From Plates to Baits: Using a Remote Video Foraging System to Study the Impact of Foraging on Fouling Non-Indigenous Species

Citation: Chebaane, S.; Canning-Clode, J.; Ramalhosa, P.; Belz, J.; Castro, N.; Órfão, I.; Sempere-Valverde, J.; Engelen, A.H.; Pais, M.P.; Monteiro, J.G. From Plates to Baits: Using a Remote Video Foraging System to Study the Impact of Foraging on Fouling Non-Indigenous Species. J. Mar. Sci. Eng. 2022,10, 611. https://doi.org/ 10.3390/jmse10050611 Academic Editor: Jean-Claude Dauvin Received: 9 April 2022 Accepted: 28 April 2022 Published: 30 April 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Journal of Marine Science and Engineering Article From Plates to Baits: Using a Remote Video Foraging System to Study the Impact of Foraging on Fouling Non-Indigenous Species Sahar Chebaane 1,2,* , João Canning-Clode 1,3 , Patrício Ramalhosa 1,4 , Janina Belz 5, Nuno Castro 1,6 , InêsÓrfão1, Juan Sempere-Valverde 1,7 , Aschwin Hillebrand Engelen 8, Miguel Pessanha Pais 6,9 and João Gama Monteiro 1,* 1MARE—Marine and Environmental Sciences Centre, Agência Regional para o Desenvolvimento da Investigação Tecnologia e Inovação (ARDITI), Edifício Madeira Tecnopolo, Caminho da Penteada, 9020-105 Funchal, Portugal; [email protected] (J.C.-C.); [email protected] (P.R.); [email protected] (N.C.); [email protected] (I.Ó.); [email protected] (J.S.-V.) 2Faculdade de Ciências, Universidade de Lisboa, Campo Grande, 1749-016 Lisboa, Portugal 3Smithsonian Environmental Research Center, 647 Contees Wharf Road, Edgewater, MD 21037, USA 4OOM—Oceanic Observatory of Madeira, Agência Regional para o Desenvolvimento da Investigação Tecnologia e Inovação, Edifício Madeira Tecnopolo, Piso 0, Caminho da Penteada, 9020-105 Funchal, Portugal 5Centre for Organismal Studies (COS), Heidelberg University, Im Neuenheimer Feld 360, 69120 Heidelberg, Germany; [email protected] 6MARE—Marine and Environmental Sciences Centre, Faculdade de Ciências da Universidade de Lisboa, Campo Grande, 1749-016 Lisboa, Portugal; [email protected] 7Laboratorio de Biología Marina, Departamento de Zoología, Facultad de Biología de la Universidad de Sevilla, Av. de la Reina Mercedes, 41012 Sevilla, Spain 8 CCMAR—Centre for Marine Sciences, University of Algarve, Campus de Gambelas, 8005-139 Faro, Portugal; [email protected] 9Departamento de Biologia Animal, Faculdade de Ciências da Universidade de Lisboa, Campo Grande, 1749-016 Lisboa, Portugal *Correspondence: [email protected] (S.C.); [email protected] (J.G.M.) Abstract: Marinas are a gateway for the introduction and establishment of non-indigenous species (NIS). In these habitats, competition and predation are crucial determinants for NIS establishment and invasiveness. However, fish trophic preferences and biotic effects inside marinas are poorly known. This study proposes a novel method that combines the deployment of settlement plates to recruit different assemblages, followed by their use as bait in remote underwater video systems. This combined approach, addressed as a remote video foraging system (RVFS), can record fish foraging behaviour, including feeding choices and their impacts on fouling assemblage composition. An experimental RVFS trial carried out in a marina of Madeira Island, Portugal (NE Atlantic), identified the Mediterranean parrotfish, Sparisoma cretense, as the most important fouling grazer in the area. S. cretense behaved as a generalist and increased the heterogeneity of fouling assemblages, which can hamper NIS dominance of the fouling and reduce the pressure of propagules from the marina to the natural environment. The RVFS tool was useful to understand the trophic links between foragers and fouling and has the potential to provide relevant information for the management of NIS introductions, establishment and spread. Keywords: marina; trophic interactions; cascade effect; biotic resistance; RVFS 1. Introduction The relocation of marine species outside their historical distribution has increased significantly in recent decades [ 1 – 4 ]. When non-indigenous species (NIS) find favourable conditions in their introduced ranges, they may become invasive. Invasive NIS can promote biotope changes, habitat modifications and loss of ecosystem services, leading to severe J. Mar. Sci. Eng. 2022,10, 611. https://doi.org/10.3390/jmse10050611 https://www.mdpi.com/journal/jmse J. Mar. Sci. Eng. 2022,10, 611 2 of 15 environmental, socioeconomic and human health impacts [ 2 , 5 ]. With a growing number of examples around the globe, marine invasions are currently one of the main drivers of biodiversity loss and an important element of global change [ 6 – 9 ]. In this context, understanding the factors underpinning NIS proliferation and success is an important issue for conservation and mitigation of global change. The proliferation of NIS is often shaped by a complex interaction of compounding factors such as climate change, environmental pollution, local community resilience and habitat degradation [ 10 , 11 ]. Although it is difficult to predict whether a NIS will succeed in a given environment, there are many hypotheses designed to explain invasion success (see Catford et al. [ 12 ]). For instance, some particular biological and ecological traits are shared by many invasive species [ 13 ]. Invasive species tend to have faster reproductive and growth rates and higher tolerances to extreme environmental conditions than native species [ 14 ]. Invasive species also often have traits that favour competition against native species and/or provide enhanced defences against predators, i.e., [ 11 , 12 , 15 , 16 ]. Some NIS with invasive traits have allelochemical defences such as toxins and non-palatable compounds that grant them a repulsive smell or taste [ 17 – 19 ], often granting them ecological advantages towards native taxa. In such cases, native consumers (predators and grazers) often reject feeding on NIS and selectively target a native-based diet. This behaviour can facilitate the proliferation and spread of NIS [ 20 – 23 ]. In contrast, local consumers that feed on NIS can control NIS dominance through trophic interactions, exerting a top-down control over potential invaders [ 24 – 27 ]. In this context, consumer–prey interactions can be particularly relevant, as predators may either facilitate or hamper a newly arrived NIS from successfully establishing and/or invading a new region by limiting its ability to spread from a site where it is already established [ 28 – 30 ]. As such, understanding whether local consumers exhibit selective or generalist feeding behaviours that favour or deter NIS dominance may represent a crucial mechanism and an essential piece of the puzzle when assessing resistance and susceptibility of local communities to biological invasions. Moreover, the presence of benthic NIS in the ecosystem can also alter the feeding preferences and foraging intensity of invertebrates and fishes [ 31 ]. Therefore, further research on trophic interactions between native and NIS communities and their biotic functional response to current and future climate scenarios is necessary to formulate an effective management strategy against NIS [32–35]. Efforts to better understand trophic interactions between local consumers and NIS communities in shallow coastal areas have primarily relied on indirect methods, such as exclusion experiments, to assess the influence of predators and grazers in shaping communities [ 36 ]. These studies manipulate consumer densities among treatments by caging and assessing the effect of predation pressure as a top-down control mechanism for NIS success. This approach does not inform which species are feeding on what, nor does it allow the selective exclusion of predator species to assert the ecological importance of a particular taxon [ 36 ]. In this regard, video surveillance methods may be a promising tool for studying trophic interactions in the marine environment, allowing researchers to identify predators and grazers, as well as foraging behaviours and prey selectivity [ 37 ]. Among video techniques, the use of baited remote underwater video (BRUV) to attract and quantify various motile taxa could be a powerful tool to study these interactions [ 38 , 39 ]. Oricchio et al. [ 40 ] provided an example of how this approach could be used to assess trophic interactions between fish and fouling recruited on caged polyvinyl chloride (PVC) panels. This method allowed identifying main predator species and which fouling species were targeted. Here, we advance this research approach one step further by combining the two methods and use video and imagery analysis for a more detailed assessment of prey–predator interactions. Specifically, we propose to assess how predator exclusion shapes different fouling assemblages (i.e., higher native vs. higher NIS “content”) and offer them as bait on a video setup so that one can inspect the imagery and the plates to identify predators/grazers, study feeding choices and assess the effects of existing preferences on the fouling community structure, diversity and abundances. J. Mar. Sci. Eng. 2022,10, 611 3 of 15 Marinas and harbours are considered as hubs for enhanced NIS recruitment, as they are under greater pressure from maritime traffic and human activities than other natural and artificial coastlines [ 41 – 43 ]. These can serve as stepping stones for NIS introduction into local natural habitats since they serve as a refuge for arriving NIS [ 14 , 44 , 45 ]. It is often in these areas that newly arrived NIS pass through several ecological filters such as prevailing temperature and salinity, pollutants in the water, and pressure from predators or grazers, before taking hold, establishing and being able to disperse to natural habitats in the vicinity [ 14 , 46 ]. With higher NIS diversity and abundance and with the facilitated access, marinas and harbours are often considered as the ground zero for marine biological invasions, where monitoring and early detection is likely to be essential for prevention practices, mitigation activities and adequate NIS management. In this context, the overall goal of this study is to (1) design, develop and test a combined method that leverages recruitment panels and a remote video foraging system (RVFS), custom designed to study trophic choices of foragers and their effects on fouling assemblages present in marinas and/or harbours; (2) provide guidelines for integrating the use of recruitment panels, RVFS and imagery analysis as a monitoring tool in marinas, and; (3) evaluate the success of RVFS as a tool to detect potential fish consumers that may provide biotic resistance against NIS in the marinas of Madeira. 2. Materials and Methods 2.1. Remote Video Foraging System We designed and built a custom made remote video foraging system (RVFS), a noninvasive system that avoids any contact or manipulation of the fish, complying with general bioethics in research and with the European Community Council Directive (Directive 2010/63/EU) on the protection of animals used for scientific purposes. The RVFS comes up as a combination of techniques used to study trophic interactions inside marinas: exclusion experiments (Figure 1A) and video experiments (Figure 1B). Exclusion experiments have been used worldwide to investigate the trophic effect of foragers on the fouling composition [ 36 , 47 , 48 ]. Video experiments can identify foraging species and their trophic choices [ 37 , 40 ]. To this matter, the proposed RVFS would help to investigate trophic choices of the fish foragers among fouling assemblages differing in variables such as fouling species composition and NIS dominance (Figure 1C). The different assemblages offered to the fish can be obtained by deploying plates in different environments or under different predation pressure. Then, plates will be used as bait units to perform the RVFS trials by offering all assemblages in a randomised blocks arrangement using a PVC structure with top and frontal cameras. The RVFS was tested in this pilot study in order to assess its applications and outline guidelines for its use in NIS monitoring and studies. 2.2. Pilot Study 2.2.1. Experimental Setup Thispilotstudywas carried out in themarina of Quinta do Lorde( 32◦44.50N, 016◦42.80W ), in Madeira Island, Portugal (Figure 2A), where the most common fish species present inside the marina had been previously catalogued based on sightings during underwater surveys (Table S1). Fouling assemblages used as bait units were recruited by haphazardly deploying 12 settlement plates [ 49 ], consisting of polyvinyl chloride (PVC) panels of 14 ×14 cm , grouped in three treatments. These were chosen to manipulate predator pressure and create artefact effects, which would provide three different assemblages. Treatments were: uncaged (T1), caged with 15 mm mesh (T2), and caged with 1 mm mesh (T3) (Figure 2B). Caged with 15 mm mesh was aimed to avoid predation, taking into account that meshes as small as 5 mm did not cause artefact effects in Madeira (see Gestoso et al. [ 48 ]), whereas caging with smaller mesh sizes (1 mm mesh) is expected to avoid predation and create an artefact effect that alters the structure of the assemblages [ 47 ]. Settlement plates were hung from the marina’s pontoons at 1 m depth and facing downwards from May to September 2018. After five months, plates were retrieved, photographed using an Olympus TG4 J. Mar. Sci. Eng. 2022,10, 611 4 of 15 camera, and checked under a binocular microscope (LEICA S8 APO, Portugal) for detailed identification of the fouling community to produce a taxa inventory. Settlement plates were also weighed individually using a scale (KERN KB 2000-2N, Portugal) (resolution of 0.01 g ) after holding them for 30 s in a vertical orientation to allow water to drain. To ensure non-destructive sampling and no loss of live coverage and biomass, the plates were kept separated from each other in tanks and submerged in seawater for less than two hours during this inspection and photo capture period. J. Mar. Sci. Eng. 2022, 10, x FOR PEER REVIEW 4 of 16 Figure 1. Traditional approaches used to study trophic effects and interactions between fouling and foragers inside marinas (A, B) and the proposed combined method (C). (A) Exclusion experiments study the recruitment and colonisation of settlement plates by fouling assemblages under a caging (predators exclusion) treatment against a procedural control (Ctrl 1), testing possible artefacts caused by the exclusion method (mesh used for caging), and uncaged plates (Ctrl 2). (B) Video experiments record fish trophic interactions with fouling species recruited on plates (e.g., caged settlement plates]). (C) Remote video foraging system (RVFS) technique developed and tested in this study is baited with different fouling assemblages (C.1 and C.2) recruited on settlement plates (e.g., caged and uncaged). 2.2. Pilot Study 2.2.1. Experimental Setup This pilot study was carried out in the marina of Quinta do Lorde (32º44.5’ N, 016º42.8’ W), in Madeira Island, Portugal (Figure 2A), where the most common fish species present inside the marina had been previously catalogued based on sightings during underwater surveys (Table S1). Fouling assemblages used as bait units were recruited by haphazardly deploying 12 settlement plates [49], consisting of polyvinyl chloride (PVC) panels of 14 × 14 cm, grouped in three treatments. These were chosen to manipulate predator pressure and create artefact effects, which would provide three different assemblages. Treatments were: uncaged (T1), caged with 15 mm mesh (T2), and caged with 1 mm mesh (T3) (Figure 2B). Caged with 15 mm mesh was aimed to avoid predation, taking into account that meshes as small as 5 mm did not cause artefact effects in Madeira (see Gestoso et al. [48]), whereas caging with smaller mesh sizes (1 mm mesh) is expected to avoid predation and create an artefact effect that alters the structure of the assemblages [47]. Settlement plates were hung from the marina’s pontoons at 1 m depth and facing downwards from May to September 2018. After five months, plates were retrieved, photographed using an Olympus TG4 camera, and checked under a binocular microscope Figure 1. Traditional approaches used to study trophic effects and interactions between fouling and foragers inside marinas ( A , B ) and the proposed combined method ( C ). ( A ) Exclusion experiments study the recruitment and colonisation of settlement plates by fouling assemblages under a caging (predators exclusion) treatment against a procedural control (Ctrl 1), testing possible artefacts caused by the exclusion method (mesh used for caging), and uncaged plates (Ctrl 2). ( B ) Video experiments record fish trophic interactions with fouling species recruited on plates (e.g., caged settlement plates]). ( C ) Remote video foraging system (RVFS) technique developed and tested in this study is baited with different fouling assemblages (C.1 and C.2) recruited on settlement plates (e.g., caged and uncaged). For RVFS trials, one randomly selected plate from each treatment (T1, T2 and T3) was offered as bait in a randomised block arrangement. A total of four RVFS trials, of 4 h each, were carried out using a custom-built structure with two Olympus TG-tracker cameras, suspended at 2 m depth from the marina pontoons (see Figure 2B). Cameras were synced and set up in time-lapse mode with 2 s intervals (0.5 frames per second). After each trial, the structure was retrieved, the plates detached, and these were re-weighted and re-photographed for analysis (see the conceptual design of the experiment in Figure 1C). Our RVFS trials were carried out in late September 2018 over a 2-day period, with one trial in the morning and one in the early afternoon. RVFS deployment was at the same location J. Mar. Sci. Eng. 2022,10, 611 5 of 15 where water temperature (measured every 4 h using a HOBO temperature data logger) ranged between 23.5 ◦C and 24.0 ◦C during the trial period. J. Mar. Sci. Eng. 2022, 10, x FOR PEER REVIEW 5 of 16 (LEICA S8 APO, Portugal) for detailed identification of the fouling community to produce a taxa inventory. Settlement plates were also weighed individually using a scale (KERN KB 2000-2N, Portugal) (resolution of 0.01 g) after holding them for 30 s in a vertical orientation to allow water to drain. To ensure non-destructive sampling and no loss of live coverage and biomass, the plates were kept separated from each other in tanks and submerged in seawater for less than two hours during this inspection and photo capture period. Figure 2. Pilot study experimental setup. (A) The study was conducted in Marina of Quinta do Lorde, Madeira Island, Portugal. (B) The fouling assemblages used as bait were obtained using caging treatments (T1: uncaged, T2: caged 15 mm mesh, and T3: caged 1 mm mesh) and were exposed to foraging in a randomised blocks arrangement on a PVC structure with top and frontal cameras. (C) Structure used for the remote video foraging system (RVFS) (top) and an image taken from the frontal camera during trials (bottom). For RVFS trials, one randomly selected plate from each treatment (T1, T2 and T3) was offered as bait in a randomised block arrangement. A total of four RVFS trials, of 4 h each, were carried out using a custom-built structure with two Olympus TG-tracker cameras, suspended at 2 m depth from the marina pontoons (see Figure 2B). Cameras were synced and set up in time-lapse mode with 2 s intervals (0.5 frames per second). After each trial, the structure was retrieved, the plates detached, and these were re-weighted and re-pho- tographed for analysis (see the conceptual design of the experiment in Figure 1C). Our RVFS trials were carried out in late September 2018 over a 2-day period, with one trial in the morning and one in the early afternoon. RVFS deployment was at the same location where water temperature (measured every 4 h using a HOBO temperature data logger) ranged between 23.5 °C and 24.0 °C during the trial period. 2.2.2. Data Acquisition and Treatment Images of the settlement plates before and after RVFS trials were labelled, colour lens corrected, perspective cropped and resized. Using CPCe V4.1 software (Coral Point Count with Excel extensions, Nova southeastern University, Florida, USA[50]), 99 points were randomly overlaid in a stratified fashion over a 3 × 3 cell grid (11 points per cell) over each image to quantify taxa relative abundance before and after exposure to predation during video trials. The same 99 hit-point grid was also used to assess which taxonomic categories had been removed from each plate. Point coordinates were maintained by using the same *.cpce file on the cropped and resized images of each plate (prior and after exposure). Figure 2. Pilot study experimental setup. ( A ) The study was conducted in Marina of Quinta do Lorde, Madeira Island, Portugal. ( B ) The fouling assemblages used as bait were obtained using caging treatments (T1: uncaged, T2: caged 15 mm mesh, and T3: caged 1 mm mesh) and were exposed to foraging in a randomised blocks arrangement on a PVC structure with top and frontal cameras. ( C ) Structure used for the remote video foraging system (RVFS) (top) and an image taken from the frontal camera during trials (bottom). 2.2.2. Data Acquisition and Treatment Images of the settlement plates before and after RVFS trials were labelled, colour lens corrected, perspective cropped and resized. Using CPCe V4.1 software (Coral Point Count with Excel extensions, Nova southeastern University, Florida, USA [ 50 ]), 99 points were randomly overlaid in a stratified fashion over a 3 × 3 cell grid (11 points per cell) over each image to quantify taxa relative abundance before and after exposure to predation during video trials. The same 99 hit-point grid was also used to assess which taxonomic categories had been removed from each plate. Point coordinates were maintained by using the same *.cpce file on the cropped and resized images of each plate (prior and after exposure). Image files of the same plate (prior and after exposure) were renamed to match during analysis on CPCe. This strategy allows the same hit points to be sampled before and after exposing the plates to predation in a repeated-measurement sampling strategy. The identified fouling species were classified by major taxa, morphology (encrusting, massive and arborescent) and status (native, cryptogenic and NIS). Species status was assigned based on the existing literature [44,51–67]. The video files obtained during RVFS were visually inspected frame by frame, and the following fish behaviours were annotated by scan sampling [ 68 ]: forager presence, forager interest and forager feeding. Forager presence was defined as the occurrence of a fish species within the frontal camera frame range (see Figure 2C). The image frames from the top and frontal cameras were used for recording interest and bites. Forager interest was defined as the percentage of present individuals displaying a curiosity action towards one of the three offered plates (e.g., swimming towards or stopping less than 5 cm from the plate and facing its surface). Forager feeding was defined as the number of bites over each offered plate. J. Mar. Sci. Eng. 2022,10, 611 6 of 15 2.2.3. Statistical Analyses For multivariate analyses, fouling species relative abundance data were square-root transformed and used to calculate a Bray–Curtis similarity matrix (Bray and Curtis, 1957). This matrix was used to test the orthogonal factors: treatment (fixed with three levels: T1 uncaged, T2 caged with 15 mm mesh and T3 caged with 1 mm mesh); and time (fixed with two levels: before and after predation) using permutational multivariate analysis of variance (PERMANOVA). The heterogeneity of multivariate dispersions was tested using permutational analysis of multivariate dispersions (PERMDISP). Univariate analyses were carried out on fouling assemblage biomass (wet weight), total biotic abundance, native species’ relative abundance and NIS relative abundance. These datasets were used to calculate Euclidean distance matrices and tested using the same design as multivariate analyses. Data on fish interest towards plates and number of bites (feeding) from RVFS trials were used to calculate Euclidean distance matrices. These were tested using PERMANOVA on a randomised block design; for details, see Anderson et al. [69] (p. 58), including on factors’ treatment (fixed, three levels) and trials (random, four levels). All non-parametric data analyses were carried out with 9999 permutations using PRIMER v7 with PERMANOVA+ add-on (PRIMER-E Ltd., Plymouth, UK) [69,70]. 3. Results The Mediterranean parrotfish Sparisoma cretense (Linnaeus, 1758) was almost the only species foraging on the settlement plates during the 16 h of remote video foraging system (RVFS) trials. This species was registered showing interest 1829 times and feeding on the settlement plates 421 times. The only exception to this occurred when a Mugilidae school approached showing interest and fed once from the settlement plates. Overall, fish foragers’ interest and feeding on plates were similar across RVFS trials (Table 1). Among treatments, foragers showed a higher interest in uncaged and caged 15 mm over caged 1 mm, although feeding was similar among treatments. Table 1. PERMANOVA results on foragers’ interest and feeding over settlement plates during remote video foraging system (RVFS) trials. df = degrees of freedom; MS = mean square sum; P(Perm) = p-values for permutation test; Tr = Treatment, 3 levels: uncaged (T1), caged 15 mm mesh (T2) and caged 1 mm mesh (T3); Trial = random: four levels. Foragers Interest Foragers Feeding Source df MS Pseudo-F P(Perm) MS Pseudo-F P(Perm) Treatment 2 826.48 70.8584 0.0172 * 541.58 2.8368 0.1381 Trials 3 155.17 10.4754 0.3144 408.75 2.141 0.1962 Residuals 6 105.17 190.92 Pair-wise (T1 = T2) > T3 * Statistically significant result. Regarding the fouling community on the settlement plates, twenty-seven taxa were recorded: six native species, eight cryptogenic species, nine NIS and three taxa unidentified to species level (see Table S2). Annelida and Bryozoa were the most abundant phyla, while the most abundant morphologies were massive and arborescent (Figure 3). Overall, the structure of fouling assemblages was unique for each treatment, which confirms that different choices of fouling were offered to the fish during RVFS trials (Figures 4and S2). The foraging activity of Sparisoma cretense reduced mean fouling biomass by more than 10% and its relative abundance by more than 40% (see Figure S2). Wet biomass among treatments was different before but equal after RVFS, while relative abundance was similar among treatments before and after RVFS (see Table 2). Similarly, the relative abundance of NIS and native species decreased after RVFS (Figure S1; Table S3). However, native species J. Mar. Sci. Eng. 2022,10, 611 7 of 15 were more abundant at the uncaged treatment (T1) before and equal to the other treatments after RVFS, while NIS abundances were similar among treatments before and after RVFS. J. Mar. Sci. Eng. 2022, 10, x FOR PEER REVIEW 7 of 16 * Statistically significant result. Regarding the fouling community on the settlement plates, twenty-seven taxa were recorded: six native species, eight cryptogenic species, nine NIS and three taxa unidentified to species level (see Table S2). Annelida and Bryozoa were the most abundant phyla, while the most abundant morphologies were massive and arborescent (Figure 3). Overall, the structure of fouling assemblages was unique for each treatment, which confirms that different choices of fouling were offered to the fish during RVFS trials (Figure 4; Figure S2). Figure 3. Mean relative abundance over settlement plates before and after remote video foraging system (RVFS) trials, biogeographic status and functional morphologies for the main phyla. Other includes taxonomic categories with a relative abundance lower than 5% (Porifera, Arthropoda and Macroalgae). Unknown includes taxa that could not be identified to the species level (uncertain biographic status). Massive: domed or flat-topped shape; Encrusting: encrusting plate-like shape; Arborescent: erect branching shape. The foraging activity of Sparisoma cretense reduced mean fouling biomass by more than 10% and its relative abundance by more than 40% (see Figure S2). Wet biomass among treatments was different before but equal after RVFS, while relative abundance was similar among treatments before and after RVFS (see Table 2). Similarly, the relative abundance of NIS and native species decreased after RVFS (Figure S1; Table S3). However, native species were more abundant at the uncaged treatment (T1) before and equal to the other treatments after RVFS, while NIS abundances were similar among treatments before and after RVFS. Table 2. PERMANOVA results for the relative abundance and wet biomass. Tr = Treatment, 3 levels: uncaged (T1), caged 15mm mesh (T2) and caged 1mm mesh (T3); Ti = Time, 2 levels: before and after remote video foraging system (RVFS) trials. df = degrees of freedom; MS = mean square sum; P(Perm) =p-values for permutation test. Relative Abundance Wet Biomass Source df MS Pseudo-F P(perm) MS Pseudo-F P(perm) Treatment 2 180.54 0.4924 0.6376 1.7689 7.4793 0.0059* Time 1 22083 60.23 0.0001* 23.764 100.48 0.0001* Tr × Ti 2 351.04 0.9575 0.419 1.9138 8.092 0.0032* Residuals 18 366.64 0.2365 Pair-wise Before > After Before: T2 > (T1 = T3) After: T1 = T2 = T3 Figure 3. Mean relative abundance over settlement plates before and after remote video foraging system (RVFS) trials, biogeographic status and functional morphologies for the main phyla. Other includes taxonomic categories with a relative abundance lower than 5% (Porifera, Arthropoda and Macroalgae). Unknown includes taxa that could not be identified to the species level (uncertain biographic status). Massive: domed or flat-topped shape; Encrusting: encrusting plate-like shape; Arborescent: erect branching shape. J. Mar. Sci. Eng. 2022, 10, x FOR PEER REVIEW 8 of 16 * Statistically significant result. Foraging activity modified the structure of the fouling assemblages, although without having a homogenising effect across treatments at the level of species composition (Figure 4, Table 3), major taxa or morphologies (Figure 3). However, in terms of multivariate dispersions, the heterogeneity of the assemblages increased after RVFS, particularly for the treatment caged with 15 mm mesh (Figure 4, Table 3). Overall, the variables with a higher contribution to the multivariate dissimilarities were: the cryptogenic bryozoan Cradoscrupocellaria bertholletii (Audouin, 1826), the NIS bryozoan Parasmittina alba Ramalho, Muricy and Taylor, 2011, and the native annelid Spirobranchus triqueter (Linnaeus, 1758). These species had 24.9%, 10.8% and 10.2% mean relative abundance before RVFS trials, respectively (see Table S2). Table 3. Results for the multivariate PERMANOVA and PERMDISP on fouling assemblage structure. Tr = Treatment, 3 levels: uncaged (T1), caged 15 mm (T2) and caged 1 mm (T3); Ti = Time, 2 levels: before and after. df = degrees of freedom; MS = mean square sum; P(perm) = p-values for permutation test; Tr = Treatment, 3 levels: uncaged (T1), caged 15mm mesh (T2) and caged 1mm mesh (T3); Ti = Time, 2 levels: before and after RVFS trials; res. df = residual degrees of freedom for the PERMDISP test. PERMANOVA PERMDISP Source df MS Pseudo-F P(perm) df res. df F P(perm) Treatment 2 4224.8 3.3431 0.0007* 2 21 1.555 0.4229 Time 1 7617.8 6.028 0.0001* 1 22 19.836 0.0002* Tr × Ti 2 1219.8 0.9652 0.4983 2 18 24.108 0.0001* Residuals 18 1263.7 18 Pair-wise Treatment: T1 ≠ T2 ≠ T3 Before: (T1 = T3) > T2 After: T1 = T2 = T3 * Statistically significant result. Figure 4. A principal coordinates ordination (PCO) for an exploratory tool of the structure of the assemblages (taxa relative abundance) to understand the components of multivariate variation. The PCO displays 49.1% of variation and, using the blue vectors, those variables with a mean relative abundance higher than 10% before remote video foraging system (RVFS) trials and Pearson correlation coefficient (Pr) higher than 0.7 with any of the ordination axes. Vector’s length indicates the Pr value, with the maximum value (Pr = 1) highlighted by blue circumference. T1 = uncaged, T2 = caged 15 mm mesh and T3 = caged 1 mm mesh. J. Mar. Sci. Eng. 2022,10, 611 8 of 15 Table 2. PERMANOVA results for the relative abundance and wet biomass. Tr = Treatment, 3 levels: uncaged (T1), caged 15 mm mesh (T2) and caged 1 mm mesh (T3); Ti = Time, 2 levels: before and after remote video foraging system (RVFS) trials. df = degrees of freedom; MS = mean square sum; P(Perm) = p-values for permutation test. Relative Abundance Wet Biomass Source df MS Pseudo-F P(Perm) MS Pseudo-F P(Perm) Treatment 2 180.54 0.4924 0.6376 1.7689 7.4793 0.0059 * Time 1 22083 60.23 0.0001 * 23.764 100.48 0.0001 * Tr ×Ti 2 351.04 0.9575 0.419 1.9138 8.092 0.0032 * Residuals 18 366.64 0.2365 Pair-wise Before > After Before: T2 > (T1 = T3) After: T1 = T2 = T3 * Statistically significant result. Foraging activity modified the structure of the fouling assemblages, although without having a homogenising effect across treatments at the level of species composition (Figure 4, Table 3), major taxa or morphologies (Figure 3). However, in terms of multivariate dispersions, the heterogeneity of the assemblages increased after RVFS, particularly for the treatment caged with 15 mm mesh (Figure 4, Table 3). Overall, the variables with a higher contribution to the multivariate dissimilarities were: the cryptogenic bryozoan Cradoscrupocellaria bertholletii (Audouin, 1826), the NIS bryozoan Parasmittina alba Ramalho, Muricy and Taylor, 2011, and the native annelid Spirobranchus triqueter (Linnaeus, 1758). These species had 24.9%, 10.8% and 10.2% mean relative abundance before RVFS trials, respectively (see Table S2). Table 3. Results for the multivariate PERMANOVA and PERMDISP on fouling assemblage structure. Tr = Treatment, 3 levels: uncaged (T1), caged 15 mm (T2) and caged 1 mm (T3); Ti = Time , 2 levels: before and after. df = degrees of freedom; MS = mean square sum; P(Perm) = p-values for permutation test; Tr = Treatment, 3 levels: uncaged (T1), caged 15 mm mesh (T2) and caged 1 mm mesh (T3); Ti = Time , 2 levels: before and after RVFS trials; res. df = residual degrees of freedom for the PERMDISP test. PERMANOVA PERMDISP Source df MS Pseudo-F P(Perm) df res. df F P(Perm) Treatment 2 4224.8 3.3431 0.0007 * 2 21 1.555 0.4229 Time 1 7617.8 6.028 0.0001 * 1 22 19.836 0.0002 * Tr ×Ti 2 1219.8 0.9652 0.4983 2 18 24.108 0.0001 * Residuals 18 1263.7 18 Pair-wise Treatment: T1 6=T2 6=T3 Before: (T1 = T3) > T2 After: T1 = T2 = T3 * Statistically significant result. 4. Discussion The integration of novel technological tools in ecological studies is paramount for enhancing current capability in collecting and processing qualitative and quantitative data needed to increase the understanding of marine systems [ 71 ]. This is the case when integrating underwater video technology and imagery analysis tools in trophic and behavioural studies. Regarding this matter, the use of a remote video foraging system (RVFS) in a new enhanced method developed for this study allowed us to record fish foragers’ feeding choices and gather knowledge on their effect on non-indigenous species (NIS) establishment and proliferation inside marinas [ 72 ]. Furthermore, the RVFS allowed us to study foragers’ preference between and within different fouling assemblage structures that can be manipulated and pre-selected. This could be relevant, as fish can visually identify and memorise different patterns in fouling and express interest towards a whole assemblage J. Mar. Sci. Eng. 2022,10, 611 9 of 15 structure (food patches) rather than a single species [ 73 ]. The data output of RVFS can be further enhanced by integrating other tools such as macro photography (i.e., high-resolution photomosaics), machine learning (e.g., automated species abundance/biomass estimation in plates, foraging behaviour and species identification in videos), environmental DNA (for plate species composition) and stable isotope analysis (e.g., to validate trophic links) (Figure 5). Additionally, the collection of samples of tissue and stomach content from identified foragers could also help to better understand their feeding behaviour and the effect of the inclusion of NIS in their diets [ 73 ]. Even though it was not the case during this pilot study, this may be of particular interest if NIS present in the study area have allelochemical defences (e.g., Caulerpa spp. Algae) which can promote adverse effects in native foragers’ health, fitness and behaviour, as well as the accumulation of toxic compounds in their tissues, which can ultimately pose risks to human health [ 74 – 76 ]. Regarding data workflow, RVFS outputs can be analysed in multiple ways. For example, cluster analysis and discriminant methods such as SIMPROF can assign fouling assemblages into different groups post-settlement, before their deployment as bait. Before and after foraging, univariate measures (e.g., total abundance or biomass, diversity index, behavioural metrics) can be analysed using repeated measures ANOVA or t-tests (or their non-parametric alternatives), while the multivariate species composition of fouling assemblages can follow a similar approach to what was used in this pilot study (Figure 5). J. Mar. Sci. Eng. 2022, 10, x FOR PEER REVIEW 10 of 16 S. cretense in the marina of Quinta do Lorde during RVFS trials matches this pattern and corroborates its importance in shaping fouling communities. However, further research is needed to better understand this species’ functional role in artificial coastal habitats and to what extent it affects fouling and sessile communities at a broader spatial scale (i.e., beyond the marina infrastructure). Figure 5. Summary of techniques compatible with remote video foraging system (RVFS), highlighting the pathway followed in this study. The section Tools includes additional tools that can be used to segregate the fouling assemblages used as bait before RVFS and for the visualisation and analysis of the final RVFS data output. Foraging fish can be attracted to novelty [73], and the PVC structure used for RVFS trials could have attracted foragers feeding over plates. Methodologically, this effect could be minimised by pre-deploying the RVFS structure before the beginning of the experiment, so it is not perceived as a novelty for the fish. However, the intense feeding registered over plates could have been influenced by the structure of the fouling assemblages. The settlement plates were hung facing downwards during recruitment and vertically during RVFS trials, which would increase illumination over plates and make the fouling noticeable to the foraging fish. Figure 5. Summary of techniques compatible with remote video foraging system (RVFS), highlighting the pathway followed in this study. The section Tools includes additional tools that can be used to segregate the fouling assemblages used as bait before RVFS and for the visualisation and analysis of the final RVFS data output.