Multi-platform study of the extreme bloom of the barrel jellyfish Rhizostoma pulmo (Cnidaria: Scyphozoa) in the northernmost gulf of the Mediterranean Sea (Gulf of Trieste) in April 2021.

. On 7 April 2021, an exceptional bloom of the scyphomedusa Rhizostoma pulmo was observed in the Gulf of Trieste (Italy). Blooms of this species in the northern Adriatic Sea have been reported since the late 1800s, however, the density of jellyfish observed in 2021 reached more than 10 specimens per square meter. In this work, we analyse the bloom from a multi-platform approach using observations and model data at different time scales. We attempt to explain the intensity of the bloom as a consequence of thermohaline and hydro-dynamical conditions in the Gulf. Meteo-oceanographic conditions that may have 5 contributed to the exceptional aggregation of jellyfish observed along the northernmost coast of the Adriatic Sea are discussed in detail. Specifically, our results indicate that this bloom was enabled by 1) the presence of a high number of jellyfish in the Gulf probably linked to the anomalously warm sea conditions in spring 2020 and winter 2021 which may have favoured a longer reproductive period and enhanced survival of adult R. pulmo respectively; 2) strong wind events, such as the Bora wind for the Gulf of Trieste, which enhanced upwelling and mixing processes in the Gulf bringing the jellyfish from the deeper 10 waters to the surface and clustering them along the coast.

Rhizostoma pulmo (Macri, 1778) (Figure 1), is one of the most abundant and largest jellyfishes in the Mediterranean Sea (Fuentes et al., 2011), where it is common in coastal areas and semi-enclosed lagoons. In recent decades, an increase in abundance and frequency of blooms of this jellyfish has been reported in southern European seas (Leoni et al., 2021a) and 20 especially in the northern Adriatic Sea (Kogovšek et al., 2018;Leoni et al., 2021a;Pestorić et al., 2021). In the Gulf of Trieste 1 Figure 1. R. pulmo clustering in the Gulf of Trieste. Photos were taken on the 8 April 2021 in the city of Trieste (left-hand and centre panels respectively) and in the city of Grado (right-hand panel) in Italy. These pictures were kindly provided by AvvistAPP users (https: //www.avvistapp.it, Tirelli et al., 2021) (GOT hereafter), R. pulmo was first reported in 1875 and has been observed since then, with the exception of the period 1930 to 1960 (Kogovšek et al., 2010): blooms were reported in five years between 1899 and 1914, in ten years from 1980 to 2010 and again in the period 2015-2017 (Pestorić et al., 2021).
The life cycle of R. pulmo begins with a pelagic, free-swimming stage (medusae) that reproduces sexually and releases 25 planulae, which in turn develop into polyps (benthic stage). The polyps of R. pulmo have never been observed in nature.
However, studies on laboratory cultures have shown that polyps reproduce asexually and multiply themselves by buds and podocysts, and, under favourable conditions, strobilate forming several ephyrae (polydisc strobilation with up to eight per polyp; Fuentes et al., 2011), which develop into new jellyfish. While R. pulmo is typically observed from summer to autumn in the central and southern Adriatic, this species is present in the GOT throughout the year (Pierson et al., 2020;Pestorić et al., 30 2021) being often very abundant also in the coldest months (Pierson et al., 2020). In early spring 2021, an extraordinary bloom (for the impressive density of jellyfish and the large size of many of them) was observed on the coast of Trieste (Italy), where R. pulmo were seen and photographed by citizens ( Figure 1) and reporters were attracted by this "anomalous event".
Jellyfish blooms are complex phenomena to study, and to date, knowledge about the location, size, and dynamics of R. pulmo, especially its polyp stage, in the Adriatic is sparse or non-existent (Pestorić et al., 2021). The lack of this information in the 35 Gulf of Trieste makes it difficult to understand how the jellyfish might respond to various environmental factors. Nevertheless, this study aims to uncover the role of hydrodynamics in the perception of jellyfish blooms in coastal areas, and provide a possible explanation for the magnitude of the April 2021 outbreak of R. pulmo. The main objective of this study is to show how hydrological properties of the water column and atmospheric forcing might have affected jellyfish aggregation in the GOT, highlighting the importance of a multidisciplinary and multi-platform approach to the study of jellyfish bloom dynamics. 40 This article is organised as follows: Section 2 presents the oceanographic characteristics of the study area. Section 3 describes the data used for this study. The data consist of observational and model data selected for their availability and temporal and spatial resolution. Section 4 is devoted to data analysis. Here we describe the sea conditions before, during, and shortly after the R. pulmo highest aggregation by comparing the above model and observational data. In section 5 the discussion of results and conclusions are given.

45
2 Oceanographic characteristics of the study area The Gulf of Trieste, the northernmost part of the Mediterranean Sea, is a mid-latitude semi-enclosed marginal basin situated in the Adriatic Sea ( Figure 2). Despite its relatively small size (∼ 20 × 20 km; average depth: ∼ 19 m, maximum depth: ∼ 25 m), the GOT together with the northernmost part of the Adriatic shelf are of great oceanographic importance as hotspots of the North Adriatic Dense Water (NAdDW) formation due to shelf convection (Pullen et al., 2007;Jeffries and Lee, 2007). The role 50 of dense shelf water formation in deep-basin processes has recently been perceived as important, changing the concept that only the open ocean plays a substantial role in driving thermohaline circulation in the global ocean and deep basins (Allen and Durrieu de Madron, 2009). The generated NAdDW flow as a dense current along the western Adriatic shelf, replacing the old waters in the middle and south Adriatic Pits (Jabuka and South Adriatic, respectively; Vilibić et al.,2004;Bensi et al., 2013;Vilibić and Mihanović, 2013). The NAdDW partly transforms into Adriatic Dense Water (AdDW) during deep-convection 55 processes in the south Adriatic, and the AdDW subsequently flows out of the Adriatic through the Strait of Otranto, sinking to the Ionian deep layers and affecting the whole deep Eastern Mediterranean (Robinson et al., 2001).
The main winds influencing the atmospheric and oceanographic circulation dynamics in the GOT are the cold-dry Bora with an ENE direction and the warmer Scirocco with a SE to SSW direction (Stravisi, 1977;Poulain and Raicich, 2001). Bora episodes are gusty and intense, blowing intermittently from land over the GOT, predominantly in autumn/winter. The Bora jet is 60 responsible for most of the mean net heat loss of the Adriatic Sea (Dorman et al., 2006;Cosoli et al., 2013;Raicich et al., 2013), as well as for the vertical mixing of the water column, dense water formation and the renewal of intermediate and bottom water masses (Querin et al., 2021). Although spring/summer bora episodes are weaker, it remains the main meteorological forcing in the GOT (Querin et al., 2006).
The GOT circulation is mostly cyclonic (counterclockwise) due to currents flowing northwards along the Istrian coast, 65 entering the gulf at its southern part. The circulation is daily modulated by the local atmospheric conditions (Querin et al., 2006) and by the inertial confinement of the Isonzo river plume to the northern coast of the GOT. From time to time, the circulation pattern changes anticyclonically (Cosoli et al., 2012;Querin et al., 2021), due to an increase in the freshwater input from the Isonzo river (yearly average flow rate varying from 100 to 200 m 3 s −1 , Covelli et al., 2004). Indeed, river discharge from the Isonzo, so as from the Dragonja, Rižana, and Timavo rivers (Figure 2), play an active role in the dynamics of the 70 gulf, collecting part of the precipitations from the land (Celio et al., 2002). During strong Bora events the water in the surface layer is removed offshore (westward), inducing a westward pressure gradient which generates a compensating eastward bottom counter-current inducing an upwelling/mixing along the eastern coastal area of the GOT to keep the mass balance (Querin et al., 2006; Ličer et al., 2016). At the open boundary, there is an intense water mass exchange between the GOT and the Adriatic Sea (Ličer et al., 2016;Francé et al., 2021).

75
The Gulf is also characterised by a shallow water column with large salinity and temperature variations (winter to summer: 10−38.5 and 4−29.2°C respectively, as described in Celio, 1997 andCozzi et al., 2020). When the water column is stratified, the surface layer is mostly wind-driven, while the lower layers exhibit the general compensating cyclonic circulation as described above. During winter the water column is completely mixed, while in spring intensified freshwater input and warming of the surface layer together contribute to the stratification which increases even more in summer (Malačič et al., 80 2001;Cozzi et al., 2020).

Data
In this study, we have analysed available data from different observational platforms in the GOT such as high-frequency radar and oceanographic buoys, CTD data from survey campaigns, satellite, model, and reanalysis data as summarised in Table 1. All types of data (except CTD profiles, hydrometric levels and satellite SST) were considered to characterise the period from 20 85 March to 20 April 2021, which spans over a period before and after the highest jellyfish aggregation, allowing the assessment of the stability conditions of the water column and its disposition for mixing. Satellite data were downloaded for the period between 1 January and 31 May 2021 and used to showcase the sea surface temperature trend in the GOT -while historical CTD ship measurements, from January 2008 to July 2021, were used to determine the climatological characteristics of the study area. Hydrometric levels were used to assess the river run-off input in the GOT.

Observational data
Half hourly high-frequency radar (HFR) combined current data (u-zonal, East-West and v-meridional, North-South) from two beamforming WERA (Gurgel et al., 1999) stations ( Figure 2, red triangles) working at 24.5 MHz and operating in the GOT on a spatial resolution of 1.5 km, were considered for the period shown in Table 1. The data were analysed by applying the quality control standards from the EU high-frequency node and averaged on an hourly basis to standardise with the temporal 95 resolution of CMEMS dataset. For the purpose of this study, the tidal component in HFR currents was not removed, as tidal forcing in the GOT is weak compared to the wind contribution and, in any case, the astronomical tide contributes negligibly to transport through small residual currents (Cosoli et al., 2012;Querin et al., 2021). The dataset are publicly available at http://150.145.136.27:8080/thredds/HF_RADAR/HFR_NAdr/HFR_NAdr_catalog.html.
Hourly surface (2 m) and bottom (15 m) temperature data from MAMBO1 buoy were analysed for the period described in 100   Table 1. The buoy is anchored at 45.692°N -13.705°E, 1.5 km away from the Italian coast in front of the Miramare Marine protected area (Figure 2, red star), and belongs to a network of oceanographic buoys in the Northern Adriatic Sea. Detailed information on the buoy and data can be visualised at http://nettuno.ogs.it/ilter/GoTTs/.
Wind speed and direction data at 10 m asl along with surface and bottom temperature (3 m and 22 m respectively) were downloaded for the period in Table 1  for the study period in Table 1. Data were then spatially averaged over the extracted area to depict the SST trend in the GOT.
This product is based on nighttime images collected by the infrared sensors mounted on different satellite platforms and can be accessed and downloaded at: https://resources.marine.copernicus.eu/ DOI: 10.48670/moi-00172 (product identifier, hereafter pi: SST_MED_SST_L4_NRT_OBSERVATIONS_010_004_c).
Further temperature and salinity measurements, collected on a prevalent monthly basis from multi-parametric sonde CTD at 115 Paloma site (∼ 45.613°N, ∼ 13.578°E, Figure 2, white star) were provided (upon request) by ARPA FVG and analysed for the whole available period in Table 1. The 13-year dataset consist of a time series of CTD data from the surface to 25 dbar of depth on intervals of ∼ 1 dbar.
Isonzo river hydrometric levels were measured every 30 minutes by the Servizio Idrografico-FVG with an ultrasonic water level CAE sensor. The data was provided upon request at a station positioned 14 km upstream of the Isonzo river mouth.

120
Data about the abundance of Rhizostoma pulmo in the Gulf of Trieste were obtained from the avvistAPP dataset (DOI: 10.13120/h127-9v54; Tirelli et al., 2021) available at EMODnet Biology (https://www.emodnet-biology.eu/data-catalog?module= dataset&dasid=7972) and in the reports that have been provided in the course of this study. Sightings were grouped in 3 intervals of abundance and values of abundance were attributed as follows: 1 corresponds to 1 individual per m −2 (hereafter ind m −2 ); 2 corresponds to 2 − 10 ind m −2 ; 3 corresponds to more than 10 ind m −2 . The daily R. pulmo abundance was obtained 125 by averaging the abundance of data collected on the same day. 6

Models and reanalysis
Hourly model outputs of current (pi: med-cmcc-currents-an-fc-h), salinity (S, pi: med-cmcc-sal-an-fc-hts) and potential temperature (θ, pi: med-cmcc-tem-an-fc-hts) with a horizontal spatial resolution of 6 (unequally spaced) vertical levels of the S and θ products (hts) cover depths from ∼ 1 to ∼ 20 m, while the 8 (unequally spaced) vertical levels of the current product (h) cover depths ranging ∼ 1 to ∼ 20 m. Surface currents (level 1, L1 ≈ 1 m), S and θ were spatially averaged over the extracted grid and potential density (σ θ ) was calculated using the MATLAB toolbox: ocean (http://mooring.ucsd.edu/software/matlab/doc/ocean/mindex_ocean.html). Model data were used to estimate the structure of the water column and its thermohaline properties, as well as the circulation in the Gulf, mainly to complete the temporal  Table 1 to assess whether the measurements from the coastal buoy VIDA are representative of the wind forcing in the centre of the GOT. For a complete description of the model see Termonia et al. (2018) and Strajnar et al. (2019).
Net surface heat flux was obtained from hourly atmospheric data of the European Centre for Medium-Range Weather Forecast (ECMWF) ERA5 (Hersbach et al., 2018) for the period resumed in Table 1 where Q SW R and Q LW R represent the radiative terms, and Q H and Q L represent the turbulent terms. Here, the positive 150 value in each flux indicates that heat is transferred from the atmosphere to the ocean (Tomita and Kubota, 2004;Tomita et al., 2021;Marullo et al., 2021).

Data analysis
Citizen science data collected with avvistAPP show that R. pulmo was seen in the GOT during the whole period of the time series in figure 3. The highest abundance (more than 10 ind m −2 ) was recorded in most sightings in April 2021. High 155 abundances of R. pulmo were occasionally reported in 2020, but the highest observed abundances during that year (figure 4) were always lower than those observed in April 2021 and represented jellyfish aggregations in a limited area, smaller than the large aggregation of R. pulmo observed in April 2021. To understand the causes behind the exceptional R. pulmo jellyfish aggregation observed in April 2021, we analysed the spatial and temporal variability of oceanographic and meteorological conditions in the GOT in the period prior to, during 160 and after the peak of jellyfish aggregation. Moreover, considering the important role played by the temperature on jellyfish physiology, sea temperature seasonality in the Gulf of Trieste over the last decade was also analysed.

Meteo-oceanographic variability
We analysed the available observed and modelled wind data (VIDA and ALADIN, in Table 1), along with the sea surface currents (HFR and CMEMS) time series from 20 March to 20 April 2021, a period encompassing the Bora and jellyfish 165 bloom events. As radar data only covered the surface layer and no current data were available along the water column, it was, therefore, necessary to use model data. To comply with the data and to understand the sea conditions before, during and shortly after the jellyfish bloom, spatially averaged sea surface currents (SSC) from CMEMS model and HFR observations were used to calculate SSC speed as shown in Figure 5. To assess the goodness of the modelled data the root-mean-squared error, RMSE, was normalized by the measure of the spread using the max-min difference. NRMSE and Pearson correlations between the 170 modelled and the observed dataset were calculated (Table 2) was stronger. Moreover, the NRMSE presents a value close to zero and the correlation coefficients are fairly good as shown in Table 2. VIDA wind measurements are consequently considered a good representation of the wind in the GOT and will be used henceforth in this study. Finally, a pronounced influence of the wind on the circulation of the GOT was supported by the significant correlation between SSC and wind, mainly along the u-component, as expected. The correlation for the with the maximum recorded at 13:30. SSC data also showed the effect of the strong Bora event, as SSC speeds increased and reached their maximums around the same time as the wind (Figure 5a shadowed area). Maximum values of SSC speeds were measured from HFR at 02:00 on 3 April with a peak value of 0.26 m s −1 . The aforementioned Bora episode lasted three days approximately followed by a wind calm on the morning of the 5 April (∼ 1 m s −1 ), and by two peaks: the first took place on 195 the evening of the same day with a wind speed of ∼ 8.37 m s −1 (∼ 30 km h −1 ), and the second on the early morning of the 6 April reaching 16.24 m s −1 (∼ 58 km h −1 ). The response of the SSC to this wind maximum is observed by the peak recorded on the morning of the 6 April (∼ 0.23 m s −1 ).  Figure 6 shows the output of CMEMS model (blue vectors) and HFR measurements (red vectors) on the 3 April at t 1 = 02:00 and t 2 = 06:00, and on the 6 April 2021 at 09:00 at 2 different levels representing surface (a, b, c, upper panels) and 200 bottom currents (d, e, f, lower panels). It can be seen that after the action of the wind, in the surface layer the water was pushed westwards, leading to offshore water removal (Figure 6a, 6b and 6c). Consequently, in the bottom layer the compensating eastward counter-current started to build up (Figure 6d, 6e and 6f). It is known that the Bora wind pushes water out of the GOT at the surface, particularly in conditions of low Isonzo river run-off (Querin et al., 2006), as it occurred during March and early April 2021 (Figure 7). To satisfy the mass balance, water from the bottom layer is pumped up vertically enhancing the 205 upwelling on the eastern side of the Gulf, thus replacing the water that was pushed away. During this process, the water masses below the surface are pulled inshore and the counter-current is set up, (Malačič and Petelin, 2009;Querin et al., 2006Querin et al., , 2021. Querin et al. (2006) showed that after 3 hours of the Bora wind onset the bottom layer separated from the surface layer and the direction of the horizontal bottom current was opposite to the wind forcing. The same process was observed in spring 2021 when on the 3 and 6 April, after a couple of hours, the wind had set up the bottom counter-current (Figures 6b, 6e and 6c, 6f).

210
The Bora displaced the surface water from the gulf, stranding many jellyfish on the coast of Grado (right-hand panel in Figure   1), and at the same time, thanks to the counter-current, it has pushed up the deeper waters transporting them towards the coast and carrying with it the jellyfish that subsequently accumulated on the shores of the city of Trieste (left and central panels of  As pointed out in the previous section, the effect of the mixing, the currents and the coastal upwelling, induced the bottom dense water of the GOT to reach the eastern flank. During this process, cooler subsurface water was brought up to the surface as can be seen from the daily satellite surface temperature (SST) maps in Figure 8a. Before the Bora event, the surface layer was warming as shown by SST maps from 31 March 2021 with values of ∼ 13°C to 14.5°C by 3 April 2021. During the Bora onset, the water column got thoroughly mixed, as shown by the SST map on the 5 April in Figure 8a, occupying the eastern 220 flank of the GOT with values ∼ 11°C. As expected, the response of the sea to the wind forcing was detected ∼ 2 days later.
After the generated upwelling and the mixing of the water column, a cold layer occupied the surface of the Gulf for several days (5 to 13 April in Figure 8a).
Satellite SST times series obtained by spatially averaging over the extracted area (Figure 8b)   During the whole Bora-bloom interval, the water column stratification disappeared (Figure 9) as an effect of the mixing 240 triggered by the heat loss and subsurface water masses being pumped vertically. While S and σ θ followed a similar pattern, θ showed two cores, one of ∼ 12°C affecting the surface to ∼ 9 m and the second one of ∼ 11°C from ∼ 9 m to the bottom. Full mixing of the water column was reached in the late evening of the 6 April and it lasted until the early hours of the 7 April for the whole basin (Figure 9b). The temperature plots depicted in Figure 8b, 9b and 9d show that a relatively cold layer (θ ≈ 11.5°C ) occupied the whole basin for several days after the Bora onset.

Temporal variability of net surface heat flux
The total heat exchange between the atmosphere and the sea, the main triggering process for the mixing of the water column, was obtained by calculating the daily net surface heat flux (Q N ET ), for the north Adriatic Sea from 20 March to 20 April 2021. The net surface heat flux ranges from −200 to 200 W m −2 , whereas for the Gulf of Trieste the maximum value was Q N ET ≤ 120 W m −2 (Figure 10). This positive Q N ET was seen from the 30 March to the 2 April indicating that the heat was 250 being transferred from the atmosphere to the sea, in agreement with the warming up of the surface layer and the stratification process during this time interval shown in section 4.2. During the Bora-bloom period, the net surface heat flux ranged from −200 to approximately +50 W m −2 in the GOT. It can be seen Q N ET decreasing to less than 50 W m −2 as the Bora strikes the GOT on the 3 April losing heat to the atmosphere. On the 4 April the surface net heat flux was negative Q N ET ≈ −10 W m −2 and the sea lost heat to the atmosphere. On 5 April, a little increment in Q N ET was observed in agreement with the 255 wind calm reported earlier in Section 4.1 on the same day. As the Bora began to strengthen, Q N ET reached its minimum in the studied period (∼ −150 W m −2 ) transferring heat from the sea to the atmosphere, an effect not only seen in the GOT but extended to the whole north Adriatic Sea (6 Apr 2021 in Figure 10). A similar pattern was reported by Dorman et al., 2006, Raicich et al., 2013and Cosoli et al., 2013 for other cases. This heat loss also contributed to the mixing of the water column which by 6 April was fully mixed (Figure 9c). Afterwards, from 7 to 12 April, Q N ET increased as the Bora wind died down 260 with values ranging from ∼ −10 to < 50 W m −2 respectively. However, another negative peak on 13 April was observed reaching values Q N ET ≤ −100 W m −2 corresponding to the wind and SSC peak on 13 April in Figure 5.

Seasonal temperature variability in the Gulf of Trieste
We analysed the seasonal temperature averages over the last 13 years of the available CTD casts described in section 3.1. The data were grouped in seasons as follows: winter-JFM (January, February, March), spring-AMJ (April, May, June), summer-265 JAS (July, August, September) and autumn-OND (October, November, December), where data were averaged for each season giving each month the same weight ( Figure 11).
The temperature time series for the 2008-2020 winter period averages between 8.8°C and 9.73°C showing two maxima in 2014 (11.98°C and 2020 (11.58°C ( Figure 11a) and a minimum in 2012 (7.5°C coinciding with the harshest winter in recent decades in the area (Bensi et al., 2013;Raicich et al., 2013). The spring period (Figure 11b), shows the trend of slightly  15% and 4% of the variance in the temperature (for winter and spring respectively) are explained by the linear trend. On the contrary, the linear regression calculated for summer and autumn is not statistically significant.

Discussion and conclusion
The life cycle of many gelatinous zooplankton, alternating sexual and asexual reproduction, implies the natural formation of large aggregations due to their rapid population growth, which typically occurs seasonally. Therefore, bloom of medusae like R. pulmo, which has a metagenic life cycle, should not be considered in itself an anomalous phenomenon. Nevertheless, the aggregation of jellyfish observed at GOT in April 2021 was particularly impressive for the amount of medusae observed (> 10 285 ind m −2 in some areas, Figure 3) and for the presence of many large specimens (Figure 1).
Rhizostoma pulmo is generally present in the GOT all around the year (Pierson et al., 2020) and this was documented also by the citizens' sightings of this species received in avvistAPP from July 2019 to August 2021 (Figure 3). The smallest individuals of R. pulmo are often observed in summer (Tirelli V., personal observation) but the lack of information about polyps at sea has prevented a complete description of the reproductive cycle of this species. Based on medusae semi-quantitative data and 290 satellite temperature data, Leoni et al. (2021a) identified the thermal window of R. pulmo in the Mediterranean sea between 15 and 22°C, while experimental observations pointed out a potential thermal niche at local scales of 13 to 29°C (Leoni et al., 2021b). This study confirms the reported range, but also shows that the thermal niche of this species is actually larger in the Adriatic Sea, where R. pulmo medusae occurred in winter at temperatures often lower than 10°C (Figure 11a).
A recent study of long-term records of R. pulmo in southern European seas (Mediterranean and Black Seas) (Leoni et al.,295 2021a) pointed out an "increase greater than expected" by the upward phase of the jellyfish population oscillation pattern assumed at the global scale (Condon et al., 2014). Specifically, the authors noted that the long-term intensity of the bloom and the biogeographic pattern of the species was determined by a latitudinal temperature gradient, but not by productivity: northern sites (low temperatures) showed less intense bloom events, while southern sites (high temperatures) showed the most intense bloom events. In this context, the observations of R. pulmo at GOT confirm the trend of increasing blooms in the last decades 300 (Kogovšek et al., 2010;Pierson et al., 2020;Pestorić et al., 2021) and at the same time represent an exception to the latitudinal temperature gradient, as it is one of the coldest areas of the Mediterranean Sea and simultaneously one of the areas where the blooms of R. pulmo were most intense, as in the period (early spring 2021) documented by this study. Based on molecular data, the northern Adriatic Sea harbours one of the three distinct populations of R. pulmo identified in the Mediterranean Sea (Faleh et al., 2017), whose adaptation to this environment remains to be investigated.

305
During 2008-2018, spring temperatures in the Mediterranean Sea were significantly correlated with the onset and duration of the R. pulmo season (defined, respectively, as the first and last observations of the species in a given area) (Leoni et al., 2021a). In the Adriatic Sea, R. pulmo showed an early occurrence during the warm spring and blooms' duration was also positively correlated with winter SST (Leoni et al., 2021a). These results support the possibility that increasing abundance of this jellyfish is related to increasing temperatures in this region: warmer springs could cause R. pulmo to appear earlier and 310 begin strobilation activity earlier, more intensely, and for a longer period of time (Purcell et al., 2012); additionally, warmer winters could favour polyps survival and lead adult stages last longer (Boero et al., 2016). This scenario seems to apply to what happened at GOT, where, according to the observed CTD data, spring and winter preceding the April 2021 bloom, were warmer than the same seasons in the 4 years before the jellyfish outbreak ( Figure 11). Furthermore, most of the jellyfish that formed the bloom were very large medusae, likely overwintering specimens born in 2020.

315
In their review of the physical relationships involved in the aggregation of gelatinous zooplankton, Graham et al. (2001) distinguish between rapid changes in jellyfish concentration due to rapid population growth (termed a "true bloom" by the authors) and those due to redistribution or redistribution of a stable population ("apparent bloom"). The numerous reports (from fishermen, citizens and scientists) of increasing occurrences of R. pulmo throughout the GOT in the weeks leading up to the April 2021 bloom, support the conclusion that the observed bloom of R. pulmo was a "true bloom" and not merely a 320 concentration of specimens from a stable population. At the same time, our study indicates that the perception of the extent of this bloom was influenced by the role played by the Bora wind in jellyfish aggregation along the coast near Trieste. After the onset of the Bora wind, we observed how the sea temperature at the surface decreased as a result of the generated upwelling and the subsequent mixing of the water column. This process is consistent with the results of Querin et al. (2006Querin et al. ( , 2021. At the surface, the water was pushed westward into the gulf towards Grado, where the jellyfish present at the surface were stranded 325 on the beach (right panel in Figure 1).
In the deeper layers, we observed the formation of a compensating counter-current, where after a few hours the surface and deep currents were completely opposing each other. Due to the effect of the Bora-enhanced mechanism (Bora -counter-current -upwelling), multi-platform temperature data altogether with the net surface heat flux showed the entire gulf was filled with a cold temperature layer for several days with the maximum heat loss as well as full vertical mixing found on the 6th of April, 330 just before the jellyfish bloom was reported by the media and citizens. Although Rhizostoma medusae cannot be considered drifting because they are actively swimming and R. octopus has been shown to even swim against the current (Fossette et al., 2015), it is very likely that the current generated by the Bora wind (> 0.25 m s −1 at the surface and > 0.15 m s −1 at the bottom) was too strong for R. pulmo to overcome. Therefore, we suggest that the upwelling generated on the eastern flank of the GOT by the counter-current, caused the numerous jellyfish hidden in this layer to gather on the coast of the city of Trieste 335 (Figure 1, left and middle panels). The quantity of jellyfish present in the area in front of the city of Trieste was absolutely impressive (> 10 ind m −2 ) and persisted in some areas for a couple of weeks: the Bora wind made the R. pulmo medusae, otherwise distributed in water column, visible directly from the wharves surrounding the city of Trieste.
Although Bora events and their dynamics have been well documented for the GOT (Cosoli et al., 2012;Querin et al., 2006;Ličer et al., 2016), to our knowledge this is the first study that contextualises jellyfish aggregation and meteo-oceanographic 340 conditions in this area. This study demonstrates the necessity of a multidisciplinary and multi-platform approach to understand jellyfish bloom dynamics. Acknowledgements. The authors would like to thank all the people whose contributions made this research possible. In particular, we would provided valuable data. We thank ARPA FVG technicians for marine monitoring in the GOT. Finally, the authors also thank the citizens who used the AvvisAPP app, which provided important data on jellyfish sightings. We acknowledge Dr. Neva Pristov and the Slovenian Environment Agency (Agencija Republike Slovenije za okolje, ARSO) for the meteorological data of the ALADIN model. This study has been partially developed in the framework of the Interreg MED Strategic Project SHAREMED, co-financed by the European Regional Development Fund under the Funding Programme Interreg MED 2014-2020. Website: https://sharemed.interreg-med.eu/.