targets (RT) were installed.If the stimulus was presented in the left stimulus field on the monitor, the harbor seal had to touch the response target on the left side and vice versa.determined from three photos taken with a digital camera (Canon Powershot V10, Canon Deutschland GmbH, Krefeld, photo resolution: 2400x1600 pixel, JPG-format) of each stimulus presented on the experimental monitor.A luminance profile was extracted three times from each photo in ImageJ.The luminance profile data was exported to OriginPro 2018b for smoothening and sigmoidal curve fitting to luminance profile.The luminance profile of four adjacent stripes in the middle of the photo was used to calculate FWHM.Using the average width of the four stripes, the spatial frequency was calculated taking the viewing distance into account (Supplement Tab. 2).

General experimental procedure for STA and GA determination
The experimental procedure described in this chapter explains the general experimental procedure during the training and STA/GA threshold determination phase.Specifics of STA determination with stationary or moving targets as well as of GA determination will be mentioned in the subchapters 2.4.1.to 2.4.3.
At the start of the experimental session, the animal was guided into the experimental chamber, where it placed its head into the hoop station.Then, the experimenter closed the door of the chamber and went into the adjacent observation room to prepare the experimental session.After a time interval of approx. 2 min, the first trial was started.When stimulus presentation was started, the animals had to indicate whether the positive stimulus (S+; see 2.3) was presented in the left or right stimulus field on the monitor according to a 2AFC task.The animal indicated its decision by removing its head from the hoop station and by touching a response target on one side of the hoop.If it touched the response target on the side corresponding to the stimulus field on the monitor, on which the S + was presented, the answer was correct, the slider was opened and the animal got a fish reward.If the animal answered incorrectly, the experimenter said "no", and no reward was given.After the feedback, the monitor turned gray, and the animal had to re-station in the hoop station waiting for a new trial to start.
The position of the S + in the left and right stimulus fields of the monitor over the course of a session, was pre-determined after Gellermann (1933).One experimental session consisted of 36 trials, allowing for each of six stimuli to be presented six times per experimental session.The order in which the six stimuli were presented within an experimental session was randomized with the RAND() function in Excel (Microsoft Office Professional Plus 2016v. 16.0.4266.1001, 2016 Microsoft Corporation).For randomization, we divided an experimental

Table 2
Overview of the various stages of STA determination with stationary and moving stimuli as well as GA determination.For all stages, the number of stimuli, the sizes of the stimuli (for STA, the size represents the diameter of the circular single target in deg; for GA, the size represents the width of one single stripe of the stripe pattern in deg), and the number of sessions needed to reach the learning criterion of ≥ 80 % correct choices to be met in two consecutive sessions (including the two sessions with which the learning criterion was fulfilled).Basic task and the stages of pretraining were conducted with full contrast (C1) stimuli for stationary and moving STA as well as with full contrast square wave gratings for GA.For stationary and moving STA, we determined STA thresholds for four contrasts in randomized order until our criterion set to end repeated threshold determination and to obtain the final STA value was fulfilled.session of 36 trials into three blocks of 12 trials each.Each of the six stimuli were shown once in the left stimulus field and once in the right during a block of 12 trials in a randomized order.This was to ensure that putative differences in motivation would not only affect performance with respect to single stimuli.
We measured pupil diameter during STA and GA determination to assess whether the changes in stimulus presentation had an impact on pupil dilation, assumed to be a critical parameter for aerial vision in harbor seals (see Introduction).Pupil diameter was measured on six photos of the right eyes of both animals taken with a Motorola moto g200 5G triple camera (108 + 2 + 13 Megapixel, photo resolution 4000x3000 pixel, Motorola Mobility LLC, Libertyville, Illinois) in two experimental sessions.The photos were imported in ImageJ (version 1.53c, Wayne Rasband, National Institutes of Health, Bethesda Maryland, USA) to measure the vertical and horizontal pupil diameter three times on each photo.A white paper square of 5 mm side length was glued next to the eye and served as a scale to convert pixel values in mm (Supplement Tab. 3).

Experimental procedure for STA determination with stationary targets
For the determination of the STA with stationary targets, the animal's task was to indicate the position of the S+, meaning the stimulus field containing the circular target (see 2.3.1):if the S + was presented in the left stimulus field, the harbor seal had to move its head to the response target on the left side of the hoop station to obtain a reward and vice versa.
STA training (for overview see Table 2) was started with a full contrast stationary stimulus (C1) clearly above threshold (basic task).After task acquisition with this stimulus, we asked the seals to generalize the procedure by presenting high contrast stimuli of different diameters first successively and then in one session within the generalization phase of pretraining (for overview see Table 2).During all pretraining phases, a new phase was started when the learning criterion, defined as a performance of ≥ 80 % correct choices in two consecutive sessions, was reached.
For STA determination, we followed the method of constant stimuli, meaning we used six single targets with preset diameters and a defined contrast, and presented these six stimuli 30 times each over the course of five sessions.The performance over the 30 stimulus presentations was summarized to obtain the psychometric function (plotting the performance of the animal averaged over the 30 presentations of the specific stimuli as a function of stimulus size in deg of visual angle) from which the threshold at a performance of 75 % correct choices was determined.The threshold was calculated by linear interpolation between the performance of the animal with respect to the last supra-and first subthreshold stimulus; the last supra-and first subthreshold stimuli refer to the targets for which the performance of the animal was just below and just above 75 % correct choices, respectively.
Once an STA threshold was determined, we started a new STA threshold determination.Altogether, we determined STA thresholds for stationary targets at least four times for each of the four contrasts for each animal (see Supplement Fig. 1).The STA thresholds for all contrasts were assessed in randomized order.Data collection was continued until the animal's performance did not improve over two consecutive STA threshold determinations, meaning the two thresholds adopted the same values or the second of the two thresholds was inferior to the first.We defined the final STA value for each contrast and each animal, which we report in the results section, as the best STA threshold reached by the animal during data collection.

Experimental procedure for STA determination with moving targets
As for the determination of STA with stationary single targets (see 2.4.1.),a head movement to the response target on the side of the hoop station that corresponded to the stimulus field shown on the monitor, which displayed the moving stimulus (S + ), was considered a correct response.
Because the animals were familiar with STA determination using stationary single targets, we could directly start STA threshold determination with a moving single target after two sessions (for overview see Table 2).STA thresholds for moving single targets were determined in the same manner as described for stationary targets (see 2.4.1.).All STA thresholds determined to obtain the final STA value for moving targets, reported in the results section, can be found in Supplement Fig. 1).

Experimental procedure for GA determination
The animal had to indicate the position of the horizontal grating (S + ) to obtain a reward: if the S + was shown in the right stimulus field, it had to move its head to the response target to the right of the hoop station and vice versa.
To train the basic task, gratings with a stripe width clearly above threshold were used (for overview see Table 2).After reaching the learning criterion as defined already in the training stages for STA determination (see 2.4.1), the animal had to generalize the procedure to gratings with different stripe widths.Therefore, the animal was first presented with gratings of a single stripe width and then of numerous stripe widths within one experimental session (for overview see Table 2).For seal Luca, well-versed with respect to optic stimuli, we skipped the generalization phase and directly started GA threshold determination due to time constraints unrelated to this study.
Once the training stages were complete, threshold determination was commenced.Threshold determination was as described for STA determination (see 2.4.1).The thresholds determined can be found in Supplement Fig. 2.

STA with stationary single targets
Both harbor seals rapidly learned the basic task in only 5 (seal Nick) and 10 (seal Luca) training sessions, respectively, and went into the generalization phase of pretraining (for overview see Table 2, learning curves in Supplement Fig. 3 and 4).This pretraining phase lasted 10 (seal Nick) and 40 (seal Luca) experimental sessions altogether before STA determination could be started.During STA determination, STAs reached a constant level within 4 to 7 STA threshold determinations (see Supplement Fig. 1).
Stationary STA for a full contrast (C1) single target was determined as 0.27 deg for both seals (Fig. 2).When contrast of the single target to the background was reduced, the STA for seal Nick slightly deteriorated to 0.30 deg for C2 single targets and 0.31 deg for C3 and C4 single targets (Fig. 2A).For seal Luca, the STAs for C2 and C3 single targets were slightly better than for C1 single targets, as a threshold value of 0.23 deg was reached.Seal Luca's STA worsened to 0.34 deg (Fig. 2B) for single targets with the lowest contrast (C4).

STA with moving single targets
The STA for moving single targets with full contrast (C1) to the background was assessed as 0.21 deg for seal Nick and 0.23 deg for seal Luca (Fig. 2).When lowering the contrast, seal Nick's performance deteriorated and reached an STA of 0.27 deg for C2 and C3 single targets (Fig. 2A).STA for the targets with lowest contrast (C4) was again slightly worse than for the other contrasts and amounted to 0.31 deg.
Moving STA for C2-C4 targets could not be determined for seal Luca as he refused to cooperate over a prolonged period of time, and it was finally decided that data collection with him would no longer continue.

GA
Both harbor seals learned the basic task in 9 (seal Nick) and 30 (seal Luca) experimental sessions, respectively (for overview see Table 2, learning curves in Supplement Fig. 5 and 6).Afterwards, the generalization phase of pretraining lasted 32 experimental sessions (seal Nick).
During threshold determination, GA using full contrast gratings was determined as 1.2 cycles/deg for seal Nick and 1.1 cycles/deg for seal Luca (Fig. 3).Thus, at threshold performance, the seals were able to perceive a single stripe subtending 0.42 deg and 0.45 deg of visual angle, respectively.

Pupillary measurements
For seal Nick, the mean horizontal pupil diameter (with standard deviation) was assessed as 1.82 ± 0.32 mm for the STA and 1.80 ± 0.30 mm for the GA experimental condition, while the mean vertical diameter (with standard deviation) was determined as 5.04 ± 0.42 mm for the STA and 5.05 ± 0.43 mm for the GA experimental condition (all pupillary measurements can be found in Supplement Tab. 4).For seal Luca, we measured a mean horizontal pupil diameter (with standard deviation) as 2.61 ± 0.12 mm for the STA and 2.64 ± 0.21 mm for the GA experimental condition.The mean vertical diameter (with standard deviation) of this harbor seal was determined as 5.79 ± 0.21 mm for the STA and 5.87 ± 0.45 mm for the GA experimental condition.Thus, the different stimulus conditions for STA and GA determination did not elicit different pupil dilation in the two harbor seals, although absolute values differed between the individuals.

Discussion
In this study, we set out to assess how small single dark objects can be to be perceivable for harbor seals.An improvement in performance was clearly visible over STA and GA threshold determinations; an effect that had already been documented in common sunfish (Lepomis gibbosus) and budgerigar STA testing (Chaib et al., 2019, Spratte et al., 2021).This change in perception with experience, which resembles 'perceptual learning', primarily defined for humans (Gibson, 1963), needs to be emphasized as it most likely also occurs in other behavioral experiments and thus needs to be considered in the experimental designs of future experiments.
We determined STA in harbor seals to be nearly unaffected by a reduction of contrast down to 0.33 (motivation 1).When lowering the contrast from C1 to C4, STA only slightly deteriorated by 0.05 deg for seal Nick and 0.07 deg for seal Luca for stationary single objects.For moving single objects, a 0.11 deg deterioration was shown for seal Nick.In contrast, Spratte et al. (2021) and Chaib et al. (2019), who had tested STA for stationary targets in common sunfish and budgerigars with different contrasts, concluded that a reduction of contrast had a clear negative impact on STA.This difference in results reflects the slightly and clearly lower CS in sunfish, as determined in bluegill sunfish (Lepomis microchirus; Northmore, Oh and Celenza, 2007), which is closely related to the common sunfish, and budgerigars (Lind and Kelber, 2011) in comparison to the CS of harbor seals (Hanke et al., 2011).As  speculated for predatory fish (Caves, Sutton and Johnson, 2017), high sensitivity might particularly be important, and possibly more so than high resolution for an aquatic predator such as a harbor seal as underwater prey might disappear from sight before falling under the resolution limit.
Our STA results from contrast reduction in harbor seals is consistent with the high-frequency cut-off of the harbor seal's CS function determined with gratings (Fig. 3; Hanke et al., 2011).This correspondence reflects that the detection of single objects is also limited by contrast sensitivity in harbor seals, as simple models suggest (O'Carroll and Wiederman, 2014).From Fig. 3, it is evident that the CS for single objects was slightly worse than the CS for gratings and that the GA determined herein is inferior than the resolution limit suggested by the CS function.Various factors such as the type of stimulus, different experimental conditions or different harbor seal individuals being tested might affect the absolute CS values and might thus be responsible for the observed difference.In humans, it was shown that for example the age of the participant (Derefeldt, Lennerstrand and Lundh, 1979) or ambient light (see e.g.De Valois, Morgan and Snodderly, 1974) could influence CS.A future study could potentially assess a full CS curve for compact targets to compare with the currently available CS curve obtained with gratings to evaluate which parameters affect CS curves and how.
When comparing moving with stationary STA, moving STA was slightly better than stationary STA (motivation 2).We would have expected to see a more pronounced difference in STA when adding motion due to the harbor seal's visual system perceiving motion well (Hanke et al., 2008;Gläser et al., 2014;Weiffen et al., 2014) and having a low brightness discrimination threshold (Scholtyssek, Kelber and Dehnhardt, 2008).Different results for stationary versus moving STA could be obtained, as stationary and moving objects are processed by different pathways with different properties at least in primates (Livingstone and Hubel, 1988).Particularly, CS, underlying STA, was found to be different for stationary versus moving stimuli in birds and humans (Haller et al., 2014;Burr and Ross, 1982;Kelly, 1979;Robson, 1966).In line with the good correspondence of moving and stationary STA assessed in budgerigars (Chaib et al., 2019;Chaib et al., 2021) and common sunfish (Spratte et al., 2021; M. Hoppe unpublished data), our results from STA determination with moving and stationary targets in harbor seals indicate that CS for moving does not strongly deviate from CS for stationary targets of high spatial frequencies; a systematic assessment of CS for various stimulus conditions could shed more light on this aspect in a future experiment.It needs to be clearly pointed out, however, that all the cited studies only analyzed STA with targets moving at a single velocity.It remains to be determined in a future experiment whether velocity or specifics of the stimulus such as its shape or movement pattern would have an impact on either moving or stationary aerial STA in harbor seals, or in any of the other species mentioned.
STA was slightly better than GA (motivation 3) when comparing the visual angle, under which a single stripe of the visual acuity grating at threshold was seen with the STA for full contrast single targets (Fig. 3).An even better correspondence between the two measures of resolution was documented for two otariids (Busch and Dücker, 1987), for budgerigars (Lind and Kelber, 2011;Chaib et al., 2019) and common sunfish (Spratte et al., 2021;M. Muck unpublished results).In contrast, a clear difference between the STA and GA was assessed for humans (Hecht, Ross and Mueller, 1947).A possible reason for the human STA values being superior to GA is the high contrast sensitivity of humans (see for example Bisti and Maffei, 1974); if contrast sensitivity is high, a single object which is far smaller than the receptive field of a retinal sampling unit is still detectable provided it has a high contrast to the background (O'Carroll and Wiederman, 2014).The lower CS in harbor seals versus humans (Hanke et al., 2011) might explain why STA was only slightly better than GA in our experimental animals.We can exclude the possibility that our results were caused by different pupillary states, as our pupillary measurements under STA and GA experimental conditions revealed no difference.
For the above comparison, we used the STA and GA assessed in this study under very similar experimental conditions as well as in the same harbor seal individuals.This approach was necessary as it was previously shown for GA (Hanke and Dehnhardt, 2009) that experimental conditions such as luminanceambient luminance and/or the luminance of the stimuli to which the animals are most likely adapted during experimentsaffect visual resolution of gratings.Moreover, visual resolution can differ between harbor seal individuals (Hanke and Dehnhardt, 2009;Weiffen et al., 2006); a phenomenon that we could however not document for our STA assessment.Due to individual differences and additional differences in for example stimulus presentation, we would not have been able to compare our STA results with GA values obtained in Hanke and Dehnhardt (2009).In general, comparisons across studies are always complicated due to numerous factors, which is the reason we adopted the direct approach.
We cannot confirm the results published by Jamieson and Fisher (1970) that suggested STA is much better than GA in harbor seals (motivation 4).The harbor seals of the former study were asked to detect a gap between two vertical bars.Thus, it is possible that long linear objects, as used by Jamieson and Fisher (1970) are detected better than compact single objects we used in our STA experiment.Two reasons may account for this phenomenon: first, for long linear objects, responses from numerous photoreceptor cells might be averaged (Hering 1899; but see Westheimer and McKee, 1977), and second, the overall difference in brightness between S + and S-is larger for long linear objects versus compact objects which could facilitate the discrimination of S + and S-.It would be interesting to ask our experimental animals to detect single lines against a bright background to assess whether we can replicate the results obtained previously for harbor seals and humans to determine whether STA was better than STA with compact single targets (Hecht, Ross and Mueller, 1947;Jamieson and Fisher, 1970).
In conclusion, harbor seals have the capacity to perceive single objects, irrespective of their contrast and motion, well and with a comparable precision as for objects within a cluttered scene at least under the light conditions of our study.The results of this study determine from which distance a single object is still perceivable by a harbor seal even under relatively low light conditions.Considering an STA of 0.3 deg, a harbor seal would be able to see a 1 m-object from a distance as large as 190 m under clear conditions.The ability to estimate viewing distances will be essential for future considerations regarding orientation and navigation with the help of for example visual landmarks (Maaß and Hanke, 2022;Maaß et al., 2022).Additionally, when extending STA determination to under water, we will be able to calculate viewing distances in the foraging context or to determine the space surrounding the head of the animal in which particles will, when passed by, elicit optic flow (Gläser et al., 2014).

Fig. 2 .
Fig. 2. STA (best value of at least four threshold determinations in deg of visual angle) for A| seal Nick and B| seal Luca for stationary (•) and moving (○) single targets of different Weber contrasts to the background (C1 = 1, C2 = 0.78, C3 = 0.53, C4 = 0.33).Thresholds with moving single targets for C2-C4 could not be recorded with seal Luca due to motivational aspects (see text).

Fig. 3 .
Fig. 3. Contrast sensitivity function (dashed line) of a harbor seal (data replotted from Hanke et al., 2011) assessed with gratings to which the GA (arrow), stationary STA (filled dots), and moving STA values (empty dots) assessed with objects of different contrast in the current study were added for A| seal Nick and B| seal Luca.

Table 1
Contrasts between single targets and background for STA and black and white stripes for GA determination as Weber contrast and Michelson contrasts.I max indicates the luminance of the white background/white stripes and I min the luminance of the single targets/black stripes.