Long-Term Monitoring of Diel and Seasonal Rhythm of Dentex dentex at an Artificial Reef

Behavioral rhythms are a key aspect of species fitness, since optimize ecological activities of animals in response to a constantly changing environment. Cabled observatories enable researchers to collect long-term biological and environmental data in real-time, providing relevant information on coastal fishes’ ecological niches and their temporal regulation (i.e., phenology). In this framework, the platform OBSEA (an EMSO Testing-Site in the NW coastal Mediterranean) was used to monitor the 24-h and seasonal occurrence of an ecologically iconic (i.e., top-predator) coastal fish species, the common dentex (Dentex dentex). By coupling image acquisition with oceanographic and meteorological data collection at a high-frequency (30 min), we compiled 8-years’ time-series of fish counts, showing daytime peaks by waveform analysis. Peaks of occurrence followed the photophase limits as an indication of photoperiodic regulation of behavior. At the same time, we evidenced a seasonal trend of counts variations under the form of significant major and minor increases in August and May, respectively. A progressive multiannual trend of counts increase was also evidenced in agreement with the NW Mediterranean expansion of the species. In GLM and GAM modeling, counts not only showed significant correlation with solar irradiance but also with water temperature and wind speed, providing hints on the species reaction to projected climate change scenarios. Grouping behavior was reported mostly at daytime. Results were discussed assuming a possible link between count patterns and behavioral activity, which may influence video observations at different temporal scales.

INTRODUCTION Diel (i.e., and seasonal biological processes of species inhabiting temperate regions, are synchronized to changes in photoperiod length and overall levels of environmental illumination (Foster and Kreitzman, 2010;Visser et al., 2010;Helm et al., 2013;Kronfeld-Schor et al., 2013). In marine coastal fishes, the photoperiod light intensity are among the most important environmental variables controlling biological rhythms and overall phenology (Naylor, 2010). For example, environmental illumination determines the timing of activity of predators and preys, that perform their ecological tasks according to a trade-off between maximum opportunities of visual-based feeding and minimum mortality risk (Daan, 1981;Reebs, 2002;Brierley, 2014;Mittelbach et al., 2014). However, the exposure of marine costal ecosystems to solar light produces a seasonal co-variation of photoperiod length with other habitat variables that also affect biological rhythms. For example, temperature can have strong effects on fishes at day-night and seasonal scales (Reebs, 2002;López-Olmeda et al., 2006). Combined photoperiod length and temperature cycles regulate physiological processes over the day-night alternation, resulting in global growth and reproduction patterns at a seasonal level (Falcón et al., 2010;Bulla et al., 2017;Cowan et al., 2017). Nevertheless, many marine species can also follow the lunar or tidal cycle to carry out their biological processes within the lunar day of 24.8-h (Naylor, 2010). In particular, tidal rhythms in marine species were related to locomotion and reproduction (Wagner et al., 2007;. The interaction of activity rhythms of all species within a marine community may affect the estimation of its overall biodiversity. This is particularly significant for ecologically important species, such as top-predators, that play a critical role in maintaining the structure and stability of communities and affect ecosystem functioning (Heithaus et al., 2008(Heithaus et al., , 2012Byrnes et al., 2021). Sampling should be repeated at a frequency sufficient to grasp the whole alternation between consecutive peak and through in population abundances as a product of massive rhythmic displacement (Aguzzi et al., 2015b). Moreover, that sampling has to be repeated in association with concomitant data collection to understand how photoperiod length, light intensity and other environmental variables modulate behavioral responses (Aguzzi et al., 2020d). Similar temporal effects exist on fish grouping behavior (Rodriguez-Pinto et al., 2020), whose strategy can be related to foraging, spawning and predator evasion (Ford and Swearer, 2013;Makris et al., 2019;Lear et al., 2021). Moreover, environmental modulation of grouping behavior of fish has been observed in association to photoperiod changes (Meager et al., 2012;Georgiadis et al., 2014). Changes on grouping behavior, driven by human activities such as fishing, could affect the ecosystem functioning, and have repercussions for biodiversity conservation and fisheries management strategies (Sbragaglia et al., 2021).
Data on the phenology of marine fishes, as a product of a variation in local abundances, can be studied by cabled observatories for their capability to perform high-frequency, continuous and long-lasting imaging along with a concomitant multiparametric oceanographic data acquisition (Snelgrove et al., 2014;Danovaro et al., 2017;Aguzzi et al., 2019Aguzzi et al., , 2020aRountree et al., 2020). In particular, cabled systems have the capacity to host many environmental sensors at high resolution, collecting many habitat variables, thus giving a better instrumental field approach to fishes' ecological niches (Hutchingson, 1957). Stand-alone or lander-based cameras are also good tools to study those aspects of species (e.g., Langlois et al., 2020;Drazen et al., 2021). But, given to energy constrains, a limited set of environmental variables is usually acquired. Each environmental variable measured by the installed sensor (e.g., essential environmental variables) can add habitat information for each imaged species (Aguzzi et al., 2020b). Time-lapse imaging studies with that technology have been efficiently used to describe diel and seasonal patterns in fish counts as a proxy for behavior rhythms, resulting in projected abundance changes at all depths of the continental margin (e.g., Juniper et al., 2013;Doya et al., 2014;Matabos et al., 2014Matabos et al., , 2015Milligan et al., 2020). In fact, in the marine three-dimensional scenario of the seabed and the water-column, day-night and seasonal shifts in populations bathymetric distributions, displacement ranges, and overall activity, influence the number of collectable animals into our sampling windows (e.g., Scapini, 2014;Chatzievangelou et al., 2021). A variation in counted animals produce changes in estimated abundances for a species in comparison to all the others (i.e., evenness; Aguzzi et al., 2015b). When rhythmic abundance changes are not carefully considered at sampling, their effect transcend to the computed biodiversity (Doya et al., 2017).
The use of cabled observatories for the monitoring of economically or ecologically important fish species is of relevance for the international conservation strategy agendas (Aguzzi et al., 2020c). Here, we used a coastal cable observatory to videomonitor the 24-h and seasonal occurrence of a top-predator, the common dentex (Dentex dentex; hereafter refers to as Dentex), at an artificial reef at high frequency over almost a decade. This species represents an iconic study case also for its value in commercial and recreational fisheries (Marengo et al., 2014;Sbragaglia et al., 2020), and a previous time-lapse study at the same artificial reef using the same cabled observatory suggested a relationship of fish presence with temperature, salinity and photoperiod length (Sbragaglia et al., 2019). Here, we moved a step forward and attempted to measure the association of count patterns over the 24-h to the photoperiod length, scaling this phenomenon over the whole seasonal cycle (i.e., photoperiodism). In doing so, we evaluated which of the measured oceanographic and meteorological variables mostly affected the reported count patterns. At the same time, we innovatively quantified the occurrence and the temporal dynamic of grouping behavior, also relating this phenomenon to the environmental variation.

The OBSEA Platform Location and Equipment
The coastal Seafloor Observatory (OBSEA 1 ) is a cabled observatory platform located at 4 km off Vilanova i la Geltrú (Catalonia, Spain) at 20 m depth within the Colls i Miralpeix Natura 2000 area Del Rio et al., 2020; Figures 1A,B). The observatory is equipped with an OPT-06 Underwater IP Camera (OpticCam), which can acquire images/footages of the surrounding environment with a resolution of 640 × 480 pixels. OBSEA is also equipped with two custom developed white LEDs (2,900 lumen; color temperature of 2,700 K), located besides the camera (with an angle of 120 • ) at 1 m distance from each other to allow image acquisition at night . A procedure controlling the ON-OFF status of lighting immediately before and after image acquisition, was performed because of the artificial photic footprint on species (e.g., Matabos et al., 2011). The lights were switched ON and OFF (lasting for 3 s) by a LabView application that also controlled their white balance.

Image Acquisition, Fish Counting and Environmental Data Processing
We acquired 70,254 images with a 30 min time-lapse mode, continuously during 8 years (2012-2019), preserving the same field of view, centered on the artificial reef at 3.5 m in front of the OBSEA (see Figure 1C). Individuals of Dentex were manually counted for each image ( Figure 1D) by a trained operator following procedures by Condal et al. (2012) and Aguzzi et al. (2013).
Temperature ( • C), salinity (PSU), and depth (m) were measured by the CTD probe installed aside the camera . Furthermore, we collected data of air temperature ( • C), wind speed (km/h), and wind direction (deg.) from the meteorological station located on SARTI (Development Center of Remote Acquisition and Information Processing Systems) rooftop in Vilanova i la Geltrú. We also gathered sun irradiance (W/m 2 ) and rain (mm) from the Catalan Meteorological Service station in San Pere de Ribes (6 km away from the OBSEA). Time series for all the environmental data compiled by selecting and extracting only readings contemporary to the timing of all acquired images.
We applied range filters for the fluctuation of environmental variables in order to remove out-layer data (i.e., due to instruments malfunctioning). Guillén et al. (2018) was referenced for water temperature and salinity (i.e., ranges of 11-28 • C and 36.80-39.67 PSU, respectively), since authors have a 10 years' time series of readings from a nearby station in Barcelona (Spain). For air temperature and wind speed (ranges of 3-31 • C and 0-60 km/h, respectively) we used an online website 2 with 30 years of hourly weather modeled data. Rain and solar irradiance were not filtered since downloaded from an Institutional and already filtered source 3 .

Multivariate Analysis
Prior to the multivariate analysis, we transformed the number of individuals of Dentex per photo into nominal presence/absence response variable (Zuur et al., 2007). In order to obtain an optimized model for fish presence/absence, we then executed a correlation analysis on the environmental variables, to group the highly-correlated ones, removing the lesser representative from further analyses (Zuur et al., 2007). We used a General Linear Model (GLM) and a General Additive Model (GAM) using a binomial distribution, to identify which selected environmental variables mostly affected fish presence/absence, and compared the results between those analyses. We tested both methods because we did not have a priori reason for using a particular model.
We proceeded with the same multivariate analyses to describe the grouping behavior of Dentex. In order to do so, we firstly ranked images depending on variable number of pictured individuals (i.e., starting from 1). That frequency of groups of individuals was compiled into a frequency histogram plot. Then, we transformed the number of individuals into a nominal variable for grouping or not grouping behavior (i.e., "0" when in the photo there was only one individual, and "1" there was more than one individual). Then, we added this column of values to the temporal variables (i.e., hours, months and years), to detect any temporal pattern for this social behavior and identify which environmental variables affect it. We interpreted the data based on ethological common use of the wording (as per the general definition Pitcher, 1983). Thus, we consider the occurrence of the grouping behavior as the co-presence of fishes in the same field of view of the camera.
The correlation analysis was carried out with the library "PerformanceAnalytics, " and GLM and GAM models were executed using the libraries "gdata" and "mgcv" of R software.

Time Series Analysis
In order to obtain a global overview of Dentex diel and seasonal behavioral rhythms across consecutive years, we first plotted the 8-years visual counts time series computing the means and standard errors (SE) values per each month of the time series. Temporal gaps in image acquisition were evidenced by line discontinuity. Time series analysis was performed separately for time series of fish counts and each relevant environmental variable for the presence/absence of Dentex evidenced by GLM and GAM modeling (see previous Section). All graphic outputs were again plotted in local time.
Waveform analysis was carried out to describe the diel and seasonal pattern of activity rhythm of the species. Waveforms computing was as follows: time series of visual counts were subdivided in 30 min time-series and averaged together over a standard 24-h period (i.e., 48 values per segment). A consensus averaged fluctuation over that standard 24-h period was then obtained by averaging all values of the different segments at the corresponding timings. The resulting means (± SE) were plotted to identify peaks and troughs in the waveform profile. The peaks temporal amplitude (i.e., the phase) was then computed according to the Midline Estimating Statistic of Rhythm (MESOR) method (Aguzzi et al., 2006), by re-averaging all waveform averages and the resulting value was represented as a threshold horizontal line superimposed onto the waveform plot. The Onset and Offset timings of activity (delimiting peaks intervals) were estimated by considering the first and the last waveform value above the MESOR. The peak was considered as continuous if no more than 3 values occurred below the MESOR (Aguzzi et al., 2020d). All waveform analyses were carried out using the library "ggplot2" of R software.
That waveform analysis was firstly conducted on the fish 8years count time series and solar irradiance data, to visualize the general peaks as a proxy for the solar-driven, behaviorally induced changes in abundance as a product of behavioral activity (i.e., the photic character of the species ecological niche). Then, the same waveform analysis was repeated for each month and each season, by joining time series counts for winter (i.e., December, January, and February), spring (i.e., March, April, and May), summer (June, July, and August), and autumn (i.e., September, October, and November), to assess peaks' timings and amplitude variations as marker of photoperiodic regulation of behavioral rhythms. Moreover, to better describe the seasonal behavior of Dentex, and its relation with the photoperiod, we plotted the mean values (± SE) and MESORs of number of counts and solar irradiance of each month of the year. Finally, the same waveform analysis was performed for those environmental variables selected by models of presence/absence data (see previous Section).
Additionally, we assessed precisely the average values of those environmental variables selected by GLM and GAM modeling for presence/absence data (see previous Section) at Dentex waveform peaks crossing MESOR (see above), in order to add information on the species multidimensional niche (sensu Pocheville, 2015). At the same time, to better describe the environmental and temporal pattern of grouping behavior, we additionally plotted conditional densities of the environmental variables selected by GLM and GAM models of grouping or not grouping behavioral data.
An integrated chart depicting the temporal relationships of waveform peaks (i.e., the phases) in fish counts and the solar irradiance was created month by month over the whole 8 years of data acquisition (Aguzzi et al., , 2015a. The values of each monthly waveform were compared with the respective MESOR through an inequality function in Excel (i.e., each waveform value per 30 min automatically resulted as "major" or "minor" in relation to the MESOR). All waveforms values identified as greater than the MESOR (i.e., the peak duration) were then plotted as horizontal continuous bar per each month. That operation was repeated for solar irradiance.

RESULTS
A total of 140257 photos should have been obtained during 8years of monitoring (i.e., one per 30 min, from 2012 to 2019), but due to several malfunctioning problems creating gaps in the time series, we were able to analyze only 7,0254 photos (50.09% out of the total expected photos). The 95.99% of analyzed images (67438 photos) contained no Dentex (i.e., has "zero" as count value), and a few observations had high abundance (e.g., in only 5 photos there were more than 8 individuals; 0.18%).
We counted a total of 3,747 individuals of Dentex. The three months with the highest number of individuals were ( Table 1)   July with 615 counts (16.41%), and finally October with 494 individuals of Dentex counted (13.18%). By compiling this time series into monthly estimates (± SE), we observed a consistent seasonal trend in Dentex counts (Figure 2). A major peak occurred in spring-summer and its height progressively increased over the consecutive years.

Multivariate Statistic
From the correlation analysis among the environmental variables, we observed a significant relationship between water and air temperatures (Correlation Index = 0.69) (Figure 3). Accordingly, we removed the air temperature as explanatory variable from the further analysis. We did not eliminate water temperature because it was considered a more biologically important variable for Dentex.
We observed that in both GLM and GAM models on presence/absence data all the variables were significant at the 5% level, except for salinity, wind direction and rain (Supplementary Table 1). Both approaches gave a model where water temperature, wind speed and solar irradiance were selected ( Table 2 and  Supplementary Table 2). So, we selected these variables for the next time series analysis.
To study the grouping behavior of Dentex, we computed the percentage on the total number of images where it was present (2816 photos; Figure 4). Mostly, it was observed as solitary (2231 photos; 79.23%), but more rarely it appeared in FIGURE 3 | Correlation chart among the environmental variables. The name of each variable is shown on the diagonal. Below the diagonal the bivariate scattered plots with the fitted line in red are displayed. Above the diagonal the value of the correlation plus the significance level as stars: to p-values of 0, 0.001, 0.1, 0.05, 0.1, and 1 correspond respectively "***", "**", "*", ".", and "". pairs or in larger groups. In particular, in 395 photos (14.03%) it occurred in pairs, in 169 photos (6%) it occurred in groups of 3-5 individuals. Finally, it was observed in groups of 6-8 or more individuals (i.e., 16 and 5 photos respectively, equal to 0.57 and 0.18%). The maximum number of individuals in a single photo has been detected during 27th July 2019 at 8:00 in a group of 11 individuals.
Afterward, we carried out correlation analysis between environmental and temporal variables observing that there was a significant relationship between water and air temperature (Correlation Index = 0.69) (Supplementary Figure 1). As before, we removed air temperature as explanatory variable.
Then, we performed GLM and GAM models on the grouping or not grouping behavioral data (respectively when Dentex was observed alone or in group of two or more individuals) with the selected environmental and temporal variables. In both models all the variables were significant at the 5% level, except for wind speed and direction, solar irradiance, rain  Table 3). We observed in both approaches that water temperature, solar irradiance, hours, months and years were selected as relevant variables for the grouping behavior of the Dentex (Table 3) (Supplementary Table 4). It has to be noted that solar irradiance and month were slightly less significant than the other variables regarding the p-values (respectively Pr (> | z|) = 1.27 * 10 −02 and Pr (> | z|) = 3.82 * 10 −03 ). We decided to report only GAMs upon GLMs results for both presence/absence and grouping or not grouping behavioral data, even if the two methods obtained same outputs, because GAMs models were considered an extension of GLMs.

Diel and Seasonal Fish Count Patterns
The waveform analysis on the 8-years' time series showed the occurrence of a solid diurnal peak, defining an increase   of occurrence in the light hours ( Figure 5). That waveform analysis repeated at the seasonal level (Figure 6) evidenced the photoperiodic regulation of occurrence with transient uni-and bimodality in counts peaks: crepuscular and diurnal peaks during respectively short and long photophases (i.e., autumn-winter versus spring-summer). Also peaks temporal limits are following irradiance temporal limits.
In the plotting of mean counts per month of Dentex vs. mean solar irradiance depicting the overall seasonal fluctuation trend in local abundance evidenced a general increase from winter to summer, with two peaks, a major on August and a minor on May (Figure 7). The increase of the solar irradiance follows a similar pattern but with a peak in June (Figure 7). In accordance, the waveforms MESORs values of Dentex for the different months (see Table 1) is increasing from February, when this average value is at the minimum, to August, when this average value is at the maximum (i.e., 0.007 and 0.122 individuals per photo, respectively). At the same time, the MESORs values of the solar irradiance are increasing from a minimum in December to a maximum in June (i.e., from 76.23 to 297.95 W/m 2 , respectively) ( Table 4).
In the integrated chart comparing Dentex waveforms peaks (i.e., means values higher than the MESOR as horizontal continuous band) (Figure 8 and Supplementary Figures 2, 3) we could observe counts increases form December to June, with onset and offset timings that shift form 8:00 and 16:30, to 5:00 and 19:00, respectively (see Table 1). For solar irradiance peaks amplitude also varied from December to May, with onset and offset at 8:00/15:00 and 7:00/16:30, respectively (see Table 4). The integrated chart (see Figure 8) indicated that Dentex counts followed the solar irradiance pattern, with values of onset and offset that could anticipate and are delayed to the irradiance onset of maximum 2 and 2.5-h, respectively.
In order to describe the photic niche of Dentex, we noted that the average values of irradiance when Dentex averaged counts start spiking (i.e., is becoming active) as the peak onset; these are between 3 and 76.3 W/m 2 (see Table 1 and Supplementary  Figures 2, 3). Inactivity (i.e., offset) occurs for average values   Table 1 and Supplementary Figures 2, 3).

Environmental Cycles
In Table 4, we reported MESOR values, onset and offset per each month of the year for the environmental variable previously selected by GLM and GAM models for presence/absence data (i.e., water temperature, wind speed, and solar irradiance). The temporal dynamic of those variables is described below, but not for solar irradiance that was already described (see previous Section).
The water temperature cycle (Supplementary Figures 4, 5) had a phase shift to early hours from January, with an onset and offset at 15:30 and 6:30 respectively, to December, with onset and offset at 0:00 and 17:00 respectively. Furthermore, we reported that the water temperature had a minimum and a maximum MESOR value in February and August (i.e., 13.11 • C and 23.14 • C, respectively). One should notice that those two months also correspond to the minimum and maximum for Dentex.
The wind speed cycle (Supplementary Figures 6, 7) followed the same pattern of solar irradiance and fish visual counts (see previous Section). Its onset anticipated its timing from In order to describe the ecological niche of Dentex, we annotated the average values of the detected relevant variables from the statistical models when Dentex started and finished its active phase (i.e., onset and offset, respectively) (see Table 1 and Supplementary Figures 2-7). Indeed, when this species started to spike (i.e., is becoming active) as the peak onset, the values of water temperature and wind speed were, respectively, between 13.1-22.96 • C and 2.47-5.06 km/h (see Table 1). Instead, inactivity (i.e., offset) occurs, in average values, between 13.12-23.3 • C water temperature, and 4.01-8.5 km/h wind speed.
Plotting conditional densities for the most important explanatory variables of the grouping or not grouping behavioral data of Dentex (i.e., the distribution of the nominal variable for the grouping behavior of Dentex given a certain value of environmental and temporal driver) (Supplementary Figure 8), we observed that Dentex form groups during the day or at dusk and down, but not during the night. Moreover. The frequency of grouping increased along the years of observation. No particular seasonal pattern along the months of the year has been observed for grouping behavior. Furthermore, we could not obtain particular information on the relationship between grouping and the environmental variables selected by the models for this behavior.

DISCUSSION
We described the occurrence of diel and seasonal behavioral patterns in a coastal marine top predator, Dentex, by analyzing 8-years of high-frequency and continuous time series of visualcounts plus concomitant multiparametric oceanographic and meteorological data. Firstly, we detected a relationship between fish counts and the solar irradiance as a proxy for rhythmic activity. Then, a seasonal variation in video-counts was evidenced with a major peak in August and a minor one in May, suggesting for local abundance changes, possibly linked to population dynamics (e.g., seasonal migration). Also, the species counts were significantly correlated to water temperature and wind speed. Finally, we detected the occurrence of grouping behavior correlated to solar irradiance and water temperature, suggesting an effect of the environment as a regulator of grouping behavior.

Limitations in Cabled Observatory Monitoring Strategies
Cabled observatories provide a spatially limited data acquisition (a single platform can provide a relatively narrow field of view of few m 2 ). Another problem is that with this methodology it is not possible to separate the influence of abundance variation from activity variation, and the first one certainly affects the results of the second. Anyway, general inferences can be made on activity rhythms with spatially limited sampling windows (Hansteen et al., 1997;Refinetti et al., 2007;Bu et al., 2016;Gaudiano et al., 2021). Even trawling, which is the more spatially representative tool, is still anyway limited in comparison to the real extent of marine species distributions (Cama et al., 2011;Sonnewald and Türkay, 2012;Ünlüoglu, 2021). Furthermore, Campos-Candela et al. (2018) recently reviewed some methods to inference abundance from visual counts with cameras stating that averaged estimates of animal density do not show any substantial improvement after an adequate sampling effort (i.e., number of cameras and deployment time).
In our monitoring, fish were observed during daytime and this could cause more observations per day in summer than in winter, being the photoperiodic difference between months the cause for an increased probability in observing fishes in a summer day rather than in a winter day. In any case, there are diurnal species that are sampled more in winter for a reason that is related to an increase in their abundance and not to the possible effect of increasing photoperiod (Condal et al., 2012;Aguzzi et al., 2015a). Here, it is difficult to methodologically distinguish this abundance/activity/photoperiod phenomenon with the present methodology.
In order to acquire more representative results on rhythmic movements and habitat use of fishes at the scale of species distribution a better spatial coverage in monitoring would be needed (Holt, 2009). Networks of cameras with synchronous image acquisition routines may be required to track the species movements across different levels of habitat heterogeneity (e.g., Doya et al., 2017;Aguzzi et al., 2020b;Rountree et al., 2020). Such a synchronous image acquisition could clarify if the peaks in video counts of Dentex in different areas are associated to a different habitat uses (e.g., preying vs. resting), and then could be used to relate this information to the activity rhythms. Inspiration on how to set the network monitoring may be drown from spatially extended surveys with camera traps, aiming at the visual census of fauna in terrestrial environments (e.g., Beaudrot et al., 2016;Norouzzadeh et al., 2018).
OBSEA data collection could be implemented with other complementary actions within the monitoring area, such as the classic visual census sampling by divers (Samoilys and Carlos, 2000;Grane-Feliu et al., 2019), and collected data could be cross-checked with information provided by telemetry. This technology allows the tracking of particular individuals over large period of times (Hussey et al., 2015;Villegas-Ríos et al., 2017;Brownscombe et al., 2020;Matley et al., 2021). Acoustic telemetry could help achieving continuous long-term tracking of single individuals to study the habitat use of fish species (Hussey et al., 2015;Dominoni et al., 2017;Lennox et al., 2017), overcoming the spatial and temporal bias of fixed-point video monitoring for a reliable evaluation of population demography and local biodiversity (Aguzzi et al., 2020a,b). It is impossible with fixed cameras imaging technologies to support for fish "site fidelity" when this area specificity is not a clear life trait of the species (e.g., territoriality, burrowing, etc.). We have no morphological tools to identify the individuals, whose position and orientation changes within the field of view. For this reason, we may need acoustic tagging coupled with imaging to enforce such a site specificity study.
Cabled observatory imaging equipment could have some monitoring footprint on coastal areas for the introduction of light at nocturnal image sampling, which can induce behavioral disturbance on the local fauna (e.g., Davies et al., 2015;Kurvers et al., 2018;Czarnecka et al., 2019;Lucena et al., 2021). Nevertheless, in our case it is unlikely that the OBSEA lightening system, active every 30 min for about 3 s, affected the reported Dentex count patterns, being the individuals of this species absent at nighttime all the yearlong (see previous Section). However, the environmental footprint of future long-term monitoring could be reduced with the use of acoustic multi-beam cameras (Aguzzi et al., 2019).
Despite the evidenced monitoring limitations, we would like to stress out that one positive aspect of cabled observatory use is the low-invasive character at data collection. For example, visual census obtained by divers implies a factor of disturbance on the organisms as human presence (Harmelin-Vivien and Francour, 1992;Januchowski-Hartley et al., 2011;Assis et al., 2013;Azzurro et al., 2013;Emslie et al., 2018;Pais and Cabral, 2018).

How to Interpret Day-Night Rhythms in Dentex Visual Counts
Counts peaks timing and amplitude followed the photophase. In our case, the interpretation of video-counts peaks in terms of increase or decrease activity should be carried out with precaution. A similar precaution is adopted when evaluating the ecological meaning of species peaks in catches or visual census; i.e., animals captures or spotting are provoked by their increased availability in the sampling area for their resting or because of their activity (Aguzzi and Bahamon, 2009;. Notwithstanding, many species of fishes display activity rhythms (e.g., Eriksson, 1978;Muller, 1978;Helfman, 1986) that drive changes in abundance between day and night in coastal areas, as detected by different sampling systems and methodologies (e.g., Aguzzi et al., 2013;Hawley et al., 2017;Schalm et al., 2021). Diurnal, nocturnal, and crepuscular activity is often described as a product of fish behavioral response to solar irradiance variations (Helfman, 1986;Coles, 2014).
In this scenario, almost no Dentex was consistently detected at nighttime over several consecutive years. This observation suggests that video-counts peaks are a product of an increase activity at daytime. Laboratory data on fish behavior and physiology may provide a first insight on this phenomenon, assuming a link between visual counts and activity. Photoperiodic regulation of fish physiology and swimming behavior occur for the modulation that light intensity and temperature exert on the production of hormones (e.g., Pavlidis et al., 1999;Cowan et al., 2017;Sánchez-Vázquez et al., 2019). Fish melatonin measures environmental light levels and, as a result, variable rates of swimming occur (Saha et al., 2019).
A daytime activity increases for Dentex resulting in the increment of video-spotting at the OBSEA can be postulated for the following reasons. First, animals rest at nighttime within Posidonia seagrass beds (Zabala et al., 1992). Second, the species has a home range of less than 1 km 2 in specific period of the year , with the exception of moments in which a migration may follow bathymetric changes related to optimal water temperature (Aspillaga et al., 2017; see next Section). Third, Dentex is a visual predator whose prey spotting is optimized during light hours (Marengo et al., 2014).

Seasonal Fluctuation in Fish Video-Counts
Here, we reported a seasonal rhythm in visual counts of Dentex, with a significant increase in August and a second minor peak in May, as consistent across multiple years. In the past study of Sbragaglia et al. (2019) at the OBSEA, the major peak in counts of Dentex on August was detected, but not the minor one of May. This points out the strategic importance of a prolonged monitoring activity at the OBSEA. That seasonal pattern has been also detected with recreational fishing data for the Italian coasts (Sbragaglia et al., 2020).
We interpreted the first large peak of August as the product of thermocline regulation on fish behavior. Dentex shows a preference for warm suprathermoclinal waters, whose shallowest depths (i.e., between 20-30 m) are usually reached in our monitoring geographic zone (i.e., the NW Mediterranean) in July and August (Aspillaga et al., 2017). The OBSEA is placed within that depth range and this fact may explain the count increase of summer.
Another explanation could be that Dentex seasonal counts increase are synchronized upon maximum abundances of its preys (see previous Section), that augment in the OBSEA area in spring-summer; e.g., D. vulgaris from June to October, O. melanura in May and June, and S. maena from May to July (Aguzzi et al., 2015a). Seasonally synchronic abundance changes may occur between fish predators and preys (Fox and Bellwood, 2011;Bustos et al., 2015;Mishra et al., 2020;Liu et al., 2021). Possibly, the presence of artificial reef structures nearby the OBSEA attract fish preys and consequently concentrate the presence of the Dentex as well.
We observed a second, minor peak of Dentex counts in May that can be discussed in relation to the phenology of breeding. If from one side, the species migrates deeper to reproduce in areas at 40-100 m depth from March to June (Marengo et al., 2014;Grau et al., 2016), from the other we did not observe a temporally concomitant drop in counts at the OBSEA location in May (as an indication for a deeper migration of individuals in that period). Possibly, some individuals that have finished the reproduction (or with no mature gonads), return (or stay) to shallower depths for foraging. In fact, regressing ovaries in females and late developing testes of Dentex were already reported during May (Grau et al., 2016).

Species Relationship With Water Temperature and Wind Speed
The oceanographic and meteorological monitoring was dedicated to understand the species tolerance to certain ranges in the variation of selected measured habitat variables, as an indication of the effects that climate change may exert on fish's phenology (Stevenson et al., 2015). Those ranges have a practical value for ecological monitoring, since indicate a roadmap to develop smart sampling procedures in marine species: i.e., the optimum time window when to expect a maximum presence of individuals, according to the fluctuation status of key environmental drivers (Aguzzi et al., 2020b).
Here, counts of Dentex were related to the water temperature, being the seasonal peak always reported above an averaged threshold of 13.1 • C. A past study at the OBSEA with 3 years' time-series detected the increase in number counts of Dentex above 20 • C (Sbragaglia et al., 2019). This highlight the importance of pursuing the monitoring activities at the OBSEA to better characterize the environmental preference of this species.
The importance of water temperature as environmental driver has already been described in many fish species (Vinagre et al., 2016;Van Der Walt et al., 2021). Temperature deeply affects fish presence (or absence), because it influences directly species physiological performance (Cussac et al., 2009;Freitas et al., 2016;Day et al., 2018;Waldock et al., 2019). Dentex can cope with temperature range above our reported threshold, as also indicated by the current trend of geographic expansion in the North Mediterranean (Orozco et al., 2011;García-Rubies et al., 2013). We confirmed that trend by a progressive increase in counts over the years (i.e., see Figure 2), which would possibly continue in the next decade, when temperature is expected to grow in the NW Mediterranean (Bahamon et al., 2020). This indicates the value of cabled-observatory assets to disclose the occurrence of progressive trends in population shifts beyond more contingent seasonal dynamics due to the climate forcing.
We found a significant relationship between Dentex counts and wind speed. This variable affects the population distribution in some fish species (Daskalov, 2003;Teo et al., 2007;Bakun and Weeks, 2008;Selleslagh and Amara, 2008;Brander, 2010;Kuparinen et al., 2010), based on upwelling nutrient inputs (Bakun and Weeks, 2008;Bellido et al., 2008;Brander, 2010) although this phenomenon is not relevant in a shallow costal area, such as the one where the OBSEA is deployed.
The changes in wind speed and direction could also affect indirectly other environmental variables, that consequently affect the marine biota. For example, it was observed that changes in wind affected salinity in the North Sea and in the Baltic Sea (Schrum, 2001), which had negative consequences on cod recruitment in both areas (Brander, 2010). In our case, salinity was not significantly associated to counts of Dentex nor to wind. Hence, the same dynamic reported for cod recruitment in the North Sea and Baltic Sea may not be valid in our case. Notwithstanding, wind speed may resuspend and mix seabed and water column nutrients at periods of blowing, hence influencing the coastal food web with the consequent overall increase of trophism at all predator levels of the trophic food web (Bellido et al., 2008). For the overall increase in pray abundance, Dentex counts may consequently increase at moments of wind blowing.

The Grouping Behavior of the Species
We reported data on the grouping behavior of Dentex, that showed a clear 24-h modulation. Here, the formation of groups of Dentex significantly occurred more during daytime (including twilight hours) than nighttime, given the broad phase relationship between all visual counts and solar irradiance as a proxy for diurnal activity rhythms (see previous Section). Differently, no peaking was reported over different seasons. A seasonality for Dentex grouping behavior was described in rocky coastal areas for juveniles during summer (Chemmam-Abdelkader et al., 2004;Sahyoun et al., 2013). We did not observe this phenomenon, but we could not resolve if our video-monitoring were composed by individuals in this stage of development, since no tools for body sizing (e.g., lasers) were present aside the camera; however, we can assume that the majority of individuals were adults. Indeed, for adults Dentex, groups of individuals may be detected during the spawning season in spring, between 40 and 100 m depth (Marengo et al., 2014), but, given the shallower depth of OBSEA deployment, we did not observe this phenomenon (see previous Section).
The grouping behavior of Dentex was associated to solar irradiance and water temperature. Grouping has been already broadly correlated to the environmental variation in previous works for different fish species (Félix-Hackradt et al., 2010;Meager et al., 2012;Georgiadis et al., 2014;Palacios-Fuentes et al., 2020). In particular, the formation of fish groups has been related to light intensity (Meager et al., 2012;Georgiadis et al., 2014) and to water temperature (Power et al., 2000;Davoren et al., 2006;Meager et al., 2012;Palacios-Fuentes et al., 2020). Also, the weak increase of grouping behavior reported across consecutive years of observations (see Supplementary Figure 8) is likely the result of the increasing abundance of this species in the OBSEA area (see also previous Section).

DATA AVAIALABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
Ethical review and approval was not required for the animal study because we did not actively sampled, nor manipulated, nor killed any animals. We just performed video imaging.

AUTHOR CONTRIBUTIONS
MF: conceptualization, formal analysis, investigation, writing-original draft, and writing-review and editing. VS: conceptualization and formal analysis. ET, JAn, IM and JP: formal analysis. JdRF: funding acquisition. MNC and DMT: data curation. JAg: conceptualization, formal analysis, and writing-review and editing. All authors gave final approval for publication.

ACKNOWLEDGMENTS
We want to thank the members of the Development Center of Remote Acquisition and Information Processing Systems (SARTI) for the maintenance of the OBSEA seafloor platform. In particular we want to thank DMT, who is included in the RESBIO project, and MNC, Matias Carandell and Enoc Martinez, who performed the tasks for the maintenance of the OBSEA structure. We also acknowledge financial support from the Spanish Ministry of Science and Innovation (Juan de la Cierva Incorporación Research Fellowship to VS #IJC2018-035389-I), plus the funding from the Spanish government through the 'Severo Ochoa Centre of Excellence' accreditation (CEX20 19-000928-S).