Time-compensated sun compass in juvenile sprat (Sprattus sprattus) reveals the onset of migratory readiness

Numerous fish species undertake migrations of tens to thousands of kilometers to reach suitable spawning areas or feeding grounds (Spiecker et al., 2021). Though critical for effective management, factors that provide proximate orientation cues for migrating fish are not well understood. Insights into which signals the fish use and which factors influence migration and its triggering provide information about ecological and evolutionary processes and deliver important information for species protection and the management of fish stocks (Chapman et al., 2012).

Sprat (Sprattus sprattus) are small (16–18 cm) pelagic, schooling fish that are commercially and ecologically important in the Baltic Sea (Bailey, 1980; Rechlin, 1975). They have a fairly short life span of about 5 years. The distribution range of sprat in the Baltic Sea extends from the Belt Sea and the western Baltic Sea to the Quark area in the north and also to the northeastern part of the Gulf of Finland (Aro, 1989) (Fig. 1). Three different sprat stocks have been determined in the Baltic Sea: one in the Belt Seas, the Western Baltic and the region of the Bornholm Basin, the second in the Gdansk Deep and Gotland Deep area, and the third in the Northwestern and Northern Baltic as well as the Gulf of Finland (Aro, 1989). Boundaries between stocks are not clear, and stocks intermingle during the summer feeding and overwintering seasons (Hinrichsen et al., 2005; Rechlin, 1975). The distribution of sprat in the Baltic Sea is mainly determined by winter temperatures and reproduction requirements (Ojaveer, 2017). The main feeding season in the central and southern Baltic Sea is from July to November, when sprat migrate after spawning to more offshore regions of the Baltic to reach good feeding grounds and form larger feeding schools (Aro, 1989; Mollmann et al., 2004). Although feeding and spawning migration routes are suggested, the extent and magnitude of migration patterns of different stocks are not well known (Aro, 1989). Sprat typically spawn in the pelagial in the deep basins in the central Baltic and coastal slopes of the deeps, additionally, in Kiel Bay, Mecklenburg Bay, Arkona Basin, Bornholm Basin and Gdansk Deep (Alheit, 1988; Aro, 1989; Grauman, 1975; Krenkel and Hinrichs, 1979; Ojaveer and Kalejs, 2009) (see Fig. 1). Spawning in the Western Baltic begins in March–April and lasts until July–August and is usually longer in the South than in the Northern Baltic (Aro, 1989; Grauman, 1975). Sprat are serial spawners with several batches per spawning season (Alheit, 1988). In spring, they spawn the pelagic eggs near the bottom and in summer they tend to spawn in the surface layers above the deep (Aro, 1989; Grauman, 1975).

Performing mid-range migrations over several hundred kilometres to get from their overwintering and feeding areas to their spawning grounds and back requires appropriate cues for orientation and navigation. Fish can use a range of sensory cues, such as olfaction (Gerlach et al., 2007), sound (Radford et al., 2011), and omnipresent cues such as the Earth's magnetic field (Bottesch et al., 2016; Cresci et al., 2019) and celestial cues, such as the sun (Mouritsen et al., 2013; Quinn, 1980; Spiecker et al., 2022). Among the various orientation mechanisms, using a sun compass to maintain a constant bearing requires a time compensation to balance the sun's movement during the day. Therefore, it is necessary that animals observe the path of the sun and link the sun's azimuth at different time points during the day to their circadian clock (Massy et al., 2021; Mouritsen, 2018, 2022; Mouritsen et al., 2013; Schmidt-Koenig et al., 1991). In this study, we were interested in whether juvenile sprat use a time-compensated sun compass for orientation, similar to other Clupeidae, such as Atlantic herring (Spiecker et al., 2022). We tested two different age groups of juvenile sprat to investigate their ability to use sun compass orientation before and directly at the beginning of their migration towards feeding and overwintering grounds.

All animal procedures were approved by the Animal Care and Use Committees of the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (LAVES, Oldenburg, Germany; Az.: 33.19-42502-04-17/2721) and performed in accordance with the relevant guidelines and regulations. Permission for scientific fishing to catch fish using fishing rods, fish traps and sink nets was granted by the Schleswig-Holstein State office for agriculture, environment and rural areas to GEOMAR.

All sprat [Sprattus sprattus (Linnaeus 1758)] were caught in July–August 2021 and July 2022 in Kiel Fjord (54°19′46.0″N, 10°08′58.1″E) with hand nets from the pier in front of the GEOMAR Helmholtz Centre for Ocean Research Kiel west shore building. Before transport to Oldenburg, where the experiments were conducted, the fish were kept in flow-through tanks with Baltic Sea water according to Kiel Fjord conditions: 15–17 PSU and 17°C for 2 weeks in 2021 and 1 week in 2022.

In Oldenburg (180 km southeast of Kiel), the fish were separated in two groups (for sun compass and time-shift sun compass experiments) and transferred into 200 liter glass aquaria with the same water conditions as their previous holding water (15–17 PSU and 17°C). A 14 h:10 h light:dark cycle was used with lights-on at 06:00 h and lights-off at 20:00 h for the control sun compass experiments, and lights-on between midnight and 14:00 h for time-shift conditions. Curtains covering the aquaria prevented the incidence of light. Fish of the time-shift group were clock-shifted by 6 h for at least 7 days before starting the experiments. Owing to the time shift, the internal clock gives the animals a different time of day than in reality; the orientation of the animals should change accordingly, which would be an indication of a time-compensated sun compass orientation. For example, if the animal was tested at 08:00 h, the fish would assume it was 14:00 h and orient itself according to this time shift.

The tanks were filled with natural seawater similar to the sampling site conditions, temperature was controlled by a TECO Aquarium Cooler TK500 and an external filter was used for continuous filtration (Eheim Professional). The fish were fed with Artemia sp. shrimp, enriched with Aqua Biotica Orange+ (Mrutzek Meeres-Aquaristik) once per day after the experiments, when lights were on for both groups.

In 2021, 21 young sprats were tested in the experiments with the control sun compass and 9 sprats were tested with a time shift. In 2022, the tests were conducted exclusively under time-shift conditions; 29 sprats were tested at the beginning of August, and 18 sprats were tested at the end of August. The experimental setup corresponded to our sun compass experiments with young herring described in Spiecker et al. (2022). During the course of the experiments, the fish were kept in bowls filled with natural seawater. Every 30 s for 20 min a picture was taken with a GoPro camera (GoPro Hero 4 silver), fixed in place underneath the bowl with a free view to the sky (Fig. 1, right corner) and aligned towards the north, resulting in 40 pictures. The positional data of fish in 40 frames for each trial were translated into compass directions. Additionally, percentage of cloud cover was independently assessed by three to four different people and categorized in 20% steps.

To verify the use of a time-compensated sun compass, fish were time-shifted 6 h backwards and tested before noon, anticipating – based on their internal clock – that it was already afternoon. In other fish species, this approach led to a counter-clockwise shift in orientation direction, showing that the animals depend on the sun as their main orientation cue (e.g. Mouritsen et al., 2013; Spiecker et al., 2022).

Each fish was tested 2–6 times; a mean orientation direction was calculated. For data analysis, the circular statistics program Oriana.4 (Kovach, 2011) as well as R (version 4.1.3; https://www.r-project.org/) with ggplot2 (Wickham, 2016) were used. First, a Rayleigh uniformity test was used to evaluate whether the individual orientation preference was statistically different from a random distribution. The second step was averaging the mean directions of all significant trials (at least two significant runs) of the same fish that were tested up to 6 times. In the third step, all orientation directions of the individual fish were taken together to obtain a mean group orientation.

Subsequently, the results of the control sun compass and the uncorrected time-shift experiment group were compared using a Mardia–Watson–Wheeler test. According to the hypothesis that sprats are able to use the sun effectively as an orientation aid, the mean orientation direction of the data from the control sun and the corrected time-shifted sun compass should match. The next step was to compare the mean orientation of the control sun compass group with that of the time-shifted fish, using the general and accurately corrected time-shift data. Two different analysis methods were used to correct the raw data according to the sun's position during the course of the day. One achieves a generalised correction whereas the other achieves an accurate correction. Both methods were needed as it is unknown whether fish presume a uniform shift of the sun azimuth or are capable of accurately measuring the azimuth of the sun. The generalised correction assumes a 90 deg shift in 6 h, resulting from a solar azimuth change of 15 deg h⁻¹ (Lambrinos et al., 2016). The accurate correction compares the exact position of the sun at a specific location, day and time of each trial run using www.sunearthtools.com (October 2020, based on Michalsky, 1988) to determine the exact sun azimuth shift at that specific point in time.

To complement our data, we also analysed whether the extent of cloud cover had an effect on the ability of the juvenile fish to orientate accurately. This accuracy is represented by the length of the mean vector (r) of each experimental run. After logit transformation of the r-vectors and Bonferroni correction for multiple comparisons, the r-vectors of the different cloud cover categories (0–20%, >20–40%, >40–60%, >60–80% and >80–100%) were statistically compared after confirming that the variances were equal and data were normally distributed.

Juvenile sprats migrate to their feeding and overwintering grounds without the help of experienced adults. Therefore, they have to rely on inherited orientation strategies, and a sun compass would be a reliable guidance mechanism.

In 2021, when tested in round bowls with the sun as a visual cue, juvenile sprats showed a significant orientation direction towards the northeast with a mean vector of 59 deg (n=21, r=0.383, Z=3.08, P=0.044, 95% confidence interval of the mean: 15–102 deg; Fig. 2A). In 2021, only 9 sprats were time-shifted; they showed a mean direction of 330 deg, but probably owing to small sample size this result was not significant (Z=1.75, P=0.176). Therefore, more time-shift experiments were conducted in August 2022 with 18 fish of a similar size and developmental stage as in 2021. These fish showed a significant orientation (n=18, mean orientation 360 deg, Z=4.63, P=0.008; Fig. 2B, right). The data from time-shifted fish of 2021 and 2022 were combined after testing that their orientation was not significantly different from each other. The time-shifted group of sprat indeed displayed a significant orientation towards the north–northwest, with a mean direction of 349 deg (n=27, r=0.465, Z=5.83, P=0.002, 95% confidence interval of the mean: 318–20 deg; Fig. 2A, right).

The control sun compass group (2021; Fig. 2A, left) and the time-shifted group (2021, 2022; Fig. 2A, right) were compared using a Mardia–Watson–Wheeler test, which showed a statistical difference between the two datasets (W=6.044, P=0.049), with a bearing difference of 70 deg between the two mean vectors. The corrected time-shift group showed an orientation direction towards the east with a mean vector of 79 deg (generalised correction) and 106 deg (accurate correction) after generalised and accurate correction for the time-shift, respectively. The difference in orientation direction between the time-shift and control groups of 70 deg points to a generalised correction. This probably means that fish can only roughly estimate the azimuth of the sun.

Juvenile sprat and juvenile Baltic herring occur at the same time in the Kiel Fjord, and are even schooling together. Spiecker et al. (2022) showed that juvenile herring from the Kiel Fjord also use a time-compensated sun compass for orientation, heading towards the east–southeast, possibly following a migration route close to the coastline and later most likely changing orientation towards the north to get to their overwintering grounds in the sound, an area between Denmark and Sweden. However, juvenile sprats oriented towards the northeast, probably heading towards Bornholm Basin (Fig. 1).

In 2022, we also obtained small (3.9±0.4 cm) sprats earlier in August 2022. Those were also time-shifted and they showed no significant orientation (Z=0.035, P=0.966; Fig. 2B, left). From the comparison of orientation behaviour in smaller and larger fish (see Fig. 2B), we conclude that the developmental stage of sprat has an influence on the migratory readiness.

The start of migration can depend on a variety of factors that can be triggered extrinsically (environmental) or intrinsically (e.g. innate, physiological) (Nabi et al., 2014). In juvenile anadromous herring, lunar phase, decreasing water temperature and increased water flow have been associated with their migration (Yako et al., 2002), but the actual factors that trigger migration are still unknown. Makris et al. (2009) showed that group migration of Atlantic herring occurs once vast shoals form and a population density reaches a critical threshold. Because sprats form dense shoals (Aro, 1989), population density could also be an indicator of the start of their migration.

Environmental parameters can trigger the onset of migration (Lucas and Baras, 2008; Smith, 1985); however, we conclude from our experiments that there are intrinsic or developmental triggers, as water temperature, light cycle and food resources were similar in all our experiments. The individuals in our first group, tested in early August, were significantly smaller (mean: 3.9±0.4 cm) than the individuals in the second group, tested in late August (mean: 4.4±0.3 cm), suggesting that body size may indicate the onset of migration. This observation reveals the onset of migratory readiness of juvenile sprats at the end of August. In contrast, Spiecker et al. (2022) tested two groups of juvenile herring for sun compass orientation, one in late August and one in early to mid September. They also took the effect of body size on the individual bearing into account (September cohort was larger and older), but neither group showed a significant difference in their orientation bearing, suggesting that both groups were already in ‘migration mode’. Cresci et al. (2020) tested Atlantic herring larvae at 14–16 and 25–28 days post-hatch in drifting in situ chambers (DISC) in coastal areas of a Norwegian fjord for their orientation ability. In contrast to sprat, both groups displayed a significant orientation direction, suggesting that larval as well as juvenile herring showed a sun compass-oriented direction. Whether young sprat need a sun compass orientation before starting their migration journey is unknown.

In birds, there is a common phenomenon called ‘Zugunruhe’ or migratory restlessness, which is an increased locomotor activity caged birds express during migration season. It is considered to be an indicator for the urge of migration and is suggested to be driven by an endogenous clock (Helm, 2006). Migratory restlessness has been observed in fish as well. Sudo and Tsukamoto (2015) showed the locomotor activity in Japanese eel (Anguilla japonica) during its migration season. In their migratory phase, these eels (silver eels) showed higher locomotor activity in aquaria during the spawning season compared with yellow eels, which were not in the migratory growth phase. Here, we showed migratory readiness based on orientation behaviour in sprats; to our knowledge, this is the first time a behaviour such as this (migratory readiness) has been observed in oceanodromous fish.

Weather conditions in the northern hemisphere often lead to dense cloud cover. Therefore, we also tested whether clouds have an impact on the orientation ability of the fish. After sorting the data from every individual experimental run and performing a logarithmic transformation, mean r-vectors (length of mean vector) of different cloud cover categories were calculated. There was a strong trend indicating that the directedness of juvenile sprat decreased with increasing cloudiness. Therefore, we performed a chi-squared test for our six categories, which showed a significant difference among them (χ²=16.2, d.f.=5, P=0.0063). Additionally, a pairwise Student’s t-test was performed, showing a significant difference between groups 0–20%, 20–40% and 40–60% and the 80–100% group (Fig. 3). After applying a Bonferroni correction for multiple comparisons, our results showed that the category 60–80% cloudiness was not significantly different from any of the other categories. We showed that a cloud cover of more than 80% decreased the directional swimming significantly. A similar dependency of cloud cover on orientation accuracy was found in larval as well as juvenile herring (Clupea harengus) (Cresci et al., 2020; Spiecker et al., 2022). When the view of the sun is completely blocked by clouds, juvenile sprat would need to rely on different orientation cues such as a magnetic compass.

We thank Felix Mittermayer, Fabian Wendt and Hanna Rudnick for sample collection, Susanne Wallenstein and Anke Müller for video analysis, and Bo Leberecht and Simon Käfer for creating the R script for the circular data plots. We also thank the GEOMAR Kiel for help, hospitality and logistic support.