The brown shrimp (Crangon crangon) is a highly abundant invertebrate in the North Sea, with its life cycle stages ranging from deep offshore spawning to shallow onshore nursery areas. To overcome the long distances between these two habitats, brown shrimp are suspected to use selective tidal stream transport (STST), moving with the cyclic tide currents towards their preferred water depths. However, it is not known which stimulus actually triggers STST behavior in brown shrimp. In this work, we determined the influence of different hyperbaric pressures on STST behavior of juvenile brown shrimp. Brown shrimp activity was recorded in a hyperbaric pressure chamber that supplied constant and dynamic pressure conditions simulating different depths, with and without a tidal cycle. Subsequent wavelet and Fourier analysis were performed to determine the periodicity in the activity data. The results of the experiments show that STST behavior in brown shrimp varies with pressure and therefore with depth. We further show that STST behavior can be initiated by cyclic pressure changes. However, an interaction with one or more other environmental triggers remains possible. Furthermore, a security ebb-tide activity was identified that may serve to avoid potential stranding in shallow waters and is ‘remembered’ by shrimp for about 1.5 days without contact with tidal triggers.
Migration and distribution patterns of marine organisms are often closely connected to oceanic and tidal currents. In addition to long-distance migrations following the predominant oceanic streams, several marine organisms have evolved behavioral mechanisms to actively make use of locally changing flow fields for directional movement. In many coastal and intertidal species, for example, vertical migrations into and out of the water column are synchronized to the tidal cycle (Holt et al., 1989; Gibson, 2003). By means of this coupling to the tidal streams, the ebb-tide, the flood-tide or both currents can be used for directional horizontal displacement. This specific type of behavior has been observed in a variety of marine invertebrates as well as fish species so far (Holt et al., 1989; Forward and Tankersley, 2001) and has been referred to as selective tidal stream transport (STST) (Jones et al., 1979; Criales et al., 2013). STST is especially common in larval and juvenile stages. Because of their limited swimming abilities, these early life stages would not be able to span great distances, migrate into favored directions or even maintain their present position on their own without using STST (Queiroga et al., 1997; Forward and Tankersley, 2001). STST behavior, however, has also been documented for adult specimens of several marine species (Forward and Tankersley, 2001).
One species that is suspected to use STST during its extended ontogenetic and seasonal migrations is the common brown shrimp [Crangon crangon (Linnaeus 1758)] (van der Baan, 1975; Cattrijsse et al., 1997; Temming and Damm, 2002; Daewel et al., 2011). Brown shrimp spawn off-shore with the eggs first being attached to the female (Ehrenbaum, 1890). After hatching, the larvae are released into the water column where they pass five zoea stages (Ehrenbaum, 1890) and drift passively towards the coastal areas (Berghahn, 1983; Boddeke et al., 1986; Daewel et al., 2011). After reaching the post-larval stage, they settle to the ground and move further to the shallow coastal nursing areas of the Wadden Sea, where they grow rapidly (Kuipers and Dapper, 1981). After obtaining a total length (TL) of 20–30 mm, the brown shrimp leave the tidal flats and gradually migrate towards deeper waters, where they become mature (Janssen and Kuipers, 1980). When water temperatures decrease during late autumn and winter, they again migrate off-shore for spawning (Boddeke, 1976).
Larval and also adult brown shrimp are relatively small, reaching a maximum body size of ∼9 cm total length (Tiews, 1967). Thus, it seems unlikely that brown shrimp cover these distances of up to 60 km (Daewel et al., 2011) during their migration by active swimming exclusively. Cattrijsse et al. (1997), based on observations made by Boddeke (1976) and also Temming and Damm (2002), speculated that juvenile and larval brown shrimp use STST to get transported to the shallow coastal nursing grounds. These assumptions were supported by several field studies reporting tidally influenced behavior in brown shrimp, which can be considered as a premise for STST. Hartsuyker (1966) reported a size-dependent horizontal movement of brown shrimp following the changing water level between low and high tide, whereas Adhub-Al and Naylor (1975) found brown shrimp emerging vertically into the water column with a tidal- and light-dependent periodicity. Hufnagl et al. (2014) observed increased activity during ebb-tide under laboratory conditions for several tidal cycles after the shrimp had been transferred from the field to an aquarium with constant water level. This indicates a natural zeitgeber stimulus – a ‘remembered’ trigger experienced in the field that persists for a certain time. However, the exact nature of the zeitgeber and the mechanisms inducing STST in brown shrimp remain speculative.
Daewel et al. (2011) assumed temperature and salinity changes during the tidal cycle as potential cues and used these triggers to simulate larval migration, including STST behavior in the North Sea. This work demonstrated that both triggers could potentially induce STST and also guide shrimp towards the coast. Both cues, however, were not considered to be sufficiently strong to trigger the larval migration exclusively. Besides temperature and salinity, changes in hydrostatic pressure are the most obvious alternating variable in coastal intertidal habitats during the tidal cycle, fluctuating constantly throughout the seasons. Several studies were able to connect hydrostatic pressure as zeitgeber and trigger of STST in fish, amphipods and crustaceans (Morgan, 1965; Gibson, 1982; Vance and Pendrey, 1997). Still, those animals have to be barosensitive for relatively small changes in hydrostatic pressure, which has been shown by Digby (1961) and Adhub-Al and Naylor (1975) for brown shrimp. A connection between this ability of brown shrimp to detect pressure changes and STST, however, has not yet been proved.
In the present study, we thus investigated whether and how varying levels of hydrostatic pressure affect vertical emergence behavior in brown shrimp. We constructed an experimental hyperbaric chamber that simulated static (constant hydrostatic pressure) and dynamic (sinusoidal pressure changes) hyperbaric conditions based on the tidal cycle in the field. In one experiment, we introduced the brown shrimp into the chamber straight from the field, exposing them to constant pressure conditions. In a second experiment, we applied dynamic pressure conditions in the chamber after the shrimp had been maintained under constant atmospheric pressure for 9 days in the dark, eliminating any potential zeitgeber from the field (Hufnagl et al., 2014).
Activity at constant pressure
Brown shrimp activity patterns showed marked differences between the three constant hydrostatic pressure treatments (Fig. 1). At 0 bar (1 bar=100kPa; Fig. 1A), three peaks in activity following the tidal rhythmicity in the field were observed. The first peak was most pronounced and started with the first ebb-tide signal. The following two peaks were less pronounced but still corresponded to the field ebb-tide signal. The fourth peak more or less merged with the third one and was not as clearly related to a specific part of the tidal cycle. The fifth peak was also not consistent with the fifth ebb tide in the field and a sixth peak for the last cycle could not be identified.
At 0.5 bar (Fig. 1B), the shrimp showed a somewhat different behavioral pattern. Compared with 0 bar, activity was markedly reduced. The smaller, but still clearly detectable peaks, however, did not correspond to the tidal signal in the field but were expressed at ∼24 h rhythmicity. At 1.0 bar (Fig. 1C), however, shrimp showed five minor peaks in activity, beginning with two small peaks that corresponded to the first two tidal cycles. The third peak was also consistent with the third tidal signal but a bit more pronounced. However the fourth and fifth peak seemed to be expressed more in a ∼24 h pattern than following a tidal rhythmicity. Between the third and fourth peak, activity increased slightly, but did not display a separate peak.
Comparing the different pressure treatments, it became apparent that activity markedly decreased with hydrostatic pressure. The activity plotted against time in combination with the tidal cycle (Fig. 1) revealed a strong decrease in activity for the shrimp exposed to 0.5 and 1.0 bar constant pressure in comparison to those that were treated with 0 bar additional pressure. The most pronounced peak in the activity of the 0 bar group reached 14 laser counts per shrimp, while the activity in the 0.5 and 1.0 bar group never exceeded about 4 counts (i.e. 71% less in activity).
Fourier and wavelet analysis of constant pressure experiments
The Fourier and wavelet analysis of brown shrimp activity at constant pressure experiments (Fig. 2 and Table 1) revealed a strong tidal periodicity at the beginning of the 0 bar experiment and diurnal as well as tidal activity patterns for the 0.5 and 1.0 bar groups.
The wavelet power spectrum for the 0 bar group (Fig. 2A) showed a clear tidal rhythm of about 12.4 h (Holt et al., 1989) for the first three activity peaks. After the third peak, no continuing rhythm was detectable until a tidal rhythm occurred again for the last 5 h of the time series. The Fourier analysis confirmed those findings, with 12.8 h being the most frequent period. Fourier analysis also identified 76.6 h as the second strongest period which is not visible at the wavelet spectrum because of a low scaling of the y-axes.
In contrast, the wavelet power spectrum of the 0.5 bar group (Fig. 2B) showed a strong diurnal periodicity, which was consistent over the whole time series and got more prominent at the end of the experiment. Also, at the end of the time series, a short tidal period was visible, in addition to the consistent diurnal period. In accordance with these observations, the two most prominent periods of the Fourier analysis were 24.4 and 12.2 h.
The wavelet spectrum for the 1.0 bar group (Fig. 2C) showed a tidal period between the second and third activity peak and a diurnal period from the third to the end of the time series. Those periods were also reflected in the Fourier analysis, with 26.4 h as the strongest period and 11.3 h as the second strongest. However, at the beginning of the time series, the wavelet analysis showed a short period around 30 h and one of about 8 h.
Activity at dynamic pressure
In the experiments with dynamic hydrostatic pressure changes, it is obvious that an artificial tidal signal immediately induced a tidal rhythmicity in activity and that the strength of activity decreased with increasing mean pressure.
In the dynamic pressure experiments, constant activity peaks are shown for all three groups (Fig. 3). In each of those groups, the shrimp activity started to follow the applied 12 h cyclic pressure immediately with an ebb-tide signal and did not increase or decrease over time. By comparing the different pressure groups, it was shown that the activity level decreased with a higher mean pressure. While the highest activity for the 0–0.35 bar group (Fig. 3A) was measured with about 11 laser counts per shrimp, the highest activity for the 0.5–0.85 bar group (Fig. 3B) was ∼9 laser counts (about 18% less activity) and the 1.0–1.35 bar group (Fig. 3C) showed a maximum activity of just 5 laser counts (about 55% less activity). In all dynamic pressure experiments, the activity peaks were located in the ebb-tide phases (Fig. 4). Especially when comparing the last activity peaks with the first ones, it became apparent, that the activity during ebb-tide became even more pronounced over time.
Fourier and wavelet analysis of dynamic pressure experiments
Results of the wavelet and Fourier analysis for the dynamic pressure groups are shown in Fig. 5 and Table 2. The wavelet analysis confirmed the previous results by showing a clear ‘tidal’ signal in the power spectrum for each group. Additionally, the Fourier analysis showed one period of about 12 h for each group that was much more prominent than every other period. In this case, the second most prominent periods seemed to be of less importance.
Analyzing activity during ebb- and flood-tide phases revealed that shrimp in the dynamic pressure conditions tend to establish an ebb-tide activity. However, this ebb-tide activity pattern was less clear than the ebb-tide signal that had been observed in newly caught shrimp, treated with low constant pressure (Fig. 1A).
For visualization, each tidal cycle was split into ebb- and flood-tide activity expressed as a percentage (Fig. 4). Thus, the activity of one pressure cycle from low water to low water was considered to be 100%. Fig. 4 shows a clear shift for all pressure groups from about 55% ebb-tide activity at the first pressure cycle to about 80% ebb-tide activity at the fourth cycle. This 80% ebb-tide activity remains relatively constant for the 0–0.35 and the 0.5–0.85 bar groups, whereas the 1–1.35 bar group ebb-tide activity decreases after the fourth cycle to about 65% and afterwards rises again to about 71%. To compare those pressure-induced ebb-tide activities with the observed activity in constant low pressure treatment, an ebb-tide activity of 91% was calculated for the first observed activity peak of the constant 0 bar group (Fig. 1A).
Constant pressure experiments
To analyze the internal zeitgeber-based activity patterns, shrimp were tested right after catch at constant pressures of 0, 0.5 and 1.0 bar for 3 days. We showed that brown shrimp at 0 bar executed very clear activity peaks during the first two tidal cycles that were consistent with the ebb-tide phases in the field. In contrast, brown shrimp at 0.5 and 1 bar showed much less activity and prominent ∼24 h rhythms with short periods of ∼12 h rhythms. This can be interpreted such that shrimp in shallow water attempt to leave these habitats via ebb-tide transport (Hartsuyker, 1966; Viegas et al., 2012) and that this behavior decreased or even ceased as hydrostatic pressure and hence depth increased. Since shrimp at 0 bar experienced just the 0.034 bar of hydrostatic pressure of the water column, a water depth of 34 cm was simulated. The brown shrimp is capable of detecting hydrostatic pressures of one bar upward (Digby, 1961), a sensitivity for pressure changes lower than that has not been proved, but may be possible. However, the water level in the field varies about 3.5 m between high and low water and thus a simulated depth of 34 cm during beginning of ebb-tide cycles implies a danger of stranding for the shrimp. This could explain the observed ebb-tide transport as a security mechanism that is only carried out if the water level is low. Since during the experiment the tested shrimps were excluded from any tidal effect, the observed activity pattern can only be explained by some kind of internal zeitgeber that the animals experienced in the field. These findings confirm the results from Hufnagl et al. (2014), who found similar activity patterns for brown shrimp and also presumed that a ‘remembered’ tidal rhythm in the shrimp was responsible for this behavior. These observations are also consistent with identified patterns in the field where shrimp tend to stay in shallow waters of the tidal zone during high tide and migrate offshore to avoid getting stranded (Hartsuyker, 1966; Adhub-Al and Naylor, 1975). The shrimp appeared to be able to identify the time of ebb-tide for three tidal cycles without exposure to any tidal triggers, although the total amount of activity decreased. After the third peak, the activity signal became less frequent and the peaks got broader and less pronounced. Thus, no clear activity pattern, specifically no tidal rhythmicity, could be identified after the third tidal peak. This indicates that brown shrimp lose their internal zeitgeber after a period of about 1.5 days after being isolated from tidal triggers.
The weaker and broader activity peaks after the first day, are more difficult to interpret. The wavelet analysis did not identify a clear and prominent pattern in activity rhythm after the first two tidal signals. However, as revealed in the time series (Fig. 1), distinct peaks were still present. It remains unresolved whether these peaks would have decreased further.
In contrast to the experimental groups subjected to 0 bar, brown shrimp exposed to 0.5 and 1 bar did not show such strong and clear patterns in tidal activity peaks. This does not necessarily imply that the internal zeitgeber was not present, but may indicate that these animals did not respond as strongly because of the different hyperbaric conditions. This observation supports the idea of an internal ‘security’ ebb-tide movement that is only executed if the shrimp detect an acute risk of stranding. Since the 0.5 and 1 bar groups represent simulated depths of 5 and 10 m and therefore no immediate risk of stranding within one ebb-tide phase, the shrimp in those groups did not have to execute a security ebb-tide movement.
Besides the much lower activity in the 0.5 and 1 bar groups, it is conspicuous that both groups show a prominent ∼24 h rhythmicity and short periods of ∼12 h activity. However, one has to consider that our time series represents averages of several experiments that were conducted with shrimp caught at different times of the day. In addition, the data were matched according to the first high-tide after the catch and not to photoperiod. Since shrimp were caught at different times of the day, it is still possible that the ∼24 h patterns depicted in the time series of the 0.5 and 1.0 bar groups might be affected by a diurnal activity pattern. This assumption is supported by the work of Del Norte-Campos and Temming (1994) and Feller (2006) who found evidence for a higher activity at night. Hufnagl et al. (2014) described a shift from emergence behavior driven by a zeitgeber to increased nocturnal activity if brown shrimp were subjected to a photoperiod. The diurnal activity rhythms shown in the present work indicate that the photo-periodical rhythm from the field was still present in the animal and might have had an effect on brown shrimp activity. The short ∼12 h patterns instead are more difficult to interpret. Assuming that C. crangon has a size-dependent depth optimum (Hartsuyker, 1966; Adhub-Al and Naylor, 1975; Temming and Damm, 2002), the absence of ebb-tide signals at the beginning of the 0.5 bar data could mean that the shrimps preferred to stay at that depth and therefore suppress STST. If that were correct, the 1 bar group should respond the same way or with a flood-tide signal to reach shallower waters. However, the 1 bar group shows three small tidal signals with slight dominance of activity during the ebb-tide phases, which would lead the shrimps to even deeper waters. It is also possible that the tidal signals in the 0.5 and 1.0 bar groups are artifacts that were generated by aligning the data according to the tidal rhythms and not the photoperiod, but consequently, these differences between the 0.5 and 1 bar groups cannot be explained in this work.
Dynamic pressure experiments
Prior to the experiments at dynamic pressures of 0–0.35, 0.5–0.85 and 1–1.35 bar, brown shrimp were kept at total darkness for 9–10 days, ensuring the loss of any internal zeitgeber (Hufnagl et al., 2014).
The results of the activity plots (Fig. 3) and the corresponding power spectra of the wavelet and Fourier transform analysis (Fig. 5, Table 2) show very clear ∼12 h activity patterns that can be solely ascribed to the dynamic pressures changes. The ∼12 h activity pattern in all groups emerged when the first pressure cycle started and the activity peaks tended to shift to the ebb-tide phase within the first three to four pressure cycles (Fig. 4). During that shift, the 12 h periodicity within the activity signal is stable, which excludes a possible shift to an inert 12.4 h rhythm. A general decrease in activity with increased pressure was clearly visible as well (Fig. 3). Thus, we provide evidence that brown shrimp is able to react on tidal pressure changes by means of changes in activity as postulated by Adhub-Al and Naylor (1975). Additionally, cyclic pressure changes evoke an activity pattern leading the shrimp into deeper waters by means of selective ebb-tide transport. This confirms the general assumption that brown shrimp is able to perform STST (Henderson and Holmes, 1987; Cattrijsse et al., 1997) and complements the work of Digby (1961) who observed a general pressure sensitivity of brown shrimp. It further supports the hypothesis of Hufnagl et al. (2014) who suggested pressure as a potential trigger for STST in brown shrimp.
Combining the results from the experiments with dynamic as well as constant pressure conditions, it becomes obvious that hydrostatic pressure affects STST in brown shrimp. However, if a cyclic signal is missing, the internal zeitgeber triggers distinct peaks in activity when hydrostatic pressure is low to avoid stranding. In deeper waters, the zeitgeber did not lead to a clear tidal activity. However, the security ebb-tide activity of the constant 0 bar pressure experiment accounted for 91% of the whole activity during the tidal cycle. The pressure-induced ebb-tide activity of the dynamic 0–0.35 bar group was lower (Fig. 5), but still accounted for 80% of the activity. This slight difference in ebb-tide activity might indicate that the ‘remembered’ STST activity of the internal zeitgeber is influenced by additional triggers interacting with pressure and enhancing accuracy in STST. Daewel et al. (2011) suggested that temperature and salinity could be additional potential triggers for STST in juvenile C. crangon. However, changes in salinity and temperature during the tides are small and it remains questionable whether these alone might induce STST. Another potential trigger was identified by Criales et al. (2013) who observed an increase in vertical activity of pink shrimp (Farfantepenaeus duorarum) with increasing turbulence. They stated that turbulence, in combination with increases in salinity, could trigger flood tide transport in these shrimp. This combination might also influence STST in brown shrimp and has to be tested in future studies. It is nevertheless striking that turbulence as a trigger for STST needs to interact with at least one additional trigger because it does not contain information about the direction of the water flow. Hydrostatic pressure instead is directly connected to the water level and is therefore a very robust indicator of the tidal cycle and the current direction. Our data show clearly that brown shrimp would be able to perform STST movements solely under the influence of tidal pressures.
The gradual decrease of activity between the dynamic pressure groups suggests the existence of a depth preference for C. crangon; maximum activity dropped from 11 counts (100%) in the 0–0.35 bar group to 9 counts (82%) in the 0.5–0.85 bar group and to 5 counts (45%) in the 1–1.35 bar group. Furthermore, the observed ebb-tide activity becomes less distinct in the 1–1.35 bar group (Fig. 5). This may indicate that the preferred depth of 30–35 mm shrimp is below 10–13.5 m, and that the shrimps would further reduce their ebb-tide activity at higher pressure settings to avoid deeper waters. This idea of a preferred depth range in larger shrimp outside the shallow nursery areas is supported by field observations of Beukema (1992) who found that shrimp leave the shallow nursery areas at 20–30 mm. Furthermore, Janssen and Kuipers (1980) as well as Hufnagl et al. (2014), observed peak abundance of 30–40 mm shrimp in deeper tidal channels adjacent to tidal flats where the abundance of that size class was low. The depth optimum is, however, unlikely to be constant but may vary with season, size, sex and stage of maturity (Boddeke, 1976; Del Norte-Campos and Temming, 1998; Spaargaren, 2000; Campos and Van der Veer, 2008). Additionally, it can be assumed that the relative decrease of ebb-tide activity in the 1–1.35 bar group marks the start of a shift to flood-tide transport, which could indicate a reversed activity pattern at higher pressures. If brown shrimp uses STST to remain in or return to its preferred depth range, it also needs to have a mechanism that triggers a migration from deeper offshore waters to shallower coastal regions. Thus, it can be speculated that with cyclic pressures above the tested maximum of 1.35 bar, the total activity would increase and shift to a flood-tide activity. Performing flood-tide activity would also be an explanation for the very low abundance of brown shrimp beyond 40 m water depth (Callaway et al., 2002; Siegel et al., 2005). However, for confirmation of these interpretations, additional experiments are needed at higher pressures and with different size classes and maturity stages.
MATERIALS AND METHODS
Animal sampling and maintenance
Brown shrimp were sampled from May to July 2012 close to a tidal channel at the coast off Büsum, Germany (54°07′N, 8°51′E) (Fig. 6). Sampling was conducted at low tide using a push net (2 mm mesh size) in approximately 1 m water depth. Following sampling, the animals were transported to the nearby laboratory of the Research and Technology Centre in Büsum (FTZ) in cooled buckets (25 l) filled with seawater from the sampling location. In the laboratory, the shrimp were measured to the nearest 5 mm total length (TL), starting from the tip of the rostrum to the end of the uropod. Shrimp of 30–35 mm TL were separated and either immediately used for the experiments (see below) or transferred to circular holding units connected to a seawater recirculation system [25 PSU (practical salinity units), 14°C]. To exclude any potential effects of external zeitgeber or circadian rhythms (Rodriguez and Naylor, 1972), the holding units were darkened using a light-tight tarpaulin. For the next 9 days, they were fed dry feed (Coppens International MariCo Advance, Helmond, The Netherlands) to apparent satiation at different times of the day.
The hyperbaric pressure chamber (Fig. 7) was constructed using a metric 90 deg PVC-U tee (GF Piping Systems, Schaffhausen, Switzerland) with a removable top at which the inflow and outflow pipes as well as a pressure sensor (STST ATM. 1ST, Sindelfingen, Germany) were mounted. At the front and the back of the tee, glass windows were inserted. Constant inflow into the pressure chamber was adjusted by using a gear pump (Ismatec RS232, Wertheim, Germany). The outflow was regulated by a two-way control valve (Belimo Automation AG TRCD24A-SR, Hinwil, Switzerland). The pressure sensor and the two-way control valve were connected to a proportional integral derivative (PID) process controller (West Control Solutions 4100+, Kassel, Germany), adjusting the opening of the valve according to the detected pressure. Water temperature inside the pressure chamber was controlled through the surrounding air with a mono block air conditioner (Bonus KlimaMonoblock 2600W, Hamburg, Germany) by cooling the whole system to approximately 14°C. The water volume inside the pressure chamber amounted to 40 l and the system was connected to a total water volume of 500 l.
In front of the glass windows, transmitter and collector consoles of a red low-level laser grid (SICKMLG1-0290F212, Waldkirch, Germany) were installed 8 cm over the bottom of the chamber. The laser grid served to monitor brown shrimp activity, i.e. emergence from the bottom of the chamber, by means of interrupting the laser beam. Each time a shrimp disrupted one or more of the laser beams, the script counted the events in 10 s intervals. The laser grid was connected with two digital boards (Velleman VM110N USB-Experiment-Interface-Board, Gavere, Belgium) that were stored in a plastic case. Gear pump, process controller and the boards were in turn connected to an operating PC. After filling and closing the chamber, the experimental setup was covered with tarpaulin for monitoring the activity of the shrimp by means of the laser grid.
All the data were integrated at the PC in a constantly running Matlab script (MATLAB) that provided a target pressure value to a given time. If the target and the actual measured pressure differed, the rotation rate of the pump was regulated via the Matlab script to match both values. In addition, the laser counts, pressure and rotation rate were exported every 10 s into a text file.
Brown shrimp were tested at three different constant hydrostatic pressures, i.e. 0, 0.5 and 1 bar, simulating water depths of 0, 5 and 10 m, respectively. For each tested pressure group, three independent runs were conducted with 10 shrimp per run i.e. 30 shrimp per pressure group. The shrimp of the constant pressure experiment were used directly after catch and were considered to still have an internal zeitgeber rhythm. Following length measurements, the shrimp were inserted into the pressure chamber, the chamber was locked and darkened and the system was adjusted to the respective pressure. Subsequently, the controlling Matlab script was started. In the following 72 h, changes in swimming activity were recorded with the installed laser grid.
Activity of brown shrimp under dynamic hydrostatic pressure was tested for three different water depths, i.e. 0, 5 and 10 m. Again, the groups were tested in a triplicate design with 10 shrimp per run i.e. 30 shrimp per pressure group. Sinusoidal pressure changes of 0.35 bar simulated the tidal cycle in the range of 0–3.5, 5–8.5 and 10–13.5 m. Upon prior maintenance for 9 days to lose the internal zeitgeber signal, the shrimp were inserted into the darkened chamber for 72 h and the system was started as described above. Afterwards, a second Matlab script that simulated tidal rhythmicity by using a sine wave function was started with either 0–0.35, 0.5–0.85 or 1–1.35 bar dynamic pressure. The simulated tidal cycle always started at the lowest pressure (low tide) and the length of one tidal cycle was simulated as 12 h. This period is slightly less than the actual average tidal length of 12.4 h (Holt et al., 1989) and was chosen so that experiments did not shift into the night. The pressure fluctuation between low and high tide was simulated with 0.35 bar, which is consistent with the average water level difference in the field of 3.5 m that was calculated with tide data from the Bundesamt für Seeschifffahrt und Hydrographie (BSH).
Data collected during the experiments represented brown shrimp activity, i.e. interruptions of the laser beam and pressure data, integrated over 10 s count data were binned for 10 min, i.e. all 60 data points of each 10 min interval. To correct for an average mortality of 18%, the count data were divided by the number of shrimp removed from the chamber alive at the end of the experiment and by half of the dead or missing shrimp, creating activity measures per individual that are averages of all shrimp activity at a given time. This activity measure considered dead and missing shrimp in general to be half as active as the surviving ones. The multiple experimental runs of each dynamic and constant pressure were adjusted to their first low tide and averaged to obtain one dataset for each pressure group. Since for the dynamic experiments the shrimp were considered to have no internal zeitgeber signal, the periodicity in applied pressures was assumed to trigger the activity. Thus, the first low tide was defined as the time, when the targeted low tide pressure was reached. For the constant experiments in which the shrimp still had their zeitgeber signal, calculations of the BSH were used to determine an averaged first low tide of the field.
Count data were analyzed following Hufnagl et al. (2014) using a fast Fourier transformation (FFT; Bracewell, 1999) as well as a wavelet analysis (Morlet, http://www.jmlilly.net/jmlsoft.html) to identify frequency components in the activity signals. In the Fourier analysis, activity data were analyzed excluding the temporal component, thus specifying the rhythm of the frequency per se (Cain et al., 1984). The wavelet analysis, however, is capable of providing time allocation and frequency information of the activity signals, simultaneously (Meyer, 1992). To achieve this, the wavelet analysis decomposes the time series in time-frequency intervals and produces a display of what frequency bands exist at what time. However, the wavelet transform, is not totally accurate and always produces a result that is a compromise between the accuracy of frequency and the accuracy of time information (Torrence and Compo, 1998). Because we are dealing with finite-length time series, errors will occur at the edges (start and end) of the wavelet power spectrum, thus the results at those points have to be interpreted with care (Meyer, 1992; Torrence and Compo, 1998). Both methods tend to identify frequencies that are longer than or as long as the time series itself by simulating one big frequency curve through all data points. Those frequencies are technically present but do not resemble any biological process.
In addition to the frequency analysis, it was defined whether the shrimp in the dynamic pressure experiments develop a cyclic activity that is more distinct at flood or ebb tide. Therefore, the periods of rising and declining pressure were separately analyzed and the activity for ebb and flood phases was calculated as a percentage.
We would like to express our great appreciation to the staff of the Bundesamt für Seeschifffahrt und Hydrographie who provided us with water level and depth measurements of the North Sea. Also we want to thank Jörg Bruns for his help in assembling the experimental setup, Eva Gogolin for her spelling, Dörte Lüdemann for her assistance in graphical design and finally Klaus Ricklefs and colleagues from the Forschungs- und Technologiezentrum Westküste for their hospitality.
A.T., J.-P.H. and S.R. designed the study, M.T. and A.E. sampled data, M.T., A.T., S.R. and M.H. analyzed and interpreted the data, wrote and revised the manuscript.
This study was partly funded by the Cluster of Excellence ‘Integrated Climate System Analysis and Prediction’ (CliSAP) of the University of Hamburg.
The authors declare no competing or financial interests.