The use of new multiparametric observatory platforms for the remote monitoring and exploration of deep-sea ecosystems at day-night and seasonal temporal scales

The use of new multiparametric observatory platforms for the remote monitoring and exploration of deep-sea ecosystems at day-night and seasonal temporal scales Carolina Doya Le Besnerais Supervisor : Dr. Jacopo Aguzzi Co-supervisor: Dr. Joan Baptista Company i Claret The use of new multiparametric observatory platforms for the remote monitoring and exploration of deep-sea ecosystems at day-night and seasonal temporal scales Carolina Doya Le Besnerais ADVERTIMENT La consulta d’aquesta tesi queda condicionada a l’acceptació de les següents condicions d'ús: La difusió d’aquesta tesi per mitjà del repositori institucional UPCommons (http://upcommons.upc.edu/tesis) i el repositori cooperatiu TDX (http://www.tdx.cat/) ha estat autoritzada pels titulars dels drets de propietat intel·lectual únicament per a usos privats emmarcats en activitats d’investigació i docència. No s’autoritza la seva reproducció amb finalitats de lucre ni la seva difusió i posada a disposició des d’un lloc aliè al servei UPCommons o TDX. No s’autoritza la presentació del seu contingut en una finestra o marc aliè a UPCommons (framing). Aquesta reserva de drets afecta tant al resum de presentació de la tesi com als seus continguts. En la utilització o cita de parts de la tesi és obligat indicar el nom de la persona autora. ADVERTENCIA La consulta de esta tesis queda condicionada a la aceptación de las siguientes condiciones de uso: La difusión de esta tesis por medio del repositorio institucional UPCommons (http://upcommons.upc.edu/tesis) y el repositorio cooperativo TDR (http://www.tdx.cat/?localeattribute=es) ha sido autorizada por los titulares de los derechos de propiedad intelectual únicamente para usos privados enmarcados en actividades de investigación y docencia. No se autoriza su reproducción con finalidades de lucro ni su difusión y puesta a disposición desde un sitio ajeno al servicio UPCommons No se autoriza la presentación de su contenido en una ventana o marco ajeno a UPCommons (framing). Esta reserva de derechos afecta tanto al resumen de presentación de la tesis como a sus contenidos. En la utilización o cita de partes de la tesis es obligado indicar el nombre de la persona autora. WARNING On having consulted this thesis you’re accepting the following use conditions: Spreading this thesis by the institutional repository UPCommons (http://upcommons.upc.edu/tesis) and the cooperative repository TDX (http://www.tdx.cat/?localeattribute=en) has been authorized by the titular of the intellectual property rights only for private uses placed in investigation and teaching activities. Reproduction with lucrative aims is not authorized neither its spreading nor availability from a site foreign to the UPCommons service. Introducing its content in a window or frame foreign to the UPCommons service is not authorized (framing). These rights affect to the presentation summary of the thesis as well as to its contents. In the using or citation of parts of the thesis it’s obliged to indicate the name of the author. The use of new multiparametric observatory platforms for the remote monitoring and exploration of deep-sea ecosystems at day-night and seasonal temporal scales Uso de la nuevas plataformas de observación submarinas para el monitoreo remoto y exploración de los ecosistemas del mar profundo a escalas temporales de día-noche y estacionales Carolina Doya Le Besnerais Tesis presentada para la obtención del título de Doctora por la Universitat Politècnica de Catalunya (UPC) Programa de Doctorado de Ciencias del Mar 2017 Director: Dr. Jacopo Aguzzi Co-director: Dr. Joan Baptista Company y Claret Dept. Recursos Marinos Renovables, Instituto de Ciencias del Mar (ICM) del Consejo Superior de Investigaciones Científicas (CSIC) “The use of new multiparametric observatory platforms for the remote monitoring and exploration of deep-sea ecosystems at day-night and tidal temporal scales” The author has been financed by a FPI pre-doctoral grant from April 2012 to April 2015 (BES-2010-042728). The research presented in this Ph.D. Thesis was carried out in the framework of the project RITFIM (CTM2010-16274). The graphic art in the cover is a reproduction of the original work by ©Autun Purser Barcelona, May 2017 i Abstract Traditional sampling technologies such as trawling but also novel ones as ROV surveying are oriented toward a high spatial coverage without repeating data collection at fixed seabed windows. The temporal repetition is often neglected so any reported difference in sampling among sites or studies may potentially be confounded with time-induced variations as a product of rhythmic population displacements within the continental margin seabed and water column 3D scenarios. Behaviour is an important life trait conditioning our perception of deep-sea biodiversity, being its rhythmic expression upon different diel (i.e. 24-h based day-night and tidal cycles) poorly known. In this context, technological step forward must be taken in order to observe community changes in deep-sea areas as a product of population behavioural patterns. Here, I studied how activity rhythms of benthic species within deep-sea communities modulate their composition, species abundances, richness, biodiversity and other life-history trait information in representative deep-sea environments through the use of multiparametric video-fixed cabled and non-cabled stations plus moving platforms. At the same time, I provided new methodological sampling hints on data collection protocols and analyses specifically tuned to the different characteristic of each observatory platform. I shed new light on the regulation that environmental cycles exert on animals’ rhythmic behavior, revealing that the main environmental rulers affecting deep-sea benthic communities are still day-night indirect or more direct tidal-oriented cycles which act on endobenthic, benthopelagic, and nektobenthic migrations. Keywords:behavioral rhythms, tidal cycles, day-night cycles, observational technologies, cabled observatories, time-lapse imaging, rovers, landers, day-night and tidal rhythms, richness, biodiversity, benthopelagic coupling, nektobenthic migrations. ii Resumen Las tecnologías de muestreo tradicionales, como la pesca de arrastre, pero también las nuevas, como los trabajos con ROV, están orientadas hacia una elevada cobertura espacial sin necesariamente repetir la recopilación de datos en ventanas fijas de los fondos marinos. Esa repetición temporal es a menudo obviada por lo que cualquier diferencia en el muestreo entre sitios o estudios puede quedar enmascarada por las variaciones a escala temporal como resultado de los desplazamientos rítmicos de la población dentro del margen continental marino y el volumen tridimensional de columna de agua. El comportamiento en los animales es un rasgo importante que condiciona nuestra percepción de la diversidad biológica de los fondos marinos, siendo su expresión rítmica diaria (es decir, ciclos día-noche y mareales) mal conocida. En este contexto, hay una necesidad de avance tecnológico y metodológico que permita la observación de las variaciones en las comunidades del mar profundo en las zonas de aguas profundas como producto de los patrones de comportamiento poblacional. Aquí estudié cómo los ritmos comportamentales de las especies bentónicas y bentopelágicas modulan composición, abundancia de especies, riqueza, diversidad biológica y otra información del ciclo biológico de dichas especies y comunidades dentro de ecosistemas representativos del mar profundo mediante el uso de sistemas de cableado submarino multiparamétrico, plataformas submarinas no cableadas y plataformas submarinas móviles. Al mismo tiempo, aporté nuevas recomendaciones metodológicas sobre protocolos de recolección y análisis de datos específicamente ajustadas a las diferentes particularidades y necesidades de las susodichas tecnologías de observación submarina. Los resultados de este trabajo han ayudado a comprender mejor la regulación que los ciclos ambientales ejercen sobre el comportamiento rítmico de los animales, revelando que los principales factores ambientales que gobiernan las comunidades bentónicas de aguas profundas siguen siendo el ciclo día-noche de forma indirecta y el mareal de forma más directa que actúan modulando los desplazamientos endobentónicos, bentopelágicos y migraciones nektobentónicas. Palabras clave: ritmos de comportamiento, ciclos mareales, ciclos día-noche, tecnologías de observación submarina, observatorios cableados, imágenes en tiempo transcurrido, rovers, landers, ritmos diurnos y de marea, riqueza, biodiversidad, acoplamiento bentopelágico, migraciones nektobénicas. iii Preface “ Les bateaux, les engins, l’argent, les hommes, moi-même, tout ça c’est de la quincaillerie. Ce qui compte, c’est l’œuvre accomplie: dans un siècle on nous aura oublié, mais on se souviendra encore de ce que nous avons filmé et dit” Jacques-Yves Cousteau Als meus pares i germans... A mis padres y hermanos... À mes parents et frères... To my parents and brothers... iv v Table of contents Abstract .......................................................................................................................................................... i Resumen ........................................................................................................................................................ ii Preface ......................................................................................................................................................... iii Table of contents ........................................................................................................................................... v List of Figures ............................................................................................................................................ viii List of Tables ............................................................................................................................................. xiv List of Appendix ......................................................................................................................................... xv Agraïments/Agradecimientos/Remerciements/Acknowledgements .......................................................... xix Introduction. State of the art, hypotheses and objectives .............................................................................. 1 1. Ocean services, major pressures, and motivation for Seafloor Observatories ................................. 3 2. Temporal bottleneck in marine sampling strategies ......................................................................... 4 3. Activity rhythms in the three-dimensional marine scenario ............................................................ 5 4. New seafloor monitoring technologies ............................................................................................ 6 5. Imaging as strategic monitoring approach linking animal behavior to ecosystem functioning ....... 7 6. Current cabled observatory network initiatives worldwide ............................................................. 8 7. Limitations to the ecological monitoring by cabled observatories imaging in the deep sea and envisaged solutions ........................................................................................................................ 10 8. Thesis hypotheses and objectives................................................................................................... 12 Chapter 1. Diel behavioral rhythms in sablefish (Anoplopoma fimbria) and other benthic species, as recorded by the deep-sea cabled observatories in Barkley canyon (NEPTUNE-Canada) .......................... 15 1.1 Abstract ....................................................................................................................................... 17 1.2 Introduction ................................................................................................................................. 18 1.3 Material and methods .................................................................................................................. 19 1.3.1 Cabled network specifications ............................................................................................ 19 1.3.2 Data acquisition and time series processing........................................................................ 20 1.3.3 Size-frequency distribution ................................................................................................. 21 1.3.4 Light effects ........................................................................................................................ 22 1.4 Results ......................................................................................................................................... 23 1.4.1 Faunal remarks .................................................................................................................... 23 1.4.2 Time series analysis ............................................................................................................ 24 1.4.3 Class-size frequency distribution ........................................................................................ 31 1.4.4 Light effects ........................................................................................................................ 31 1.5 Discussion ................................................................................................................................... 33 1.5.1 Sablefish behavioral rhythms .............................................................................................. 33 1.5.2 The hagfish and crab behavior ............................................................................................ 36 1.5.3 The population structure of sablefish .................................................................................. 36 1.5.4 Biases in cabled observatory video recording ..................................................................... 37 1.6 Conclusions ................................................................................................................................. 38 1.7 Acknowledgments ....................................................................................................................... 39 Chapter 2. The seasonal use of small-scale space by benthic species in a transiently hypoxic area .... 41 2.1 Abstract ....................................................................................................................................... 43 2.2 Introduction ................................................................................................................................. 44 2.3 Materials and methods ................................................................................................................. 45 2.3.1 The platform, the study area, and data acquisition .............................................................. 45 2.3.2 Data treatment ..................................................................................................................... 47 xii Fig.23. Biodiversity, richness and megafaunal species visual counts (i.e. from 14th February 2013 to 14th April 2014) time series, as reported by crawler video-imaging. Biodiversity have been reported as mean per month Shannon Index (C) and Simpson Index (A). Similarly, we reported the species richness (D). Abundance (B) was normalized for the maximum transects of the study period (i.e. 48). Horizontal dashed lines correspond to the Midline Estimated Statistic of Rhythm (MESOR). .......... 83 Fig.24. Map showing the deployment site of the juvenile sperm whale SW off Sagami Bay (empty triangle) at 500 m depth and in relation to the seafloor cabled video observatory (filled triangle) at 1100 m depth (Aguzzi et al. 2010). The deployment site was about. B. 3D laser scan image of the deployment site shows the whale carcass relative to the ballast used to sink it, the camera and ADCP instrumented landers (courtesy of B. Thorton and A. Bodemann). ............................................................ 97 Fig. 25. Images of the taxonomical units as components of the local megafaunal assemblage detected at whale carcass positioning by time-lapse camera imaging (see also Appendix G): A. Pannychia moseleyi; B. Echinothurioida; C. Solaster paxillatus; D. Ophiurida; E. Buccinum yoroianum; F. Bathynomus doederleini; G. Macrocheira kaempferi; H. Hexanchus griseus; I. Cephaloscyllium umbratile; L. Coelorinchus sp.; M. Pterothrissus gissu; N. Zoarcidae; O. Physiculus japonicus; P. Simenchelys parasiticus; Q. Helicolenus hilgendorfii; and finally, R. Eptatretus deani. ........................ 101 Fig.26. Temporal changes in the whale carcass positioning with respect to the camera field of view as detected by time-lapse imaging. A. 06.15.2012 to 07.15.2012; B. 07.15.2012 to 07.26.2012; C. 07.26.2012 to 08.07.2012; and finally, D. 08.07.2102 to 08.24.2012. In A, two individuals of Solaster paxillatus, and single individuals of Echinothurioidea, Ophiuridea, and Buccinum yoroianum are visible. In C and D, two ribs appear on the seafloor.............................................................................................. 102 Fig.27. Temperature at sea bottom, water pressure, vertical and East-West current component as recorded over 72 days by ADCP close to the whale carcass deployment.Whale carcass position changes (see Figure 25) are indicated by vertical light grey dotted lines. .............................................................. 104 Fig.28. Waveform analysis outputs for the visual count time series of those species showing significant 24 hrs. periodicity patterns in the periodogram analysis (see Appendix G for A. S. parasiticus, B. M. kaempferi, and C. P.gissu. Note the varying Y-axis abundance scales. Up and down arrows indicate onset and offset of significant increments in abundance peaks, respectively. They also indicate the first and last calculated mean abundance values above the Midline Estimated Statistic Of Rhythm (MESOR as dashed horizontal lines) line as re-average of all waveform data. MESORs are: A, 4.76; B, 1.18; C, 0.46. The correspondent integrated diurnal displacements (m) along the E-W axis, computed from the FFT results, are superimposed, showing a 2 hrs. delay in the animal periodicity. ........................................... 106 Fig.29. Cluster analysis of the mean abundance values by day of species recorded at the whale fall for factor “week”. Black lines indicate significant separation at p<0.05 according to SIMPROF test; xiii B. Bubble nMDS plots with over imposed cluster analysis results (at 60% of resemblance level) of two of the dominant taxa of each phase (weeks 1-6 vs. 7-11): Macrocheira kaempferi and ophiuroids. ............ 108 Fig.30. Scatter plot of the first two axes of the CANOCO. Species and trophic categories (i.e. scavengers, predators, and detritivores) were represented by black points and environmental variables were represented by red vectors. Environmental variables considered are: Date of the observation as “Date”, spatial components of current velocity as “EW_Comp_V” and “Vert_Comp_V”, as well as Depth, Temperature and the time shifted. B, Bar chart of the loadings relative to the first latent vector (explained variance = 77.44%) for both environmental and species blocks (no. 1 and 2, respectively). . 110 xiv List of Tables Table 1Complete list of faunal observations (number plus relative percentages) as reported by NEPTUNE-Canada color video cameras installed in Barkley canyon, including the targeted species.That list is complete only for platforms POD 1 and POD 3 on the canyon flank. Such a list is incomplete for platform POD 4, given the lower resolution of its video camera, which in most cases prevented the identification of smaller animals ................................................................................................................. 24 Table 2.Interspecific dominance relationships, presented as the percentage of significantly more (+) or less (−) occupied parcels over the total (i.e., 81), as observed over a month. All significant (*) comparisons (p b 0.0001) are indicated ...................................................................................................... 56 Table 3: Kruskal-Wallis test among the 14 months of the study for the 8 most abundant species with 13 degrees of freedom ......................................................................................................................... 80 Table 4Results of the SIMPER analysis carried out on the two periods (weeks 1-6 vs. 7-11) evidenced by multivariate analysis ........................................................................................................... 109 xv List of Appendix Appendix A.Annual Pearson’s goodness of fit test for each of the three groups under the null hypothesis of equal probability of parcel occupation.Only significantly more occupied parcels (critical value (p =0.001) = sqrt (X20.001,80/81) = 1.24) are shown. ................................................................... 161 Appendix B.Monthly heat maps calculations for the three groups of benthic animals. See Figure 6 for specifications on grey scale (i.e. occupation) and onred and yellow outlining (occupation significance). ............................................................................................................................................. 162 Appendix C.Monthly Segregation Index (S) values between pairs of groups (F, slender sole; L - S, large and small squat lobsters, respectively). Missing values occurred when one of the species was absent. In bold are reported intraspecific S values representing the degree of aggregation between conspecifics, whereas normal-text S values reports the degree of segregation between different groups 165 Appendix D.Monthly total visual counts and percentages for the different megafaunal species studied during the 14 months of video acquisition. .................................................................................. 166 Appendix E.Date and time of the reported ethological remarks. (A) Rockfish (Sebatidae) agonistic display against the crawler (i.e. approaching the camera with the open mouth and then escaped). (B) Sablefish (Anoplopoma fimbria) swimming close to the crawler. (C) Male scarlet king crab (Lithodes couesi) feeding behavior and agonistic interaction with a grooved tanner crab. (D) Grooved tanner crab (Chionnecetes tanneri) agonistic display against the (i.e. an elevated body posture and chelipeds forward projection). (E) Grooved tanner crab reproduction behaviour. (F) Female grooved tanner crabs carrying eggs. .......................................................................................................................................................... 167 Appendix F.Photo-mosaic of all species portrayed with the camera installed on the crawler during the 14 months of video acquisition.Individuals in images are: (A) Eelpout (Licenchelys spp.). (B) Dover sole (Microstomus pacificus). (C) Deep sea sole (Embassichthys bathybius). (D) Pacific halibut (Hippoglossus stenolepis). (E) Rattail (Coryphaenoides spp.). (F) Rockfish (Sebastidae). (G) Rockfish (Sebastidae). (H) Rockfish (Sebastidae). (I) Blackfin poacher (Bathyagonus nigripinnis). (J) Sablefish (Anoplopoma fimbria). (K) Hagfish. (L) Dumbo octopus (Grimpoteuthis spp.). (M) Squid. (N) Buccinids (Neptunidae). (O) Brittle star (Ophiuroidea). (P) Starfish (Asteroidea). (Q) Holoturian. (R) Hermit crab. (S) Scarlet king crab. (T) Grooved tanner crab (Chionoecetes tanneri). (U) Male (left) and female (right) of grooved tanner crab facing each other as a part of their reproduction behaviour. (V) Male (left) carrying out a female (right) of grooved tanner crab as a part of their reproduction behaviour. (W) Small crabs, probably small individuals of grooved tanner crab. (X) Ctenophore (Bolinopsis infundibulum). (Y) Scyphomedusa (Poralia rufescens). (Z) Dinner plate jelly (Solmissus spp.). (AA) Traquimedusa (Voragonema pedunculata). ..................................................................................................................... 168 xvi Appendix G. Taxonomical assemblage detected over 72 days starting on 14 June 2012 are hierarchically listed by phylogeny and grouped according to behavioral traits (type of movement and life habit as relationship with the seabed), along with bibliographic sources used to derive such an information. The number of individuals per species (N), their relative abundances (%), periodogram analysis outputs as significant (p < 0.05) periodicities(P) in minutes (min) and hours (hrs.) are reported along with periodogram peak variance (Var., %) as proxy of rhythm strength. In the periodicity column, we also reported in parenthesis significant sub-periodicities as a proxy of weak tidal patterning. Total animal abundance by functional groups with respect to the 3 trophic groups is alsoprovided. ................ 169 Appendix H.Seventy-two days-time series of visual counts for the different macrofauna taxonomical units identified during the scavenging process of the sunken whale carcass (following the same taxonomical order of Figure 2). Vertical dashed grey lines are the whale change positioning (see Figure 3). ................................................................................................................................................... 171 Appendix I.Time series of horizontal current data from the ADCP, at 5 m, 32 m, 62 m and 92 m above sea bottom, applying a 25 hrs.moving average window. Red and blue lines represent North-South and East-West component respectively. Black lines show the fluctuations of current velocity. .............. 172 Appendix J.Stick diagrams of horizontal current data from the ADCP, at 5 m, 32 m, 62 m and 92 m above sea bottom, applying a 25 hrs. moving average window. Southeastward flows are dominant in the first 30 m above the sea bottom, whereas northwestward flow characterizes the water column from 32 m to 92 m above the sea bottom, creating a two-layer system. ................................................................ 173 Appendix K.Time series of vertical current data from the ADCP, at 5 m, 32 m, 62 m and 92 m above sea bottom, applying a 25 hrs.moving average window. Dominant negative values indicate prominent downward flow in each layer, especially close to the bottom. ................................................ 174 Appendix L.Power Spectra of water temperature, East-West, North-South and Vertical components of the velocity current data from the ADCP, at 5 m above sea bottom.Spectra of current data clearly show peak periods for 24, 12, and 6 hrs., whereas peaks for temperature data are only evident at 12 and 6 hrs. The flow near the sea bottom is affected by tidal periodicity. This rhythm may affect the behavior of marine organism .................................................................................................................... 175 Appendix M.Waveform analysis outputs for ADCP data. Time series in A. have been partitioned in 12 hrs.segments, while time series in B. are at 24 hrs.length to show both diurnal and semidiurnal tides. The MESOR is the dashed horizontal line. The semidiurnal and diurnal Fourier component curves are superimposed for vertical and East-West current components (the dash-dotted lines), showing general accordance with the waveform analysis. ............................................................ 176 Appendix N.Waveform analysis outputs for visual counts time series of A. scavengers (and facultative predators),B. detritivores, and C. predators, indicating the occurrence of significant time xvii shifted peaks (i.e. values above the MESOR; as horizontal dashed line) in the former (i.e. nocturnal) and the latter group (i.e. diurnal). Conversely, detritivores do not show any defined tendency (although a weak peaking occurs in the morning). MESORs are: A. 6.48; B. 0.99; C. 097. ................................................ 177 xviii xix Agraïments/Agradecimientos/Remerciements/Acknowledgements Recordaré siempre la primera vez que vi un ctenóforo en mis vídeos. Sus cepillos iridiscentes ondulando al unísono y su baile etéreo recordando a una bailarina de vals en sus mejores galas mientras se adentraba en lo impresivo de la oscuridad absoluta. Ahí entendí que esto iba a ser una aventura y que me atraparía irremediablemente por su enigmática y bella naturaleza. En esta aventura me han acompañado muchas personas a las que quisiera mencionar. Seguramente olvide alguna, espero no hacerlo. Primeramente, quisiera agradecer a mi director Dr. Jacopo Aguzzi y mi co-director Dr. Joan Baptista Company (Batis), la oportunidad que me disteis de entrar en el fascinante mundo de la ecología marina y de las nuevas tecnologías de observación submarina. Vuestra ayuda ha sido inestimable. A Jacopo, los conocimientos que me has transmitido, a nivel científico y existencial. Gracias por tu paciencia, sentido del humor, energía y disponibilidad. A Batis, gràcies pel teu pragmatisme, empatia i per estirar del braç en els moments difícils. També voldria agrair a en Dr.Domènech Lloris, el meu primer contacte a l’ICM, el que sempre estigués disponible per orientar-me des de aquell dia que vaig demanar-li assessorament per meu treball de final de carrera. I would like to thank all the researchers that I met during my short stays in Canada and Japan. My first contact and supervisor during my stay in Department of Biology of the University of Victoria, BC; Canada, Prof. Verena Tunnicliffe, not only for the astonishing knowledge she shared with me but the way she does it by pointing the finger to the correct doors without opening them. I would like to thank you to be an example of woman in science because of your personality and values. Thanks to the Biology Department team to make feel like home in Victoria. In this context, I would like to specially mention Dr. Jackson Chu and Amalis. Thank you Prof. Kim Juniper toopen the doors of the Ocean Networks Canada (ONC) during my stay in Victoria and to invite me to the expedition Wiring the Abyss in which I could understand for real the tools and habitats I was working on. Also thank your advice about how to cope with my wallyland study. The friends I make during the cruise, especially to Maëva and Reyna. That was a gift! I am also enormously grateful to my supervisor during my stay in Japan, Dr. Yoshihiro Fujiwara for his helpfulness during my stay in Japan. Many thanks for your kind assistance in all the aspects, for willing to listen and advice and for all the received knowledge about whale-fall ecosystems. Thank you to give me the opportunity to ship and collaborate in your deep-sea shark project. I keep in mind this moment as a treasure as the bird shooting in Aveiro. Thanks to all the BioGeoscience team of the Japan Agency for Marine Sciences (JAMSTEC) to share great after-work moments and guide me in the amazing Japanese culture and country. I also would like to especially thank you to my dear friends Mitsuko, Kumiko and Yoshie. Me gustaría dar las gracias a todos los investigadores y estudiantes con los que he tenido la suerte de colaborar. Gracias a M.S. María Pardo por tantas horas invertidas contando bichitos y el trabajo xx bien hecho. Gracias al Damianos, por darme la oportunidad de ser su codirectora de tesis de máster, por su sentido del humor y por darme la alegría de un buen resultado. ¡Gracias por las horas invertidas en este proyecto y por los paseos para airear las ideas que siempre terminaban en un “Eureka!”. Al Dr. Nixon Bahamon, por toda la ayuda que me ha brindado con sus amplios conocimientos en estadística que me ayudaron a domesticar el deep sea. Sorprende la modestia, se agradece la amabilidad y disponibilidad para las largas charlas al teléfono. Thank you Dr. Corrado Costa for your help in statistics too. A special thank you to Prof. Laurenz Thomsen to give to me the opportunity to drive your crawler “Wally I and have faith in our project”. It has been one of the best experiences of my Ph.D. Thank you Autun Purser to teach me how to drive the crawler and for your helpfulness from the sampling to the publishing process and of course for the amazing cover that you draw for this thesis. Thanks to my co-authors Fabio De Leo, Marjolaine Matabos, Steve Mihaly, Miquel Canals, Emanuela Faneli, Tiziana Ciuffardi, AntonioSchirone, MasaruKawato, Makoto Miyazaki and Furushima Yasuo for your valuable help. Agradecer también al resto de personas del Instituto de Ciencias del Mar que han contribuido a que en mi estancia en el Departamento me haya sentido tan a gusto: la Isabel i el seu somriure, sempre disposada a un consell o un cop de mà i el Miquel Àngel que m’ha tret les castanyes del foc quan l’ordenador ha fet de les seves. La Nuria Angosto i la Conchita Borruel por su disponibilidad y profesionalidad. No puedo dejar de mencionar a mis compañeros de pasillo y departamento (becarios, técnicos, personal laboral, estudiantes de máster y grado), por todos los momentos vividos a nivel profesional (presentaciones, consejos, cables) como en nuestros fuera del instituto (cervezas, almuerzos, cenas, calçotades, escapadas):Aihnoa, Alba, Alejandro, Alfredo, Amalia, Anabel, Ariadna, Catarina, Chiara, Claudio, Cristina, Dafni, Dani, Daniela, David, Federico, Fernando, Iván, John, José Antonio, Laia, Marta, Morane, Noelia, Paula, Sonia, Susana, Tabita, Ulla, Valeria, Vanesa, Xavi…hicisteis que esta estancia fuera para recordar y me llevo buenas amistades y recuerdos increíbles! Quisiera mencionar especialmente a mis compañeros de despacho Giulia, no sólo compañera de despacho, también de maullidos gatunos y de aventuras varias y a Valerio por ser ejemplo a seguir en tantos aspectos. También a los que los han sustituido, Marta, Marina y Marc por acompañarme último período con su buen humor. A los científicos deportistas con los que hemos compartido grandes momentos: Raquel, Marta, María, Angel, Pau, Raül, y Teresa. Quisiera también agradecer a mi tutor de la UPC, Dr. Vicente Gracia, el haber aceptado tutelar mi tesis, y al coordinador del programa de doctorado de Ciencias del Mar, Dr. Agustín Sánchez-Arcilla y a Genoveva Comas, por su disponibilidad y por toda la ayuda proporcionada Ya fuera del ambiente laboral, quisiera agradecer a mis amigos. Gracias Walter por aguantarme con paciencia de santo. A mi compañero de piso y amigo John, Ava y Rachel por los momentos de relax de copa de tinto en terraza, las escapadas, salidas, vuestro sentido del humor y el compañerismo. Gracias por hacer que la casa fuera un hogar. A totes les amistats universitàries que s’han mantingut tants anys, pels plans junts i els bons consells, mercès Albert Fernández, Albert Rivas, Alejandro, Cidalina i Joan. xxi Als meus amics de tota la vida, mercès pel vostre exemple i pels moments de tant necessària desconnexió: Felip, Jess, Marta i Nuria. Ester, mercès tot això i pel cop de mà amb el formatat de la tesis. I finalment, als meus pares, Josep Lluís i Martine, pel vostre recolzament incondicional, pels valors que m’heu après i per creure en mi. Gracies per mostrar-vos sempre orgullosos de la vostra filla. Merci pour vos conseils et encore merci pour m’aider à faire réalité mon rêve. Al meu germà, Pau, per fer màgia amb la teva agenda i ser la meva línea telefónica resuelve-marrones 24/24h. Mercès per aquestes sorpreses que em fas quan no m’ho espero. Merci à ma sœur Eva, pour nos rires, pour être un exemple de sœur, femme et amie, agencia de viajes Salazar i coach. I també a la Ruth i al Marc, dos amores de mis amores. Sense vosaltres això no hagués estat possible, per això aquesta tesis també és molt vostra. Mencionar als Doya et aux Le Besnerais pour être toujours là bien sûr. A todos y todas: ¡Muchas gracias! Moltes Mercès! Thank you! Merci! Introduction 6 (i.e. twilight) zones (Daan and Aschoff 2001,Sbragaglia et al. 2015). But as sunlight disappears (i.e. below the twilight zone end) with last photons being detected as deep as 1 km depending from local oligotrophic conditions (Herring 2002), other geophysical cycles are expected to govern these clocks (Aguzzi et al. 2010). Tides are known to be important synchronizers in certain marine species and depths (Sbragaglia et al. 2015). In spite of that, the transmission of energy from shallower winds and tides in a complex manner to due to the effect of frictional dragging with seafloor geomorphologies (Garrett 2003,Babanin 2009)in the deep sea give origin to inertial waves (with or without tidal frequency). Complex hydrodynamics may be an important synchronizer in the deep sea (Aguzzi 2009)but the knowledge of their effects on animals is still insufficient. Populations react to those deterministic changes by performing rhythmic displacements in a tridimensional space. Diel vertical migrations (DVM) represent the biggest animal migration on Earth in terms of biomass (Hays 2003). Animals travel up from deeper zones into worldwide richer shallower waters. When these bathymetric migrations take place along the seabed those are known as nektobenthic displacements. Lastly, endobenthic movements consist in periodic concealment and emergence from the sediment (Aguzzi et al. 2010). Thus, marine communities present rhythmic composition changes that imply a modification in intra and interspecific relationships, which ends up into spatio (depth) -temporal (tidal, day-night and seasonal) modifications in ecosystem functioning, redistribution of nutrients and organic matter in the ocean. 4. New seafloor monitoring technologies There is the need of new monitoring technologies adding a solid temporal dimension to marine discovery and monitoring. Most of nowadays multiparametric platforms used for marine exploration bear video cameras as items that can approach the study of ecosystems at the complexity level of fauna (Board 2000,Favali and Beranzoli 2006, Matabos et al. 2017). Platforms such as ROVs, AUVs and landers lack of unlimited power autonomy to sustain a broad set of sensors, large imaging and environmental data storage capacity, as well as real-time inspection possibility, all features needed to monitor ecosystems at 360 degrees over long-term scales encompassing months, seasons, and consecutive years. During the last 30 years, we have witnessed the appearance of cabled seafloor observatories all around the world, as platforms able to overcome the limitations of other ship-assisted sampling methods. A seafloor observatory is defined as a multiparametric platform at a fixed site, feed of power and supply command, control and communications to sensors located above, on or below the seabed itself, as well as spreading with a cluster of other interconnected stations into a geographic network (NRC 2000,Favali and Beranzoli 2006). The observatory node may also support mobile vehicles, such as Rovers as benthic mobile robots (Thomsen et al. 2015). Among these, there are the crawlers which represent one of the newest Internet Operated Vehicles [IOV] which are tethered to a cabled node, allowing higher spatial coverage around it (Purser 2015,Thomsen et al. 2015). Introduction 7 Cabled observatories are vocationally interdisciplinary, integrating biological, geological, and oceanographic sensors (Favali and Beranzoli 2006; Lampitt et al. 2010), all specifications that make of them potentially important tools for the study of behavioral rhythms in association of concomitant environmental changes(Aguzzi et al. 2012). Their monitoring capacity is sustained by continuous data flow and sufficient power to operate environmental sensors at a high frequency (seconds) over prolonged durations (from days to years) in a remote and real-time fashion. This interdisciplinary monitoring approach would then allow to efficiency scale marine community composition changes at different temporal scales down to the behavioral response of individuals to cyclic and stochastic changes in their eco-field (Danovaro et al. 2017). 5. Imaging as strategic monitoring approach linking animal behavior to ecosystem functioning Observatory imaging capacities can be used to highlight the relationships between rhythmic behavior of species and community composition spatiotemporal changes. The chance of observing an individual of a species within the framework of a fixed camera fluctuates in response to changes in its rate of displacement activity (which include several behavioral motivations such as for example, feeding, mating, territorial control and homing). In time-lapse photography and footage acquisition, fluctuations in visual counts are proxies of the animals’ behavioral rhythms(Costa et al. 2009,Chabanet et al. 2012; Doya et al. 2014). Accordingly, there is a direct correlation between visual counts at a fixed camera field of view and changes in local population abundance due to rhythmic water column seabed displacements (Aguzzi et al. 2010). Thus visual counts monitoring represents a straight ahead method detecting rhythmic behaviours although not the only one. Virtually, any visually detectable rhythmic behavior can be monitored with time-lapse photography any footage acquisition. When cameras are added to other environmental monitoring sensors, one can use an observatory, not only to properly monitor the response of species and communities to environmental changes (Aguzzi et al. 2012,Condal et al. 2012,Matabos 2012), but also to monitor other interesting behavior such as intra and interspecific interactions. Imaging is therefore a powerful tool of abiotic factors and global change research, and the procedures for best image data analyses will likely be a trending line within future developments of cabled observatory technology(Vardaro et al. 2013). Time-lapse image capture should ideally be designed at temporal frequencies that maximize the detection of the entire range of species behavior we are interested in. Particularly, such methodology allows discerning between short-term variations in behavior (and local abundance) due to diel and seasonal rhythms from other more long-lasting trend of change for the progressive adjustment of biotic communities to a shifting climate. When data are collected through enough consecutive years, animals’ Introduction 8 responses to ongoing local and global changes can be disentangled from infrannual cycles (Glover et al. 2010). 6. Current cabled observatory network initiatives worldwide A network of cabled video-observatories dispersing across different marine areas of the world brings new possibilities to track large-scale changes in the marine environment. Many large-scale projects have implemented permanent seafloor networks at International level. Canada, Europe, Japan, and USA are the major actors (Fig. I). Those initiatives increase the number of worldwide connected camera nodes, providing progressively higher geographic and bathymetric coverage. Improved access to study real-time or near-real time data (open data policy) of a wide variety of sensors will provide synergy to the researchers’ efforts anywhere in the word to understand ocean functioning promoting the interdisciplinary collaborations worldwide. Concomitantly, a wide range of opportunities to public access and participation to the ongoing science and raw data would increase education and public environmental awareness (Matabos et al. 2017). Figure I. Seafloor video-observatory networks that has been or are being implemented at different depths of the continental margin, form coastal areas to the deep sea, having the chance to be used to monitor different areas experiencing anthropic impacts (from Danovaro et al. 2017). Introduction 9 The main efforts in the implementation of ocean observatories are led by the ONC (http://www.oceannetworks.ca), an initiative of the Victoria University (British Columbia). The major components are NEPTUNE (North East Pacific Time-series Underwater Networked Experiments, (http://www.neptunecanada.ca) and VENUS (Victoria Experimental Network Under the Sea, (http://www.oceannetworks.ca/installations/observatories/venus-salish-sea). VENUS is located in a seasonally hypoxic fjord in Vancouver Island and was the first of his kind in the country. NEPTUNE is a plate-scale multidisciplinary video-observatory network on the Juan de Fuca Plate, located off the west coast Vancouver Island. NEPTUNE is the largest of his kind worldwide and aims to solve a wide span of research questions, such as earthquake dynamics and tsunami hazards, fluid fluxes in the ocean crust and sediments, gas hydrate deposits; ocean and climate dynamics, acidification, nutrient fluxes and impacts on biota; deep-sea ecosystems dynamics; and applied engineering and computer science research (Barnes et al. 2013; Favali 2015). VENUS houses ocean and biological and delta dynamics studies in waters to 300-m depth (Barnes et al. 2013). OOI US initiative sponsored by the National Science Foundation (NSF) is also an ambitious monitoring program already encompassing or aiming to install cabled observatories from the high latitudes of the North Pacific Station Papa, the North Atlantic component into the Irminger Sea, the Southern Chile and the Argentine Basin Array through the Pioneer located off the coast of New England, south of Martha’s Vineyard and the Endurance Array into the coastal upwelling region of the Oregon and Washington coasts (http://oceanobservatories.org/). Out of the umbrella of the OOI, the Monterey Accelerated Research System (MARS; http://www.mbari.org/at-sea/cabled-observatory/), and the University of Hawaii ALOHA Cabled Observatory(ACO; http://aco-ssds.soest.hawaii.edu/) are two cabled observatories also sponsored by the NSF. Japan its cabled seafloor observatories are not yet fully devoted to the ecological monitoring since chiefly used as seismic and tsunami warning system (Aguzzi et al. 2012). At present, the DONET (Dense Ocean floor Network system for Earthquakes and Tsunamis; http://jamstec.go.jp/donet), managed by JAMSTEC (Japan Agency for Marine-Earth Science and TEChnology) is chiefly represented by observing points in the Nankai trough seismogenic zone, and includes several seismometers and pressure gauges in its sensors. DONET has anyway a cabled video-station, in Sagami bay (1100 m depth) to be used for studies on reproductive behavior of Calyptogena clam fields in hydrocarbon seepages(Fujiwara et al. 1998). The sole other but shallower cabled video-platform is the reef observing OCEANCUBE held by theOkinawa Institute of Science and Technology (OIST) in Okinawa at 20 m depth. In Europe, the European Multidisciplinary Seafloor and water-column Observatory (EMSO, http://www.emso-eu.org/;Best et al. 2014)is a transnational infrastructure aiming to long-term monitoring the interaction between the geosphere, biosphere, and hydrosphere. It is composed of several stations deployed in geologically or ecologically relevant shallow and deep-sea areas, from the Arctic to the Atlantic, through the Mediterranean, to the Black Sea. Introduction 10 The Center for Ocean, Seafloor, and Marine Observing Systems (COSMOS) located in the Norwegian margin is the northern EMSO video-observatory. Near this observatory there is the LOVE deployed by Statoil ASA and the Institute of Marine Research (IMR) bearing a still camera. Of particular relevance are the sole shallow waters two platforms bearing cameras: the OBSEA (http://www.obsea.es/www.obsea.es) and the SMARTBAY subsea real-time observatory (http://www.smartbay.ie), both deployed at 20 m depth. Other, much deeper platforms bear special imaging equipment that could be indirectly used for the study of behavioral rhythms and communities’ turnover, but whose primary vocation is related to particle physics. These are Neutrino telescopes endowed with photomultiplier tubes capable of detecting not only the passage of quantum sub-atomic particles but also bioluminescence. The ANTARES (Astronomy with a Neutrino Telescope and Abyss environmental RESearch; http://antares.in2p3.fr/) at 2300 m depth off Marseille, the NEutrino Mediterranean Observatory (NEMO), off Capopassero (Catania) at 2200 m depth as well as the Neutrino Extended Submarine Telescope with Oceanographic Research (NESTOR) to be installed off Pylos (Greece), at similar depth, represent the protagonist of the Kilometer Cube Network (KM3-NeT). 7. Limitations to the ecological monitoring by cabled observatories imaging in the deep sea and envisaged solutions Fixed cameras installed on cabled observatories are only capable of imaging a relatively small area of the marine ecosystem, usually a few square meters of seafloor and overlying water column (Aguzzi et al. 2011). This spatial coverage limitation prevents from extrapolating abundance data to larger areas (Assis et al. 2013). Because of these spatial limitations, researchers need to rely on bibliography or perform laboratory tests to determine if peaks in visual counts at specific locations correspond to peaks in activity (Aguzzi et al. 2015). For example, an increase in visual counts for a certain species may be a proxy of a decrease in the rate of swimming rather than an increase in species activity. This often occurs in species that aggregate close to observatory structures (and therefore observatory cameras) during daytime. Seafloor observatories would greatly benefit from the spatial extension of their monitoring borders by establishing spatial network of functionally coordinated nodes (Aguzzi et al. 2012). Videoing and time-lapse monitoring permit to analyze a wide range of behaviours, including those with an associated periodicity (Aguzzi et al. 2012) but special care must be done with their intrinsic limitations. Powerful lights are needed to illuminate in oligophotic or aphotic depths. Previous studies found out that these lights can attract some fishes biasing animal counts (Longcore and Rich 2004). A solution to consider in certain cases can be the use of infrared light which is not detectable by most marine organisms. Unfortunately, its use is only possible when distances to be enlightened are short due to the rapid attenuation of this wavelength within saltwater (Widder et al. 2005). Biases in animals’ counts can also occur as product of recounting the same individuals as they swim in and out of the field of Introduction 11 view, or when the same individual lingers within the field of view in different images or video frames (Mazzei et al. 2014). Time-lapse image acquisition, at minute and hourly frequencies mitigates this problem somewhat, though for species prone to resident behavior (e.g., scorpaenids) observations might still produce erroneously high estimations of abundance. The technical constraints in time-lapse image monitoring, combined with the bias caused by attraction of some species to the artificial observatory structures, reduce the applicability of using cabled observatories to record unbiased abundance data, but does not wholly prevent their use in the study of activity rhythms (and community changes). Recounting of individuals may be proportional to the differential rate of swimming, and therefore correlate with activity and passivity behavioral phases (Aguzzi et al. 2010). Downward-oriented cameras provide a more uniform field of view for animal counting than oblique imaging(Smith et al. 1993)which would be appropriate for the detecting of walkers and crawlers but would be less advantageous in capturing and identifying active and passive swimming species(Matabos et al. 2014). A strong variability in overall species abundance may also be the result of habitat heterogeneity(Harvey et al. 2012). At a local scale of meters, different animal counts with the same camera, depending on the field of view to which the camera is aimed (e.g., toward a rocky zone rather than to the adjacent water column; Condal et al. 2012,Purser et al. 2013). Therefore, the careful selection of the appropriate field of view and consideration of potential over or under abundance estimations are required. Other solutions, such as the use of cameras equipped with pan/tilt solutions, require the programming of operational routines for rotatory image acquisition and the standardization of counts across local habitat heterogeneity (Pelletier et al. 2012,Mallet et al. 2014). ROV-video transects should ideally be conducted around observatory still camera systems to evaluate the effect of habitat heterogeneity on locally detected species assemblages. ROV surveys focus on spatial rather than temporal abundance variability, allowing large areas of the seabed to be investigated though the surveys are often short and sporadic (De Leo et al. 2010, Grange and Smith 2013). Despite the clear advantage in terms of spatial coverage provided by moving imaging techniques, such photographic or video censuses are not necessarily more reliable for estimation of fish population densities than observations from fixed camera systems. Cumulated multiparametric data produced over long periods of time demand Artificial Intelligence advances for the statistical treatment to reduce human-based time processing costs (Costa et al. 2011). Unfortunately, universal customization of automated video-imaging scripts is highly improbable (Aguzzi et al. 2015)since automated species recognitions depend in too many particular case factors such local species (Aguzzi et al. 2012), heterogeneous background and illumination (Beauxisaussalet et al. 2013). Citizen Science represents an alternative to overcome this issue (Matabos et al. 2017). This term refers to the participation of the public in scientific programs taking advantage of new Introduction 12 technologies, such as the Internet, for easy data collection and sharing (Del Rio et al. 2013)with volunteers who collect and/or process data (Silvertown 2009,Dickinson 2010). 8. Thesis hypotheses and objectives Oscillations as biological rhythms are a pervasive phenomenon in nature (Patel et al. 2015). In extreme environments, such as those of the deep-water continental margin and the deep sea activity rhythms are poorly known. I proposed here three hypotheses dedicated to filling the gap of knowledge on important life traits for species such as behaviour and its spatiotemporal modulation resulting in community changes that will be discussed throughout the work: 1. Tidal, day-night and seasonal behavioral rhythms can result in spatial shifts in populations distributions, and therefore on measurable faunal composition, richness, species abundances and evenness. 2. Inclusion of behavioral components in the spatiotemporal analysis of a community can alter the communities composition, based on species alternations at different continental margin locations, being this phenomenon of paramount importance to understand ecosystem functioning. 3. New multiparametric video-observatory platforms represent a relevant tool to improve the overall knowledge several still unknown life traits of deep-water and deep-sea species. To prove these hypotheses, the general objective of this Ph.D. project was focused to evidence how complex populational rhythms can biases our perception of the composition of local communities faunal composition, richness, species abundances and evenness, through the use of novel but to date still underused observational technologies.The specific aims areto: 1. Examine the interplay between diel, tidal and seasonal cycles in regulating rhythmic behavior of animals with different locomotorbehavior and life stages. 2. Provide valuable ecological information about community composition, dynamics and behavior of species residing deep-sea environments, as a validation of observational technologies and methodologies used here. 3. Provide methodological solutions to the intrinsic and actual limitations of fix cameras to assess animal counting. In my Thesis, I highlight the importance and expand the knowledge on behavioral rhythms in different depth strata and deep-sea ecosystems. To do so, I took notice on threerelevant deep-water and deep-sea environments (i.e. a canyon, a hypoxic fjord and a whale-fall) and carried out studies spanning from few months up to a year with the use of different observational technologies. In addition, I providednew information onhow behavioral rhythms may affect community dynamics. Finally, I specifically tuned protocols for data collection according to the different technical character of platforms Introduction 13 (fixed or mobile) highlighting methodological requirements and mentioning constraints for the appropriate measurement of species behavior over various tidal, day-night and seasonal temporal scales. In this context, I contribute with new directions for the appropriate use and development of observatory technology in relevant deep-sea environments. Introduction 14 Chapter 1. Diel behavioral rhythms in sablefish (Anoplopoma fimbria) and other benthic species, as recorded by the deep-sea cabled observatories in Barkley canyon (NEPTUNE-Canada) Chapter 1 22 1.3.4 Light effects The presence of lighting may potentially alter the behavior of deep-sea fishes, sablefish in particular (Widder et al. 2005), although usually it does not impair the successful detection of population rhythms (Aguzzi et al. 2010)and behavior (Aguzzi et al. 2012). In order to evaluate the potential perturbing effect of punctual lighting during the footage acquisition, a regression analysis was carried out on fish counted at the POD 1, POD 3 and POD 4 in relation to time. This analysis was performed by subdividing the video recording time in two standard halves: 0–25 s and 26–50 s. We randomly selected 15 sequences taken during the first five days of the survey from the three cameras and summed together for each recording half the counted animals per second. We then did separate regression analysis for the number of observed animals in each standard recording half against the observation duration with the software SPSS 20.0.0. Chapter 1 23 1.4 Results We obtained a total of 1487 videos for POD 1 within the canyon axis and 1462 for both POD 3 and POD 4 on the canyon flank. Excluding missing videos due to lighting malfunction, the studies were carried out for 97.7% of POD1 videos and 98.1% of both POD 3 and POD 4 videos. 1.4.1 Faunal remarks Sablefish was the most abundant species observed, with a total number of 1246, 786, and 1735 individuals counted at POD 1, POD 3, and POD 4, respectively. Two other taxa showed relatively high abundances, although much reduced in comparison to the sablefish. Hagfish, Eptatretus spp., total count values were 52, 62, and 50 at POD 1, POD 3, and POD 4, respectively. Crabs, with no taxonomic distinction between species, occurred in lower but still appreciable number, with 27, 63, and 9 individuals at POD 1, POD 3, and POD 4, respectively. We observed other benthopelagic and benthic species that are listed in Table 1. Chapter 1 24 Table 1Complete list of faunal observations (number plus relative percentages) as reported by NEPTUNE- Canada color video cameras installed in Barkley canyon, including the targeted species.That list is complete only for platforms POD 1 and POD 3 on the canyon flank. Such a list is incomplete for platform POD 4, given the lower resolution of its video camera, which in most cases prevented the identification of smaller animals 1.4.2 Time series analysis We plotted a 1-month time series of visual counts for sablefish, hagfish and crabs determined by summing data for the three cameras (see Fig. 2). While noticeable differences in the overall detected abundance among the selected taxa were readily visible, rhythmic fluctuations were appreciable only for the relatively abundant sablefish. The scarcity of visual observations for the other two groups created excessively intermittent and squared fluctuation patterns, with no apparent defined periodicity. In particular hagfish were often observed as coiled within the camera field of view. By considering the total of animals detected at the three cameras those observed as stationary (i.e. coiled) amounted to 69%, those Chapter 1 25 drifting (i.e. displacing without active movement) were 23%, and the remaining 8% were observed as actively swimming. Harmonic analysis of the tidal currents in the Barkley canyon area indicated a mixed diurnal and semidiurnal regime. Fluctuations in the tidal components (Fig.2) showed markedly different amplitude, being larger in the North–south (i.e. V) component than in the East–west (i.e. U) component. The significant tidal components based on signal to noise ratio were: K1 at 0.0417807 c/h (23.9-h); N2 at 0.0789992 c/h (12.6-h); M2 at 0.0805114 c/h (12.4-h); and finally, S2 at 0.083333c/h (12-h). The periodogram analysis did not detect any significant (p ˂ 0.05) diel periodicity in sablefish, hagfish and crab time series (Fig. 3). For hagfish and crabs, this was possibly due to their small abundance (see Figure 2). Since there was no such paucity for sablefish, the periodogram analysis was repeated for this species on the data sets derived from each camera separately (see Figure 3). Results showed the occurrence of population rhythmic activity for POD 1 and POD 4, with percentage of variance levels below 10% while, again, no rhythmicity was visible at POD 3. At POD 4, mixed diel rhythmicity occurred with a day–night or diurnal tidal-based periodicity of about 1440 min (24-h), along with a weaker one at a semidiurnal frequency of about 750 min (12.4-h). This tidal signal was also present into the periodogram plot, with a non-significant larger periodicity at around 1490 min (~25-h). Chapter 1 26 Fig.2.14th October to 14th November 2011 time series in visual counts for the three most abundant taxa (i.e. sablefish, hagfish and crabs), as reported by summing data of the three cameras deployed in Barkley canyon by NEPTUNE-Canada.These fluctuations have been reported along with fluctuations in the northerly and easterly components of the tidal currents (gray, north–south; black, east–west). These currents, made up of the combined effect of the K1, M2, N2, and S2 tidal constituents, are considered markers for the local mixed diurnal/semidiurnal tidal regime. Note the differences in fluctuation amplitude of both tidal components and in the abundance values for the three analyzed taxa. Vertical lines delimit consecutive 24-h cycles. Chapter 1 27 Fig.3. Periodogram analysis outputs on time series of visual counts for the three taxa, as obtained from video recording with the three NEPTUNE-Canada cameras within Barkley canyon.Periodogram results for the time series of Fig. 2 are to the left. For the much abundant sablefish this analysis was repeated for each camera separately (on the right), which showed different significant rhythmicity (reported in min by the percentage of variance peak crossing the significant threshold set at p ˂0.05). Waveform analysis (Fig. 4) for visual count time series for the three species summed over the three cameras showed small phase amplitudes in agreement with periodogram results (see Figure 3). Sablefish displayed a bimodal and nocturnal phase with two main peaks (i.e. values above the MESOR) near sunset and sunrise (respective maxima at 18:00 and 7:00). Conversely, for hagfish there was no Chapter 1 28 discernible diel pattern in their visual count fluctuation, since several peaks appeared at both day and nighttime. Crabs showed a sparse nocturnal increase, with a main peak observed after sunset (centered at 20:00) and other two maxima prior to sunrise (at 3:00 and 8:00). Waveform analysis outputs for water speed data (Fig. 4) were obtained only for the significant North–south tidal component (i.e. V), since it showed the strongest oscillations (see Figure 2). Water speed showed a mix of diurnal and semidiurnal oscillations, with a larger nocturnal phase (maximum at 2:00) and a smaller peaking at daytime (maximum at 14:00). The superposition of waveforms for fish visual counts and water speed seems to indicate that animal detections increase when current velocity decreases. The peak in crab abundance occurred in a similar way and the peak was also in antiphase with tidal water speed maxima. For sablefish, waveform analysis was repeated separately for the time series from each camera (Fig. 5). In the canyon axis (POD 1), a single nocturnal peak at 22:00 was observed in the visual counts. Whereas at POD 3 on the canyon flank, no phase appeared, this result could be attributable to a lowering of signal to noise ratio due to weaker currents and lower animal counts relative to the canyon axis. At POD 4, also in the canyon flank, a different rhythmicity occurred, with a bimodal waveform phase, showing temporal shifts with respect to POD 1. At the location of POD 4, two peaks were visible, a broad one at sunset (18:00) and a smaller one during the day (7:00). Chapter 1 29 Fig.4. Waveform analysis output plots for visual counts and oceanographic time series for the three taxa, as obtained from video recording with the three NEPTUNE-Canada cameras within Barkley canyon.Results for species depicted in Fig. 2 (in black) and for current velocities (along the significant tidal North–south component U; in gray) account for the absence of any strong diel fluctuation in the former. Only sablefish seems to present a bimodal pattern, with a major and a minor peak (i.e. values above the MESOR) at nighttime and in antiphase with water speed increases. MESOR values (dashed lines) account for the overall differences in the abundance of the three taxa and are equal to 5.06 for sablefish, 0.23 for hagfish and 0.14 for crabs. The gray square accounts for the average night duration during the month of October. Chapter 1 30 Fig.5. Waveform analysis output plots on visual count time series for sablefish separately for each NEPTUNE-Canada camera within Barkley canyon (i.e. POD 1, POD 3 and POD 4).Again, MESOR values (dashed lines; POD 1 = 1.7; POD 3 = 1.1; POD 4 = 2.3) account for markedly different abundances and timing of phases, shifting from uni-modal at night-time (POD 1) to crepuscular bimodal (POD 4) or arrhythmic (POD 1). Chapter 1 31 1.4.3 Class-size frequency distribution 195 sablefish were sized during our video recordings (Fig. 6). The total class-size frequency distribution ranged from 35 to 95 cm with a peak at 60–65 cm and average (±SD) length of 63.6 ± 10.4 cm. The repetition of this analysis by each POD camera revealed similar class-size frequency shapes. Average length (±SD) for these class sizes was: POD1 = 66.9 ± 9.9; POD 3 = 62.3 ± 9.1; POD 4 = 62.4 ± 10.6. Although a larger proportion of bigger individuals were reported at the deeper camera (POD 1), no conclusive remark on that difference can be reported given the low sample size. Fig.6. Class-size frequency distribution representing total body length of measured sablefish (N = 195) at the three NEPTUNE-Canada cameras installed in Barkley canyon.That class-size analysis has been repeated by representing total body length of measured sablefish separately for each POD camera. The average length (±SD) was 63.6 ± 10.4 cm and the three frequency distributions showed similar shapes. 1.4.4 Light effects The regression analysis indicated significant perturbing effects of lighting on sablefish visual count (Fig. 7). Considering the 15 randomly selected videos together, we counted in the video recording intervals of 0–25 s and 26–50 s a total of 108 and 68 animals, respectively. This suggests that a significant light attraction effect was reported at the beginning of video recording (p ˂ 0.005), followed by a departure (p˂0.0001) in the second half of the video sequences. Chapter 1 38 1.6 Conclusions NEPTUNE-Canada video cameras enabled to study the activity rhythms of a sablefish (A. fimbria) population in Barkley canyon, west of Vancouver Island. Sablefish is a highly motile marine organism of high ecological and economical relevance in the North Pacific. Synchronous video recording in the canyon axis and flank within the depth range of 892–984 m elucidated a complex behavioral rhythm of displacement for this species that potentially involves these two canyon environments. These results, along with others being obtained in other deep-sea areas of the planet, stress how complexpopulational rhythms can be, thus deeply biasing our perception of the composition of local communities, especially when sampling technologies and strategies produce too scattered results. Chapter 1 39 1.7 Acknowledgments This research was funded by project RITFIM (ref. CTM2010-16274), of the Spanish national RTD program. The Canada Foundation for Innovation and the British Columbia Knowledge Development Fund provided funding for the instrumentation of Barkley canyon. The Spanish DOS MARES project (ref. CTM2010-21810-C03) and Generalitat de Catalunya support to CRG Marine Geosciences (ref. 2009 SGR 1305) are also acknowledged. We thank the ROPOS ROV team and the crews of the CCGS John P. Tully and R/V Thomas G. Thompson for their assistance in the field. J. Aguzzi is a postdoctoral Fellow of the Spanish Ramón y Cajal Program. Ms. Carol Doya is a FPI Ph.D. student. We would like to thank Mrs. V. Radovanovic for editing advice during the manuscript preparation. Chapter 1 40 Chapter 2. The seasonal use of small-scale space by benthic species in a transiently hypoxic area Chapter 2 42 Chapter 2 43 2.1 Abstract The use of small-scale space by benthos and its variation over the seasons in transiently hypoxic zones is poorly known. In this study, we examined the reciprocal spatial dispersion of the squat lobster (Munida quadrispina) and the slender sole (Lyopsetta exilis) according to oxygen concentrations at a VENUS platform of Ocean Networks Canada (ONC). This platform is located in a seasonally hypoxic zone at 96mdepth in the fjord of Saanich Inlet (British Columbia, Canada). We counted and located small as well as large squat lobsters and slender soles in digital still images during 1 year (2012–2013) also concomitantly obtained oxygen data. Images were subdivided in a squared grid to obtain relative density maps as a proxy for surface occupation and spatial autocorrelation. Pearson's chi-squared tests at a yearly scale, along with Dixon's spatial segregation index (S) for each possible pair among the studied groups, showed a significant absence of overlap. The same analyses by month and cross-correlation between oxygen and S showed that while the dispersion patterns of the large squat lobsters seemed to be driven mainly by the morphology of the seafloor, an effect of hypoxia was found in the small squat lobsters and the slender soles levels of aggregation. Small squat lobster sought seabed protrusions, such as sponges, to reach more oxygenated water. The slender sole's space occupation decreased significantly, being forced to retreat when the squat lobsters' abundance peaked as a result of what appeared to be a seasonal reproduction event in early summer. Our results contribute to the understanding of the ways in which oxygen levels modulate substrate use by benthic species in the framework of a global expansion of hypoxia in coastal and ocean areas. Keywords:Fjord, Saanich Inlet, Munida quadrispina; Lyopsettaexilis; Dispersion; Imaging techniques; Hypoxia; Seasonal variations Chapter 2 44 2.2 Introduction The study of the ecological niches regarding the fauna of continental margins requires a detailed knowledge of animal presence and activity in space and time (Aguzzi et al., 2013, Kraan et al. 2013). The spacing of individuals affects the population dynamics, genetics, and evolution of species (Brown and Orians 1970, Dixon 2002). Accordingly, it is of special interest to study the spatio-temporal variability of those factors affecting animal spatial dispersion (i.e., as the position of individuals relative to one another; Ricklefs 1979). Dispersion patterns of benthic megafauna at different scales may result from biotic factors such as reproductive processes, intraspecific social behavior, interspecific coactive interactions, abiotic parameters, and stochastic events (Jumars and Eckman 1983). Therefore, animal dispersion patterns in an area should also be studied in an integrated fashion, along with those environmental parameters that play a key role in the spatial and temporal regulation of these patterns. In particular, special attention should be dedicated to the study of those factors affecting the dispersion of dominant species because the niche occupation by the individuals of these species as a response to local environmental changes may directly influence the presence of other subordinate species and the overall diversity of the community (Aiken and Navarrete 2014). Sediment composition and seabed morphology influence the distribution and abundance of benthos (McArthur et al. 2010). For example, epibenthic walking species are more dependent upon sediment typology than epibenthic swimmers, which are also influenced by currents (Aguzzi and Company 2010, Cnudde et al. 2013, Doya et al. 2014). Such lifestyles would determine the dispersion patterns of deep-water benthos via a differential response of animals to biotic factors such as predation pressure (determining an aggregated pattern), territoriality and competition (a regular pattern), or none of the aforementioned (a random pattern) (Brown and Orians 1970). Other putative factors influencing species dispersion could be related to seasonal changes in environmental conditioning, such as, for example, oxygen levels (Riedel et al. 2014). Thus, an analysis of the consequences of the seasonal variability of such factors for the spatial patterns should be conducted in the context of animal-sediment relationships as a core feature of different lifestyles. In the past years, dissolved oxygen has emerged as amajor environmental driver that modulates the presence of benthos in deep-water areas (Levin et al. 2009). According to Grimes (2014), dissolved oxygen concentration is considered an essential ocean variable (EOV) due to the high impact of variation in its concentrations on biotic systems that inform stated scientific and societal issues. Hypoxia is increasing in coastal and oceanicwaters (Gilbert et al. 2010, Rabalais et al. 2010), due to eutrophication and global warming (Díaz and Rosenberg 2008, Zhang et al. 2010, Rabalais et al. 2014). Shoaling of hypoxic strata is producing noticeable effects in benthic species such as mass mortality (Breitburg 2002) and shifts in average depths of distribution (Keister and Tuttle 2013, Pihl et al. 1991), posing a threat to marine biodiversity (Díaz and Rosenberg 1995, Kodama et al. 2010). The depth expansion of oxygen minimum zones has been associatedwith the decline in important fisheries as a result of habitat loss for Chapter 2 45 targeted species (Craig and Crowder, 2005, Keller et al., 2010). Whilemany studies have addressed the effects of hypoxia on the distribution of benthic communities (Levin et al. 2009), fewhave focused on effects of oxygen levels on the animals in situ (Villnäs et al. 2012), due to the lack of suitable coordinated image and oxygen sampling. Cabled seafloor observatories can provide science from high-frequency (minutes) and prolonged (years) datasets on both benthic communities and water characteristics (Aguzzi et al. 2012,Favali and Beranzoli 2006, Puillat et al. 2012). Here, we used the VENUS observatory located in a seasonally hypoxic fjord (Macoun et al. 2010, Tunnicliffe et al. 2003) to test the hypothesis that the dispersion patterns within and between two abundant benthic species, the galatheid squat lobster Munida quadrispina, and the slender sole Lyopsetta exilis, are random and do not depend on oxygen changes. These species were selected because they dominate the epibenthic community of Saanich Inlet throughout the year and show overall resistance to hypoxia (Dinning and Metaxas 2013, Keller et al. 2010, Matabos 2012). Changes in these dominant and hypoxia-tolerant species are of relevance because they can affect the structure and dynamics of the entire community, and, therefore, strategies of fisheries management and conservation in a scenario of expanding hypoxia. A 1-year time-lapse imaging survey with coordinated dissolved oxygen measurements was designed to assess any changes in the abundance and dispersion patterns of the two selected study species under seasonally transient hypoxia. Only short-range phenomena were examined in detail with statistical tests at small scale to highlight direct inter-individual interactions but also because we were limited by the per-quadrat numbers being generally ones and zeros. Small-scale sampling is of importance for reliably documenting ecologicallysignificant changes at larger geographic scales because mesoscale sampling can miss contrasting environmental shifts at smaller scales (Blanchard and Heder 2014). In literature, small scale is inconsistently defined. In this manuscript, small scale is understood being between 10s of cm and over 100 cm, after Benedetti-Cecchi (2001). 2.3 Materials and methods 2.3.1 The platform, the study area, and data acquisition The VENUS cabled observatory (http://www.oceannetworks.ca) supports a cabled array of instruments in Saanich Inlet (Fig. 9; 48° 39.054' N, 123° 29.203' W; British Columbia, Canada). This siting enables imaging studies of the local benthic community in response to seasonally transient hypoxia(Aguzzi et al. 2011). Saanich Inlet has a maximum depth of 234 m (Herlinveaux 1962) with a shallow sill (70 m) that isolates its basin from the adjacent Satellite Channel. This configuration restricts seasonal renewal of the deep water and promotes oxygen depletion in the basin (Anderson and Devol, 1973, Stucchi and Giovando 1984). Oxygen concentrations b1.0 ml/l occur below100mthroughout the year, and the depth and extent of the hypoxic and anoxic interfaces vary over consecutive months (Tunnicliffe 1981). Chapter 2 46 Fig.9. Map of Saanich Inlet, showing the location of the camera, as part of the VENUS Cabled Observatory infrastructure (black triangle). A 1-year image acquisition program (starting at 00:00 h onMarch 2, 2012, and continuing until 22:00 h on February 26, 2013; UTC) was conducted with a modified Olympus® C8080 wide zoom camera with a high-resolution (8 megapixels) CCD. A ROS PT-25FB pan and tilt unit allowed the selection of an appropriate field of view. The camera tilt was fixed at 45° to record a constant seabed surface of approximately 7200 cm2. The substratum is characterized by consolidated muds around outcropping bedrock in contrast with the predominantly softsubstratum, soupy mud of the lower reaches of the bay (Chu and Tunnicliffe 2015). The field of view encompassed a black and white scale bar marked at 10 cm and a large sponge (upper center; Fig. 10). Illumination was not constant but activated only at the camera shooting times via Ikelite 200 Ws flashes producing light pulses of 2 ms duration. Image acquisitionwas automated so that the camera occupied the same position and sampling occurred at preset intervals. The imaging survey was conducted in a time-lapsemode by taking images every 3 days each 2 h (i.e., 12 JPEG images per 24 h). Chapter 2 47 Fig.10. The selected VENUS camera field of view, with individuals for the three groups: large (A) and small M. quadrispina (B) plus L. exilis (C). Themain seabed structureswithin the area are the PVC scale and the sponges (indicated by the arrows). The image was taken at 18:48 h UTC, on September 29, 2012. Dissolved oxygen concentrationsweremeasured using an Aanderaa Optode 4175 (S/N 0580) mounted on the camera frame with the sensor head 20 cm above the bottom. This sensor has an accuracy of b±0.17 ml/l or ±5% (whichever is greater). Oxygen data were taken at 1 Hz. We focused on the two most abundant, low-oxygen tolerant, epibenthic megafaunal species in the area: the squat lobster Munida quadrispina and the slender sole Lyopsetta exilis. We considered two size classes (i.e., large and small) of squat lobsters. We used key morphological characteristics (Matabos 2012; see Figure10) to define these classes: large individuals were greater than 10 cm, and they possessed thick hairs on their chelipeds. Recruits (i.e., individuals with no visible walking legs) were excluded from analysis because they were too small to be properly identified and counted. Slender sole (see Figure 10) were identified based on Eschmeyer et al. (1983). 2.3.2 Data treatment An overall seasonal screening of biological data was obtained by plotting time series of visual counts for the three groups (i.e., large and small Munida and Lyopsetta) along with dissolved oxygen concentrations. Oxygen concentration data were averaged at a 2 h frequency for the 2 h prior to image acquisition in order to facilitate the visual interpretation of the time series analysis. Chapter 2 54 The relative density map for large squat lobsters (see Figure14) showed a significant aggregation in the farthest part of the field of view (i.e., 3 superior parcel rows), coinciding with the location of a sponge and other structures (also visible as a trend in the majority of the monthly relative density map analysis outputs; see Appendix B). Animals also aggregated in the middle left of the field of view around a rocky complexand in the lower right side, where another isolated sponge is located. The annual relative density map for the small squat lobsters (see Figure14) showed a significant aggregation on two main structures: large sponges and the PVC pipe (see Figure10). The monthly relative density maps (see Appendix B) showed that animals significantly aggregate in high densities along the PVC pipe during the first 3 months of the study (from March to May), when oxygen showed high concentrations (see Figure12). In June, the small Munida significantly aggregated along the PVC pipe but also around the large sponge. From August to November, and after a sharp reduction of oxygen in August, the occupation displayed by these individuals gradually shifted from the parcels corresponding to the pipe to those in the large sponge area. After this period, no small individual was detected until the next February (see also Figure 12).The slender sole occupied the flatter central part of the field of view (see Figure14), an area bounded by the bar and the sponges (see Figure10).However, some of the monthly relative density maps (see Appendix B), mainly corresponding to months with only a few counts, did not reveal this tendency. Fig. 15presents Moran's I spatial autocorrelation for the three faunal groups. There was significantly positive spatial autocorrelation in all cases for distances smaller than 2 density map parcels, thus, patchiness existed in all cases. Spatial autocorrelation was strong for large squat lobsters. Together with that, the first maximum negative Moran's Icorrelation (which indicates the size of patches, “range of influence” or the distance between zones of high and low concentrations), for large squat lobsters occurred for distances greater than 6 parcels, with the corresponding low for slender sole occurring in a distance class close to 4 parcels. Small squat lobsters did not show any significantly negative spatial autocorrelation. Taken together, these results indicate the more robust shape of the area that large squat lobsters occupied in comparison with the elongated area occupied by small conspecifics and slender sole. These shapes are in agreement with the annual density maps (see also Figure14). All correlograms were significant, since none of the coefficients out of the grey area changed in significance after the Bonferroni correction was applied (α′ = 0.00625). Chapter 2 55 Fig.15. Moran's I correlogram for the number counts of A. large M. quadrispina; B. small M. quadrispina; and C. L. exilis. Black dots represent spatial autocorrelation coefficients significant at theα=5% level, before applying the Bonferroni correction; white dots are nonsignificant values. Coefficients for the larger distance values (grey zones in correlograms) should not be interpreted, because they are based on a small number of pair (test with low power) and only include the pairs of points bordering the surface. The horizontal red line helps to identify the sign of the spatial autocorrelation coefficient. 2.4.3 Interspecific relationships in space use The Pearson test of independence examined the degree of dominance in each parcel by each of the groups in relation to the others, highlighting the groups that preferentially occupied given parcels. The results are presented for each group in terms of the percentage of parcels in which the group was significantly (p ˂ 0.0001)more or less prevalent (Table 2). From the beginning of our survey until November, the interspecific relationship between the large squat lobsters and the slender sole changed from a significant under-occupation to significant dominance by the latter. However, both groups prevailed significantly against the small squat lobsters during the whole year except in June, when counts of the small squat lobsters peaked. Chapter 2 56 Table 2.Interspecific dominance relationships, presented as the percentage of significantly more (+) or less (−) occupied parcels over the total (i.e., 81), as observed over a month. All significant (*) comparisons (p b 0.0001) are indicated 2.4.4 The segregation index (S) In the months when only isolated individuals were observed in all images, pairwise comparison was not possible. Large (L) squat lobsters (Fig. 16A) showed an overall opposite trend compared with their small (S) conspecifics and the slender sole (flatfish; F). The S for large squat lobsters was always close to 0, i.e., they did not show a strong tendency to be distant from or closer to other individuals of any of the groups;thus, they were randomly dispersed. The Segregation Index showed that intraspecific aggregations of small Munida(Fig. 16B) occurred throughout the year, with a minimum value observed in August (SSS = 0.29). During this month, the separation of individuals into two groups was more evident as a product of their gradual displacement from the PVC pipe towards the large sponge (see monthly relative density maps in Appendix B). On the other hand, small and large squat lobsters were constantly dissociated (as shown by the negative SSL value obtained throughout the year; see Appendix C). This finding is in agreement with the dispersion pattern of both groups as presented in the annual relative density maps (see Figure 14). Slender sole was spatially aggregated through the year (i.e., SFF N 0), having the highest value in July (SFF=2.75) among all the 9 possible permutations. At that time, individuals occurred as aggregated with small but dissociated from large squat lobsters (SFS = 1.43 and SFL = −1.41, respectively). While SFL and SLF values had similar trends indicating similar reciprocal relations, large squat lobsters more often occurred close to a slender sole than the reverse (having S values closer to 0). In other words, slender sole were less prone to be found near large squat lobsters than the reverse. Chapter 2 57 Fig.16. Segregation index (S) over time for each group of animals with respect to the nearest: A. large M. quadrispina; B. small M. quadrispina; and C. L. exilis. On each graph, large M. quadrispina (dark gray line), small M. quadrispina (light gray) and L. exilis (black) are shown. Gaps in S time series represent absence of intraspecific neighbors. 2.4.5 Correlation with oxygen The linear regressions between segregation index values and dissolved oxygen levels over the year indicated a significant negative trend (p˂ 0.05) only between small and large (i.e., SSL) Munida(Fig. 17) and very close to being significant when S was computed between slender sole and large Munida (i.e., SFL). A negative non-significanttrend was reported in the interspecific S values for large squat lobsters. Chapter 2 58 Fig.17. Linear regression depicting the effect of oxygen levels on the monthly S index for the small—large M. quadrispina pair at VENUS during the study. The SFL cross-correlation with oxygen concentration showed a significant and strong negative correlation at lag 0, −1, and −2 meaning that a particular SFL value can be explained by the oxygen concentrations of the same month but also with the oxygen concentrations of the previous 2 months (Fig. 18 A). In contrast, an almost identical response to oxygen concentration was found for the SFF and SSF (Fig. 18B and C, respectively) showing a rising significant positive correlation from lag 0 to lag−2. This means that both S were more affected by oxygen concentration from the previous 2 months than the contemporary one. Cross-correlation analysis between SSL and oxygen concentration is in agreement with the related linear regression results showing a significant and strong negative relationship (p˂ 0.05) at lag 0with a R2 of 0.59, meaning that the 59% of a particular SSL value can be explained by the oxygen concentrations of the same month. In addition to that, there was a non-significant correlation between a particular SSL and oxygen of previous months (negative lags) indicating that SSL was only affected by contemporary oxygen concentration (Fig. 18 D). In general, the three groups were associated with their respective conspecifics when the oxygen levels were high, thus presenting a clustering behavior as the oxygengot higher, even though this tendency was not significant (data not shown). The principal outcome of these tests was that the small individuals and the slender sole were dissociated from the larger ones when the oxygen concentrations increased and that only the SSL was exclusively correlated with the contemporary oxygen concentration. Chapter 2 59 Fig.18. Cross-correlation between: A. SFL; B.SFF; C.SSF; D.SSL and oxygen concentrations. The maximum (or minimum if the signals are negatively correlated) of the cross-correlation function indicates the point in time where the signals are best aligned. The blue dashed line represents the confidence intervals of the cross-correlation. 2.5 Discussion We observed how benthic Munida quadrispina and Lyopsetta exilis vary their use of the seafloor with respect to small-scale seabed morphology and their reciprocal presence. These relationships appear to be modified by dissolved oxygen levels, as hypoxia gradually builds up in the Saanich Fjord area along spring and summer. Our results indicated that animals occupy different zones within the VENUS camera field of view, Munida in rougher and Lyopsetta in flatter parcels. The seasonal fluctuation of the oxygen concentration is caused by two renewal events associated with a combination of neap tides: river discharge, and the spring phytoplankton bloom (Manning et al. 2010). That oceanographic process deeply modified the spatial arrangement of squat lobsters. Size class segregation occurred in Munida, with small individuals significantly clustering near the base of seabed structures at normoxic conditions and then atop those structures under severe hypoxia. Lyopsetta were mostly displaced by large Munida (replaced in the field of view), when a large increase of squat lobsters occurred during early summer (i.e., see June in Figure12). Burd and Brinkhurst (1984) reported a similar aggregation in fall, as shoaling hypoxia forced an upward migration of squat lobsters in Saanich. Chapter 2 60 However, we do not believe that hypoxia was the cause of the appearance of the aggregation in our case, as we did not see any corresponding decrease in oxygen. 2.5.1 Use of space by Munida Throughout the first 10 days of June, the squat lobster counts showed a sharp increase with densities exceeding 100 individuals/m2 (see Figure12). This early summer aggregation is noteworthy for its timing. A similar seasonal phenomenon is recurrent in Saanich although this case occurred in the fall and is related to the oxygen concentration cycle (Burd and Brinkhurst 1984), which produces a seasonal rise of the OMZ through the depth at which the VENUS node of this study is located (Macoun et al. 2010, Tunnicliffe et al. 2003). However, the lack of correspondence between the rise in Munida abundance and oxygen levels suggests that the origins of the observed aggregation in June are different. In another study 2 years earlier (i.e., in 2010) but 10mdeeper, a maximum in squat lobster densities was also noted in June but was not related to oxygen levels (Matabos 2012). Here, the rise of the hypoxic layer as the cause of the observed peak has been reviewed in view of various aspects of the behavior of the species and with special reference to reproduction, which occurs after molting in some decapod crustaceans (Raviv et al. 2008). We detected the presence of shed exosqueletons of squat lobsters (not quantified) throughout the period of increased animal density and afterwards. This observation raises the question whether these fragments were due to aggressive intraspecific behavior or resulted from synchronized molting. The lack of regular spacing among squat lobsters expected in a framework of intraspecific aggressive behavior (see segregation index values SLL in Figure16) together with the fact that transiently aggressive behavior in M. quadrispina rarely escalates to a real fight (Antonsen and Paul 1997) suggest that agonistic interactions were not the reason for the presence of shed exosqueletons. Shed exosqueletons were neither related with a hypoxia-induced migration towards shallower depths since hypoxic conditions did not prevail here or deeper in the inlet prior to June in 2012; in contrast, retreat of anoxia to deeper depths allowed the squat lobsters population to expand deeper in the basin between May and September of the following year (Chu and Tunnicliffe 2015). The latter must be interpreted considering that year-to-year fluctuations in the anoxic and hypoxic water will influence behavior, causing different patterns to emergence. Our data suggest that the aggregation peak could be related to another phenomenon, such as the reproductive cycle. In areas with seasonal changes in food, reproduction in galatheids is linked to shallower waters (Sanz-Brau et al. 1998) where the survival of larvae may be higher due to enhanced conditions (Palma and Arana 1997). According to that, it is plausible that our single peak in squat lobster densities Chapter 2 61 represents an annual reproductive event and mating might occur outside the range of the camera. That would be in agreementwith the fact that we could not detect any copulatory pairs in images. In this context, the shed exosqueletons found in the field of view in mid to late June could be the result of a synchronous molt before mating, as documented in various crustacean species such as hermit crabs (Wada et al. 2007), squat lobsters (i.e., Munidopsis polymorpha; Thiel and Lovrich 2011), and king crabs (Espinoza-Fuenzalida et al. 2012). Further studies including several years are needed to confirm this annual reproduction hypothesis, but to the best of our knowledge, this is the first report of similar behavior for M. quadrispina and represents a clue to understand the population dynamics for this species. In the framework of a global spread of hypoxic conditions, understanding the population dynamics of a hypoxia-tolerant and dominant species is important to model future scenarios. In our study, the large M. quadrispina were generally epibenthic, but they concealed under seabed structures at times. Large squat lobsters primarily occurred in the farthest regions of the field of view (see Figure 10), as observed in the relative density maps (see Figure 14). Accordingly, the results of the spatial autocorrelation analysis indicate the presence of a single wide patch or “zone of influence” (see Figure 15). The preference for rough areas could be related to feeding behavior. Galatheids are carnivores and scavengers (Garm and Høeg 2000), occasionally capturing swimming zooplankton (e.g., M. quadrispina; Burd and Brinkhurst 1984), preying on krill or foraging in detritus (Aguzzi and Company 2010). Little is known about the behavior of the small squat lobsters in relation to seasonally changing oxygen levels. We have observed these individuals progressively concentrating on prominent seabed structures (e.g., the sponges) as the oxygen concentration dropped (see the relative density map analysis for August in Appendix B). The oxygen concentration decrease in August may have been forcing small (thusless tolerant to hypoxia; Burd, 1985) M. quadrispina to move up onto seabed projections in an attempt to gain access to the layers of more highly oxygenated water above them. An identical response has also been found in various epibenthic crustacean species (Haselmair et al. 2010, Riedel et al., 2014). Similarly, the burrowing crustacean decapod Norway lobster (Nephrops norvegicus) suppresses its retraction behavior, constantly remaining exposed on the more oxygenated seabedwhen oxygen drops in eutrophic North Sea areas (reviewed by Aguzzi and Sardà 2008). In contrast, when oxygen was not a limiting factor, small Munida were observed hiding, primarily under the PVC pipe. Therefore, one may speculate that the hiding behavior of small individuals under normoxic conditions is the result of predation pressure and of avoidance of intraspecific aggression (by large individuals). That explanation is consistent with the significantly negative linear regression between oxygen levels and the intraspecific segregation index (SSL) and with the strongly significant negative cross-correlation in which the 59% of the variability is explained by the oxygen concentration of the Chapter 2 62 SSL.In fact, predation pressure affects the small-scale spatial distribution of benthic macrofauna (Serpetti et al. 2013), as the need for protection against predators is a behavioral driver for strong aggregations of epibenthic decapod crustaceans (Corgos et al. 2010). In agreement with the aforementioned distribution of small individuals, we could only detect a very narrow patch or “zone of influence” from the spatial autocorrelation analysis. In general, despite this oxygen-induced movement, small Munida remained aggregated in patches throughout the whole year, as revealed by the constantly positive intraspecific SSS value. However, large Munida did not showa similar trend, with the SLL close to 0, indicating a random intraspecific dispersion pattern. The existence of a size-dependent tendency has also been found in three different invertebrate phyla in certain Atlantic fjords, where small species were highly aggregated, in contrast with the larger ones (Kristensen et al. 2013). Findings similar to our results have been obtained in spider crabs. Adult spider crabs do not display a clear spatial structure, whereas small individuals are found in strong aggregations (Corgos et al. 2010). 2.5.2 Use of space by Lyopsetta The biological time series (see Figure12) highlighted that the slender sole rarely increases in number when the squat lobsters do, and viceversa. This opposite trend is confirmed by the interspecific dominancerelationships determined by the Pearson test of independence (see Table 1). The absence of spatial overlap among benthic species might be caused by reciprocal avoidance, competition, territoriality, different preferences for food supply or habitat, or physical barriers (Brown and Orians, 1970). It is well known that Pleuronectiformes such as Lyopsetta emerge from the sediment in the benthic boundary layer to feed and then re-bury themselves(Holmes and Gibson 1983, Yahel et al. 2008). This lifestyle restricts their presence to flat seabed areas inwhich their burying behavior is feasible. Therefore, the spatial autocorrelation analysis indicates a narrowdistance between zones of high and low concentrations also in agreement with the annual relative density maps outputs (see Figure 6). The preference for different substrates was not the only reason explaining the absence of overlap with Munida. Simultaneously with the seasonal peak in densities of the squat lobsters, we detected a sharp reduction in the counts of the slender sole. Given that Lyopsetta is very resistant to hypoxia (i.e., animals are reported to occur down to oxygen levels of 0.1 ml/l,Matabos 2012), the concentrations recorded in the summer of 2012 could not be the reason behind the aforementioned decrease. The negative segregation index SFL values (see Appendix C) indicate that the slender sole rarely occurred close to large Munida and suggest the occurrence of strong interspecific competition for space that results in the displacement of the flatfish. Conversely, large squat lobsters are less affected by the slender sole (see less negative SLF values in Appendix C). The significant and negative cross-correlation between SFL may be interpreted a Chapter 2 63 clustering behavior of slender sole in relation to large Munida when oxygen concentration was high. Although these species are highly resistant to hypoxia, their optimal range of oxygen concentration may entrain this behavior. Taken together, these results indicate that slender sole aggregates in optimal oxygen conditions and their life habits limit its presence to flat areas (i.e., for burying) but that spatial limitation becomes even stronger during the Munida aggregation. This spatial study at small scale using imaging of high-tolerant to hypoxia species is of importance to understand community dynamics at different levels, including the role of large scale movements. Additionally, probable scenarios in the framework of expanding hypoxia may be inferred from our results on the reciprocally different spatial influencebetween slender sole and large squat lobsters when oxygen is low (although those results must be carefully interpreted as our study is based in a single location). 2.5.3 Methodology remarks The presence of the observatory platform could have biasing effects on animal detection (and counting) for a phenomenon of attraction. We assume that these effects were constant throughout the study, so that the observations were related exclusively to external factors (Aguzzi et al. 2010, 2013). Another factor that could have been affecting the accuracy in counts was transient turbidity. During periods of low visibility caused by resuspended sediment (Yahel et al. 2008), smaller squat lobsters located in the distant field of view might have been miscounted. Despite that, the high frequency of sampling (12 images/day every 3 days) minimized the effect of the aforementioned considerations on the-results. Chapter 3 70 3.3 Materials and Methods 3.3.1 The NEPTUNE network and the Barkley Canyon node Authorization for installation of the infrastructure supporting this research was provided by the Transport Canada (http://www.tc.gc.ca), after Fisheries and Oceans Canada (http://www.dfompo.gc.ca)assessed that the cabled installation would not have a negative impact on fish habitat. Field studies did not involve endangered or protected species. The NEPTUNE cabled observatory network, off Vancouver Island (BC, Canada) operated by Ocean Networks Canada (ONC; http://www.oceannetworks.ca; Barnes et a. 2015) supports continuous multiparametric and video observations from coastal to deep-sea habitats, providing power and data connectivity through a 840-km looped fiber-optical cable (Fig. 19). Chapter 3 71 Fig.19.Overview map of the cold-seep site in Barkley Canyon (northeast Pacific, Canada). (A) The cabled observatory network, The Barkley Canyon node, site of Wally I crawler deployment is highlighted by a black dot. (B) High-resolution bathymetric map showing the region of Barkley Canyon investigated in the current study. The black dots within the canyon axis represent the crawler node with the nearby mid-canyon POD4 platform and other deeper nodes. (C) Image demonstrating the crawler Field Of View (FOV) as used to acquire the faunal data. (D) Map showing position of navigation waypoints arranged around the hydrate outcrop, the crawler (near waypoint no. 12 in this schematic) and the route from waypoint no. 10 to 14, representing the survey transect analysis area. The cold seep community was distributed on a soft bottom zone with no apparent emerging rocks. At the end of the transect analysis area, a 5 m depth cliff (dashed grey line) separated our study area from a from the gas-hydrate mound (E) The crawler in operation. The observatory network is powered by a shore station located in Port Alberni, Vancouver Island (see Figure 19 A). This sophisticated interdisciplinary monitoring infrastructure supports a node in Barkley Canyon at ~890 m depth (Fig. 19 B; 48°18.89′ N, 129°03.48′ W) , a node which has become an Chapter 3 72 important hub for in-situ study of the environmental drivers modulating ecosystem functioning of deepsea cold-seep communities (Belley et al. 2016, Thomsen et al. 2012). For the duration of the study described herein, the Wally I crawler was in operation on the western flank of the main canyon, connected by an umbilical power and data cable to the node (Fig. 19 B). The node is also within the local Oxygen Minimum Zone (OMZ), and thus a site of further global interest as a location for ongoing monitoring during global environmental change. A full description of the study site (Figs 19 C and 19 D) the images recorded by the crawler (Fig. 19 E) installed in this area (Fig. 19 F) is provided in (Purser et al. 2013). 3.3.2 The ‘Wally I’ crawler deployment, connectivity, operations and sensor systems For the duration of the current study, the mobile ‘Wally I’ crawler was connected via a 70-m fiber-optic tetherto the Barkley Canyon node (Purser et al. 2015) with syntactic foam floats arrayed every 3 m along the tether to insure buoyancy and to preclude entanglement with the crawler tracks. Crawler driving control and data acquisition were carried out in real-time via a custom web interface, with sensor data and imagery automatically stored to the NEPTUNE archive at the sampling resolution of each device (Purser et al. 2015). The ‘Wally I’ crawlers’ mobility is provided by a pair of caterpillar tracks, which may be operated together, or independently to drive the vehicle forward, backward or to rotate it on the seafloor. Movement commands are made directly from the home laboratory over an internet connection. Movement of the caterpillar tracks (and therefore, the vehicle) creates a footprint on the seafloor that could potentially disturb infauna. To prevent these impacts influencing the collected results transects aimed at collecting time series data are usually run over the same track, hence reducing the impacts of the tracks on the ecosystem. Such an operational plan was carried out in the current study (see Figure 19 E). One of the primary sampling systems on the ‘Wally I’ crawler is the forward looking camera system. Video footage can be acquired using the 470 Line ROS Inspector low-light, colour camera, equipped with an 18X optical zoom whenever suitable illumination is also provided. For the current study, 20 m transects were driven every 4 hours, over the first 5 days of each month, from February 14th 2013 to April 14th 2014, with video data recorded on every occasion. For all transects, the camera was always oriented with a fixed tilt angle of 45º from the horizontal, from a mounting position roughly 1m from the seafloor at the front of the crawler. During operations, the crawler’s motion was kept at a near constant speed (0.02 m/s). Light was provided by two deep-Sea power and light-variable intensity lamps (Purser et al. 2015). Lights were switched on during transect driving and turned off immediately after, in order to minimize photic contamination in the area. The total observed seafloor area recorded within each transect was approximately 60 m2. Chapter 3 73 The ‘Wally I’ crawler can mount a plethora of sensor systems, and for the duration of the current study an upward facing 2-MHz ADCP (Nortek Aquadopp Profiler AQD 9917), measuring water velocity and temperature at 1 mab and a Seapoint Turbidity Meter, measuring at approx. 0.2-0.3 mab the turbidity were equipped. An adjacent observatory (the POD4 NEPTUNE node), situated at a distance of ~500 m from the site of crawler operations, within the mid-canyon at 896 m depth (48°18.8923′ N, 126°03.4804′W; see Figure 19 B) was simultaneously recording pressure, water density data and oxygen, as collected respectively by a CTD (Sea-Bird SeaCAT SBE16plus V2 7027) and an oxygen sensor (Sea- Bird SBE 63 630111). Collected data at 1 min frequency throughout the entire survey period, both at times coinciding to the collection of video transect data as well as during the periods of immobility between crawler transects. These data are freely available online (http://dmas.uvic.ca/DataSearch). Additionally, the Bakun Index, as a proxy of upwelling and downwelling processes, was also computed at 48N, 125W from National Oceanic and Atmospheric Administration (NOAA) / National Marine Fisheries Service(NMFS) / Pacific Fisheries Environmental Group (PFEG) data. 3.3.3 Faunal data collection and treatment Following collection of the transect video data, one user made visual counts of the megafauna individuals present within each transect. The same user analysed all collected videos, identifying individuals to the lowest taxonomic level as possible using the NEPTUNE Canada Fauna Identification Guide (Gervais et al. 2012). All video recorded were archived and are available online through the Oceans 2.0 portal (http://dmas.uvic.ca/DataSearch). Although a near constant crawler speed and camera angle were maintained throughout the duration of the study, occasional variabilities in field of view recorded did result from small differences in seafloor relief across the transect survey length. To account for this, we analysed only the portions of video data where 75% of the field of view encompassed the seabed, and with sufficient water transparency for animal classification and counting. Slight variabilities in transect duration also occurred, due to slight differences in seafloor angle, firmness and due to technical driving and internet connectivity issues (e.g. interruption of signal send / return from our working office in Barcelona to the Pacific). Therefore, to make the biological data recorded as comparable as possible over the video transects collected, we standardized visual counts for the each megafaunal taxa and transect to the maximum video duration recorded within the study, which was 23 min. In order to highlight species count patterns as proxies of seasonality, we plotted time series of average visual counts (± SD) for each month, based on all transects recorded during the 5 days of monitoring at the start of each calendar month throughout the 14 month period. In the resulting plots, we superimposed horizontal dashed lines representing the ‘Midline Estimated Statistic of Rhythm Chapter 3 74 (MESOR;Aguzzi et al. (2015), computed by re-averaging all monthly values. Their overlaying onto the monthly plot allows the identification of significant seasonal increases or decreases in abundance. Current velocity component-vectors (i.e. North-South and East-West, xy direction data collected via the ADCP system) were transformed to flow magnitude (m/s). All environmental data collected by the crawler and the POD4 node (see 2.2) were averaged into 4 h bins to match the video-sampling frequency. Bakun Index values, the daily-averages of wind-driven cross-shore transport, were computed from Fleet Numerical and Meteorology Oceanographic Centered (FNMOC) 6-hourly surface pressure analyses. 3.3.4 Analysis of the community structure linkage with oceanographic variables A Kruskal Wallis test was carried out using the R statistical language (R Development Core Team 2008), to detect significant differences in the number of megafaunal visual counts (of each species or taxa logged) between months, following the methodology presented in (Zar 2010). A Kruskal Wallis test was carried out using the R statistical language (R development core Team, 2008), to detect significant differences in the number of megafaunal visual counts (of each species or taxa logged) between months, following the methodology presented in (Zar 2010). A Nonmetric Multidimensional Scaling (NMDS;Minchin 1987) analysis was performed in the R library vegan (Oksanen et al. 2006) to visualize the level of similarity among species presence and visual counts (i.e. assemblage structure) together with correlated environmental vectors into a Cartesian plane. The Bray-Curtis dissimilarity Index was used to quantify the dissimilarity between megafaunal species based on the time series visual counts, while aWisconsin double standardization was performed (since this improves the gradient detection ability of dissimilarity indices; Oksanen et al. (2015)). The significance of the estimated determination coefficients (r2) of the environmental variables fitting onto the species ordination, produced by the NMDS analysis, was estimated using a permutation test. Results of the species ordination using NMDS and the correlation with the environmental variables are shown in a plot, with environmental vectors (arrows) showing significant (p< 0.05) maximum correlation with the species ordination. Results are shown only for the environmental variables significantly correlated to the species ordination. Biodiversity was calculated with two indices presented as mean diversity per month: the Shannon index (H’) and the Simpson Index, presented as 1-D. In a similar manner, we presented the species richness (S). The monthly abundance (number of individuals of all species) was normalized to the maximum number of transects per month (i.e. 48). Herein we refer to these means as biodiversity indices, Shannon index, Simpson Index or abundance as appropriate. Chapter 3 75 3.3.5 Ethological remarks Behavioural observations were logged for the motile fauna imaged during the crawler surveys. These, though opportunistic, were described in order to increase general information on the ethology of several deep-sea species, and may represent pivotal aspects of the functioning of some ecosystems, particularly in terms of inter- and intraspecific relationships (Aguzzi et al. 2012),and also to help support the second hypothesis of the current study, that IOV technology has application as a solid faunistic monitoring tool. . Chapter 3 76 3.4 Results 438 video transects, representing a total of 92 h of footage of acceptable quality for analysis was collected during the 14-month monitoring period. No video data was collected during March 2014 due technical problems. 7698 megafauna individuals, belonging to 6 phyla were observed (Appendix E). These included representatives of 26 taxa, 12 of which could be identified to species level, with remaining individuals classified to higher taxonomical levels (S1 Fig). We detected considerable differences in megafaunal abundances, between species and temporally, throughout the study period (see Appendix E). The most abundant species within the study period were the sablefish (Anoplopoma fimbria) with 2214 visual counts, representing 29% of the total observed megafauna. Buccinids (Buccinoidea) were also abundant with 1318 individuals representing 17% of the total megafauna. The third most abundant taxon was hagfish (Eptatretus stoutii; i.e. 852, as 11%). The fourth was abundant was Scyphomedusa (Poralia rufescens) with 789 individuals (10%). Next were rockfish (i.e. from the family Sebastidae) with 573 filmed individuals (7%) and small crabs (of undefined taxon) showing a similar abundance (N=535, 7%). The final significantly abundant taxon, with 449 records, was ctenophores (Bolinopsis infundibulum), representing 6% of recordings of fauna. 3.4.1 Seasonal patterns Time series analysis was carried out for the 7 most abundant megafaunal species (Fig. 20, plotted to show the occurrence of any monthly variation as a proxy of seasonal fluctuation). Grooved tanner crab count time series were also plotted to show the particular variabilities in their abundance resulting from their reproduction dynamics. Since no video data was collected during March 2014, we assumed significant values during this month when the concomitant months were significant. An overview of the time series analysis showed species peaking frequently from February to August 2013 and from January to April 2014. A general lack of species occurred during the central months of the study (autumn and December 2013). Chapter 3 77 Fig.20. Visual counts time series (i.e. from 14th February 2013 to 14th April 2014) for the 7 most abundant megafaunal species and the particular case of the grooved tanner crab, as reported by crawler videoimaging. Average values (the grey dots) have been reported in order to better highlight the occurrence of seasonal trends. From left to right and from top to bottom taxonomic units are: Rockfish (i.e. Sebastidae); Sablefish (Anoplopoma fimbria); Hagfish (Eptatretus stoutii); Buccinid (Neptunidae); Small crab; Grooved tanner crab (Chionnecetes tanneri); Ctenophore (Bolinopsis infundibulum); and finally, Scyphomedusa (Poralia rufescens). Horizontal blue dashed lines correspond to the Midline Estimated Statistic of Rhythm (MESOR) and horizontal blue lines identify significant seasonal increases visual counts increases. Blue asterisks correspond to a month with no data due technical problems. A close-up of Figure 20 showed that rockfish were present in roughly consistent densities throughout the observation period, though significantly higher (i.e. above the MESOR) numbers of them were observed during spring and August 2013 and from January to April 2014. We observed a significant sharp peak in sablefish visual counts during late spring (March) and summer 2013 reaching their maximum abundance in July, with a mean of 29.1±23.8 SD individuals per transect. Hagfish were also Chapter 3 78 uniformly abundant throughout the study, with a modest but significant count peak in February-March and July 2013, and another one of similar magnitude in January 2014. Buccinids showed low but consistent abundances during much of the survey period with a mean of 3.3±4.46 SD individuals per transect. Though a moderate but significant increase occurred from May to June 2013 and a sharper one occurred in February 2013 (i.e. 18.3±18.9 SD individuals per transect). Small crabs were rarely present during the study period, although their numbers were reaching significant levels intermittently from August 2013 to April 2014 (i.e. August, October and from December 2013 to April 2014), reaching a maximum in January (i.e. 4.5±6.7 SD individuals per transect). Grooved tanner crabs were present all year-round with visual counts increasing moderately but significantly during March and April 2013 (1.6±3.7 and 1.6±2.1 SD individual per transect, respectively) and again in April 2014, with 1.1±2 SD individuals per transect being observed (coinciding with reproduction; see the Behavioural Remarks Section below). Another significant peak was observed during early and mid-summer 2013 (June and July). Ctenophore visual counts displayed clear seasonal patterning. A sharp increase in numbers occurred during autumn with a maximum density observed in October 2013 (i.e. mean density of 6.9±6.4 SD individuals per transect). Finally, the Scyphomedusa was significantly present during almost all months of the monitoring period (i.e. from February to June and from November to December 2013 and February 2014) with a moderate count decrease from mid-summer to mid-autumn. Fig. 21shows the time series data for the investigated oceanographic parameters, as measured by the crawler on board sensors and the nearby POD4 network node. There were a few occasional gaps in data acquisition produced by sensor malfunctions or data network issues. A patterning in seasonality is evident in water density, temperature and turbidity measurements. Additionally, the Bakun Index and oxygen concentrations measured during the 14 months of the study showed seasonality in the values measured. Velocity was highly related to the diurnal and semi-diurnal tidal cycle. The Bakun Index presented the lower values during February 2013, indicating predominant downwelling processes ongoing during that month. A progressive increase in the Bakun Index led to positive values as sustained pulses (i.e. upwelling events) from mid-April to August 2013. Coinciding with these upwelling events, we detected the greatest recorded variations in turbidity. Elevated turbidity levels were coincident with the first shallow water incursion in the local deep-sea area in August, which brought in warmer and less salty waters. Oxygen levels ranged from 0.24 (October 2013) to 0.29 ml/l (March 2014), with a mean of 0.26 ml/l. Oxygen concentrations gradually decreased in August, reaching their minimum values in September-October. Thereafter, oxygen progressively increased to return to the initial concentrations by the end of the study. This lower oxygen period coincides in time with Bakun Index reaching values close to zero. Water density also showed a weak seasonal trend, with generally warmer and fresher water masses present at the study site from July to October. Chapter 3 79 Fig.21. Oceanographic parameters time series (i.e. from 14th February 2013 to 14th April 2014), as recorded by crawler and the nearby POD4 platform within the Barkley Canyon. Average values (the grey dots) have been reported in order to better highlight the occurrence of seasonal trends. Gaps in data acquisition were due to instrument malfunctioning or downtime for the cabled infrastructure. These parameters are: (A) Velocity; (B) Water Density; (C) Temperature; (D) Turbidity; (E) Bakun Index; and finally, (F) Oxygen. We used up-looking and down-looking arrows to highlight upwelling and downwelling periods respectively. Chapter 3 86 Conversely, less information exists on deep-sea populations’ responses to seasonal oceanographic variations, especially where benthic habitats occur within OMZs (i.e. depth strata with oxygen concentrations < 0.5 ml/l; (Levin et al. 2002). The depth of our study site corresponds to the core of the OMZ in the North Pacific (Juniper et al. 2013). The high-frequency variations in water mass physical properties (i.e., water density, O2 levels, current velocity and temperature) detected at this coldseep site were relatively minor in comparison with those measured at shallower depths of the canyon in (Juniper et al. 2013). This suggests that the region surveyed by the crawler in the current study, at ~890 m depth, was below the depths affected by shelf-edge upwelling, as observed at similar depths. Our results in the seasonal patterns in abundances, as well as the NMDS analyses suggest that some of our assessed species are indirectly influenced by upwelling (sablefish) and downwelling (rockfish and buccinids). In Juniper et al. (2013), the authors suggested that an inverse correlation of dissolved oxygen concentration and temperature could be an indication of water mass changes, as seems also to have been the case in the area investigated in the current study. Sablefish day-night nektobenthic migrations in Barkley Canyon are in antiphase with current speed, a possible compromise strategy between search for prey and energy saving due to physiological limitations(Doya et al. 2014, Chatzievangelou et al. 2016).Our data suggest that sablefish could follow a similar strategy also at a seasonal scale with animals avoiding shallower depths affected by upwelling. This would explain the peak in sablefish visual counts, the peak in total megafaunal abundance and the sharp decrease of biodiversity in July. Significant seasonality in sablefish abundance has been previously reported in SE Alaska, with the local population associated with deeper waters during the winter period (Csepp et al. 2011). Buccinids exhibited a sharp increase in densities during February 2014. These gastropods are both predators and scavengers, known to form aggregations either to exploit prey(Aguzzi et al. 2012, Lapointe 1992)or to mate (Himmelman et al. 1993). They were not affected by currents above the seafloor, yet their maximum abundances coincided with the most pronounced downwelling period of the year. NMDS showed a positive correlation of buccinid abundances with increasing oxygen levels. In contrast, gastropods are known to be particularly resistant to hypoxia (Theede et al. 1973). This suggests an indirect effect of downwelling on the abundance of this species in our location. In agreement with this, (Matabos et al. 2014)found that B. viridum migrated from shallower depths to a location ~200 m from our study area, presumably to avoid enhanced currents. Highest rockfish densities also coincided with the strongest downwelling period (from January to April 2014). The NMDS indicated that their abundances were affected by the same environmental factors that influenced buccinid abundances. Although rockfish has often observed to be resident at fixed locations (Marliave et al. 2013,Reynolds et al. 2010, Yoshiyama et al. 1992), our findings suggest that at least some individuals move into this area to avoid enhanced currents in a similar strategy to the buccinids. This hypothesis needs to be confirmed with further studies. Although buccinids and rockfish were also observed outside of the strongest downwelling periods, their numbers were close to the significant threshold and coincided with Bakun Index values close to zero. Chapter 3 87 Additionally, rockfish are the major contributors to sablefish diet (Liadig et al. 1997). This could explain the decrease in rockfish counts during June and July when sablefish abundances were the highest of the study period. Hagfish and the Scyphomedusa increased in abundance coinciding with an increase of oxygen concentration following the period of low oxygen concentrations from August 2013 to January 2014. NMDS outputs suggest that these species have their ideal oxygen threshold above the minimum levels observed in our study. Data on small crab abundance suggest a more complicated picture, with both seasonal (month-scale) fluctuations and diel trends negatively correlated with oxygen levels (Chatzievangelou et al. 2016). A seasonally-related change in the depth zonation of the grooved tanner crab has been reported for the US West coast, with local densities at particular depths varying temporally and as a consequence of their growth and reproductive cycle (Keller et al. 2012). Sexes are also completely segregated by depth during spring and summer, whereas males and females mix in autumn and winter months, when males move into deeper waters(Pereyra et al. 1966). The observed peaks in visual counts of individuals within the current study, during March-April and June-July 2013, together with the subsequent sharp decrease and maintenance of low abundance maintained until the next spring, could be related to a similar seasonal displacement. Here, high visual counts of small crabs were made in December 2013, with abundances remaining high throughout winter. To our knowledge, there have been no previous studies of seasonal recruitment of Chionoecetes tanneri, so this potential evidence of a synchronous recruitment event should be considered with caution. According to (Jamieson et al. 1990)female Tanner crabs, which live at deeper water depths than males, move shallower for egg release and mating in March and April. Their higher abundance here in March-April, during higher oxygen levels, could be explained by a requirement of eggcarrying females to reduce the energy investment for oxygenating their eggs, as has been observed for other crustaceans (Fernández et al. 2003). To summarize, abundance trends and ethological observations throughout the study period, suggest that a massive reproductive aggregation of grooved tanner crab occurred during March and April 2013, following a previous migration from shallower waters (males) and deeper waters (females). This may have resulted in an associated peak in recruits observed in from December 2013 to February 2014. Ctenophores were present during the autumn months of 2013. In Juniper et al. (2013), the authors also detected an increase in ctenophore visual counts in October-December at a nearby midcanyon site. After occupying shallower depths near the surface ocean in spring and summer, a period coinciding with the maximum surface chlorophyll levels, (Båmstedt et al. 2015)found larger ctenophore individuals overwintering in deep waters, where they prey on copepods. Such vertical displacement of larger zooplankton represents a trade-off between feeding efficiency and predation avoidance (Darnis et al. 2014). Ctenophores are able to perform vertical movements at various temporal scales (i.e., diel, tidal, and seasonal migrations) thanks to their high tolerance of transient hypoxia (Thuesen et al. 2005). Our Chapter 3 88 time series counts and NMDS outputs support the occurrence of a similar phenomenon at the cold-seep site investigated here. We detected a subsequent increase in the visual counts of the Scyphomedusa, a species previously reported as being associated with OMZs (Osborn et al. 2007). Although little is known about the diet preferences of the species, other Scyphomedusa prey on Bolinopsis infundibulum, which could explain the concurrent peaks of these two gelatinous species (Martinussen et al. 1999,Båmstedt et al. 2015). The same occurrence pattern for this jellyfish has been also observed in the mid-Barkley Canyon (Juniper et al. 2013). As previously mentioned, this species was not present in video-data during the periods with the lowest oxygen levels, likely due to physiological limitation. Although community composition differed throughout the study period, changes in species richness did not reach an order of magnitude, with a similar number of species replacing some of the previously existing ones. This was also reflected in the two biodiversity indices, which maintained a relatively constant level apart from 3 sharp decreases leading to low points (i.e. June-July, September- October 2013 and February 2014). In these particular periods, a single species (i.e. sablefish, ctenophores and buccinids, respectively) was disproportionally abundant, resulting in low biodiversity through low evenness. Taken together, these findings confirm our first hypothesis that abundance, richness, and biodiversity annual changes in relation to oceanographic conditions are directly driven by oxygen, current velocity, and upwelling/downwelling, and indirectly driven by seasonal benthopelagic and nektobenthic migrations along the canyon. 3.5.2 Methodological remarks Crawler motion produced a transient sediment clouds in the water column behind the vehicle. Since all monitoring and environmental sensors were mounted at the front of the crawler, this is not expected to have affected any of the data values measured by the sensors. Though capable of surveying similar areas of seafloor as ROVs, crawlers (particularly IOVs such as ‘Wally I’) do not need ship assistance to operate, and in this study a crawler was efficiently used to perform a long-duration, spatially extensive study. In addition, ROVs are commonly perceived as “foreign objects” to the local or periodically visiting fauna, whereas permanently deployed, slower moving IOVs are permanent elements of the local benthic panorama rapidly accepted by the resident fauna. Crawlers may therefore be considered a less intrusive ethological monitoring technology (e.g., for monitoring reproduction and making natural situation feeding and agonistic behaviour observations), since they are likely “accepted” by the local fauna. The disturbance caused by crawler lights on the behaviour of resident animals has not been well investigated. Sablefish have been previously found to be attracted by observatory lights for short periods of time and then to leave the field of view of cameras mounted on the NEPTUNE nodes (Doya et al. 2014). This short-duration light attraction suggests that animals are unlikely follow the crawler during transects and therefore unlikely to be recounted erroneously as successively imaged individuals. The Chapter 3 89 remainder of the mobile species detected didn’t react or showed only minor reactions to crawler presence, avoiding it only when it was in extremely close proximity. In any case, bias generated by the presence of the crawler would have been maintained as constant throughout the study period given the uniform methodology employed in collecting the transect video data, and therefore not prevent the detection of seasonal patterns in megafauna abundance or biodiversity, or the successful collection of ethological observations. Therefore, our second hypothesis, that IOV technology represents a reliable faunal monitoring tool for work in the deep sea is also confirmed. Chapter 3 90 3.6 Acknowledgements This work was conducted within the framework of Ocean Networks Canada (ONC) scientific activities, being led by Dr. J. Aguzzi “Science Theme Leader” for “Life in the North-East Pacific”. C. Doya is a PhD Student with a FPI grant by Spanish Industry, Economy, and Competitiveness Ministry (MINECO). The authors would like to thank the shipboard teams and staff scientists of Ocean Networks Canada and OceanLab Bremen. Dr. S. Mihály provided CTD and fluorometer data. The study was supported by ROBEX project of the Helmholtz Society. Chapter 4. Faunal activity rhythms influencing early community succession of an implanted whale carcass offshore Sagami Bay, Japan Chapter 4 92 Chapter 4 93 4.1 Abstract The fate of large whale carcasses sinking to the deep-sea environments has intrigued scientists in the last three decades. Whale carcasses provide a surplus in food supply to demersal animals from a vast array of life strategies, serving as habitats for a specialized fauna that represents the counterpart of chemosynthetic symbiont-bearing taxa, generally inhabiting hydrothermal vents and hydrocarbon seeps. A distinctive pattern in benthic community succession resulting from animal scavenging activity has been established but previous studies relied on punctual sampling and observations conducted by submersibles and ROVs. The contribution of diel (i.e. 24 hrs. based) activity rhythms within internal tidal scenarios to the dynamism of that community succession has never been evaluated before. Here, we used an autonomous time-lapse video-imaging system mounted on a lander at 2 hrs. frequency, in order to count and identify all benthic and demersal fauna, near an artificially implanted juvenile sperm whale carcass at 500 m depth off Sagami Bay in Japan, over a 2.5-month period. We also acquired concomitant water current speed and temperature data, through an Acoustic Doppler Current Profiler, showing that the local diurnal internal tidesare mixed with a dominant semi-diurnal periodicity of currents and water temperatures in the whale deployment area. Notwithstanding, with chronobiologic time-series analysis protocols (i.e. periodogram and waveform analyses), we discovered that visual count patterns of most abundant megafauna mostly followed a 24 hrs. fluctuation pattern (i.e., the snubnosed parasitic eel, Simenchelys parasiticus, the Japanese spider crab Macrocheira kaempferi, andfinally the Japanese gissu Pterothrissus gissu). Only one species (the Japanese codling, Physiculus japonicus), showed a semidiurnal rhythmicity. Our results highlighted the early occurrence of scavengers and the later appearance of predators and detritivores. The diel alternation of species at the scavenging site and the temporal succession recorded suggest the occurrence of temporal partitioning in the access to whale carcasses among deep-sea animals as ecological strategy to reduce interspecific competition and predatorymediated mortality risks. Keywords: scavenging activity, whale carcass, benthic communities, high-frequency analysis, behavioral rhythms, internal tides. Chapter 4 94 4.2 Introduction The fate of large whale carcasses sinking to the seafloor in both shallow and deep-sea environments has intrigued scientists in several research fields spanning from community ecology, evolution, and biogeography for almost three decades (Smith et al. 2015, Sumida et al. 2016). The term ‘whale-fall’, coined after the incidental discovery and observations of a balaenopterid whale skeleton at 1240 m depth in Santa Catalina basin (California, USA), has been since used to generally define these oases-like organic-enriched seafloor environments (Smith et al. 1989,Smith and Baco, 2003). In the extremely poor food-supply environments such as the deep sea, a 40-ton sunken whale approximately provides an enormous input of organic matter (~2 X 106 g C), that may be equivalent to over 200 years of particulate organic matter arriving the seafloor via vertical and lateral fluxes in the form of marine snow aggregates and phytodetritus (Smith and Baco 2003). Whale carcasses not only provide a surplus in food supply to benthic and demersal animals belong to a vast array of trophic guilds, but they also serve as habitats for a specialized fauna that represents the counterpart of chemosynthetic symbiont-bearing taxa, inhabiting other reducing environments such as hydrothermal vents and seeps (Smith et al. 1989). For that resemblance, particularly the sharing of at least 20 species with hydrocarbon seeps, whale falls have been hypothesized to act as dispersal stepping stone habitats for vent and seep faunas (Baco et al. 1999, Smith and Baco 2003, Smith et al. 2014, Sumida et al. 2016). At evolutionary time scales, whale carcasses have also been postulated to support the invasion of deep-sea chemosynthetic habitats by shallow water taxa, also in a stepping stone fashion (Baco et al. 1999, Fujiwara et al. 2010, Lorion et al. 2013). After almost three decades of studies around naturally occurring and artificially implanted whale carcasses on the seafloor, a distinctive pattern in benthic community succession has been established, with four well defined stages (see Smith and Baco 2003,Smith et al. 2015 for details): i. a mobile-scavenger stage, lasting months up to 1.5 years, ii. an enrichment-opportunistic stage, also lasting months up to 4.5 years, iii. a sulphophilic stage, lasting for decades, and finally iv. a reef stage, potentially lasting also for several years. Duration of each stage is ultimately dependent upon whale carcass size and seafloor environmental conditions such as depth, currents, temperature and dissolved oxygen. Cyclic changes in the environmental conditions around whale carrion may deeply influence the succession dynamic, by generating a rhythmic turnover in local species composition. In the aphotic deep sea, changes in water flow speed and temperature (Wagner et al. 2007,Aguzzi et al. 2011), together with seasonal food input from the euphotic zone (Billet et al. 1983, Graf 1989, Fanelli et al. 2011), are the main environmental drivers in temporizing behavioral rhythms of benthos. In fact, rhythmic population displacements into the water column-seabed three-dimensional scenario occurs more typically at a diel (i.e. 12 and 24 hrs.) base into the water column (as Diel Vertical Migrations or DVMs), horizontal Chapter 4 95 displacements along the seabed (as nektobenthic movements across shelves and slopes), or in and out from the sediment (as endobenthic) (Aguzzi and Company 2010). The role of those population displacement rhythms on whale scavenging dynamic in relation to overt environmental cycles has been poorly investigated to date. The behavioral activity of scavenging species and their access to the whale carrion could be temporally regulated upon internal tidal motions, as well as upon the rhythmic presence of benthopelagic predators into the local benthic boundary layer (Aguzzi et al. 2015). Other factors regulating species turnover could be related to intraspecific competition pressure which may provoke a partitioning of time of access to the same food resource by species with overlapping trophic niches (Kronfeld-Schor and Dayan 2003), namely the carrion in the case of facultative versus obligated scavengers. Previous studies on whale fall communities have relied on punctual sampling and observations conducted by submersibles and remote operated vehicles (ROVs) (Smith et al., 1989; Lundsten et al., 2010a, b; Smith et al., 2014). More long-term and high-frequency in-situ observations and sampling are currently lacking due to logistical constraints, mostly associated with high operational costs in accessing remote deep-sea environments and/or difficulty in whale availability and sampling. However, multiparametric benthic platforms such as landers can be efficiently used to study the contribution of species activity rhythms in the scavenging dynamics, the time scales of the transitions between successional stages, as well as the role of interspecific relationships among species competing for carrion resources, including predator-prey interactions (De Leo et al., 2016). These autonomous platforms enable high-frequency, continuous and prolonged time-lapse imaging of benthic communities protracted over consecutive months along with concomitant oceanographic monitoring to assess the environmental control on behavior(Aguzzi et al. 2012a, 2015a; De Leo et al., 2016). The present contribution offers insights on how multiparametric landers time-lapse videoimaging can be used to gather data about whale fall species turnover, resulting scavenging dynamic, and the environmental control, by applying time series analyses fitting chronobiology standards and multivariate analysis. Chapter 4 102 The snubnosed parasitic eel, Simenchelys parasiticus, was the most abundant species, representing more than half of total observations (56.5 %, 4052 individuals) (see Appendix G). The Japanese spider crab, Macrocheira kaempferi, was the second most abundant species, present in 14.0 % of the observations and with 1002 individuals, followed by another bony fish, Pterothrissus gissu with 5.4% in frequency of occurrence and 390 individuals (see Appendix G). Finally, the isopod Bathynomus doederleinii was the fourth most abundant species with 4.9% (N=349) of occurrence. The detection at the beginning of the time-lapse sequence of a large individual of the bluntnose six-gill shark, Hexanchus griseus, feeding on the whale carcass (see Figure 25 H) was noteworthy. Also, the whole whale carcass slightly shifted its position at the beginning of the survey (i.e. 6:30 a.m., on June 15th). Then, it suffered other 3 major shifts during the experiment, at 8:30 a.m. on July 15th, 4:30 a.m. on July 26th, and finally at 2:39 a.m. of the on August 7th(Fig. 26) with no apparent co-detection of other shark individuals. Fig.26. Temporal changes in the whale carcass positioning with respect to the camera field of view as detected by time-lapse imaging. A. 06.15.2012 to 07.15.2012; B. 07.15.2012 to 07.26.2012; C. 07.26.2012 to 08.07.2012; and finally, D. 08.07.2102 to 08.24.2012. In A, two individuals of Solaster paxillatus, and single individuals of Echinothurioidea, Ophiuridea, and Buccinum yoroianum are visible. In C and D, two ribs appear on the seafloor. A B C D Chapter 4 103 In terms of the biological data, the visual counts time series for all species indicate different patterns of occurrence near the carcass (Appendix H) depicting a noticeable change in community composition particularly marked by the slow replacement of the crab M. kaempferi by the fish P. gissu around the 34th day of the experiment (between the 4th and the 5th week). While some species occurred during the entire period of observations (e.g. S. parasiticus), others peaked toward the middle (e.g. the isopod B. doederleini) or the end (e.g. ophiuroids and the sea cucumber Pannychia moseley) of observations. 4.4.2 Time series analysis outputs Oceanographic time series of water temperature, pressure, and current flow speed (East-West and vertical components) at sea bottom, re-sampled at 2 hrs. intervals, are presented in Figure 27. A mixed diurnal-semidiurnal tidal cycle occurred in the area, as indicated by the FFT analysis. In all these series, the semi-diurnal cycle is dominant with lower cycles present at 6 and 24 hrs. periods. Average (±SD), maximum and minimum values for the measured oceanographic variables were: temperature (°C) = 5.73 ± 0.44, max. = 7.00, min. = 4.23; pressure (dBar) = 493 ± 0.49, max. = 493.68, min. = 491.31; vertical velocity component (cm/s) = -1.89 ± 1.2, max. = 3.7, min. = -5.6; east-west velocity component (cm/s) = 1.47 ± 13.5, max. = 51.2, min. = -30.3. Chapter 4 104 Fig.27. Temperature at sea bottom, water pressure, vertical and East-West current component as recorded over 72 days by ADCP close to the whale carcass deployment.Whale carcass position changes (see Figure 25) are indicated by vertical light grey dotted lines. -6 -4 -2 0 2 4 6 14/6 15/6 16/6 17/6 18/6 19/6 20/6 21/6 22/6 23/6 24/6 25/6 26/6 27/6 28/6 29/6 30/6 1/7 2/7 3/7 4/7 5/7 6/7 7/7 8/7 9/7 10/7 11/7 12/7 13/7 14/7 15/7 16/7 17/7 18/7 19/7 20/7 21/7 22/7 23/7 24/7 25/7 26/7 27/7 28/7 29/7 30/7 31/7 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 9/8 10/8 11/8 12/8 13/8 14/8 15/8 16/8 17/8 18/8 19/8 20/8 21/8 22/8 -60 -40 -20 0 20 40 60 14/6 15/6 16/6 17/6 18/6 19/6 20/6 21/6 22/6 23/6 24/6 25/6 26/6 27/6 28/6 29/6 30/6 1/7 2/7 3/7 4/7 5/7 6/7 7/7 8/7 9/7 10/7 11/7 12/7 13/7 14/7 15/7 16/7 17/7 18/7 19/7 20/7 21/7 22/7 23/7 24/7 25/7 26/7 27/7 28/7 29/7 30/7 31/7 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 9/8 10/8 11/8 12/8 13/8 14/8 15/8 16/8 17/8 18/8 19/8 20/8 21/8 22/8 490 491 492 493 494 495 14/6 15/6 16/6 17/6 18/6 19/6 20/6 21/6 22/6 23/6 24/6 25/6 26/6 27/6 28/6 29/6 30/6 1/7 2/7 3/7 4/7 5/7 6/7 7/7 8/7 9/7 10/7 11/7 12/7 13/7 14/7 15/7 16/7 17/7 18/7 19/7 20/7 21/7 22/7 23/7 24/7 25/7 26/7 27/7 28/7 29/7 30/7 31/7 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 9/8 10/8 11/8 12/8 13/8 14/8 15/8 16/8 17/8 18/8 19/8 20/8 21/8 22/8 4.0 4.5 5.0 5.5 6.0 6.5 7.0 14/6 15/6 16/6 17/6 18/6 19/6 20/6 21/6 22/6 23/6 24/6 25/6 26/6 27/6 28/6 29/6 30/6 1/7 2/7 3/7 4/7 5/7 6/7 7/7 8/7 9/7 10/7 11/7 12/7 13/7 14/7 15/7 16/7 17/7 18/7 19/7 20/7 21/7 22/7 23/7 24/7 25/7 26/7 27/7 28/7 29/7 30/7 31/7 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 9/8 10/8 11/8 12/8 13/8 14/8 15/8 16/8 17/8 18/8 19/8 20/8 21/8 22/8 Pressure (dBar) Time (day/month) Temperature (°C) East-West current component (cm/s) Vertical current component (cm/s) Chapter 4 105 Using the FFT diel component of East-West current it is possible to evaluate the water displacement associated to this tidal motion: by integrating in time the current we obtain exactly this displacement. This allows estimating both the distance reached in this movement (ranging between 0 to 440 m) and the time phase of this process. Time series and stick diagrams of daily mean horizontal currents (shown in Appendix I-J) allowed describing their temporal and spatial evolution during the period of observation. The currents were not vertically homogeneous: southeastward flow was dominant up to the 10th ADCP level (i.e. about 32 m above sea bottom), whereas northwestward flow was characteristic from 30 to 60 m above the bottom, indicating different transportation processes and water mass origin for the two layers. This difference in the pattern was also consistent with the results of the complex correlation analysis which reports lower value coefficients moving from lower to upper layers. In relation to the vertical components analysis, data show that a downward flow was prominent across the investigated water column, being stronger near the bottom and basically following the semidiurnal period, as shown by the pressure time series and in Appendix K(i.e. weak downward flows correspond to neap tides, represented by the lowest pressure excursion). To detect the occurrence of tidal motions and to analyze their temporal evolution, timedependent complex frequency analysis was performed on the horizontal and vertical components time series (shown in Appendix L). The predominance period of the spectrum was observed for 24, 12 (the highest peak), and 6 hrs., confirming tidal (diurnal and semidiurnal) periodicity for the flow around the sea bottom. Same results have been obtained for water temperature data (but with a less evident peak in correspondence of the 24 hrs. period). Periodogram analysis identified the occurrence of significant day-night related (i.e. 24 hrs.) visual count fluctuation patterns for 3 species (see Appendix G): S. parasiticus (swimmer; P=24.0 hrs.; Var.=15.79%), P. gissu (swimmer; P=24.0 hrs.; Var.=22.64%), and finally M. kaempferi (walker; P=23.8 h; Var.=7.76%). A single species showed a tidal-related periodicity, the fish Physiculus japonicus (swimmer; P=12.3 hrs.; Var.= 7.78%; see Appendix G). Interestingly, no other significant tidal periodicity was revealed for the remaining species, and only Coelorinchus sp. showed a different periodicity related to inertial currents (P=19.5 hrs.; Var. =8.56%). The waveform analysis conducted for S. parasiticus, M. kaempferiand finally P. gissu, as species showing significant 24 hrs. (Fig. 28 A, B, C respectively), from count patterns in previous periodogram analysis (see Appendix G), indicated the occurrence of temporally coherent phases (i.e. a continuous series of values above the MESOR). The phase timing comparison through species indicates the occurrence of a progressive phase shift from night (S. parasiticus) to daytime (P. gissu). The correspondent integrated diurnal displacements along the East-West axis, computed from the FFT results, Chapter 4 106 are superimposed with a 2 hrs. delay to match the S. parasiticus periodicity, while for the M. kaempferi the delay is of about 8 hours. Fig.28. Waveform analysis outputs for the visual count time series of those species showing significant 24 hrs. periodicity patterns in the periodogram analysis (see Appendix G for A. S. parasiticus, B. M. kaempferi, and C. P.gissu. Note the varying Y-axis abundance scales. Up and down arrows indicate onset and offset of significant increments in abundance peaks, respectively. They also indicate the first and last calculated mean abundance values above the Midline Estimated Statistic Of Rhythm (MESOR as dashed horizontal lines) line as re-average of all waveform data. MESORs are: A, 4.76; B, 1.18; C, 0.46. The correspondent integrated diurnal displacements (m) along the E-W axis, computed from the FFT results, are superimposed, showing a 2 hrs. delay in the animal periodicity. 0.7 0.9 1.1 1.3 1.5 1.7 16:30 18:30 20:30 22:30 0:30 2:30 4:30 6:30 8:30 10:30 12:30 14:30 0.0 0.2 0.4 0.6 0.8 1.0 1.2 16:30 18:30 20:30 22:30 0:30 2:30 4:30 6:30 8:30 10:30 12:30 14:30 2.0 3.0 4.0 5.0 6.0 7.0 8.0 16:30 18:30 20:30 22:30 0:30 2:30 4:30 6:30 8:30 10:30 12:30 14:30 A B C Time (hrs) Mean (±SD) no. animals A Displacement (m) -500 -400 -300 -200 -100 0 100 200 300 400 500 600 2.0 3.0 4.0 5.0 6.0 7.0 8.0 16:30 18:30 20:30 22:30 00:30 02:30 04:30 06:30 08:30 10:30 12:30 14:30 Chapter 4 107 Waveform analysis was repeated for the vertical and the East-West current components (Appendix M), partitioned once in 12 hours and then in 24 hrs. segments, to show both diurnal and semidiurnal tides. Vertical current speed values were in average negative (i.e. downward) and its tidal components (diurnal and semidiurnal) were in phase with the East-West components. Hence both components produced oscillations along the local topography. The correspondent diurnal and semidiurnal Fourier component curves (the dash-dotted lines superimposed in Appendix M), confirmed the same trends. Both the vertical and East-West current components showed an evident bimodal fluctuation, typical of mixed diurnal and semidiurnal regime, even if the semidiurnal cycle was clearly dominant (about tenfold the diurnal cycle). The waveform analysis conducted separately for scavengers, detritivores and predators, indicated the presence of a temporal segregation of species according to their food preferences (Appendix N). Values above the MESOR indicate significant nocturnal and diurnal increases, respectively for scavengers (and facultative predators) and predators. Conversely, no clear phase is reported for detritivores, likely due to a scarcity of visual counts. These results were sustained by periodogram analysis outputs (see Appendix G), which identified a 24 h significant rhythmicity in respective data sets, with arrhythmia for detritivores. 4.4.3 Multivariate analysis outputs Multivariate analyses partially confirmed and added information to previous results: two main phases were identified by cluster analysis, with weeks 1-6 significantly separated by weeks 7-11 (accordingly to SIMPROF test at p<0.05, black lines in Figure29A), at 60% of resemblance level. This separation (weeks 1-6 vs. 7-11) resulted highly significant also according to PERMANOVA test (F1,10=11.34, p<0.001). Consistently, SIMPER analysis evidenced those species which mostly typified each phase (Table 4), with S. parasiticus abundant in both phases and M. kaempferi and ophiuroids being the taxa that most typifying phase 1 (weeks 1-6) and phase 2 (weeks 7-11), respectively. The 2-D bubble nMDS (Fig.29B) showed the abundance of these latter taxa in the two different phases: the overlaid clusters at 60% of similarity mostly divided weeks 1-6 from 7-11. Chapter 4 108 Fig.29. Cluster analysis of the mean abundance values by day of species recorded at the whale fall for factor “week”. Black lines indicate significant separation at p<0.05 according to SIMPROF test; B. Bubble nMDS plots with over imposed cluster analysis results (at 60% of resemblance level) of two of the dominant taxa of each phase (weeks 1-6 vs. 7-11): Macrocheira kaempferi and ophiuroids. Chapter 4 109 Table 4Results of the SIMPER analysis carried out on the two periods (weeks 1-6 vs. 7-11) evidenced by multivariate analysis The ordination on the first two axes of the CANOCO (Fig. 30 A) showed the association of species (represented by point, block 2) with the environmental variables (represented as vectors, block 1). The first (horizontal) axis represents consistent patterning in species ordination, from positive (right) to negative (left) side with a variable time of appearance according to the “Date” variable (see Figure 30 A). From right to left that ordering is made by scavengers then predators, and finally detritivores. 2-block PLS (Fig. 30 B) showed how the most important variables on the first axis of the block 1 (88.5% Covariance) were mainly Date and then, the Horizontal Component V; variables on the first axis of the second block for this reason, are associated with species succession from the earlier (positive side) to the later (negative side). Chapter 4 110 Fig.30. Scatter plot of the first two axes of the CANOCO. Species and trophic categories (i.e. scavengers, predators, and detritivores) were represented by black points and environmental variables were represented by red vectors. Environmental variables considered are: Date of the observation as “Date”, spatial components of current velocity as “EW_Comp_V” and “Vert_Comp_V”, as well as Depth, Temperature and the time shifted. B, Bar chart of the loadings relative to the first latent vector (explained variance = 77.44%) for both environmental and species blocks (no. 1 and 2, respectively). Chapter 4 111 4.5 Discussion We reported for the first time to our best knowledge, the occurrence of a diel turnover in species appearance at a slope whale fall site (i.e. about 500 m depth), which accounts for the importance of behavioral rhythms in regulating processes in deep-sea environments. Interestingly, despite the local internal-tidal diurnal mixed with a dominant semi diurnal regime of currents, visual counts fluctuations of most abundant macro- and megafauna (as a proxy of population rhythms; Aguzzi et al. 2010)followed a day-night like patterning. No information on the light intensity at slope depths exists for this area. Notwithstanding, our behavioral data indicate that light temporal axis, together with a weak, but not negligible, diurnal tidal cycle, may still indirectly exert a role in regulating slope benthic communities’ temporal turnover based on species rhythmic behavior in the dark deep-sea environment. Currents FFT analysis shows peaks with periods of 24, 12 and 6 hours (shown in Appendix L) and, coherently, the correspondent integrated displacements along the E-W axis in each semi-period are: 440, 940, and 140 m, respectively. These lengths may be considered as the average distances reached by each cyclic movement. The presence of that diurnal tidal cycle, even if not dominant, may explain, together with a still possible day-night effect, the evidence of the 24 hrs. based local community turnover in most videoabundant species (i.e. Simenchelys parasiticus, Macrocheira kaempferi, and Pterothrissus gissu), despite their progressive trends in overall detection over the 72-days monitoring period (see Appendix H), in agreement with successional stages at whale scavenging sites (Smith and Baco 2003). This alternation of species may be temporally sustained at aphotic deep-sea depths by species undertaking massive displacements along bathymetric gradients involving photic and aphotic strata at their extremes (Bird and Au 2012). Animals swarming as isolated or into schools within the Benthic Boundary Layer or BBL (nektobenthic mode) or, vertically into the water column (benthopelagic mode) may be rhythmically present in the local BBL despite the overt tidal flows drift Aguzzi et al. 2010, 2011). That movement may convey day-night timing information ruling communities’ turnover into aphotic benthic zones in an indirect manner, adding another layer of temporal complexity to deep-sea ecosystems (Aguzzi et al. 2010). Our observations agree with day-night changes in deep-sea species demersal assemblages at 500 m depth in central Japan, as reported through temporally-scheduled trawling and are consistent with the idea of benthopelagic and nektobenthic displacements (Suetsugu and Ohta 2005). Specific examples can be given by interpreting our results for some of the species here analyzed, in the light of different rhythmic behavior according to their life traits (i.e., locomotor and/or feeding behaviors see Appendix G).For example, diel current appears in phase with the walking Japanese spider crab M. kaempferi (see Fig. 28B): its abundance increases when the East-West flux reachesits minimum values (Appendix Mpart B) and about 8 hours after the maximum correspondent displacement (i.e. possibly, the time to the odor Discussion 118 Discussion 119 The general objective of this Thesis was to understand how activity rhythms of benthic species within deep-water and deep-sea communitiesmodulate overall richness and evenness, hence biasing biodiversity measured in our seabed sampling windows,through the use of novel remote, multiparametric and continuous observational multiparametric video-technology, highlighting at the same time, methodological sampling problems associated to the different fixed or mobile characterof platforms. My major goal was to scale observed changes in community composition down to the behaviour of individuals which show temporal expression in response to cyclic environmental changes. In large-scale studies, different locations are usually sampled at different times of the day and/or different days throughout a sampling trawling or ROV cruise (e.g. Young et al. 1986, Olivar 1990, Sabatés 1990, 1990,Gray, 1993). Consequently, any reported difference among sites or studies may potentially be confounded with species behaviourally time-induced variations (e.g. Morrisey et al, 1992, Trush et al.1994). There is particularly scant information in case of deep-water and deep-sea areas, where behavioural rhythms of residing species and their nektobenthic or benthopelagic mode of displacement are still largely undescribed. This is due to the difficulties of performing direct and long-lasting observations in such extreme and distant locations at frequencies disclosingabundance changes and resulting community dynamics. Nevertheless, a scientific approach considering this time-dependent constrains, is of paramount importance, since the biodiversity we can perceive in our limited sampling windows is directly related upon the movement of populations within a three-dimensional water columnseabed environment (Aguzzi and Company, et al. 2011). Therefore, the first aim of this Ph.D. thesis was to move a step forward, shedding new light on the regulation that environmental cycles exerts to animals’ rhythmic behavior within three iconic deep-water and deep-sea ecosystems: a large canyon, a cold-seep, a temporal hypoxic fjord (as a proxy of a temporal hypoxic deep-sea area) and finally, an ephemeral whale fall site. In this context, the second specific objective of this Ph.D. dissertation was to provide information about faunal composition, richness, species abundances and evenness, and other valuable ecological information in other to date still unknown species life traits within the abovementionedthree deep-water and deep-sea ecosystems. The difficulties to perform high-frequency and long-lasting studies at those sampling frequencies matching the expression of rhythmic displacements of populations are inherent to the present limitations of marine observational technology. Fortunately, enhanced and novel subsea monitoring platform engineering occurred in the last 30 years, nowadays allowing us to fill gaps of ecological knowledge impossible to achieve in earlier in time. Accordingly, the third specificand last specific objectiveof this Ph.D. thesis was to establish a methodology to solve operational difficulties in sampling and data analysis due to the peculiar and different fixed and motile nature of the used platforms. Discussion 120 1. Internal and external validity of the thesis 1.1 First objective:time as ruler of deep-sea community changes To date, the use of chronobiology approaches and data analysis practices only received poor attention in deep-sea ecology. While most of data have been derived by direct trawl sampling, and some with by ROV or AUV technologies, only few of them could achieve a temporally structured sampling. Additionally, only few studies addressed the behavioural activity and admit regulation upon habitat variations (e.g., fishes and internal tides by ROV). In the present work, I applied concepts and data treatment methodologies from the chronobiology field to time series of biological and environmental data acquired with new observational technologies, hence highlighting species abundance variations as a putative product of populational rhythmic displacement in and out from the sampled seabed window. In Chapters 1 and 4, I used periodogram and waveform analyses to characterize behavioural rhythms in deep-sea twilight zone ranges and determine the influence of different environmental factors. By using these methods, I succeeded in the determination of significant day-night and tidal rhythms of sablefish by cabled video-stations, but I couldn’t find any significant rhythm in hagfish and crabs presumably due to the scarcity of their counts. My results were recently confirmed by Chatzievangelou et al (2016) who also found those behavioral rhythms with the use of mobile crawler. Different rhythms were heighted in shallower but still disphotic zones of the Japanese continental margin. Most abundant whale-fall-ecosystem associated macrofaunal species (the snubnosed parasitic eel, Simenchelys parasiticus, the Japanese spider crab Macrocheira kaempferi, andfinally the Japanese gissu Pterothrissus gissu) followed a 24-hday-night oriented rhythmicitywhile only one species, the Japanese codling(Physiculus japonicus) followed a semi-diurnal (tidal) rhythmicity (see Chapter 4). Both day-night oriented and tidal rhythms seem to be a widespread phenomenon occurring in the deep-sea waters of all oceans, showing the pervasive character of photic and hydrodynamic cycles in shaping ecosystem functioning in oceans worldwide. Apparently, species preforming vertically extensive diel benthopelagic vertical migrations reaching shallower photic stratamay convey temporizing information to benthic deepsea communities into disphotic and aphotic zones when touching the seabed. Ocean tides are recognized to be one of the main factors controlling intertidal communities (Connell, 1972) but their temporizing action in deep-sea environments is still very unclear. The determination of tidal rhythms at nearly 1 km depth (Chapter 1) demonstrated how important can be this factor all over the continental margin extensions. Recently, Lelièvre et al. (2017) confirmed the influence of tides at 2 km depth in animal’s behavioral rhythms therefore, supporting my results. In addition to ocean tides, wind-generated forcing such as that brought by storms can have a strong influence on bottom-currents ocean dynamics (Lelièvre et al. 2017) by generating downward-propagating inertial waves and low-frequency currents by pressure fluctuations (Thomson et al., 1990; D’Asaro et al. 1995). Discussion 121 In contrast, in Chapters 1 and 4, periodogram analysis screening periods within the circa-tidal and day-night range could not find a single species showing a significant periodicity related to the inertial motion cycle. This indicates that inertial-currents might not be an important environmental synchronizer of deep-sea animals since blurred by stronger tidal bulges. In fact, in the Mediterranean where these latter are weak and the former are hence much appreciable in current seed data sets, abyssopelagic species show more often inertial-oriented periodicities. This is the case of bioluminescence time series of data as recorded by the Italian neutrino deep-sea telescope highlighting 20-h periodicity in biogenic light peaking due to the clashing of descending deep-scattering layers of organisms against the infrastructure (Aguzzi et al., 2017 Sci Rept. Km3 NET bioluminescence). Nevertheless, behavioral rhythms of particular deep-sea communities such as hydrothermal vents can substantially disrupted by local hydrodynamics that substantially change local chemical conditions(Lelièvre et al. 2017). In these communities, behavioral rhythms are more influenced by tides and inertial-waves than for day-night cycle, though this seems to be an exception to the general rule. Three years of monitoring would ensure the discernment of seasonal cycles though, must of the seasonal studies are based on scattered sampling along few years being their results subject to a lack of redundancy. In this sense, our high-frequency studies can partially counteract the lack interannuality. The utility of long-term observatories in the detection of deep-sea reproductive biology such as the detected in the present work (Chapters 2 and 3 reproductive migration of squat lobsters and grooved tanner crab, respectively) is corroborated by other studies such as the one of Fujiwara et al. (2007), that could find short-term and rare synchronized gamete emission by Calyptogena spp. thanks to the high-frequency sampling of a long-term in situ observatory. Reproductive patterns were also detected for deep-sea fish communities in more physiological studies at seasonal base for other Galatehids in the NW Mediterranean continental margin (Aguzzi et al., 2013) as well as for fishes (Fernandez-Arcaya et al., 2012, 2013) being also some reproductive rhythm phased the upon lunar motion(Mercier et al. 2011). For all that, the assessment of biological rhythms and the broad ecological monitoring represent a fast-growing motivation in the development of the “cabled observatory science” (sensuFavali and Beranzoli 2006,Danovaro et al. 2017). In this framework, present results on biological rhythms represent a small but important fraction of what is needed to generalize the observed temporal dynamics to other deep-sea environments. As conclusion of this sub-section I determined that the main environmental cycles entraining behavioral rhythms are day-night, tidal and seasonal cycles. Inertial-currents might not be as important as environmental synchronizer for biological rhythms in deep-waters in comparison with the tides. This is true with some exemptions such as hydrothermal vents in which inertial currents can deeply modify the local chemical environment(Lelièvre et al. 2017). The detection of biological rhythms can reveal the existence of deep-sea mesoscale features and their variability. Migrations along the water column in deepsea areas stress how complex populational rhythms can be thus highlight that the sampling methodologies Discussion 122 that produce too scattered results might have a deep biased. Therefore, long-term in situ observatories represent and ideal tool for seasonal studies permitting an otherwise impossible temporal resolution. 1.2 The Second objective: community composition, dynamics and behavior of species in deep-sea environments The response of species to environmental changes may directly affect intra- and interspecific relationships, overall diversity and being hence of relevance for the evaluation of important ecosystem services such as fisheries (Danovaro et al. 2017). In Chapter 2, I demonstrated that an Essential Ocean Variable, EOV(i.e. oxygen) may force the local spatial arrangement of different benthic species exposing them to predating pressure and competition for the substrate. Good evidence of this was seen in the squat lobster Munida quadrispina and the slender sole Lyopsetta exilis, the number of which (as a proxy of local population density)showed important seasonal modifications according to increasing hypoxia. My results contribute to the understanding of the modulation of substrate use by benthic species in the framework of a global expansion of hypoxia in coastal and ocean areas named as Oxygen Minimum Zones. Chapter 2 species spatial arrangement was also conditioned by the small-scale seabed morphology, which might also be the reason behind the differences of sablefish counts between POD 3 and 4 (see Chapter 1). In fact, recently Lelièvre et al. (2017) highlighted the importance of local favourable conditions in the small-scale distribution of animals supporting our research results of Chapter2 supporting my findings. Also in Chapter 4, I found a temporal lag in diel increases of visual counts for the 3 most abundant species (i.e. the snubnosed eelSimenchelys parasiticus, the Japanese spider crab Macrocheira kaempferi, andfinally the Japanese gissu Pterothrissus gissu), which I interpreted as the occurrence of interspecific competition among species acceding to the feeding resource at the study site. The use of observational technologies allows documenting punctual events impossible to report otherwise.This is the case of the reproductive behavior of the grooved tanner crab videoed for the first time in Chapter 3. In this footage I could observe individuals engaged in a particular moment of their mating, in which a large male hold a smaller female above the ground with its left clawed. These technologies can potentially witness also other behavioural activities in relation to interactionsamong species, as for example in the case of crabs feeding on a clam or the seal feeding on a hagfish at Pod 4 camera in Barkley Canyon (depth: 894 m) in January 2013. This latter event was detected within the framework of the Digital Fishers citizen science project (Matabos et al. 2017). In relation to that, the other punctual but important observations refer to the activity of large predators as in the case of the six-gill shark (see Chapter 4) heating chunks ofa sperm whale carcass. Also in chapter 3, we demonstrated the Discussion 123 existence of monthly fluctuations of abundance, biodiversity and richness of megafauna with the mobile platform Wally I. As a conclusion of this section,I determined that the factors affecting the spatial but also temporal distribution of deep-sea species and their populations are driven by the competition for the substrate and feeding as key ecological resources within the framework of intra- and interspecific relationships (e.g. reproduction behaviorand predation)as well as in response to the variation of environmental key drivers such as the dissolved oxygen. Additionally, increasing our knowledge of animal’s behavior within a changing deep-sea environment through the non-invasive monitoring has also socio-economic implications and I determined here that observational technologies represent an appropriate tool to reach this goal. Nevertheless, physical collection is irreplaceable for robust record of speciesecological traits requiring the taxonomical determination of the species beyond the resolution capacity of imagery, as required to monitor changes on species level. Thenoveltyofthe present ecologicalresearch lies inhowthe information mainstreaming obtained bymeansofnew underwater observational technologies applies inan integrative mannertoformanideaofspecificnichewithinacommunity for an specific species. Anapproachthatis hardly achievable with laboratory studies. 1.3 Third objective: methodology remarks Some ecosystems are difficult to study with the traditional physical sampling strategies (e.g. rough bottom ecosystems with demersal trawls or dredges; (Beisiegel 2017). Besides the fact of being intrinsically destructive and disruptive, these tools do not allow a high frequency in sampling repetition for evident technological difficulties and also drive to ulterior misleading samplings outcomes in the case of repetition in a certain area, since this has been previously impacted. Actually, one should assume that, some parts of the ecosystems are also difficult to investigate with the novelobservational technologies presentedin this Thesis. Notwithstanding, one should bear in mind that for an evident reduction of observed seabed area, a dramatic increase in time coverage can be achieved. One limitation of the observational technologies proposed here is that the resolution of the used camera will prevent the observation of the entire community. In Chapter 3, I proposed a methodology that will notably reduce this flawless(see “recommendations and future researches” in the Section below). Together with that, it’s a fact that the scientific community only sampled a tiny fraction of the deep sea (5%; Ramirez-Llodra et al. 2010), hence, regardless of the used technology and methodology we are still far from being able to reach a complete description of deep-water and deep-sea ecosystems’ structures and dynamics. The continuous discovering of new habitats is far to be completed(Danovaro et al. 2017), with a number of species being continuously described and leaving accumulation curves unsaturated(Roberto Danovaro et al. 2010).In this context, the scientific community should not despise the considerable information that can be extracted from the use of observational technologies in terms of Discussion 124 perceived richness and evenness and how these respond upon the environmental modulation, if we formulate the specific questions in a correct manner. Although multiparametric fixed video-stations may lack of sufficient spatial coverage, their detailed imaging may represent a true advantage. For instance, these may help to understand community dynamics in relation with key changing habitat factors, elucidating new aspects of species ecological niches in continental margins. In Chapter 2, I focussed on small scale seabed surface use in moments of environmental crisis such as transient hypoxia and I could determine that the massive aggregation of squat lobster was not due to shoaling hypoxia forcing an upward migration only with a single small-scale fixed observatory platform and the support of bibliography although in other occasions this species has been reported to migrates due to shoaling hypoxia (Burd and Brinkhurst 1984, Chu and Tunnicliffe, 2015). I highlighted the value of observational technologies in contraposition with other sampling methodologies, such as trawling which, although presenting much larger space coverage (therefore with apparently less limited ecological representativeness), are non-useful to describe phenomena such as massive aggregations for depth oriented migrations of benthic fauna in response to shoaling hypoxia. The species intra- and interspecific competition for space use at small-scale was described during the upward migration thanks to spatially calibrated methods for counting (i.e. with metric bars within a fixed and constant field of view, as deployed by ROV). I also succeeded in describing the dominance hierarchy in the spatial occupancy between two species and in relation to decreasing dissolved oxygen. Such a methodology could be extended to other key habitats in Pacific continental margins (e.g. Oxygen Minimum Zones) and considering other essential ocean variables (EOV; Grimes 2014) for which data on a biodiversity crisis driven by upward benthos migrations are still scant. In Chapter 1 I stresses how to overcome the limited representational power of ecological data from fixed observational nodes by integrating them into monitoring-coordinated spatial networks. This can be of utility to detect and follow benthopelagic and nektobenthic migrations. Good evidence of this could be find in the phase delay in the 24-h averaged number of counts of sablefish (Anoplopoma fimbria) between cameras at different depths along the canyon axis, as a putative product of a nektobenthic migration. Sablefish migrations along the canyon have been corroborated by adding another observing station (i.e. the crawler) in a shallower canyon head position by Chatzievangelou et al. (2016). Here, I also demonstrated the potential applicability of observational technologies to extract demographic indices for benthic species, by disentangling the population structure of sablefish by using coordinated, video-based, class-sizing of individuals appearing as parallel within the field of view. This methodology potentially has high socio-economic implications if the target species are directly studied that way to improve fishery management. Networks could be implemented by adding moving platforms such as crawlers, conferring an adaptive character to the network geometry itself. Discussion 125 The heterogeneity of physical and chemical conditions may lead to a small-scale spatial distribution of species that is an indicator of animal’s physiological tolerance to abiotic factors, resource ability and biotic factors (Lelièvre et al., 2017). Therefore, there is a need of a technology and methodology that is able to achieve enough spatial coverage without rescind from long-term temporal resolution. In Chapter 3, the rover used corresponds to a new generation of semi mobile multiparametric video-platforms (i.e. Internet Operated Vehicles, IOVs) as hydrides in between epibenthic moving ROVs and fully benthic and fixed cabled observatories or landers. IOVs could provide “higher-spatial” (60 m2) coverage compared to observatories, still with at high-temporal resolution (i.e. hourly data by 14 months). One difficulty was related to the lack of automation in crawler nested routines, which obliged the operator to drive it by certain time intervals, hence liming the frequency for data acquisition to central data within each week. In the future, the elimination of the tether (platform recharge and data transference will occur in an inductive fashion) and the automation in driving, will surely increase the spatiotemporal representation power of acquired data. Further, I acquire highly valuable ecological data that endorse this technology and methodology for ecological studies in problematic areas associated with deep sea environments, in this particular case, in a deep-sea and cold seep environment. A detailed explanation of the achieved ecological goal achieved is provided under the sub-section 1.2. Another technical limitation of the used monitoring technologies is that imaging stillrequireslighting, which may have an impact on the number of counts (Aguzzi et al. 2015; Doya et al. 2014). I observed a light-biasing effect on sablefish counts in Chapter 1 although globally, both attraction and subsequent retreat seem to occur over a temporal scale of few tens of seconds and hence likely compensate each other, so an overall biasing may be absent in our video samples (i.e. footages lasted 30 s). Additionally, this was a constant input along the study, thus, I assumed that any observed variations in the number of counts would be attributable to external factors and not to an environmental memory (i.e. blurred by the 30 min time-lapse acquisition procedure implying light on only at moment of filming). In any case, lighting on did not disrupt the occurrence of populational rhythms as we could appreciate in all the Chapters of this Thesis, being able to detect several behavioral temporal patterns (see sub-section 2.3.). In Chapter 3, we noticed a similar behavior of sablefish with the rover, thus we limited the light contamination by switching off the lights when the rover was not moving and performed similar lighting effect analysis of those performed in Chapter 1 (data not shown). An important methodological limitation was related with the scarcity of animals within the field of view which could limitingthe quality of time series for the disclosure of behavioral rhythms. When counts for a species were scant, behavioral rhythms analysis could be anyway performed by joining counts into larger time-spans (i.e. averaging or summing) per day. Since each species has its own behavioural rhythm, it is difficult to determine a minimum number of counts needed to ensure a successful analysis with periodogram and waveform. Anyway, some species concentrate their counts in a certain phase of the day, being the time-series analysis efficient in discriminating “squared” datasets with Discussion 126 clear peaks from troughs. Nevertheless, the data processing using in a combined fashion different methodologies inspired from chronobiology (seechapter 1) can make hint on activity rhythms, when there is a scares number of counts for a species. It was the case of the undetermined crabs (see chapter 1), with which the waveform analysis could detect a nocturnal activity that was not detected in the periodogram analysis virtually due to their reduced counts. A problem of small scale imaging analyses is the distortion of the field of view depth due to the angle of cameras which is oblique (around 45º). I focussed on that aspect in Chapter 2. I solved the perspective issue by creating a digital grid from a real grid temporarily placed on site and obtaining a correction factor for the vertical component since in this particular case, the horizontal component of the image was considered minimal. This methodology is appropriate as long as the camera either doesn’t change its orientation or perform a predefined path. Finally, another factor that could be affecting the accuracy in counts was transient turbidity. I solved that in two ways, either increasing the number of sampled images in a first screening of the images (in the case of Chapter 2) or eliminating fractions of the videoed samples (like in Chapter 3). As a conclusion of the aforementioned sub-section (1) with minimal habitat disturbance imagery provides valuable cost-efficient assessment of general ecosystem functioning at large temporal scales (2) observational technologies improve their spatial resolution by an appropriate coordination with each other into a network and (3) despite poor taxonomic resolution of image-derived data the sampling methodology is sufficient to infer aunal composition, richness, species abundances and evenness, and other valuable ecological information of mobile megafauna (although there are no captured specimens concomitantly, comparisons with guides and previous sampling surveys can be done).In this sense, present findings represent a step forward to settle and systematize an adequate methodology to monitor biological rhythms in the deep sea highlighting the problems and advantages of such methodologies to solve how important is to consider temporal dependent universal phenomena in marine ecosystem. Finally, automatization of data processing using the methodology inspired by chronobiology here exposed (i.e. periodogram, waveform and phase integrated chart analyses…) is of relevance for the realtime and continuous monitoring of deep-sea diel and seasonal dynamics. 2. Recommendations and future researches Although it would be desirable to perform behavioral,physiologic, and chronobiologic studies in laboratory conditions, the reality is that is not possible to maintain in captivity must of marine species. In addition, laboratory studies are reductionist and lacking of a functional vision, with very few testing rhythms responses with few environmental conditions. Also in the case of modelling only certain conditions are tested, though in essence, animals are not found in laboratories or in mathematical approximations but in the field which is a much complex scenario. Big ecological data approach with the Discussion 127 use of new transversal subsea observatories will not replace other disciplines and methodologies used nowadays but will be an important complement to later deepen our knowledge with more focused laboratory approaches. Taking into account the enormous benefices of observational technologies mentioned in the present work, and the difficulties of getting permanent funding for the stablishing long-standing monitoring programs, I adhere to thediscovery and monitoring strategy protocol proposed by Juniper and Tunnicliffe (1997)made by consecutive phases: determining the composition and richness of the system, determining its communities structure and finally the dynamics of the system. In that sense, I would add that being marine ecological monitoring a matter of economic resources, the selection of surveyed ecosystems must be carefully agreed with relevant end-users such as the industry, which will provide those deployment and technological assets allowing scientists to extend their observation capability form the continental margins to the abyssal plains(Danovaro et al. 2017).In order to pursuit this approach, testing of methodologies and technologies is needed and results must be shared in an inverse pyramid from research centres and universities to national and international organizations that will envisage the creation of protocols for environmental evaluation of impact, supervising their application at the development on maritime activities. The major impediment for that is the high economic effort to be done in the construction and maintenance of observational technologies. Nowadays, fishing, mining, oil and gas industries are the main source of income for this purpose being ecological research considered as subservient to overall economic interest. Besides these difficulties, we are facing highly performant technology not only concerning the purely scientific and industry interests but also a virtually endless source of didactic and citizen awareness. In this context, anaware society (1) puts pressure on industry and governments to prevent the indiscriminate exploitation of marine services (2) is much more prone to invest in research and conservation purpose than an unconscious population. Because of that, and ideally considering the independence of research from particular economic interest as a priority of general interest, future researches would need to concomitantly achieve scientific goals and promote the general public’s awareness. A prove of the remarkable pow“power of the image” can be seen in the use of images of Chapter 4, the imaging material of which has been obtained for a Discovery Channel science documentary. Additionally, this same whale has been monitored using 3D technology to perform high-spatial resolution, long term ecological studies. My recommendation is to use this kind of images for virtual reality combined with long-term time-lapses in science disclosure within the context of an increasing naturalization of technologies in our societies. Regarding the scientific goals for future research, main efforts must be directed towards solving the problems of photic contamination at filmingand spatial underrepresentation, depending form used observational technologies. The researcher must pay special attention to each species reaction to different