sign in sign up
<<
Home , Channel (geography), Coastal fish

Research 219 (2019) 105318

Contents lists available at ScienceDirect

Fisheries Research

journal homepage: www.elsevier.com/locate/fishres

Fish movements and schooling behavior across the tidal channel in a Mediterranean coastal : An automated approach using acoustic T imaging ⁎ Fabrizio Capoccionia, Chiara Leoneb,e, , Domitilla Pulcinia, Massimo Cecchettic, Alessandro Rossid, Eleonora Ciccottib a Centro di ricerca “Zootecnia e Acquacoltura”, Consiglio per la Ricerca in Agricoltura e l'Analisi dell'Economia Agraria (CREA), Via Salaria, 31, 00015, Monterotondo, Rome, Italy b Dipartimento di Biologia, Università degli Studi di Roma “Tor Vergata”, Via della Ricerca Scientifica snc, 00133, Roma, Italy c Reparto Carabinieri Biodiversità di Fogliano, Str. di Fogliano, 04100, Latina, LT, Italy d Crisel Instruments S.r.l., Via Mattia Battistini, 177, 00167, Roma, Italy e CoNISMa, Piazzale Flaminio 9, 00196, Roma, Italy

ARTICLE INFO ABSTRACT

Handled by: George A. Rose A method for fish monitoring by high-frequency acoustic camera, coupled with an automated method for fi Keywords: counting sh, was implemented that allowed to analyze very large amounts of data with reduced costs. The counting acoustic camera was mounted in the tidal channel of the Caprolace lagoon (central Italy) and programmed for Acoustic camera 12 h. continuous recording at night time, between October 2016 and February 2017, for a total of 413 h, that were automatically processed by a specific software routine. A total of 266,717 fishes passed across the acoustic Fish schools cone of the camera, and two typologies of schooling events were documented, based on fish numbers making up Automation the groups. Large (n° ≥ 10 fish) school events (n = 884) maximum density occurred in early morning hours, in Monitoring coincidence with low ; on the other hand, small (n° ≥ 5 and < 10 fish) schools were more numerous (n = 5,394), and did not show a clear distribution pattern related to night hour and tidal phase. The individual acoustic area was different in the two schooling typologies, average single fish size being significantly higher in large schools. Acoustic images recording effectiveness was tested by evaluating fish movements dynamics be- tween a Mediterranean coastal lagoon and the adjacent sea. The method proved to be an efficient, versatile and portable system for fish monitoring in coastal and in general in transitional environments, and it allowed to describe schooling behavior and abundance of marine migratory fish during the winter migration season.

1. Introduction over 400 coastal lagoons located in the Mediterranean region, that range from very large to extremely small in size, for a total area of The quantification of fish abundance through space and time, and 640,000 ha (Aalto et al., 2015; Cataudella et al., 2015). All provide the understanding of fish behavior patterns have long been a concern nursery areas, feeding grounds and migration pathways for many for biologists and fisheries managers (Marini et al., 2018). This is par- marine fish species (Pihl et al., 2002; Franco et al., 2006), that enter ticularly true for transitional environments, such as coastal lagoons, lagoons as juveniles to feed and grow in a habitat where competition where the high supports fish yields, that are typically used and predation are minimized (Franco et al., 2008; Pérez-Ruzafa et al., by traditional aquaculture and capture fisheries in many parts of the 2010). Attaining maturity, they migrate to the adjacent sea coastal world. areas for reproduction, often using to facilitate swimming (Se- The ecological features of these environments, and primarily lective tidal transport – STST; Gibson, 2003), thus contributing exchange dynamics with the adjacent open sea, are of the utmost im- to sustain offshore stocks of many valuable species, thereby supporting portance in determining lagoon water quality and trophic state, as well coastal fisheries in most Mediterranean zones (Pérez-Ruzafa et al., as influencing composition and abundance of biota, and in particular of 2011). fish communities (Pérez-Ruzafa et al., 2007). This is also the case of the Knowledge on life cycles features of fish in relation to their shifts

⁎ Corresponding author at: Dipartimento di Biologia, Università degli Studi di Roma “Tor Vergata”, Via della Ricerca Scientifica snc, 00133, Roma, Italy. E-mail address: [email protected] (C. Leone). https://doi.org/10.1016/j.fishres.2019.105318 Received 5 April 2019; Received in revised form 20 June 2019; Accepted 21 June 2019 0165-7836/ © 2019 Published by Elsevier B.V. F. Capoccioni, et al. Fisheries Research 219 (2019) 105318 between lagoon and sea for colonization and escapement, as well as the exploiting the lagoon living resources on a sustainable basis, was de- understanding of the role of environmental parameters influencing finitively closed in 2007 (Fusari et al., 2006). The Caprolace lagoon has these movements (Milardi et al., 2018), is extensive (Tournois et al., a rectangular shape due to reclamation interventions (1926–1935), and 2017). is oriented along the coastal line. Its area covers 226 ha, its mean depth Investigations aimed at highlighting and describing fish movements is 1.30 m (maximum depth around 2.90 m) and its average annual are usually based on the use of field tracking systems (Hussey et al., temperature and salinity ( ± SD) are respectively 19.4 ± 5.8 °C and 2015), but also of passive fishing gears that can intercept and sample 40.2 ± 2.6 PSU. Caprolace is directly connected to the sea by a tidal fish on movement routes or shifts (Milardi et al., 2018). These fishery- channel (250 m long and 28 m width) equipped with a fishing barrier, independent studies rely on costly research programs which generally called lavoriero, now dismantled, and a hydraulic weir for water ex- have a limited coverage in space, as well as in time (in terms of sea- change control. This channel ensures water exchanges between the la- sonality as well as in the number of years of available data), and hence goon and sea, and guarantees the movements of living organisms to and could lead to biased estimations because of unrepresentative sampling from the lagoon and the open sea. Tides appear as a 12 h cycle, not- (Hilborn and Walters, 1992). Quantitative information on the fish withstanding their reduced amplitude (about 45 cm, Leone et al., community structure and on fish behavior in coastal lagoons can also be 2016). derived from fishery-dependent surveys, that can provide long time Caprolace fish community is consistent with those typical of hy- series of data on a large variety of target species, often ensuring a wide peraline lagoons characterized, and consisting of a relatively high spatial coverage all year-round (Lunn and Dearden, 2006; Pennino number of marine migrants (Anguilla anguilla, Sparus aurata, et al., 2016). In this case, bias can be due to constraints imposed by Dicentharchus labrax, Lithognatus mormyrus, Diplodus spp., fi ve species of misreporting of catches or management options (Pennino et al., 2016). the Mugilidae), and marine stragglers (Epinephelus marginatus, Sardina The latter may represent a strong limiting factor in many transitional pilchardus, Engraulis encrasicolus, Trachinotus ovatus, Solea solea, Spicara habitats of the Mediterranean that today are included in protected flexuosa), along with resident species such as Atherina boyeri, Syngnatus areas, where fishing activities are therefore strongly limited or for- abaster, Aphanius fasciatus, Knipowitschia panizzae and other gobids bidden. (Mariani, 2001; Prato et al., 2014). In the last 20 years, hydro-acoustic methods have been widely used in fishery management to monitor fish stocks and movements because 2.2. Field surveys they are quantitative, non-invasive, fast and synoptic (Foote, 2009; Martignac et al., 2015). Despite the improvements achieved in the use In order to design an appropriate methodology to evaluate fish of acoustic techniques (e.g. single-beam, split-beam techniques, dual- migrations between the lagoon and the sea, a survey was set up aimed frequency identification sonar; DIDSON), the interpretation and clas- at monitoring acoustic images from a fixed location at a single point on sification of collected data are often challenging, requiring extensive the channel. The migration frequency was measured continuously for experience and effort (Jing et al., 2018). When dealing with noisy data several nights during six months - from mid-October 2016 to mid- or attempting to enumerate or identify fish near scattering boundaries February 2017, thus focusing on the period when most marine mi- (Holmes et al., 2006; Boswell et al., 2007a), often a complex post- gratory species return to the sea, after their residence in the lagoon processing is needed (Simmonds and MacLennan, 2005; Holmes et al., (Gonzalez and Gerlotto, 1998). Although fish activity can be influenced 2006; Boswell et al., 2007b). Petreman et al. (2014) sustain that post- by cycle, not all species are affected likewise (Katselis et al., processing simple fish counts from DIDSON raw data is labor-intensive 2007; Milardi et al., 2018; Kara and Quignard, 2018), and therefore and costly, an issue that is currently minimized through subsampling monitoring and samplings were carried out in coincidence with dark approaches in long study periods. Because acoustic cameras are con- moon, to minimize its effect. Night-time sampling was chosen, con- sidered powerful devices for detecting and quantifying fish movements, sistent with the observation that fish movements between brackish and appropriate methodologies for effective, automated processing are of marine mainly occur during night hours (Helfman, 1986). At utmost interest in managing large amounts of data, that can be col- daylight fish activity was not detectable acoustically because larger lected over a time range that could span from seconds to hours and even fishes tend to dwell near the bottom (Lucas and Baras, 2001). months. A temporary laboratory, powered from a Honda 1000 W generator Up to now, acoustic devices have been widely used mostly in riv- and equipped with a laptop for acoustic video storage, was set up in a erine environments, to evaluate fish migrations (Baumgartner et al., fishing house near the sea channel. A BlueView 900-130 Imaging 2006; Crossman et al., 2011), and enumerate Salmonids (Nyqvist et al., SONAR (© 2003–2006 BlueView Technologies, Inc. SubSeaTech s.r.l.) 2016; Jing et al., 2018), (Piper et al., 2015; Lenihan et al., 2018), was mounted close to the weir in the tidal channel and oriented per- lamprey (Keefer et al., 2017), cyprinids and percids (Lilja et al., 2003) pendicularly to the water surface (90° from the horizontal plane) swimming through dam bypasses. Little use of such devices has been (Fig. 1). This multibeam acoustic camera operates at 0.9 MHz fre- made in coastal lagoons, and scarce is the evaluation of their perfor- quency, and images a sample volume 130° wide by 20° deep with a fan mances in such environments (Gonzalez and Gerlotto, 1998; Brehmer of 768 individual beams, 1° wide and 20° deep. The acoustic camera et al., 2011, 2013). was mounted on an adjustable stabilized bracket that provided precise Against this background, the aim of the present study was the de- manual control of depth, roll angle and tilt angle. The sonar head was velopment of a novel, standardized acoustic video-automated proce- always placed at 1 m below the water surface, about half depth of the dure for the investigation of fish movements in coastal lagoons, that channel, where it was neutrally buoyant in water. In this position, could reduce the effort associated with fish counting in complex tran- movements of the transducer due to the , which could result in sitional habitats. blurred images, were prevented, and its orientation allowed to reduce the interference of fish acoustic shadows against the bottom. 2. Materials and methods Real-time videos were recorded to evaluate fish vertical distribution and passage with high resolution images. Records were also taken in 2.1. Study site dark and/or turbid water conditions that often occur in transitional water habitats, that conversely severely affect the use and reliability of Caprolace is a coastal lagoon located in central Italy (41°20′ N- conventional fish sampling techniques (e.g. electric fishing or fyke nets) 12°58′ E). Since 1975, it has been included within the Circeo National (Peirson and Frear, 2003). Range window length was set at 7 m from Park, and is managed by the Park Authority, that addresses habitat the transducer. In order to reduce, as much as possible, fish passage protection and conservation. A local artisanal fishery, aimed at outside the conical beam, fish counts were made at a distance of about

2 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Fig. 1. Acoustic camera location and ar- rangement: Sea channel at Caprolace lagoon photo (A) and top view picture of acoustic camera orientation under water surface (ca. 1 m depth) (B). Schematic top (C) and lateral view (D) of the sampling site, showing the deployment of the mechanical structure of the acoustic camera image system and diagram along the horizontal/lateral plane of the high- frequency sonar.

4 m from the channel , just in front of the weir opening - where considered, based on fish numbers: small schools (n° ≥ 5 and < 10) depth is about 2 m and fish passage is concentrated across the channel. and large schools (n° ≥ 10). Within large and small schools, individual The acoustic camera was programmed for 12 h. continuous re- fish size was also measured in pixels. Direction of fish movement was cording at night time (from 7:30 p.m. to 7:30 a.m.), for 37 days of easily discriminated in the recording, but no filtering of acoustic re- observation, about 8 nights per month between October 2016 and cordings was carried out because most movements were in the same February 2017, excluding short periods when the BlueView camera was direction, fish coming from the inner side of the lagoon and dwelling stopped for logistical reasons. across the channel. In concurrence with the acoustic survey, a Hydrolab 4a Datasonde By using MetaMorph software, 12 h. of recorded acoustic images (Hydrolab INC., USA) was installed in the tidal channel to collect hourly (about 13,000 frames) can be analyzed and exported in approximately water temperatures (˜ 1 m depth). Tidal periods were defined according less than one hour. The output Excel file contained fish numbers au- to the local tide calendar. tomatically counted per each frame analyzed, individual fish area di- mension (in pixels) and fish area intensity. The method, on the other fi 2.3. Post processing and data analysis hand, did not allow to identify sh species. Linear Pearson’s correlation was calculated to test relationship be- fi All video data were collected using the ProViewer 4.2 Software tween total sh counts, schooling events and water temperature. ff (Teledyne BlueView, 2014). The post-processing sampling procedure Di erences between single acoustic area of small and large schools – consisted of deleting any series of data blurred by suspension due to were analyzed using the Mann Whitney test. storms. During real-time acoustic video acquisition, a frame image, every three seconds, was recorded automatically. Based on eye video 3. Results analysis of fish passage and considering the average speed of the water in front of the device, this interval was set to capture the acoustic signal During the 37 night-time surveys, a total of 413 h of continuous of the same fish only once, in a pre-defined region of interest (ROI). A acoustic video images were recorded. A total of 30 h was discarded, 7% rectangular ROI (118 × 360 pxl), with the same position and dimen- of total video images acquired, due to background noise caused by sion, was set in all data analysis to obtain the best acoustic signal raided by heavy storms for three days of sampling. corresponding to a cross section of the tidal channel, ahead of the The automated process estimated, from the acoustic images, a total transponder, just in front of the hydraulic weir passage (Fig. 1A). This of 266,717 fishes passing in the tidal channel. Numbers ranged from procedure is a step of an automated protocol (“journal routine”) set up 2.45 to 34 fish/hr. The highest daily fish movements were observed using MetaMorph software for image analysis (Molecular Devices, San during the night between 11th and 12th October 2016, when about Jose, CA). At first, the software removed static background objects 38,443 fishes crossed the monitoring site ( 1). using a frame manually selected per each recording day, in order to Moreover, two typologies of schooling events, recorded during eliminate groups of pixels that would otherwise be erroneously detected acoustic surveys, were identified: a) small schools - when in the same as fish. Then, the pre-determined ROI was cropped and used for the acoustic frame, the number of fishes automatically counted ranged identification and count of items/fishes. This procedure is auto- between 5 and 9; b) large schools, characterized by a number of counts matically carried out by means of the “count nuclei” function. This tool ≥10 fishes. allows to set threshold and intensity settings to maximize contrast be- Their distribution, observed during the sonar surveys, was not tween fishes and background, and to produce the best raw image in homogeneous. High numbers of small schooling events occurred at each most situations. The parameters chosen were 10, 150 and 60 respec- day of maximum total counts, especially in 12 samplings with more tively for “Approximate min width”, “Approximate max width” and than 100 schooling events counted, all along the study period. During “Intensity above local background”. Such values were defined for the October, the highest number of small schools was recorded (2603). entire analysis to allow fish detection with a minimum distance be- Considering large groups, schooling behavior was noticeable in the late tween two items (corresponding to about 5 cm) (Fig. 2A). Since sa- autumn season: the majority of the days with high numbers of recorded turation of acoustic camera device can occur when multiple targets are large schooling events occurred in a very narrow period between mid- in the beam range, recognized items with pixel area > 800, mean in- November and mid-December (Table 1 and Fig. 3). tensity < 40 and integrated intensity < 3200 (intensity arbitrary units) Water temperature and total daily fish counts were negatively cor- were discarded. related (r = −0.380, p = 0.020), while the correlation between water The automated count protocol was able to detect and count temperature and small schools was positive (r = 0.427, p = 0.008). schooling events, i.e. when multiple fishes were automatically counted Conversely, large schooling events did not show any correlation with within the same acoustic frame image. Two school typologies were water temperature.

3 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Fig. 2. Elaboration process of acoustic images. A: load of raw image from video and selection of ROI (region of interest). B: ROI crop, removal of a static background objects by background image subtraction, and identification of acoustic items/fish with “count nuclei” function of Metamorph Image System software (Universal Imaging, Downington, PA).

Concerning schooling behavior, both group dimension typologies This result seems relevant considering that the hydro-acoustic methods were analyzed with respect to mean frequency distribution of schooling used in ecological monitoring studies are still at the beginning of their events during acoustic surveys (a standardization of schools counts over application history (Martignac et al., 2015). Although a variety of the effective recorded time) during night hours and tidal phases multibeam systems ― in particular dual-frequency identification sonar (Fig. 4). (DIDSON) ― are beginning to be included in monitoring of fish Large schools maximum density occurred in early morning hours (5, passages, their cost in terms of money and time is still high in com- 6 and 7 a.m.), whilst small schools were more evenly distributed all parison to optical systems, and need highly standardized conditions be along the 12 h. interval (Fig. 4A). Considering tidal phases, higher in order to produce reliable results (Petreman et al., 2014). frequencies of large schools were evident during low tide (outflow) although no significant differences were found for both schooling 4.1. Methodological aspects groups (Mann-Whitney pairwise comparison) (Fig. 4B). The analysis of single fish acoustic patch dimension revealed that individuals of large The system used in this study proved to be compact, lightweight, schools showed a significantly higher median area (in pixels) compared easy to position and effective in a transitional environment. In addition, to small ones (Fig. 5). the system is relatively low in cost when compared with some of its competitors. The quality of the images is partially limited by the phy- 4. Discussion sical nature of the instruments, and hence the challenge of species identification is still not possible. On the contrary, the customized As a non-invasive method, the acoustic camera system and the au- software routine set up in this study has allowed a reliable and fast tomated approach tested in this study provided quantitative and qua- counting of fishes passing through the sea channel, based on the con- litative information about fish movements in the sea channel of a siderable amount of acoustic data recorded (more than 400 h). As far as Mediterranean coastal lagoon during a considerable period of 4 months. we are aware of, no robust procedures for automatically counting fish

4 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Table 1 Data records of each sampling day including, mean water temperature, fish numbers, mean counting frequency and number of large and small schooling events recorded.

Acoustic sampling day Acoustic sampling date Automated Fish count Mean water temperature (1 m) Fish/hr Small schools ≥ 5 < 10 Large schools ≥ 10

1 11-Oct-16 38,443 20.3 3,785 2,618 15 2 22-Oct-16 28,487 20.2 2,448 1,394 4 3 25-Oct-16 16,306 20.1 1,436 28 0 4 27-Oct-16 2,678 19.8 238 2 0 5 29-Oct-16 8,133 19.4 691 11 0 6 30-Oct-16 2,268 19.2 216 4 0 7 31-Oct-16 6,786 19.2 623 15 0 8 02-Nov-16 4,757 19.7 445 16 0 9 05-Nov-16 19,898 17.6 1,813 861 14 10 13-Nov-16 3,561 15.5 306 30 1 11 18-Nov-16 1,895 16.3 406 47 1 12 21-Nov-16 12,422 16.9 1,070 435 76 13 24-Nov-16 17,096 16.8 1,518 556 36 14 26-Nov-16 24,537 16.1 2,062 905 27 15 28-Nov-16 5,016 14.3 445 64 10 16 29-Nov-16 14,417 13.2 1,225 603 185 17 30-Nov-16 7,224 13.6 652 191 44 18 02-Dec-16 15,111 13.8 1,263 621 218 19 04-Dec-16 5,064 13.9 437 139 92 20 11-Dec-16 3,586 13.5 321 15 0 21 20-Dec-16 6,761 13.0 574 337 131 22 23-Dec-16 4,549 12.6 405 180 15 23 26-Dec-16 1,271 12.6 111 0 0 24 28-Dec-16 1,488 11.3 126 8 0 25 29-Dec-16 383 10.7 36 4 0 26 30-Dec-16 728 11.5 64 7 0 27 01-Jan-17 996 12.5 83 0 0 28 04-Jan-17 2,329 10.9 199 37 4 29 15-Jan-17 1,695 9.6 145 13 0 30 20-Jan-17 489 10.3 41 15 1 31 23-Jan-17 3,142 10.4 274 161 10 32 25-Jan-17 553 10.8 47 3 0 33 27-Jan-17 561 11.1 53 0 0 34 29-Jan-17 1,222 11.2 106 22 0 35 31-Jan-17 403 12.1 34 0 0 36 02-Feb-17 1,053 12.4 89 59 0 37 12-Feb-17 1,409 12.1 140 1 0 using acoustic cameras are currently available (Rakowitz, 2009; advantage, coupled to the robustness of the software routine, opens Martignac et al., 2015), which means that only manual, or semi-auto- perspectives for its application to a number of issues of interest in mated, data processing have been carried out until now. The accuracy fisheries and fish management and conservation in coastal and inland of the software routine in counting was remarkable, and such to allow waterbodies. Such a method could allow to implement fish counters in the characterization of fish schools in terms of number of fishes and fish passages in , or be used for estimating fish stocks individual size. In the case under study, results enabled to infer some returning to sea from larger lagoons. information on behavioral patterns of movement across the tidal channel of a small-size lagoon, owing to the absence of disturbance due to the physical presence of sampling devices (Lilja et al., 2003). This

Fig. 3. Single fish and schooling events automated acoustic counts: Single fishes (grey peaks; first y axis) and schooling events counts (small schools, white histogram; large schools, black his- togram. Number of events are reported above) at Caprolace la- goon sea channel from 11th October 2016 to 12th February 2017. Second axis shows mean daily water temperature values.

5 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Fig. 4. Schooling behavior: Time (A) and tidal phase (B) fish schooling events frequency (number of counted events per hour) distribution, from 11th October 2016 to 12th February 2017.

Gibson, 2003), even if the degree to which tides influence fish activity and movements at this interface has not been completely understood. Recently Verhelst et al. (2018) found strong evidence that silver Eur- opean apply STST to reduce energy loss when they leave tidal systems to start the spawning migration. Therefore, the acoustic ima- ging method implemented might be of use in detecting and describing STST by fish in transitional waters. Finally, analysis of the acoustic characteristics of fishes belonging to the two schooling typologies, confirms that the individual acoustic area was different, average single fish size being significantly higher in large schools. Aggregated movements of larger fishes in restricted time in- tervals, coincident with the ebb tide, stand for the hypothesis that large schools are formed by a single species, whose specific movement pat- tern is related to season, hour of the night and tide (Kara and Quignard, 2018).

4.3. Species interpretation Fig. 5. Single fish acoustic area: Box plots showing the median, 25th,75th quartiles for detected acoustic area of fishes (in pixels) belonging to the two The monitoring based on the use of the acoustic devices tested in schooling groups. Whiskers indicate 1% and 99%. Different letters indicate this study does not allow to distinguish or identify the fish species on significant difference in medians. the basis of the images detected. Such an inference can only be drawn by knowledge on the life cycles of marine migratory fishes using the lagoon, integrated by empirical information contributed by fishery 4.2. Schooling behavior data. Historic data of fish yields from 1990 to 2006 at Caprolace lagoon Number of single fish counts increased with decreasing water tem- fixed barriers, provided by the National Park Authority, highlight that perature, conversely small schools were positively correlated. Also, gilthead sea bream, Sparus aurata (L. 1758), has always been the most large and small schools occurred in different moments during the important species in landings. Moreover, average monthly commercial winter season: the first in a restricted period of about one month (from fishery yields reveal how gilthead seabream catch, and hence its pas- mid-November to mid-December), while the second has been recorded sage through the tidal channel towards the sea, for more than 15 years over the entire span of the monitoring period. was concentrated in the last three months of the year (Fig. 6). This is The two different schooling behaviors also showed a different dis- consistent with the biological knowledge of the life cycle of this fish in tribution in relation to night-time and tidal phase: large schooling the Mediterranean area, where the main movement pattern on a sea- events were recorded mainly during first hours of the morning (5-6-7 sonal basis is the spawning migration (Katselis et al., 2007), whose a.m.) and during low tide periods (outgoing waters flow); on the other dynamics are related to water temperature and atmospheric pressure. hand, small schools did not show a clear distribution pattern related to Environmental cues, as photoperiod and water temperature, are often night hour and tidal phase. the trigger for fish to start the spawning migration (Bromage and Results then testify a major swimming activity and a specific Roberts, 1995). In brackish environments, temperature gradients be- schooling pattern during low tide, that facilitate fish in passing the tween transitional and off waters can also induce climatic fish channel towards the sea coastal area, and such a pattern during out- migrations (Jones and Miller, 1966). Katselis et al. (2007), in an ex- migration of fish from the lagoon seems driven by a selective tidal- tensive study, found that most of the biomass, over of 77%, of annual transport (STST). Concerning this aspect, the use of tidal transport has migrating S. aurata, was caught during its winter migration. Therefore, been described for both juvenile and large size adult fish, moving be- when fish biomass is dominated by a single species, that can be inferred tween transitional waters and the sea (Forward and Tankersley, 2001; by the integration with fishery data, species identification is then

6 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Fig. 6. Historic commercial landings: Average monthly commercial fish catches at fixed barrier (lavoriero) of Caprolace sea channel (1990–2007). Landings were calculated for each species in kg per month focusing on the four most abundant species present in the coastal lagoon. possible (e.g. Horne, 2000; Romakkaniemi et al., 2000; Lilja et al., References 2003). Aalto, E., Capoccioni, F., Terradez Mas, J., Schiavina, M., Leone, C., De Leo, G.A., Ciccotti, E., 2015. Quantifying sixty years of declining European eel (Anguilla anguilla L., 5. Conclusion 1758) fishery yields in Mediterranean coastal lagoons. ICES J. Mar. Sci. 73 (1), 101–110. Baumgartner, L.J., Reynoldson, N., Cameron, L., Stanger, J., 2006. Assessment of a Dual Intensive long-term monitoring of fish movements can offer op- frequency Identification Sonar (DIDSON) for application in fish migration studies, 84. portunities for understanding the biological and environmental factors NSW Department of Primary Industries – Fisheries Final Report Series. 33 pp. Boswell, K.M., Miller, M.W., Wilson, C.A., 2007a. A lightweight transducer platform for controlling their migratory behavior, also allowing to deal with quan- use in stationary shallow-water horizontal-aspect acoustic surveys. Fish. Res. 85, titative assessment of local fish stocks. Such opportunities could be of 291–294. extreme interest in providing evaluations and advice for coastal en- Boswell, K.M., Wilson, M.P., Wilson, C.A., 2007b. Hydroacoustics as a tool for assessing fi vironment management, thus implementing interactions between sci- sh biomass and size distribution associated with discrete habitats in a Louisiana . Estuar. Coast. 30, 607–617. entists and managers. Coastal lagoons and in general transitional en- Brehmer, P., Guillard, J., Pinzon, P.I.C., Bach, P., 2011. Exploratory and instantaneous vironment are increasingly considered crucial habitats for swimming speeds of Amphidromous Fish school in shallow-water coastal lagoon – conservation, as well as important habitats contributing to marine channels. Estuar. Coast. 34, 739 744. ff fi fi Brehmer, P., Laugier, T., Kantoussan, J., Galgani, F., Moullot, D., 2013. Does coastal la- o coastal sh stocks, and hence the evaluation of sh-related goon habitat quality affect fish growth rate and their recruitment? Insights from ecological services is important (Marcos et al., 2019). fishing and acoustic surveys. Estuar. Coast. Shelf Sci. 126, 1–6. The technological properties of acoustic cameras, such as video Bromage, N.J., Roberts, R.J., 1995. Broodstock Management and Egg and Larval Quality. Blackwell Science, Oxford. viewing to obtain data, can thus greatly improve monitoring of dia- Cataudella, S., Crosetti, D., Massa, F., 2015. Mediterranean coastal lagoons: sustainable dromous fish populations (abundance, distribution and behavior). Even management and interactions among aquaculture, capture fisheries and environ- if the challenge of species identification remains for a single frequency ment. General Fisheries Commission for the Mediterranean. Studies and Reviews N. 95. FAO, Rome 288 pp. camera such as BLUEVIEW tech., the combined tool (acoustic camera Crossman, J.A., Martel, G., Johnson, P.N., Bray, K., 2011. The use of Dual- frequency and software routine) developed in this study could represent an effi- IDentification SONar (DIDSON) to document white sturgeon activity in the Columbia cient, versatile and portable system for fish monitoring. Moreover, it River, Canada. J. Appl. Ichthyol. 27 (s2), 53–57. Foote, K.G., 2009. Acoustic methods: brief review and prospects for advancing fisheries could be applied to improve DIDSON acoustic images analysis which research. In: Beamish, R.J., Rothschild, B.J. (Eds.), The Future of in are more resolute and suitable for species discrimination (Langkau North America. Fish & Fisheries Series, pp. 313–342. et al., 2012). Future work shall focus on such implementations, as Forward Jr, R.B., Tankersley, R.A., 2001. Selective tidal-stream transport of marine ani- – species identification shall allow to gain further insight on the mi- mals. Oceanogr. Mar. Biol. 39, 305 353. Franco, A., Franzoi, P., Malavasi, S., Riccato, F., Torricelli, P., Mainardi, D., 2006. Use of gratory behavior of fish, its space and time dynamics and relationship shallow water habitats by fish assemblages in a Mediterranean coastal lagoon. Estuar. with environmental factors, an issue of paramount interest in Coast. Shelf Sci. 66, 67–83. https://doi.org/10.1016/j.ecss.2005.07.020. fi and coastal lagoons (Kara and Quignard, 2018). Thus, the possibility to Franco, A., Elliott, M., Franzoi, P., Torricelli, P., 2008. Life strategies of shes in European estuaries: the functional guild approach. Mar. Ecol. Prog. Ser. 354 (219-), 228. record and analyze very large amounts of data at reduced costs, could Fusari, A., Cecchetti, M., De Angelis, P., 2006. La pesca nei laghi costieri: aspetti storici e pave the way for incorporating acoustic technology into routine surveys situazione attuale. In: Zerunian, S. (Ed.), Pesci e pesca nei laghi costieri. Ufficio – – in transitional environments, also having a potential for quantitative territoriale per la biodiversità di Fogliano Parco Nazionale del Circeo, pp. 89 104. Gibson, R.N., 2003. Go with the flow: tidal migration in marine animals. Hydrobiologia assessments. 503, 153–161. Gonzalez, L., Gerlotto, F., 1998. Observation of fish migration between the sea and a Mediterranean lagoon (Etang de l’Or, France) using multibeam sonar and split beam Acknowledgements echo sounder. Fish. Res. 35, 15–22. Helfman, G.S., 1986. Fish behaviour by day, night and twilight. In: Pitcher, T.J. (Ed.), Behaviour of . Springer, pp. 479–512. https://doi.org/10.1007/978-94- This work was supported by the Italian Ministry of Agricultural, 011-1578-0_14. Food and Forestry Policy and Turism" within the research project Hilborn, R., Walters, C.J., 1992. Quantitative Fisheries . Chapman & “ Hall, New York XV, 570 pp. LAGURES Valutazione della capacità restituiva degli ambienti lagunari Holmes, J.A., Cronkite, G.M.W., Enzenhofer, H.J., Mulligan, T.J., 2006. Accuracy and a sostegno di stock ittici locali costieri: aspetti quantitativi e qualitativi precision of fish-count data from a “dual-frequency identification sonar’’ (DIDSON) nei siti pilota dei laghi costieri di Fogliano e Caprolace (Parco Nazionale imaging system. ICES J. Mar. Sci. 63, 543e555. https://doi.org/10.1016/j.icesjms. del Circeo, Lazio)”. The Authors are grateful to the Circeo National Park 2005.08.015. Horne, J.K., 2000. Acoustic approaches to remote species identification: a review. Fish. authorities and to the Ufficio Territoriale Carabinieri per la Biodiversità Oceanogr. 9, 356–371. di Fogliano (UTCB) for their support during the field activities, and to Hussey, N.E., Kessel, S.T., Aarestrup, K., Cooke, S.J., Cowley, P.D., Fisk, A.T., Harcourt, Andrea Fusari and Luigi Loffredi for their invaluable assistance during R.G., Holland, K.N., Iverson, S.J., Kocik, J.F., Mills-Flemming, J.E., Whoriskey, F.G., 2015. telemetry: a panoramic window into the underwater world. all the phases of the research.

7 F. Capoccioni, et al. Fisheries Research 219 (2019) 105318

Science 348, 6240. Santos, T., 2016. Migratory delay leads to reduced passage success of Atlantic Jing, D., Han, J., Zhang, J.A., 2018. Method to track targets in three-dimensional space smolts at a hydroelectric dam. Ecol. Freshw. Fish 26 (4), 707–718. https://doi.org/ using an imaging sonar. Sensors 18, 1. 10.1111/eff.12318. Jones, D., Miller, P.J., 1966. Seasonal migrations of the , Peirson, G., Frear, P.A., 2003. Fixed location hydroacoustic monitoring of fish populations microps (Krøyer), in Morecambe . Hydrobiologia 27, 515–528. in the tidal River Hull, north-east England, in relation to water quality. Fish. Manage. Kara, M.H., Quignard, J.-P., 2018. Fishes in Lagoons and Estuaries in the Mediterranean: Ecol. 10, 1–12. Diversity, Bioecology and Exploitation. Wiley, Hoboken 269 pp. Pennino, M.G., Conesa, D., López-Quílez, A., Muñoz, F., Fernández, A., Bellido, J.M., Katselis, G., Koukou, K., Dimitriou, E., Koutsikopoulos, C., 2007. Short-term seaward fish 2016. -dependent and -independent data lead to consistent estimations of migration in the Messolonghi-Etoliko lagoons (Western Greek coast) in relation to essential habitats. ICES J. Mar. Sci. 73 (9), 2302–2310. https://doi.org/10.1093/ climatic variables and the lunar cycle. Estuar. Coast. Shelf Sci. 73, 571–582. https:// icesjms/fsw062. doi.org/10.1016/j.ecss.2007.02.010. Pérez-Ruzafa, A., Mompeán, M.C., Marcos, C., 2007. Hydrographic, geomorphologic and Keefer, M.L., Christopher, C., Caudill, E.L., Johnson, T.S., Clabough, C.T., Boggs, P.N., fish assemblage relationships in coastal lagoons. Hydrobiologia 577, 107–125. Johnson, Nagy, W.T., 2017. Inter-observer Bias in fish classification and enumeration Pérez-Ruzafa, A., Marcos, C., Pérez-Ruzafa, I.M., Pérez-Marcos, M., 2010. Coastal la- using dual-frequency identification sonar (DIDSON): a pacific lamprey case study. goons: “transitional ” between transitional and coastal waters. J. Coast. Northwest Sci. 91 (1). Conserv. 15, 369–392. Langkau, M.C., Balk, H., Schmidt, M.B., Borcherding, J., 2012. Can acoustic shadows Pérez-Ruzafa, A., Marcos, C., Pérez-Ruzafa, I.M., 2011. Mediterranean coastal lagoons in identify fish species? A novel application of imaging sonar data. Fish. Manage. Ecol. an and aquatic resources management context. Phys. Chem. 36, 19, 313–322. 160–166. https://doi.org/10.1016/j.pce.2010.04.013. Lenihan, E.S., McCarthy, T.K., Lawton, C., 2018. Use of an acoustic camera to monitor Petreman, I.C., Jones, N.E., Milne, S.W., 2014. Observer bias and subsampling efficiencies seaward migrating silver-phase eels (Anguilla anguilla) in a regulated river. for estimating the number of migrating fish in rivers using dual-frequency identifi- Ecohydrol. Hydrobiol. 1–7. https://doi.org/10.1016/j.ecohyd.2018.07.001. cation sonar (DIDSON). Fish. Res. 155, 160–167. Leone, C., Zucchetta, M., Capoccioni, F., Gravina, M.F., Franzoi, P., Ciccotti, E., 2016. Pihl, L., Cattrijsse, A., Codling, I., Mathieson, S., McLusky, D.S., Roberts, C., 2002. Habitat Stage-specific distribution models can predict eel occurrence during settlement in use by fishes in estuaries and other brackish areas. In: Elliott, M., Hemingway, K.L. coastal lagoons. Estuar. Coast. Shelf Sci. 170, 123–133. https://doi.org/10.1016/j. (Eds.), Fishes in Estuaries. Blackwell Science, Oxford, pp. 10–53. ecss.2016.01.005. Piper, A.T., Manes, C., Siniscalchi, F., Marion, A., Wright, R., Kemp, P.S., 2015. Response Lilja, J., Keskinen, T., Marjomäki, T., Valkeajärvi, P., Karjalainen, J., 2003. Upstream of seaward-migrating European eel (Anguilla anguilla) to manipulated flow fields. migration activity of cyprinids and percids in a channel, monitored by a horizontal Proc. R. Soc. Lond. B Biol. Sci. 282, 20151098. split-beam echosounder. Aquat. Living Resour. 16 (3), 185–190. https://doi.org/10. Prato, S., La Valle, P., De Luca, E., Lattanzi, L., Migliore, G., Morgana, J.G., Munari, C., 1016/S0990-7440(03)00016-0. Nicoletti, L., Izzo, G., Mistri, M., 2014. The “one-out, all-out” principle entails the risk Lucas, M.C., Baras, E., 2001. Migration of Freshwater Fishes. Blackwell Science, Cornwall of imposing unnecessary restoration costs: a study case in two Mediterranean coastal 420 pp. lakes. Mar. Pollut. Bull. 80, 30–40. Lunn, K.E., Dearden, P., 2006. Monitoring small‐scale marine fisheries: an example from Rakowitz, G., 2009. Nafplion, GreeceDidson Counting - Manual or Automatic? Thailand’s Ko Chang . Fish. Res. 77, 60–71. Proceedings of the Underwater Acoustic MeaureMents: Technologies and Marcos, C., Gamito, S., Umgiesser, G., Pérez-Ruzafa, A., 2019. Vindicating the biological Results2009. Didson Counting - Manual or Automatic? Proceedings of the and socioeconomic importance of coastal lagoons and transitional waters. Estuar. Underwater Acoustic MeaureMents: Technologies and Results 6 pp. Coast. Shelf Sci. 216, 1–3. Romakkaniemi, A., Lilja, J., Nykänen, M., Marjomäki, T.J., Jurvelius, J., 2000. Spawning Mariani, S., 2001. Can spatial distribution of ichthyofauna describe marine influence on run of Atlantic salmon (Salmo salar) in the river Tornionjoki monitored by horizontal coastal lagoons? A central Mediterranean case study. Estuar. Coast. Shelf Sci. 52 (2), split beam echosounding. Aquat. Living Resour. 13, 349–354. 261–267. Simmonds, J., MacLennan, D., 2005. : Theory and Practice, 2nd edition. Marini, S., Fanelli, E., Sbragaglia, V., Azzurro, E., Fernandez, J.D.R., Aguzzi, J., 2018. Blackwell Press, Oxford. Tracking fi sh abundance by underwater image recognition. Sci. Rep. 8 (1), 13748. Tournois, J., Darnaude, A.M., Ferraton, F., Aliaume, C., Mercier, L., McKenzie, D.J., 2017. Martignac, F., Daroux, A., Bagliniere, J.L., Ombredane, D., Guillard, J., 2015. The use of Lagoon nurseries make a major contribution to adult populations of a highly prized acoustic cameras in shallow waters: New hydroacoustic tools for monitoring mi- coastal fish. Limnol. Oceanogr. 62 (3), 1219–1233. gratory fish population. A review of DIDSON technology. Fish Fish. https://doi.org/ Verhelst, P., Bruneel, S., Reubens, J., Coeck, J., Goethals, P., Oldoni, D., Moens, T., 10.1111/faf.12071. Mouton, A., 2018. Selective tidal stream transport in silver European eel (Anguilla Milardi, M., Lanzoni, M., Gavioli, A., Fano, E.A., Castaldelli, G., 2018. Tides and moon anguilla L.)—migration behaviour in a dynamic estuary. Estuar. Coast. Shelf Sci. 213, drive fish movements in a brackish lagoon. Estuar. Coast. Shelf Sci. 215, 207–214. 260–268. Nyqvist, D., Greenberg, L.A., Goerig, E., Calles, O., Bergman, E., Ardren, W.R., Castro-

8